Post on 29-Apr-2023
COHORT PROFILE
Cohort Profile The Diabetes Study of NorthernCalifornia (DISTANCE)mdashobjectives and designof a survey follow-up study of social healthdisparities in a managed care populationy
Howard H Moffet1 Nancy Adler2 Dean Schillinger3 Ameena T Ahmed1 Barbara Laraia4
Joe V Selby1 Romain Neugebauer5 Jennifer Y Liu1 Melissa M Parker1 Margaret Warton1
and Andrew J Karter16
Accepted 11 February 2008
How did the study come aboutOne of the challenges of the national initiative HealthyPeople 20101 is to support interventions that willreduce social disparities in health While social dispa-rities such as differences in education income race orethnicity may affect health the mechanisms are poorlyunderstood If social disparities in health originate inchildhood are current social disparities in healthmodifiable and are they the responsibility of a medicalprovider or health plan Nonetheless modifiable factorsmay exist at the individual neighbourhood or systemlevel that mediate (explain) social disparities in healthand that may be suitable targets for interventions
aiming to reduce disparities Our aim was to surveyand prospectively follow a large diverse and well-characterized population with diabetes and to collectdata on risk factors which may affect diabetes healthoutcomes but which may differ substantively inprevalence or effect size across ethnic groups oreducational levels
The 2002 Institute of Medicine report lsquoUnequalTreatment Confronting Racial and Ethnic Disparitiesin Health Carersquo2 detailed the socioeconomic fragmen-tation of health care quality and access and itsdifferential negative impact on minorities As accessto health care is an important determinant of healthoutcomes and is in turn associated with ethnicity andsocioeconomic position it represents a potent sourceof confounding bias in population-based studies ofsocial health disparities
At the population level identifying antecedents anddeterminants of social differences in disease progres-sion remains a challenge There are nationallyrepresentative surveys (eg National Health andNutrition Examination Survey) that identify membersof the general population with diabetes and that canestimate risk factors that are associated with diabetescomplications but these cohorts are often cross-sectional or panel studies and have limited ability totrack continuous changes in individual-level healthcare outcomes and risk factors over timeLongitudinal population-based studies often havetoo few minority subjects with diabetes to revealracial or ethnic differences
The relatively uniform access to care in a managed carehealth plan such as Kaiser Permanente providesan advantageous setting in which to conduct a long-itudinal study of social disparities in health and studyfindings may be compared with population-based
y Additional information about DISTANCE will be published athttpdistancesurveyorg
Corresponding author Kaiser Permanente ndash Division ofResearch 2000 Broadway Oakland CA 94612 USAE-mail HowardHMoffetkporg
1 Kaiser Permanente ndash Division of Research 2000 BroadwayOakland CA 94612 USA
2 Department of Psychiatry Department of Pediatrics andCenter for Health and Community University of California3333 California Street Suite 465 San Francisco CA 94118USA
3 University of California San Francisco School of MedicineCenter for Vulnerable Populations San Francisco GeneralHospital San Francisco CA 94110 USA
4 Department of Medicine and Center for Health andCommunity University of California 3333 California StreetSuite 465 San Francisco CA 94118 USA
5 Division of Biostatistics School of Public Health University ofCalifornia Berkeley CA 94704 USA
6 Department of Epidemiology School of Public Health ampCommunity Medicine University of Washington Seattle WA98195 USA
Published by Oxford University Press on behalf of the International Epidemiological Association
The Author 2008 all rights reserved Advance Access publication 7 March 2008
International Journal of Epidemiology 20093838ndash47
doi101093ijedyn040
38
studies where health care access and quality may varyby socioeconomic position
We established the Kaiser Permanente NorthernCalifornia Diabetes Registry (lsquoRegistryrsquo) in 1993 usingstandardized criteria (Table 1) to identify andprospectively follow members with diabetes tomeasure prevalence and incidence of diabetes andits co-morbidities to understand factors associatedwith disease progression and complications and toevaluate health care processes and outcomes TheRegistry has an estimated sensitivity of 99 based onchart review validation (unpublished results) Weconducted the first survey of the Registry in 1994ndash97(Diabetes Registry Questionnaire) among all Registrymembers over 19 years of age The primary goal ofthat survey was to capture individual-level informa-tion on the clinical characteristics of diabetes age atdiagnosis ethnicity education health-related behav-iours and diabetes family history There were 77 726respondents (83 response rate among eligiblemembers) and that survey cohort has been the basisfor numerous publications regarding the epidemiolo-gic and health services aspects of diabetes3ndash9
We previously reported findings regarding ethnicdisparities in the incidence of myocardial infarctionstroke congestive heart failure end-stage renaldisease and lower-extremity amputation among dia-betic African American Asian Latino and Caucasianmembers of Kaiser Permanente Northern California(lsquoKaiserrsquo)7 a population with uniform access tocare10 Socioeconomic disparities in diabetic complica-tions based on educational attainment and incomehave been reported in other populations1112
The National Institutes of Health provided fundingto the Kaiser Division of Research and the Universityof California San Francisco School of Medicine toconduct the Diabetes Study of Northern California(DISTANCE) This study was approved by theirrespective Institutional Review Boards
What does it coverWe developed and implemented DISTANCE as asurvey follow-up cohort study13 to assess a widerange of social and behavioural factors that wehypothesized to be potentially confounding moderat-ing or mediating factors associated with socialdisparities in diabetes-related outcomes TheDISTANCE Survey consisted of 184 questions andthe domains included demographics socioeconomicsclinical profile health behaviours treatment adher-ence diabetes knowledge psychosocial characteristicspatientndashprovider relationship and quality and accessto care (Table 2) Every effort was made to create asurvey with language font and layout that would beaccessible to non-English speakers and persons of lowliteracy or visual acuity1415 The complete survey isavailable as Supplementary data and at httpdistancesurveyorg
Survey contentOur two primary exposuresattributes of interest wereself-reported raceethnicity and education (years ofeducation and degrees earned) Other social factorssurveyed included country of birth acculturationand language fluency subjective socioeconomic posi-tion1617 and several objective measures of socioeco-nomic position educational attainment of self1819
parents and spouse20 individual employment2122
household income23 assets2324 marital status andfamily size
The survey included extensive questions on healthbehaviours and self-reported symptoms not readilyavailable in administrative data diet25 physicalactivity26 smoking21 alcohol consumption27 self-monitoring of blood glucose (SMBG)25 oralhealth2829 self-examination of feet25 medicationadherence253031 multivitamin use32 TV watching20
attitudes and beliefs about diabetes23 internal vs
Table 1 Inclusion and exclusion criteria for the Kaiser Permanente Northern California Diabetes Registrya
Inclusion in the Registry is based on
(1) Self-report from Kaiser member health surveys or(2) Pharmacy utilization of any insulin or oral hypoglycemic agents or(3) Laboratory results of glycosylated haemoglobin A1C (A1C) 570 or(4) Two or more abnormal glucose values (either fasting glucose5126 mgdl or random glucose5200 mgdl) or(5) Outpatient utilization (5 2 visits for diabetes) or(6) Inpatient hospital utilization (primary hospital discharge ICD9250xx)
Exclusion criteria are
(1) Having no diabetes inclusion indicators during a sum total of two years of Kaiser membership after the date theywere identified in the Registry or
(2) Identified only by a ICD9 code of 6488 (gestational diabetes) or(3) Identified due to pharmacy utilization of metformin or thiazolidinedione only (no other indicators) and also
diagnosed with polycystic ovary syndrome HIV lipodystrophy metabolic syndrome pre-diabetes or reproductiveproblems
aThe Registry is updated annually
COHORT PROFILE DISTANCE 39
Table 2 Clinical behavioral psychosocial and quality and access indicators
Domain Survey scales or variables
Demographic Age
Sex
Race
Nativity
Language
Socioeconomic Education
years of school
highest degree attained
quality-weighted attainment
functional health literacy
Income
Occupation
Clinical profile Laboratory values
Blood pressure
Body mass index
Inpatient and outpatient utilization
Pharmacy utilization
Type of diabetes
Co-morbidity Index
Self-reported clinical characteristics
Health behaviors AdherenceKnowledge
Self-care behaviors
smoking
alcohol
exercise
medication refill adherence and discontinuation
self-monitoring of blood glucose
insulin injection frequency
appointment-keeping adherence
health screening adherence
sleep adequacy
Diabetes knowledge
Psychosocial Characteristics Big Five personality facets of conscientiousness and neuroticism
Perceived stress
Depressive and anxiety disorders
Social support and social networks
Discrimination
Socioeconomic Position (SEP) Individual-level socioeconomic indicators
Neighbourhood SEP contextual variables
Subjective measures of SEP
Financial barriers
Patient-provider relationship Patient-provider communication
Primary care provider characteristics
Quality and access to care Processes of care
Referrals for specialty care
40 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
external locus of control3334 social support2335ndash37
health status3839 stress40 depression2241ndash43 sleepquality44ndash46 chronic pain38 personality traits47ndash49
health literacy50ndash52 diabetes knowledge5354 medicaland dental coverage medical costs and ability topay55 medical visit travel time43 language barriersdelayed treatment56 provider recommendationsregarding aspirin43 and SMBG43 foot exam byphysician43 quality of care43 discrimination5758
provider communication and interactions5960 trustin provider6162 and self-reported height weight43
symptoms complications43 erectile dysfunction orfemale urinary incontinence
Survey modesThe survey included four modes of administration (i)a computer-assisted telephone interview (CATI)administered by a third party (ii) a password-enabled internet-accessible survey (lsquoweb surveyrsquo)maintained on a secure server at the Kaiser Divisionof Research (iii) a self-administered written surveyor (iv) a short version of the written survey Thecontent of each survey mode was identical except forslight adjustments in wording and logic patterns asneeded and was estimated to take 45ndash60 min tocomplete the short written version was abridged andcontained 40 questions The written and web surveyswere in English only but the CATI was available inEnglish Spanish Cantonese Mandarin and Tagalogusing certified translations of an English script andwas intended to maximize accessibility to those withlanguage barriers or limited English literacy orfluency
Other baseline dataWe geocoded the home addresses of all respondentsand non-respondents and linked their census blockgroup and tract to 2000 Census data These censuscharacteristics were then factor-analysed and wedeveloped an eight-item lsquodeprivation indexrsquo as a wayof summarizing the many indicators The deprivationindex has excellent internal reliability (Cronbachalphafrac14 093) with each of the eight items loadingalmost equally onto the underlying construct
Baseline clinical characterizations were obtainedfrom the extensive Kaiser administrative databasesincluding smoking status body mass index currentpharmacotherapy utilization and adherence63 labora-tory findings history of co-morbid events andprocedures use of emergency room outpatient andinpatient health services costs outpatient and inpa-tient risk scores based on health care utilization andseverity of disease64 and end-stage renal diseaseregistry (linked to the United States Renal DataSystem) These data are available for all subjectsboth respondents and non-respondents and allow fora substantive assessment of survey response bias
Who is in the sampleThis study took place within the Kaiser FoundationHealth Plan the largest not-for-profit managed healthcare company in the United States65 Study subjectswere members of Kaiser Permanente NorthernCalifornia (KPNC) which provides comprehensivemedical services to over 32 million members (as ofJanuary 2005) in the San Francisco Bay andSacramento metropolitan areas or 25ndash30 of theregionrsquos population KPNC members are predomi-nantly employed or retired individuals and theirfamilies and closely approximate the general popula-tion ethnically and socioeconomically except forthe extreme tails of income distribution66667 TheDiabetes Registry consisted of 199 123 members as ofJanuary 1 2005 and from this we selected anethnically stratified random sample of 40 735 healthplan members aged 30ndash75 at baseline to receive theDISTANCE Survey 6871 African American (17)11 197 Asian (27) 4233 Caucasian (10) 7018Latino (17) and 11 417 members of unknownethnicity (28) The DISTANCE Survey was in thefield from May 5 2005 until December 31 2006
The survey was completed by 20 188 persons(Table 3) The participation by ethnicity was 3420African Americans (169) 2312 Asian (114) 4602Caucasians (228) 2404 Filipinos (119) 3717Latinos (184) 2222 multi-racial (110) and 1511South Asian Pacific Islander Native America Eskimoor otherunknown (75) The distribution by modewas 10 429 CATI (517) 4288 written survey(212) 2393 short version (118) and 3078 web(152) Using an algorithm endorsed by the Councilof American Survey Research Organizations (CASRO)if persons unable to be contacted had the same rate ofeligibility as those contacted and were counted in thedenominator the survey response rate was 62 Ofthe 20 188 respondents to this current survey 4524subjects (22) had also responded to the priorDiabetes Registry Questionnaire (1994ndash97)
What has been measuredThe entire cohort has been characterized usingdemographic clinical and behavioural data fromadministrative records and census data
African American respondents were more likely tobe female to have had English as their preferredlanguage to have had elevated levels of low-densitylipoprotein (LDL) uncontrolled blood pressure werecurrent smokers with higher co-morbidity scores andliving in a deprived neighbourhood
Asian respondents were more likely to be male over60 years of age to not have had English as theirpreferred language to have had lower levels of LDLbetter blood pressure control lower missed appoint-ment rate better medication adherence were non-smokers not practicing SMBG with lower co-morbidity
COHORT PROFILE DISTANCE 41
Ta
ble
3C
hara
cter
isti
cso
fsu
rvey
resp
on
den
tsan
dn
on
-res
po
nd
ents
Re
spo
nd
en
tsn
()
or
Me
an
plusmn(S
D)
No
n-r
esp
on
de
nts
Afr
ica
nA
me
rica
n
Asi
an
(Ch
ine
se
Ko
rea
n
Vie
tna
me
se
Ja
pa
ne
se)
Ca
uca
sia
nF
ilip
ino
La
tin
oM
ult
ira
cia
lO
the
rau
nk
no
wn
All
resp
on
de
nts
No
nre
spo
nd
en
tsN
um
be
r(
)o
rM
ea
nplusmn
(SD
)
n(r
ow
)
34
20
(16
9)
23
12
(11
4)
46
02
(22
8)
24
04
(11
9)
37
17
(18
4)
22
22
(11
0)
15
11
(75
)2
01
88
(10
0)
20
54
7(1
00
)
Su
rve
yM
od
e
Tel
eph
on
eIn
terv
iew
18
06
(52
8)
97
1(4
20
)2
05
1(4
46
)8
46
(35
2)
22
29
(60
0)
15
66
(70
5)
96
0(6
35
)1
04
29
(51
7)
0
Wri
tten
-lo
ng
77
1(2
25
)5
75
(24
9)
97
3(2
11
)7
23
(30
1)
67
6(1
82
)3
46
(15
6)
22
4(1
48
)4
28
8(2
12
)0
Wri
tten
-sh
ort
43
9(1
28
)2
59
(11
2)
46
9(1
02
)5
16
(21
5)
44
8(1
20
)1
08
(49
)1
54
(10
2)
23
93
(11
8)
0
Web
40
4(1
18
)5
07
(21
9)
11
09
(24
1)
31
9(1
33
)3
64
(98
)2
02
(91
)1
73
(11
4)
30
78
(15
2)
0
Fem
ale
19
39
(56
7)
10
19
(44
1)
20
35
(44
2)
12
46
(51
8)
18
86
(50
7)
11
22
(50
5)
59
3(3
93
)9
84
0(4
87
)9
53
7(4
64
)
Ag
e(y
ea
rs)
30
ndash4
42
74
(80
)1
37
(59
)3
82
(83
)1
68
(70
)5
64
(15
2)
25
3(1
14
)1
89
(12
5)
19
67
(97
)2
58
0(1
26
)
45
ndash5
91
34
7(3
94
)9
34
(40
4)
17
94
(39
0)
11
06
(46
0)
16
39
(44
1)
94
3(4
24
)6
95
(46
0)
84
58
(41
9)
87
55
(42
6)
60thorn
17
99
(52
6)
12
41
(53
7)
24
26
(52
7)
11
30
(47
0)
15
14
(40
7)
10
26
(46
2)
62
7(4
15
)9
76
3(4
84
)9
21
2(4
48
)
Lan
gu
age
pre
fere
nce
En
gli
shsp
eak
ing
32
21
(94
2)
13
76
(59
5)
43
94
(95
5)
17
49
(72
8)
22
17
(59
6)
16
91
(76
1)
11
64
(77
0)
15
81
2(7
83
)1
72
12
(83
8)
Mea
nA
1C5
7
16
55
(54
0)
10
87
(50
1)
19
15
(46
8)
13
52
(60
4)
19
57
(58
6)
10
98
(56
2)
77
9(5
79
)9
84
3(5
41
)8
34
3(5
13
)
Mea
nL
DL5
10
01
25
0(4
17
)6
26
(28
9)
13
91
(34
6)
63
4(2
86
)1
20
2(3
71
)7
36
(38
1)
48
2(3
63
)6
32
1(3
53
)6
90
9(4
24
)
Un
con
tro
lled
blo
od
pre
ssu
re(S4
13
0o
rD4
80
)1
96
6(6
14
)9
13
(42
8)
22
64
(54
7)
10
41
(46
5)
17
46
(51
2)
10
89
(54
2)
68
1(4
95
)9
70
0(5
24
)9
58
1(5
43
)
Mis
sed
ap
po
intm
ent
rate
(per
year)
01
7plusmn
(01
8)
01
0plusmn
(01
4)
01
3plusmn
(01
6)
01
3plusmn
(01
7)
01
8plusmn
(01
9)
01
7plusmn
(01
9)
01
8plusmn
(01
9)
01
5plusmn
(01
7)
01
8plusmn
(02
2)
Po
or
med
icati
on
ad
her
ence
b9
55
(37
1)
46
1(2
51
)9
61
(28
2)
59
2(2
93
)1
24
7(4
28
)6
36
(36
4)
42
4(3
51
)5
27
6(3
36
)5
10
8(3
90
)
Sm
ok
ing
(cu
rren
t)3
48
(10
2)
93
(40
)4
10
(90
)1
47
(61
)2
45
(66
)2
00
(91
)1
05
(70
)1
54
8(7
7)
18
29
(90
)
Pra
ctic
esS
MB
G1
57
3(4
60
)1
00
4(4
34
)2
24
6(4
88
)1
11
5(4
64
)1
62
2(4
36
)1
02
5(4
61
)6
23
(41
2)
92
08
(45
6)
65
32
(31
8)
Ou
tpati
ent
risk
sco
rec
24
(21
)1
8(1
1)
21
(17
)1
8(1
0)
19
(14
)2
2(1
6)
19
(14
)2
0(1
6)
18
(15
)
Inp
ati
ent
risk
sco
rec
26
(42
)1
5(2
0)
22
(33
)1
4(2
1)
18
(30
)2
2(3
9)
17
(29
)2
0(3
2)
18
(34
)
Lin
gu
isti
call
yis
ola
ted
nei
gh
bo
urh
oo
dd
10
5(3
1)
11
5(5
0)
55
(12
)1
14
(48
)2
81
(76
)1
13
(51
)6
2(4
1)
84
5(4
2)
10
35
(51
)
Eco
no
mic
all
yd
epri
ved
nei
gh
bo
urh
oo
de
82
7(2
43
)1
24
(54
)3
46
(76
)1
48
(62
)6
68
(18
2)
32
3(1
47
)1
67
(11
2)
26
03
(13
0)
27
24
(13
4)
Wo
rkin
gcl
ass
nei
gh
bo
urh
oo
df
14
40
(42
2)
40
1(1
74
)1
16
8(2
58
)8
84
(37
0)
17
49
(47
6)
86
8(3
94
)4
64
(31
0)
69
74
(34
9)
75
66
(37
2)
aIn
clu
des
Paci
fic
Isla
nd
ers
Nati
veA
mer
ican
sS
ou
thA
sian
o
ther
un
spec
ifie
dan
du
nk
no
wn
(did
no
tan
swer
)b5
20
co
nti
nu
ou
sm
edic
ati
on
gap
s6
3
c Th
eyare
wei
gh
ted
by
pati
entrsquo
sh
ealt
hca
reu
tili
zati
on
an
dse
veri
tyo
fd
isea
se6
4
d5
25
o
fh
ou
seh
old
sin
cen
sus
blo
ckgro
up
are
lin
gu
isti
call
yis
ola
ted
e5
20
o
fh
ou
seh
old
sin
cen
sus
blo
ckgro
up
are
livi
ng
bel
ow
po
vert
yli
ne
f 56
6
of
per
son
sin
cen
sus
blo
ckgro
up
are
emp
loye
din
wo
rkin
gcl
ass
occ
up
ati
on
42 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
scores and not living in an economically deprived orworking class neighbourhood
Caucasian respondents were more likely to completethe survey online to have had lower mean A1Cpracticing SMBG and not living in linguisticallyisolated neighbourhood
Filipino respondents were more likely to participatevia written survey to have been older than 45 yearsof age to have had lower levels of LDL and lowerco-morbidity scores
Latino respondents were more likely to participatevia CATI to have been younger than 45 years of ageto not have English as their preferred language tohave had higher mean A1C higher rates of missedappointments poorer medication adherence andliving in a linguistically isolated or working classneighbourhood
What outcomes will be measuredduring follow-upThis cohort will be the basis for longitudinal evalua-tions of a wide range of clinical outcomes associatedwith diabetes and are powered to evaluate the ethnicand educational disparities in diabetes-related com-plication rates (eg myocardial infarction strokeheart failure kidney failure and amputation) andmortality after a 25-year follow-up Clinical andadministrative follow-up data will be captured fromthese same sources as baseline plus deaths from thestate mortality files Data from all sources combinedmay offer insights into how differences at the patient-level (eg differences in behaviours type of therapyintermediate health status and psychosocial factorsand competing demands) provider-level (eg trust inprovider racial concordance and patient-providercommunication) and system-level (eg differentialimpact of cost-sharing) may lead to social disparitiesin diabetes health end-stage diabetes complicationsand mortality
What is the anticipated attritionOne of the strengths of this diabetes registry is thevery low turnover rate (5 discontinue membershipeach year) affording minimal loss to follow-up Onaverage the duration of Kaiser membership amongour diabetes cohort members is 89 years (SDfrac14 66)
What are the strengths andweaknessesA significant weakness is that written and websurveys were only offered in English Howevertelephone interviewers made the initial efforts tocontact subjects who had the opportunity to complete
interviews in Spanish Mandarin Cantonese orTagalog Offering the survey by oral interview inmultiple languages was intended to mitigate thelanguage andor literacy barriers but we still observedlower participation rates by those with less educationand modest response differences by race On a morefavourable note 84 of non-respondents were identi-fied by the health plan as having English as theirpreferred language suggesting that language alonewas not a major barrier to participation
The response rate was as expected lower than ourprevious Diabetes Research Questionnaire (1994ndash97)A recent survey conducted in the Translating ResearchInto Action for Diabetes (TRIAD) study amongmanaged care populations across the United Statesincluding Kaiser43 had a response rate of 50 TheBehavioural Risk Factor Surveillance System alsoexperienced a decline in median response ratesfrom 632 in 1996 to 535 in 200168 Privacyconcerns and competition from telemarketing haveprobably played a role in declining response ratesHowever if generally lower response rates have beenobserved in minority populations the response ratefor this survey may be considered favourable giventhat 77 of our sample were racial or ethnic minorityhealth plan members
One strength of this study is the relatively uniformaccess to care provided by membership in thisintegrated health plan Thus this study largely avoidsdisparities in access and quality which are potentsources of confounding bias in population-basedstudies of social health disparities Another strength isthe quality and richness of the clinical follow-up dataAdministrative and clinical data are available directlyfrom the health plan and are considered to be reliableand complete Each telephone interview was conductedby professional interviewers employed trained andsupervised by the Public Health Institute SurveyResearch Group in Sacramento CA which has itsown internal quality controls Written surveys (long-and short-forms) were coded and edited using anextensive data dictionary and detailed coding rules tostandardize and clarify responses before being sentto the Data Entry department at the KaiserDivision of Research which conducted double dataentry The survey data quality is limited by beingself-reported
The diversity size and wealth of data included inthe DISTANCE cohort make it suitable for theprospective study of a wide range of social disparitiesin the processes and outcomes of diabetes health careInformation about childhood socioeconomic positionand educational attainment will facilitate a life courseapproach6970
Analysis of response biasResponse bias is a concern in any survey Surveys aresubject to response bias if participation is associatedwith the risk factors (exposures) andor diseases
COHORT PROFILE DISTANCE 43
(outcomes) under investigation In this study popula-tion we have substantial data on non-respondentsincluding Kaiser administrative data co-morbidityand census data on neighbourhood characteristicsallowing us to compare hypothesized relationships inrespondents vs non-respondents and adjust for thispotential response bias71 While participation wassomewhat lower among minorities and those withless education overall respondents and non-respon-dents were quite similar
We had pre-baseline reported race data for 29 319(72) members of the DISTANCE cohort andobserved differences in response rates by raceAfrican Americans 52 Asians 46 Caucasians63 Latinos 53 unknown 45 (P lt 00001) Wealso had educational attainment data for 4524subjects (22) who had completed the previoussurvey in 1994ndash97 and observed that subjects with ahigh school education or less participated at a lowerrate than those with more than a high schooleducation (54 vs 60) (P lt 00001)
A more complete view of response bias was obtainedfrom health plan administrative data and census datawhich was available for all members of the cohortWe found few notable differences between respon-dents and non-respondents in demographic clinicalbehavioural or neighbourhood characteristics Com-pared with respondents non-respondents had highermean LDL levels (993 vs 943 mgdl) and were lesslikely to practice SMBG (32 vs 46) However aportion of these unadjusted differences are likely dueto age differences in respondents vs non-respondents
We conducted an assessment of response bias bycomparing differences in A1C across ethnic groupsseparately among respondents and non-respondentsWe specified a logistic regression model of poorglycemic control which regressed the outcome A1C47 on ethnicity age and sex and an interactionterm (ethnicity survey response) Although respon-dents consistently had higher A1C than non-respon-dents ethnic differences in the relationship betweenethnicity and A1C did not differ between respondentsand non-respondents (Pfrac14 055) We conducted asimilar analysis with the subjects for whom we hadeducational attainment data from a previous surveyand again found no substantive response bias(Pfrac14 028) While characteristics differ betweenrespondents and non-respondents (response bias)the associations with outcomes of interest (slopes)are much less vulnerable to such bias7273
Statistical methodsAnalyses of DISTANCE data will focus on associationsrather than descriptive characterization We willemploy a modelling approach using inverse weightingestimation of marginal structural models72 to inves-tigate causal relationships in this study Thesemodels will include weights to adjust for the non-proportionate sampling fractions (eg over-sampling
of the minority ethnic groups) and response bias(eg giving greater weight to respondents whohave characteristics more similar to the non-respondents) Thus the proposed estimating approachwill adjust for confounding and selection biassimultaneously73ndash77
How can I collaborate Where canI find out moreWe are not free to release participantsrsquo personal dataunder our promise to them regarding confidentialityHowever the DISTANCE steering committee is inter-ested in collaborations with external researchersDISTANCE investigators are particularly interested incomparing social disparities observed in this insuredpopulation with uniform access to care to disparitiesobserved in population-based samples where qualityand access to care vary widely by social strataRequests for collaboration must be submitted to thedirector of the central coordinating centre (corre-sponding author) Before collaborations can beinitiated proposals require review and approval bythe DISTANCE Publications and Presentations (PampP)Committee This committee was formed to (i) Ensureaccurate uniform timely and high quality reportingof DISTANCE findings (ii) Preserve the scientificintegrity of the study and (iii) Safeguard the rightsand confidentiality of participants
Supplementary DataSupplementary data are available at IJE online
AcknowledgementsFunds were provided by National Institute ofDiabetes Digestive and Kidney Diseases [R01DK65664] and National Institute of Child Healthand Human Development [R01 HD046113] Dr DS isalso supported by a National Institutes of HealthMentored Clinical Scientist Award [K-23 RR16539]The Public Health InstituteSurvey Research Groupconducted our telephone interviews Gerard andAssociates provided certified translations of all patientmaterials and interview scripts KP Corporation (notaffiliated with Kaiser Permanente) provided printingand mailing services for the written surveys andother communications We thank Marcia Ewing andCarol Rabello for their diligent work on surveyoperations Most of all we thank all the studyparticipants
Conflict of interest None declared
References1 Healthy People 2010 U S Department of Health and
Human Services 2005 Available at httpwwwhealthypeoplegov(Accessed August 2 2007)
