A Statewide Collaboration to Monitor Diabetes Quality Improvement Among Wisconsin Health Plans

Objective: The Wisconsin Collaborative Diabetes QualityImprovement Project was initiated in 1999 by the WisconsinDepartment of Health and Family Services, Diabetes Preventionand Control Program to monitor quality of diabetes care among thestate's health plans.

Study Design: Prospective observational.

Methods: Annual invitations were mailed to all Wisconsin managedcare plans to participate in the project. Collaborators submittedComprehensive Diabetes Care Health Plan Employer Data andInformation Set (HEDIS) measures, as well as other selected HEDISmeasures. Data were summarized and reported anonymously backto the collaborators at quarterly forums and in annual reports.

Results: Five of the 6 Comprehensive Diabetes Care HEDISmeasures have improved significantly in Wisconsin since 1999.Despite this improvement, measure variation across health plansremains high. Collaborators have continued to share resources andbest practices at quarterly forums and through statewide initiatives.

Conclusions: This project is an example of an ongoingstatewide collaborative quality improvement effort among otherwisecompeting health plans. Collaboration at regular forums, sharingof HEDIS data to assess quality of diabetes care in health plans,and sharing of resources and best practices may have contributedto improvements in the quality of diabetes care in Wisconsin.

(Am J Manag Care. 2005;11:332-336)

Research has demonstrated that improving theclinical care of diabetes mellitus reduces overallcosts, reduces secondary health issues, andimproves the quality and length of life of individualswith diabetes.^1,2The Wisconsin Collaborative DiabetesQuality Improvement Project was initiated in 1999 bythe Wisconsin Department of Health and FamilyServices, Diabetes Prevention and Control Program as acollaborative effort to build on this research locally. Aquality collaborative is a relatively new form of intervention,the goal of which is to create a rapid healthcarechange by incorporating multiple organizations into theendeavor.³This project includes staff from health planquality improvement departments, the state healthdepartment, and a university partner. The collaborativeproject was established as a forum to share resources,population-based strategies, and best practices, in orderto improve the quality of diabetes care.

An important part of the project is the annual collectionand reporting of the Health Plan Employer Dataand Information Set (HEDIS) Comprehensive DiabetesCare measures. The collaborators chose the HEDISmeasures to evaluate quality for several reasons. First,these data must be collected to obtain accreditationfrom the National Committee for Quality Assurance(NCQA). Because many of the health plans werealready collecting these data, the burden of data collectionwas not increased, encouraging initial participation.Furthermore, the data are collected annuallyand offer standardized and consistent measurement atthe population level.

A central objective of the project is the discussionand sharing of ideas among health plan collaborators.All collaborators are invited to regular quarterly forums.Forum discussion topics have included rotation of diabetesmeasures, stratification of hemoglobin A_1c (HbA_1c)and low-density lipoprotein (LDL) data, diabetes registries,and efforts to improve eye examination rates.Time during each forum is reserved for collaborators toshare their ongoing projects and initiatives to improvethe quality of diabetes care in their organization.Diabetes Prevention and Control Program staff is in frequentcontact with the collaborators, providing technicalassistance and updated information.

Through this project, the collaborators are able tomonitor progress in the quality of diabetes care inhealth plans across the state of Wisconsin. Healthplans gain knowledge about best practices, strategies,and resources in addition to unique feedback on theirrelative performance over time on HEDIS measures.

METHODS

Each year, from 1999 through 2002, all Wisconsinhealth maintenance organizations (HMOs) and severalother health systems, including an intertribal healthcaresystem, were invited to participate in theWisconsin Collaborative Diabetes Quality ImprovementProject. A small stipend was provided in the first yearonly to health plans that volunteered to participate. Atotal of 22 plans participated in 1 or more years; 15 plansparticipated for all 4 years and were included in thisanalysis. These 15 plans were similar to the other plansin Wisconsin regarding size and location in the state.

