THE VIRGINIA HEALTH

OUTCOMES

PARTNERSHIP

Reducing the use of health care services and their associated costsby educating pharmacists, physicians, and nurses on communication

techniques and new advances in clinical practice.

Participants:

The Williamson InstituteThe Schools of Pharmacy, Medicine, and Nursing

of theMedical College of Virginia

Virginia Commonwealth UniversityRichmond, Virginia

The Virginia Department of Medical Assistance ServiceRichmond, Virginia

The Degge Group, Ltd.Arlington, Virginia

COMSORTBaltimore, Maryland

Sponsored by:The National Pharmaceutical Council

Reston, Virginia

Copyright © The Williamson Institute, Virginia Commonwealth University and The NationalPharmaceutical Council, 1997. No part of the material protected by this copyright notice may bereproduced or utilized in any form or by any means without permission from the copyright owners.

EXECUTIVE SUMMARY

Virginia Health Outcomes Partnership (VHOP) is a model program in whichthe state partnered with its health professionals to improve health outcomes forVirginia Medicaid patients. VHOP was implemented as part of VirginiaMedicaid’s Medallion program, a fee-for-service, primary care case manage-ment program; the enrollees consisted mostly of children and women of child-bearing age. The Medallion program was implemented under a waiver grantedby HCFA.

Cost containment of pharmaceutical expenditures usually involves variousefforts to control drug utilization. These efforts typically are aimed at physicianproviders, pharmacist providers, or the pharmaceutical industry. They haveincluded the establishment of drug formularies and prior authorization pro-grams. Federal legislation has had an important role in shaping these efforts.

A relatively new approach to cost containment is disease management. Thegoal of disease management programs is to optimize therapy. Optimizingtherapy should improve outcomes and decrease overall expenditures associ-ated with a disease.

The VHOP approach involves a patient-centered disease management strat-egy. With this approach, better disease management results from patients’optimal adherence to prescribed therapies. Increasing providers’ communica-tions skills, as well as their disease-specific knowledge, can result in betterchoice of treatments and improved patient compliance, leading to improvedoutcomes and lower costs.

This publication provides general background information on: 1) the methodsused in the disease selection process; 2) the choice of asthma as the first diseaseto be studied; and, 3) implementation of the intervention and research compo-nents of VHOP. VHOP did not have an a priori disease selected for intervention.One challenge for the investigators was determining what diseases to selectinitially for intervention since the selection would shape the content of themedical treatment guidelines taught to health care providers.

The first disease selected was asthma, and its selection was based primarily onthe prevalence and cost of the disease to Virginia Medicaid, but also on thepotential for improving patient outcomes in Medicaid managed care programs.The disease selection process incorporated empirical claims data from VirginiaMedicaid and a theoretical framework to identify relevant disease candidates.Additionally, multiple procedures were used to ensure that the most appropri-ate diseases were selected for the program.

This publication describes the disease selection process and the implementa-tion of the pilot component of VHOP in general terms. A manual outlining howto implement a project like VHOP in a state will be available in late 1997.

18

Copyright © The Williamson Institute, Virginia Commonwealth University and The NationalPharmaceutical Council, 1997. No part of the material protected by this copyright notice may bereproduced or utilized in any form or by any means without permission from the copyright owners.

TABLE OF CONTENTS

VHOP Program Contacts ........................................................................................... i

Abbreviations .............................................................................................................. ii

Introduction ..................................................................................................................1

VHOP Program History .............................................................................................1

VHOP Program Structure...........................................................................................2

The Disease Selection Process ....................................................................................2The Value of Empirical Data to Measure Outcomes ......................................................... 2The Framework for Disease Selection ................................................................................. 3Three Key Procedures ............................................................................................................ 4Procedure 1: Using Medicaid (DMAS) Data to Identify

Disease Selection Candidates ................................................................................... 4Procedure 2: Weighted Scoring of Diseases ....................................................................... 4Procedure 3: Survey of Medicaid DUR Committee Members ......................................... 4

The Intervention: HUB and SPOKE Model .............................................................6Using Physician Focus Groups for Feedback on the Intervention .................................. 6Recruiting Physicians............................................................................................................. 8Insights on the Recruitment Process .................................................................................... 9Insights on Multidisciplinary Training ............................................................................... 9

The Research.................................................................................................................9Measuring Health Status of Patients Using a Household Survey ..................................10Measuring Medical Care Utilization by Using Claims Data ...........................................10

Summary ..................................................................................................................... 11

Appendix: Evaluation of Specific Diseases Using Actual Data In Virginia ...... 12Asthma ....................................................................................................................................12Diabetes Mellitus ...................................................................................................................12Otitis Media ............................................................................................................................13

Hurley RE, Rossiter LF. Final Report of the Evaluation of the Medallion Program.Richmond, VA. Virginia Commonwealth University; April 1993.

