Comparative Effectiveness of PeerLeaders and Community HealthWorkers in Diabetes Self-management Support: Results of aRandomized Controlled TrialDOI: 10.2337/dc13-2161

OBJECTIVE

To compare a peer leader (PL) versus a community health worker (CHW) tele-phone outreach intervention in sustaining improvements in HbA1c over 12monthsafter a 6-month diabetes self-management education (DSME) program.

RESEARCH DESIGN AND METHODS

One hundred and sixteen Latino adults with type 2 diabetes were recruited from afederally qualified health center and randomized to 1) a 6-month DSME programfollowed by 12 months of weekly group sessions delivered by PLs with telephoneoutreach to those unable to attend or 2) a 6-month DSME program followed by 12months of monthly telephone outreach delivered by CHWs. The primary outcomewas HbA1c. Secondary outcomes were cardiovascular disease risk factors, diabe-tes distress, and diabetes social support. Assessments were conducted at base-line, 6, 12, and 18 months.

RESULTS

After DSME, the PL group achieved a reduction in mean HbA1c (8.2–7.5% or 66–58mmol/mol, P < 0.0001) that wasmaintained at 18months (20.6% or26.6mmol/molfrombaseline [P = 0.009]). The CHWgroup also showed a reduction in HbA1c (7.8 vs.7.3% or 62 vs. 56 mmol/mol, P = 0.0004) post–6 month DSME; however, it wasattenuated at 18 months (20.3% or 23.3 mmol/mol from baseline, within-groupP = 0.234). Only the PL groupmaintained improvements achieved in blood pressureat 18 months. At the 18-month follow-up, both groups maintained improvementsin waist circumference, diabetes support, and diabetes distress, with no significantdifferences between groups.

CONCLUSIONS

Both low-cost maintenance programs led by either a PL or a CHW maintainedimprovements in key patient-reported diabetes outcomes, but the PL interven-tion may have additional benefit in sustaining clinical improvements beyond 12months.

1University of British Columbia Department ofMedicine, Vancouver, British Columbia, Canada2University of Michigan Department of MedicalEducation, Ann Arbor, MI3University of Michigan School of Social Work,Ann Arbor, MI4Community Health and Social Services Center,Detroit, MI5University of Michigan Department of InternalMedicine, Ann Arbor, MI6Ann Arbor VA Center for Clinical ManagementResearch, Ann Arbor, MI

Corresponding author: Tricia S. Tang, tricia.tang@vch.ca.

Received 16 September 2013 and accepted 29January 2014.

Clinical trial reg. no. NCT01517919, clinicaltrials.gov.

© 2014 by the American Diabetes Association.See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

Tricia S. Tang,1 Martha Funnell,2

Brandy Sinco,3 Gretchen Piatt,2

Gloria Palmisano,4 Michael S. Spencer,3

Edith C. Kieffer,3 and Michele Heisler 5,6

Diabetes Care 1

DIABETES

CARESYM

POSIU

MARTIC

LE

Diabetes Care Publish Ahead of Print, published online April 10, 2014©

To help improve their diabetes out-comes, adults with diabetes often needeffective diabetes self-management ed-ucation (DSME) followed by diabetesself-management support (DSMS). TheNational Standards for Diabetes Self-Management Education defines DSMEas “the ongoing process of facilitatingthe knowledge, skill, and ability neces-sary for pre-diabetes and diabetes self-care,” and DSMS consists of “activitiesthat assist the person with pre-diabetesor diabetes in implementing and sus-taining the behaviors needed tomanagehis or her condition on an ongoing basisbeyond or outside formal self-manage-ment training” (1).DSME programs have been evaluated

extensively (2–5). Yet, to date there islittle evidence to guide subsequentDSMS efforts to maintain gains fromDSME, especially in low-resource healthcare systems. Without ongoing DSMS,improvements achieved from DSMEprograms are short-lived (6 months orless) (4,5). This is especially the casefor low-income ethnic and racial mi-norities in resource-poor settingssuch as inner-city communities whereadults with diabetes face multiple chal-lenges to self-management (6). Feder-ally qualified health centers servingthese communities lack resources tomaintain professionally staffed care-management programs that can pro-vide between-clinic visit outreach topatients who have completed short-term DSME programs. Thus, in the faceof resource constraints, it is critically im-portant to develop and evaluate low-cost, flexible, and sustainable DSMSapproaches that do not rely on healthcare professionals to maintain gainsachieved through DSME.Intervention studies of DSMS have as-

sessed delivery modalities includinggroup-based support sessions, periodiceducational reinforcement, telephoneoutreach, peer support groups, inter-net-based discussion boards, and auto-mated telephone reminders (7–11).These programs have recruited a varietyof DSMS interventionists including cer-tified diabetes educators, psychologists,dietitians, care coordinators, physicians,health educators, and peers. To ourknowledge, however, there are no pub-lished studies comparing two differentmodels of ongoing DSMS delivered bynonprofessionals. Moreover, those few

studies that have compared profes-sional staff such as Certified DiabetesEducators and peer supporters havenot examined long-term DSMS beyond6months (11). Accordingly, we compareda peer leader (PL) DSMS interventionwith a community health worker (CHW)DSMS intervention as two possible ap-proaches to maintain over a 12-monthperiod health-related gains achievedthrough Journey to Health (JTH), anevidence-based, CHW-led, 6-monthDSME program (12). In our prior trials,including one randomized controlled trial(RCT), JTH has been found to significantlyimprove short-term glycemic controlcompared with usual care (13,14).

RESEARCH DESIGN AND METHODS

Setting and Identification of PatientsThis study was approved by the Univer-sity of Michigan Institutional ReviewBoard and was developed and imple-mented using community-based partic-ipatory research principles (15). It wasconducted in partnership with the Com-munity Health and Social Services Cen-ter (CHASS), Inc., a federally qualifiedcommunity health center located inSouthwest Detroit, and the Racial andEthnic Approaches to CommunityHealth (REACH) Detroit partnership, acoalition of community organizations,academic institutions, and health caresystems working to improve diabetesoutcomes and care in east and south-west Detroit. Approximately 70% of res-idents of southwest Detroit are Latino ofMexican origin with an annual medianhousehold income of 27,248–31,097USD (depending on zip code) and highrates of diabetes and obesity (16).