44 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
2 Smedley BD Stith AY Nelson AR Unequal TreatmentConfronting Racial and Ethnic Disparities in Health CareWashington DC Institute of Medicine NationalAcademy Press 2002
3 Karter AJ Newman B Rowell S et al Large-scalecollection of family history data and recruitment ofinformative families for genetic analysis J Reg Manag1998257ndash12
4 Karter AJ Ferrara A Darbinian J Ackerson LM Selby JVSelf-monitoring of blood glucose language and financialbarriers in a managed care population with diabetesDiabetes Care 200023477ndash83
5 Karter AJ Moffet HH Liu J et al Achieving goodglycemic control initiation of new antihyperglycemictherapies in patients with type 2 diabetes from theKaiser Permanente Northern California Diabetes RegistryAm J Manag Care 200511262ndash70
6 Karter AJ Ackerson LM Darbinian JA et al Self-monitoring of blood glucose levels and glycemic controlthe Northern California Kaiser Permanente Diabetesregistry Am J Med 20011111ndash9
7 Karter AJ Ferrara A Liu JY Moffet HH Ackerson LMSelby JV Ethnic disparities in diabetic complications inan insured population JAMA 20022872519ndash27
8 Karter AJ Parker MM Moffet HH et al Missedappointments and poor glycemic control an opportunityto identify high-risk diabetic patients Med Care200442110ndash15
9 Selby JV Karter AJ Ackerson LM Ferrara A Liu JDeveloping a prediction rule from automated clinicaldatabases to identify high-risk patients in a largepopulation with diabetes Diabetes Care 2001241547ndash55
10 Martin TL Selby JV Zhang D Physician and patientprevention practices in NIDDM in a large urbanmanaged-care organization Diabetes Care 1995181124ndash32
11 Chaturvedi N Jarrett J Shipley MJ Fuller JHSocioeconomic gradient in morbidity and mortality inpeople with diabetes cohort study findings from theWhitehall Study and the WHO Multinational Study ofVascular Disease in Diabetes BMJ 1998316100ndash5
12 Booth GL Hux JE Relationship between avoidablehospitalizations for diabetes mellitus and income levelArch Intern Med 2003163101ndash6
13 Kleinbaum DG Kupper LL Morgenstern H EpidemiologicResearch Principles and Quantitative Methods New York VanNorstrand Reinhold 1982
14 Dillman DA Mail and Internet Surveys The Tailored DesignMethod 2nd edn New York John Wiley 2000
15 Rudd RE Comings JP Hyde JN Leave no one behindimproving health and risk communication through att-ention to literacy J Health Commun 20038 (Suppl 1)104ndash15
16 Singh-Manoux A Adler NE Marmot MG Subjectivesocial status its determinants and its association withmeasures of ill-health in the Whitehall II study Soc SciMed 2003561321ndash33
17 John D Catherine T MacArthur Research Network onSocioeconomic Status and Health The MacArthur Scale ofSubjective Social Status San Francisco University ofCalifornia 1999 Available at httpwwwmacsesucsfeduResearchPsychosocialnotebooksubjectivehtml(Accessed July 12 2007)
18 Ross CE Wu CL Education age and the cumulativeadvantage in health J Health Soc Behav 199637104ndash20
19 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
20 White S Dillow S Key Concepts and Features of the 2003National Assessment of Adult Literacy (NAALS) (NCES2006ndash471) Washington DC Department of EducationNational Center for Education Statistics 2005
21 Botman SL Moore TF Moriarity CL Parsons VL Designand Estimation for the National Health Interview Survey(NHIS) 1995ndash2004 National Center for Health StatisticsVital Health Stat 2 20001301ndash31
22 Friedman GD Cutter GR Donahue RP et al CARDIAstudy design recruitment and some characteristics of theexamined subjects J Clin Epidemiol 1988411105ndash16
23 Ross CE Wu CL The links between education and healthAm Sociol Rev 199560719ndash45
24 Robert S House JS SES differentials in health by ageand alternative indicators of SES J Aging Health19968359ndash88
25 Toobert DJ Hampson SE Glasgow RE The summary ofdiabetes self-care activities measure results from 7studies and a revised scale Diabetes Care 200023943ndash50
26 Craig CL Marshall AL Sjostrom M et al Internationalphysical activity questionnaire 12-country reliability andvalidity Med Sci Sports Exerc 2003351381ndash95
27 Gordon AJ Maisto SA McNeil M et al Three questionscan detect hazardous drinkers J Fam Pract 200150313ndash20
28 National Health and Nutrition Examination Survey(NHANES) Questionnaire (2001-2002) Centers for DiseaseControl and Prevention (CDC) National Center for HealthStatistics (NCHS) Hyattsville MD US Department ofHealth and Human Services Centers for Disease Controland Prevention 2002
29 Behavioral Risk Factor Surveillance System (BRFSS)Centers for Disease Control and Prevention (CDC) NationalCenter for Health Statistics (NCHS) Hyattsville MD USDepartment of Health and Human Services Centers forDisease Control and Prevention 2003
30 Chien-Wen T Phillips RL Green LA Fryer GE Dovey SMWhat physicians need to know about seniors and limitedprescription benefits and why Am Fam Physician200266212
31 Morisky DE Green LW Levine DM Concurrent andpredictive validity of a self-reported measure of medica-tion adherence Med Care 19862467ndash74
32 Barringer TA Kirk JK Santaniello AC Foley KLMichielutte R Effect of a multivitamin and mineralsupplement on infection and quality of life A random-ized double-blind placebo-controlled trial Ann Intern Med2003138365ndash71
33 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
34 Peyrot M Rubin RR Structure and correlates of diabetes-specific locus of control Diabetes Care 199417994ndash1001
35 Wagenknecht LE Mayer EJ Rewers M et al The insulinresistance atherosclerosis study (IRAS) objectivesdesign and recruitment results Ann Epidemiol19955464ndash72
COHORT PROFILE DISTANCE 45
36 King DE Mainous AG III Pearson WS C-reactiveprotein diabetes and attendance at religious servicesDiabetes Care 2002251172ndash76
37 Berkman LF Syme SL Social networks host resistanceand mortality a nine-year follow-up study of AlamedaCounty residents Am J Epidemiol 1979109186ndash204
38 Ware JE Kosinski M Keller SD SF-12 How to Score theSF-12 Physical and Mental Health Summary Scales 2nd ednBoston The Health Institute New England MedicalCenter 1995
39 Ware JE Kosinksi M Dewey JE Gandek B How to Scoreand Interpret Single-Item Health Status Measures A Manual forUsers of the SF-8 Health Survey Lincoln RI QualityMetricIncorporated 2001
40 Cohen S Kamarck T Mermelstein R A global measure ofperceived stress J Health Soc Behav 198324385ndash96
41 Kroenke K Spitzer RL Williams JB The PHQ-9 validityof a brief depression severity measure J Gen Intern Med200116606ndash13
42 Spitzer RL Kroenke K Williams JB Validation and utilityof a self-report version of PRIME-MD the PHQ primarycare study Primary care evaluation of mental disordersPatient Health Questionnaire JAMA 19992821737ndash44
43 The TRIAD Study Group The translating research intoaction for diabetes (TRIAD) study a multicenter study ofdiabetes in managed care Diabetes Care 200225386ndash89
44 Ayas NT White DP Al Delaimy WK et al A prospectivestudy of self-reported sleep duration and incidentdiabetes in women Diabetes Care 200326380ndash84
45 Sickel AE Moore PJ Adler NE Williams DR Jackson JSThe differential effects of sleep quality and quantity onthe relationship between SES and health Ann NY Acad Sci1999896431ndash34
46 Moore PJ Adler NE Williams DR Jackson JSSocioeconomic status and health the role of sleepPsychosom Med 200264337ndash44
47 Benet-Martinez V John OP Los Cinco Grandes acrosscultures and ethnic groups multitrait multimethodanalyses of the Big Five in Spanish and English J PersSoc Psychol 199875729ndash50
48 Srivastava S John OP Gosling SD Potter J Developmentof personality in early and middle adulthood set likeplaster or persistent change J Pers Soc Psychol2003841041ndash53
49 Gosling SD Rentfrow PJ Swann WB A very briefmeasure of the Big-Five personality domains J Res Pers200337504ndash28
50 Schillinger D Grumbach K Piette J et al Association ofhealth literacy with diabetes outcomes JAMA2002288475ndash82
51 Schillinger D Barton LR Karter AJ Wang F Adler NDoes literacy mediate the relationship between educationand health outcomes A study of a low-income popula-tion with diabetes Public Health Rep 2006121245ndash54
52 Chew LD Bradley KA Boyko EJ Brief questions toidentify patients with inadequate health literacy Fam Med200436588ndash94
53 The American Association of Clinical EndocrinologistsMedical Guidelines for the Management of DiabetesMellitus The AACE System of Intensive DiabetesSelf-Management ndash 2002 Update Endocr Pract 20028(Suppl 1)40ndash82
54 Fitzgerald JT Funnell MM Hess GE et al The reliabilityand validity of a brief diabetes knowledge test DiabetesCare 199821706ndash10
55 Piette JD Heisler M Wagner TH Problems paying out-of-pocket medication costs among older adults withdiabetes Diabetes Care 200427384ndash91
56 Epidemiology of Diabetes Interventions andComplications (EDIC) Design implementation andpreliminary results of a long-term follow-up of thediabetes control and complications trial cohort DiabetesCare 19992299ndash111
57 Krieger N Sidney S Coakley E Racial discrimination andskin color in the CARDIA study implications for publichealth research Coronary artery risk development inyoung adults Am J Public Health 1998881308ndash13
58 Krieger N Smith K Naishadham D Hartman CBarbeau EM Experiences of discrimination validity andreliability of a self-report measure for population healthresearch on racism and health Soc Sci Med 2005611576ndash96
59 CAHPS 20 Questionnaires Crofton C AHCPR Pub No97-R079 Washington DC Agency for Health Care Policyand Research 1998
60 Hays RD Shaul JA Williams VS et al Psychometricproperties of the CAHPS 10 survey measures Consumerassessment of health plans study Med Care 199937MS22ndash31
61 Anderson LA Dedrick RF Development of the trust inphysician scale a measure to assess interpersonal trust inpatient-physician relationships Psychol Rep 1990671091ndash100
62 Thom DH Ribisl KM Stewart AL Luke DA Furthervalidation and reliability testing of the Trust in PhysicianScale The Stanford trust study physicians Med Care199937510ndash17
63 Steiner JF Koepsell TD Fihn SD Inui TS A generalmethod of compliance assessment using centralizedpharmacy records Description and validation Med Care198826814ndash23
64 Clark DO Von Korff M Saunders K Baluch WMSimon GE A chronic disease score with empiricallyderived weights Med Care 199533783ndash95
65 Kaiser Permanente Business Intelligence Report 2006Research and Markets 2006 Available at httpwwwresearchandmarketscomreportinfoaspreport_idfrac14451423(Accessed July 9 2007)
66 Krieger N Overcoming the absence of socioeconomic datain medical records validation and application of acensus-based methodology Am J Public Health 199282703ndash10
67 Hiatt RA Friedman GD Characteristics of patientsreferred for treatment of end-stage renal diseasein a defined population Am J Public Health 198272829ndash33
68 Porter S Jackson K Trosclaire A Pederson LL Office onsmoking and Health NCfCDPaHPC Prevalence of currentcigarette smoking among adults and changes in pre-valence of current and some day smoking ndash UnitedStates 1996ndash2001 MMWR 200352303ndash7
69 Smith GD Hart C Blane D Gillis C Hawthorne VLifetime socioeconomic position and mortality prospec-tive observational study BMJ 1997314547ndash52
46 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
70 Lynch JW Kaplan GA Shema SJ Cumulative impact ofsustained economic hardship on physical cognitivepsychological and social functioning N Engl J Med19973371889ndash95
71 Melton LJ3 Dyck PJ Karnes JL OrsquoBrien PC Service FJNon-response bias in studies of diabetic complicationsthe Rochester Diabetic Neuropathy Study J Clin Epidemiol199346341ndash48
72 Robins JM Association causation and marginal struc-tural models Synthese 1999121151ndash79
73 Hernan MA Hernandez-Diaz S Robins JM A structuralapproach to selection bias Epidemiology 200415615ndash25
74 Bodnar LM Davidian M Siega-Riz AM Tsiatis AAMarginal structural models for analyzing causal effects
of time-dependent treatments an application in perinatalepidemiology Am J Epidemiol 2004159926ndash34
75 Bryan JF Yu Z van der Laan M Analysis of longitudinalmarginal structural models U C Berkeley Division ofBiolstatistics Working Paper Series Paper 1201ndash29 2002
76 Neugebauer R van der Laan M Why prefer double robustestimators in causal inference J Stat Plan Inference2005129405ndash26
77 Wang Y Petersen ML Bangsberg D van der Laan MDiagnosing bias in the inverse probability of treatmentweighted estimator resulting from violation ofexperimental treatment assignment U C BerkeleyDivision of Biolstatistics Working Paper Series Paper 2111ndash25 2006
COHORT PROFILE DISTANCE 47
studies where health care access and quality may varyby socioeconomic position
We established the Kaiser Permanente NorthernCalifornia Diabetes Registry (lsquoRegistryrsquo) in 1993 usingstandardized criteria (Table 1) to identify andprospectively follow members with diabetes tomeasure prevalence and incidence of diabetes andits co-morbidities to understand factors associatedwith disease progression and complications and toevaluate health care processes and outcomes TheRegistry has an estimated sensitivity of 99 based onchart review validation (unpublished results) Weconducted the first survey of the Registry in 1994ndash97(Diabetes Registry Questionnaire) among all Registrymembers over 19 years of age The primary goal ofthat survey was to capture individual-level informa-tion on the clinical characteristics of diabetes age atdiagnosis ethnicity education health-related behav-iours and diabetes family history There were 77 726respondents (83 response rate among eligiblemembers) and that survey cohort has been the basisfor numerous publications regarding the epidemiolo-gic and health services aspects of diabetes3ndash9
We previously reported findings regarding ethnicdisparities in the incidence of myocardial infarctionstroke congestive heart failure end-stage renaldisease and lower-extremity amputation among dia-betic African American Asian Latino and Caucasianmembers of Kaiser Permanente Northern California(lsquoKaiserrsquo)7 a population with uniform access tocare10 Socioeconomic disparities in diabetic complica-tions based on educational attainment and incomehave been reported in other populations1112
The National Institutes of Health provided fundingto the Kaiser Division of Research and the Universityof California San Francisco School of Medicine toconduct the Diabetes Study of Northern California(DISTANCE) This study was approved by theirrespective Institutional Review Boards
What does it coverWe developed and implemented DISTANCE as asurvey follow-up cohort study13 to assess a widerange of social and behavioural factors that wehypothesized to be potentially confounding moderat-ing or mediating factors associated with socialdisparities in diabetes-related outcomes TheDISTANCE Survey consisted of 184 questions andthe domains included demographics socioeconomicsclinical profile health behaviours treatment adher-ence diabetes knowledge psychosocial characteristicspatientndashprovider relationship and quality and accessto care (Table 2) Every effort was made to create asurvey with language font and layout that would beaccessible to non-English speakers and persons of lowliteracy or visual acuity1415 The complete survey isavailable as Supplementary data and at httpdistancesurveyorg
Survey contentOur two primary exposuresattributes of interest wereself-reported raceethnicity and education (years ofeducation and degrees earned) Other social factorssurveyed included country of birth acculturationand language fluency subjective socioeconomic posi-tion1617 and several objective measures of socioeco-nomic position educational attainment of self1819
parents and spouse20 individual employment2122
household income23 assets2324 marital status andfamily size
The survey included extensive questions on healthbehaviours and self-reported symptoms not readilyavailable in administrative data diet25 physicalactivity26 smoking21 alcohol consumption27 self-monitoring of blood glucose (SMBG)25 oralhealth2829 self-examination of feet25 medicationadherence253031 multivitamin use32 TV watching20
attitudes and beliefs about diabetes23 internal vs
Table 1 Inclusion and exclusion criteria for the Kaiser Permanente Northern California Diabetes Registrya
Inclusion in the Registry is based on
(1) Self-report from Kaiser member health surveys or(2) Pharmacy utilization of any insulin or oral hypoglycemic agents or(3) Laboratory results of glycosylated haemoglobin A1C (A1C) 570 or(4) Two or more abnormal glucose values (either fasting glucose5126 mgdl or random glucose5200 mgdl) or(5) Outpatient utilization (5 2 visits for diabetes) or(6) Inpatient hospital utilization (primary hospital discharge ICD9250xx)
Exclusion criteria are
(1) Having no diabetes inclusion indicators during a sum total of two years of Kaiser membership after the date theywere identified in the Registry or
(2) Identified only by a ICD9 code of 6488 (gestational diabetes) or(3) Identified due to pharmacy utilization of metformin or thiazolidinedione only (no other indicators) and also
diagnosed with polycystic ovary syndrome HIV lipodystrophy metabolic syndrome pre-diabetes or reproductiveproblems
aThe Registry is updated annually
COHORT PROFILE DISTANCE 39
Table 2 Clinical behavioral psychosocial and quality and access indicators
Domain Survey scales or variables
Demographic Age
Sex
Race
Nativity
Language
Socioeconomic Education
years of school
highest degree attained
quality-weighted attainment
functional health literacy
Income
Occupation
Clinical profile Laboratory values
Blood pressure
Body mass index
Inpatient and outpatient utilization
Pharmacy utilization
Type of diabetes
Co-morbidity Index
Self-reported clinical characteristics
Health behaviors AdherenceKnowledge
Self-care behaviors
smoking
alcohol
exercise
medication refill adherence and discontinuation
self-monitoring of blood glucose
insulin injection frequency
appointment-keeping adherence
health screening adherence
sleep adequacy
Diabetes knowledge
Psychosocial Characteristics Big Five personality facets of conscientiousness and neuroticism
Perceived stress
Depressive and anxiety disorders
Social support and social networks
Discrimination
Socioeconomic Position (SEP) Individual-level socioeconomic indicators
Neighbourhood SEP contextual variables
Subjective measures of SEP
Financial barriers
Patient-provider relationship Patient-provider communication
Primary care provider characteristics
Quality and access to care Processes of care
Referrals for specialty care
40 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
external locus of control3334 social support2335ndash37
health status3839 stress40 depression2241ndash43 sleepquality44ndash46 chronic pain38 personality traits47ndash49
health literacy50ndash52 diabetes knowledge5354 medicaland dental coverage medical costs and ability topay55 medical visit travel time43 language barriersdelayed treatment56 provider recommendationsregarding aspirin43 and SMBG43 foot exam byphysician43 quality of care43 discrimination5758
provider communication and interactions5960 trustin provider6162 and self-reported height weight43
symptoms complications43 erectile dysfunction orfemale urinary incontinence
Survey modesThe survey included four modes of administration (i)a computer-assisted telephone interview (CATI)administered by a third party (ii) a password-enabled internet-accessible survey (lsquoweb surveyrsquo)maintained on a secure server at the Kaiser Divisionof Research (iii) a self-administered written surveyor (iv) a short version of the written survey Thecontent of each survey mode was identical except forslight adjustments in wording and logic patterns asneeded and was estimated to take 45ndash60 min tocomplete the short written version was abridged andcontained 40 questions The written and web surveyswere in English only but the CATI was available inEnglish Spanish Cantonese Mandarin and Tagalogusing certified translations of an English script andwas intended to maximize accessibility to those withlanguage barriers or limited English literacy orfluency
Other baseline dataWe geocoded the home addresses of all respondentsand non-respondents and linked their census blockgroup and tract to 2000 Census data These censuscharacteristics were then factor-analysed and wedeveloped an eight-item lsquodeprivation indexrsquo as a wayof summarizing the many indicators The deprivationindex has excellent internal reliability (Cronbachalphafrac14 093) with each of the eight items loadingalmost equally onto the underlying construct
Baseline clinical characterizations were obtainedfrom the extensive Kaiser administrative databasesincluding smoking status body mass index currentpharmacotherapy utilization and adherence63 labora-tory findings history of co-morbid events andprocedures use of emergency room outpatient andinpatient health services costs outpatient and inpa-tient risk scores based on health care utilization andseverity of disease64 and end-stage renal diseaseregistry (linked to the United States Renal DataSystem) These data are available for all subjectsboth respondents and non-respondents and allow fora substantive assessment of survey response bias
Who is in the sampleThis study took place within the Kaiser FoundationHealth Plan the largest not-for-profit managed healthcare company in the United States65 Study subjectswere members of Kaiser Permanente NorthernCalifornia (KPNC) which provides comprehensivemedical services to over 32 million members (as ofJanuary 2005) in the San Francisco Bay andSacramento metropolitan areas or 25ndash30 of theregionrsquos population KPNC members are predomi-nantly employed or retired individuals and theirfamilies and closely approximate the general popula-tion ethnically and socioeconomically except forthe extreme tails of income distribution66667 TheDiabetes Registry consisted of 199 123 members as ofJanuary 1 2005 and from this we selected anethnically stratified random sample of 40 735 healthplan members aged 30ndash75 at baseline to receive theDISTANCE Survey 6871 African American (17)11 197 Asian (27) 4233 Caucasian (10) 7018Latino (17) and 11 417 members of unknownethnicity (28) The DISTANCE Survey was in thefield from May 5 2005 until December 31 2006
The survey was completed by 20 188 persons(Table 3) The participation by ethnicity was 3420African Americans (169) 2312 Asian (114) 4602Caucasians (228) 2404 Filipinos (119) 3717Latinos (184) 2222 multi-racial (110) and 1511South Asian Pacific Islander Native America Eskimoor otherunknown (75) The distribution by modewas 10 429 CATI (517) 4288 written survey(212) 2393 short version (118) and 3078 web(152) Using an algorithm endorsed by the Councilof American Survey Research Organizations (CASRO)if persons unable to be contacted had the same rate ofeligibility as those contacted and were counted in thedenominator the survey response rate was 62 Ofthe 20 188 respondents to this current survey 4524subjects (22) had also responded to the priorDiabetes Registry Questionnaire (1994ndash97)
What has been measuredThe entire cohort has been characterized usingdemographic clinical and behavioural data fromadministrative records and census data
African American respondents were more likely tobe female to have had English as their preferredlanguage to have had elevated levels of low-densitylipoprotein (LDL) uncontrolled blood pressure werecurrent smokers with higher co-morbidity scores andliving in a deprived neighbourhood
Asian respondents were more likely to be male over60 years of age to not have had English as theirpreferred language to have had lower levels of LDLbetter blood pressure control lower missed appoint-ment rate better medication adherence were non-smokers not practicing SMBG with lower co-morbidity
COHORT PROFILE DISTANCE 41
Ta
ble
3C
hara
cter
isti
cso
fsu
rvey
resp
on
den
tsan
dn
on
-res
po
nd
ents
Re
spo
nd
en
tsn
()
or
Me
an
plusmn(S
D)
No
n-r
esp
on
de
nts
Afr
ica
nA
me
rica
n
Asi
an
(Ch
ine
se
Ko
rea
n
Vie
tna
me
se
Ja
pa
ne
se)
Ca
uca
sia
nF
ilip
ino
La
tin
oM
ult
ira
cia
lO
the
rau
nk
no
wn
All
resp
on
de
nts
No
nre
spo
nd
en
tsN
um
be
r(
)o
rM
ea
nplusmn
(SD
)
n(r
ow
)
34
20
(16
9)
23
12
(11
4)
46
02
(22
8)
24
04
(11
9)
37
17
(18
4)
22
22
(11
0)
15
11
(75
)2
01
88
(10
0)
20
54
7(1
00
)
Su
rve
yM
od
e
Tel
eph
on
eIn
terv
iew
18
06
(52
8)
97
1(4
20
)2
05
1(4
46
)8
46
(35
2)
22
29
(60
0)
15
66
(70
5)
96
0(6
35
)1
04
29
(51
7)
0
Wri
tten
-lo
ng
77
1(2
25
)5
75
(24
9)
97
3(2
11
)7
23
(30
1)
67
6(1
82
)3
46
(15
6)
22
4(1
48
)4
28
8(2
12
)0
Wri
tten
-sh
ort
43
9(1
28
)2
59
(11
2)
46
9(1
02
)5
16
(21
5)
44
8(1
20
)1
08
(49
)1
54
(10
2)
23
93
(11
8)
0
Web
40
4(1
18
)5
07
(21
9)
11
09
(24
1)
31
9(1
33
)3
64
(98
)2
02
(91
)1
73
(11
4)
30
78
(15
2)
0
Fem
ale
19
39
(56
7)
10
19
(44
1)
20
35
(44
2)
12
46
(51
8)
18
86
(50
7)
11
22
(50
5)
59
3(3
93
)9
84
0(4
87
)9
53
7(4
64
)
Ag
e(y
ea
rs)
30
ndash4
42
74
(80
)1
37
(59
)3
82
(83
)1
68
(70
)5
64
(15
2)
25
3(1
14
)1
89
(12
5)
19
67
(97
)2
58
0(1
26
)
45
ndash5
91
34
7(3
94
)9
34
(40
4)
17
94
(39
0)
11
06
(46
0)
16
39
(44
1)
94
3(4
24
)6
95
(46
0)
84
58
(41
9)
87
55
(42
6)
60thorn
17
99
(52
6)
12
41
(53
7)
24
26
(52
7)
11
30
(47
0)
15
14
(40
7)
10
26
(46
2)
62
7(4
15
)9
76
3(4
84
)9
21
2(4
48
)
Lan
gu
age
pre
fere
nce
En
gli
shsp
eak
ing
32
21
(94
2)
13
76
(59
5)
43
94
(95
5)
17
49
(72
8)
22
17
(59
6)
16
91
(76
1)
11
64
(77
0)
15
81
2(7
83
)1
72
12
(83
8)
Mea
nA
1C5
7
16
55
(54
0)
10
87
(50
1)
19
15
(46
8)
13
52
(60
4)
19
57
(58
6)
10
98
(56
2)
77
9(5
79
)9
84
3(5
41
)8
34
3(5
13
)
Mea
nL
DL5
10
01
25
0(4
17
)6
26
(28
9)
13
91
(34
6)
63
4(2
86
)1
20
2(3
71
)7
36
(38
1)
48
2(3
63
)6
32
1(3
53
)6
90
9(4
24
)
Un
con
tro
lled
blo
od
pre
ssu
re(S4
13
0o
rD4
80
)1
96
6(6
14
)9
13
(42
8)
22
64
(54
7)
10
41
(46
5)
17
46
(51
2)
10
89
(54
2)
68
1(4
95
)9
70
0(5
24
)9
58
1(5
43
)
Mis
sed
ap
po
intm
ent
rate
(per
year)
01
7plusmn
(01
8)
01
0plusmn
(01
4)
01
3plusmn
(01
6)
01
3plusmn
(01
7)
01
8plusmn
(01
9)
01
7plusmn
(01
9)
01
8plusmn
(01
9)
01
5plusmn
(01
7)
01
8plusmn
(02
2)
Po
or
med
icati
on
ad
her
ence
b9
55
(37
1)
46
1(2
51
)9
61
(28
2)
59
2(2
93
)1
24
7(4
28
)6
36
(36
4)
42
4(3
51
)5
27
6(3
36
)5
10
8(3
90
)
Sm
ok
ing
(cu
rren
t)3
48
(10
2)
93
(40
)4
10
(90
)1
47
(61
)2
45
(66
)2
00
(91
)1
05
(70
)1
54
8(7
7)
18
29
(90
)
Pra
ctic
esS
MB
G1
57
3(4
60
)1
00
4(4
34
)2
24
6(4
88
)1
11
5(4
64
)1
62
2(4
36
)1
02
5(4
61
)6
23
(41
2)
92
08
(45
6)
65
32
(31
8)
Ou
tpati
ent
risk
sco
rec
24
(21
)1
8(1
1)
21
(17
)1
8(1
0)
19
(14
)2
2(1
6)
19
(14
)2
0(1
6)
18
(15
)
Inp
ati
ent
risk
sco
rec
26
(42
)1
5(2
0)
22
(33
)1
4(2
1)
18
(30
)2
2(3
9)
17
(29
)2
0(3
2)
18
(34
)
Lin
gu
isti
call
yis
ola
ted
nei
gh
bo
urh
oo
dd
10
5(3
1)
11
5(5
0)
55
(12
)1
14
(48
)2
81
(76
)1
13
(51
)6
2(4
1)
84
5(4
2)
10
35
(51
)
Eco
no
mic
all
yd
epri
ved
nei
gh
bo
urh
oo
de
82
7(2
43
)1
24
(54
)3
46
(76
)1
48
(62
)6
68
(18
2)
32
3(1
47
)1
67
(11
2)
26
03
(13
0)
27
24
(13
4)
Wo
rkin
gcl
ass
nei
gh
bo
urh
oo
df
14
40
(42
2)
40
1(1
74
)1
16
8(2
58
)8
84
(37
0)
17
49
(47
6)
86
8(3
94
)4
64
(31
0)
69
74
(34
9)
75
66
(37
2)
aIn
clu
des
Paci
fic
Isla
nd
ers
Nati
veA
mer
ican
sS
ou
thA
sian
o
ther
un
spec
ifie
dan
du
nk
no
wn
(did
no
tan
swer
)b5
20
co
nti
nu
ou
sm
edic
ati
on
gap
s6
3
c Th
eyare
wei
gh
ted
by
pati
entrsquo
sh
ealt
hca
reu
tili
zati
on
an
dse
veri
tyo
fd
isea
se6
4
d5
25
o
fh
ou
seh
old
sin
cen
sus
blo
ckgro
up
are
lin
gu
isti
call
yis
ola
ted
e5
20
o
fh
ou
seh
old
sin
cen
sus
blo
ckgro
up
are
livi
ng
bel
ow
po
vert
yli
ne
f 56
6
of
per
son
sin
cen
sus
blo
ckgro
up
are
emp
loye
din
wo
rkin
gcl
ass
occ
up
ati
on
42 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
scores and not living in an economically deprived orworking class neighbourhood
Caucasian respondents were more likely to completethe survey online to have had lower mean A1Cpracticing SMBG and not living in linguisticallyisolated neighbourhood
Filipino respondents were more likely to participatevia written survey to have been older than 45 yearsof age to have had lower levels of LDL and lowerco-morbidity scores
Latino respondents were more likely to participatevia CATI to have been younger than 45 years of ageto not have English as their preferred language tohave had higher mean A1C higher rates of missedappointments poorer medication adherence andliving in a linguistically isolated or working classneighbourhood
What outcomes will be measuredduring follow-upThis cohort will be the basis for longitudinal evalua-tions of a wide range of clinical outcomes associatedwith diabetes and are powered to evaluate the ethnicand educational disparities in diabetes-related com-plication rates (eg myocardial infarction strokeheart failure kidney failure and amputation) andmortality after a 25-year follow-up Clinical andadministrative follow-up data will be captured fromthese same sources as baseline plus deaths from thestate mortality files Data from all sources combinedmay offer insights into how differences at the patient-level (eg differences in behaviours type of therapyintermediate health status and psychosocial factorsand competing demands) provider-level (eg trust inprovider racial concordance and patient-providercommunication) and system-level (eg differentialimpact of cost-sharing) may lead to social disparitiesin diabetes health end-stage diabetes complicationsand mortality