Jang R. Medicaid formularies: A critical review of the literature. J Pharm MarketingManage 1988;2:39–61.

Kaplan SH, Greenfield S, Ware JE. Assessing the Effects of Physician-Patient Interac-tions on the Outcomes of Chronic Disease. Med Care 1989; 27:S110–127.

Kligman EW. Treatment of otitis media. Am Fam Phys 1992; 45:242–50.

Kozma CM, Reeder CE, Lingle EW. Expanding Medicaid drug formulary coverage:Effects on utilization of related services. Med Care 1990;28:963–77.

Krakauer H, Bailey RC. Epidemiological oversight of the medical care provided toMedicare beneficiaries. Stat Med 1991;10:521–40.

Lipton HL, Bird JA. Drug utilization review in ambulatory settings: state of the scienceand directions for outcomes research. Med Care 1993;31:1069–82.

Lipton HL, Bird JA. Drug utilization review: state of the art from an academicperspective. Clin Pharmacol Ther 1991;50:616–9.

Maish M, Sagraves R. Childhood asthma. US Pharmacist 1993; Jan:36–110.

Maklan CW, Greene R, Cummings MA. Methodological challenges and innovations inpatient outcomes research. Med Care 1994;32:JS13–21.

Malveaus FJ, Diamond E. Deaths from asthma by race, sex and age, 1979-1983. J AllergyClin Immunol 1987; 79:183.

Maser RD, Steenkiste AR, Dorman JS, et al. Epidemiologic correlates of diabeticneuropathy: Report from Pittsburgh Epidemiology of Diabetes Complications Study.Diabetes 1989; 8:1456–61.

Moore WJ. Medicaid drug utilization review: a critical appraisal. Med Care Rev1994;51:3–37.

Moore WJ, Newman RJ. Drug formulary restrictions as a cost-containment policy inMedicaid programs. J Law Econ 1993;36:79–97.

Moore WJ, Newman RJ. US Medicaid drug formularies: Do they work?Pharmacoeconomics 1992;1(suppl1):28–31.

Oxman Ad, Sackett DL, Guyatt GH. User’s guides to the medical literature. How to getstarted. The Evidence-Based Medicine Working Group. JAMA 1993;270:2093–5.

Pearce MJ, Begg EJ. A review of limited lists and formularies: Are they cost-effective?PharmacoEconomics 1992;1:191–202.

Penberthy L, Rossiter L, Whitehurst-Cook M, et al. Research Design Report. Richmond,VA: Virginia Health Outcomes Partnership Project; December 22, 1994.

Physician Payment Review Commission, Annual Report to Congress, Washington,D.C.: U. S. Congress, 1996

Roper WL, Winkenwerder W, Hackbarth GM, Krakauer H. Effectiveness in health care:An initiative to evaluate and improve medical practice. N Engl J Med 1988;319:1197–202.

Sheffer AL, Taggart VS. National Asthma Education Program: Expert Panel and ReportGuidelines for the Diagnosis and Management of Asthma. Med Care 1993; 31:MS20–8.

Sloan FA, Gordon GS, Cocks DL. Hospital drug formularies and use of hospitalservices. Med Care 1993;31:851–867.

Smalley WE, Griffin MR, Fought RL, Sullivan L, Ray WA. Effect of a prior-authoriza-tion requirement on the use of nonsteroidal anti-inflammatory drugs by Medicaidpatients. N Engl J Med 1995;332:1612–7.

16

Soumerai SB, Avorn J. Principles of educational outreach (‘academic detailing’) toimprove clinical decision making. JAMA 1990;263:549–56.