From November 2009 to July 2011,we identified participants from a com-puter-generated list of potentially eligi-ble patients who were receiving medicalcare at CHASS with physician-diagnosedtype 2 diabetes, were at least 21 yearsold, had a regular health care provider,and self-identified as Latino.We excludedindividuals who had physical limitationspreventing participation, terminal healthconditions, serious psychiatric illness, andself-reported excessive alcohol or illicitdrug use.

Recruitment and Randomization ofthe PL Versus CHW TelephoneOutreach StudyThe patients in this study represent asubsample of patients recruited for an

RCT that continued recruitment beyondthe time period of this substudy. Thatstudy compared outcomes between alarger group of participants receivingthe 6-month JTH program and an armthat received enhanced usual care.

We contacted potentially eligible pa-tients by telephone and screened for el-igibility via phone or in the clinic. Eligiblepatients were invited to a group infor-mation session in which they were toldthat if they were randomized to the6-month DSME, at the end of the DSMEthey would be offered either a CHW or apeer-led maintenance program over 12months. Those interested in participatingsigned informed consent documentsand enrolled in the study. Participantscompleted a baseline survey, and theirHbA1c, lipid panel, blood pressure (BP),height, weight, BMI, and waist circum-ference were measured. Participantswere then randomized to one of twoarms for the purpose of this substudy:1) a 6-month CHW-led JTH Program fol-lowed by 12 months of the peer supportintervention or 2) the 6-month CHW-ledJTH program followed by 12 months ofmonthly CHW telephone outreach.

Random sequence generation andtreatment group assignment were de-termined centrally just prior to the ini-tial session. Sequence was concealeduntil interventions were assigned. Pa-tients, research staff, and caremanagerswere blinded to randomization resultsuntil completion of baseline assess-ments. Data assessors remained blindedto group assignment throughout thestudy. Participants received a stipendof 40 USD upon completion of each ofthe baseline, 6-month, 12-month, and18-month assessments.

PLsThe PLs were volunteers and receivedonly a modest stipend to defray costsof participation (e.g., transportation,child care). To be eligible for the study,PL candidates had to 1) have diabetes, 2)be a resident of the southwest Detroitcommunity, 3) be $21 years old, 4) bebilingual in Spanish and English, 5) be agraduate of JTH, 6) have transportationto attend training, and 7) be willing tocommit to 3 months of training. EligiblePL candidates underwent a 46-h trainingprogram over 12 weeks (17). The trainingconsisted of three main components:review of basic diabetes information;

2 Peer Compared with Community Health Worker Follow-up Diabetes Care

©

communication, facilitation, and behav-ior modification skills; and practice ap-plying skills in experiential learningscenarios. Specifically, PL candidateswere trained to help patients build mo-tivation for making lifestyle changes,use basic motivational interviewing skills(e.g., active listening and making reflec-tions), apply empowerment-based facil-itation strategies, set goals, developaction plans, and problem solve. Tograduate successfully, candidates hadto meet the pre-established compe-tency criteria for four domains: dia-betes knowledge, active listening,empowerment-based facilitation, andself-efficacy.

CHWsThe CHWs had an average of 6 years’ ex-perience leading DSME at CHASS. Theywere all employees of the health clinicand received a salary. They all camefrom southwest Detroit, had a high schooldiploma or GED, and were fluent inSpanish and of Latino origin. There wasno requirement for them to have diabe-tes. They had completed 160 h of com-munity outreach training, 80h of diabeteseducation, home visit experiences, humansubjects tutorial; had training in behaviormodification strategies, JTH curriculum,cultural competency, motivational inter-viewing techniques, and community-based participatory research; and hadbasic computer and internet skills.

Description of Intervention

DSME Program

Both PL and CHW groups participatedin the 6-month JTH program (12) de-livered by CHWs trained in patient-empowerment approaches (18) andmotivational interviewing (19). JTH con-sisted of three components: eleven 2-hculturally tailored interactive group self-management classes, two home visits(60 min in length) per month to helppatients set and follow up on specificself-care goals (action plans), and onevisit with the participant and his or herprimary care provider. This 6-monthprogram in our prior RCT among adultswith diabetes at CHASS led to mean de-creases of 0.8% HbA1c compared with a0.0% point drop among participants inthe usual-care group (20).

PL DSMSAdapted from the Lifelong Manage-ment program of Tang et al. (7), the PL

intervention (delivered largely in Span-ish) was designed to provide patientswith ongoing emotional and behavioralsupport by PLs through group-based ses-sions and follow-up telephone contacts.Group-based DSMS sessions were of-fered weekly, with participants encour-aged to attend sessions as often asneeded. Based on patient-empowermentprinciples (18), discussion topics weredriven by patients’ self-identified priori-ties, questions, and concerns. The PLsdid seek to complete at each sessionthe following five tasks: discuss recentself-management challenges, share feel-ings about these challenges and otheraspects of living with diabetes, engagein group-based problem solving, addressquestions about diabetes and its care,and set self-management goals. The PLshelped participants set goals using thefive-step goal-setting model, whichincludes: 1) exploring a participant-identified problem, 2) discussing theemotional impact of the problem, 3)selecting a self-management goal, 4) de-veloping an action plan, and 5) executingand evaluating the action plan (20). PLsalso provided support to participants bydiscussing psychosocial concerns, identi-fying facilitators and barriers to behaviorchange, taking inventory of supportsources, anddeveloping strategies to nav-igate the health care system.

To ensure regular contact with eachparticipant, PLs made a telephone sup-port call to any participant who had notattended a DSMS session over threeconsecutive weeks. During the tele-phone support calls, PLs facilitated aconversation that closely mirrored sup-port activities conducted in the groupsetting.