What is the anticipated attritionOne of the strengths of this diabetes registry is thevery low turnover rate (5 discontinue membershipeach year) affording minimal loss to follow-up Onaverage the duration of Kaiser membership amongour diabetes cohort members is 89 years (SDfrac14 66)
What are the strengths andweaknessesA significant weakness is that written and websurveys were only offered in English Howevertelephone interviewers made the initial efforts tocontact subjects who had the opportunity to complete
interviews in Spanish Mandarin Cantonese orTagalog Offering the survey by oral interview inmultiple languages was intended to mitigate thelanguage andor literacy barriers but we still observedlower participation rates by those with less educationand modest response differences by race On a morefavourable note 84 of non-respondents were identi-fied by the health plan as having English as theirpreferred language suggesting that language alonewas not a major barrier to participation
The response rate was as expected lower than ourprevious Diabetes Research Questionnaire (1994ndash97)A recent survey conducted in the Translating ResearchInto Action for Diabetes (TRIAD) study amongmanaged care populations across the United Statesincluding Kaiser43 had a response rate of 50 TheBehavioural Risk Factor Surveillance System alsoexperienced a decline in median response ratesfrom 632 in 1996 to 535 in 200168 Privacyconcerns and competition from telemarketing haveprobably played a role in declining response ratesHowever if generally lower response rates have beenobserved in minority populations the response ratefor this survey may be considered favourable giventhat 77 of our sample were racial or ethnic minorityhealth plan members
One strength of this study is the relatively uniformaccess to care provided by membership in thisintegrated health plan Thus this study largely avoidsdisparities in access and quality which are potentsources of confounding bias in population-basedstudies of social health disparities Another strength isthe quality and richness of the clinical follow-up dataAdministrative and clinical data are available directlyfrom the health plan and are considered to be reliableand complete Each telephone interview was conductedby professional interviewers employed trained andsupervised by the Public Health Institute SurveyResearch Group in Sacramento CA which has itsown internal quality controls Written surveys (long-and short-forms) were coded and edited using anextensive data dictionary and detailed coding rules tostandardize and clarify responses before being sentto the Data Entry department at the KaiserDivision of Research which conducted double dataentry The survey data quality is limited by beingself-reported
The diversity size and wealth of data included inthe DISTANCE cohort make it suitable for theprospective study of a wide range of social disparitiesin the processes and outcomes of diabetes health careInformation about childhood socioeconomic positionand educational attainment will facilitate a life courseapproach6970
Analysis of response biasResponse bias is a concern in any survey Surveys aresubject to response bias if participation is associatedwith the risk factors (exposures) andor diseases
COHORT PROFILE DISTANCE 43
(outcomes) under investigation In this study popula-tion we have substantial data on non-respondentsincluding Kaiser administrative data co-morbidityand census data on neighbourhood characteristicsallowing us to compare hypothesized relationships inrespondents vs non-respondents and adjust for thispotential response bias71 While participation wassomewhat lower among minorities and those withless education overall respondents and non-respon-dents were quite similar
We had pre-baseline reported race data for 29 319(72) members of the DISTANCE cohort andobserved differences in response rates by raceAfrican Americans 52 Asians 46 Caucasians63 Latinos 53 unknown 45 (P lt 00001) Wealso had educational attainment data for 4524subjects (22) who had completed the previoussurvey in 1994ndash97 and observed that subjects with ahigh school education or less participated at a lowerrate than those with more than a high schooleducation (54 vs 60) (P lt 00001)
A more complete view of response bias was obtainedfrom health plan administrative data and census datawhich was available for all members of the cohortWe found few notable differences between respon-dents and non-respondents in demographic clinicalbehavioural or neighbourhood characteristics Com-pared with respondents non-respondents had highermean LDL levels (993 vs 943 mgdl) and were lesslikely to practice SMBG (32 vs 46) However aportion of these unadjusted differences are likely dueto age differences in respondents vs non-respondents
We conducted an assessment of response bias bycomparing differences in A1C across ethnic groupsseparately among respondents and non-respondentsWe specified a logistic regression model of poorglycemic control which regressed the outcome A1C47 on ethnicity age and sex and an interactionterm (ethnicity survey response) Although respon-dents consistently had higher A1C than non-respon-dents ethnic differences in the relationship betweenethnicity and A1C did not differ between respondentsand non-respondents (Pfrac14 055) We conducted asimilar analysis with the subjects for whom we hadeducational attainment data from a previous surveyand again found no substantive response bias(Pfrac14 028) While characteristics differ betweenrespondents and non-respondents (response bias)the associations with outcomes of interest (slopes)are much less vulnerable to such bias7273
Statistical methodsAnalyses of DISTANCE data will focus on associationsrather than descriptive characterization We willemploy a modelling approach using inverse weightingestimation of marginal structural models72 to inves-tigate causal relationships in this study Thesemodels will include weights to adjust for the non-proportionate sampling fractions (eg over-sampling
of the minority ethnic groups) and response bias(eg giving greater weight to respondents whohave characteristics more similar to the non-respondents) Thus the proposed estimating approachwill adjust for confounding and selection biassimultaneously73ndash77
How can I collaborate Where canI find out moreWe are not free to release participantsrsquo personal dataunder our promise to them regarding confidentialityHowever the DISTANCE steering committee is inter-ested in collaborations with external researchersDISTANCE investigators are particularly interested incomparing social disparities observed in this insuredpopulation with uniform access to care to disparitiesobserved in population-based samples where qualityand access to care vary widely by social strataRequests for collaboration must be submitted to thedirector of the central coordinating centre (corre-sponding author) Before collaborations can beinitiated proposals require review and approval bythe DISTANCE Publications and Presentations (PampP)Committee This committee was formed to (i) Ensureaccurate uniform timely and high quality reportingof DISTANCE findings (ii) Preserve the scientificintegrity of the study and (iii) Safeguard the rightsand confidentiality of participants
Supplementary DataSupplementary data are available at IJE online
AcknowledgementsFunds were provided by National Institute ofDiabetes Digestive and Kidney Diseases [R01DK65664] and National Institute of Child Healthand Human Development [R01 HD046113] Dr DS isalso supported by a National Institutes of HealthMentored Clinical Scientist Award [K-23 RR16539]The Public Health InstituteSurvey Research Groupconducted our telephone interviews Gerard andAssociates provided certified translations of all patientmaterials and interview scripts KP Corporation (notaffiliated with Kaiser Permanente) provided printingand mailing services for the written surveys andother communications We thank Marcia Ewing andCarol Rabello for their diligent work on surveyoperations Most of all we thank all the studyparticipants
Conflict of interest None declared
References1 Healthy People 2010 U S Department of Health and
Human Services 2005 Available at httpwwwhealthypeoplegov(Accessed August 2 2007)
44 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
2 Smedley BD Stith AY Nelson AR Unequal TreatmentConfronting Racial and Ethnic Disparities in Health CareWashington DC Institute of Medicine NationalAcademy Press 2002
3 Karter AJ Newman B Rowell S et al Large-scalecollection of family history data and recruitment ofinformative families for genetic analysis J Reg Manag1998257ndash12
4 Karter AJ Ferrara A Darbinian J Ackerson LM Selby JVSelf-monitoring of blood glucose language and financialbarriers in a managed care population with diabetesDiabetes Care 200023477ndash83
5 Karter AJ Moffet HH Liu J et al Achieving goodglycemic control initiation of new antihyperglycemictherapies in patients with type 2 diabetes from theKaiser Permanente Northern California Diabetes RegistryAm J Manag Care 200511262ndash70
6 Karter AJ Ackerson LM Darbinian JA et al Self-monitoring of blood glucose levels and glycemic controlthe Northern California Kaiser Permanente Diabetesregistry Am J Med 20011111ndash9
7 Karter AJ Ferrara A Liu JY Moffet HH Ackerson LMSelby JV Ethnic disparities in diabetic complications inan insured population JAMA 20022872519ndash27
8 Karter AJ Parker MM Moffet HH et al Missedappointments and poor glycemic control an opportunityto identify high-risk diabetic patients Med Care200442110ndash15
9 Selby JV Karter AJ Ackerson LM Ferrara A Liu JDeveloping a prediction rule from automated clinicaldatabases to identify high-risk patients in a largepopulation with diabetes Diabetes Care 2001241547ndash55
10 Martin TL Selby JV Zhang D Physician and patientprevention practices in NIDDM in a large urbanmanaged-care organization Diabetes Care 1995181124ndash32
11 Chaturvedi N Jarrett J Shipley MJ Fuller JHSocioeconomic gradient in morbidity and mortality inpeople with diabetes cohort study findings from theWhitehall Study and the WHO Multinational Study ofVascular Disease in Diabetes BMJ 1998316100ndash5
12 Booth GL Hux JE Relationship between avoidablehospitalizations for diabetes mellitus and income levelArch Intern Med 2003163101ndash6
13 Kleinbaum DG Kupper LL Morgenstern H EpidemiologicResearch Principles and Quantitative Methods New York VanNorstrand Reinhold 1982
14 Dillman DA Mail and Internet Surveys The Tailored DesignMethod 2nd edn New York John Wiley 2000
15 Rudd RE Comings JP Hyde JN Leave no one behindimproving health and risk communication through att-ention to literacy J Health Commun 20038 (Suppl 1)104ndash15
16 Singh-Manoux A Adler NE Marmot MG Subjectivesocial status its determinants and its association withmeasures of ill-health in the Whitehall II study Soc SciMed 2003561321ndash33
17 John D Catherine T MacArthur Research Network onSocioeconomic Status and Health The MacArthur Scale ofSubjective Social Status San Francisco University ofCalifornia 1999 Available at httpwwwmacsesucsfeduResearchPsychosocialnotebooksubjectivehtml(Accessed July 12 2007)
18 Ross CE Wu CL Education age and the cumulativeadvantage in health J Health Soc Behav 199637104ndash20
19 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
20 White S Dillow S Key Concepts and Features of the 2003National Assessment of Adult Literacy (NAALS) (NCES2006ndash471) Washington DC Department of EducationNational Center for Education Statistics 2005
21 Botman SL Moore TF Moriarity CL Parsons VL Designand Estimation for the National Health Interview Survey(NHIS) 1995ndash2004 National Center for Health StatisticsVital Health Stat 2 20001301ndash31
22 Friedman GD Cutter GR Donahue RP et al CARDIAstudy design recruitment and some characteristics of theexamined subjects J Clin Epidemiol 1988411105ndash16
23 Ross CE Wu CL The links between education and healthAm Sociol Rev 199560719ndash45
24 Robert S House JS SES differentials in health by ageand alternative indicators of SES J Aging Health19968359ndash88
25 Toobert DJ Hampson SE Glasgow RE The summary ofdiabetes self-care activities measure results from 7studies and a revised scale Diabetes Care 200023943ndash50
26 Craig CL Marshall AL Sjostrom M et al Internationalphysical activity questionnaire 12-country reliability andvalidity Med Sci Sports Exerc 2003351381ndash95
27 Gordon AJ Maisto SA McNeil M et al Three questionscan detect hazardous drinkers J Fam Pract 200150313ndash20
28 National Health and Nutrition Examination Survey(NHANES) Questionnaire (2001-2002) Centers for DiseaseControl and Prevention (CDC) National Center for HealthStatistics (NCHS) Hyattsville MD US Department ofHealth and Human Services Centers for Disease Controland Prevention 2002
29 Behavioral Risk Factor Surveillance System (BRFSS)Centers for Disease Control and Prevention (CDC) NationalCenter for Health Statistics (NCHS) Hyattsville MD USDepartment of Health and Human Services Centers forDisease Control and Prevention 2003
30 Chien-Wen T Phillips RL Green LA Fryer GE Dovey SMWhat physicians need to know about seniors and limitedprescription benefits and why Am Fam Physician200266212
31 Morisky DE Green LW Levine DM Concurrent andpredictive validity of a self-reported measure of medica-tion adherence Med Care 19862467ndash74
32 Barringer TA Kirk JK Santaniello AC Foley KLMichielutte R Effect of a multivitamin and mineralsupplement on infection and quality of life A random-ized double-blind placebo-controlled trial Ann Intern Med2003138365ndash71
33 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
34 Peyrot M Rubin RR Structure and correlates of diabetes-specific locus of control Diabetes Care 199417994ndash1001
35 Wagenknecht LE Mayer EJ Rewers M et al The insulinresistance atherosclerosis study (IRAS) objectivesdesign and recruitment results Ann Epidemiol19955464ndash72
COHORT PROFILE DISTANCE 45
36 King DE Mainous AG III Pearson WS C-reactiveprotein diabetes and attendance at religious servicesDiabetes Care 2002251172ndash76
37 Berkman LF Syme SL Social networks host resistanceand mortality a nine-year follow-up study of AlamedaCounty residents Am J Epidemiol 1979109186ndash204
38 Ware JE Kosinski M Keller SD SF-12 How to Score theSF-12 Physical and Mental Health Summary Scales 2nd ednBoston The Health Institute New England MedicalCenter 1995
39 Ware JE Kosinksi M Dewey JE Gandek B How to Scoreand Interpret Single-Item Health Status Measures A Manual forUsers of the SF-8 Health Survey Lincoln RI QualityMetricIncorporated 2001
40 Cohen S Kamarck T Mermelstein R A global measure ofperceived stress J Health Soc Behav 198324385ndash96
41 Kroenke K Spitzer RL Williams JB The PHQ-9 validityof a brief depression severity measure J Gen Intern Med200116606ndash13
42 Spitzer RL Kroenke K Williams JB Validation and utilityof a self-report version of PRIME-MD the PHQ primarycare study Primary care evaluation of mental disordersPatient Health Questionnaire JAMA 19992821737ndash44
43 The TRIAD Study Group The translating research intoaction for diabetes (TRIAD) study a multicenter study ofdiabetes in managed care Diabetes Care 200225386ndash89
44 Ayas NT White DP Al Delaimy WK et al A prospectivestudy of self-reported sleep duration and incidentdiabetes in women Diabetes Care 200326380ndash84
45 Sickel AE Moore PJ Adler NE Williams DR Jackson JSThe differential effects of sleep quality and quantity onthe relationship between SES and health Ann NY Acad Sci1999896431ndash34
46 Moore PJ Adler NE Williams DR Jackson JSSocioeconomic status and health the role of sleepPsychosom Med 200264337ndash44
47 Benet-Martinez V John OP Los Cinco Grandes acrosscultures and ethnic groups multitrait multimethodanalyses of the Big Five in Spanish and English J PersSoc Psychol 199875729ndash50
48 Srivastava S John OP Gosling SD Potter J Developmentof personality in early and middle adulthood set likeplaster or persistent change J Pers Soc Psychol2003841041ndash53
49 Gosling SD Rentfrow PJ Swann WB A very briefmeasure of the Big-Five personality domains J Res Pers200337504ndash28
50 Schillinger D Grumbach K Piette J et al Association ofhealth literacy with diabetes outcomes JAMA2002288475ndash82
51 Schillinger D Barton LR Karter AJ Wang F Adler NDoes literacy mediate the relationship between educationand health outcomes A study of a low-income popula-tion with diabetes Public Health Rep 2006121245ndash54
52 Chew LD Bradley KA Boyko EJ Brief questions toidentify patients with inadequate health literacy Fam Med200436588ndash94
53 The American Association of Clinical EndocrinologistsMedical Guidelines for the Management of DiabetesMellitus The AACE System of Intensive DiabetesSelf-Management ndash 2002 Update Endocr Pract 20028(Suppl 1)40ndash82
54 Fitzgerald JT Funnell MM Hess GE et al The reliabilityand validity of a brief diabetes knowledge test DiabetesCare 199821706ndash10
55 Piette JD Heisler M Wagner TH Problems paying out-of-pocket medication costs among older adults withdiabetes Diabetes Care 200427384ndash91
56 Epidemiology of Diabetes Interventions andComplications (EDIC) Design implementation andpreliminary results of a long-term follow-up of thediabetes control and complications trial cohort DiabetesCare 19992299ndash111
57 Krieger N Sidney S Coakley E Racial discrimination andskin color in the CARDIA study implications for publichealth research Coronary artery risk development inyoung adults Am J Public Health 1998881308ndash13
58 Krieger N Smith K Naishadham D Hartman CBarbeau EM Experiences of discrimination validity andreliability of a self-report measure for population healthresearch on racism and health Soc Sci Med 2005611576ndash96
59 CAHPS 20 Questionnaires Crofton C AHCPR Pub No97-R079 Washington DC Agency for Health Care Policyand Research 1998
60 Hays RD Shaul JA Williams VS et al Psychometricproperties of the CAHPS 10 survey measures Consumerassessment of health plans study Med Care 199937MS22ndash31
61 Anderson LA Dedrick RF Development of the trust inphysician scale a measure to assess interpersonal trust inpatient-physician relationships Psychol Rep 1990671091ndash100
62 Thom DH Ribisl KM Stewart AL Luke DA Furthervalidation and reliability testing of the Trust in PhysicianScale The Stanford trust study physicians Med Care199937510ndash17
63 Steiner JF Koepsell TD Fihn SD Inui TS A generalmethod of compliance assessment using centralizedpharmacy records Description and validation Med Care198826814ndash23
64 Clark DO Von Korff M Saunders K Baluch WMSimon GE A chronic disease score with empiricallyderived weights Med Care 199533783ndash95
65 Kaiser Permanente Business Intelligence Report 2006Research and Markets 2006 Available at httpwwwresearchandmarketscomreportinfoaspreport_idfrac14451423(Accessed July 9 2007)
66 Krieger N Overcoming the absence of socioeconomic datain medical records validation and application of acensus-based methodology Am J Public Health 199282703ndash10
67 Hiatt RA Friedman GD Characteristics of patientsreferred for treatment of end-stage renal diseasein a defined population Am J Public Health 198272829ndash33
68 Porter S Jackson K Trosclaire A Pederson LL Office onsmoking and Health NCfCDPaHPC Prevalence of currentcigarette smoking among adults and changes in pre-valence of current and some day smoking ndash UnitedStates 1996ndash2001 MMWR 200352303ndash7
69 Smith GD Hart C Blane D Gillis C Hawthorne VLifetime socioeconomic position and mortality prospec-tive observational study BMJ 1997314547ndash52
46 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
70 Lynch JW Kaplan GA Shema SJ Cumulative impact ofsustained economic hardship on physical cognitivepsychological and social functioning N Engl J Med19973371889ndash95
71 Melton LJ3 Dyck PJ Karnes JL OrsquoBrien PC Service FJNon-response bias in studies of diabetic complicationsthe Rochester Diabetic Neuropathy Study J Clin Epidemiol199346341ndash48
72 Robins JM Association causation and marginal struc-tural models Synthese 1999121151ndash79
73 Hernan MA Hernandez-Diaz S Robins JM A structuralapproach to selection bias Epidemiology 200415615ndash25
74 Bodnar LM Davidian M Siega-Riz AM Tsiatis AAMarginal structural models for analyzing causal effects
of time-dependent treatments an application in perinatalepidemiology Am J Epidemiol 2004159926ndash34
75 Bryan JF Yu Z van der Laan M Analysis of longitudinalmarginal structural models U C Berkeley Division ofBiolstatistics Working Paper Series Paper 1201ndash29 2002
76 Neugebauer R van der Laan M Why prefer double robustestimators in causal inference J Stat Plan Inference2005129405ndash26
77 Wang Y Petersen ML Bangsberg D van der Laan MDiagnosing bias in the inverse probability of treatmentweighted estimator resulting from violation ofexperimental treatment assignment U C BerkeleyDivision of Biolstatistics Working Paper Series Paper 2111ndash25 2006
COHORT PROFILE DISTANCE 47
Table 2 Clinical behavioral psychosocial and quality and access indicators
Domain Survey scales or variables
Demographic Age
Sex
Race
Nativity
Language
Socioeconomic Education
years of school
highest degree attained
quality-weighted attainment
functional health literacy
Income
Occupation
Clinical profile Laboratory values
Blood pressure
Body mass index
Inpatient and outpatient utilization
Pharmacy utilization
Type of diabetes
Co-morbidity Index
Self-reported clinical characteristics
Health behaviors AdherenceKnowledge
Self-care behaviors
smoking
alcohol
exercise
medication refill adherence and discontinuation
self-monitoring of blood glucose
insulin injection frequency
appointment-keeping adherence
health screening adherence
sleep adequacy
Diabetes knowledge
Psychosocial Characteristics Big Five personality facets of conscientiousness and neuroticism
Perceived stress
Depressive and anxiety disorders
Social support and social networks
Discrimination
Socioeconomic Position (SEP) Individual-level socioeconomic indicators
Neighbourhood SEP contextual variables
Subjective measures of SEP
Financial barriers
Patient-provider relationship Patient-provider communication
Primary care provider characteristics
Quality and access to care Processes of care
Referrals for specialty care
40 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
external locus of control3334 social support2335ndash37
health status3839 stress40 depression2241ndash43 sleepquality44ndash46 chronic pain38 personality traits47ndash49
health literacy50ndash52 diabetes knowledge5354 medicaland dental coverage medical costs and ability topay55 medical visit travel time43 language barriersdelayed treatment56 provider recommendationsregarding aspirin43 and SMBG43 foot exam byphysician43 quality of care43 discrimination5758
provider communication and interactions5960 trustin provider6162 and self-reported height weight43
symptoms complications43 erectile dysfunction orfemale urinary incontinence
Survey modesThe survey included four modes of administration (i)a computer-assisted telephone interview (CATI)administered by a third party (ii) a password-enabled internet-accessible survey (lsquoweb surveyrsquo)maintained on a secure server at the Kaiser Divisionof Research (iii) a self-administered written surveyor (iv) a short version of the written survey Thecontent of each survey mode was identical except forslight adjustments in wording and logic patterns asneeded and was estimated to take 45ndash60 min tocomplete the short written version was abridged andcontained 40 questions The written and web surveyswere in English only but the CATI was available inEnglish Spanish Cantonese Mandarin and Tagalogusing certified translations of an English script andwas intended to maximize accessibility to those withlanguage barriers or limited English literacy orfluency
Other baseline dataWe geocoded the home addresses of all respondentsand non-respondents and linked their census blockgroup and tract to 2000 Census data These censuscharacteristics were then factor-analysed and wedeveloped an eight-item lsquodeprivation indexrsquo as a wayof summarizing the many indicators The deprivationindex has excellent internal reliability (Cronbachalphafrac14 093) with each of the eight items loadingalmost equally onto the underlying construct
Baseline clinical characterizations were obtainedfrom the extensive Kaiser administrative databasesincluding smoking status body mass index currentpharmacotherapy utilization and adherence63 labora-tory findings history of co-morbid events andprocedures use of emergency room outpatient andinpatient health services costs outpatient and inpa-tient risk scores based on health care utilization andseverity of disease64 and end-stage renal diseaseregistry (linked to the United States Renal DataSystem) These data are available for all subjectsboth respondents and non-respondents and allow fora substantive assessment of survey response bias
Who is in the sampleThis study took place within the Kaiser FoundationHealth Plan the largest not-for-profit managed healthcare company in the United States65 Study subjectswere members of Kaiser Permanente NorthernCalifornia (KPNC) which provides comprehensivemedical services to over 32 million members (as ofJanuary 2005) in the San Francisco Bay andSacramento metropolitan areas or 25ndash30 of theregionrsquos population KPNC members are predomi-nantly employed or retired individuals and theirfamilies and closely approximate the general popula-tion ethnically and socioeconomically except forthe extreme tails of income distribution66667 TheDiabetes Registry consisted of 199 123 members as ofJanuary 1 2005 and from this we selected anethnically stratified random sample of 40 735 healthplan members aged 30ndash75 at baseline to receive theDISTANCE Survey 6871 African American (17)11 197 Asian (27) 4233 Caucasian (10) 7018Latino (17) and 11 417 members of unknownethnicity (28) The DISTANCE Survey was in thefield from May 5 2005 until December 31 2006
The survey was completed by 20 188 persons(Table 3) The participation by ethnicity was 3420African Americans (169) 2312 Asian (114) 4602Caucasians (228) 2404 Filipinos (119) 3717Latinos (184) 2222 multi-racial (110) and 1511South Asian Pacific Islander Native America Eskimoor otherunknown (75) The distribution by modewas 10 429 CATI (517) 4288 written survey(212) 2393 short version (118) and 3078 web(152) Using an algorithm endorsed by the Councilof American Survey Research Organizations (CASRO)if persons unable to be contacted had the same rate ofeligibility as those contacted and were counted in thedenominator the survey response rate was 62 Ofthe 20 188 respondents to this current survey 4524subjects (22) had also responded to the priorDiabetes Registry Questionnaire (1994ndash97)
What has been measuredThe entire cohort has been characterized usingdemographic clinical and behavioural data fromadministrative records and census data
African American respondents were more likely tobe female to have had English as their preferredlanguage to have had elevated levels of low-densitylipoprotein (LDL) uncontrolled blood pressure werecurrent smokers with higher co-morbidity scores andliving in a deprived neighbourhood
Asian respondents were more likely to be male over60 years of age to not have had English as theirpreferred language to have had lower levels of LDLbetter blood pressure control lower missed appoint-ment rate better medication adherence were non-smokers not practicing SMBG with lower co-morbidity
COHORT PROFILE DISTANCE 41
Ta
ble
3C
hara
cter
isti
cso
fsu
rvey
resp
on
den
tsan
dn
on
-res
po
nd
ents
Re
spo
nd
en
tsn
()
or
Me
an
plusmn(S
D)
No
n-r
esp
on
de
nts
Afr
ica
nA
me
rica
n
Asi
an
(Ch
ine
se
Ko
rea
n
Vie
tna
me
se
Ja
pa
ne
se)
Ca
uca
sia
nF
ilip
ino
La
tin
oM
ult
ira
cia
lO
the
rau
nk
no
wn
All
resp
on
de
nts
No
nre
spo
nd
en
tsN
um
be
r(
)o
rM
ea
nplusmn
(SD
)
n(r
ow
)
34
20
(16
9)
23
12
(11
4)
46
02
(22
8)
24
04
(11
9)
37
17
(18
4)
22
22
(11
0)
15
11
(75
)2
01
88
(10
0)
20
54
7(1
00
)
Su
rve
yM
od
e
Tel
eph
on
eIn
terv
iew
18
06
(52
8)
97
1(4
20
)2
05
1(4
46
)8
46
(35
2)
22
29
(60
0)
15
66
(70
5)
96
0(6
35
)1
04
29
(51
7)
0
Wri
tten
-lo
ng
77
1(2
25
)5
75
(24
9)
97
3(2
11
)7
23
(30
1)
67
6(1
82
)3
46
(15
6)
22
4(1
48
)4
28
8(2
12
)0
Wri
tten
-sh
ort
43
9(1
28
)2
59
(11
2)
46
9(1
02
)5
16
(21
5)
44
8(1
20
)1
08
(49
)1
54
(10
2)
23
93
(11
8)
0
Web
40
4(1
18
)5
07
(21
9)
11
09
(24
1)
31
9(1
33
)3
64
(98
)2
02
(91
)1
73
(11
4)
30
78
(15
2)
0
Fem
ale
19
39
(56
7)
10
19
(44
1)
20
35
(44
2)
12
46
