DOI: 10.1542/peds.2009-0233 2009;124;1579-1586; originally published online Nov 23, 2009; Pediatrics

Linda Y. Fu, Nuala Cowan, Rosie McLaren, Ryan Engstrom and Stephen J. Teach With Medicaid

Spatial Accessibility to Providers and Vaccination Compliance Among Children

http://www.pediatrics.org/cgi/content/full/124/6/1579located on the World Wide Web at:

The online version of this article, along with updated information and services, is

rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275. Grove Village, Illinois, 60007. Copyright © 2009 by the American Academy of Pediatrics. All and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elkpublication, it has been published continuously since 1948. PEDIATRICS is owned, published, PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

. Provided by NYS Dept Of Health on February 23, 2010 www.pediatrics.orgDownloaded from

Spatial Accessibility to Providers and VaccinationCompliance Among Children With Medicaid

WHAT’S KNOWN ON THIS SUBJECT: The geospatial availability ofprimary care is a known determinant of health outcomes such asasthma and mental health. State-level data have suggested this

that may also be true for vaccinations.

WHAT THIS STUDY ADDS: Spatial accessibility to vaccinationproviders is associated with vaccination compliance amongchildren from low-income families who live in Washington, DC.

abstractOBJECTIVE: We examined the relationship between spatial accessibil-ity to pediatric immunization providers and vaccination compliance ina low-income, urban population of children.

METHODS: In 2007, we accessed the Washington, DC, ImmunizationInformation System (IIS) to collect data on the immunization statusesand residential addresses of children who were aged 19 to 35 monthsand had Medicaid insurance. In addition, we calculated each child’sspatial accessibility to pediatric vaccination providers by assessingthe provider-to-population ratio at each residential address. Spatialaccessibility was divided into tertiles (low, medium, and high) of ac-cess. The relationship between spatial accessibility to providers andvaccination compliance was examined by using logistic regressionanalysis adjusting for age, type of vaccination provider, and enrollmentin child care status.

RESULTS: Overall for our cohort of 4195 children, 80.5% of the childrenwere up-to-date with vaccinations. Vaccination coverage ranged from61.6% to 100% (median: 79.2%) among different neighborhoods. Havingthe highest level of access to pediatric vaccination providers was as-sociated with 36% higher odds of being up-to-date as compared withhaving the lowest level of access. The middle tertile of access wasassociated with 25% higher odds of being up-to-date.

CONCLUSIONS: Within our low-income, urban population, childrenwith higher spatial accessibility to pediatric vaccination providerswere more likely to be up-to-date with vaccinations. This associationmay guide future studies and efforts to ensure adequate immunizationcoverage for children regardless of where they live. Pediatrics 2009;124:1579–1586

AUTHORS: Linda Y. Fu, MD, MS,a,b Nuala Cowan, MA,c

Rosie McLaren, MS,d Ryan Engstrom, PhD,c and StephenJ. Teach, MD, MPHb

aGoldberg Center for Community Pediatric Health and bCenterfor Clinical and Community Research, Children’s NationalMedical Center, Washington, DC; cDepartment of Geography,George Washington University, Washington, DC; and dNationalCenter for Immunization and Respiratory Diseases, Centers forDisease Control and Prevention, Atlanta, Georgia

KEY WORDSgeography, immunizations, vaccines

ABBREVIATIONSIIS—Immunization Information SystemPCV7—7-valent pneumococcal conjugate vaccineFTE—full-time equivalent

www.pediatrics.org/cgi/doi/10.1542/peds.2009-0233

doi:10.1542/peds.2009-0233

Accepted for publication Jul 9, 2009

Address correspondence to Linda Y. Fu, MD, MS, Goldberg Centerfor Community Pediatric Health, Children’s National MedicalCenter, 111 Michigan Ave, NW, Washington, DC 20010. E-mail:lfu@cnmc.org

PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).

Copyright © 2009 by the American Academy of Pediatrics

FINANCIAL DISCLOSURE: The authors have indicated they haveno financial relationships relevant to this article to disclose.