The HEDIS Comprehensive Diabetes Care measureswere first piloted in 1998. The baseline was establishedthe following year, and the Comprehensive Diabetes Caremeasures have been repeated yearly since then. Years referencedin this report reflect the year that care was provided(eg, when 1999 is mentioned, we are referring to1999 as the year that care was provided, which is a HEDIS2000 measure). HEDIS Comprehensive Diabetes Caremeasures are limited to individuals with diabetes aged 18to 75 years. The eligible population was defined as anyindividual with a diabetes diagnosis meeting the continuousenrollment definition in a managed care organizationor healthcare delivery system. Individuals with diabeteswere identified using criteria defined by the NCQA.⁴

The 6 HEDIS Comprehensive Diabetes Care measuresare LDL cholesterol (LDL-C) screening, LDL-Ccontrolled (LDL <130 mg/dL), nephropathy monitoring,HbA_1c poorly controlled (>9.5%), HbA_1c testing, and eyeexamination. Higher percentages are desired for allmeasures, except for HbA_1c poorly controlled. Decreasingpercentages in the HbA_1c poorly controlledmeasure demonstrates improvement in control of HbA_1c.

Participating organizations provided data for theircommercial populations only. Medicare and Medicaidbeneficiaries were not included, to facilitate comparisonover time, among plans (eg, some plans do not insureMedicare patients), and with national data. Participantsreported their data collection methods (administrativeor hybrid), eligible populations, sample sizes, and thepercentage of patients in the sample size meeting therequirements for each HEDIS measure.

To maintain confidentiality among health plans, collaboratorswere each assigned a unique, confidential codenumber. They could then see their relative ranking withinthe group, but would not know which code numberscorresponded to specific plans, other than their own.

The State of Health Care Quality

2003: Industry Trends and Analysis,

Each health plan reported a percentage for eachmeasure. For each year, the mean of these reported percentageswas calculated to determine the aggregateaverage percentage for all 15 health plans that participatedduring all 4 years of the project. Data published inthe NCQA publication ⁵were derivedfrom similar methods, including calculations of themean rather than the median. This method permitscomparison of Wisconsin data with national and regionaldata. Because patients from the intertribal health systemmay have been included in one or more HMOs,aggregate calculations included only HMO data.

Trends over time and variation between plans wereexamined. Each plan's 4-year average percentage wascalculated for each of the 6 measures. The range andmean for each measure illustrates which diabetes caremeasures were consistent or varied between plans. Theintertribal health system was included in these calculations.Trends over time, from 1999 through 2002, werecalculated for Wisconsin and the United States. Toaccount for ceiling effects, percent improvement wascalculated as a decrease in failure rate.⁶This calculationrepresents the reduction in the proportion not havingthe indicator (eg, an increase from 80% to 90% representsa 50% reduction in the proportion without theindicator). The Wilcoxon signed rank test was used tocompare 1999 averages to 2002 averages for Wisconsin;a comparison of means with equal variance was used tocompare US averages from 1999 through 2002.

Representatives of the health plans met quarterly toshare and discuss findings from the HEDIS data and toshare resources, best practices, strategies, and barriersencountered in their organizations. The quarterly forumsgenerally included brainstorming, as well as discussionabout data collection issues, data analysis, and resultsrelating to the project. The forums were generally wellattended, with much of the group representing qualityimprovement managers; total attendance, including individualsparticipating by teleconference, was generallynumbered in the 20s. Two separate educational sessionsprovided an opportunity for collaborators to learn moreabout HEDIS and allowed them to ask specific questions.Ongoing communication through several methodswas essential to the success of this collaboration.^7-12Collaborator e-mail discussions and updates occurredmonthly; presentations at local and national conferenceswere also frequent. Participation in the project, definedas the submission of HEDIS data for that year, was evaluatedyearly. Despite the high turnover rate in qualityimprovement managers, most (15 of 22) health planscontinued to submit their HEDIS ComprehensiveDiabetes Care measures annually for the 4 years of thisreport, as well as attend and participate in forums.

Collaborators also strived to improve the quality ofdiabetes care through statewide collaborative qualityimprovement initiatives. In 2001, the collaboratorsdeveloped the Diabetes Eye Care Initiative. Collaboratorsraised concerns that the dilated eye examination isthe only diabetes measure performed outside the primarycare office system. This factor may decreasepatient adherence and adds a barrier in communicatingresults from the eye care specialist to the primary carephysician. A lack of communication provides challengesin collecting accurate data. The objectives of this initiativeincluded increasing the percentage of individualswith diabetes receiving eye examinations as well asimproving communication and reporting of results andrecommendations.

Using joint letterhead to show a united message, theDiabetes Eye Care Initiative targeted all primary carephysicians and eye care specialists in the state.Physicians were encouraged to use an "Eye ExamCommunication Form," which was created to reportresults from the eye care specialist to the primary carephysician.

The second initiative was the Cardiovascular RiskReduction Initiative, the goal of which was to preventcardiovascular events in persons at risk or with establishedcardiovascular disease. Materials included 1-pageguidelines for healthcare providers and patient walletcards to assist individuals in tracking their own lipidand blood pressure results.