(51
8)
18
86
(50
7)
11
22
(50
5)
59
3(3
93
)9
84
0(4
87
)9
53
7(4
64
)
Ag
e(y
ea
rs)
30
ndash4
42
74
(80
)1
37
(59
)3
82
(83
)1
68
(70
)5
64
(15
2)
25
3(1
14
)1
89
(12
5)
19
67
(97
)2
58
0(1
26
)
45
ndash5
91
34
7(3
94
)9
34
(40
4)
17
94
(39
0)
11
06
(46
0)
16
39
(44
1)
94
3(4
24
)6
95
(46
0)
84
58
(41
9)
87
55
(42
6)
60thorn
17
99
(52
6)
12
41
(53
7)
24
26
(52
7)
11
30
(47
0)
15
14
(40
7)
10
26
(46
2)
62
7(4
15
)9
76
3(4
84
)9
21
2(4
48
)
Lan
gu
age
pre
fere
nce
En
gli
shsp
eak
ing
32
21
(94
2)
13
76
(59
5)
43
94
(95
5)
17
49
(72
8)
22
17
(59
6)
16
91
(76
1)
11
64
(77
0)
15
81
2(7
83
)1
72
12
(83
8)
Mea
nA
1C5
7
16
55
(54
0)
10
87
(50
1)
19
15
(46
8)
13
52
(60
4)
19
57
(58
6)
10
98
(56
2)
77
9(5
79
)9
84
3(5
41
)8
34
3(5
13
)
Mea
nL
DL5
10
01
25
0(4
17
)6
26
(28
9)
13
91
(34
6)
63
4(2
86
)1
20
2(3
71
)7
36
(38
1)
48
2(3
63
)6
32
1(3
53
)6
90
9(4
24
)
Un
con
tro
lled
blo
od
pre
ssu
re(S4
13
0o
rD4
80
)1
96
6(6
14
)9
13
(42
8)
22
64
(54
7)
10
41
(46
5)
17
46
(51
2)
10
89
(54
2)
68
1(4
95
)9
70
0(5
24
)9
58
1(5
43
)
Mis
sed
ap
po
intm
ent
rate
(per
year)
01
7plusmn
(01
8)
01
0plusmn
(01
4)
01
3plusmn
(01
6)
01
3plusmn
(01
7)
01
8plusmn
(01
9)
01
7plusmn
(01
9)
01
8plusmn
(01
9)
01
5plusmn
(01
7)
01
8plusmn
(02
2)
Po
or
med
icati
on
ad
her
ence
b9
55
(37
1)
46
1(2
51
)9
61
(28
2)
59
2(2
93
)1
24
7(4
28
)6
36
(36
4)
42
4(3
51
)5
27
6(3
36
)5
10
8(3
90
)
Sm
ok
ing
(cu
rren
t)3
48
(10
2)
93
(40
)4
10
(90
)1
47
(61
)2
45
(66
)2
00
(91
)1
05
(70
)1
54
8(7
7)
18
29
(90
)
Pra
ctic
esS
MB
G1
57
3(4
60
)1
00
4(4
34
)2
24
6(4
88
)1
11
5(4
64
)1
62
2(4
36
)1
02
5(4
61
)6
23
(41
2)
92
08
(45
6)
65
32
(31
8)
Ou
tpati
ent
risk
sco
rec
24
(21
)1
8(1
1)
21
(17
)1
8(1
0)
19
(14
)2
2(1
6)
19
(14
)2
0(1
6)
18
(15
)
Inp
ati
ent
risk
sco
rec
26
(42
)1
5(2
0)
22
(33
)1
4(2
1)
18
(30
)2
2(3
9)
17
(29
)2
0(3
2)
18
(34
)
Lin
gu
isti
call
yis
ola
ted
nei
gh
bo
urh
oo
dd
10
5(3
1)
11
5(5
0)
55
(12
)1
14
(48
)2
81
(76
)1
13
(51
)6
2(4
1)
84
5(4
2)
10
35
(51
)
Eco
no
mic
all
yd
epri
ved
nei
gh
bo
urh
oo
de
82
7(2
43
)1
24
(54
)3
46
(76
)1
48
(62
)6
68
(18
2)
32
3(1
47
)1
67
(11
2)
26
03
(13
0)
27
24
(13
4)
Wo
rkin
gcl
ass
nei
gh
bo
urh
oo
df
14
40
(42
2)
40
1(1
74
)1
16
8(2
58
)8
84
(37
0)
17
49
(47
6)
86
8(3
94
)4
64
(31
0)
69
74
(34
9)
75
66
(37
2)
aIn
clu
des
Paci
fic
Isla
nd
ers
Nati
veA
mer
ican
sS
ou
thA
sian
o
ther
un
spec
ifie
dan
du
nk
no
wn
(did
no
tan
swer
)b5
20
co
nti
nu
ou
sm
edic
ati
on
gap
s6
3
c Th
eyare
wei
gh
ted
by
pati
entrsquo
sh
ealt
hca
reu
tili
zati
on
an
dse
veri
tyo
fd
isea
se6
4
d5
25
o
fh
ou
seh
old
sin
cen
sus
blo
ckgro
up
are
lin
gu
isti
call
yis
ola
ted
e5
20
o
fh
ou
seh
old
sin
cen
sus
blo
ckgro
up
are
livi
ng
bel
ow
po
vert
yli
ne
f 56
6
of
per
son
sin
cen
sus
blo
ckgro
up
are
emp
loye
din
wo
rkin
gcl
ass
occ
up
ati
on
42 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
scores and not living in an economically deprived orworking class neighbourhood
Caucasian respondents were more likely to completethe survey online to have had lower mean A1Cpracticing SMBG and not living in linguisticallyisolated neighbourhood
Filipino respondents were more likely to participatevia written survey to have been older than 45 yearsof age to have had lower levels of LDL and lowerco-morbidity scores
Latino respondents were more likely to participatevia CATI to have been younger than 45 years of ageto not have English as their preferred language tohave had higher mean A1C higher rates of missedappointments poorer medication adherence andliving in a linguistically isolated or working classneighbourhood
What outcomes will be measuredduring follow-upThis cohort will be the basis for longitudinal evalua-tions of a wide range of clinical outcomes associatedwith diabetes and are powered to evaluate the ethnicand educational disparities in diabetes-related com-plication rates (eg myocardial infarction strokeheart failure kidney failure and amputation) andmortality after a 25-year follow-up Clinical andadministrative follow-up data will be captured fromthese same sources as baseline plus deaths from thestate mortality files Data from all sources combinedmay offer insights into how differences at the patient-level (eg differences in behaviours type of therapyintermediate health status and psychosocial factorsand competing demands) provider-level (eg trust inprovider racial concordance and patient-providercommunication) and system-level (eg differentialimpact of cost-sharing) may lead to social disparitiesin diabetes health end-stage diabetes complicationsand mortality
What is the anticipated attritionOne of the strengths of this diabetes registry is thevery low turnover rate (5 discontinue membershipeach year) affording minimal loss to follow-up Onaverage the duration of Kaiser membership amongour diabetes cohort members is 89 years (SDfrac14 66)
What are the strengths andweaknessesA significant weakness is that written and websurveys were only offered in English Howevertelephone interviewers made the initial efforts tocontact subjects who had the opportunity to complete
interviews in Spanish Mandarin Cantonese orTagalog Offering the survey by oral interview inmultiple languages was intended to mitigate thelanguage andor literacy barriers but we still observedlower participation rates by those with less educationand modest response differences by race On a morefavourable note 84 of non-respondents were identi-fied by the health plan as having English as theirpreferred language suggesting that language alonewas not a major barrier to participation
The response rate was as expected lower than ourprevious Diabetes Research Questionnaire (1994ndash97)A recent survey conducted in the Translating ResearchInto Action for Diabetes (TRIAD) study amongmanaged care populations across the United Statesincluding Kaiser43 had a response rate of 50 TheBehavioural Risk Factor Surveillance System alsoexperienced a decline in median response ratesfrom 632 in 1996 to 535 in 200168 Privacyconcerns and competition from telemarketing haveprobably played a role in declining response ratesHowever if generally lower response rates have beenobserved in minority populations the response ratefor this survey may be considered favourable giventhat 77 of our sample were racial or ethnic minorityhealth plan members
One strength of this study is the relatively uniformaccess to care provided by membership in thisintegrated health plan Thus this study largely avoidsdisparities in access and quality which are potentsources of confounding bias in population-basedstudies of social health disparities Another strength isthe quality and richness of the clinical follow-up dataAdministrative and clinical data are available directlyfrom the health plan and are considered to be reliableand complete Each telephone interview was conductedby professional interviewers employed trained andsupervised by the Public Health Institute SurveyResearch Group in Sacramento CA which has itsown internal quality controls Written surveys (long-and short-forms) were coded and edited using anextensive data dictionary and detailed coding rules tostandardize and clarify responses before being sentto the Data Entry department at the KaiserDivision of Research which conducted double dataentry The survey data quality is limited by beingself-reported
The diversity size and wealth of data included inthe DISTANCE cohort make it suitable for theprospective study of a wide range of social disparitiesin the processes and outcomes of diabetes health careInformation about childhood socioeconomic positionand educational attainment will facilitate a life courseapproach6970
Analysis of response biasResponse bias is a concern in any survey Surveys aresubject to response bias if participation is associatedwith the risk factors (exposures) andor diseases
COHORT PROFILE DISTANCE 43
(outcomes) under investigation In this study popula-tion we have substantial data on non-respondentsincluding Kaiser administrative data co-morbidityand census data on neighbourhood characteristicsallowing us to compare hypothesized relationships inrespondents vs non-respondents and adjust for thispotential response bias71 While participation wassomewhat lower among minorities and those withless education overall respondents and non-respon-dents were quite similar
We had pre-baseline reported race data for 29 319(72) members of the DISTANCE cohort andobserved differences in response rates by raceAfrican Americans 52 Asians 46 Caucasians63 Latinos 53 unknown 45 (P lt 00001) Wealso had educational attainment data for 4524subjects (22) who had completed the previoussurvey in 1994ndash97 and observed that subjects with ahigh school education or less participated at a lowerrate than those with more than a high schooleducation (54 vs 60) (P lt 00001)
A more complete view of response bias was obtainedfrom health plan administrative data and census datawhich was available for all members of the cohortWe found few notable differences between respon-dents and non-respondents in demographic clinicalbehavioural or neighbourhood characteristics Com-pared with respondents non-respondents had highermean LDL levels (993 vs 943 mgdl) and were lesslikely to practice SMBG (32 vs 46) However aportion of these unadjusted differences are likely dueto age differences in respondents vs non-respondents
We conducted an assessment of response bias bycomparing differences in A1C across ethnic groupsseparately among respondents and non-respondentsWe specified a logistic regression model of poorglycemic control which regressed the outcome A1C47 on ethnicity age and sex and an interactionterm (ethnicity survey response) Although respon-dents consistently had higher A1C than non-respon-dents ethnic differences in the relationship betweenethnicity and A1C did not differ between respondentsand non-respondents (Pfrac14 055) We conducted asimilar analysis with the subjects for whom we hadeducational attainment data from a previous surveyand again found no substantive response bias(Pfrac14 028) While characteristics differ betweenrespondents and non-respondents (response bias)the associations with outcomes of interest (slopes)are much less vulnerable to such bias7273
Statistical methodsAnalyses of DISTANCE data will focus on associationsrather than descriptive characterization We willemploy a modelling approach using inverse weightingestimation of marginal structural models72 to inves-tigate causal relationships in this study Thesemodels will include weights to adjust for the non-proportionate sampling fractions (eg over-sampling
of the minority ethnic groups) and response bias(eg giving greater weight to respondents whohave characteristics more similar to the non-respondents) Thus the proposed estimating approachwill adjust for confounding and selection biassimultaneously73ndash77
How can I collaborate Where canI find out moreWe are not free to release participantsrsquo personal dataunder our promise to them regarding confidentialityHowever the DISTANCE steering committee is inter-ested in collaborations with external researchersDISTANCE investigators are particularly interested incomparing social disparities observed in this insuredpopulation with uniform access to care to disparitiesobserved in population-based samples where qualityand access to care vary widely by social strataRequests for collaboration must be submitted to thedirector of the central coordinating centre (corre-sponding author) Before collaborations can beinitiated proposals require review and approval bythe DISTANCE Publications and Presentations (PampP)Committee This committee was formed to (i) Ensureaccurate uniform timely and high quality reportingof DISTANCE findings (ii) Preserve the scientificintegrity of the study and (iii) Safeguard the rightsand confidentiality of participants
Supplementary DataSupplementary data are available at IJE online
AcknowledgementsFunds were provided by National Institute ofDiabetes Digestive and Kidney Diseases [R01DK65664] and National Institute of Child Healthand Human Development [R01 HD046113] Dr DS isalso supported by a National Institutes of HealthMentored Clinical Scientist Award [K-23 RR16539]The Public Health InstituteSurvey Research Groupconducted our telephone interviews Gerard andAssociates provided certified translations of all patientmaterials and interview scripts KP Corporation (notaffiliated with Kaiser Permanente) provided printingand mailing services for the written surveys andother communications We thank Marcia Ewing andCarol Rabello for their diligent work on surveyoperations Most of all we thank all the studyparticipants
Conflict of interest None declared
References1 Healthy People 2010 U S Department of Health and
Human Services 2005 Available at httpwwwhealthypeoplegov(Accessed August 2 2007)
44 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
2 Smedley BD Stith AY Nelson AR Unequal TreatmentConfronting Racial and Ethnic Disparities in Health CareWashington DC Institute of Medicine NationalAcademy Press 2002
3 Karter AJ Newman B Rowell S et al Large-scalecollection of family history data and recruitment ofinformative families for genetic analysis J Reg Manag1998257ndash12
4 Karter AJ Ferrara A Darbinian J Ackerson LM Selby JVSelf-monitoring of blood glucose language and financialbarriers in a managed care population with diabetesDiabetes Care 200023477ndash83
5 Karter AJ Moffet HH Liu J et al Achieving goodglycemic control initiation of new antihyperglycemictherapies in patients with type 2 diabetes from theKaiser Permanente Northern California Diabetes RegistryAm J Manag Care 200511262ndash70
6 Karter AJ Ackerson LM Darbinian JA et al Self-monitoring of blood glucose levels and glycemic controlthe Northern California Kaiser Permanente Diabetesregistry Am J Med 20011111ndash9
7 Karter AJ Ferrara A Liu JY Moffet HH Ackerson LMSelby JV Ethnic disparities in diabetic complications inan insured population JAMA 20022872519ndash27
8 Karter AJ Parker MM Moffet HH et al Missedappointments and poor glycemic control an opportunityto identify high-risk diabetic patients Med Care200442110ndash15
9 Selby JV Karter AJ Ackerson LM Ferrara A Liu JDeveloping a prediction rule from automated clinicaldatabases to identify high-risk patients in a largepopulation with diabetes Diabetes Care 2001241547ndash55
10 Martin TL Selby JV Zhang D Physician and patientprevention practices in NIDDM in a large urbanmanaged-care organization Diabetes Care 1995181124ndash32
11 Chaturvedi N Jarrett J Shipley MJ Fuller JHSocioeconomic gradient in morbidity and mortality inpeople with diabetes cohort study findings from theWhitehall Study and the WHO Multinational Study ofVascular Disease in Diabetes BMJ 1998316100ndash5
12 Booth GL Hux JE Relationship between avoidablehospitalizations for diabetes mellitus and income levelArch Intern Med 2003163101ndash6
13 Kleinbaum DG Kupper LL Morgenstern H EpidemiologicResearch Principles and Quantitative Methods New York VanNorstrand Reinhold 1982
14 Dillman DA Mail and Internet Surveys The Tailored DesignMethod 2nd edn New York John Wiley 2000
15 Rudd RE Comings JP Hyde JN Leave no one behindimproving health and risk communication through att-ention to literacy J Health Commun 20038 (Suppl 1)104ndash15
16 Singh-Manoux A Adler NE Marmot MG Subjectivesocial status its determinants and its association withmeasures of ill-health in the Whitehall II study Soc SciMed 2003561321ndash33
17 John D Catherine T MacArthur Research Network onSocioeconomic Status and Health The MacArthur Scale ofSubjective Social Status San Francisco University ofCalifornia 1999 Available at httpwwwmacsesucsfeduResearchPsychosocialnotebooksubjectivehtml(Accessed July 12 2007)
18 Ross CE Wu CL Education age and the cumulativeadvantage in health J Health Soc Behav 199637104ndash20
19 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
20 White S Dillow S Key Concepts and Features of the 2003National Assessment of Adult Literacy (NAALS) (NCES2006ndash471) Washington DC Department of EducationNational Center for Education Statistics 2005
21 Botman SL Moore TF Moriarity CL Parsons VL Designand Estimation for the National Health Interview Survey(NHIS) 1995ndash2004 National Center for Health StatisticsVital Health Stat 2 20001301ndash31
22 Friedman GD Cutter GR Donahue RP et al CARDIAstudy design recruitment and some characteristics of theexamined subjects J Clin Epidemiol 1988411105ndash16
23 Ross CE Wu CL The links between education and healthAm Sociol Rev 199560719ndash45
24 Robert S House JS SES differentials in health by ageand alternative indicators of SES J Aging Health19968359ndash88
25 Toobert DJ Hampson SE Glasgow RE The summary ofdiabetes self-care activities measure results from 7studies and a revised scale Diabetes Care 200023943ndash50
26 Craig CL Marshall AL Sjostrom M et al Internationalphysical activity questionnaire 12-country reliability andvalidity Med Sci Sports Exerc 2003351381ndash95
27 Gordon AJ Maisto SA McNeil M et al Three questionscan detect hazardous drinkers J Fam Pract 200150313ndash20
28 National Health and Nutrition Examination Survey(NHANES) Questionnaire (2001-2002) Centers for DiseaseControl and Prevention (CDC) National Center for HealthStatistics (NCHS) Hyattsville MD US Department ofHealth and Human Services Centers for Disease Controland Prevention 2002
29 Behavioral Risk Factor Surveillance System (BRFSS)Centers for Disease Control and Prevention (CDC) NationalCenter for Health Statistics (NCHS) Hyattsville MD USDepartment of Health and Human Services Centers forDisease Control and Prevention 2003
30 Chien-Wen T Phillips RL Green LA Fryer GE Dovey SMWhat physicians need to know about seniors and limitedprescription benefits and why Am Fam Physician200266212
31 Morisky DE Green LW Levine DM Concurrent andpredictive validity of a self-reported measure of medica-tion adherence Med Care 19862467ndash74
32 Barringer TA Kirk JK Santaniello AC Foley KLMichielutte R Effect of a multivitamin and mineralsupplement on infection and quality of life A random-ized double-blind placebo-controlled trial Ann Intern Med2003138365ndash71
33 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
34 Peyrot M Rubin RR Structure and correlates of diabetes-specific locus of control Diabetes Care 199417994ndash1001
35 Wagenknecht LE Mayer EJ Rewers M et al The insulinresistance atherosclerosis study (IRAS) objectivesdesign and recruitment results Ann Epidemiol19955464ndash72
COHORT PROFILE DISTANCE 45
36 King DE Mainous AG III Pearson WS C-reactiveprotein diabetes and attendance at religious servicesDiabetes Care 2002251172ndash76
37 Berkman LF Syme SL Social networks host resistanceand mortality a nine-year follow-up study of AlamedaCounty residents Am J Epidemiol 1979109186ndash204
38 Ware JE Kosinski M Keller SD SF-12 How to Score theSF-12 Physical and Mental Health Summary Scales 2nd ednBoston The Health Institute New England MedicalCenter 1995
39 Ware JE Kosinksi M Dewey JE Gandek B How to Scoreand Interpret Single-Item Health Status Measures A Manual forUsers of the SF-8 Health Survey Lincoln RI QualityMetricIncorporated 2001
40 Cohen S Kamarck T Mermelstein R A global measure ofperceived stress J Health Soc Behav 198324385ndash96
41 Kroenke K Spitzer RL Williams JB The PHQ-9 validityof a brief depression severity measure J Gen Intern Med200116606ndash13
42 Spitzer RL Kroenke K Williams JB Validation and utilityof a self-report version of PRIME-MD the PHQ primarycare study Primary care evaluation of mental disordersPatient Health Questionnaire JAMA 19992821737ndash44
43 The TRIAD Study Group The translating research intoaction for diabetes (TRIAD) study a multicenter study ofdiabetes in managed care Diabetes Care 200225386ndash89
44 Ayas NT White DP Al Delaimy WK et al A prospectivestudy of self-reported sleep duration and incidentdiabetes in women Diabetes Care 200326380ndash84
45 Sickel AE Moore PJ Adler NE Williams DR Jackson JSThe differential effects of sleep quality and quantity onthe relationship between SES and health Ann NY Acad Sci1999896431ndash34
46 Moore PJ Adler NE Williams DR Jackson JSSocioeconomic status and health the role of sleepPsychosom Med 200264337ndash44
47 Benet-Martinez V John OP Los Cinco Grandes acrosscultures and ethnic groups multitrait multimethodanalyses of the Big Five in Spanish and English J PersSoc Psychol 199875729ndash50
48 Srivastava S John OP Gosling SD Potter J Developmentof personality in early and middle adulthood set likeplaster or persistent change J Pers Soc Psychol2003841041ndash53
49 Gosling SD Rentfrow PJ Swann WB A very briefmeasure of the Big-Five personality domains J Res Pers200337504ndash28
50 Schillinger D Grumbach K Piette J et al Association ofhealth literacy with diabetes outcomes JAMA2002288475ndash82
51 Schillinger D Barton LR Karter AJ Wang F Adler NDoes literacy mediate the relationship between educationand health outcomes A study of a low-income popula-tion with diabetes Public Health Rep 2006121245ndash54
52 Chew LD Bradley KA Boyko EJ Brief questions toidentify patients with inadequate health literacy Fam Med200436588ndash94
53 The American Association of Clinical EndocrinologistsMedical Guidelines for the Management of DiabetesMellitus The AACE System of Intensive DiabetesSelf-Management ndash 2002 Update Endocr Pract 20028(Suppl 1)40ndash82
54 Fitzgerald JT Funnell MM Hess GE et al The reliabilityand validity of a brief diabetes knowledge test DiabetesCare 199821706ndash10
55 Piette JD Heisler M Wagner TH Problems paying out-of-pocket medication costs among older adults withdiabetes Diabetes Care 200427384ndash91
56 Epidemiology of Diabetes Interventions andComplications (EDIC) Design implementation andpreliminary results of a long-term follow-up of thediabetes control and complications trial cohort DiabetesCare 19992299ndash111
57 Krieger N Sidney S Coakley E Racial discrimination andskin color in the CARDIA study implications for publichealth research Coronary artery risk development inyoung adults Am J Public Health 1998881308ndash13
58 Krieger N Smith K Naishadham D Hartman CBarbeau EM Experiences of discrimination validity andreliability of a self-report measure for population healthresearch on racism and health Soc Sci Med 2005611576ndash96
59 CAHPS 20 Questionnaires Crofton C AHCPR Pub No97-R079 Washington DC Agency for Health Care Policyand Research 1998
60 Hays RD Shaul JA Williams VS et al Psychometricproperties of the CAHPS 10 survey measures Consumerassessment of health plans study Med Care 199937MS22ndash31
61 Anderson LA Dedrick RF Development of the trust inphysician scale a measure to assess interpersonal trust inpatient-physician relationships Psychol Rep 1990671091ndash100
62 Thom DH Ribisl KM Stewart AL Luke DA Furthervalidation and reliability testing of the Trust in PhysicianScale The Stanford trust study physicians Med Care199937510ndash17
63 Steiner JF Koepsell TD Fihn SD Inui TS A generalmethod of compliance assessment using centralizedpharmacy records Description and validation Med Care198826814ndash23
64 Clark DO Von Korff M Saunders K Baluch WMSimon GE A chronic disease score with empiricallyderived weights Med Care 199533783ndash95
65 Kaiser Permanente Business Intelligence Report 2006Research and Markets 2006 Available at httpwwwresearchandmarketscomreportinfoaspreport_idfrac14451423(Accessed July 9 2007)
66 Krieger N Overcoming the absence of socioeconomic datain medical records validation and application of acensus-based methodology Am J Public Health 199282703ndash10
67 Hiatt RA Friedman GD Characteristics of patientsreferred for treatment of end-stage renal diseasein a defined population Am J Public Health 198272829ndash33
68 Porter S Jackson K Trosclaire A Pederson LL Office onsmoking and Health NCfCDPaHPC Prevalence of currentcigarette smoking among adults and changes in pre-valence of current and some day smoking ndash UnitedStates 1996ndash2001 MMWR 200352303ndash7
69 Smith GD Hart C Blane D Gillis C Hawthorne VLifetime socioeconomic position and mortality prospec-tive observational study BMJ 1997314547ndash52
46 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
70 Lynch JW Kaplan GA Shema SJ Cumulative impact ofsustained economic hardship on physical cognitivepsychological and social functioning N Engl J Med19973371889ndash95
71 Melton LJ3 Dyck PJ Karnes JL OrsquoBrien PC Service FJNon-response bias in studies of diabetic complicationsthe Rochester Diabetic Neuropathy Study J Clin Epidemiol199346341ndash48
72 Robins JM Association causation and marginal struc-tural models Synthese 1999121151ndash79
73 Hernan MA Hernandez-Diaz S Robins JM A structuralapproach to selection bias Epidemiology 200415615ndash25
74 Bodnar LM Davidian M Siega-Riz AM Tsiatis AAMarginal structural models for analyzing causal effects
of time-dependent treatments an application in perinatalepidemiology Am J Epidemiol 2004159926ndash34
75 Bryan JF Yu Z van der Laan M Analysis of longitudinalmarginal structural models U C Berkeley Division ofBiolstatistics Working Paper Series Paper 1201ndash29 2002
76 Neugebauer R van der Laan M Why prefer double robustestimators in causal inference J Stat Plan Inference2005129405ndash26
77 Wang Y Petersen ML Bangsberg D van der Laan MDiagnosing bias in the inverse probability of treatmentweighted estimator resulting from violation ofexperimental treatment assignment U C BerkeleyDivision of Biolstatistics Working Paper Series Paper 2111ndash25 2006
COHORT PROFILE DISTANCE 47
external locus of control3334 social support2335ndash37
health status3839 stress40 depression2241ndash43 sleepquality44ndash46 chronic pain38 personality traits47ndash49
health literacy50ndash52 diabetes knowledge5354 medicaland dental coverage medical costs and ability topay55 medical visit travel time43 language barriersdelayed treatment56 provider recommendationsregarding aspirin43 and SMBG43 foot exam byphysician43 quality of care43 discrimination5758
provider communication and interactions5960 trustin provider6162 and self-reported height weight43
symptoms complications43 erectile dysfunction orfemale urinary incontinence
Survey modesThe survey included four modes of administration (i)a computer-assisted telephone interview (CATI)administered by a third party (ii) a password-enabled internet-accessible survey (lsquoweb surveyrsquo)maintained on a secure server at the Kaiser Divisionof Research (iii) a self-administered written surveyor (iv) a short version of the written survey Thecontent of each survey mode was identical except forslight adjustments in wording and logic patterns asneeded and was estimated to take 45ndash60 min tocomplete the short written version was abridged andcontained 40 questions The written and web surveyswere in English only but the CATI was available inEnglish Spanish Cantonese Mandarin and Tagalogusing certified translations of an English script andwas intended to maximize accessibility to those withlanguage barriers or limited English literacy orfluency
Other baseline dataWe geocoded the home addresses of all respondentsand non-respondents and linked their census blockgroup and tract to 2000 Census data These censuscharacteristics were then factor-analysed and wedeveloped an eight-item lsquodeprivation indexrsquo as a wayof summarizing the many indicators The deprivationindex has excellent internal reliability (Cronbachalphafrac14 093) with each of the eight items loadingalmost equally onto the underlying construct
Baseline clinical characterizations were obtainedfrom the extensive Kaiser administrative databasesincluding smoking status body mass index currentpharmacotherapy utilization and adherence63 labora-tory findings history of co-morbid events andprocedures use of emergency room outpatient andinpatient health services costs outpatient and inpa-tient risk scores based on health care utilization andseverity of disease64 and end-stage renal diseaseregistry (linked to the United States Renal DataSystem) These data are available for all subjectsboth respondents and non-respondents and allow fora substantive assessment of survey response bias
Who is in the sampleThis study took place within the Kaiser FoundationHealth Plan the largest not-for-profit managed healthcare company in the United States65 Study subjectswere members of Kaiser Permanente NorthernCalifornia (KPNC) which provides comprehensivemedical services to over 32 million members (as ofJanuary 2005) in the San Francisco Bay andSacramento metropolitan areas or 25ndash30 of theregionrsquos population KPNC members are predomi-nantly employed or retired individuals and theirfamilies and closely approximate the general popula-tion ethnically and socioeconomically except forthe extreme tails of income distribution66667 TheDiabetes Registry consisted of 199 123 members as ofJanuary 1 2005 and from this we selected anethnically stratified random sample of 40 735 healthplan members aged 30ndash75 at baseline to receive theDISTANCE Survey 6871 African American (17)11 197 Asian (27) 4233 Caucasian (10) 7018Latino (17) and 11 417 members of unknownethnicity (28) The DISTANCE Survey was in thefield from May 5 2005 until December 31 2006