Soumerai SB, Avorn J, Ross-Degnan D, Gortmaker S. Payment restrictions for prescrip-tion drugs under Medicaid: effects on therapy, cost, and equity. N Engl J Med1987;317:550–6.

Soumerai SB, McLaughlin TJ, Ross-Degnan D, et al. Effects of limiting Medicaid drug-reimbursement benefits on the use of psychotropic agents and acute mental healthservices by patients with schizophrenia. N Engl J Med 1994;331:650–655.

Soumerai SB, McLaughlin TJ, Avorn J. Improving drug prescribing in primary care: acritical analysis of the experimental literature. Milbank Q 1989;67(2):268–317.

Soumerai SB, Ross-Degnan D. Experience of state drug benefit programs. Health Aff(Millwood) 1990;9(3):36–54.

Soumerai SB, Ross-Degnan D, Avorn J, McLaughlin TJ, Choodnovskiy I. Effects ofMedicaid drug-payment limits on admission to hospitals and nursing homes. N EnglJ Med 1991;325:1072–7.

Soumerai SB, Ross-Degnan D, Fortess EE, Abelson J. A critical analysis of studies ofstate drug reimbursement policies: research in need of discipline. Milbank Q1993;71(2):217–52.

Steinwachs DM, Wu AW, Skinner EA. How will outcomes management work? HealthAff (Millwood) 1994;13:153–62.

Stool SE, Berg AO, Berman S, et al. Managing otitis media with effusion in youngchildren: Quick reference guide for clinicians. AHCPR Publication No. 94-0623,Rockville, MD, July 1994.

Tengs TO, Adams ME, Pliskin JS, Safran DG, Siegel JE, Weinstein MC, et al. Five-hundred life-saving interventions and their cost-effectiveness. Anal Intern Med1995;15:369–89.

Ware JE, Snow KK, Kosinski M, Gandek B. SF-36 Health Survey: Manual & Interpre-tation Guide. Boston: The Health Institute, New England Medical Center, 1993.

Whitehurst-Cook M, Roberts M, Nelson M, et al. Educational Planning Design Report.Richmond, VA: Virginia Health Outcomes Partnership Project; December 22, 1994.

Wickizer TM. The effect of utilization review on hospital use and expenditures: areview of the literature and an update on recent findings. Med Care Rev 1990;47:327–63.

Wilson CN. Medicaid moving to managed care. Hospital Pharm 1995; 30;214–20.

17

Louis F. Rossiter, Ph.D.Co-Principal InvestigatorThe Williamson Institute for Health StudiesMedical College of VirginiaVirginia Commonwealth UniversityP.O. Box 980203Richmond, VA 23298-0203(804) 828-5223

Michelle Y. Whitehurst-Cook, M.D.Co-Principal InvestigatorSchool of MedicineMedical College of VirginiaVirginia Commonwealth UniversityP.O. Box 980216Richmond, VA 23298-0216(804) 828-9626

Ralph E. Small, Pharm.D.Co-Principal InvestigatorSchool of PharmacyMedical College of VirginiaVirginia Commonwealth UniversityP.O. Box 980533Richmond, VA 23298-0533(804) 828-6333

Charles A. Shasky, R.Ph., M.B.A.Project Manager, VHOPThe Williamson Institute for Health StudiesMedical College of VirginiaVirginia Commonwealth UniversityP.O. Box 980203Richmond, VA 23298-0203(804) 828-0172

Dave Shepherd, R.Ph.Pharmacy SupervisorProgram OperationsVirginia Department of Medical Assistance Services600 E. Broad St., Suite 1300Richmond, VA 23219(804) 225-2773

Judith K. Jones, M.D., Ph.D.PresidentThe Degge Group, Ltd.1616 North Fort Myer DriveSuite 1430Arlington, VA 22209-3109(703) 276-0067

John W. HawksPresidentCOMSORT2300 North Charles StreetBaltimore, MD 21218(410) 467-1100

Richard A. Levy, Ph.D.Vice President, Scientific AffairsNational Pharmaceutical Council1894 Preston White DriveReston, VA 20191-5433(703) 620-6390

VHOP PROGRAM CONTACTS

Jeann Lee Gillespie, Pharm.D., M.S.Director, Scientific AffairsNational Pharmaceutical Council1894 Preston White DriveReston, VA 20191-5433(703) 620-6390

i

Specific Measures for Assessing Otitis MediaInterventions

The following are measures available to assessinterventions in otitis media patients:

• Percent utilization of first- versus second- orthird-generation of antibiotics in patients withno prior history of otitis (no diagnosis of otitisin the previous 6 months)

• Percentage of patients placed on prophylac-tic therapy after a diagnosis of 3 episodes ofacute otitis media within a 6-month period or4 episodes within a 12-month period

• Rates of follow-up office visits after treatmentand relapse rates

• Rates and costs of tympanotomy tube re-placement overall and among those treated

prophylactically; significant medical com-plications and persistent symptoms of dis-equilibrium (vertigo or ataxia), validated bymedical records; include documented de-velopmental delays due to language diffi-culties

• Rates of medical complications: hearing loss,perforation of the eardrum, cholesteatoma(often requiring multiple surgical procedures),acute mastoiditis, atelectasis of the eardrum,retraction pockets, and ossicular discontinu-ity and fixation; other behavioral manifesta-tions (e.g., irritability, disturbed sleep, de-creased responsiveness, social withdrawal)

• Rates of referral for speech and languagedevelopmental lags

• Rates of side effects from medications

14

ABBREVIATIONS

AHCPR Agency for Health Care Policy and Research

CME Continuing Medical Education

COPD Chronic Obstructive Pulmonary Disease

DMAS Department of Medical Assistance Services (i.e., Virginia Medicaid)

DUR Drug Utilization Review

FY Fiscal Year

HCFA Health Care Financing Administration

NIH National Institutes of Health

PA/VHOP Prior Authorization/Virginia Health Outcomes Partnership(i.e., Virginia Medicaid’s Prior Authorization/Virginia Health OutcomesPartnership Committee)

UTI Urinary Tract Infection

VCU Virginia Commonwealth University

VHOP Virginia Health Outcomes Partnership

ii

services representing 178 individuals for an av-erage cost of $133. Among Medallion patients,803 had at least one physician office visit with adiagnosis of diabetes appearing on the physi-cian claim. The average cost of these visits was$38. Of all patients diagnosed with diabetes, 662had claims for laboratory services for an averagelaboratory claim charge of $108.

Feasibility of Voluntary AuthorizationGuidelinesIntensive diabetes therapy is designed to achievenormal blood glucose values through the use ofthree or more insulin injections per day or insulinadministration with an external pump. Thistherapy prevented development and retardedprogression of diabetic retinopathy and neuropa-thy, albeit at the expense of a 3-fold increase in thenumber of severe hypoglycemic events (DiabetesControl and Complications Trial Research Group,1993). Intensive therapy was guided by frequentself-monitoring of blood glucose levels, leadingto systematic insulin dose adjustments (DiabetesControl and Complications Trial Research Group,1993).

Specific Measures for Assessing DiabetesInterventions• Rates of diabetic retinopathy

• Rates of cardiovascular disease

Otitis MediaPrevalence and Associated Drug UseOtitis media (inflammation of the middle ear) isthe most frequent primary diagnosis for chil-dren less than 15 years of age based on physicianoffice recordings. Almost all children experi-ence one or more episodes of otitis media by age6 (Stool et al., 1994). The incidence of otitismedia is highest in children between 6 monthsand 3 years of age. A smaller peak incidenceoccurs between 4 and 7 years. Infections areuncommon after age 8. It has been estimatedthat 76% to 95% of all children will suffer at leastone episode during their childhood. Approxi-mately 50% of children who have experiencedotitis media have had 3 or more episodes, while25% have had 6 or more episodes. Nationally,otitis media is estimated to be responsible formore than 30 million outpatient visits per year ata cost that exceeds $1 billion. Another $1 to $2

billion is spent annually on surgical treatmentssuch as tube replacement (Kligman, 1992).

In Virginia, based on the ICD-9-CM codes (codes381 and 382) for otitis appearing on the physi-cian invoice, 73,608 separate individuals pre-sented with otitis during FY 1993. The annualnonduplicated number of Medicaid recipientsunder the age of 21 during FY 1993 was 279,531.However, the number of claims and total cost ofthese claims were not available.

