Veterans Affairs Primary Care Organizational Characteristics Associated With Better Diabetes Control

Objective: To examine organizational features of VeteransAffairs (VA) primary care programs hypothesized to be associatedwith better diabetes control, as indicated by hemoglobin A_1c (HbA_1c) levels.

Study Design: Cross-sectional cohort.

Methods: We established a cohort of 224 221 diabetic patientsusing the VA Diabetes Registry and Dataset and VA corporate databases.The 1999 VHA (Veterans Health Administration) Survey ofPrimary Care Practices results were combined with individualpatient data. A 2-level hierarchical model was used to determinethe relationship between organizational characteristics and HbA_1c levels in 177 clinics with 82 428 cohort members.

P

Results: The following attributes were associated with lower(better) HbA_1c and were statistically significant at < .05: greaterauthority to establish or implement clinical policies (lower by0.21%), greater staffing authority (0.28%), computerized diabetesreminders (0.17%), notifying all patients of their assigned provider(0.21%), hiring needed new staff during fiscal year 1999 (0.18%),having nurses that report only to the program (0.16%), and being alarge academic practice (0.27%). Associated with higher (worse)HbA_1c were programs reporting that patients almost always seetheir assigned provider (greater by 0.18%), having a qualityimprovement program involving all nurses without all physicians(0.38%), having general internal medicine physicians report onlyto the program (0.20%), and being located at an acute care hospital(0.20%).

Conclusion: Programs that are associated with better diabetescontrol simultaneously have teams that actively involve physiciansin quality improvement, use electronic health information systems,have authority to respond to staffing and programmatic issues, andengage patients in care.

(Am J Manag Care. 2005;11:225-237)

The quality and outcomes of healthcare in theUnited States are far from optimal.^1-4 The chroniccare model⁵ posits that optimum chronic illnessmanagement requires a healthcare system thatprovides self-management support to patients, decisionsupport, clinical information systems that support care,and effective connections with community resources.Such a system can improve interactions betweenproviders and patients and, in turn, enhance patient outcomes.The model proposes a basic road map for combiningthese elements to achieve optimal patientoutcomes.^5-12

Several features of the Veterans Health Administration(VHA) make it an ideal system in which to testfeatures of the chronic care model. First, the VHA operatesthe largest integrated healthcare delivery systemin the United States.¹³ Second, in the mid-to-late1990s, the VHA was transformed from a system with aninpatient specialty treatment focus to one that emphasizesprimary care and integrated practice teams.^14-16Third, the VHA's extensive electronic medical recordsystem facilitates care and provides the opportunity toexamine the relationship between the organization ofprimary care and chronic disease outcomes. Therefore,study of Veterans Affairs (VA) health systems offerslarge-scale sources of data for understanding managedcare delivery.^17-19

Diabetes mellitus is an excellent disease for testingthe chronic care model. Diabetes mellitus is prevalentin the United States: 12 million adults have diagnoseddiabetes mellitus, and another 6 million have undiagnoseddiabetes mellitus.²⁰ Persons with diabetes mellitusare at increased risk for macrovascular andmicrovascular complications that compromise patients'quality of life and result in extensive use of healthcareservices.^20-24 Given its prevalence, morbidity, and associatedhealth services use, diabetes mellitus is estimatedto cost the United States $132 billion annually.²⁵ Thesignificance of diabetes mellitus in the VHA mirrors orexceeds that in the rest of the United States.^26-34 Forexample, diabetes mellitus accounts for 25% of all pharmacycosts and 1.7 million inpatient bed days each yearin the VHA.³¹

The morbidity and costs of diabetes mellitus can bereduced by tight control of blood sugar.^35-38 However,adequate blood sugar control is often not achievedamong adults with diabetes mellitus.^39-41 Furthermore,the VHA External Peer Review Program and the NationalCommittee on Quality Assurance's HealthEmployer Data and Information Set (HEDIS) includeprovision of hemoglobin A_1c (HbA_1c) tests and control ofHbA_1c among performance measures.⁴²

The present study seeks to identify organizationalcharacteristics of primary care programs that are associatedwith glycemic control and, hence, a reduced riskof long-term diabetes complications. Although not allcharacteristics of the chronic care model were measured,we hypothesized that those programs that moreclosely approximated this type of optimal system wouldhave patients with better glycemic control.

METHODS

We studied the effect of primary care organizationduring fiscal year (FY) 1999, which covers October 1,1998 through September 30, 1999. In FY 1999, the VHAincluded 172 hospitals, more than 600 outpatient clinics,132 nursing homes, 40 domiciliaries, 73 comprehensivehome care programs, and 206 counselingcenters.¹⁶ That year, 3.7 million people received carethrough the VHA.⁴³ The institutional review board ofthe Durham Veterans Affairs Medical Center approvedthe study protocol.