CHW DSMSSimilar to the PL intervention, the CHWsupport is also based on empowermentprinciples and involved monthly tele-phone outreach initiated by a CHWover a period of 12months. Themonthlycalls were structured around the five-step behavioral goal-setting model de-scribed above. During this call, CHWsalso offered emotional support andhelped participants understand how toeffectively use the health care resourcesavailable to them. CHWs and partici-pants were also encouraged to ex-change e-mail communication whenneeded.

Outcomes and MeasurementsThe primary clinical outcomewas HbA1c,as measured with a Bayer DCA2000+Analyzer (21). This assay has a test co-efficient of variation,5% as required bythe National Diabetes Data Group. Sec-ondary clinical outcomes included a lipidpanel (total cholesterol, LDL, and HDL),BP, BMI, and waist circumference. Tomeasure lipids, we used the CholestechLDX (Cholestech Corp., Hayward, CA)point-of-care machine (3), which hasbeen found to meet National Choles-terol Education Program guidelines formeasuring total cholesterol (22). Sys-tolic and diastolic BP were taken withtwo readings on a Welch Allyn SpeidelKeller sphygmomanometer; the averagereadings were used in the analysis. Allclients were weighed on an Ever Weighlithium digital scale. Heights and waistcircumference were measured by thesame technician at each time point.We used weight and height measure-ments to calculate BMI as weight in kilo-grams divided by the square of height inmeters. Waist circumference was mea-sured using the Tech-Med model cat.no. 4414 measuring tape and was mea-sured at the umbilical waist.

All other measures were adminis-tered orally in patients’ preferredlanguage (English or Spanish). We as-sessed diabetes-related distress usingthe Diabetes Distress Scale (DDS), a 17-item instrument that assesses emo-tional distress and functioning specificto living with diabetes, with higherscores indicating higher levels of distress(23). We assessed diabetes-specific so-cial support with an adapted versionof the Diabetes Support Scale, a six-item instrument that assesses patient-perceived social support as it relatesto meeting emotional needs, seekingadvice, and obtaining information,with higher scores indicating more sup-port (24).

Statistical AnalysisThe sample size was estimated for lon-gitudinal analysis with a linear mixedmodel to detect a difference of 0.6%,or 6.6 mmol/mol, in HbA1c change be-tween groups with 80% power and atwo-sided a of 0.05. Although the UKProspective Diabetes Study foundthat a 0.5%, or 5.5 mmol/mol, mean dif-ference in HbA1c translates into a 2.8%absolute risk reduction in diabetes

care.diabetesjournals.org Tang and Associates 3

©

events over a 10-year period (25), forrecruitment concerns, our study waspowered to detect a difference of0.6%, or 6.6 mmol/mol. The power cal-culation was based on an SD of 1.75%,or 19.1 mmol/mol, and an intraclasscorrelation coefficient of 0.5 for re-peated measures over time on thesame individual (26).For our analyses of our primary out-

come of HbA1c, we examined 1) whetherwithin-group HbA1c gains achieved after6 months of the JTH program were sus-tained at 18 months after receipt oftheir assigned DSMS intervention and2) whether there were differencesin HbA1c between the two groups at18 months. Success of either arm inmaintaining gains is indicated whenP values , 0.05 are observed betweenbaseline and follow-up time points, in-dicating that improvements in HbA1c atmonth 6 were sustained at months 12and 18. Secondary outcomes includedsustained improvement at 18 monthsof changes between baseline and6 months in cardiovascular disease riskfactors, including total cholesterol, LDLcholesterol, HDL cholesterol, BP, BMI,and waist circumference, and the self-reported psychosocial measures.All analyses were intention to treat.

All continuous measures, except diabe-tes duration, were compared betweenthe PL intervention and the CHW tele-phone outreach intervention with theStudent t test. Diabetes duration wasanalyzed with the log-rank test . Tocheck for differences between groupsin categorical variables, the Fisher exacttest was used if the expected count inany cell was under 5; Pearson x2 testwas used for all other categoricalvariables.All longitudinal outcomes, except the

DDS, were analyzed by using a linearmixed model, except for the DDS, whichwas analyzed using a generalized esti-mating equation. Both the generalizedestimating equation and linear mixedmodel allow for correlation among ob-servations on the same person and en-able participants to be included in theanalysis if they had data at one or moretime points (27,28). All models were ad-justed for time (6, 12, and 18 months),study group, a time 3 study group in-teraction, and the baseline value. For alloutcomes, the “intervention effect”wasestimated as a contrast between the

changes from baseline to follow-up be-tween the two intervention groups.

Treatment of Missing DataWe also conducted sensitivity analysesthat included variables for diabetes,cholesterol, and blood pressure medica-tion intensification to ensure that inter-vention effects were not principally dueto medication intensification. As therewas no change in the results, we reportthe unadjusted results.

We followed CONSORT (ConsolidatedStandards of Reporting Trials) guidelinesfor analyses and reporting (29).

RESULTS

Participant Flow and Baseline DataOf the 756 potentially eligible patients,54% (n = 406) did not meet inclusioncriteria, and 23% (n = 177) could notbe contacted (Fig. 1). Of the 173 eligiblepatients, 4% (n = 7) consented but werenot randomized and 29% (n = 50) de-clined to participate. Of 116 randomizedpatients, 96 had HbA1c data at the 6-month assessment, 83 at the 12-monthpost–JTH assessment, and 69 at the 18-month assessment (attrition rate 41%)(Fig. 1). Loss to follow-up was not differ-ent between the two groups and wasnot associated with clinical or demo-graphic variables. When comparingstudy participants (n = 116) with theeligible population (n = 173) to deter-mine generalizability, there were nostatistical differences in demographicvariables.