Otitis Media in the Medallion PopulationThe total number of otitis media patients in theMedallion population in FY 1994 was 18,469.Only 32 of these patients were hospitalized withthis diagnosis at an average cost per hospitaliza-tion of $1,505. Claims for outpatient services(n=11,714) representing 7,969 individuals in-volved an average payment of $74. The averagephysician office visit cost $36, but with 31,940visits from 16,434 Medallion recipients the totalamount paid was more than $1.1 million.

Feasibility of Voluntary AuthorizationGuidelinesPneumatic otoscopy is recommended for assess-ment of the middle ear because it combines thevisualization of the tympanic membrane with atest of membrane mobility and results in anaccurate diagnosis of otitis media in up to ap-proximately 75% of patients.

Longitudinal studies of otitis media with effu-sion demonstrate high rates (>50%) of spontane-ous resolution within 3 months of symptomonset. Surgery is recommended only for thechild who has had bilateral effusion for 3 monthsand who has a bilateral hearing deficiency (de-fined as a 20-decibel hearing threshold level orworse in the better-hearing ear). In this case,bilateral myringotomy with tube insertion be-comes an additional treatment option. Place-ment of tympanotomy tubes is recommendedafter a total of 4–6 months of bilateral effusionwith a bilateral hearing deficit.

Antibiotic therapy resulted in a 14% improve-ment in clearance of effusion at 1 month. How-ever, nausea, vomiting, and diarrhea are sideeffects 2% to 32% of the time, depending on theantibiotic type and dosage. Also, there is thepotential for developing resistant strains of bac-teria with this treatment.

13

INTRODUCTIONIn early 1996, approximately 7.8 million peopleacross the country (23% of all Medicaid recipi-ents) were enrolled in managed care programs(Physician Payment Review Commission, 1996).Nearly 3 million people were enrolled in pri-mary care case management systems. In mostof these systems, physicians are paid case man-agement fees (typically $3 per recipient permonth) in addition to their regular fee-for-ser-vice payments for the primary care servicesthey provide.

Virginia established the Medallion program, itsprimary care case management program, in theearly 1990s. This program was implementedunder a waiver granted by HCFA in an effort tocontrol costs. By 1996, nearly 500,000 recipientswere enrolled with primary care physicians ascase managers in Virginia Medicaid statewide.As the primary care case management programwas being implemented in the early 1990s, thestate also was looking for ways to control risingdrug costs in the Medicaid program.

Although a number of mandatory prior autho-rization programs were already in place in otherstates, no voluntary prior authorization pro-gram had been implemented. Mandatory priorauthorization programs have been shown to beexpensive to administer and have not proven toresult in overall Medicaid cost savings. Vir-ginia had some success with a voluntary priorauthorization program in conjunction with theVirginia Pharmacy Association to reduce theutilization of H2-antagonists and save costs(American Pharmacy, 1994).

VHOP is a voluntary prior authorization pro-gram in which the physician self-authorizesprescriptions. This program has helped Vir-ginia Medicaid decrease service utilization, aswell as maintain and invest in its relationshipwith Medicaid health care providers.

VHOP PROGRAM HISTORYIn May 1993, VHOP was proposed by Virginia’sDepartment of Medical Assistance (DMAS),also commonly known as Virginia Medicaid.The project was directed at health care provid-

ers and patients enrolled in Virginia Medicaid’sMedallion program.

The Medallion program was developed to offera primary care physician to each enrollee in thehopes of improving access while lowering Med-icaid costs (Physician Payment Review Com-mission, 1996; Wilson, 1995). The primary careprovider agrees to become the case manager orgatekeeper for the enrollee. Included in thisresponsibility is the obligation to be available toprovide and/or authorize all nonemergency carefor the enrollee.

The primary goals of the VHOP–Medallion col-laboration are:

• to improve patient health outcomes by im-proving the general and disease-specific com-munication skills of physicians and otherhealth care providers; and

• to increase physicians’ use of established prac-tice guidelines and appropriate pharmaco-therapy for specific disease states.

In 1994, the Williamson Institute for Health Stud-ies of VCU was chosen by DMAS to administerthe VHOP program. A working coalition wasformed among DMAS; the Schools of Medicine,Pharmacy, and Nursing of the Medical Collegeof Virginia; and the National PharmaceuticalCouncil. Early support was obtained from pro-fessional societies including the Virginia Acad-emy of Family Physicians, the Medical Society ofVirginia, the Old Dominion PharmaceuticalAssociation, and the DMAS–DUR Committee.Many of the major health professional societiesin Virginia have expressed support for VHOP.