Data Sources

1999 VHA Survey of Primary Care Practices.

Information on the organization of VHA primary careclinics was obtained from the 1999 VHA Survey ofPrimary Care Practices.^44,45 Between October andDecember 1999, the survey was sent to clinical leadersof primary care programs at individual VHA facilitieswith at least 4000 unique patients and at least 20 000outpatient visits in FY 1998. This included 170 VA medicalcenters and 65 community-based outpatient clinics.The objective was to evaluate the organizational featuresof primary care programs so as to elucidate therelationships among organizational features and clinicaland economic outcomes. There were 219 responses(160 medical centers and 59 community-based outpatientclinics), a response rate of 93%.^44,45

VA Diabetes Registry and Dataset.

pharmacy file

laboratory

file

vital signs file

The VA DiabetesRegistry and Dataset was used to establish the cohort ofdiabetic patients and obtain outcome data.⁴⁶ During FY1999, 503 371 patients were in the registry.⁴⁷ The registrycontains 3 files. The has informationon patients who have ever had prescriptions filledat a VA pharmacy for insulin, oral hypoglycemic agents,or blood glucose monitoring supplies. The contains test results directly related to diabetescare. The has information on patientheight, weight, body mass index, blood pressure, andreceipt of influenza and pneumonia vaccines.⁴⁶

Veterans Affairs Corporate Databases.

Data onpatient demographics, healthcare use, comorbidities,and vital status were obtained from multiple VA corporatedatabases housed at the VA Austin AutomationCenter, Austin, Tex. Most databases maintained at theVA Austin Automation Center result from downloadsfrom individual Veterans Information SystemTechnologies Architecture health information systemslocated at VHA facilities.^48,49 The VA Austin AutomationCenter has data on all VHA inpatient and outpatientencounters since 1980.^50,51 Vital status of veterans isrecorded in the Beneficiary Identification and RecordLocator System.⁵² Information on patient comorbiditiesused for risk adjustment is located in The Johns HopkinsAdjusted Clinical Groups Case-Mix System files.⁵³

Subjects

Using the VA Diabetes Registry and Dataset, we identifiedpatients who met the following inclusion criteriaduring FY 1999: (1) had at least 2 non-mental healthoutpatient visits with an associated diagnosis of diabetesmellitus, or any non-mental health inpatient dischargewith an associated diagnosis of diabetes mellitus; (2)had a prescription filled for insulin, an oral hypoglycemicagent, or blood glucose monitoring supplies(from the VA Diabetes Registry and Dataset pharmacyfile); and (3) had at least 1 outpatient visit to a VA primarycare clinic. Exclusion criteria were age youngerthan 18 years, switching primary VA facilities duringthe study period, any VA endocrinology visit during FY1999 through FY 2001, and pregnancy.

International

Classification of Diseases, Ninth Revision,

Clinical Modification

Encounter data were obtained from the inpatient bedsection and outpatient event files. Specific codes used to identify patientswith diabetes mellitus were obtained from the 2003 versionof the Clinical Classifications Software from theAgency for Healthcare Research and Quality.⁵⁴

Individuals were considered primary care patients ofthe VA facility where they made the greatest number ofprimary care visits during FY 1999. In case of a tie, thepatient was randomly assigned to one of the tied locations.Once patients' primary care clinics were determined,a random sample of 800 diabetic patientsmeeting the inclusion criteria was drawn from each VAfacility operating in FY 1999. If a location did not have800 patients meeting the criteria, all eligible patientswere retained. The Figure illustrates the process ofdeveloping the diabetic cohort.

Outcomes

Our outcome was HbA_1c, which measures mean glucosecontrol during approximately 120 days.⁵⁵ Resultsfrom facilities that measured total glycosylated hemoglobin(an older measure of glycemic control) were convertedto HbA_1c using the laboratory equipmentmanufacturer's conversion formula. During the studyperiod, the VA considered HbA_1c of 7.0% or higher to beelevated to a concerning level; levels of 9.0% or higherwere considered markedly elevated.²⁷ Specifically, thestudy modeled the associations among organization-leveland individual-level independent variables and thelast HbA_1c value obtained during the study period of FY2000 through FY 2001 (October 1, 1999, throughSeptember 30, 2001).

Organizational-level Variables

Gatekeeping

Preauthorization

Profiling

guideline

scale

education scale

The present study used a hypothesis-driven modelbuilding approach that captures the elements of thechronic care model that we hypothesized to be relatedto better diabetes control. From the 1999 VHA Surveyof Primary Care Practices, we used 5 scales related tothe implementation of managed care, defined asarrangements "implemented to influence clinicians'healthcare service utilization decisions."^56,p1 measures the degree to which primary careproviders maintain a level of control over referrals forspecialty services, tests, and treatments (eg, patientshaving the ability to self-refer for specialty services indicatesless use of gatekeeping than not allowing suchreferrals). measures the degree towhich providers must formally request approval to providecertain types of specialized services. isthe degree to which clinics monitor and report use ofspecific services by individual providers. The measures the degree to which clinics use evidence-based guidelines and mechanisms to implementthe guidelines. The indicates the levelof use of educational programs aimed at encouraging cliniciansto use evidence-based or cost-effective services.

clinical

practice autonomy scale

Primary care staffing authority

primary care

organizational influence and ability to form interorganizational

alliances scale

The survey also contains 3 scales that measure thedegree of primary care program authority.⁵⁷ The represents the authority ofthe program to set clinical policies and implement careguidelines. indicatesthe authority of the program to select, evaluate, andarrange the work of staff members. The addresses the authority ofthe program to establish relationships with other groupswithin and outside the VA facility.

Beyond these scales, the 1999 VHA Survey of PrimaryCare Practices assesses the following organizationalcharacteristics: (1) use of computerized providerreminders for diabetes care; (2) presence of multidisciplinaryteams that meet at least once a week to makepatient care decisions; (3) use of special-purpose teamsor protocols developed for designated clinical conditions(eg, diabetes mellitus) or patient groups (eg,homeless persons); (4) notification of all or almost allpatients of their assigned primary care provider; (5)indication that patients almost always see their assignedprovider (defined by the survey respondent at eachsite); (6) use of some type of provider recognition forquality care; and (7) indication that in the year beforethe survey (FY 1999) the clinic hired needed new staff.