Patient characteristics are presentedin Table 1. There were no significant dif-ferences in baseline characteristics be-tween groups (Table 1), except for aborderline significant difference in dia-betes social support (PL 4.0 vs. CHW 4.4,P = 0.053).

Glycemic ControlAt 6 months, all participants experi-enced on average a significant reductionin HbA1c levels (PL 20.7% or 27.7mmol/mol, P , 0.0001; CHW 20.5%or 25.5 mmol/mol, P = 0.004) (Table2; Fig. 2). These improvements weremaintained in both groups at 12 months(PL20.6% or26.6mmol/mol, P = 0.001;CHW 20.4% or 24.4 mmol/mol, P =0.011). However, at 18 months, the PLgroup sustained their initial HbA1c im-provement (20.6% or 26.6 mmol/mol,P = 0.009), while the CHW group’s aver-age HbA1c levels began to rise (20.3% or

23.3 mmol/mol, P = 0.234). There wereno significant differences betweengroups at any time point (Table 2), andneither group returned to baseline.

Cardiovascular Risk Factors

LDL Cholesterol

At 6 and 12 months, there were no sta-tistically significant changes in LDL cho-lesterol levels for either group. At 18months, PL group participants experi-enced significant declines in mean LDLlevels, by close to 14 mg/dL at 18months, while the CHW group experi-enced no improvements (PL 214.3mg/dL, P = 0.009; CHW 28.4 mg/dL,P = 0.103). There were no significant be-tween-group differences at any timepoint for LDL (Table 2).

Blood Pressure

Within the PL group, average systolicand diastolic BPs declined by 6.6mmHg (P = 0.001) and 3.9 mmHg (P =0.001), respectively, from baseline to6 months, and the results were sustainedat 12 months (systolic BP 26.4 mmHg,P = 0.003; diastolic BP 4.1 mmHg, P =0.002) and 18 months (systolic BP 25.8mmHg, P = 0.010; diastolic BP 23.4mmHg, P = 0.013) (Table 2). None of theimprovements in blood pressure in theCHW group were statistically significant(Table 2).

Waist Circumference

Average waist circumference decreasedsignificantly in both groups at 6 months(PL 21.8 inches, P , 0.001; CHW 21.4inches, P = 0.001), with no significantdifferences between groups. At 18months, both groups sustained a 21.3inch reduction (P = 0.001) in waist cir-cumference. There were no significantlong-term changes in HDL or in BMI.

Psychosocial Outcomes

Diabetes Social Support

Table 2 also shows changes in perceiveddiabetes social support over time forboth groups. From baseline to 6months,levels of social support increased withinboth groups (PL 1.0, P , 0.0001; CHW0.6, P, 0.0001). Both groups sustainedthe improvements in social support at12 months (PL 0.8, P , 0.0001; CHW0.4, P = 0.001) and at 18 months (PL0.6, P = 0.0001; CHW 0.3, P = 0.050);however, the intervention effects wereonly significant between groups at6 months (P = 0.0004) and at 12 months(P = 0.025) (Table 2).

4 Peer Compared with Community Health Worker Follow-up Diabetes Care

©

Diabetes Distress

Significant decreases in the proportion ofindividuals who reported high diabetesdistress occurred at the 6-month follow-up in each group. However, the reductions

in the proportion of individuals with highdiabetes distress were not sustained be-yond6months in either group. In contrast,the proportion of individuals who re-ported moderate diabetes distress levels

in the CHW group decreased from 28.6%to 14.5% at the 6-month follow-up, P =0.013, and was sustained at 12 months(16.2%, P = 0.003) and at 18 months(18.8%, P = 0.030). Within the PL group,

Figure 1—Consort flow diagram for Detroit, 2012. (A high-quality color representation of this figure is available in the online issue.)

care.diabetesjournals.org Tang and Associates 5

©

the proportion of individuals who re-ported moderate diabetes distress de-clined but not significantly until 18months (baseline 18.3%, 18 month 7.0%,P = 0.026).

Frequency of PL-participant andCHW-participant contactsAll contact data are based on participantself-report. Total number of contacts wascalculated by adding number of groupsessions participants reported attendingplus number of telephone support callsreceived for the PL group and total num-ber of telephone support calls receivedplus number of emails exchanged for

the CHW group. In the PL group, 45.0%(n = 27) had at least one contact with his/her PL between 6- and 18-month assess-ments, with an average of 3.67 contacts.In the CHW group, 53.6% (n = 30) had atleast one contact with his/her CHW be-tween 6- and 18- month assessments,with an average of 2.88 contacts. No sig-nificant differences in mean or percent-age contacts were found between the PLand CHW groups.

CONCLUSIONS

Among this sample of low-income, pre-dominantly Spanish-speaking adults with

diabetes who receive care at a federallyqualified health center, both the PL andCHW interventions were effective in sus-taining achieved HbA1c improvementsfrom a 6-month CHW-led DSME programup to and including 12-month follow-up.At 18-month follow-up, the groups di-verged as the PL group sustained improve-ment in HbA1c, while the CHW group didnot. However, it should be noted thatthere were no significant between-groupdifferences in theamountof improvementobserved between the groups.

Unlike the CHW group, the PL groupsustained statistically and clinically

Table 1—Baseline characteristics of CHASS participants (N = 116)

Characteristic PL group CHW group Total P for between-group difference

n 60 56 116

Age (years), mean (SD) 50.2 (11.1) 48.4 (10.8) 49.3 (11.0) 0.36a

Men, n (%) 28 (46.7) 20 (35.7) 48 (41.4) 0.23b

Spanish speaking, n (%) 49 (81.7) 51 (91.1) 100 (86.2) 0.14b

Education, n (%) 0.33c

#Some high school 43 (71.7) 46 (83.6) 89 (77.4)High school graduate or GED 10 (16.7) 6 (10.9) 16 (13.9)Some college/technical/vocational training 7 (11.7) 3 (5.5) 10 (8.7)$College graduate 0 (0.0) 0 (0.0) 0 (0.0)