The survey was completed by 20 188 persons(Table 3) The participation by ethnicity was 3420African Americans (169) 2312 Asian (114) 4602Caucasians (228) 2404 Filipinos (119) 3717Latinos (184) 2222 multi-racial (110) and 1511South Asian Pacific Islander Native America Eskimoor otherunknown (75) The distribution by modewas 10 429 CATI (517) 4288 written survey(212) 2393 short version (118) and 3078 web(152) Using an algorithm endorsed by the Councilof American Survey Research Organizations (CASRO)if persons unable to be contacted had the same rate ofeligibility as those contacted and were counted in thedenominator the survey response rate was 62 Ofthe 20 188 respondents to this current survey 4524subjects (22) had also responded to the priorDiabetes Registry Questionnaire (1994ndash97)
What has been measuredThe entire cohort has been characterized usingdemographic clinical and behavioural data fromadministrative records and census data
African American respondents were more likely tobe female to have had English as their preferredlanguage to have had elevated levels of low-densitylipoprotein (LDL) uncontrolled blood pressure werecurrent smokers with higher co-morbidity scores andliving in a deprived neighbourhood
Asian respondents were more likely to be male over60 years of age to not have had English as theirpreferred language to have had lower levels of LDLbetter blood pressure control lower missed appoint-ment rate better medication adherence were non-smokers not practicing SMBG with lower co-morbidity
COHORT PROFILE DISTANCE 41
Ta
ble
3C
hara
cter
isti
cso
fsu
rvey
resp
on
den
tsan
dn
on
-res
po
nd
ents
Re
spo
nd
en
tsn
()
or
Me
an
plusmn(S
D)
No
n-r
esp
on
de
nts
Afr
ica
nA
me
rica
n
Asi
an
(Ch
ine
se
Ko
rea
n
Vie
tna
me
se
Ja
pa
ne
se)
Ca
uca
sia
nF
ilip
ino
La
tin
oM
ult
ira
cia
lO
the
rau
nk
no
wn
All
resp
on
de
nts
No
nre
spo
nd
en
tsN
um
be
r(
)o
rM
ea
nplusmn
(SD
)
n(r
ow
)
34
20
(16
9)
23
12
(11
4)
46
02
(22
8)
24
04
(11
9)
37
17
(18
4)
22
22
(11
0)
15
11
(75
)2
01
88
(10
0)
20
54
7(1
00
)
Su
rve
yM
od
e
Tel
eph
on
eIn
terv
iew
18
06
(52
8)
97
1(4
20
)2
05
1(4
46
)8
46
(35
2)
22
29
(60
0)
15
66
(70
5)
96
0(6
35
)1
04
29
(51
7)
0
Wri
tten
-lo
ng
77
1(2
25
)5
75
(24
9)
97
3(2
11
)7
23
(30
1)
67
6(1
82
)3
46
(15
6)
22
4(1
48
)4
28
8(2
12
)0
Wri
tten
-sh
ort
43
9(1
28
)2
59
(11
2)
46
9(1
02
)5
16
(21
5)
44
8(1
20
)1
08
(49
)1
54
(10
2)
23
93
(11
8)
0
Web
40
4(1
18
)5
07
(21
9)
11
09
(24
1)
31
9(1
33
)3
64
(98
)2
02
(91
)1
73
(11
4)
30
78
(15
2)
0
Fem
ale
19
39
(56
7)
10
19
(44
1)
20
35
(44
2)
12
46
(51
8)
18
86
(50
7)
11
22
(50
5)
59
3(3
93
)9
84
0(4
87
)9
53
7(4
64
)
Ag
e(y
ea
rs)
30
ndash4
42
74
(80
)1
37
(59
)3
82
(83
)1
68
(70
)5
64
(15
2)
25
3(1
14
)1
89
(12
5)
19
67
(97
)2
58
0(1
26
)
45
ndash5
91
34
7(3
94
)9
34
(40
4)
17
94
(39
0)
11
06
(46
0)
16
39
(44
1)
94
3(4
24
)6
95
(46
0)
84
58
(41
9)
87
55
(42
6)
60thorn
17
99
(52
6)
12
41
(53
7)
24
26
(52
7)
11
30
(47
0)
15
14
(40
7)
10
26
(46
2)
62
7(4
15
)9
76
3(4
84
)9
21
2(4
48
)
Lan
gu
age
pre
fere
nce
En
gli
shsp
eak
ing
32
21
(94
2)
13
76
(59
5)
43
94
(95
5)
17
49
(72
8)
22
17
(59
6)
16
91
(76
1)
11
64
(77
0)
15
81
2(7
83
)1
72
12
(83
8)
Mea
nA
1C5
7
16
55
(54
0)
10
87
(50
1)
19
15
(46
8)
13
52
(60
4)
19
57
(58
6)
10
98
(56
2)
77
9(5
79
)9
84
3(5
41
)8
34
3(5
13
)
Mea
nL
DL5
10
01
25
0(4
17
)6
26
(28
9)
13
91
(34
6)
63
4(2
86
)1
20
2(3
71
)7
36
(38
1)
48
2(3
63
)6
32
1(3
53
)6
90
9(4
24
)
Un
con
tro
lled
blo
od
pre
ssu
re(S4
13
0o
rD4
80
)1
96
6(6
14
)9
13
(42
8)
22
64
(54
7)
10
41
(46
5)
17
46
(51
2)
10
89
(54
2)
68
1(4
95
)9
70
0(5
24
)9
58
1(5
43
)
Mis
sed
ap
po
intm
ent
rate
(per
year)
01
7plusmn
(01
8)
01
0plusmn
(01
4)
01
3plusmn
(01
6)
01
3plusmn
(01
7)
01
8plusmn
(01
9)
01
7plusmn
(01
9)
01
8plusmn
(01
9)
01
5plusmn
(01
7)
01
8plusmn
(02
2)
Po
or
med
icati
on
ad
her
ence
b9
55
(37
1)
46
1(2
51
)9
61
(28
2)
59
2(2
93
)1
24
7(4
28
)6
36
(36
4)
42
4(3
51
)5
27
6(3
36
)5
10
8(3
90
)
Sm
ok
ing
(cu
rren
t)3
48
(10
2)
93
(40
)4
10
(90
)1
47
(61
)2
45
(66
)2
00
(91
)1
05
(70
)1
54
8(7
7)
18
29
(90
)
Pra
ctic
esS
MB
G1
57
3(4
60
)1
00
4(4
34
)2
24
6(4
88
)1
11
5(4
64
)1
62
2(4
36
)1
02
5(4
61
)6
23
(41
2)
92
08
(45
6)
65
32
(31
8)
Ou
tpati
ent
risk
sco
rec
24
(21
)1
8(1
1)
21
(17
)1
8(1
0)
19
(14
)2
2(1
6)
19
(14
)2
0(1
6)
18
(15
)
Inp
ati
ent
risk
sco
rec
26
(42
)1
5(2
0)
22
(33
)1
4(2
1)
18
(30
)2
2(3
9)
17
(29
)2
0(3
2)
18
(34
)
Lin
gu
isti
call
yis
ola
ted
nei
gh
bo
urh
oo
dd
10
5(3
1)
11
5(5
0)
55
(12
)1
14
(48
)2
81
(76
)1
13
(51
)6
2(4
1)
84
5(4
2)
10
35
(51
)
Eco
no
mic
all
yd
epri
ved
nei
gh
bo
urh
oo
de
82
7(2
43
)1
24
(54
)3
46
(76
)1
48
(62
)6
68
(18
2)
32
3(1
47
)1
67
(11
2)
26
03
(13
0)
27
24
(13
4)
Wo
rkin
gcl
ass
nei
gh
bo
urh
oo
df
14
40
(42
2)
40
1(1
74
)1
16
8(2
58
)8
84
(37
0)
17
49
(47
6)
86
8(3
94
)4
64
(31
0)
69
74
(34
9)
75
66
(37
2)
aIn
clu
des
Paci
fic
Isla
nd
ers
Nati
veA
mer
ican
sS
ou
thA
sian
o
ther
un
spec
ifie
dan
du
nk
no
wn
(did
no
tan
swer
)b5
20
co
nti
nu
ou
sm
edic
ati
on
gap
s6
3
c Th
eyare
wei
gh
ted
by
pati
entrsquo
sh
ealt
hca
reu
tili
zati
on
an
dse
veri
tyo
fd
isea
se6
4
d5
25
o
fh
ou
seh
old
sin
cen
sus
blo
ckgro
up
are
lin
gu
isti
call
yis
ola
ted
e5
20
o
fh
ou
seh
old
sin
cen
sus
blo
ckgro
up
are
livi
ng
bel
ow
po
vert
yli
ne
f 56
6
of
per
son
sin
cen
sus
blo
ckgro
up
are
emp
loye
din
wo
rkin
gcl
ass
occ
up
ati
on
42 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
scores and not living in an economically deprived orworking class neighbourhood
Caucasian respondents were more likely to completethe survey online to have had lower mean A1Cpracticing SMBG and not living in linguisticallyisolated neighbourhood
Filipino respondents were more likely to participatevia written survey to have been older than 45 yearsof age to have had lower levels of LDL and lowerco-morbidity scores
Latino respondents were more likely to participatevia CATI to have been younger than 45 years of ageto not have English as their preferred language tohave had higher mean A1C higher rates of missedappointments poorer medication adherence andliving in a linguistically isolated or working classneighbourhood
What outcomes will be measuredduring follow-upThis cohort will be the basis for longitudinal evalua-tions of a wide range of clinical outcomes associatedwith diabetes and are powered to evaluate the ethnicand educational disparities in diabetes-related com-plication rates (eg myocardial infarction strokeheart failure kidney failure and amputation) andmortality after a 25-year follow-up Clinical andadministrative follow-up data will be captured fromthese same sources as baseline plus deaths from thestate mortality files Data from all sources combinedmay offer insights into how differences at the patient-level (eg differences in behaviours type of therapyintermediate health status and psychosocial factorsand competing demands) provider-level (eg trust inprovider racial concordance and patient-providercommunication) and system-level (eg differentialimpact of cost-sharing) may lead to social disparitiesin diabetes health end-stage diabetes complicationsand mortality
What is the anticipated attritionOne of the strengths of this diabetes registry is thevery low turnover rate (5 discontinue membershipeach year) affording minimal loss to follow-up Onaverage the duration of Kaiser membership amongour diabetes cohort members is 89 years (SDfrac14 66)
What are the strengths andweaknessesA significant weakness is that written and websurveys were only offered in English Howevertelephone interviewers made the initial efforts tocontact subjects who had the opportunity to complete
interviews in Spanish Mandarin Cantonese orTagalog Offering the survey by oral interview inmultiple languages was intended to mitigate thelanguage andor literacy barriers but we still observedlower participation rates by those with less educationand modest response differences by race On a morefavourable note 84 of non-respondents were identi-fied by the health plan as having English as theirpreferred language suggesting that language alonewas not a major barrier to participation
The response rate was as expected lower than ourprevious Diabetes Research Questionnaire (1994ndash97)A recent survey conducted in the Translating ResearchInto Action for Diabetes (TRIAD) study amongmanaged care populations across the United Statesincluding Kaiser43 had a response rate of 50 TheBehavioural Risk Factor Surveillance System alsoexperienced a decline in median response ratesfrom 632 in 1996 to 535 in 200168 Privacyconcerns and competition from telemarketing haveprobably played a role in declining response ratesHowever if generally lower response rates have beenobserved in minority populations the response ratefor this survey may be considered favourable giventhat 77 of our sample were racial or ethnic minorityhealth plan members
One strength of this study is the relatively uniformaccess to care provided by membership in thisintegrated health plan Thus this study largely avoidsdisparities in access and quality which are potentsources of confounding bias in population-basedstudies of social health disparities Another strength isthe quality and richness of the clinical follow-up dataAdministrative and clinical data are available directlyfrom the health plan and are considered to be reliableand complete Each telephone interview was conductedby professional interviewers employed trained andsupervised by the Public Health Institute SurveyResearch Group in Sacramento CA which has itsown internal quality controls Written surveys (long-and short-forms) were coded and edited using anextensive data dictionary and detailed coding rules tostandardize and clarify responses before being sentto the Data Entry department at the KaiserDivision of Research which conducted double dataentry The survey data quality is limited by beingself-reported
The diversity size and wealth of data included inthe DISTANCE cohort make it suitable for theprospective study of a wide range of social disparitiesin the processes and outcomes of diabetes health careInformation about childhood socioeconomic positionand educational attainment will facilitate a life courseapproach6970
Analysis of response biasResponse bias is a concern in any survey Surveys aresubject to response bias if participation is associatedwith the risk factors (exposures) andor diseases
COHORT PROFILE DISTANCE 43
(outcomes) under investigation In this study popula-tion we have substantial data on non-respondentsincluding Kaiser administrative data co-morbidityand census data on neighbourhood characteristicsallowing us to compare hypothesized relationships inrespondents vs non-respondents and adjust for thispotential response bias71 While participation wassomewhat lower among minorities and those withless education overall respondents and non-respon-dents were quite similar
We had pre-baseline reported race data for 29 319(72) members of the DISTANCE cohort andobserved differences in response rates by raceAfrican Americans 52 Asians 46 Caucasians63 Latinos 53 unknown 45 (P lt 00001) Wealso had educational attainment data for 4524subjects (22) who had completed the previoussurvey in 1994ndash97 and observed that subjects with ahigh school education or less participated at a lowerrate than those with more than a high schooleducation (54 vs 60) (P lt 00001)
A more complete view of response bias was obtainedfrom health plan administrative data and census datawhich was available for all members of the cohortWe found few notable differences between respon-dents and non-respondents in demographic clinicalbehavioural or neighbourhood characteristics Com-pared with respondents non-respondents had highermean LDL levels (993 vs 943 mgdl) and were lesslikely to practice SMBG (32 vs 46) However aportion of these unadjusted differences are likely dueto age differences in respondents vs non-respondents
We conducted an assessment of response bias bycomparing differences in A1C across ethnic groupsseparately among respondents and non-respondentsWe specified a logistic regression model of poorglycemic control which regressed the outcome A1C47 on ethnicity age and sex and an interactionterm (ethnicity survey response) Although respon-dents consistently had higher A1C than non-respon-dents ethnic differences in the relationship betweenethnicity and A1C did not differ between respondentsand non-respondents (Pfrac14 055) We conducted asimilar analysis with the subjects for whom we hadeducational attainment data from a previous surveyand again found no substantive response bias(Pfrac14 028) While characteristics differ betweenrespondents and non-respondents (response bias)the associations with outcomes of interest (slopes)are much less vulnerable to such bias7273
Statistical methodsAnalyses of DISTANCE data will focus on associationsrather than descriptive characterization We willemploy a modelling approach using inverse weightingestimation of marginal structural models72 to inves-tigate causal relationships in this study Thesemodels will include weights to adjust for the non-proportionate sampling fractions (eg over-sampling
of the minority ethnic groups) and response bias(eg giving greater weight to respondents whohave characteristics more similar to the non-respondents) Thus the proposed estimating approachwill adjust for confounding and selection biassimultaneously73ndash77
How can I collaborate Where canI find out moreWe are not free to release participantsrsquo personal dataunder our promise to them regarding confidentialityHowever the DISTANCE steering committee is inter-ested in collaborations with external researchersDISTANCE investigators are particularly interested incomparing social disparities observed in this insuredpopulation with uniform access to care to disparitiesobserved in population-based samples where qualityand access to care vary widely by social strataRequests for collaboration must be submitted to thedirector of the central coordinating centre (corre-sponding author) Before collaborations can beinitiated proposals require review and approval bythe DISTANCE Publications and Presentations (PampP)Committee This committee was formed to (i) Ensureaccurate uniform timely and high quality reportingof DISTANCE findings (ii) Preserve the scientificintegrity of the study and (iii) Safeguard the rightsand confidentiality of participants
Supplementary DataSupplementary data are available at IJE online
AcknowledgementsFunds were provided by National Institute ofDiabetes Digestive and Kidney Diseases [R01DK65664] and National Institute of Child Healthand Human Development [R01 HD046113] Dr DS isalso supported by a National Institutes of HealthMentored Clinical Scientist Award [K-23 RR16539]The Public Health InstituteSurvey Research Groupconducted our telephone interviews Gerard andAssociates provided certified translations of all patientmaterials and interview scripts KP Corporation (notaffiliated with Kaiser Permanente) provided printingand mailing services for the written surveys andother communications We thank Marcia Ewing andCarol Rabello for their diligent work on surveyoperations Most of all we thank all the studyparticipants
Conflict of interest None declared
References1 Healthy People 2010 U S Department of Health and
Human Services 2005 Available at httpwwwhealthypeoplegov(Accessed August 2 2007)
44 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
2 Smedley BD Stith AY Nelson AR Unequal TreatmentConfronting Racial and Ethnic Disparities in Health CareWashington DC Institute of Medicine NationalAcademy Press 2002
3 Karter AJ Newman B Rowell S et al Large-scalecollection of family history data and recruitment ofinformative families for genetic analysis J Reg Manag1998257ndash12
4 Karter AJ Ferrara A Darbinian J Ackerson LM Selby JVSelf-monitoring of blood glucose language and financialbarriers in a managed care population with diabetesDiabetes Care 200023477ndash83
5 Karter AJ Moffet HH Liu J et al Achieving goodglycemic control initiation of new antihyperglycemictherapies in patients with type 2 diabetes from theKaiser Permanente Northern California Diabetes RegistryAm J Manag Care 200511262ndash70
6 Karter AJ Ackerson LM Darbinian JA et al Self-monitoring of blood glucose levels and glycemic controlthe Northern California Kaiser Permanente Diabetesregistry Am J Med 20011111ndash9
7 Karter AJ Ferrara A Liu JY Moffet HH Ackerson LMSelby JV Ethnic disparities in diabetic complications inan insured population JAMA 20022872519ndash27
8 Karter AJ Parker MM Moffet HH et al Missedappointments and poor glycemic control an opportunityto identify high-risk diabetic patients Med Care200442110ndash15
9 Selby JV Karter AJ Ackerson LM Ferrara A Liu JDeveloping a prediction rule from automated clinicaldatabases to identify high-risk patients in a largepopulation with diabetes Diabetes Care 2001241547ndash55
10 Martin TL Selby JV Zhang D Physician and patientprevention practices in NIDDM in a large urbanmanaged-care organization Diabetes Care 1995181124ndash32
11 Chaturvedi N Jarrett J Shipley MJ Fuller JHSocioeconomic gradient in morbidity and mortality inpeople with diabetes cohort study findings from theWhitehall Study and the WHO Multinational Study ofVascular Disease in Diabetes BMJ 1998316100ndash5
12 Booth GL Hux JE Relationship between avoidablehospitalizations for diabetes mellitus and income levelArch Intern Med 2003163101ndash6
13 Kleinbaum DG Kupper LL Morgenstern H EpidemiologicResearch Principles and Quantitative Methods New York VanNorstrand Reinhold 1982
14 Dillman DA Mail and Internet Surveys The Tailored DesignMethod 2nd edn New York John Wiley 2000
15 Rudd RE Comings JP Hyde JN Leave no one behindimproving health and risk communication through att-ention to literacy J Health Commun 20038 (Suppl 1)104ndash15
16 Singh-Manoux A Adler NE Marmot MG Subjectivesocial status its determinants and its association withmeasures of ill-health in the Whitehall II study Soc SciMed 2003561321ndash33
17 John D Catherine T MacArthur Research Network onSocioeconomic Status and Health The MacArthur Scale ofSubjective Social Status San Francisco University ofCalifornia 1999 Available at httpwwwmacsesucsfeduResearchPsychosocialnotebooksubjectivehtml(Accessed July 12 2007)
18 Ross CE Wu CL Education age and the cumulativeadvantage in health J Health Soc Behav 199637104ndash20
19 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
20 White S Dillow S Key Concepts and Features of the 2003National Assessment of Adult Literacy (NAALS) (NCES2006ndash471) Washington DC Department of EducationNational Center for Education Statistics 2005
21 Botman SL Moore TF Moriarity CL Parsons VL Designand Estimation for the National Health Interview Survey(NHIS) 1995ndash2004 National Center for Health StatisticsVital Health Stat 2 20001301ndash31
22 Friedman GD Cutter GR Donahue RP et al CARDIAstudy design recruitment and some characteristics of theexamined subjects J Clin Epidemiol 1988411105ndash16
23 Ross CE Wu CL The links between education and healthAm Sociol Rev 199560719ndash45
24 Robert S House JS SES differentials in health by ageand alternative indicators of SES J Aging Health19968359ndash88
25 Toobert DJ Hampson SE Glasgow RE The summary ofdiabetes self-care activities measure results from 7studies and a revised scale Diabetes Care 200023943ndash50
26 Craig CL Marshall AL Sjostrom M et al Internationalphysical activity questionnaire 12-country reliability andvalidity Med Sci Sports Exerc 2003351381ndash95
27 Gordon AJ Maisto SA McNeil M et al Three questionscan detect hazardous drinkers J Fam Pract 200150313ndash20
28 National Health and Nutrition Examination Survey(NHANES) Questionnaire (2001-2002) Centers for DiseaseControl and Prevention (CDC) National Center for HealthStatistics (NCHS) Hyattsville MD US Department ofHealth and Human Services Centers for Disease Controland Prevention 2002
29 Behavioral Risk Factor Surveillance System (BRFSS)Centers for Disease Control and Prevention (CDC) NationalCenter for Health Statistics (NCHS) Hyattsville MD USDepartment of Health and Human Services Centers forDisease Control and Prevention 2003
30 Chien-Wen T Phillips RL Green LA Fryer GE Dovey SMWhat physicians need to know about seniors and limitedprescription benefits and why Am Fam Physician200266212
31 Morisky DE Green LW Levine DM Concurrent andpredictive validity of a self-reported measure of medica-tion adherence Med Care 19862467ndash74
32 Barringer TA Kirk JK Santaniello AC Foley KLMichielutte R Effect of a multivitamin and mineralsupplement on infection and quality of life A random-ized double-blind placebo-controlled trial Ann Intern Med2003138365ndash71
33 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
34 Peyrot M Rubin RR Structure and correlates of diabetes-specific locus of control Diabetes Care 199417994ndash1001
35 Wagenknecht LE Mayer EJ Rewers M et al The insulinresistance atherosclerosis study (IRAS) objectivesdesign and recruitment results Ann Epidemiol19955464ndash72
COHORT PROFILE DISTANCE 45
36 King DE Mainous AG III Pearson WS C-reactiveprotein diabetes and attendance at religious servicesDiabetes Care 2002251172ndash76
37 Berkman LF Syme SL Social networks host resistanceand mortality a nine-year follow-up study of AlamedaCounty residents Am J Epidemiol 1979109186ndash204
38 Ware JE Kosinski M Keller SD SF-12 How to Score theSF-12 Physical and Mental Health Summary Scales 2nd ednBoston The Health Institute New England MedicalCenter 1995
39 Ware JE Kosinksi M Dewey JE Gandek B How to Scoreand Interpret Single-Item Health Status Measures A Manual forUsers of the SF-8 Health Survey Lincoln RI QualityMetricIncorporated 2001
40 Cohen S Kamarck T Mermelstein R A global measure ofperceived stress J Health Soc Behav 198324385ndash96
41 Kroenke K Spitzer RL Williams JB The PHQ-9 validityof a brief depression severity measure J Gen Intern Med200116606ndash13
42 Spitzer RL Kroenke K Williams JB Validation and utilityof a self-report version of PRIME-MD the PHQ primarycare study Primary care evaluation of mental disordersPatient Health Questionnaire JAMA 19992821737ndash44
43 The TRIAD Study Group The translating research intoaction for diabetes (TRIAD) study a multicenter study ofdiabetes in managed care Diabetes Care 200225386ndash89
44 Ayas NT White DP Al Delaimy WK et al A prospectivestudy of self-reported sleep duration and incidentdiabetes in women Diabetes Care 200326380ndash84
45 Sickel AE Moore PJ Adler NE Williams DR Jackson JSThe differential effects of sleep quality and quantity onthe relationship between SES and health Ann NY Acad Sci1999896431ndash34
46 Moore PJ Adler NE Williams DR Jackson JSSocioeconomic status and health the role of sleepPsychosom Med 200264337ndash44
47 Benet-Martinez V John OP Los Cinco Grandes acrosscultures and ethnic groups multitrait multimethodanalyses of the Big Five in Spanish and English J PersSoc Psychol 199875729ndash50
48 Srivastava S John OP Gosling SD Potter J Developmentof personality in early and middle adulthood set likeplaster or persistent change J Pers Soc Psychol2003841041ndash53
49 Gosling SD Rentfrow PJ Swann WB A very briefmeasure of the Big-Five personality domains J Res Pers200337504ndash28
50 Schillinger D Grumbach K Piette J et al Association ofhealth literacy with diabetes outcomes JAMA2002288475ndash82
51 Schillinger D Barton LR Karter AJ Wang F Adler NDoes literacy mediate the relationship between educationand health outcomes A study of a low-income popula-tion with diabetes Public Health Rep 2006121245ndash54
52 Chew LD Bradley KA Boyko EJ Brief questions toidentify patients with inadequate health literacy Fam Med200436588ndash94
53 The American Association of Clinical EndocrinologistsMedical Guidelines for the Management of DiabetesMellitus The AACE System of Intensive DiabetesSelf-Management ndash 2002 Update Endocr Pract 20028(Suppl 1)40ndash82
54 Fitzgerald JT Funnell MM Hess GE et al The reliabilityand validity of a brief diabetes knowledge test DiabetesCare 199821706ndash10
55 Piette JD Heisler M Wagner TH Problems paying out-of-pocket medication costs among older adults withdiabetes Diabetes Care 200427384ndash91
56 Epidemiology of Diabetes Interventions andComplications (EDIC) Design implementation andpreliminary results of a long-term follow-up of thediabetes control and complications trial cohort DiabetesCare 19992299ndash111
57 Krieger N Sidney S Coakley E Racial discrimination andskin color in the CARDIA study implications for publichealth research Coronary artery risk development inyoung adults Am J Public Health 1998881308ndash13
58 Krieger N Smith K Naishadham D Hartman CBarbeau EM Experiences of discrimination validity andreliability of a self-report measure for population healthresearch on racism and health Soc Sci Med 2005611576ndash96
59 CAHPS 20 Questionnaires Crofton C AHCPR Pub No97-R079 Washington DC Agency for Health Care Policyand Research 1998
60 Hays RD Shaul JA Williams VS et al Psychometricproperties of the CAHPS 10 survey measures Consumerassessment of health plans study Med Care 199937MS22ndash31
61 Anderson LA Dedrick RF Development of the trust inphysician scale a measure to assess interpersonal trust inpatient-physician relationships Psychol Rep 1990671091ndash100
62 Thom DH Ribisl KM Stewart AL Luke DA Furthervalidation and reliability testing of the Trust in PhysicianScale The Stanford trust study physicians Med Care199937510ndash17
63 Steiner JF Koepsell TD Fihn SD Inui TS A generalmethod of compliance assessment using centralizedpharmacy records Description and validation Med Care198826814ndash23
64 Clark DO Von Korff M Saunders K Baluch WMSimon GE A chronic disease score with empiricallyderived weights Med Care 199533783ndash95
65 Kaiser Permanente Business Intelligence Report 2006Research and Markets 2006 Available at httpwwwresearchandmarketscomreportinfoaspreport_idfrac14451423(Accessed July 9 2007)
66 Krieger N Overcoming the absence of socioeconomic datain medical records validation and application of acensus-based methodology Am J Public Health 199282703ndash10
67 Hiatt RA Friedman GD Characteristics of patientsreferred for treatment of end-stage renal diseasein a defined population Am J Public Health 198272829ndash33
68 Porter S Jackson K Trosclaire A Pederson LL Office onsmoking and Health NCfCDPaHPC Prevalence of currentcigarette smoking among adults and changes in pre-valence of current and some day smoking ndash UnitedStates 1996ndash2001 MMWR 200352303ndash7
69 Smith GD Hart C Blane D Gillis C Hawthorne VLifetime socioeconomic position and mortality prospec-tive observational study BMJ 1997314547ndash52
46 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
70 Lynch JW Kaplan GA Shema SJ Cumulative impact ofsustained economic hardship on physical cognitivepsychological and social functioning N Engl J Med19973371889ndash95
71 Melton LJ3 Dyck PJ Karnes JL OrsquoBrien PC Service FJNon-response bias in studies of diabetic complicationsthe Rochester Diabetic Neuropathy Study J Clin Epidemiol199346341ndash48
72 Robins JM Association causation and marginal struc-tural models Synthese 1999121151ndash79
73 Hernan MA Hernandez-Diaz S Robins JM A structuralapproach to selection bias Epidemiology 200415615ndash25
74 Bodnar LM Davidian M Siega-Riz AM Tsiatis AAMarginal structural models for analyzing causal effects
of time-dependent treatments an application in perinatalepidemiology Am J Epidemiol 2004159926ndash34
75 Bryan JF Yu Z van der Laan M Analysis of longitudinalmarginal structural models U C Berkeley Division ofBiolstatistics Working Paper Series Paper 1201ndash29 2002
76 Neugebauer R van der Laan M Why prefer double robustestimators in causal inference J Stat Plan Inference2005129405ndash26
77 Wang Y Petersen ML Bangsberg D van der Laan MDiagnosing bias in the inverse probability of treatmentweighted estimator resulting from violation ofexperimental treatment assignment U C BerkeleyDivision of Biolstatistics Working Paper Series Paper 2111ndash25 2006
COHORT PROFILE DISTANCE 47
Ta
ble
3C
hara
cter
isti
cso
fsu
rvey
resp
on
den
tsan
dn
on
-res
po
nd
ents
Re
spo
nd
en
tsn
()
or
Me
an
plusmn(S
D)
No
n-r
esp
on
de
nts
Afr
ica
nA
me
rica
n
Asi
an
(Ch
ine
se
Ko
rea
n
Vie
tna
me
se
Ja
pa
ne
se)
Ca
uca
sia
nF
ilip
ino
La
tin
oM
ult
ira
cia
lO
the
rau
nk
no
wn
All
resp
on
de
nts
No
nre
spo
nd
en
tsN
um
be
r(
)o
rM
ea
nplusmn
(SD
)
n(r
ow
)
34
20
(16
9)
23
12
(11
4)
46
02
(22
8)
24
04
(11
9)
37
17
(18
4)
22
22
(11
0)
15
11
(75
)2
01
88
(10
0)
20
54
7(1
00
)
Su
rve
yM
od
e
Tel
eph
on
eIn
terv
iew
18
06
(52
8)
97
1(4
20
)2
05
1(4
46
)8
46
(35
2)
22
29
(60
0)
15
66
(70
5)
96
0(6
35
)1
04
29
(51
7)
0
Wri
tten
-lo
ng
77
1(2
25
)5
75
(24
9)
97
3(2
11
)7
23
(30
1)
67
6(1
82
)3
46
(15
6)
22
4(1
48
)4
28
8(2
12
)0
Wri
tten
-sh
ort
43
9(1
28
)2
59
(11
2)
46
9(1
02
)5
16
(21
5)
44
8(1
20
)1
08
(49
)1
54
(10
2)
23
93
(11
8)
0
Web
40
4(1