ARTICLES

PEDIATRICS Volume 124, Number 6, December 2009 1579. Provided by NYS Dept Of Health on February 23, 2010 www.pediatrics.orgDownloaded from

Despite improvements in overall child-hood immunization rates in the UnitedStates in the past decade, gains in cer-tain populations have been limited. Un-derimmunized children aremore likelyto be poor,1,2 from single-parent house-holds,1 black,2–5 and from inner cit-ies.6–8 These demographic characteris-tics also describe children who areless likely to have a regular source ofprimary care.9 The geospatial avail-ability of primary care providers is aknown determinant of health out-comes, especially for preventive ser-vices.9–12 It is likely that lower spatialproximity to primary care providershas a disproportionate affect on at-risk, lower income populations withless access to private transportation.In surveys of at-risk populations, par-ents have cited transportation difficul-ties as a significant obstacle to havingtheir children fully immunized.4,13,14

Conversely, having a provider nearbymay increase adherence with medicalrecommendations. In a study of wel-fare recipients, spatial proximity tosupport services increased their use.15

Because children from low-incomefamilies are more likely to have par-ents with transportation difficultiesand because they are also at higherrisk for underimmunization, it is espe-cially important to understand howspatial accessibility to primary carepediatric providers affects this popu-lation’s vaccination rates. The aim ofthis study was to determine whetherspatial accessibility to pediatric immuni-zation providers is associated with vac-cination compliance among childrenfrom low-income families inWashington,DC, as determined by Medicaid insur-ance enrollment.

METHODS

Vaccination Coverage

Childhood immunization rates wereassessed by using the DC Departmentof Health Immunization Information

System (IIS), a population-based, com-puterized registry. The IIS consolidatesvaccination data from (1) health careproviders, including all providers whovaccinate children with Medicaid in-surance; (2) all licensed child care pro-viders; and (3) all DC public schools.Submission to the IIS is a requirementfor Medicaid reimbursement underthe Vaccines for Children Program. Weextracted vaccination data on childrenwho were aged 19 to 35 months of ageas of March 1, 2007. Children were in-cluded in this study when they residedin DC, had Medicaid insurance, and, inaccordance with the Centers for Dis-ease Control and Prevention’s methodfor assessing IIS participation rates,had at least 2 immunizations capturedin the IIS.

Immunizations for each child were as-sessed for completion of the 4:3:1:3:3:1combined series (�4 doses of vaccinecontaining diphtheria and tetanus tox-oids with or without pertussis; �3doses of poliovirus vaccine;�1 dose ofmeasles, mumps, and rubella vaccine;�3 doses of Haemophilus influenzaetype b vaccine;�3 doses of hepatitis Bvaccine; and�1 dose of varicella vac-cine) and the 4:3:1:3:3:1:3 series (the4:3:1:3:3:1 series plus �3 doses of7-valent pneumococcal conjugate vac-cine [PCV7]). Two outcomes (with andwithout PCV7) were chosen becausevaccination with PCV7 for some of thechildren in our cohort may have beenreduced by the national shortage thatoccurred from February to September2004.16 The 4:3:1:3:3:1 and 4:3:1:3:3:1:3series were 2 of the key series as-sessed by the National ImmunizationSurvey, a population-based surveysponsored by the Centers for DiseaseControl and Prevention in 2007 for chil-dren aged 19 to 35 months.17

In addition to vaccination information,the following variables were extractedfrom the IIS: age, gender, home ad-dress, type of provider who adminis-

tered the child’s last listed vaccine(public health department, federallyqualified health center [ie, center re-ceiving federal funding to care for un-derserved populations], other publichealth provider, private practice, pri-vate hospital, or other private entity)and whether the child attended a li-censed child care center. It is compul-sory for licensed child care centers inDC to ensure that their attendees arecompliant with vaccinations and tosubmit documentation to the IIS. Insti-tutional review boards at the DC De-partment of Health and at Children’sNational Medical Center approved thisstudy.

Using ArcGIS 9.2 (ESRI Inc, Redlands,CA), we created a map of residentialaddresses for the children in thestudy, geocoded to latitude and longi-tude. For addresses that we were un-able to geocode with the ArcGIS auto-mated system, we manually searchedandmatched using GoogleMaps (www.maps.google.com) when possible. Thismethod of interactive geocoding in-creases match rates and decreasesthe potential for selection bias as a re-sult of incomplete geocoding.18 We in-cluded in our analysis all addresseswith an ArcGIS minimum match scoreof 70. A perfect match yields a score of100. The default minimummatch scoresetting in ArcGIS is 60, but we set ahigher minimum for greater confi-dence in address locations.