Data regarding physician and nurse involvement inquality improvement (QI) were captured by 3 variablesthat indicate whether a clinic reported that (1) allphysicians and all nurses participate in QI initiatives,(2) all physicians but not all nurses participate, or (3) allnurses but not all physicians participate. The referentgroup includes clinics in which neither all nurses nor allphysicians participate.

Two variables captured the reporting relationshipbetween staff and the primary care program. One variableindicated whether the program has general internalmedicine physicians who report only to the program.The other indicated whether the program has nurseswho report only to the program. The relationshipbetween these organizational attributes and the chroniccare model is outlined in Table 1.

Finally, in addition to characteristics that may besensitive to administrative and clinical actions, otherorganizational attributes were used to adjust results forthe clinical environment. These included (1) region ofthe country, (2) number of reported primary careprovider full-time equivalents(attending physicians,resident physicians, andmidlevel practitioners), (3)whether the clinic is a largeacademic practice withmore than 30 internal medicinehouse officers, and (4)whether the clinic is basedat an acute care hospital.

Individual-level Variables

Risk adjustment forcomorbidities was doneusing aggregated diagnosticgroups, which are part ofThe Johns Hopkins AdjustedClinical Groups Case-MixSystem.^53,58 In addition,results were adjusted for thefollowing patient-level characteristics:age, sex, race orethnicity, marital status,body mass index (on theclosest date to October 1,1999), VA eligibility meanstest category (partial proxyfor socioeconomic status),service-connected disabilityrating, number of visits tothe assigned clinic duringthe FY in which the outcomeHbA_1c value was recorded,whether the patient filled aprescription for insulinduring the FY in which theoutcome HbA_1c value wasrecorded, and whether thepatient filled a prescriptionfor an oral hypoglycemicagent during the FY in whichthe outcome HbA_1c valuewas recorded.

Data Analysis

The present study dealtspecifically with patients offacilities that replied tothe 1999 VHA Survey ofPrimary Care Practices andprovided complete data onthe organizational characteristicsunder study. Threeof these clinics were not included because of the potentialfor incomplete downloads of HbA_1c data.

Hierarchical linear modeling was used to examine theassociations among the last HbA_1c value during the studyperiod of FY 2000 through FY 2001 (outcome [dependent]variable) and independent variables occurring at the organizationaland individuallevels. Organizational variablesare the focus of thestudy, and individual-levelvariables provide an opportunityto risk-adjust the organizational-level results.For each scale, the clinicsincluded in the analysiswere categorized by quartiles.The model includesvariables that allow eachscale quartile to be comparedwith the lowest quartile.Hierarchical modelingallows one to simultaneouslyconsider the effects ofgroup-level and individual-levelvariables on individual-level outcomes.^59-62 Thehierarchical model was fitusing the MIXED procedureof SAS version 8.2.⁶³The 2-level model consideredpatients (individuallevel) to be clustered withinclinics (organizationallevel).

The model reported forthe present study doesnot allow slope coefficientsfor individual-levelindependent variables tovary across clinics. Whenthe model was run allowingcoefficients to varyacross facilities, almostnone of the randomeffects were statisticallysignificant. Those thatwere did not have a practicaleffect on model fit.

RESULTS

The analysis included177 primary care programs and 82 428 patients,with a meanHbA_1c of 7.6%. The mean number ofpatients per clinic was 466 (range, 34-690). Table 2provides characteristics of the primary care programs,and Table 3 describes attributes of studypatients.

R²

P

Table 4 includes detailedresults for theassociation between organizationalstructure andprocesses and HbA_1c.Differences across clinicsaccounted for 5.04% of thevariance in HbA_1c (notequivalent to an for thecovariants). The followingattributes were associatedwith lower (better) HbA_1c and were statistically significantat < .1: greaterauthority to establish orimplement clinical policies(lower by 0.21% [thirdscale quartile comparedwith the lowest]), greaterstaffing authority (0.28%[highest scale quartilecompared with the lowest]),greater use of gatekeeping(0.16% [highestscale quartile comparedwith the lowest]), computerizeddiabetes reminders(0.17%), special teams orprotocols to respond toclinical issues (0.13%),weekly meetings of multidisciplinaryclinical teams(0.12%), all physicians butnot all nurses involved inQI (0.25%), notifying allpatients of their assignedprovider (0.21%), hiringneeded new staff duringFY 1999 (0.18%), havingnurses that report only tothe program (0.16%), andbeing a large academicpractice (0.27%). Associatedwith higher (worse)HbA_1c were programsreporting that patientsalmost always see theirassigned provider (greaterby 0.18%), having a QIprogram involving allnurses without all physicians(0.38%), having generalinternal medicinephysicians report only to the program (0.20%), beinglocated at an acute care hospital (0.20%), and having agreater number of reported providers (0.020% per additional10 provider full-time equivalents).