Employed, n (%) 26 (43.3) 25 (44.6) 51 (44.0) 0.89b

Have health insurance, n (%) 100 at federally qualified healthcenter

N/A

Household income (USD), n (%) 0.13b

,20,000 55 (96.5) 46 (92.0) 101 (94.4)20,000–49,999 5 (8.8) 10 (20.0) 15 (14.0)$50,000 0 (0.0) 0 (0.0) 0 (0.0)

Social support, n (%)Married or partnered 45 (75.0) 38 (67.9) 83 (71.6) 0.39b

Diabetes social support, mean (SD) 4.0 (1.2) 4.4 (1.0) 4.2 (1.1) 0.053a

Antihyperglycemic medication, n (%) 0.78c

No medications 3 (5.0) 2 (3.6) 5 (4.3)Only oral diabetes medication 45 (75.0) 40 (71.4) 85 (73.3)Insulin with or without medication 12 (20.0) 14 (25.0) 26 (22.4)

Self-rated fair or poor general health, n (%) 44 (73.3) 45 (80.4) 89 (76.7) 0.37b

Minimal depression, n (%)d 14 (23.3) 11 (19.6) 25 (21.6) 0.63b

DDSe 0.36b

Little or no distress 37 (61.7) 28 (50.0) 65 (56.0)Moderate distress 11 (18.3) 16 (28.6) 27 (23.3)High distress 12 (20.0) 12 (21.4) 24 (20.7)

Physiological measures, mean (SD)

HbA1c (%) 8.2 (2.2) 7.8 (1.7) 8.0 (2.0) 0.25a

HbA1c (mmol/mol) 66.0 (24.0) 62.0 (18.6) 64.0 (21.9) 0.25a

LDL cholesterol (mg/dL) 102.1 (35.3) 95.5 (29.9) 98.9 (32.8) 0.33a

HDL cholesterol (mg/dL) 40.5 (16.8) 40.7 (13.8) 40.6 (15.3) 0.95a

Systolic BP (mmHg) 134.8 (17.8) 131.6 (18.2) 133.3 (18.0) 0.34a

Diastolic BP (mmHg) 81.2 (10.1) 78.7 (10.8) 80.0 (10.5) 0.19a

BMI (kg/m2) 33.0 (7.6) 32.0 (4.6) 32.5 (6.3) 0.28a

Waist circumference (inches) 41.8 (6.3) 40.6 (4.8) 41.2 (5.6) 0.26a

Diabetes duration (years) 6.7 (5.8) 6.4 (6.1) 6.6 (5.9) 0.80f

at test. bPearson x2 test. cFisher exact test. dMinimal depression indicated by PHQ (Primary Care Evaluation of Mental Disorders)$3. eDDS. DDS,2,little or no distress; 2 # DDS , 3, moderate distress; DDS $3, high distress. fLog-rank test.

6 Peer Compared with Community Health Worker Follow-up Diabetes Care

©

significant improvements in HbA1c andBP to 18 months. By design, the PL in-tervention of weekly face-to-face group

sessions supplemented by telephonesupport calls offers greater intensityand frequency of support to participants

than the CHW intervention of monthlytelephone support calls supplementedby emails. Therefore, any advantages

Table 2—Changes in clinical and psychosocial outcomes over time

Outcome and time point Baseline 6 months–baseline 12 months–baseline 18 months –baseline