We also created a density map of ourstudy population by using kernel esti-mation. Kernel density calculates thenumber of point features per areawithin each output raster cell. The out-put cell size is chosen by the analyst,and the final product can be repre-sented as a smooth contour map.19,20

Conceptually, a smoothly curved sur-face is fitted over each point. The sur-face value is highest at the location ofthe point and diminishes with increas-ing distance from the point, reaching 0

1580 FU et al. Provided by NYS Dept Of Health on February 23, 2010 www.pediatrics.orgDownloaded from

at the search radius distance from thepoint. The volume under the surfaceequals the population field value forthe point. The density of children wascalculated within 0.1 mi2 cells andsmoothed with a Gaussian kernel den-sity function to a distance of 1 mi. Tocreate a ratio raster map of percent-age of children who were fully immu-nized, we created a second densitylayer of all children in our study whose4:3:1:3:3:1 vaccine series was com-plete, again using 0.1-mi cells and a1-mi smoothing radius. This layer ofchildren who had complete vaccina-tion was divided by the layer of the en-tire study population to obtain a layerof percentage immunized at all pointsin DC. We repeated this process for 4:3:1:3:3:1:3 vaccination completion statusto create a second ratio raster map.

Pediatric Vaccination ProviderSpatial Accessibility

The methods for creating the providerspatial accessibility map have beendescribed previously.12,19,20 Briefly, in2004, primary care pediatric providerlocations in DC were compiled frommultiple sources, including the Ameri-can Medical Association membershiproster and practitioner lists from allDC Medicaid managed care organiza-tions. Recordswere limited to pediatri-cians, family practitioners, and gen-eral practitioners who listed theirprimary activity as “direct patientcare” or “resident.” Each point sourceof care was contacted to determinethe number and type of practitionerswho worked at that address, as well asthe number of hours per week thateach practitioner devoted to primarycare services. Using guidelines fromthe American Academy of Pediatrics,we made adjustments for the type oftraining (pediatrics [weight: 1.00] ver-sus family practice [weight: 0.75]),training level (attending [weight: 1.00]versus resident/fellow [weight: 0.35]),and type of provider (physician [weight:

1.00], nurse practitioner [weight: 0.75],or physician assistant [weight: 0.75]).21

In addition we included 5 DC Depart-ment of Health immunization-only cen-ters which were staffed by nurses whogave shots full-time. After comparingthe average number of vaccines ad-ministered per nurse at immunization-only centers with the number of vac-cines given per primary care providerat various health centers in DC, eachnurse at each immunization-only cen-ter was assigned a weight of 1.00 forthe hours of operation. A total numberof providers who were capable of or-dering/delivering vaccines was gener-ated for each point source of care andexpressed as a number of full-time

equivalent (FTE) providers on the basisof a 40-hour work week treating pedi-atric patients.

Spatial accessibility to vaccinationproviders was calculated as the ratioof pediatric provider FTEs to pediatricpopulation.19,20 To do this, we created 2density maps of DC. The provider-density layer was created using theprovider (weighted by FTE) practice lo-cations smoothed with a Gaussian ker-nel density function of 3 mi. The use ofa 3-mi radius was based on a surveythat determined the distance thatblack city residents typically travel tocommunity clinics.12,22 The population-density layer was created from block

FIGURE 1Household locations in Washington, DC, for the study population (N� 5209).

ARTICLES

PEDIATRICS Volume 124, Number 6, December 2009 1581. Provided by NYS Dept Of Health on February 23, 2010 www.pediatrics.orgDownloaded from

group centroids weighted by the num-ber of children who were younger than18 years and living in the block group(using 2000 US census data) andsmoothed over a 1-mi radius.23,24 We di-vided the provider-density layer by thepopulation-density layer to create amap of provider-to-population ratios.From this, we determined the ratio atthe residential address of each childincluded in this study and defined thatratio as the child’s spatial accessibilityto pediatric vaccination providers.

Statistical Analysis

Means and frequencies were deter-mined for sociodemographic andclinical characteristics of the studypopulation. The relationship of each in-dependent variable on the 2 outcomes,completion of the 4:3:1:3:3:1 and 4:3:1:3:3:1:3 vaccination series, was exam-ined by using the �2 test or unpaired ttest. A crude odds ratio with 95%confidence interval was calculatedusing logistic regression. To examinewhether vaccination coverage variedamong DC neighborhoods, we calcu-lated vaccination rates for each of the131 neighborhoods in DC as defined bythe DC government. Themedian and in-terquartile range for vaccination rateby neighborhood was determined forthe 80 neighborhoods with �10 studychildren.