Combinations of Organization-levelCharacteristics

Compared with clinics having all characteristics atthe baseline level, clinics that have the attributes justdescribed that are directly related to the chronic caremodel and are associated with lower HbA_1c and haveother characteristics at the referent level would beexpected to have patients with HbA_1c values that arelower by 1.87%. Clinics with all of the chronic caremodel elements related to lower HbA_1c and none of thecharacteristics associated with higher HbA_1c would beexpected to have patients with HbA_1c values that arelower by 2.63%, compared with clinics with all attributesrelated to higher HbA_1c and none of the characteristicsassociated with lower HbA_1c.

In addition to considering the clinical significance ofthe organizational characteristics, the statistical significance of including the block of characteristics thatcapture attributes of the chronic care model wasassessed by calculating models with and without thecharacteristics. The -2 log likelihood values were compared.The findings indicated that the chronic caremodel attributes led to a better explanation of the variancein HbA_1c.

Sensitivity Analysis

P

P

To address the issue that many VA patients receivecare within and outside the VHA system^64,65 and atmultiple VA facilities, we performed a sensitivity analysisin which the minimum number of visits to the individual'smain VA clinic was varied. Although the resultsare not reported herein, the relationships among clinicfeatures and HbA_1c generally did not differ from thosepresented in the article. Having teams only for specificclinical and administrative problems was no longer statisticallysignificant at < .10 when patients had tohave had more than 2 visits during the year in whichthe outcome was recorded. The variable was significantat < .11.

When including patients with 3 or more visits, thegatekeeping and specific team associations were not statisticallysignificant. However, the point estimates werein the same direction as the reported results.

DISCUSSION

Although associations between HbA_1c and singleorganizational characteristics were not generally clinicallysignificant, when taken together, the associationbetween organizational factors and HbA_1c controlamong diabetic patients may have important clinicalimplications. The UK Prospective Diabetes Study resultsindicate that a reduction of 1% in HbA_1c leads to a 21%reduction in the risk of diabetes-related complicationsand deaths.³⁵ As already indicated, the combinedeffect of characteristics would be expected to begreater than 1%.

Although this study does not establish causal relationships,the findings point toward the following actionsthat, when taken together, may be associated with aclinically meaningful and beneficial difference in HbA_1c.Future research is needed to determine whether theseassociations reflect causal relationships and to ascertainwhat other factors predict their presence.

Characteristics Associated WithBetter Diabetes Control

Integrating Computerized Health Information Systems

Into the Care of Persons With Chronic Illness.

The finding that computerized diabetes reminders areassociated with better control of chronic disease is consistentwith other studies.^66,67

Developing Multidisciplinary Clinical Teams and

Other Specific Teams to Address Clinical Concerns.

As hypothesized, both of these attributes were associatedwith lower HbA_1c.

Actively Involving Physicians in Quality Improvement

Efforts.

Lack of strong physician involvement inQI initiatives was associated with worse HbA_1c control.This may reflect that physicians provide most hands-oncare in primary care settings, but it may also reflect theimportance of having physicians model appropriatequality-of-care behaviors for other providers. Theimportance of physicians'modeling behavior has beennoted in QI efforts such as programs to increase handwashing.⁶⁸

Monitoring the Potential Effect of Staff-

reporting

Relationships on Outcomes.

Having nurses reportingonly to the primary care program was associated withlower HbA_1c. However, the opposite was true for generalinternists. It can be hypothesized that the nurse associationcould happen because of greater involvementwith the primary care team. The physician result is difficultto interpret because of the variety of relationshipsbetween physicians and VA facilities.

Giving Primary Care Programs Greater Authority

to Implement Clinical Initiatives and Develop Staffing

Arrangements.

Although this study cannot indicatewhy greater authority over these areas was associatedwith lower HbA_1c, one hypothesis is that such clinicsmay be able to better organize into care teams andrespond to clinical concerns.

Notifying Patients About Decisions Affecting the

Delivery of Their Healthcare.

Clinics reporting thatthey notify patients who their primary care provider ishad patients with lower HbA_1c.

Although the association is of borderline statistical significance,greater use of gatekeeping within clinics wasassociated with lower HbA_1c. Because of the role of gatekeepingin attempting to reduce the use of specialty services,the word has developed a negative connotationamong patients and providers.^69-71 If gatekeeping systemslimit access to appropriate, timely care, one wouldhypothesize that it would be associated with worse outcomes.However, an ideal gatekeeping system would leadto providers and patients collaborating to identify andchoose needed services.⁷⁰ Such coordination is in linewith the chronic care model. It is therefore reasonablethat the highest level of gatekeeping is associated withlower HbA_1c compared with the lowest use of the process.

Contrary to the hypothesis that continuity shouldlead to better chronic illness outcomes, clinics reportingthat patients almost always see their assigned providerare associated with higher HbA_1c. When consideredalong with a study⁷² indicating that part-time physiciansof a large group-model health maintenance organizationmay provide slightly better diabetes care thanfull-time physicians, these results point to the need tofurther study the effect on diabetes control of systemsthat cause multiple primary care providers to beinvolved with an individual's management of the illness.A hypothesis proposed by Goldzweig et al⁷³regarding similar results relating to breast and cervicalcancer screening rates is that overly stringent assignmentof providers may reduce access to other servicesneeded to provide comprehensive care.

Limitations and Considerations

This study reports the relationship among broadlydefined structural and delivery process characteristicsand a specific intermediate diabetes outcome. Theresults do not prove a causal relationship between theattributes and better chronic illness management.However, findings indicate associations that provideadditional evidence for the importance of a coordinated,team-based approach to primary care.