18
)5
07
(21
9)
11
09
(24
1)
31
9(1
33
)3
64
(98
)2
02
(91
)1
73
(11
4)
30
78
(15
2)
0
Fem
ale
19
39
(56
7)
10
19
(44
1)
20
35
(44
2)
12
46
(51
8)
18
86
(50
7)
11
22
(50
5)
59
3(3
93
)9
84
0(4
87
)9
53
7(4
64
)
Ag
e(y
ea
rs)
30
ndash4
42
74
(80
)1
37
(59
)3
82
(83
)1
68
(70
)5
64
(15
2)
25
3(1
14
)1
89
(12
5)
19
67
(97
)2
58
0(1
26
)
45
ndash5
91
34
7(3
94
)9
34
(40
4)
17
94
(39
0)
11
06
(46
0)
16
39
(44
1)
94
3(4
24
)6
95
(46
0)
84
58
(41
9)
87
55
(42
6)
60thorn
17
99
(52
6)
12
41
(53
7)
24
26
(52
7)
11
30
(47
0)
15
14
(40
7)
10
26
(46
2)
62
7(4
15
)9
76
3(4
84
)9
21
2(4
48
)
Lan
gu
age
pre
fere
nce
En
gli
shsp
eak
ing
32
21
(94
2)
13
76
(59
5)
43
94
(95
5)
17
49
(72
8)
22
17
(59
6)
16
91
(76
1)
11
64
(77
0)
15
81
2(7
83
)1
72
12
(83
8)
Mea
nA
1C5
7
16
55
(54
0)
10
87
(50
1)
19
15
(46
8)
13
52
(60
4)
19
57
(58
6)
10
98
(56
2)
77
9(5
79
)9
84
3(5
41
)8
34
3(5
13
)
Mea
nL
DL5
10
01
25
0(4
17
)6
26
(28
9)
13
91
(34
6)
63
4(2
86
)1
20
2(3
71
)7
36
(38
1)
48
2(3
63
)6
32
1(3
53
)6
90
9(4
24
)
Un
con
tro
lled
blo
od
pre
ssu
re(S4
13
0o
rD4
80
)1
96
6(6
14
)9
13
(42
8)
22
64
(54
7)
10
41
(46
5)
17
46
(51
2)
10
89
(54
2)
68
1(4
95
)9
70
0(5
24
)9
58
1(5
43
)
Mis
sed
ap
po
intm
ent
rate
(per
year)
01
7plusmn
(01
8)
01
0plusmn
(01
4)
01
3plusmn
(01
6)
01
3plusmn
(01
7)
01
8plusmn
(01
9)
01
7plusmn
(01
9)
01
8plusmn
(01
9)
01
5plusmn
(01
7)
01
8plusmn
(02
2)
Po
or
med
icati
on
ad
her
ence
b9
55
(37
1)
46
1(2
51
)9
61
(28
2)
59
2(2
93
)1
24
7(4
28
)6
36
(36
4)
42
4(3
51
)5
27
6(3
36
)5
10
8(3
90
)
Sm
ok
ing
(cu
rren
t)3
48
(10
2)
93
(40
)4
10
(90
)1
47
(61
)2
45
(66
)2
00
(91
)1
05
(70
)1
54
8(7
7)
18
29
(90
)
Pra
ctic
esS
MB
G1
57
3(4
60
)1
00
4(4
34
)2
24
6(4
88
)1
11
5(4
64
)1
62
2(4
36
)1
02
5(4
61
)6
23
(41
2)
92
08
(45
6)
65
32
(31
8)
Ou
tpati
ent
risk
sco
rec
24
(21
)1
8(1
1)
21
(17
)1
8(1
0)
19
(14
)2
2(1
6)
19
(14
)2
0(1
6)
18
(15
)
Inp
ati
ent
risk
sco
rec
26
(42
)1
5(2
0)
22
(33
)1
4(2
1)
18
(30
)2
2(3
9)
17
(29
)2
0(3
2)
18
(34
)
Lin
gu
isti
call
yis
ola
ted
nei
gh
bo
urh
oo
dd
10
5(3
1)
11
5(5
0)
55
(12
)1
14
(48
)2
81
(76
)1
13
(51
)6
2(4
1)
84
5(4
2)
10
35
(51
)
Eco
no
mic
all
yd
epri
ved
nei
gh
bo
urh
oo
de
82
7(2
43
)1
24
(54
)3
46
(76
)1
48
(62
)6
68
(18
2)
32
3(1
47
)1
67
(11
2)
26
03
(13
0)
27
24
(13
4)
Wo
rkin
gcl
ass
nei
gh
bo
urh
oo
df
14
40
(42
2)
40
1(1
74
)1
16
8(2
58
)8
84
(37
0)
17
49
(47
6)
86
8(3
94
)4
64
(31
0)
69
74
(34
9)
75
66
(37
2)
aIn
clu
des
Paci
fic
Isla
nd
ers
Nati
veA
mer
ican
sS
ou
thA
sian
o
ther
un
spec
ifie
dan
du
nk
no
wn
(did
no
tan
swer
)b5
20
co
nti
nu
ou
sm
edic
ati
on
gap
s6
3
c Th
eyare
wei
gh
ted
by
pati
entrsquo
sh
ealt
hca
reu
tili
zati
on
an
dse
veri
tyo
fd
isea
se6
4
d5
25
o
fh
ou
seh
old
sin
cen
sus
blo
ckgro
up
are
lin
gu
isti
call
yis
ola
ted
e5
20
o
fh
ou
seh
old
sin
cen
sus
blo
ckgro
up
are
livi
ng
bel
ow
po
vert
yli
ne
f 56
6
of
per
son
sin
cen
sus
blo
ckgro
up
are
emp
loye
din
wo
rkin
gcl
ass
occ
up
ati
on
42 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
scores and not living in an economically deprived orworking class neighbourhood
Caucasian respondents were more likely to completethe survey online to have had lower mean A1Cpracticing SMBG and not living in linguisticallyisolated neighbourhood
Filipino respondents were more likely to participatevia written survey to have been older than 45 yearsof age to have had lower levels of LDL and lowerco-morbidity scores
Latino respondents were more likely to participatevia CATI to have been younger than 45 years of ageto not have English as their preferred language tohave had higher mean A1C higher rates of missedappointments poorer medication adherence andliving in a linguistically isolated or working classneighbourhood
What outcomes will be measuredduring follow-upThis cohort will be the basis for longitudinal evalua-tions of a wide range of clinical outcomes associatedwith diabetes and are powered to evaluate the ethnicand educational disparities in diabetes-related com-plication rates (eg myocardial infarction strokeheart failure kidney failure and amputation) andmortality after a 25-year follow-up Clinical andadministrative follow-up data will be captured fromthese same sources as baseline plus deaths from thestate mortality files Data from all sources combinedmay offer insights into how differences at the patient-level (eg differences in behaviours type of therapyintermediate health status and psychosocial factorsand competing demands) provider-level (eg trust inprovider racial concordance and patient-providercommunication) and system-level (eg differentialimpact of cost-sharing) may lead to social disparitiesin diabetes health end-stage diabetes complicationsand mortality
What is the anticipated attritionOne of the strengths of this diabetes registry is thevery low turnover rate (5 discontinue membershipeach year) affording minimal loss to follow-up Onaverage the duration of Kaiser membership amongour diabetes cohort members is 89 years (SDfrac14 66)
What are the strengths andweaknessesA significant weakness is that written and websurveys were only offered in English Howevertelephone interviewers made the initial efforts tocontact subjects who had the opportunity to complete
interviews in Spanish Mandarin Cantonese orTagalog Offering the survey by oral interview inmultiple languages was intended to mitigate thelanguage andor literacy barriers but we still observedlower participation rates by those with less educationand modest response differences by race On a morefavourable note 84 of non-respondents were identi-fied by the health plan as having English as theirpreferred language suggesting that language alonewas not a major barrier to participation
The response rate was as expected lower than ourprevious Diabetes Research Questionnaire (1994ndash97)A recent survey conducted in the Translating ResearchInto Action for Diabetes (TRIAD) study amongmanaged care populations across the United Statesincluding Kaiser43 had a response rate of 50 TheBehavioural Risk Factor Surveillance System alsoexperienced a decline in median response ratesfrom 632 in 1996 to 535 in 200168 Privacyconcerns and competition from telemarketing haveprobably played a role in declining response ratesHowever if generally lower response rates have beenobserved in minority populations the response ratefor this survey may be considered favourable giventhat 77 of our sample were racial or ethnic minorityhealth plan members
One strength of this study is the relatively uniformaccess to care provided by membership in thisintegrated health plan Thus this study largely avoidsdisparities in access and quality which are potentsources of confounding bias in population-basedstudies of social health disparities Another strength isthe quality and richness of the clinical follow-up dataAdministrative and clinical data are available directlyfrom the health plan and are considered to be reliableand complete Each telephone interview was conductedby professional interviewers employed trained andsupervised by the Public Health Institute SurveyResearch Group in Sacramento CA which has itsown internal quality controls Written surveys (long-and short-forms) were coded and edited using anextensive data dictionary and detailed coding rules tostandardize and clarify responses before being sentto the Data Entry department at the KaiserDivision of Research which conducted double dataentry The survey data quality is limited by beingself-reported
The diversity size and wealth of data included inthe DISTANCE cohort make it suitable for theprospective study of a wide range of social disparitiesin the processes and outcomes of diabetes health careInformation about childhood socioeconomic positionand educational attainment will facilitate a life courseapproach6970
Analysis of response biasResponse bias is a concern in any survey Surveys aresubject to response bias if participation is associatedwith the risk factors (exposures) andor diseases
COHORT PROFILE DISTANCE 43
(outcomes) under investigation In this study popula-tion we have substantial data on non-respondentsincluding Kaiser administrative data co-morbidityand census data on neighbourhood characteristicsallowing us to compare hypothesized relationships inrespondents vs non-respondents and adjust for thispotential response bias71 While participation wassomewhat lower among minorities and those withless education overall respondents and non-respon-dents were quite similar
We had pre-baseline reported race data for 29 319(72) members of the DISTANCE cohort andobserved differences in response rates by raceAfrican Americans 52 Asians 46 Caucasians63 Latinos 53 unknown 45 (P lt 00001) Wealso had educational attainment data for 4524subjects (22) who had completed the previoussurvey in 1994ndash97 and observed that subjects with ahigh school education or less participated at a lowerrate than those with more than a high schooleducation (54 vs 60) (P lt 00001)
A more complete view of response bias was obtainedfrom health plan administrative data and census datawhich was available for all members of the cohortWe found few notable differences between respon-dents and non-respondents in demographic clinicalbehavioural or neighbourhood characteristics Com-pared with respondents non-respondents had highermean LDL levels (993 vs 943 mgdl) and were lesslikely to practice SMBG (32 vs 46) However aportion of these unadjusted differences are likely dueto age differences in respondents vs non-respondents
We conducted an assessment of response bias bycomparing differences in A1C across ethnic groupsseparately among respondents and non-respondentsWe specified a logistic regression model of poorglycemic control which regressed the outcome A1C47 on ethnicity age and sex and an interactionterm (ethnicity survey response) Although respon-dents consistently had higher A1C than non-respon-dents ethnic differences in the relationship betweenethnicity and A1C did not differ between respondentsand non-respondents (Pfrac14 055) We conducted asimilar analysis with the subjects for whom we hadeducational attainment data from a previous surveyand again found no substantive response bias(Pfrac14 028) While characteristics differ betweenrespondents and non-respondents (response bias)the associations with outcomes of interest (slopes)are much less vulnerable to such bias7273
Statistical methodsAnalyses of DISTANCE data will focus on associationsrather than descriptive characterization We willemploy a modelling approach using inverse weightingestimation of marginal structural models72 to inves-tigate causal relationships in this study Thesemodels will include weights to adjust for the non-proportionate sampling fractions (eg over-sampling
of the minority ethnic groups) and response bias(eg giving greater weight to respondents whohave characteristics more similar to the non-respondents) Thus the proposed estimating approachwill adjust for confounding and selection biassimultaneously73ndash77
How can I collaborate Where canI find out moreWe are not free to release participantsrsquo personal dataunder our promise to them regarding confidentialityHowever the DISTANCE steering committee is inter-ested in collaborations with external researchersDISTANCE investigators are particularly interested incomparing social disparities observed in this insuredpopulation with uniform access to care to disparitiesobserved in population-based samples where qualityand access to care vary widely by social strataRequests for collaboration must be submitted to thedirector of the central coordinating centre (corre-sponding author) Before collaborations can beinitiated proposals require review and approval bythe DISTANCE Publications and Presentations (PampP)Committee This committee was formed to (i) Ensureaccurate uniform timely and high quality reportingof DISTANCE findings (ii) Preserve the scientificintegrity of the study and (iii) Safeguard the rightsand confidentiality of participants
Supplementary DataSupplementary data are available at IJE online
AcknowledgementsFunds were provided by National Institute ofDiabetes Digestive and Kidney Diseases [R01DK65664] and National Institute of Child Healthand Human Development [R01 HD046113] Dr DS isalso supported by a National Institutes of HealthMentored Clinical Scientist Award [K-23 RR16539]The Public Health InstituteSurvey Research Groupconducted our telephone interviews Gerard andAssociates provided certified translations of all patientmaterials and interview scripts KP Corporation (notaffiliated with Kaiser Permanente) provided printingand mailing services for the written surveys andother communications We thank Marcia Ewing andCarol Rabello for their diligent work on surveyoperations Most of all we thank all the studyparticipants
Conflict of interest None declared
References1 Healthy People 2010 U S Department of Health and
Human Services 2005 Available at httpwwwhealthypeoplegov(Accessed August 2 2007)
44 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
2 Smedley BD Stith AY Nelson AR Unequal TreatmentConfronting Racial and Ethnic Disparities in Health CareWashington DC Institute of Medicine NationalAcademy Press 2002
3 Karter AJ Newman B Rowell S et al Large-scalecollection of family history data and recruitment ofinformative families for genetic analysis J Reg Manag1998257ndash12
4 Karter AJ Ferrara A Darbinian J Ackerson LM Selby JVSelf-monitoring of blood glucose language and financialbarriers in a managed care population with diabetesDiabetes Care 200023477ndash83
5 Karter AJ Moffet HH Liu J et al Achieving goodglycemic control initiation of new antihyperglycemictherapies in patients with type 2 diabetes from theKaiser Permanente Northern California Diabetes RegistryAm J Manag Care 200511262ndash70
6 Karter AJ Ackerson LM Darbinian JA et al Self-monitoring of blood glucose levels and glycemic controlthe Northern California Kaiser Permanente Diabetesregistry Am J Med 20011111ndash9
7 Karter AJ Ferrara A Liu JY Moffet HH Ackerson LMSelby JV Ethnic disparities in diabetic complications inan insured population JAMA 20022872519ndash27
8 Karter AJ Parker MM Moffet HH et al Missedappointments and poor glycemic control an opportunityto identify high-risk diabetic patients Med Care200442110ndash15
9 Selby JV Karter AJ Ackerson LM Ferrara A Liu JDeveloping a prediction rule from automated clinicaldatabases to identify high-risk patients in a largepopulation with diabetes Diabetes Care 2001241547ndash55
10 Martin TL Selby JV Zhang D Physician and patientprevention practices in NIDDM in a large urbanmanaged-care organization Diabetes Care 1995181124ndash32
11 Chaturvedi N Jarrett J Shipley MJ Fuller JHSocioeconomic gradient in morbidity and mortality inpeople with diabetes cohort study findings from theWhitehall Study and the WHO Multinational Study ofVascular Disease in Diabetes BMJ 1998316100ndash5
12 Booth GL Hux JE Relationship between avoidablehospitalizations for diabetes mellitus and income levelArch Intern Med 2003163101ndash6
13 Kleinbaum DG Kupper LL Morgenstern H EpidemiologicResearch Principles and Quantitative Methods New York VanNorstrand Reinhold 1982
14 Dillman DA Mail and Internet Surveys The Tailored DesignMethod 2nd edn New York John Wiley 2000
15 Rudd RE Comings JP Hyde JN Leave no one behindimproving health and risk communication through att-ention to literacy J Health Commun 20038 (Suppl 1)104ndash15
16 Singh-Manoux A Adler NE Marmot MG Subjectivesocial status its determinants and its association withmeasures of ill-health in the Whitehall II study Soc SciMed 2003561321ndash33
17 John D Catherine T MacArthur Research Network onSocioeconomic Status and Health The MacArthur Scale ofSubjective Social Status San Francisco University ofCalifornia 1999 Available at httpwwwmacsesucsfeduResearchPsychosocialnotebooksubjectivehtml(Accessed July 12 2007)
18 Ross CE Wu CL Education age and the cumulativeadvantage in health J Health Soc Behav 199637104ndash20
19 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
20 White S Dillow S Key Concepts and Features of the 2003National Assessment of Adult Literacy (NAALS) (NCES2006ndash471) Washington DC Department of EducationNational Center for Education Statistics 2005
21 Botman SL Moore TF Moriarity CL Parsons VL Designand Estimation for the National Health Interview Survey(NHIS) 1995ndash2004 National Center for Health StatisticsVital Health Stat 2 20001301ndash31
22 Friedman GD Cutter GR Donahue RP et al CARDIAstudy design recruitment and some characteristics of theexamined subjects J Clin Epidemiol 1988411105ndash16
23 Ross CE Wu CL The links between education and healthAm Sociol Rev 199560719ndash45
24 Robert S House JS SES differentials in health by ageand alternative indicators of SES J Aging Health19968359ndash88
25 Toobert DJ Hampson SE Glasgow RE The summary ofdiabetes self-care activities measure results from 7studies and a revised scale Diabetes Care 200023943ndash50
26 Craig CL Marshall AL Sjostrom M et al Internationalphysical activity questionnaire 12-country reliability andvalidity Med Sci Sports Exerc 2003351381ndash95
27 Gordon AJ Maisto SA McNeil M et al Three questionscan detect hazardous drinkers J Fam Pract 200150313ndash20
28 National Health and Nutrition Examination Survey(NHANES) Questionnaire (2001-2002) Centers for DiseaseControl and Prevention (CDC) National Center for HealthStatistics (NCHS) Hyattsville MD US Department ofHealth and Human Services Centers for Disease Controland Prevention 2002
29 Behavioral Risk Factor Surveillance System (BRFSS)Centers for Disease Control and Prevention (CDC) NationalCenter for Health Statistics (NCHS) Hyattsville MD USDepartment of Health and Human Services Centers forDisease Control and Prevention 2003
30 Chien-Wen T Phillips RL Green LA Fryer GE Dovey SMWhat physicians need to know about seniors and limitedprescription benefits and why Am Fam Physician200266212
31 Morisky DE Green LW Levine DM Concurrent andpredictive validity of a self-reported measure of medica-tion adherence Med Care 19862467ndash74
32 Barringer TA Kirk JK Santaniello AC Foley KLMichielutte R Effect of a multivitamin and mineralsupplement on infection and quality of life A random-ized double-blind placebo-controlled trial Ann Intern Med2003138365ndash71
33 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
34 Peyrot M Rubin RR Structure and correlates of diabetes-specific locus of control Diabetes Care 199417994ndash1001
35 Wagenknecht LE Mayer EJ Rewers M et al The insulinresistance atherosclerosis study (IRAS) objectivesdesign and recruitment results Ann Epidemiol19955464ndash72
COHORT PROFILE DISTANCE 45
36 King DE Mainous AG III Pearson WS C-reactiveprotein diabetes and attendance at religious servicesDiabetes Care 2002251172ndash76
37 Berkman LF Syme SL Social networks host resistanceand mortality a nine-year follow-up study of AlamedaCounty residents Am J Epidemiol 1979109186ndash204
38 Ware JE Kosinski M Keller SD SF-12 How to Score theSF-12 Physical and Mental Health Summary Scales 2nd ednBoston The Health Institute New England MedicalCenter 1995
39 Ware JE Kosinksi M Dewey JE Gandek B How to Scoreand Interpret Single-Item Health Status Measures A Manual forUsers of the SF-8 Health Survey Lincoln RI QualityMetricIncorporated 2001
40 Cohen S Kamarck T Mermelstein R A global measure ofperceived stress J Health Soc Behav 198324385ndash96
41 Kroenke K Spitzer RL Williams JB The PHQ-9 validityof a brief depression severity measure J Gen Intern Med200116606ndash13
42 Spitzer RL Kroenke K Williams JB Validation and utilityof a self-report version of PRIME-MD the PHQ primarycare study Primary care evaluation of mental disordersPatient Health Questionnaire JAMA 19992821737ndash44
43 The TRIAD Study Group The translating research intoaction for diabetes (TRIAD) study a multicenter study ofdiabetes in managed care Diabetes Care 200225386ndash89
44 Ayas NT White DP Al Delaimy WK et al A prospectivestudy of self-reported sleep duration and incidentdiabetes in women Diabetes Care 200326380ndash84
45 Sickel AE Moore PJ Adler NE Williams DR Jackson JSThe differential effects of sleep quality and quantity onthe relationship between SES and health Ann NY Acad Sci1999896431ndash34
46 Moore PJ Adler NE Williams DR Jackson JSSocioeconomic status and health the role of sleepPsychosom Med 200264337ndash44
47 Benet-Martinez V John OP Los Cinco Grandes acrosscultures and ethnic groups multitrait multimethodanalyses of the Big Five in Spanish and English J PersSoc Psychol 199875729ndash50
48 Srivastava S John OP Gosling SD Potter J Developmentof personality in early and middle adulthood set likeplaster or persistent change J Pers Soc Psychol2003841041ndash53
49 Gosling SD Rentfrow PJ Swann WB A very briefmeasure of the Big-Five personality domains J Res Pers200337504ndash28
50 Schillinger D Grumbach K Piette J et al Association ofhealth literacy with diabetes outcomes JAMA2002288475ndash82
51 Schillinger D Barton LR Karter AJ Wang F Adler NDoes literacy mediate the relationship between educationand health outcomes A study of a low-income popula-tion with diabetes Public Health Rep 2006121245ndash54
52 Chew LD Bradley KA Boyko EJ Brief questions toidentify patients with inadequate health literacy Fam Med200436588ndash94
53 The American Association of Clinical EndocrinologistsMedical Guidelines for the Management of DiabetesMellitus The AACE System of Intensive DiabetesSelf-Management ndash 2002 Update Endocr Pract 20028(Suppl 1)40ndash82
54 Fitzgerald JT Funnell MM Hess GE et al The reliabilityand validity of a brief diabetes knowledge test DiabetesCare 199821706ndash10
55 Piette JD Heisler M Wagner TH Problems paying out-of-pocket medication costs among older adults withdiabetes Diabetes Care 200427384ndash91
56 Epidemiology of Diabetes Interventions andComplications (EDIC) Design implementation andpreliminary results of a long-term follow-up of thediabetes control and complications trial cohort DiabetesCare 19992299ndash111
57 Krieger N Sidney S Coakley E Racial discrimination andskin color in the CARDIA study implications for publichealth research Coronary artery risk development inyoung adults Am J Public Health 1998881308ndash13
58 Krieger N Smith K Naishadham D Hartman CBarbeau EM Experiences of discrimination validity andreliability of a self-report measure for population healthresearch on racism and health Soc Sci Med 2005611576ndash96
59 CAHPS 20 Questionnaires Crofton C AHCPR Pub No97-R079 Washington DC Agency for Health Care Policyand Research 1998
60 Hays RD Shaul JA Williams VS et al Psychometricproperties of the CAHPS 10 survey measures Consumerassessment of health plans study Med Care 199937MS22ndash31
61 Anderson LA Dedrick RF Development of the trust inphysician scale a measure to assess interpersonal trust inpatient-physician relationships Psychol Rep 1990671091ndash100
62 Thom DH Ribisl KM Stewart AL Luke DA Furthervalidation and reliability testing of the Trust in PhysicianScale The Stanford trust study physicians Med Care199937510ndash17
63 Steiner JF Koepsell TD Fihn SD Inui TS A generalmethod of compliance assessment using centralizedpharmacy records Description and validation Med Care198826814ndash23
64 Clark DO Von Korff M Saunders K Baluch WMSimon GE A chronic disease score with empiricallyderived weights Med Care 199533783ndash95
65 Kaiser Permanente Business Intelligence Report 2006Research and Markets 2006 Available at httpwwwresearchandmarketscomreportinfoaspreport_idfrac14451423(Accessed July 9 2007)
66 Krieger N Overcoming the absence of socioeconomic datain medical records validation and application of acensus-based methodology Am J Public Health 199282703ndash10
67 Hiatt RA Friedman GD Characteristics of patientsreferred for treatment of end-stage renal diseasein a defined population Am J Public Health 198272829ndash33
68 Porter S Jackson K Trosclaire A Pederson LL Office onsmoking and Health NCfCDPaHPC Prevalence of currentcigarette smoking among adults and changes in pre-valence of current and some day smoking ndash UnitedStates 1996ndash2001 MMWR 200352303ndash7
69 Smith GD Hart C Blane D Gillis C Hawthorne VLifetime socioeconomic position and mortality prospec-tive observational study BMJ 1997314547ndash52
46 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
70 Lynch JW Kaplan GA Shema SJ Cumulative impact ofsustained economic hardship on physical cognitivepsychological and social functioning N Engl J Med19973371889ndash95
71 Melton LJ3 Dyck PJ Karnes JL OrsquoBrien PC Service FJNon-response bias in studies of diabetic complicationsthe Rochester Diabetic Neuropathy Study J Clin Epidemiol199346341ndash48
72 Robins JM Association causation and marginal struc-tural models Synthese 1999121151ndash79
73 Hernan MA Hernandez-Diaz S Robins JM A structuralapproach to selection bias Epidemiology 200415615ndash25
74 Bodnar LM Davidian M Siega-Riz AM Tsiatis AAMarginal structural models for analyzing causal effects
of time-dependent treatments an application in perinatalepidemiology Am J Epidemiol 2004159926ndash34
75 Bryan JF Yu Z van der Laan M Analysis of longitudinalmarginal structural models U C Berkeley Division ofBiolstatistics Working Paper Series Paper 1201ndash29 2002
76 Neugebauer R van der Laan M Why prefer double robustestimators in causal inference J Stat Plan Inference2005129405ndash26
77 Wang Y Petersen ML Bangsberg D van der Laan MDiagnosing bias in the inverse probability of treatmentweighted estimator resulting from violation ofexperimental treatment assignment U C BerkeleyDivision of Biolstatistics Working Paper Series Paper 2111ndash25 2006
COHORT PROFILE DISTANCE 47
scores and not living in an economically deprived orworking class neighbourhood
Caucasian respondents were more likely to completethe survey online to have had lower mean A1Cpracticing SMBG and not living in linguisticallyisolated neighbourhood
Filipino respondents were more likely to participatevia written survey to have been older than 45 yearsof age to have had lower levels of LDL and lowerco-morbidity scores
Latino respondents were more likely to participatevia CATI to have been younger than 45 years of ageto not have English as their preferred language tohave had higher mean A1C higher rates of missedappointments poorer medication adherence andliving in a linguistically isolated or working classneighbourhood
What outcomes will be measuredduring follow-upThis cohort will be the basis for longitudinal evalua-tions of a wide range of clinical outcomes associatedwith diabetes and are powered to evaluate the ethnicand educational disparities in diabetes-related com-plication rates (eg myocardial infarction strokeheart failure kidney failure and amputation) andmortality after a 25-year follow-up Clinical andadministrative follow-up data will be captured fromthese same sources as baseline plus deaths from thestate mortality files Data from all sources combinedmay offer insights into how differences at the patient-level (eg differences in behaviours type of therapyintermediate health status and psychosocial factorsand competing demands) provider-level (eg trust inprovider racial concordance and patient-providercommunication) and system-level (eg differentialimpact of cost-sharing) may lead to social disparitiesin diabetes health end-stage diabetes complicationsand mortality
What is the anticipated attritionOne of the strengths of this diabetes registry is thevery low turnover rate (5 discontinue membershipeach year) affording minimal loss to follow-up Onaverage the duration of Kaiser membership amongour diabetes cohort members is 89 years (SDfrac14 66)
What are the strengths andweaknessesA significant weakness is that written and websurveys were only offered in English Howevertelephone interviewers made the initial efforts tocontact subjects who had the opportunity to complete
interviews in Spanish Mandarin Cantonese orTagalog Offering the survey by oral interview inmultiple languages was intended to mitigate thelanguage andor literacy barriers but we still observedlower participation rates by those with less educationand modest response differences by race On a morefavourable note 84 of non-respondents were identi-fied by the health plan as having English as theirpreferred language suggesting that language alonewas not a major barrier to participation
The response rate was as expected lower than ourprevious Diabetes Research Questionnaire (1994ndash97)A recent survey conducted in the Translating ResearchInto Action for Diabetes (TRIAD) study amongmanaged care populations across the United Statesincluding Kaiser43 had a response rate of 50 TheBehavioural Risk Factor Surveillance System alsoexperienced a decline in median response ratesfrom 632 in 1996 to 535 in 200168 Privacyconcerns and competition from telemarketing haveprobably played a role in declining response ratesHowever if generally lower response rates have beenobserved in minority populations the response ratefor this survey may be considered favourable giventhat 77 of our sample were racial or ethnic minorityhealth plan members