To examine vaccine completion ratesas a function of spatial accessibility toproviders, we created 2 multivariatelogistic regression models. The spatialaccessibility measure was divided intotertiles for ease of interpretation ofthe results (low, medium, and high ac-cessibility). Because we were examin-ing spatial accessibility for DC, a juris-diction of only 61 mi2, we believed thatgreater partitioning of the spatial ac-cessibility variable into more catego-ries (or examining it as a continuousvariable) would be misleading. Theoutcomes of interest for the 2 models

were completion of the 4:3:1:3:3:1 and4:3:1:3:3:1:3 vaccination series, respec-tively. Each model was adjusted forage, type of vaccination provider, andenrollment in child care status. Thesevariables were checked for correla-tion with each other and the spatialaccessibility variable by using Pearsoncorrelation coefficients, which rangedfrom 0 to 0.27 for all pairings. Be-cause of a relatively high percentageof missing values for the variable typeof vaccination provider (3.6%), we cre-ated a dummy variable for both mod-els for the missing values of this vari-able. The Hosmer and Lemeshowgoodness-of-fit test was used to deter-

mine the fit of the models to the data.All analyses were conducted by usingSAS 9.2 (SAS Institute, Inc, Cary, NC).Test levels for significance wereP � .05.

RESULTS

As of March 1, 2007, 5310 children 19 to35 months of age with Medicaid insur-ance had�2 vaccinations listed in theDC IIS. Of these, 101 (1.9%) were ex-cluded from analysis because they hadaddresses of record that we were un-able to convert to latitudinal and longi-tudinal positions for 1 of the followingreasons: mapped to a nonresidentialsite, contained a nonexistent street

FIGURE 2Percentage of the study population who had completed the 4:3:1:3:3:1 vaccination series inWashington, DC.

1582 FU et al. Provided by NYS Dept Of Health on February 23, 2010 www.pediatrics.orgDownloaded from

number or name, or lacked criticalstreet information so that the addresscould not be identified as a single, dis-crete location. The rest of the sample(n � 5209) had addresses that wewere able to convert to ordinal posi-tions and were included in the analysis(Fig 1). As is seen in Fig 1, the vast ma-jority of our study population residedin the northeast and southeast quad-rants of DC.

Overall, for our cohort, 4195 (80.5%)children were complete for the 4:3:1:3:3:1 vaccination series and 3508(67.3%) children were complete forthe 4:3:1:3:3:1:3 series. Vaccinationcoverage rates varied greatly acrossthe city (Fig 2). Among neighborhoodswith �10 study children, vaccinecompletion rates for the 4:3:1:3:3:1series ranged from 61.6% to 100%(median: 79.2% [interquartile range:75.0%–84.4%]) and 41.7% to 94.5%(median: 76.9% [interquartile range:70.0%–81.7%]) for the 4:3:1:3:3:1:3series.

The density of pediatric providers ascompared with pediatric populationalso varied across the city (Fig 3).Almost uniformly, the more socio-economically advantaged northwestquadrant had the highest category ofconcentration of providers (�90 pro-viders per 100 000 children). By con-trast, coverage of the more socioeco-nomically disadvantaged southeastquadrant25 dipped to �30 providersper 100 000 children in 1 large area.

Before adjustment, completion of boththe 4:3:1:3:3:1 and 4:3:1:3:3:1:3 vaccina-tion series was related to multiple fac-tors examined (Tables 1 and 2). As ex-pected, older age was associated withvaccination completion. Each 1-monthincrement in age was associated withan 8% increase in odds of 4:3:1:3:3:1vaccination series completion and a6% increase in odds of 4:3:1:3:3:1:3completion. Child care attendance wasstrongly associated with compliance

with vaccinations: those who attendedchild care had�8 times higher odds of4:3:1:3:3:1 series completion and al-most 4 times the odds of 4:3:1:3:3:1:3series completion. The type of vaccina-tion provider was also significant. Re-ceiving immunizations from a federallyqualified health center, another type ofpublic health provider, a private hospi-tal, or other private entity was associ-ated with vaccination completion,compared with receiving vaccinationsfrom a private practitioner, althoughprivate practitioners as a group caredfor almost half (46.0%) of the studypopulation. Finally, higher spatial ac-cessibility to vaccination providerswas associated with vaccination com-pletion. Children who were in the high-

est tertile of spatial accessibility had1.42 higher odds of being 4:3:1:3:3:1 se-ries complete and 1.33 higher odds ofbeing 4:3:1:3:3:1:3 series complete ascompared with the lowest tertile. Alsoincreased odds of vaccination comple-tion were associated with the middleversus the lowest tertile of spatial ac-cessibility (1.28 and 1.18 for the 2 re-spective series).