Because of the observational nature of this study,there is the potential that our findings are confoundedby unmeasured characteristics. Variables included inthis study capture important aspects of the chroniccare model or were considered probable confoundersbased on known predictors of glycemic control. Ourchoice of hierarchical modeling and inclusion ofknown clinical predictors serve to reduce but not eliminatethis concern.

The present study takes advantage of the opportunityto combine data from a clinical database (VADiabetes Registry and Dataset) and VA administrativedata. As in other studies using administrative data asopposed to data collected specifically for researchstudies, there is a reduced ability to measure somepossible covariates, potential for missing or unknowndata, and possibility that clinics may code diagnosesdifferently.^74,75

In our study, validity concerns are partially mitigatedby the fact that laboratory values from the VADiabetes Registry and Dataset⁷⁶ and diagnosis codesfrom the VA administrative databases^77-79 have beenshown to have high levels of validity. This combinationof data sets allowed for the use of prescription andencounter criteria to determine patients eligible forinclusion in the study cohort. The result is increasedinternal validity because the cohort patients are morelikely to actually have diabetes mellitus.

As with all surveys, the issues of social desirability(ie, wanting to appear supportive of VA initiatives) andself-report may have affected the validity of the surveyresults. For example, having physicians and nursesinvolved in QI was a VA focus at the time of the survey.Therefore, some clinics may have overstated involvementin QI projects. Most organizational characteristicscame from answers to the 1999 VHA Survey ofPrimary Care Practices. These were based on self-reportand were not independently verified.

The VHA is different from most other healthcaredelivery organizations. It resembles a staff-modelhealth maintenance organization and promulgates asingle diabetes practice guideline.^80,81 Patients aremostly men who are sicker, older, and of lower socioeconomicstatus than the general population.^13,34,82Furthermore, the facilities in this study were largerthan most VHA operations. Despite these differences, itis likely that many of the observed organizationaleffects are applicable in other practice settings.

Finally, data were not available on patients' primarycare providers. Because of the variety of staffingarrangements in the VHA, we were unable to useadministrative data to determine who actually cared forindividual patients. However, evidence from the VHAindicates that, compared with facilities, providersaccount for a negligible percentage of the variance inHbA_1c.⁸³

CONCLUSIONS

Our results suggest that programs simultaneouslyleading to integrated teams that actively involvephysicians in QI, use electronic health informationsystems, provide programs with authority to respondto staffing and programmatic issues, and informpatients of important decisions have better glycemiccontrol among their diabetic patients. These conclusionsare based on clinics that face the variety of challengesencountered in the everyday practice ofprimary care.

Investigations designed to address the biopsychosocialneeds of patients with chronic illnesses should alsoconsider the environment into which interventions areplaced. The present study indicates that decisionsabout how to structure the system in which care is providedmay assist in improving clinical outcomes.Hence, the combination of data from managerial andclinical sources may be used to better inform the workof administrators and providers.

Acknowledgments

We thank the following individuals for their assistance during theresearch process: Cynthia Coffman, PhD; Scott B. Cribb; Jennifer A. Davis,MHSA; Andrew B. Lanto, MA; Brian S. Mittman, PhD; Michael E. Monger,MS; Denis Repke, PhD; and Lynn Soban, RN, MPH.

From the Health Services Research & Development Center of Excellence for HealthServices Research in Primary Care, Durham Veterans Affairs Medical Center (GLJ, DE, TKD,MW), and Division of General Internal Medicine, Duke University (GLJ, DE), Durham, NC,and Department of Epidemiology (MAI [emeritus], TSC, KEH), Cecil G. Sheps Center forHealth Services Research (TSC, S-YDL, KEH), Division of General Medicine and ClinicalEpidemiology (TSC), Department of Health Policy and Administration (S-YDL, MW), andDepartment of Obstetrics and Gynecology (KEH), University of North Carolina at ChapelHill; Health Services Research & Development Center of Excellence for the Study ofHealthcare Provider Behavior, Veterans Affairs Greater Los Angeles Healthcare System, andDepartment of Health Services, University of California at Los Angeles (EMY); HealthServices Research & Development Center of Excellence for Practice Management andOutcomes Research, Ann Arbor Veterans Affairs Medical Center, and Division of GeneralMedicine, University of Michigan, Ann Arbor (SLK); and Department of Epidemiology, TheJohns Hopkins University, Baltimore, Md (MAI).

Dr Jackson is a National Research Service Award-Agency for Healthcare Research andQuality Postdoctoral Fellow (institutional training grant 2T32HS000079-06 to DukeUniversity). Dr Weinberger is a Veterans Affairs Health Services Research & DevelopmentSenior Career Scientist awardee.

The views expressed in this article are those of the authors and do not necessarily representthe views of the Department of Veterans Affairs.

Address correspondence to: George L. Jackson, PhD, MHA, Health Services Research& Development Center of Excellence for Health Services Research in Primary Care, DurhamVeterans Affairs Medical Center, HSR&D Service (152), 508 Fulton Street, Durham, NC27705. E-mail: george_jackson@alumni.unc.edu.

JAMA.

1. Jencks SF, Huff ED, Cuerdon T. Change in the quality of care delivered toMedicare beneficiaries, 1998-1999 to 2000-2001. 2003;289:305-312.

Health Aff.

2. Kerr EA, McGlynn EA, Adams J, Keesey J, Asch SM. Profiling the quality of carein twelve communities: results of the CQI Study. 2004;23(3):247-256.