One strength of this study is the relatively uniformaccess to care provided by membership in thisintegrated health plan Thus this study largely avoidsdisparities in access and quality which are potentsources of confounding bias in population-basedstudies of social health disparities Another strength isthe quality and richness of the clinical follow-up dataAdministrative and clinical data are available directlyfrom the health plan and are considered to be reliableand complete Each telephone interview was conductedby professional interviewers employed trained andsupervised by the Public Health Institute SurveyResearch Group in Sacramento CA which has itsown internal quality controls Written surveys (long-and short-forms) were coded and edited using anextensive data dictionary and detailed coding rules tostandardize and clarify responses before being sentto the Data Entry department at the KaiserDivision of Research which conducted double dataentry The survey data quality is limited by beingself-reported
The diversity size and wealth of data included inthe DISTANCE cohort make it suitable for theprospective study of a wide range of social disparitiesin the processes and outcomes of diabetes health careInformation about childhood socioeconomic positionand educational attainment will facilitate a life courseapproach6970
Analysis of response biasResponse bias is a concern in any survey Surveys aresubject to response bias if participation is associatedwith the risk factors (exposures) andor diseases
COHORT PROFILE DISTANCE 43
(outcomes) under investigation In this study popula-tion we have substantial data on non-respondentsincluding Kaiser administrative data co-morbidityand census data on neighbourhood characteristicsallowing us to compare hypothesized relationships inrespondents vs non-respondents and adjust for thispotential response bias71 While participation wassomewhat lower among minorities and those withless education overall respondents and non-respon-dents were quite similar
We had pre-baseline reported race data for 29 319(72) members of the DISTANCE cohort andobserved differences in response rates by raceAfrican Americans 52 Asians 46 Caucasians63 Latinos 53 unknown 45 (P lt 00001) Wealso had educational attainment data for 4524subjects (22) who had completed the previoussurvey in 1994ndash97 and observed that subjects with ahigh school education or less participated at a lowerrate than those with more than a high schooleducation (54 vs 60) (P lt 00001)
A more complete view of response bias was obtainedfrom health plan administrative data and census datawhich was available for all members of the cohortWe found few notable differences between respon-dents and non-respondents in demographic clinicalbehavioural or neighbourhood characteristics Com-pared with respondents non-respondents had highermean LDL levels (993 vs 943 mgdl) and were lesslikely to practice SMBG (32 vs 46) However aportion of these unadjusted differences are likely dueto age differences in respondents vs non-respondents
We conducted an assessment of response bias bycomparing differences in A1C across ethnic groupsseparately among respondents and non-respondentsWe specified a logistic regression model of poorglycemic control which regressed the outcome A1C47 on ethnicity age and sex and an interactionterm (ethnicity survey response) Although respon-dents consistently had higher A1C than non-respon-dents ethnic differences in the relationship betweenethnicity and A1C did not differ between respondentsand non-respondents (Pfrac14 055) We conducted asimilar analysis with the subjects for whom we hadeducational attainment data from a previous surveyand again found no substantive response bias(Pfrac14 028) While characteristics differ betweenrespondents and non-respondents (response bias)the associations with outcomes of interest (slopes)are much less vulnerable to such bias7273
Statistical methodsAnalyses of DISTANCE data will focus on associationsrather than descriptive characterization We willemploy a modelling approach using inverse weightingestimation of marginal structural models72 to inves-tigate causal relationships in this study Thesemodels will include weights to adjust for the non-proportionate sampling fractions (eg over-sampling
of the minority ethnic groups) and response bias(eg giving greater weight to respondents whohave characteristics more similar to the non-respondents) Thus the proposed estimating approachwill adjust for confounding and selection biassimultaneously73ndash77
How can I collaborate Where canI find out moreWe are not free to release participantsrsquo personal dataunder our promise to them regarding confidentialityHowever the DISTANCE steering committee is inter-ested in collaborations with external researchersDISTANCE investigators are particularly interested incomparing social disparities observed in this insuredpopulation with uniform access to care to disparitiesobserved in population-based samples where qualityand access to care vary widely by social strataRequests for collaboration must be submitted to thedirector of the central coordinating centre (corre-sponding author) Before collaborations can beinitiated proposals require review and approval bythe DISTANCE Publications and Presentations (PampP)Committee This committee was formed to (i) Ensureaccurate uniform timely and high quality reportingof DISTANCE findings (ii) Preserve the scientificintegrity of the study and (iii) Safeguard the rightsand confidentiality of participants
Supplementary DataSupplementary data are available at IJE online
AcknowledgementsFunds were provided by National Institute ofDiabetes Digestive and Kidney Diseases [R01DK65664] and National Institute of Child Healthand Human Development [R01 HD046113] Dr DS isalso supported by a National Institutes of HealthMentored Clinical Scientist Award [K-23 RR16539]The Public Health InstituteSurvey Research Groupconducted our telephone interviews Gerard andAssociates provided certified translations of all patientmaterials and interview scripts KP Corporation (notaffiliated with Kaiser Permanente) provided printingand mailing services for the written surveys andother communications We thank Marcia Ewing andCarol Rabello for their diligent work on surveyoperations Most of all we thank all the studyparticipants
Conflict of interest None declared
References1 Healthy People 2010 U S Department of Health and
Human Services 2005 Available at httpwwwhealthypeoplegov(Accessed August 2 2007)
44 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
2 Smedley BD Stith AY Nelson AR Unequal TreatmentConfronting Racial and Ethnic Disparities in Health CareWashington DC Institute of Medicine NationalAcademy Press 2002
3 Karter AJ Newman B Rowell S et al Large-scalecollection of family history data and recruitment ofinformative families for genetic analysis J Reg Manag1998257ndash12
4 Karter AJ Ferrara A Darbinian J Ackerson LM Selby JVSelf-monitoring of blood glucose language and financialbarriers in a managed care population with diabetesDiabetes Care 200023477ndash83
5 Karter AJ Moffet HH Liu J et al Achieving goodglycemic control initiation of new antihyperglycemictherapies in patients with type 2 diabetes from theKaiser Permanente Northern California Diabetes RegistryAm J Manag Care 200511262ndash70
6 Karter AJ Ackerson LM Darbinian JA et al Self-monitoring of blood glucose levels and glycemic controlthe Northern California Kaiser Permanente Diabetesregistry Am J Med 20011111ndash9
7 Karter AJ Ferrara A Liu JY Moffet HH Ackerson LMSelby JV Ethnic disparities in diabetic complications inan insured population JAMA 20022872519ndash27
8 Karter AJ Parker MM Moffet HH et al Missedappointments and poor glycemic control an opportunityto identify high-risk diabetic patients Med Care200442110ndash15
9 Selby JV Karter AJ Ackerson LM Ferrara A Liu JDeveloping a prediction rule from automated clinicaldatabases to identify high-risk patients in a largepopulation with diabetes Diabetes Care 2001241547ndash55
10 Martin TL Selby JV Zhang D Physician and patientprevention practices in NIDDM in a large urbanmanaged-care organization Diabetes Care 1995181124ndash32
11 Chaturvedi N Jarrett J Shipley MJ Fuller JHSocioeconomic gradient in morbidity and mortality inpeople with diabetes cohort study findings from theWhitehall Study and the WHO Multinational Study ofVascular Disease in Diabetes BMJ 1998316100ndash5
12 Booth GL Hux JE Relationship between avoidablehospitalizations for diabetes mellitus and income levelArch Intern Med 2003163101ndash6
13 Kleinbaum DG Kupper LL Morgenstern H EpidemiologicResearch Principles and Quantitative Methods New York VanNorstrand Reinhold 1982
14 Dillman DA Mail and Internet Surveys The Tailored DesignMethod 2nd edn New York John Wiley 2000
15 Rudd RE Comings JP Hyde JN Leave no one behindimproving health and risk communication through att-ention to literacy J Health Commun 20038 (Suppl 1)104ndash15
16 Singh-Manoux A Adler NE Marmot MG Subjectivesocial status its determinants and its association withmeasures of ill-health in the Whitehall II study Soc SciMed 2003561321ndash33
17 John D Catherine T MacArthur Research Network onSocioeconomic Status and Health The MacArthur Scale ofSubjective Social Status San Francisco University ofCalifornia 1999 Available at httpwwwmacsesucsfeduResearchPsychosocialnotebooksubjectivehtml(Accessed July 12 2007)
18 Ross CE Wu CL Education age and the cumulativeadvantage in health J Health Soc Behav 199637104ndash20
19 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
20 White S Dillow S Key Concepts and Features of the 2003National Assessment of Adult Literacy (NAALS) (NCES2006ndash471) Washington DC Department of EducationNational Center for Education Statistics 2005
21 Botman SL Moore TF Moriarity CL Parsons VL Designand Estimation for the National Health Interview Survey(NHIS) 1995ndash2004 National Center for Health StatisticsVital Health Stat 2 20001301ndash31
22 Friedman GD Cutter GR Donahue RP et al CARDIAstudy design recruitment and some characteristics of theexamined subjects J Clin Epidemiol 1988411105ndash16
23 Ross CE Wu CL The links between education and healthAm Sociol Rev 199560719ndash45
24 Robert S House JS SES differentials in health by ageand alternative indicators of SES J Aging Health19968359ndash88
25 Toobert DJ Hampson SE Glasgow RE The summary ofdiabetes self-care activities measure results from 7studies and a revised scale Diabetes Care 200023943ndash50
26 Craig CL Marshall AL Sjostrom M et al Internationalphysical activity questionnaire 12-country reliability andvalidity Med Sci Sports Exerc 2003351381ndash95
27 Gordon AJ Maisto SA McNeil M et al Three questionscan detect hazardous drinkers J Fam Pract 200150313ndash20
28 National Health and Nutrition Examination Survey(NHANES) Questionnaire (2001-2002) Centers for DiseaseControl and Prevention (CDC) National Center for HealthStatistics (NCHS) Hyattsville MD US Department ofHealth and Human Services Centers for Disease Controland Prevention 2002
29 Behavioral Risk Factor Surveillance System (BRFSS)Centers for Disease Control and Prevention (CDC) NationalCenter for Health Statistics (NCHS) Hyattsville MD USDepartment of Health and Human Services Centers forDisease Control and Prevention 2003
30 Chien-Wen T Phillips RL Green LA Fryer GE Dovey SMWhat physicians need to know about seniors and limitedprescription benefits and why Am Fam Physician200266212
31 Morisky DE Green LW Levine DM Concurrent andpredictive validity of a self-reported measure of medica-tion adherence Med Care 19862467ndash74
32 Barringer TA Kirk JK Santaniello AC Foley KLMichielutte R Effect of a multivitamin and mineralsupplement on infection and quality of life A random-ized double-blind placebo-controlled trial Ann Intern Med2003138365ndash71
33 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
34 Peyrot M Rubin RR Structure and correlates of diabetes-specific locus of control Diabetes Care 199417994ndash1001
35 Wagenknecht LE Mayer EJ Rewers M et al The insulinresistance atherosclerosis study (IRAS) objectivesdesign and recruitment results Ann Epidemiol19955464ndash72
COHORT PROFILE DISTANCE 45
36 King DE Mainous AG III Pearson WS C-reactiveprotein diabetes and attendance at religious servicesDiabetes Care 2002251172ndash76
37 Berkman LF Syme SL Social networks host resistanceand mortality a nine-year follow-up study of AlamedaCounty residents Am J Epidemiol 1979109186ndash204
38 Ware JE Kosinski M Keller SD SF-12 How to Score theSF-12 Physical and Mental Health Summary Scales 2nd ednBoston The Health Institute New England MedicalCenter 1995
39 Ware JE Kosinksi M Dewey JE Gandek B How to Scoreand Interpret Single-Item Health Status Measures A Manual forUsers of the SF-8 Health Survey Lincoln RI QualityMetricIncorporated 2001
40 Cohen S Kamarck T Mermelstein R A global measure ofperceived stress J Health Soc Behav 198324385ndash96
41 Kroenke K Spitzer RL Williams JB The PHQ-9 validityof a brief depression severity measure J Gen Intern Med200116606ndash13
42 Spitzer RL Kroenke K Williams JB Validation and utilityof a self-report version of PRIME-MD the PHQ primarycare study Primary care evaluation of mental disordersPatient Health Questionnaire JAMA 19992821737ndash44
43 The TRIAD Study Group The translating research intoaction for diabetes (TRIAD) study a multicenter study ofdiabetes in managed care Diabetes Care 200225386ndash89
44 Ayas NT White DP Al Delaimy WK et al A prospectivestudy of self-reported sleep duration and incidentdiabetes in women Diabetes Care 200326380ndash84
45 Sickel AE Moore PJ Adler NE Williams DR Jackson JSThe differential effects of sleep quality and quantity onthe relationship between SES and health Ann NY Acad Sci1999896431ndash34
46 Moore PJ Adler NE Williams DR Jackson JSSocioeconomic status and health the role of sleepPsychosom Med 200264337ndash44
47 Benet-Martinez V John OP Los Cinco Grandes acrosscultures and ethnic groups multitrait multimethodanalyses of the Big Five in Spanish and English J PersSoc Psychol 199875729ndash50
48 Srivastava S John OP Gosling SD Potter J Developmentof personality in early and middle adulthood set likeplaster or persistent change J Pers Soc Psychol2003841041ndash53
49 Gosling SD Rentfrow PJ Swann WB A very briefmeasure of the Big-Five personality domains J Res Pers200337504ndash28
50 Schillinger D Grumbach K Piette J et al Association ofhealth literacy with diabetes outcomes JAMA2002288475ndash82
51 Schillinger D Barton LR Karter AJ Wang F Adler NDoes literacy mediate the relationship between educationand health outcomes A study of a low-income popula-tion with diabetes Public Health Rep 2006121245ndash54
52 Chew LD Bradley KA Boyko EJ Brief questions toidentify patients with inadequate health literacy Fam Med200436588ndash94
53 The American Association of Clinical EndocrinologistsMedical Guidelines for the Management of DiabetesMellitus The AACE System of Intensive DiabetesSelf-Management ndash 2002 Update Endocr Pract 20028(Suppl 1)40ndash82
54 Fitzgerald JT Funnell MM Hess GE et al The reliabilityand validity of a brief diabetes knowledge test DiabetesCare 199821706ndash10
55 Piette JD Heisler M Wagner TH Problems paying out-of-pocket medication costs among older adults withdiabetes Diabetes Care 200427384ndash91
56 Epidemiology of Diabetes Interventions andComplications (EDIC) Design implementation andpreliminary results of a long-term follow-up of thediabetes control and complications trial cohort DiabetesCare 19992299ndash111
57 Krieger N Sidney S Coakley E Racial discrimination andskin color in the CARDIA study implications for publichealth research Coronary artery risk development inyoung adults Am J Public Health 1998881308ndash13
58 Krieger N Smith K Naishadham D Hartman CBarbeau EM Experiences of discrimination validity andreliability of a self-report measure for population healthresearch on racism and health Soc Sci Med 2005611576ndash96
59 CAHPS 20 Questionnaires Crofton C AHCPR Pub No97-R079 Washington DC Agency for Health Care Policyand Research 1998
60 Hays RD Shaul JA Williams VS et al Psychometricproperties of the CAHPS 10 survey measures Consumerassessment of health plans study Med Care 199937MS22ndash31
61 Anderson LA Dedrick RF Development of the trust inphysician scale a measure to assess interpersonal trust inpatient-physician relationships Psychol Rep 1990671091ndash100
62 Thom DH Ribisl KM Stewart AL Luke DA Furthervalidation and reliability testing of the Trust in PhysicianScale The Stanford trust study physicians Med Care199937510ndash17
63 Steiner JF Koepsell TD Fihn SD Inui TS A generalmethod of compliance assessment using centralizedpharmacy records Description and validation Med Care198826814ndash23
64 Clark DO Von Korff M Saunders K Baluch WMSimon GE A chronic disease score with empiricallyderived weights Med Care 199533783ndash95
65 Kaiser Permanente Business Intelligence Report 2006Research and Markets 2006 Available at httpwwwresearchandmarketscomreportinfoaspreport_idfrac14451423(Accessed July 9 2007)
66 Krieger N Overcoming the absence of socioeconomic datain medical records validation and application of acensus-based methodology Am J Public Health 199282703ndash10
67 Hiatt RA Friedman GD Characteristics of patientsreferred for treatment of end-stage renal diseasein a defined population Am J Public Health 198272829ndash33
68 Porter S Jackson K Trosclaire A Pederson LL Office onsmoking and Health NCfCDPaHPC Prevalence of currentcigarette smoking among adults and changes in pre-valence of current and some day smoking ndash UnitedStates 1996ndash2001 MMWR 200352303ndash7
69 Smith GD Hart C Blane D Gillis C Hawthorne VLifetime socioeconomic position and mortality prospec-tive observational study BMJ 1997314547ndash52
46 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
70 Lynch JW Kaplan GA Shema SJ Cumulative impact ofsustained economic hardship on physical cognitivepsychological and social functioning N Engl J Med19973371889ndash95
71 Melton LJ3 Dyck PJ Karnes JL OrsquoBrien PC Service FJNon-response bias in studies of diabetic complicationsthe Rochester Diabetic Neuropathy Study J Clin Epidemiol199346341ndash48
72 Robins JM Association causation and marginal struc-tural models Synthese 1999121151ndash79
73 Hernan MA Hernandez-Diaz S Robins JM A structuralapproach to selection bias Epidemiology 200415615ndash25
74 Bodnar LM Davidian M Siega-Riz AM Tsiatis AAMarginal structural models for analyzing causal effects
of time-dependent treatments an application in perinatalepidemiology Am J Epidemiol 2004159926ndash34
75 Bryan JF Yu Z van der Laan M Analysis of longitudinalmarginal structural models U C Berkeley Division ofBiolstatistics Working Paper Series Paper 1201ndash29 2002
76 Neugebauer R van der Laan M Why prefer double robustestimators in causal inference J Stat Plan Inference2005129405ndash26
77 Wang Y Petersen ML Bangsberg D van der Laan MDiagnosing bias in the inverse probability of treatmentweighted estimator resulting from violation ofexperimental treatment assignment U C BerkeleyDivision of Biolstatistics Working Paper Series Paper 2111ndash25 2006
COHORT PROFILE DISTANCE 47
(outcomes) under investigation In this study popula-tion we have substantial data on non-respondentsincluding Kaiser administrative data co-morbidityand census data on neighbourhood characteristicsallowing us to compare hypothesized relationships inrespondents vs non-respondents and adjust for thispotential response bias71 While participation wassomewhat lower among minorities and those withless education overall respondents and non-respon-dents were quite similar
We had pre-baseline reported race data for 29 319(72) members of the DISTANCE cohort andobserved differences in response rates by raceAfrican Americans 52 Asians 46 Caucasians63 Latinos 53 unknown 45 (P lt 00001) Wealso had educational attainment data for 4524subjects (22) who had completed the previoussurvey in 1994ndash97 and observed that subjects with ahigh school education or less participated at a lowerrate than those with more than a high schooleducation (54 vs 60) (P lt 00001)
A more complete view of response bias was obtainedfrom health plan administrative data and census datawhich was available for all members of the cohortWe found few notable differences between respon-dents and non-respondents in demographic clinicalbehavioural or neighbourhood characteristics Com-pared with respondents non-respondents had highermean LDL levels (993 vs 943 mgdl) and were lesslikely to practice SMBG (32 vs 46) However aportion of these unadjusted differences are likely dueto age differences in respondents vs non-respondents
We conducted an assessment of response bias bycomparing differences in A1C across ethnic groupsseparately among respondents and non-respondentsWe specified a logistic regression model of poorglycemic control which regressed the outcome A1C47 on ethnicity age and sex and an interactionterm (ethnicity survey response) Although respon-dents consistently had higher A1C than non-respon-dents ethnic differences in the relationship betweenethnicity and A1C did not differ between respondentsand non-respondents (Pfrac14 055) We conducted asimilar analysis with the subjects for whom we hadeducational attainment data from a previous surveyand again found no substantive response bias(Pfrac14 028) While characteristics differ betweenrespondents and non-respondents (response bias)the associations with outcomes of interest (slopes)are much less vulnerable to such bias7273
Statistical methodsAnalyses of DISTANCE data will focus on associationsrather than descriptive characterization We willemploy a modelling approach using inverse weightingestimation of marginal structural models72 to inves-tigate causal relationships in this study Thesemodels will include weights to adjust for the non-proportionate sampling fractions (eg over-sampling
of the minority ethnic groups) and response bias(eg giving greater weight to respondents whohave characteristics more similar to the non-respondents) Thus the proposed estimating approachwill adjust for confounding and selection biassimultaneously73ndash77
How can I collaborate Where canI find out moreWe are not free to release participantsrsquo personal dataunder our promise to them regarding confidentialityHowever the DISTANCE steering committee is inter-ested in collaborations with external researchersDISTANCE investigators are particularly interested incomparing social disparities observed in this insuredpopulation with uniform access to care to disparitiesobserved in population-based samples where qualityand access to care vary widely by social strataRequests for collaboration must be submitted to thedirector of the central coordinating centre (corre-sponding author) Before collaborations can beinitiated proposals require review and approval bythe DISTANCE Publications and Presentations (PampP)Committee This committee was formed to (i) Ensureaccurate uniform timely and high quality reportingof DISTANCE findings (ii) Preserve the scientificintegrity of the study and (iii) Safeguard the rightsand confidentiality of participants
Supplementary DataSupplementary data are available at IJE online
AcknowledgementsFunds were provided by National Institute ofDiabetes Digestive and Kidney Diseases [R01DK65664] and National Institute of Child Healthand Human Development [R01 HD046113] Dr DS isalso supported by a National Institutes of HealthMentored Clinical Scientist Award [K-23 RR16539]The Public Health InstituteSurvey Research Groupconducted our telephone interviews Gerard andAssociates provided certified translations of all patientmaterials and interview scripts KP Corporation (notaffiliated with Kaiser Permanente) provided printingand mailing services for the written surveys andother communications We thank Marcia Ewing andCarol Rabello for their diligent work on surveyoperations Most of all we thank all the studyparticipants
Conflict of interest None declared
References1 Healthy People 2010 U S Department of Health and
Human Services 2005 Available at httpwwwhealthypeoplegov(Accessed August 2 2007)
44 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
2 Smedley BD Stith AY Nelson AR Unequal TreatmentConfronting Racial and Ethnic Disparities in Health CareWashington DC Institute of Medicine NationalAcademy Press 2002
3 Karter AJ Newman B Rowell S et al Large-scalecollection of family history data and recruitment ofinformative families for genetic analysis J Reg Manag1998257ndash12
4 Karter AJ Ferrara A Darbinian J Ackerson LM Selby JVSelf-monitoring of blood glucose language and financialbarriers in a managed care population with diabetesDiabetes Care 200023477ndash83
5 Karter AJ Moffet HH Liu J et al Achieving goodglycemic control initiation of new antihyperglycemictherapies in patients with type 2 diabetes from theKaiser Permanente Northern California Diabetes RegistryAm J Manag Care 200511262ndash70
6 Karter AJ Ackerson LM Darbinian JA et al Self-monitoring of blood glucose levels and glycemic controlthe Northern California Kaiser Permanente Diabetesregistry Am J Med 20011111ndash9
7 Karter AJ Ferrara A Liu JY Moffet HH Ackerson LMSelby JV Ethnic disparities in diabetic complications inan insured population JAMA 20022872519ndash27
8 Karter AJ Parker MM Moffet HH et al Missedappointments and poor glycemic control an opportunityto identify high-risk diabetic patients Med Care200442110ndash15
9 Selby JV Karter AJ Ackerson LM Ferrara A Liu JDeveloping a prediction rule from automated clinicaldatabases to identify high-risk patients in a largepopulation with diabetes Diabetes Care 2001241547ndash55
10 Martin TL Selby JV Zhang D Physician and patientprevention practices in NIDDM in a large urbanmanaged-care organization Diabetes Care 1995181124ndash32
11 Chaturvedi N Jarrett J Shipley MJ Fuller JHSocioeconomic gradient in morbidity and mortality inpeople with diabetes cohort study findings from theWhitehall Study and the WHO Multinational Study ofVascular Disease in Diabetes BMJ 1998316100ndash5
12 Booth GL Hux JE Relationship between avoidablehospitalizations for diabetes mellitus and income levelArch Intern Med 2003163101ndash6
13 Kleinbaum DG Kupper LL Morgenstern H EpidemiologicResearch Principles and Quantitative Methods New York VanNorstrand Reinhold 1982
14 Dillman DA Mail and Internet Surveys The Tailored DesignMethod 2nd edn New York John Wiley 2000
15 Rudd RE Comings JP Hyde JN Leave no one behindimproving health and risk communication through att-ention to literacy J Health Commun 20038 (Suppl 1)104ndash15
16 Singh-Manoux A Adler NE Marmot MG Subjectivesocial status its determinants and its association withmeasures of ill-health in the Whitehall II study Soc SciMed 2003561321ndash33
17 John D Catherine T MacArthur Research Network onSocioeconomic Status and Health The MacArthur Scale ofSubjective Social Status San Francisco University ofCalifornia 1999 Available at httpwwwmacsesucsfeduResearchPsychosocialnotebooksubjectivehtml(Accessed July 12 2007)
18 Ross CE Wu CL Education age and the cumulativeadvantage in health J Health Soc Behav 199637104ndash20
19 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
20 White S Dillow S Key Concepts and Features of the 2003National Assessment of Adult Literacy (NAALS) (NCES2006ndash471) Washington DC Department of EducationNational Center for Education Statistics 2005
21 Botman SL Moore TF Moriarity CL Parsons VL Designand Estimation for the National Health Interview Survey(NHIS) 1995ndash2004 National Center for Health StatisticsVital Health Stat 2 20001301ndash31
22 Friedman GD Cutter GR Donahue RP et al CARDIAstudy design recruitment and some characteristics of theexamined subjects J Clin Epidemiol 1988411105ndash16
23 Ross CE Wu CL The links between education and healthAm Sociol Rev 199560719ndash45
24 Robert S House JS SES differentials in health by ageand alternative indicators of SES J Aging Health19968359ndash88
25 Toobert DJ Hampson SE Glasgow RE The summary ofdiabetes self-care activities measure results from 7studies and a revised scale Diabetes Care 200023943ndash50
26 Craig CL Marshall AL Sjostrom M et al Internationalphysical activity questionnaire 12-country reliability andvalidity Med Sci Sports Exerc 2003351381ndash95
27 Gordon AJ Maisto SA McNeil M et al Three questionscan detect hazardous drinkers J Fam Pract 200150313ndash20
28 National Health and Nutrition Examination Survey(NHANES) Questionnaire (2001-2002) Centers for DiseaseControl and Prevention (CDC) National Center for HealthStatistics (NCHS) Hyattsville MD US Department ofHealth and Human Services Centers for Disease Controland Prevention 2002
29 Behavioral Risk Factor Surveillance System (BRFSS)Centers for Disease Control and Prevention (CDC) NationalCenter for Health Statistics (NCHS) Hyattsville MD USDepartment of Health and Human Services Centers forDisease Control and Prevention 2003