HbA1c (%)*PL 8.2 (7.7–8.8) 20.7 (21.0 to 20.4),

P , 0.000120.6 (20.9 to 20.3),

P = 0.00120.6 (21.0 to 20.2),

P = 0.009CHW 7.8 (7.4–8.3) 20.5 (20.8 to 20.3),

P = 0.000420.4 (20.7 to 20.1),

P = 0.01120.3 (20.7 to 0.2),

P = 0.234CHW vs. PL** 0.253 0.883 0.867 0.725

HbA1c (mmol/mol)*PL 66.0 (61.0–73.0) 27.7 (210.9 to 24.4),

P , 0.000126.6 (29.8 to 23.3),

P = 0.00126.6 (210.9 to 22.2),

P = 0.009CHW 62.0 (57.0–67.0) 25.5 (28.7 to 23.3),

P = 0.000424.4 (27.7 to 21.1),

P = 0.01123.3 (27.7 to 2.2),

P = 0.234CHW vs. PL** 0.253 0.883 0.867 0.725

LDL cholesterol (mg/dL)*PL 102.1 (92.0–112.2) 2.1 (26.9 to 11.0),

P = 0.6470.8 (28.7 to 10.3),

P = 0.866214.3 (224.8 to 23.7),

P = 0.009CHW 95.5 (86.8–104.2) 3.9 (24.8 to 12.6),

P = 0.3791.2 (28.5 to 10.8),

P = 0.81128.4 (218.4 to 1.6),

P = 0.103CHW vs. PL** 0.327 0.775 0.958 0.428

HDL cholesterol (mg/dL)*PL 40.5 (36.0–45.0) 0.4 (22.4 to 3.2),

P = 0.7983.0 (20.1 to 6.0),

P = 0.0570.6 (22.5 to 3.7),

P = 0.706CHW 40.6 (36.9–44.4) 1.0 (21.8 to 3.8),

P = 0.4822.1 (20.9 to 5.2),

P = 0.16721.4 (24.6 to 1.7),

P = 0.371CHW vs. PL** 0.955 0.754 0.710 0.366

Systolic BP (mmHg)*PL 134.8 (130.2–139.4) 26.6 (210.5 to 22.7),

P = 0.00126.4 (210.6 to 22.2),

P = 0.00325.8 (210.2 to 21.4),

P = 0.010CHW 131.6 (126.7–136.5) 23.8 (27.8 to 0.2),

P = 0.06123.1 (27.3 to 1.2),

P = 0.15821.0 (25.7 to 3.6),

P = 0.661CHW vs. PL** 0.344 0.335 0.278 0.148

Diastolic BP (mmHg)*PL 81.2 (78.6–83.8) 23.9 (26.3 to 21.6),

P = 0.00124.1 (26.6 to 21.6),

P = 0.00223.4 (26.1 to 20.7),

P = 0.013CHW 78.7 (75.8–81.6) 20.7 (23.1 to 1.8),

P = 0.59622.0 (24.5 to 0.6),

P = 0.13520.7 (23.6 to 2.1),

P = 0.623CHW vs. PL** 0.194 0.058 0.247 0.178

BMI (kg/m2)*PL 33.1 (31.2–35.1) 0.1 (20.2 to 0.4),

P = 0.4620.2 (20.2 to 0.6),

P = 0.3710.3 (20.2 to 0.8),

P = 0.247CHW 31.9 (30.6–33.1) 20.4 (20.8 to 20.1),

P = 0.00720.4 (20.8 to 0.0),

P = 0.07420.2 (20.7 to 0.4),

P = 0.553CHW vs. PL** 0.284 0.015 0.058 0.221

Waist circumference (inches)*PL 41.8 (40.2–43.4) 21.8 (22.5 to 21.1),

P , 0.000121.7 (22.6 to 20.8),

P = 0.000421.3 (22.1 to 20.6),

P = 0.001CHW 40.6 (39.4–41.9) 21.4 (22.1 to 20.8),

P , 0.000121.8 (22.7 to 20.9),

P = 0.000321.4 (22.1 to 20.6),

P = 0.001CHW vs. PL** 0.263 0.460 0.903 0.985

Diabetes Support Scale*PL 4.0 (3.7–4.3) 1.0 (0.8–1.1),

P , 0.00010.8 (0.6–1.1),P , 0.0001

0.6 (0.3–0.9),P = 0.0001

CHW 4.4 (4.1–4.7) 0.6 (0.4–0.7),P , 0.0001

0.4 (0.2–0.7),P = 0.001

0.3 (0.0–0.6),P = 0.050

CHW vs. PL** 0.053 0.0004 0.025 0.165

Data are estimates for means (95% CI). Linear mixed model for all clinical outcomes and diabetes support. Baseline to 18 month follow-up (N = 116:n = 56 CHW, n = 60 PL). All difference scores adjusted for baseline values. *P values for within-group difference from given time point to baseline.**Between-group P values. Baseline tests from Student t test; difference scores relative to baseline from linear mixed model implemented in SASProc Mixed.

care.diabetesjournals.org Tang and Associates 7

©

in long-term health maintenance by thePL group should be interpreted withcaution. However, since there was nodifference in the mean number of PLparticipant and CHW participant con-tacts, these findings cannot be attrib-uted to a treatment dosage effect. It ispossible that the PL intervention out-performed the CHW telephone inter-vention because participants felt acloser connection and greater identifica-tion with PLs who also live with the chal-lenges of having diabetes. This questionshould be explored in future studies.As originally designed, the core sup-

port mechanism of the PL interventionwas intended to be weekly group sup-port sessions. Although regular logswere not kept, the PLs all reported hav-ing substantially more telephone sup-port contact with participants thangroup-based face-to-face contact. Whilewe expected variations in preferredcommunication modality across partici-pants, we did not predict that the orig-inally designed group interventionwould essentially transform into a tele-phone outreach intervention. Upon fur-ther investigation, we learned that itrequired great effort for some patientsto attend sessions (e.g., taking two dif-ferent bus lines to reach the clinic).Other participants had other priorities(e.g., child care, employment) thattook precedence over attending aweekly peer-led session. Consistentwith these reports, in a qualitative studyof 37 Latino adults with personal or fa-milial experience with diabetes, Baig

et al. (30) found lack of time a majordeterrent to attending regular group-based, peer-led self-managementinterventions. Consequently, we didnot actually test two different deliverymodalities as the primary supportmech-anism, as both interventions consistedof telephone outreach. Notwithstand-ing, we were still able to compare twosimilar interventions delivered by twodifferent types of interventionists (PLvs. CHW) who were not formal healthcare professionals.

To date, all RCTs investigating CHW-led diabetes self-management interven-tions targeting the Latino communityhave focused on comparing CHWs withusual care (31,32), CHWs with CHWsusing different interventions (33),CHWs versus attention control (34),and CHWs versus wait-list usual care(35,36). The current study is the first touse a rigorously designed RCT to com-pare the effectiveness of two differenttypes of nonprofessional delivery mod-els in sustaining diabetes-related healthoutcomes. That both interventions pro-duced positive clinical and psychosocialoutcomes has important implicationsfor health systems in low-resource set-tings. It should be noted that not allLatino communities find a peer supportmodel culturally acceptable. However,with respect to the current study, themajority of participants agreed thathaving a PL was important (81%) andthat the PLs played a critical role in theintervention (100%).

Like other studies using nonprofes-sional delivery models (37), we did notinclude a formal cost-effectiveness anal-ysis. The few CHW-led interventionsthat have examined cost-related varia-bles have reported favorable results(38–40). While not comprehensive, wedid document some basic data thatallows us to compare the relativelabor-related costs of the PL and CHWinterventions. PLs in this study receiveda small stipend for training and inter-vention delivery (annual stipend of~1,000 USD), while the CHWs were sal-aried employees receiving benefits(annual salary ~29,120 USD). Comparedwith CHWs, training PLs to provide ongo-ing DSMS might thus result in substantialcost savings, especially in settings inwhich CHWs are not already employed.For communities that do not have thefinancial resources to even hire CHWs,recruiting and training PLs could be a vi-able evidence-based option.

Some limitations need mentioning.First, as noted, in the absence of a for-mal cost-effectiveness analysis we can-not demonstrate any cost savings orquality-adjusted life-years associatedwith the two interventions. Second,given that the PL and CHW interventionswere designed to offer different treat-ment intensity levels, comparing thetwo interventions may not have beenequitable. Fortunately, our analysis ofmean number of participant-reportedcontacts revealed that intervention dos-age was similar for both groups. Third,our follow-up rate of 59% was low

Figure 2—Trajectory of HbA1c means from baseline to 18 months. Data are presented as % (mmol/mol). (A high-quality color representation of thisfigure is available in the online issue.)