N Engl J Med.

3. McGlynn EA, Asch SM, Adams J, et al. The quality of health care delivered toadults in the United States. 2003;348:2635-2645.

Ann Intern Med.

4. Saaddine JB, Engelgau MM, Beckles GL, Gregg EW, Thompson TJ, NarayanKMV. A diabetes report card for the United States: quality of care in the 1990s.2002;136:565-574.

JAMA.

5. Bodenheimer T, Wagner EH, Grumbach K. Improving primary care for patientswith chronic illness. 2002;288:1775-1779.

Health Serv

Res.

6. Bonomi AE, Wagner EH, Glasgow RE, Von Korff M. Assessment of ChronicIllness Care (ACIC): a practical tool to measure quality improvement. 2002;37:791-820.

Am J Manag Care.

7. Heisler M, Wagner EH. Improving diabetes treatment quality in managed careorganizations: some progress, many challenges. 2004;10:115-117.

Ann Intern Med.

8. Rothman AA, Wagner EH. Chronic illness management: what is the role of primarycare? 2003;138:256-261.

Eff Clin Pract.

9. Wagner EH. Chronic disease management: what will it take to improve care forchronic illness? 1998;1:2-4.

Health Aff.

10. Wagner EH, Austin BT, Davis C, Hindmarsh M, Schaefer J, Bonomi A.Improving chronic illness care: translating evidence into action. 2001;20(6):64-78.

Milbank Q.

11. Wagner EH, Austin BT, Von Korff M. Organizing care for patients with chronicillness. 1996;74:511-544.

Manag Care Q.

12. Wagner EH, Davis C, Schaefer J, Von Korff M, Austin A. A survey of leadingchronic disease management programs: are they consistent with the literature?1999;7:56-66.

13. Department of Veterans Affairs Web site. VA fact sheets. April 2003.Available at: http://www1.va.gov/opa/fact/vafacts.html. Accessed April 28, 2004.

N Engl J Med.

14. Iglehart JK. Reform of the Veterans Affairs health care system. 1996;335:1407-1411.

N Engl J Med.

15. Jha AK, Perlin JB, Kizer KW, Dudley RA. Effect of the transformation of theVeterans Affairs health care system on the quality of care. 2003;348:2218-2227.

Am J Med Qual.

16. Kizer KW. The "new VA": a national laboratory for health care quality management.1999;14:3-20.

Hosp Health Serv Adm.

17. Flynn K, McGlynn G, Young G. Transferring managed care principles to theVA. 1997;42:323-338.

Hosp Health Serv Adm.

18. Halverson PK, Kaluzny AD, Young GJ. Strategic alliances in healthcare: opportunitiesfor the Veterans Affairs healthcare system. 1997;42:383-410.

Med Care.

19. Kizer KW, Demakis JG, Feussner JR. Reinventing VA health care: systematizingquality improvement and quality innovation. 2000;38(suppl 1):I7-I16.

National Diabetes Fact Sheet:

General Information and National Estimates on Diabetes in the United States,

2003.

20. Centers for Disease Control and Prevention. Atlanta, Ga: Division of Diabetes Translation, Centers for Disease Controland Prevention, US Dept of Health and Human Services; November 2003.

Diabetes Care.

21. American Diabetes Association. Standards of medical care in diabetes.2004;27(suppl 1):S15-S35.

JAMA.

22. Beckman JA, Creager MA, Libby P. Diabetes and atherosclerosis: epidemiology,pathophysiology, and management. 2002;287:2570-2581.

Diabetes in America.

23. National Diabetes Data Group. Bethesda, Md: NationalInstitute of Diabetes and Digestive and Kidney Diseases; 1995. NIH publication95-1468.

Am J Manag

Care.

24. Vinik AI, Vinik E. Prevention of the complications of diabetes. 2003;9(suppl):S63-S84.

Diabetes Care.

25. American Diabetes Association. Economic costs of diabetes in the US in 2002.2003;26:917-932.

Am J Manag Care.

26. Ashton CM, Septimus J, Petersen NJ, et al. Healthcare use by veterans treatedfor diabetes mellitus in the Veterans Affairs medical care system. 2003;9:145-150.

Med Care.

27. Krein SL, Hayward RA, Pogach L, BootsMiller BJ. Department of VeteransAffairs' Quality Enhancement Research Initiatives for Diabetes Mellitus. 2000;38(suppl 1):I38-I48.

Diabetes Care.

28. Maciejewski ML, Maynard C. Diabetes-related utilization and costs for inpatientand outpatient services in the Veterans Administration. 2004;27(suppl 2):B69-B73.

Diabetes Care.

29. Miller DR, Safford MM, Pogach LM. Who has diabetes? best estimates of diabetesprevalence in the Department of Veterans Affairs based on computerizedpatient data. 2004;27(suppl 2):B10-B21.

Diabetes Care.

30. Pogach LM, Hawley G, Weinstock R, et al. Diabetes prevalence and pharmacyuse in the Veterans Health Administration (1994): use of an ambulatory carepharmacy-derived database. 1998;21:368-373.

31. Quality Enhancement Research Initiative (QUERI) Web site. Diabetes mellitus.November 2003. Available at: http://www.hsrd.research.va.gov/publications/internal/querifactsheetDM.pdf. Accessed April 28, 2004.

Diabetes Care.