Potential disease candidates were systematicallyevaluated, and asthma was chosen as the firstdisease for the pilot program. The pilot inter-vention and research designs were completed inDecember 1994. The asthma intervention andresearch program were implemented in October1995. The intervention for asthma was com-pleted in Central Virginia in September 1996,and the data are currently being analyzed, com-paring the intervention to the usual care in popu-lation centers outside of Central Virginia. Theasthma intervention is expected to be imple-mented throughout Virginia by September 1997.

1

APPENDIXEvaluation of Specific Diseases Using Actual Data in Virginia

AsthmaPrevalence and Associated Drug UseNationally, pediatric asthma appears to be onthe rise, particularly among minority popula-tions. Childhood asthma (i.e., asthma in indi-viduals <21 years of age) in Virginia had a preva-lence of approximately 5% in 1993 (Maish &Sagraves, 1993). For adult and childhood asthma,approximately 4.25% of Virginia’s budget wasdirected to payment of pharmacy claims. Basedon drug use, DMAS reimbursed $6,899,213 dur-ing FY 1993 for 294,142 claims made by 63,115individuals (who had to have at least two claimsfor asthmatic drugs to be eligible). However,these drugs also are used in the treatment ofCOPD. In addition, these claims apply through-out the Medicaid system rather than being re-stricted to the Medallion program. Thus, thereare problems in interpretation. More reliable isa diagnosis based on the ICD-9-CM codes ap-pearing on the physician invoice. According tothis criterion, 21,711 individuals were diagnosedwith asthma. This discrepancy between asth-matic patients identified by drug use versusphysician forms involves approximately 66% ofthe patients and points to the need for multiplecriteria to ensure the valid selection of patients.

Asthma in the Medallion PopulationThe statewide population of asthmatic individu-als in Medallion in FY 1994 was 6,405. Therewere 729 hospitalizations for asthma represent-ing 630 individuals. The average payment perhospitalization was $1,964 (ranging from $394 to$12,409). The total amount paid during the yearfor hospital claims with a diagnosis of asthmawas $1,431,802. There were 4,422 claims foroutpatient asthma clinic services during FY 1994for 2,851 Medallion patients. The total cost foroutpatient visits was $582,194, which averagesout to $132 per visit. These outpatient clinicservices are not necessarily services rendered byemergency departments. There were 5,784 Me-dallion recipients with a diagnosis of asthmalisted on physician claim forms, for a total of12,737 claims. The total cost for these visits was$579,789, an average cost of $46 per office visit.Among the asthma patients, 2,782 had claims forlaboratory services, for a total of 15,600 claims

(about 5.6 claims per patient). The total cost was$179,062, an average cost of $64 per laboratoryclaim (DMAS, 1995).

Feasibility of Voluntary AuthorizationGuidelinesPart of the problem in treating asthma is relatedto a shift in the perceived cause of asthma byspecialists. The conventional view was thatasthmatic episodes were caused by broncho-spasm (airway constriction) that progresses toairway obstruction. However, the recent ap-proach is to view asthma as an inflammatorydisease as well, in that inflammation is critical toairway hyper-responsiveness (Djukanovic et al.,1990). Therefore, anti-inflammatory agents arebelieved to contribute to the optimal care of suchpatients (Sheffer & Taggart, 1993); this view isreflected in current NIH and other guidelines.

Specific Measures for Assessing AsthmaInterventionsTwo categories of measures are available:

• Assessing quality of care (e.g., medical recordsabstraction for peak air flow readings or tele-phone interview assessing whether and towhat extent patients were counseled and edu-cated by physicians and pharmacists)

• Assessing actual clinical outcomes (e.g., emer-gency room visits or hospitalizations and/orlength of stay)

Diabetes MellitusPrevalence and Associated Drug UseDiabetes mellitus is the leading cause of periph-eral neuropathy in developed nations. Periph-eral neuropathy affects at least 15% of diabeticpatients overall (Dyck & O’Brien, 1989) and 37%of persons 18 years and older with insulin-de-pendent diabetes mellitus (Maser et al., 1989).