8 Peer Compared with Community Health Worker Follow-up Diabetes Care

©

compared with the 63–93% range re-ported in other studies targeting theLatino community (10,32). However,this 18-month intervention was length-ier (32) and used a stricter definition for“dropouts” (31) than some other stud-ies. While we used effective retentionstrategies such as requesting contactinformation from multiple friends orfamily members and accommodatingparticipants’ availability for assess-ments, we were not able to use morecostly methods such as providing paidtaxis, conducting assessments at partic-ipants’ homes, or hiring staff dedicatedexclusively to addressing participantretention. Clearly, for very low-income,non–English speaking populations,future research needs to explore moreeffective and sustainable yet low-costapproaches to keeping participantsengaged.In conclusion, among these low-

income inner-city Latino adults with di-abetes, both PL-led and CHW-led DSMSmodels resulted in maintenance of gainsachieved in an evidence-based DSMEprogram. The sustained improvementsin HbA1c that we observed are equiva-lent to those achieved in more resource-intensive health professional–led caremanagement programs. This is encour-aging news to health care centers in low-resource settings facing significantfinancial constraints. It suggests that topromote long-term ongoing diabetesself-management efforts, we may needonly to look toward individuals in ourown communities as valuable sourcesfor self-management support.

Acknowledgments. The authors thank thevery dedicated CHWs and PLs who participatedin this intervention.Funding. This research was supported by aPeers for Progress grant from the AmericanAssociation of Family Physicians Foundation (grantP30DK092926) (Michigan Center for DiabetesTranslational Research), by the National Instituteof Diabetes and Digestive and Kidney (NIDDK)Diseases (grant R18DK0785501A1), and by theCenters for Disease Control and Prevention(cooperative agreement no. U50/CCU417409).The funding sources had no role in the studydesign; data collection; administration of theinterventions; analysis, interpretation, or report-ing of data; or decision to submit the findings forpublication.Duality of Interest. No potential conflicts ofinterest relevant to this article were reported.Author Contributions. T.S.T. and M.H. con-tributed to study conception and design, dataanalysis and interpretation, and manuscript

preparation and were the principal investiga-tors. M.F. contributed to study conception anddesign, data interpretation, and manuscriptpreparation. B.S. contributed to study concep-tion and design; data collection, analysis, andinterpretation; and manuscript preparation.G.Pi. contributed to study conception and design,data analysis and interpretation, and manu-script preparation. G.Pa. contributed to studyconception and design, data collection, studyimplementation, and manuscript preparation.M.S.S. and E.C.K. contributed to study design,data interpretation, and manuscript prepara-tion. T.T. and M.H. are the guarantors of thiswork and, as such, had full access to all thedata in the study and take responsibility forthe integrity of the data and the accuracy ofthe data analysis.Prior Presentation. This study has beenaccepted as a symposium presentation for the74th Scientific Sessions of the American DiabetesAssociation, San Francisco, CA, 13–17 June 2014.

References1. Haas L, Maryniuk M, Beck J, et al.; 2012Standards Revision Task Force. National stand-ards for diabetes self-management educationand support. Diabetes Care 2012;35:2393–24012. Gary TL, Genkinger JM, Guallar E, Peyrot M,Brancati FL. Meta-analysis of randomized edu-cational and behavioral interventions in type 2diabetes. Diabetes Educ 2003;29:488–5013. Steed L, Cooke D, Newman S. A systematicreview of psychosocial outcomes following ed-ucation, self-management and psychological in-terventions in diabetes mellitus. Patient EducCouns 2003;51:5–154. Norris SL, Engelgau MM, Narayan KMV. Ef-fectiveness of self-management training in type2 diabetes: a systematic review of randomizedcontrolled trials. Diabetes Care 2001;24:561–5875. Norris SL, Lau J, Smith SJ, Schmid CH,Engelgau MM. Self-management education foradults with type 2 diabetes: a meta-analysis ofthe effect on glycemic control. Diabetes Care2002;25:1159–11716. Heisler M, Spencer M, Forman J, et al. Par-ticipants’ assessments of the effects of a com-munity health worker intervention on theirdiabetes self-management and interactionswith healthcare providers. Am J Prev Med2009;37(Suppl. 1):S270–S2797. Tang TS, Funnell MM, Noorulla S, Oh M,Brown MB. Sustaining short-term improve-ments over the long-term: results from a 2-yeardiabetes self-management support (DSMS) in-tervention. Diabetes Res Clin Pract 2012;95:85–928. Trento M, Passera P, Bajardi M, et al. Life-style intervention by group care prevents dete-rioration of type II diabetes: a 4-yearrandomized controlled clinical trial. Diabetolo-gia 2002;45:1231–12399. Glasgow RE, Kurz D, King D, et al. Twelve-month outcomes of an Internet-based diabetesself-management support program. PatientEduc Couns 2012;87:81–9210. Lorig K, Ritter PL, Villa F, Piette JD. Spanishdiabetes self-management with and without