32. Reiber GE, Koepsell TD, Maynard C, Haas LB, Boyko EJ. Diabetes in nonveterans,veterans, and veterans receiving Department of Veterans Affairs health care.2004;27(suppl 2):B3-B9.

Diabetes

Care.

33. Young BA, Maynard C, Boyko EJ. Racial differences in diabetic nephropathy,cardiovascular disease, and mortality in a national population of veterans. 2003;26:2392-2399.

Med Care Res Rev.

34. Yu W, Ravelo A, Wagner TH, et al. Prevalence and costs of chronic conditionsin the VA health care system. 2003;60(suppl):146S-167S.

BMJ.

35. Stratton IM, Adler AI, Neil HAW, et al. Association of glycaemia withmacrovascular and microvascular complications of type 2 diabetes (UKPDS 35):prospective observational study. 2000;321:405-412.

Lancet.

Lancet.

36. UK Prospective Diabetes Study (UKPDS) Group. Intensive blood-glucose controlwith sulphonylureas or insulin compared with conventional treatment and riskof complications in patients with type 2 diabetes (UKPDS 33) [published correctionappears in 1999;354:602]. 1998;352:837-853.

Lancet.

Lancet.

37. UK Prospective Diabetes Study (UKPDS) Group. Effect of intensive blood-glucosecontrol with metformin on complications in overweight patients with type 2diabetes (UKPDS 34) [published correction appears in 1998;352:1557].1998;352:854-865.

JAMA.

38. Wagner EH, Sandhu N, Newton KM, McCulloch DK, Ramsey SD, GrothausLC. Effect of improved glycemic control on health care costs and utilization. 2001;285:182-189.

Diabetes Care.

39. Harris MI, Eastman RC, Cowie CC, Flegal KM, Eberhardt MS. Racial and ethnicdifferences in glycemic control of adults with type 2 diabetes. 1999;22:403-408.

Diabetes Care.

40. Koro CE, Bowlin SJ, Bourgeois N, Fedder DO. Glycemic control from 1988-2000 among US adults diagnosed with type 2 diabetes. 2004;27:17-20.

JAMA.

41. Saydah SH, Fradkin J, Cowie CC. Poor control of risk factors for vascular diseaseamong adults with previously diagnosed diabetes. 2004;291:335-342.

Am J Med Qual.

42. Singh H, Kalavar J. Quality of care for hypertension and diabetes in federal-versus commercial-managed care organizations. 2004;19:19-24.

Ann Emerg Med.

43. Kizer KW, Cushing TS, Nishimi RY. The Department of Veterans Affairs' role infederal emergency management. 2002;36:255-261.

1999 VHA Survey of

Primary Care Practices.

44. Yano EM, Simon B, Canelo I, Mittman B, Rubenstein LV. Sepulveda, Calif: Veterans Affairs Health Services Research& Development Center of Excellence for the Study of Healthcare ProviderBehavior; 2000 August. VA HSR&D Technical Monograph 00-MC12.

J Rural Health.

45. Weeks WB, Yano EM, Rubenstein LV. Primary care practice management inrural and urban Veterans Health Administration settings. 2002;18:298-303.

VA Diabetes Registry and Dataset

46. Healthcare Analysis and Information Group, Quality Enhancement ResearchInitiative-Diabetes Mellitus. [fact sheet]. AnnArbor, Mich: VA Ann Arbor QUERI-DM Research Coordinating Center; November22, 2002.

47. Hawley G. VIReC briefing: HAIG diabetes projects. June 20, 2001. Availableat: http://www.hsrd.ann-arbor.med.va.gov/queri/HAIGdiabetes2001.pdf. AccessedDecember 10, 2004.

Int J Med Inform.

48. Brown SH, Lincoln MJ, Groen PJ, Kolodner RM. VistA: US Department ofVeterans Affairs national-scale HIS. 2003;69:135-156.

Cancer.

49. Fletcher RD, Dayhoff RE, Wu CM, Graves A, Jones RE. Computerized medicalrecords in the Department of Veterans Affairs. 2001;91:1603-1606.

Eff Clin Pract

50. Murphy PA, Cowper DC, Seppala G, Stroupe KT, Hynes DM. Veterans HealthAdministration inpatient and outpatient care data: an overview. [serialonline]. 2002;5. Available at: http://www.acponline.org/journals/ecp/mayjun02/murphy.htm. Accessed April 28, 2004.

Diabetes Care.

51. Maynard C, Chapko MK. Data resources in the Department of Veterans Affairs.2004;27(suppl 2):B22-B26.

Ann Epidemiol.

52. Cowper DC, Kubal JD, Maynard C, Hynes DM. A primer and comparativereview of major US mortality databases. 2002;12:462-468.

VIReC Insights

53. Rosen AK, Trivedi P, Amuan M, Montez M. The Johns Hopkins AdjustedClinical Groups (ACGs) Case-Mix System: a risk-adjustment methodology currentlyavailable at the VA Austin Automation Center. [serial online].2003;4:1-10. Available at: http://www.virec.research.med.va.gov/References/VirecInsights/Insights-v04n1.pdf. Accessed April 28, 2004.

Clinical Classifications Software

(CCS) for

54. Agency for Healthcare Research and Quality. ICD-9-CM. February 2003. Available at: http://www.hcupus.ahrq.gov/toolssoftware/ccs/ccs.jsp. Accessed September 10, 2004.

Diabetes Care.