The relationship between spatial ac-cessibility to pediatric vaccinationproviders and vaccination compliancewas additionally examined after ad-justment for age, type of vaccinationprovider, and enrollment in child carestatus (Table 3). After adjustment, hav-ing the highest access to providers

FIGURE 3Distribution of pediatric vaccination providers (in terms of FTEs) in Washington, DC.

ARTICLES

PEDIATRICS Volume 124, Number 6, December 2009 1583. Provided by NYS Dept Of Health on February 23, 2010 www.pediatrics.orgDownloaded from

was associated with 1.36 higher oddsof being 4:3:1:3:3:1 series complete and1.23 higher odds of being 4:3:1:3:3:1:3series complete. The middle tertile of

access was associated with 1.25 and1.11 higher odds of being complete forthe 4:3:1:3:3:1 and 4:3:1:3:3:1:3 series,respectively, although the relationship

was significant only for the 4:3:1:3:3:1series.

DISCUSSION

We found that with increasing spatialaccessibility, the odds of vaccinationcompletion are also increased even af-ter adjustment for age, type of vaccina-tion provider, and enrollment in childcare. This study enhances the growingevidence that geographic availabilityof providers is important for a widerange of conditions best managed byiterative, longitudinal health care, in-cluding asthma,12 prenatal care,26 andmental health.15

Immunization rates for our studypopulation for the 4:3:1:3:3:1 and 4:3:1:3:3:1:3 series (80.5% and 67.3%, re-spectively) were similar to butslightly lower than the 2007 coveragerates in DC reported by the NationalImmunization Survey (81.6% and77.3%, respectively).17 We expectedcoverage rates for children in ourstudy to be slightly lower, becausewe included only children from lowerincome families.

In our study, some neighborhoods hada much lower rate of vaccination com-pliance than others; vaccination com-pliance rates varied 38.4% amongneighborhoods for the 4:3:1:3:3:1 se-ries and 52.8% for the 4:3:1:3:3:1:3 se-ries. Characteristics such as predomi-nant languages spoken and culturalnorms may differ between people who

TABLE 1 Characteristics of Study Population as Related to Completion of the 4:3:1:3:3:1 VaccinationSeries (N� 5209)

Variable Series Incomplete(n� 1014)

Series Complete(n� 4195)

Unadjusted OR(95% CI) for

Series Completion

Age, mean� SD, mo 25.7� 5.0 27.6� 4.8 1.08 (1.07–1.10)a

Gender, %Female 48.7 49.4 1.03 (0.90–1.18)Male 51.3 50.6 ReferenceType of immunization provider, %b

Private hospital or other privateentity

13.9 20.4 1.80 (1.47–2.21)

DC Department of Healthvaccination-only clinic

3.9 1.7 0.55 (0.37–0.83)

Federally qualified health center 19.0 25.0 1.61 (1.34–1.94)Other public health provider 9.0 8.8 1.21 (0.94–1.56)Private practice 54.2 44.1 ReferenceEnrolled in licensed child care, %Yes 2.5 17.0 8.06 (5.41–12.05)No 97.5 83.0 ReferenceTertile of spatial accessibility ofvaccination providers, %Highest 29.3 34.4 1.42 (1.20–1.68)Middle 31.3 33.1 1.28 (1.09–1.51)Lowest 39.4 32.5 Reference

OR indicates odds ratio; CI, confidence interval; SD, standard deviation.a Odds reported for each 1-month increase in age.b Missing 187 responses.