automated telephone reinforcement: two ran-domized trials. Diabetes Care 2008;31:408–41411. Siminerio L, Ruppert KM, Gabbay RA. Whocan provide diabetes self-management supportin primary care? Findings from a randomizedcontrolled trial. Diabetes Educ 2013;39:705–71312. Feathers JT, Kieffer EC, Palmisano G, et al.The development, implementation, and processevaluation of the REACH Detroit Partnership’sDiabetes Lifestyle Intervention. Diabetes Educ2007;33:509–52013. Spencer MS, Rosland AM, Kieffer EC, et al.Effectiveness of a community health worker in-tervention among African American and Latinoadults with type 2 diabetes: a randomized con-trolled trial. Am J Public Health 2011;101:2253–226014. Two Feathers J, Kieffer EC, Palmisano G,et al. Racial and Ethnic Approaches to Commu-nity Health (REACH) Detroit partnership: im-proving diabetes-related outcomes amongAfrican American and Latino adults. Am J PublicHealth 2005;95:1552–156015. Israel BA, Eng E, Shultz AJ, Parker EA.Meth-ods in Community-Based Participatory Researchfor Health. San Francisco, Jossey-Bass, 200516. U.S. Census Bureau. American Factfinder [in-ternet], 2010. Available from http://factfinder2.census.gov/faces/nav/jsf/index.xhtml. Accessed30 July 201317. Tang TS, Funnell MM, Gillard M, NwankwoR, Heisler M. Training peers to provide ongoingdiabetes self-management support (DSMS): re-sults from a pilot study. Patient Educ Couns2011;85:160–16818. Anderson RM, Funnell MM, Arnold MS. Us-ing the empowerment approach to help pa-tients change behavior. In Practical Psychologyfor Diabetes Clinicians. 2nd ed. Anderson B,Rubin R, Eds. Alexandria, VA, American DiabetesAssociation, 2002, p. 3–1219. Emmons KM, Rollnick S. Motivational inter-viewing in health care settings. Opportunitiesand limitations. Am J Prev Med 2001;20:68–7420. Funnell MM, Anderson RM. Behaviorchange strategies. In Medical Management ofType 2 Diabetes. 5th ed. Burant CF, Ed. Alexandria,VA, American Diabetes Association, 2004,p. 124–12921. ArsieMP,Marchioro L, Lapolla A, et al. Eval-uation of diagnostic reliability of DCA 2000 forrapid and simple monitoring of HbA1c. Acta Di-abetol 2000;37:1–722. Rogers EJ, Misner L, Ockene IS, Nicolosi RJ.Evaluation of seven Cholestech L.D.X analyzersfor total cholesterol determinations. Clin Chem1993;39:860–86423. Polonsky WH, Fisher L, Earles J, et al. As-sessing psychosocial distress in diabetes: devel-opment of the diabetes distress scale. DiabetesCare 2005;28:626–63124. Barrera M Jr, Glasgow RE, McKay HG, BolesSM, Feil EG. Do Internet-based support inter-ventions change perceptions of social support?:An experimental trial of approaches for sup-porting diabetes self-management. Am J Com-munity Psychol 2002;30:637–65425. UK Prospective Diabetes Study (UKPDS)Group. Intensive blood-glucose control with sul-phonylureas or insulin compared with conven-tional treatment and risk of complications in

care.diabetesjournals.org Tang and Associates 9

©

patients with type 2 diabetes (UKPDS 33). Lan-cet 1998;352:837–85326. Diggle PJ, Heagerty P, Liang K, Zeger SL.Analysis of Longitudinal Data. New York, OxfordUniversity Press, 200227. Little RJA, Rubin DB. Statistical AnalysisWith Missing Data. 2nd ed. New York, JohnWiley & Sons, 200228. Rubin DB. Multiple Imputation for Nonre-sponse in Surveys. New York, John Wiley &Sons, 198729. Schultz KF, Altman DG, Moher D; CONSORTGroup. CONSORT 2010 statement: updatedguidelines for reporting parallel group rando-mised trials. Ann Int Med 2010;152:726–73230. Baig AA, Locklin CA, Wilkes AE, et al. “OneCan Learn From Other People’s Experiences”: La-tino adults’ preferences for peer-based diabetesinterventions. Diabetes Educ 2012;38:733–74131. Rosal MC, Ockene IS, Restrepo A, et al. Ran-domized trial of a literacy-sensitive, culturallytailored diabetes self-management interven-tion for low-income latinos: latinos en control.Diabetes Care 2011;34:838–844

32. Philis-Tsimikas A, Fortmann A, Lleva-OcanaL, Walker C, Gallo LC. Peer-led diabetes educa-tion programs in high-risk Mexican Americansimprove glycemic control compared with stan-dard approaches: a Project Dulce promotorarandomized trial. Diabetes Care 2011;34:1926–193133. Brown SA, Blozis SA, Kouzekanani K, GarciaAA, Winchell M, Hanis CL. Dosage effects of di-abetes self-management education for MexicanAmericans: the Starr County Border Health Ini-tiative. Diabetes Care 2005;28:527–53234. Rothschild SK, Martin MA, Swider SM, et al.The Mexican-American Trial of CommunityHealth workers (MATCH): design and baselinecharacteristics of a randomized controlled trialtesting a culturally tailored community diabetesself-management intervention. Contemp ClinTrials 2012;33:369–37735. Brown SA, Garcia AA, Kouzekanani K, HanisCL. Culturally competent diabetes self-manage-ment education for Mexican Americans: theStarr County border health initiative. DiabetesCare 2002;25:259–268

36. Sixta CS, Ostwald S. Texas-Mexico borderintervention by promotores for patients withtype 2 diabetes. Diabetes Educ 2008;34:299–30937. Shah M, Kaselitz E, Heisler M. The role ofcommunity health workers in diabetes: updateon current literature. Curr Diab Rep 2013;13:163–17138. Fedder DO, Chang RJ, Curry S, Nichols G.The effectiveness of a community health workeroutreach program on healthcare utilization ofwest Baltimore City Medicaid patients with di-abetes, with or without hypertension. Ethn Dis2003;13:22–2739. LawlorMS, Blackwell CS, Isom SP, et al. Costof a group translation of the Diabetes Preven-tion Program: Healthy Living Partnerships toPrevent Diabetes. Am J Prev Med 2013;44(Suppl. 4):S381–S38940. Brown HS, Wilson KJ, Pagan JA, et al. Cost-effectiveness analysis of a community healthworker intervention for low-income Hispanicadults with diabetes. Prev Chronic Dis 2012;9:120074

10 Peer Compared with Community Health Worker Follow-up Diabetes Care

©