RESULTS

Of the 20 HMOs in Wisconsin, 15 HMOs submittedconsecutive yearly data from 1999 through 2002 forthe HEDIS Comprehensive Diabetes Care measures. Improvementswere noted for all 6 HEDIS ComprehensiveDiabetes Care measures during the 4-year period,although most measures varied widely across HMOs.

Figure 1 illustrates the overall improvement acrossComprehensive Diabetes Care measures. From 1999through 2002, percentages of patients with LDL-Cscreening, LDL-C controlled, and nephropathy monitoringimproved consistently across all years. Poorly controlledHbA_1c showed improvement, decreasing from 30%in 1999 to 21% in 2002. Hemoglobin A_1c testingimproved after the first year, but leveled off at about 90%by 2002. Similarly, eye examination rates improved afterthe first year, but leveled off at about 67%.

Figure 2 presents the variation in data for theComprehensive Diabetes Care measures submittedfrom the various health plans from 1999 through 2002.A 4-year overall mean for each measure was calculatedusing averages from each system; the range shows thepercentages from the highest-and lowest-performingplans over the 4 years of data collection. For example,one health plan had a 4-year average of 89% of itsenrollees with diabetes receiving eye examinations,whereas another health plan had a 4-year average of33%. These individual means, along with those of theother plans, combined for an overall mean of 64%.Reporting the data by range and mean more clearlyindicates areas for potential improvement.Nephropathy monitoring also had a wide variationamong the health plans, with a range of38%. The measures of HbA_1c testing and HbA_1c poorly controlledshowed much less variability, with ranges of 12% and23%, respectively.

When looking atplans individually,most plans improvedfrom 1999 through2002 (data notshown). All plans improvedtheir LDL-Cscreening and LDL-Ccontrolled rates, and81% of plans improvedtheir nephropathymonitoring rates.Ninety-four percent ofplans improved theirrates for HbA_1c poorlycontrolled and 81%improved their ratesfor HbA_1c testing. Sixty-nine percentof health plans improvedtheir eye examination rates (datanot shown).

The Table compares with theperformance of the HMOs participatingin the collaborative nationalaverages for care provided in 1999and 2002. Although the Wisconsinaverages improved slightlymore than the national averagesfor 5 of the 6 of the HEDISComprehensive Diabetes Caremeasures (based on the decreasein failure rate), these differenceswere not statistically significant.

DISCUSSION

Results from the Wisconsin collaborativeDiabetes Quality Improvement Project demonstratethat all of the HEDIS Comprehensive DiabetesCare measures improved in Wisconsin's health plansfrom 1999 through 2002. Most health plans saw improvementin most of the HEDIS indicators over time.Although the rates of improvement for most of the diabetesquality indicators were slightly greater in Wisconsincompared with the United States during this same time,these differences were not statistically significant. In fact,the differences observed between Wisconsin and theUnited States are likely due to other factors, becausequality of care was already higher in Wisconsin in 1999.Other researchers have found similar results when comparingWisconsin's performance with national averagesin the Medicare population.⁶

Despite overall improvements seen in ComprehensiveDiabetes Care measures, there is room for health planimprovement in quality diabetes care. The wide variabilityin data for some measures, such as eye examinationand nephropathy monitoring, suggest that somehealth plans can benefit from focusing on certain qualityimprovement measures. The diabetes measures areused to monitor progress in general; in reality, individualhealth plan quality and economic incentives, as wellas national trends, can contribute to the improvement.

The Wisconsin Collaborative Diabetes QualityImprovement Project differs from typical quality improvementefforts in several important ways. This project providescollaborators with local benchmarking data to assesstheir plan's quality of diabetes care, and allows the opportunityto identify potential quality improvement. Thequarterly forums offer the opportunity for collaborators toconvene to distribute new research and resources, sharelessons learned, address common barriers, and coordinatesharing of quality improvement strategies. The projectalso provides valuable surveillance data to the WisconsinDiabetes Prevention and Control Program to monitor diabetescare in Wisconsin. The consistent improvement ofdiabetes quality indicators over time and across healthplans suggests that diabetes care is improving in the state,with reduced risk of complications for the future.