TABLE 2 Characteristics of Study Population as Related to Completion of the 4:3:1:3:3:1:3Vaccination Series (N� 5209)

Variable Series Incomplete(n� 1153)

Series Complete(n� 4056)

Unadjusted OR(95% CI) for

Series Completion

Age, mean� SD, mo 26.1� 5.1 27.7� 4.8 1.06 (1.05–1.07)a

Gender, %Female 49.0 49.3 1.01 (0.89–1.16)Male 51.0 50.7 ReferenceType of immunization provider, %b

Private hospital or other privateentity

13.9 20.4 2.01 (1.65–2.45)

DC Department of Healthvaccination-only clinic

3.9 1.7 0.49 (0.33–0.72)

Federally qualified health center 19.0 25.0 1.81 (1.52–2.17)Other public health provider 9.0 8.8 1.35 (1.06–1.73)Private practice 54.2 44.1 ReferenceEnrolled in licensed child care, %Yes 5.1 16.7 3.72 (2.82–4.90)No 94.9 83.3 ReferenceTertile of spatial accessibility ofvaccination providers, %Highest 29.9 34.4 1.33 (1.13–1.56)Middle 32.2 32.9 1.18 (1.01–1.39)Lowest 37.9 32.7 Reference

OR indicates odds ratio; CI, confidence interval; SD, standard deviation.a Odds reported for each 1-month increase in age.b Missing 187 responses.

TABLE 3 Vaccine Series Completion asRelated to Tertile of SpatialAccessibility of VaccinationProviders

Tertile ofSpatial

Accessibility ofVaccinationProviders

Adjusted OR(95% CI) for4:3:1:3:3:1Series

Completiona

Adjusted OR(95% CI) for4:3:1:3:3:1:3Series

Completiona

Highest 1.36 (1.14–1.62) 1.23 (1.04–1.45)Middle 1.25 (1.05–1.49) 1.11 (0.94–1.31)Lowest Reference Reference

Odds ratios (ORs) are adjusted for age, type of vaccinationprovider, and enrollment in child care status. CI indicatesconfidence interval.

1584 FU et al. Provided by NYS Dept Of Health on February 23, 2010 www.pediatrics.orgDownloaded from

live in adjacent but distinct neighbor-hoods. The ability to identify at theneighborhood level areas with low vac-cination coverage may facilitate tailor-ing interventions to the specific needsof a neighborhood. It may also guidegovernmental allocation of limited re-sources to neighborhoods where im-munization improvement is neededmost.

The results of our study may enhanceunderstanding of a study by LeBaron etal,27 who found that, overall, a higherratio of pediatricians to infants in astate was associated with higher statevaccination coverage rates; however,DC had the highest density of pediatri-cians in the nation and a vaccinationcoverage rate below the median. Apossible explanation is that there isgreater socioeconomic heterogeneityin DC than in other states. DC hasmorechildren living in poverty than the na-tional average, but the median incomeis also higher than average.28 In ourstudy, whichwas limited to low-incomefamilies, pediatric provider distribu-tion was associated with vaccinationcoverage rates.

Our results contrast with previousfindings of Baumgardner et al,29 whofound no association in a population of151 predominantly low-income chil-dren between travel distance to amed-ical clinic and vaccination compliance.The contrast in findings might be ex-plained by differences in methods. Ourstudy looked at spatial accessibility toall vaccination providers in DC, ratherthan just to 1 site. This may be an es-sential difference if families with sig-nificant transportation barriers tendto choose health care providers who

are closest to their homes, because itis unknown whether the clinic in thestudy by Baumgardner et al was thenearest center for children in thatstudy. Families who were involved inthe study by Baumgardner et al mayhave bypassed closer providers be-cause of language barriers: 85% oftheir clinic population was Hispanic.Our study was likely composed pre-dominately of black children, becauseour populationwas limited toMedicaidrecipients and 97% of all children whoreceived welfare in DC in 2007 wereblack.30

Characteristics of a vaccination pro-vider in addition to geographic proxim-ity affect access to care and vaccina-tion compliance. Access to a source ofpediatric care may be reduced by lan-guage barriers, difficulties making ap-pointments,4,14 long waiting times,14

and lack of after-hours services.10 Inaddition, the provider’s use of remind-er/recall systems and other immuniza-tion best practices influences the like-lihood that vaccinations are deliveredappropriately.31 In our study, therewasan association between type of vacci-nation provider and vaccination com-pliance. If these nonspatial factors re-lated to access differ on the wholebetween various types of providers,then it might explain the relationshipbetween vaccination provider typeand immunization compliance that wefound.

We reported our analysis assumingthat families would travel up to 3mi forvaccines. Of note, the results of analy-sis that used a 2-mi radius were essen-tially the same (not reported). Ourstudy population was intentionally lim-

ited to urban, low-income children be-cause poverty and living in a city bothare risk factors for noncompliancewith vaccinations.1,2,6,8 Our results,therefore, should not be extrapolatedto higher income, suburban or ruralfamilies, whomay travel farther for pe-diatric health care services.