55. Sacks DB, Bruns DE, Goldstein DE, Maclaren NK, McDonald JM, Parrott M.Guidelines and recommendations for laboratory analysis in the diagnosis and managementof diabetes mellitus. 2002;25:750-786.

Adoption

of Managed Care Practice Arrangements by US Department of Veterans Affairs

Healthcare Delivery Organizations.

56. Mittman BS, Yano EM, Simon BA, Lee ML, Rubenstein LV, Kerr EA. Berkeley: Health Care OrganizationsConference 4, University of California at Berkeley; 2002.

Primary Care Authority Scales

57. Yano EM. [unpublished data]. Sepulveda, Calif:Veterans Affairs Health Services Research & Development Center of Excellence forthe Study of Healthcare Provider Behavior; 2003.

58. Johns Hopkins University ACG Case-Mix System Web site. Available at:http://www.acg.jhsph.edu. Accessed April 28, 2004.

Annu Rev Public

Health.

59. Bingenheimer JB, Raudenbush SW. Statistical and substantive inferences inpublic health: issues in the application of multilevel models. 2004;25:53-77.

Annu Rev Public

Health.

60. Diez-Roux AV. Multilevel analysis in public health research. 2000;21:171-192.

J Educ Behav Stat.

61. Singer JD. Using SAS PROC MIXED to fit multilevel models, hierarchical models,and individual growth models. 1998;24:323-355.

Stat Med.

62. Sullivan LM, Dukes KA, Losina E. Tutorial in biostatistics: an introduction tohierarchical linear modeling. 1999;18:855-888.

63. SAS System [computer program]. Version 8.2. Cary, NC: SAS Institute Inc;1999.

J Gen

Intern Med.

64. Borowsky SJ, Cowper DC. Dual use of VA and non-VA primary care. 1999;14:274-280.

Med Care Res Rev.

65. Shen Y, Hendricks A, Zhang S, Kazis LE. VHA enrollees' health care coverageand use of care. 2003;60:253-267.

JAMA.

66. Bodenheimer T, Grumbach K. Electronic technology: a spark to revitalize primarycare? 2003;290:259-264.

Int J Med Inform.

67. Tierney WM. Improving clinical decisions and outcomes with information: areview. 2001;62:1-9.

Emerg Infect Dis.

68. Lankford MG, Zembower TR, Trick WE, Hacek DM, Noskin GA, PetersonLR. Influence of role models and hospital design on hand hygiene of health careworkers. 2003;9:217-223.

JAMA.

69. Bodenheimer T, Lo B, Casalino L. Primary care physicians should be coordinators,not gatekeepers. 1999;281:2045-2049.

BMJ.

70. Forrest CB. Primary care gatekeeping and referrals: effective filter or failedexperiment? 2003;326:692-695.

Annu Rev Public Health.

71. Wickizer TM, Lessler D. Utilization management: issues, effects, and futureprospects. 2002;23:233-254.

J Gen Intern Med.

72. Parkerton PH, Wagner EH, Smith DG, Straley HL. Effect of part-time practiceon patient outcomes. 2003;18:717-724.

Am J Manag Care.

73. Goldzweig CL, Parkerton PH, Washington DL, Lanto AB, Yano EM. Primarycare practice and facility quality orientation: influence on breast and cervical cancerscreening rates. 2004;10:265-272.

Ann Intern Med.

74. Iezzoni LI. Assessing quality using administrative data. 1997;127:666-674.

Annu Rev Public Health.

75. Virnig BA, McBean M. Administrative data for public health surveillance andplanning. 2001;22:213-230.

Jt Comm J

Qual Improv.

76. Kerr EA, Smith DM, Hogan MM, et al. Comparing clinical automated, medicalrecord, and hybrid data sources for diabetes quality measures. 2002;28:555-565.

Am J Epidemiol.

77. Boyko EJ, Koepsell TD, Gaziano JM, Horner RD, Feussner JR. USDepartment of Veterans Affairs medical care system as a resource to epidemiologists.2000;151:307-314.

Med Care.

78. Kashner TM. Agreement between administrative files and written medicalrecords: a case of the Department of Veterans Affairs. 1998;36:1324-1336.

Am J Manag Care.

79. Szeto HC, Coleman RK, Gholami P, Hoffman BB, Goldstein MK. Accuracy ofcomputerized outpatient diagnoses in a Veterans Affairs general medicine clinic.2002;8:37-43.

Clin Ther.

80. Clark MJ Jr, Sterrett JJ, Carson DS. Diabetes guidelines: a summary comparisonof the recommendations of the American Diabetes Association, VeteransHealth Administration, and American Association of Clinical Endocrinologists.2000;22:899-910.

Diabetes Care.

81. Pogach LM, Brietzke SA, Cowan CL Jr, Conlin P, Walder DJ, Sawin CT;VA/DoD Diabetes Guideline Development Group. Development of evidence-basedclinical practice guidelines for diabetes: the Department of VeteransAffairs/Department of Defense guidelines initiative. 2004;27(suppl2):B82-B89.

Health Aff.

82. Wilson NJ, Kizer KW. The VA health care system: an unrecognized nationalsafety net. 1997;16(4):200-204.

Health Serv Res.

83. Krein SL, Hofer TP, Kerr EA, Hayward RA. Whom should we profile? examiningdiabetes care practice variation among primary care providers, providergroups, and health care facilities. 2002;37:1159-1180.

Download PDF: Veterans Affairs Primary Care Organizational Characteristics Associated With Better Diabetes Control