Several limitations to the project deserve mention.The evidence is limited that collaboratives, in general,have successful and cost-effective results.³Only a minorityof groups report "significant improvements"; manygroups struggle with direction, roles, and motivation.Although this report presents population-level data, thefocus is on health plans, and may not reflect trends incare statewide. Although the health plans studied represent98% of the 1.5 million individuals enrolled in HMOsin Wisconsin, Medicare and Medicaid populations, theuninsured, and others insured by other means are notincluded in this analysis. Therefore, these data do notprovide a complete picture of diabetes care in the stateof Wisconsin.

More importantly, HEDIS measures are only a proxyfor more direct quality measurement. The HEDISComprehensive Diabetes Care measures do not evaluatewhether plans have followed good diabetes careguidelines. Although an enrollee with diabetes mighthave LDL-C controlled as defined by HEDIS, this factdoes not necessarily translate to that patient havingreceived optimal care. The measures are a mix of testsperformed (eg, LDL-C screening, nephropathy monitoring,HbA_1c, and eye examination) and test results (eg,LDL-C controlled and HbA_1c poorly controlled).

An additional limitation is that HEDIS measures arecollected by separate health plans and introduce variationin data collection. Plans use different methods toextract data, especially for eye examinations, whichmay lead to varying rates that might not exactly reflectthe true situation. Furthermore, errors in data collectioncan occur, because different entities are collectingand compiling the data. Ongoing communication withthe collaborators can minimize these limitations.

The results of this collaborative project demonstratethat the state's diverse health plans are willing to collaboratewith multiple partners and the state healthdepartment on quality improvement projects. The projectalso illustrates that an ongoing communicationforum is essential to distribute new resources, promotedynamic brainstorming, coordinate the sharing of qualityimprovement strategies, and offer opportunities touse the data to initiate new quality improvement initiatives.These successful results were achieved because ofa high level of commitment by all partners to the collaborativeeffort. Collaboration is key to the continuedsuccesses of this project.

Acknowledgments

We would like to recognize the following organizations for their interestand participation in this project: Advanced Health Care, Atrium HealthPlan, Inc, CompcareBlue, Dean Health Plan, Inc, Great Lakes Inter-TribalCouncil, Inc., Group Health Cooperative of Eau Claire, Group HealthCooperative of South Central Wisconsin, Gundersen Lutheran Health Plan,Health Tradition Health Plan, Humana WHO, Managed Health Services,Medica Health Plan, Medical Associates, MercyCare Insurance Company,Network Health Plan, Physicians Plus Insurance Corporation, Prevea HealthPlan, Security Health Plan of Wisconsin, Touchpoint Health Plan, UnitedHealthcare of Wisconsin, Inc., Unity Health Plans Insurance Corporation,Valley Health Plan, and State of Wisconsin Employee Trust Funds.

Many individuals made this project possible: Leah Ludlum, RN, BSN,CDE, Pat Zapp, Nancy Chudy, MPH, Judy Wing, and Peter Rumm, MD, of theDepartment of Health and Family Services, Division of Public Health; JudyFrisch, RN, MBA, Jay Gold, MD, JD, and Eric Streicher, MD, of MetaStar, Inc;Richard M. Carr, MD, MS, and Sandra Mahkorn, MD, MPH, MS, of theDepartment of Health and Family Services, Division of Health Care Financing;Faye Gohre, RN, BSN, of Gohre Consulting, LLC; Irene Golembiewski, MA, ofthe University of Wisconsin Medical School; and the members of theWisconsin Collaborative Diabetes Quality Improvement Project.

JAMA.

1. Writing Team for the Diabetes Control and Complications Trial/Epidemiologyof Diabetes Interventions and Complications Research Group. Sustained effect ofintensive treatment of type 1 diabetes mellitus on development and progression ofdiabetic nephropathy: the Epidemiology of Diabetes Interventions andComplications (EDIC) study. 2003;290:2159-2167.

BMJ.

2. Gray A, Clarke P, Farmer A, Holman R; United Kingdom Prospective DiabetesStudy (UKPDS) Group. Implementing intensive control of blood glucose concentrationand blood pressure in type 2 diabetes in England: cost analysis (UKPDS 63).2002;325:860.

Qual Saf Health Care.

3. Ovretveit J, Bate P, Cleary P, et al. Quality collaboratives: lessons fromresearch. 2002;11:345-351.

HEDIS

2003: Health Plan

Employer Data and Information Set. Volume 2: Technical Specifications.

4. The National Committee for Quality Assurance. ® Washington, DC: NCQA; 2002:94-103.

The State of Health Care

Quality 2003: Industry Trends and Analysis.