Because our data are cross-sectional, causality cannot be deter-mined. A next step would be to com-pare perceptions about the role oftransportation difficulties in vacci-nation noncompliance among resi-dents of neighborhoods with similarsocioeconomic and demographiccharacteristics but different spatialaccessibility to providers.

CONCLUSIONS

Our study demonstrated that in-creased spatial proximity to pediatricvaccination providers was positivelyassociated with vaccination comple-tion rates among Medicaid recipientsin Washington, DC. This associationmay guide future studies and efforts toensure immunization coverage of chil-dren regardless of where they live.Such interventions may need to findways to increase transportation op-tions for low-income families to reachproviders or to incentivize more pro-viders to practice in underservedareas.

ACKNOWLEDGMENTSThis project was supported in part bythe Academic Pediatric AssociationYoung Investigator Grant for Immuni-zation Research and by Robert WoodJohnson Foundation grant 043508.

REFERENCES

1. Bates AS, Fitzgerald JF, Dittus RS, WolinskyFD. Risk factors for underimmunization inpoor urban infants. JAMA. 1994;272(14):1105–1110

2. Hughart N, Strobino D, Holt E, et al. The re-lation of parent and provider characteris-

tics to vaccination status of children in pri-vate pract ices and managed careorganizations in Maryland. Med Care. 1999;37(1):44–55

3. Chu SY, Barker LE, Smith PJ. Racial/ethnicdisparities in preschool immunizations:

United States, 1996 –2001. Am J PublicHealth. 2004;94(6):973–977

4. Morrow AL, Rosenthal J, Lakkis HD, et al. Apopulation-based study of access to immu-nization among urban Virginia childrenserved by public, private, and military

ARTICLES

PEDIATRICS Volume 124, Number 6, December 2009 1585. Provided by NYS Dept Of Health on February 23, 2010 www.pediatrics.orgDownloaded from

health care systems. Pediatrics. 1998;101(2). Available at: www.pediatrics.org/cgi/content/full/101/2/e5

5. Shefer A, Smith PJ. Improving the immuni-zation and health status of children in theWomen, Infants, and Children (WIC) Pro-gram. J Health Care Poor Underserved.2004;15(1):127–140

6. Smith PJ, Hoaglin DC, BattagliaMP, KhareM,Barker LE. Statistical methodology of theNational Immunization Survey, 1994–2002.Vital Health Stat 2. 2005;(138):1–55

7. Centers for Disease Control and Prevention.National, state, and urban area vaccinationcoverage among children aged 19 –35months—United States, 2005. MMWR MorbMortal Wkly Rep. 2006;55(36):988–993

8. Zell ER, Dietz V, Stevenson J, Cochi S, BruceRH. Low vaccination levels of US preschooland school-age children. Retrospective as-sessments of vaccination coverage,1991–1992. JAMA. 1994;271(11):833–839

9. Starfield B, Shi L. The medical home, accessto care, and insurance: a review of evi-dence. Pediatrics. 2004;113(5 suppl):1493–1498

10. Forrest CB, Starfield B. Entry into primarycare and continuity: the effects of access.Am J Public Health. 1998;88(9):1330–1336

11. McLafferty S, Grady S. Prenatal care needand access: a GIS analysis. J Med Syst. 2004;28(3):321–333

12. Teach SJ, GuagliardoMF, Crain EF, et al. Spa-tial accessibility of primary care pediatricservices in an urban environment: associa-tion with asthma management and out-come. Pediatrics. 2006;117(4 pt 2):S78–S85

13. Thomas M, Kohli V, King D. Barriers to child-hood immunization: findings from a needsassessment study. Home Health Care ServQ. 2004;23(2):19–39

14. Lannon C, Brack V, Stuart J, et al. Whatmothers say about why poor children fallbehind on immunizations. A summary of fo-cus groups in North Carolina. Arch PediatrAdolesc Med. 1995;149(10):1070–1075

15. Allard SW, Tolman RM, Rosen D. Proximity toservice providers and service utilizationamong welfare recipients: the interactionof place and race. J Policy Anal Manage.2003;22(4):599–613