30 Chien-Wen T Phillips RL Green LA Fryer GE Dovey SMWhat physicians need to know about seniors and limitedprescription benefits and why Am Fam Physician200266212
31 Morisky DE Green LW Levine DM Concurrent andpredictive validity of a self-reported measure of medica-tion adherence Med Care 19862467ndash74
32 Barringer TA Kirk JK Santaniello AC Foley KLMichielutte R Effect of a multivitamin and mineralsupplement on infection and quality of life A random-ized double-blind placebo-controlled trial Ann Intern Med2003138365ndash71
33 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
34 Peyrot M Rubin RR Structure and correlates of diabetes-specific locus of control Diabetes Care 199417994ndash1001
35 Wagenknecht LE Mayer EJ Rewers M et al The insulinresistance atherosclerosis study (IRAS) objectivesdesign and recruitment results Ann Epidemiol19955464ndash72
COHORT PROFILE DISTANCE 45
36 King DE Mainous AG III Pearson WS C-reactiveprotein diabetes and attendance at religious servicesDiabetes Care 2002251172ndash76
37 Berkman LF Syme SL Social networks host resistanceand mortality a nine-year follow-up study of AlamedaCounty residents Am J Epidemiol 1979109186ndash204
38 Ware JE Kosinski M Keller SD SF-12 How to Score theSF-12 Physical and Mental Health Summary Scales 2nd ednBoston The Health Institute New England MedicalCenter 1995
39 Ware JE Kosinksi M Dewey JE Gandek B How to Scoreand Interpret Single-Item Health Status Measures A Manual forUsers of the SF-8 Health Survey Lincoln RI QualityMetricIncorporated 2001
40 Cohen S Kamarck T Mermelstein R A global measure ofperceived stress J Health Soc Behav 198324385ndash96
41 Kroenke K Spitzer RL Williams JB The PHQ-9 validityof a brief depression severity measure J Gen Intern Med200116606ndash13
42 Spitzer RL Kroenke K Williams JB Validation and utilityof a self-report version of PRIME-MD the PHQ primarycare study Primary care evaluation of mental disordersPatient Health Questionnaire JAMA 19992821737ndash44
43 The TRIAD Study Group The translating research intoaction for diabetes (TRIAD) study a multicenter study ofdiabetes in managed care Diabetes Care 200225386ndash89
44 Ayas NT White DP Al Delaimy WK et al A prospectivestudy of self-reported sleep duration and incidentdiabetes in women Diabetes Care 200326380ndash84
45 Sickel AE Moore PJ Adler NE Williams DR Jackson JSThe differential effects of sleep quality and quantity onthe relationship between SES and health Ann NY Acad Sci1999896431ndash34
46 Moore PJ Adler NE Williams DR Jackson JSSocioeconomic status and health the role of sleepPsychosom Med 200264337ndash44
47 Benet-Martinez V John OP Los Cinco Grandes acrosscultures and ethnic groups multitrait multimethodanalyses of the Big Five in Spanish and English J PersSoc Psychol 199875729ndash50
48 Srivastava S John OP Gosling SD Potter J Developmentof personality in early and middle adulthood set likeplaster or persistent change J Pers Soc Psychol2003841041ndash53
49 Gosling SD Rentfrow PJ Swann WB A very briefmeasure of the Big-Five personality domains J Res Pers200337504ndash28
50 Schillinger D Grumbach K Piette J et al Association ofhealth literacy with diabetes outcomes JAMA2002288475ndash82
51 Schillinger D Barton LR Karter AJ Wang F Adler NDoes literacy mediate the relationship between educationand health outcomes A study of a low-income popula-tion with diabetes Public Health Rep 2006121245ndash54
52 Chew LD Bradley KA Boyko EJ Brief questions toidentify patients with inadequate health literacy Fam Med200436588ndash94
53 The American Association of Clinical EndocrinologistsMedical Guidelines for the Management of DiabetesMellitus The AACE System of Intensive DiabetesSelf-Management ndash 2002 Update Endocr Pract 20028(Suppl 1)40ndash82
54 Fitzgerald JT Funnell MM Hess GE et al The reliabilityand validity of a brief diabetes knowledge test DiabetesCare 199821706ndash10
55 Piette JD Heisler M Wagner TH Problems paying out-of-pocket medication costs among older adults withdiabetes Diabetes Care 200427384ndash91
56 Epidemiology of Diabetes Interventions andComplications (EDIC) Design implementation andpreliminary results of a long-term follow-up of thediabetes control and complications trial cohort DiabetesCare 19992299ndash111
57 Krieger N Sidney S Coakley E Racial discrimination andskin color in the CARDIA study implications for publichealth research Coronary artery risk development inyoung adults Am J Public Health 1998881308ndash13
58 Krieger N Smith K Naishadham D Hartman CBarbeau EM Experiences of discrimination validity andreliability of a self-report measure for population healthresearch on racism and health Soc Sci Med 2005611576ndash96
59 CAHPS 20 Questionnaires Crofton C AHCPR Pub No97-R079 Washington DC Agency for Health Care Policyand Research 1998
60 Hays RD Shaul JA Williams VS et al Psychometricproperties of the CAHPS 10 survey measures Consumerassessment of health plans study Med Care 199937MS22ndash31
61 Anderson LA Dedrick RF Development of the trust inphysician scale a measure to assess interpersonal trust inpatient-physician relationships Psychol Rep 1990671091ndash100
62 Thom DH Ribisl KM Stewart AL Luke DA Furthervalidation and reliability testing of the Trust in PhysicianScale The Stanford trust study physicians Med Care199937510ndash17
63 Steiner JF Koepsell TD Fihn SD Inui TS A generalmethod of compliance assessment using centralizedpharmacy records Description and validation Med Care198826814ndash23
64 Clark DO Von Korff M Saunders K Baluch WMSimon GE A chronic disease score with empiricallyderived weights Med Care 199533783ndash95
65 Kaiser Permanente Business Intelligence Report 2006Research and Markets 2006 Available at httpwwwresearchandmarketscomreportinfoaspreport_idfrac14451423(Accessed July 9 2007)
66 Krieger N Overcoming the absence of socioeconomic datain medical records validation and application of acensus-based methodology Am J Public Health 199282703ndash10
67 Hiatt RA Friedman GD Characteristics of patientsreferred for treatment of end-stage renal diseasein a defined population Am J Public Health 198272829ndash33
68 Porter S Jackson K Trosclaire A Pederson LL Office onsmoking and Health NCfCDPaHPC Prevalence of currentcigarette smoking among adults and changes in pre-valence of current and some day smoking ndash UnitedStates 1996ndash2001 MMWR 200352303ndash7
69 Smith GD Hart C Blane D Gillis C Hawthorne VLifetime socioeconomic position and mortality prospec-tive observational study BMJ 1997314547ndash52
46 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
70 Lynch JW Kaplan GA Shema SJ Cumulative impact ofsustained economic hardship on physical cognitivepsychological and social functioning N Engl J Med19973371889ndash95
71 Melton LJ3 Dyck PJ Karnes JL OrsquoBrien PC Service FJNon-response bias in studies of diabetic complicationsthe Rochester Diabetic Neuropathy Study J Clin Epidemiol199346341ndash48
72 Robins JM Association causation and marginal struc-tural models Synthese 1999121151ndash79
73 Hernan MA Hernandez-Diaz S Robins JM A structuralapproach to selection bias Epidemiology 200415615ndash25
74 Bodnar LM Davidian M Siega-Riz AM Tsiatis AAMarginal structural models for analyzing causal effects
of time-dependent treatments an application in perinatalepidemiology Am J Epidemiol 2004159926ndash34
75 Bryan JF Yu Z van der Laan M Analysis of longitudinalmarginal structural models U C Berkeley Division ofBiolstatistics Working Paper Series Paper 1201ndash29 2002
76 Neugebauer R van der Laan M Why prefer double robustestimators in causal inference J Stat Plan Inference2005129405ndash26
77 Wang Y Petersen ML Bangsberg D van der Laan MDiagnosing bias in the inverse probability of treatmentweighted estimator resulting from violation ofexperimental treatment assignment U C BerkeleyDivision of Biolstatistics Working Paper Series Paper 2111ndash25 2006
COHORT PROFILE DISTANCE 47
2 Smedley BD Stith AY Nelson AR Unequal TreatmentConfronting Racial and Ethnic Disparities in Health CareWashington DC Institute of Medicine NationalAcademy Press 2002
3 Karter AJ Newman B Rowell S et al Large-scalecollection of family history data and recruitment ofinformative families for genetic analysis J Reg Manag1998257ndash12
4 Karter AJ Ferrara A Darbinian J Ackerson LM Selby JVSelf-monitoring of blood glucose language and financialbarriers in a managed care population with diabetesDiabetes Care 200023477ndash83
5 Karter AJ Moffet HH Liu J et al Achieving goodglycemic control initiation of new antihyperglycemictherapies in patients with type 2 diabetes from theKaiser Permanente Northern California Diabetes RegistryAm J Manag Care 200511262ndash70
6 Karter AJ Ackerson LM Darbinian JA et al Self-monitoring of blood glucose levels and glycemic controlthe Northern California Kaiser Permanente Diabetesregistry Am J Med 20011111ndash9
7 Karter AJ Ferrara A Liu JY Moffet HH Ackerson LMSelby JV Ethnic disparities in diabetic complications inan insured population JAMA 20022872519ndash27
8 Karter AJ Parker MM Moffet HH et al Missedappointments and poor glycemic control an opportunityto identify high-risk diabetic patients Med Care200442110ndash15
9 Selby JV Karter AJ Ackerson LM Ferrara A Liu JDeveloping a prediction rule from automated clinicaldatabases to identify high-risk patients in a largepopulation with diabetes Diabetes Care 2001241547ndash55
10 Martin TL Selby JV Zhang D Physician and patientprevention practices in NIDDM in a large urbanmanaged-care organization Diabetes Care 1995181124ndash32
11 Chaturvedi N Jarrett J Shipley MJ Fuller JHSocioeconomic gradient in morbidity and mortality inpeople with diabetes cohort study findings from theWhitehall Study and the WHO Multinational Study ofVascular Disease in Diabetes BMJ 1998316100ndash5
12 Booth GL Hux JE Relationship between avoidablehospitalizations for diabetes mellitus and income levelArch Intern Med 2003163101ndash6
13 Kleinbaum DG Kupper LL Morgenstern H EpidemiologicResearch Principles and Quantitative Methods New York VanNorstrand Reinhold 1982
14 Dillman DA Mail and Internet Surveys The Tailored DesignMethod 2nd edn New York John Wiley 2000
15 Rudd RE Comings JP Hyde JN Leave no one behindimproving health and risk communication through att-ention to literacy J Health Commun 20038 (Suppl 1)104ndash15
16 Singh-Manoux A Adler NE Marmot MG Subjectivesocial status its determinants and its association withmeasures of ill-health in the Whitehall II study Soc SciMed 2003561321ndash33
17 John D Catherine T MacArthur Research Network onSocioeconomic Status and Health The MacArthur Scale ofSubjective Social Status San Francisco University ofCalifornia 1999 Available at httpwwwmacsesucsfeduResearchPsychosocialnotebooksubjectivehtml(Accessed July 12 2007)
18 Ross CE Wu CL Education age and the cumulativeadvantage in health J Health Soc Behav 199637104ndash20
19 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
20 White S Dillow S Key Concepts and Features of the 2003National Assessment of Adult Literacy (NAALS) (NCES2006ndash471) Washington DC Department of EducationNational Center for Education Statistics 2005
21 Botman SL Moore TF Moriarity CL Parsons VL Designand Estimation for the National Health Interview Survey(NHIS) 1995ndash2004 National Center for Health StatisticsVital Health Stat 2 20001301ndash31
22 Friedman GD Cutter GR Donahue RP et al CARDIAstudy design recruitment and some characteristics of theexamined subjects J Clin Epidemiol 1988411105ndash16
23 Ross CE Wu CL The links between education and healthAm Sociol Rev 199560719ndash45
24 Robert S House JS SES differentials in health by ageand alternative indicators of SES J Aging Health19968359ndash88
25 Toobert DJ Hampson SE Glasgow RE The summary ofdiabetes self-care activities measure results from 7studies and a revised scale Diabetes Care 200023943ndash50
26 Craig CL Marshall AL Sjostrom M et al Internationalphysical activity questionnaire 12-country reliability andvalidity Med Sci Sports Exerc 2003351381ndash95
27 Gordon AJ Maisto SA McNeil M et al Three questionscan detect hazardous drinkers J Fam Pract 200150313ndash20
28 National Health and Nutrition Examination Survey(NHANES) Questionnaire (2001-2002) Centers for DiseaseControl and Prevention (CDC) National Center for HealthStatistics (NCHS) Hyattsville MD US Department ofHealth and Human Services Centers for Disease Controland Prevention 2002
29 Behavioral Risk Factor Surveillance System (BRFSS)Centers for Disease Control and Prevention (CDC) NationalCenter for Health Statistics (NCHS) Hyattsville MD USDepartment of Health and Human Services Centers forDisease Control and Prevention 2003
30 Chien-Wen T Phillips RL Green LA Fryer GE Dovey SMWhat physicians need to know about seniors and limitedprescription benefits and why Am Fam Physician200266212
31 Morisky DE Green LW Levine DM Concurrent andpredictive validity of a self-reported measure of medica-tion adherence Med Care 19862467ndash74
32 Barringer TA Kirk JK Santaniello AC Foley KLMichielutte R Effect of a multivitamin and mineralsupplement on infection and quality of life A random-ized double-blind placebo-controlled trial Ann Intern Med2003138365ndash71
33 Ross CE Mirowsky J Refining the association betweeneducation and health the effects of quantity credentialand selectivity Demography 199936445ndash60
34 Peyrot M Rubin RR Structure and correlates of diabetes-specific locus of control Diabetes Care 199417994ndash1001
35 Wagenknecht LE Mayer EJ Rewers M et al The insulinresistance atherosclerosis study (IRAS) objectivesdesign and recruitment results Ann Epidemiol19955464ndash72
COHORT PROFILE DISTANCE 45
36 King DE Mainous AG III Pearson WS C-reactiveprotein diabetes and attendance at religious servicesDiabetes Care 2002251172ndash76
37 Berkman LF Syme SL Social networks host resistanceand mortality a nine-year follow-up study of AlamedaCounty residents Am J Epidemiol 1979109186ndash204
38 Ware JE Kosinski M Keller SD SF-12 How to Score theSF-12 Physical and Mental Health Summary Scales 2nd ednBoston The Health Institute New England MedicalCenter 1995
39 Ware JE Kosinksi M Dewey JE Gandek B How to Scoreand Interpret Single-Item Health Status Measures A Manual forUsers of the SF-8 Health Survey Lincoln RI QualityMetricIncorporated 2001
40 Cohen S Kamarck T Mermelstein R A global measure ofperceived stress J Health Soc Behav 198324385ndash96
41 Kroenke K Spitzer RL Williams JB The PHQ-9 validityof a brief depression severity measure J Gen Intern Med200116606ndash13
42 Spitzer RL Kroenke K Williams JB Validation and utilityof a self-report version of PRIME-MD the PHQ primarycare study Primary care evaluation of mental disordersPatient Health Questionnaire JAMA 19992821737ndash44
43 The TRIAD Study Group The translating research intoaction for diabetes (TRIAD) study a multicenter study ofdiabetes in managed care Diabetes Care 200225386ndash89
44 Ayas NT White DP Al Delaimy WK et al A prospectivestudy of self-reported sleep duration and incidentdiabetes in women Diabetes Care 200326380ndash84
45 Sickel AE Moore PJ Adler NE Williams DR Jackson JSThe differential effects of sleep quality and quantity onthe relationship between SES and health Ann NY Acad Sci1999896431ndash34
46 Moore PJ Adler NE Williams DR Jackson JSSocioeconomic status and health the role of sleepPsychosom Med 200264337ndash44
47 Benet-Martinez V John OP Los Cinco Grandes acrosscultures and ethnic groups multitrait multimethodanalyses of the Big Five in Spanish and English J PersSoc Psychol 199875729ndash50
48 Srivastava S John OP Gosling SD Potter J Developmentof personality in early and middle adulthood set likeplaster or persistent change J Pers Soc Psychol2003841041ndash53
49 Gosling SD Rentfrow PJ Swann WB A very briefmeasure of the Big-Five personality domains J Res Pers200337504ndash28
50 Schillinger D Grumbach K Piette J et al Association ofhealth literacy with diabetes outcomes JAMA2002288475ndash82
51 Schillinger D Barton LR Karter AJ Wang F Adler NDoes literacy mediate the relationship between educationand health outcomes A study of a low-income popula-tion with diabetes Public Health Rep 2006121245ndash54
52 Chew LD Bradley KA Boyko EJ Brief questions toidentify patients with inadequate health literacy Fam Med200436588ndash94
53 The American Association of Clinical EndocrinologistsMedical Guidelines for the Management of DiabetesMellitus The AACE System of Intensive DiabetesSelf-Management ndash 2002 Update Endocr Pract 20028(Suppl 1)40ndash82
54 Fitzgerald JT Funnell MM Hess GE et al The reliabilityand validity of a brief diabetes knowledge test DiabetesCare 199821706ndash10
55 Piette JD Heisler M Wagner TH Problems paying out-of-pocket medication costs among older adults withdiabetes Diabetes Care 200427384ndash91
56 Epidemiology of Diabetes Interventions andComplications (EDIC) Design implementation andpreliminary results of a long-term follow-up of thediabetes control and complications trial cohort DiabetesCare 19992299ndash111
57 Krieger N Sidney S Coakley E Racial discrimination andskin color in the CARDIA study implications for publichealth research Coronary artery risk development inyoung adults Am J Public Health 1998881308ndash13
58 Krieger N Smith K Naishadham D Hartman CBarbeau EM Experiences of discrimination validity andreliability of a self-report measure for population healthresearch on racism and health Soc Sci Med 2005611576ndash96
59 CAHPS 20 Questionnaires Crofton C AHCPR Pub No97-R079 Washington DC Agency for Health Care Policyand Research 1998
60 Hays RD Shaul JA Williams VS et al Psychometricproperties of the CAHPS 10 survey measures Consumerassessment of health plans study Med Care 199937MS22ndash31
61 Anderson LA Dedrick RF Development of the trust inphysician scale a measure to assess interpersonal trust inpatient-physician relationships Psychol Rep 1990671091ndash100
62 Thom DH Ribisl KM Stewart AL Luke DA Furthervalidation and reliability testing of the Trust in PhysicianScale The Stanford trust study physicians Med Care199937510ndash17
63 Steiner JF Koepsell TD Fihn SD Inui TS A generalmethod of compliance assessment using centralizedpharmacy records Description and validation Med Care198826814ndash23
64 Clark DO Von Korff M Saunders K Baluch WMSimon GE A chronic disease score with empiricallyderived weights Med Care 199533783ndash95
65 Kaiser Permanente Business Intelligence Report 2006Research and Markets 2006 Available at httpwwwresearchandmarketscomreportinfoaspreport_idfrac14451423(Accessed July 9 2007)
66 Krieger N Overcoming the absence of socioeconomic datain medical records validation and application of acensus-based methodology Am J Public Health 199282703ndash10
67 Hiatt RA Friedman GD Characteristics of patientsreferred for treatment of end-stage renal diseasein a defined population Am J Public Health 198272829ndash33
68 Porter S Jackson K Trosclaire A Pederson LL Office onsmoking and Health NCfCDPaHPC Prevalence of currentcigarette smoking among adults and changes in pre-valence of current and some day smoking ndash UnitedStates 1996ndash2001 MMWR 200352303ndash7
69 Smith GD Hart C Blane D Gillis C Hawthorne VLifetime socioeconomic position and mortality prospec-tive observational study BMJ 1997314547ndash52
46 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
70 Lynch JW Kaplan GA Shema SJ Cumulative impact ofsustained economic hardship on physical cognitivepsychological and social functioning N Engl J Med19973371889ndash95
71 Melton LJ3 Dyck PJ Karnes JL OrsquoBrien PC Service FJNon-response bias in studies of diabetic complicationsthe Rochester Diabetic Neuropathy Study J Clin Epidemiol199346341ndash48
72 Robins JM Association causation and marginal struc-tural models Synthese 1999121151ndash79
73 Hernan MA Hernandez-Diaz S Robins JM A structuralapproach to selection bias Epidemiology 200415615ndash25
74 Bodnar LM Davidian M Siega-Riz AM Tsiatis AAMarginal structural models for analyzing causal effects
of time-dependent treatments an application in perinatalepidemiology Am J Epidemiol 2004159926ndash34
75 Bryan JF Yu Z van der Laan M Analysis of longitudinalmarginal structural models U C Berkeley Division ofBiolstatistics Working Paper Series Paper 1201ndash29 2002
76 Neugebauer R van der Laan M Why prefer double robustestimators in causal inference J Stat Plan Inference2005129405ndash26
77 Wang Y Petersen ML Bangsberg D van der Laan MDiagnosing bias in the inverse probability of treatmentweighted estimator resulting from violation ofexperimental treatment assignment U C BerkeleyDivision of Biolstatistics Working Paper Series Paper 2111ndash25 2006
COHORT PROFILE DISTANCE 47
36 King DE Mainous AG III Pearson WS C-reactiveprotein diabetes and attendance at religious servicesDiabetes Care 2002251172ndash76
37 Berkman LF Syme SL Social networks host resistanceand mortality a nine-year follow-up study of AlamedaCounty residents Am J Epidemiol 1979109186ndash204
38 Ware JE Kosinski M Keller SD SF-12 How to Score theSF-12 Physical and Mental Health Summary Scales 2nd ednBoston The Health Institute New England MedicalCenter 1995
39 Ware JE Kosinksi M Dewey JE Gandek B How to Scoreand Interpret Single-Item Health Status Measures A Manual forUsers of the SF-8 Health Survey Lincoln RI QualityMetricIncorporated 2001
40 Cohen S Kamarck T Mermelstein R A global measure ofperceived stress J Health Soc Behav 198324385ndash96
41 Kroenke K Spitzer RL Williams JB The PHQ-9 validityof a brief depression severity measure J Gen Intern Med200116606ndash13
42 Spitzer RL Kroenke K Williams JB Validation and utilityof a self-report version of PRIME-MD the PHQ primarycare study Primary care evaluation of mental disordersPatient Health Questionnaire JAMA 19992821737ndash44
43 The TRIAD Study Group The translating research intoaction for diabetes (TRIAD) study a multicenter study ofdiabetes in managed care Diabetes Care 200225386ndash89
44 Ayas NT White DP Al Delaimy WK et al A prospectivestudy of self-reported sleep duration and incidentdiabetes in women Diabetes Care 200326380ndash84
45 Sickel AE Moore PJ Adler NE Williams DR Jackson JSThe differential effects of sleep quality and quantity onthe relationship between SES and health Ann NY Acad Sci1999896431ndash34
46 Moore PJ Adler NE Williams DR Jackson JSSocioeconomic status and health the role of sleepPsychosom Med 200264337ndash44
47 Benet-Martinez V John OP Los Cinco Grandes acrosscultures and ethnic groups multitrait multimethodanalyses of the Big Five in Spanish and English J PersSoc Psychol 199875729ndash50
48 Srivastava S John OP Gosling SD Potter J Developmentof personality in early and middle adulthood set likeplaster or persistent change J Pers Soc Psychol2003841041ndash53
49 Gosling SD Rentfrow PJ Swann WB A very briefmeasure of the Big-Five personality domains J Res Pers200337504ndash28
50 Schillinger D Grumbach K Piette J et al Association ofhealth literacy with diabetes outcomes JAMA2002288475ndash82
51 Schillinger D Barton LR Karter AJ Wang F Adler NDoes literacy mediate the relationship between educationand health outcomes A study of a low-income popula-tion with diabetes Public Health Rep 2006121245ndash54
52 Chew LD Bradley KA Boyko EJ Brief questions toidentify patients with inadequate health literacy Fam Med200436588ndash94
53 The American Association of Clinical EndocrinologistsMedical Guidelines for the Management of DiabetesMellitus The AACE System of Intensive DiabetesSelf-Management ndash 2002 Update Endocr Pract 20028(Suppl 1)40ndash82
54 Fitzgerald JT Funnell MM Hess GE et al The reliabilityand validity of a brief diabetes knowledge test DiabetesCare 199821706ndash10
55 Piette JD Heisler M Wagner TH Problems paying out-of-pocket medication costs among older adults withdiabetes Diabetes Care 200427384ndash91
56 Epidemiology of Diabetes Interventions andComplications (EDIC) Design implementation andpreliminary results of a long-term follow-up of thediabetes control and complications trial cohort DiabetesCare 19992299ndash111
57 Krieger N Sidney S Coakley E Racial discrimination andskin color in the CARDIA study implications for publichealth research Coronary artery risk development inyoung adults Am J Public Health 1998881308ndash13
58 Krieger N Smith K Naishadham D Hartman CBarbeau EM Experiences of discrimination validity andreliability of a self-report measure for population healthresearch on racism and health Soc Sci Med 2005611576ndash96
59 CAHPS 20 Questionnaires Crofton C AHCPR Pub No97-R079 Washington DC Agency for Health Care Policyand Research 1998
60 Hays RD Shaul JA Williams VS et al Psychometricproperties of the CAHPS 10 survey measures Consumerassessment of health plans study Med Care 199937MS22ndash31
61 Anderson LA Dedrick RF Development of the trust inphysician scale a measure to assess interpersonal trust inpatient-physician relationships Psychol Rep 1990671091ndash100
62 Thom DH Ribisl KM Stewart AL Luke DA Furthervalidation and reliability testing of the Trust in PhysicianScale The Stanford trust study physicians Med Care199937510ndash17
63 Steiner JF Koepsell TD Fihn SD Inui TS A generalmethod of compliance assessment using centralizedpharmacy records Description and validation Med Care198826814ndash23
64 Clark DO Von Korff M Saunders K Baluch WMSimon GE A chronic disease score with empiricallyderived weights Med Care 199533783ndash95
65 Kaiser Permanente Business Intelligence Report 2006Research and Markets 2006 Available at httpwwwresearchandmarketscomreportinfoaspreport_idfrac14451423(Accessed July 9 2007)
66 Krieger N Overcoming the absence of socioeconomic datain medical records validation and application of acensus-based methodology Am J Public Health 199282703ndash10
67 Hiatt RA Friedman GD Characteristics of patientsreferred for treatment of end-stage renal diseasein a defined population Am J Public Health 198272829ndash33
68 Porter S Jackson K Trosclaire A Pederson LL Office onsmoking and Health NCfCDPaHPC Prevalence of currentcigarette smoking among adults and changes in pre-valence of current and some day smoking ndash UnitedStates 1996ndash2001 MMWR 200352303ndash7
69 Smith GD Hart C Blane D Gillis C Hawthorne VLifetime socioeconomic position and mortality prospec-tive observational study BMJ 1997314547ndash52
46 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
70 Lynch JW Kaplan GA Shema SJ Cumulative impact ofsustained economic hardship on physical cognitivepsychological and social functioning N Engl J Med19973371889ndash95
71 Melton LJ3 Dyck PJ Karnes JL OrsquoBrien PC Service FJNon-response bias in studies of diabetic complicationsthe Rochester Diabetic Neuropathy Study J Clin Epidemiol199346341ndash48
72 Robins JM Association causation and marginal struc-tural models Synthese 1999121151ndash79
73 Hernan MA Hernandez-Diaz S Robins JM A structuralapproach to selection bias Epidemiology 200415615ndash25
74 Bodnar LM Davidian M Siega-Riz AM Tsiatis AAMarginal structural models for analyzing causal effects
of time-dependent treatments an application in perinatalepidemiology Am J Epidemiol 2004159926ndash34
75 Bryan JF Yu Z van der Laan M Analysis of longitudinalmarginal structural models U C Berkeley Division ofBiolstatistics Working Paper Series Paper 1201ndash29 2002
76 Neugebauer R van der Laan M Why prefer double robustestimators in causal inference J Stat Plan Inference2005129405ndash26
77 Wang Y Petersen ML Bangsberg D van der Laan MDiagnosing bias in the inverse probability of treatmentweighted estimator resulting from violation ofexperimental treatment assignment U C BerkeleyDivision of Biolstatistics Working Paper Series Paper 2111ndash25 2006
COHORT PROFILE DISTANCE 47
70 Lynch JW Kaplan GA Shema SJ Cumulative impact ofsustained economic hardship on physical cognitivepsychological and social functioning N Engl J Med19973371889ndash95
71 Melton LJ3 Dyck PJ Karnes JL OrsquoBrien PC Service FJNon-response bias in studies of diabetic complicationsthe Rochester Diabetic Neuropathy Study J Clin Epidemiol199346341ndash48
72 Robins JM Association causation and marginal struc-tural models Synthese 1999121151ndash79
73 Hernan MA Hernandez-Diaz S Robins JM A structuralapproach to selection bias Epidemiology 200415615ndash25
74 Bodnar LM Davidian M Siega-Riz AM Tsiatis AAMarginal structural models for analyzing causal effects
of time-dependent treatments an application in perinatalepidemiology Am J Epidemiol 2004159926ndash34
75 Bryan JF Yu Z van der Laan M Analysis of longitudinalmarginal structural models U C Berkeley Division ofBiolstatistics Working Paper Series Paper 1201ndash29 2002
76 Neugebauer R van der Laan M Why prefer double robustestimators in causal inference J Stat Plan Inference2005129405ndash26
77 Wang Y Petersen ML Bangsberg D van der Laan MDiagnosing bias in the inverse probability of treatmentweighted estimator resulting from violation ofexperimental treatment assignment U C BerkeleyDivision of Biolstatistics Working Paper Series Paper 2111ndash25 2006
COHORT PROFILE DISTANCE 47