5. The National Committee for Quality Assurance. Washington, DC: NCQA; 2003.

JAMA.

6. Jencks SF, Huff ED, Cuerdon T. Change in the quality of care delivered toMedicare beneficiaries, 1998-1999 to 2000-2001 [published correction appears in2002;289:2649]. 2003;289:305-312.

The Wisconsin Collaborative

Diabetes Quality Improvement Project.

7. University of Wisconsin-Madison, Department of Population Health Sciences,Wisconsin Public Health & Health Policy Institute. June 2002. Available at: http://dhfs.wisconsin.gov/health/diabetes/pdf_files/CDQIProjectRep2002.pdf. Accessed January 12, 2005.

The Wisconsin Collaborative

Diabetes Quality Improvement Project: 2003.

8. University of Wisconsin-Madison, Department of Population Health Sciences,Wisconsin Public Health & Health Policy Institute. Available at:http://dhfs.wisconsin.gov/health/diabetes/pdf_files/DiabHMO_Collab2003.pdf.Accessed January 12, 2005.

The Wisconsin Collaborative

Diabetes Quality Improvement Project: 2004.

9. University of Wisconsin-Madison, Department of Population Health Sciences,Wisconsin Public Health & Health Policy Institute. Available at: http://dhfs.wisconsin.gov/health/diabetes/pdf_files/DiabCollab2004.pdf. Accessed January 12, 2005.

10. Siomos E. The success of collaboration: Wisconsin's Diabetes QualityImprovement Project, 1999-2001. Paper presented at: CDC Diabetes TranslationConference; May 11-14, 2004; Chicago, Ill.

Wisconsin Public Health and

Health Policy Institute Issue Brief.

11. Dawson K, Stone-Newsom R, Gohre F. The Wisconsin Collaborative DiabetesQuality Improvement Project: a collaborative success. May 2002;3(3). Available at: http://www.pophealth.wisc.edu/wphi/publications/briefs/may02brief.htm. Accessed January 12, 2005.

Wisconsin Medical Journal.

12. Quenan L, Remington P, Gohre F, Zapp P. The Wisconsin Collaborative DiabetesQuality Improvement Project. 2000;99:48-55.

From the University of Wisconsin Population Health Institute (EES, RSN, PLR) and theDepartment of Population Health Sciences (PLR), University of Wisconsin Medical School,Madison, Wis; and the Diabetes Prevention and Control Program, Division of PublicHealth, Wisconsin Department of Health and Family Services, Madison, Wis (JC).

This study was supported by Cooperative Agreement Number U321/CCU522717-01for the Centers for Disease Control and Prevention. This project was also supported by theWisconsin Department of Health and Family Services, Diabetes Prevention and ControlProgram and the Diabetes Advisory Group.

Address correspondence to: Patrick L. Remington, MD, MPH, UW Population HealthInstitute, 610 Walnut Street, Room 760, Madison, WI 53726. E-mail: plreming@wisc.edu.

JAMA.

BMJ.

Qual Saf Health Care.

3. Ovretveit J, Bate P, Cleary P, et al. Quality collaboratives: lessons fromresearch. 2002;11:345-351.

HEDIS

2003: Health Plan

Employer Data and Information Set. Volume 2: Technical Specifications.

4. The National Committee for Quality Assurance. ® Washington, DC: NCQA; 2002:94-103.

The State of Health Care

Quality 2003: Industry Trends and Analysis.

5. The National Committee for Quality Assurance. Washington, DC: NCQA; 2003.

JAMA.

6. Jencks SF, Huff ED, Cuerdon T. Change in the quality of care delivered toMedicare beneficiaries, 1998-1999 to 2000-2001 [published correction appears in2002;289:2649]. 2003;289:305-312.

The Wisconsin Collaborative

Diabetes Quality Improvement Project.

The Wisconsin Collaborative

Diabetes Quality Improvement Project: 2003.

The Wisconsin Collaborative

Diabetes Quality Improvement Project: 2004.

10. Siomos E. The success of collaboration: Wisconsin's Diabetes QualityImprovement Project, 1999-2001. Paper presented at: CDC Diabetes TranslationConference; May 11-14, 2004; Chicago, Ill.

Wisconsin Public Health and

Health Policy Institute Issue Brief.

Wisconsin Medical Journal.

12. Quenan L, Remington P, Gohre F, Zapp P. The Wisconsin Collaborative DiabetesQuality Improvement Project. 2000;99:48-55.