16. Centers for Disease Control and Prevention.Notice to readers: pneumococcal conjugatevaccine shortage resolved. MMWR MorbMortal Wkly Rep. 2004;53(36):851–852

17. Centers for Disease Control and Prevention.National, state, and local area vaccinationcoverage among children aged 19 –35months: United States, 2007. MMWR MorbMortal Wkly Rep. 2008;57(35):961–966

18. Gilboa SM, Mendola P, Olshan AF, et al. Com-parison of residential geocoding methodsin population-based study of air quality andbirth defects. Environ Res. 2006;101(2):256–262

19. Guagliardo MF. Spatial accessibility of pri-mary care: concepts, methods and chal-lenges. Int J Health Geogr. 2004;3(1):3

20. Guagliardo MF, Ronzio CR, Cheung I, ChackoE, Joseph JG. Physician accessibility: an ur-ban case study of pediatric providers.Health Place. 2004;10(3):273–283

21. American Academy of Pediatrics. PhysicianWorkforce: Ratios for Child Health. Elk GroveVillage, IL: American Academy of Pediatrics;1998

22. Shannon GW, Bashshur RL, Spurlock CW.The search for medical care: an explorationof urban black behavior. Int J Health Serv.1978;8(3):519–530

23. McLafferty S, Williamson D, McGuire P. Iden-tifying crime hot spots using kernel

smoothing in analyzing crime patterns. In:Goldsmith V, McGuire P, Mollenkopf J, RossT, eds. Analyzing Crime Patterns: Frontiersof Practice. Thousand Oaks, CA: Sage; 1999:77–85

24. Longley P, Goodchild M, Maguire D, Rhind D.Geographic Information Systems and Sci-ence. Chichester, NY: Wiley; 2001

25. Neighborhood Profiles. The Urban Instituteand Washington DC local initiatives supportcorporation, 2008. Available at: www.neighborhoodinfodc.org/nclusters/nclusters.html#ANCs. Accessed October 13, 2008

26. McLafferty S, Grady S. Immigration and geo-graphic access to prenatal clinics in Brook-lyn, NY: a geographic information systemsanalysis. Am J Public Health. 2005;95(4):638–640

27. LeBaron CW, Massoudi M, Stevenson J,Lyons B. Vaccination coverage and physi-cian distribution in the United States, 1997.Pediatrics. 2001;107(3). Available at: www.pediatrics.org/cgi/content/full/107/3/e31

28. US Census Bureau. 2007 American Commu-nity Survey. Available at: www.census.gov/acs. Accessed October 2, 2008

29. Baumgardner DJ, Halsmer SE, Steber DL,Shah DS, Mundt MP. Does proximity to clinicaffect immunization rates and blood pres-sure? Int J Psychiatry Med. 2006;36(2):199–209

30. DC Kids Count Collaborative. Every Kid Countsin the District of Columbia: 14th Annual FactBook. Washington, DC: DC Children’s TrustFund; 2007

31. Briss PA, Rodewald LE, Hinman AR, et al. Re-views of evidence regarding interventionsto improve vaccination coverage in chil-dren, adolescents, and adults. Task Forceon Community Preventive Services. Am JPrev Med. 2000;18(1 suppl):97–140

1586 FU et al. Provided by NYS Dept Of Health on February 23, 2010 www.pediatrics.orgDownloaded from

DOI: 10.1542/peds.2009-0233 2009;124;1579-1586; originally published online Nov 23, 2009; Pediatrics

Linda Y. Fu, Nuala Cowan, Rosie McLaren, Ryan Engstrom and Stephen J. Teach With Medicaid

Spatial Accessibility to Providers and Vaccination Compliance Among Children

& ServicesUpdated Information

http://www.pediatrics.org/cgi/content/full/124/6/1579including high-resolution figures, can be found at:

References

http://www.pediatrics.org/cgi/content/full/124/6/1579#BIBLat: This article cites 23 articles, 8 of which you can access for free

Subspecialty Collections

http://www.pediatrics.org/cgi/collection/office_practice Office Practice

following collection(s): This article, along with others on similar topics, appears in the

Permissions & Licensing

http://www.pediatrics.org/misc/Permissions.shtmltables) or in its entirety can be found online at: Information about reproducing this article in parts (figures,

Reprints http://www.pediatrics.org/misc/reprints.shtml

Information about ordering reprints can be found online:

. Provided by NYS Dept Of Health on February 23, 2010 www.pediatrics.orgDownloaded from