Diabetes in the Medallion PopulationBased on limited data in 1994, only 871 patientswere diagnosed with diabetes statewide. Amongthese patients, there were 65 hospitalizations(involving 61 unique individuals), with an aver-age payment per hospitalization of $3,151. Therewere 324 claims for outpatient hospital clinic

12

VHOP PROGRAMSTRUCTUREThe VHOP program has three main compo-nents: the disease selection process, the interven-tion, and the research to determine the program’seffectiveness. First a process was developed toidentify the most appropriate diseases for theprogram. Once a disease is selected, an inter-vention is developed. The intervention is de-signed around the HUB and SPOKE model ofhealth care provider-patient communications.The HUB consists of basic communication skills,and the SPOKE consists of treatment guidelinesfor a specific disease. The research componentevaluates the economic, service utilization, pa-tient and provider satisfaction, and outcomes ofthe intervention. Each of these three compo-nents is discussed in this publication.

THE DISEASE SELECTIONPROCESSA key element of VHOP was to develop a sys-tematic process to identify the most appropriatediseases for the program. The process had toutilize both empirical data from the MedicaidManagement Information System (i.e., DMAS’sclaims data) and clinical knowledge about thepatterns of disease and treatment (i.e., a theoreti-cal framework) to compare attributes of diseasecandidates for intervention.

Disease candidates were chosen by identifyingsources of information about them and then devel-oping standard methods for using this informa-tion. Primary source data were provided by DMAS,and the patient-specific information came directlyfrom claims records and enrollment files. Informa-tion describing medical providers came fromDMAS provider record files. All of this informa-tion was confidential and used for research pur-poses only. This highly specific information wasthe basis for most of the decisions related to selec-tion of a disease SPOKE. When appropriate, theprimary data were supplemented with data fromother sources (e.g., other studies and surveys).

Information about disease treatment was obtainedfrom national consensus-based sources (e.g., NIH,AHCPR). Reviews of treatment guidelines weresought at the state and local levels to ensure

applicability to community practice. The expertopinions of the clinicians providing services tothe Medallion patient population were used toassess community perceptions of the areas ofneed, problems, practical barriers, and potentialfor improvement.

Final disease selection was made by a grouprepresenting all parties working on the imple-mentation of VHOP: the Williamson Institute,the Schools of Pharmacy, Medicine, and Nurs-ing of the Medical College of Virginia, DMAS,the Degge Group, Ltd., and the National Phar-maceutical Council. The decision was based onthe results of the many analyses described in thissection; key factors were the prevalence and costof the disease to Virginia Medicaid, and thepotential for improving patient outcomes inMedicaid managed care programs. All repre-sentatives in the group agreed to the selection ofasthma as the first disease in the pilot program.

The Value of Empirical Data toMeasure OutcomesIdeally, successful outcomes management re-quires detailed population-based data, data onhealth care service use, and clinical outcomes.Service use and cost data are typically availablefrom medical claims information traditionallycollected in Medicaid populations for the past 20years or more. For each Medicaid recipient inMedallion, billing data include age, sex, race,county, inpatient hospitalization and outpatientdiagnoses, outpatient drugs and procedures (e.g.,laboratory, radiologic), and number and costs ofphysician office and emergency department vis-its. Pharmacy records—including prescriptionname, amount, dose dispensed, and estimateddays of therapy—also are collected. In an inter-nal analysis of pharmacy usage among Medal-lion recipients (n=2,399), 93% used only one ortwo pharmacies, whereas only 7% used three ormore pharmacies (Miller, 1994). Matching pa-tient data is facilitated by the lack of variance inthe number of pharmacies, so assessing compli-ance with certain drug regimens is possible.

As is true for all claims data, these data arecollected for billing and so are free of somebiases (e.g., interview, recall, reporting). How-ever, the data are of uncertain validity, espe-

2

research, although the formats were cost-ef-fective for storing billing data.

• Medicaid claims data sets are very large, anddata analysis strained the technical capacitiesof the DMAS computer systems. DMAS storesmost of its data on tapes, which results inslower processing.

• Researchers had limited access to the datadue to competing demands for administra-tive and other research purposes. For ex-ample, access to DMAS storage tapes waslimited to overnight hours.

• There has been no independent verificationof the validity of the data.

VHOP is developing a comprehensive data dic-tionary and has migrated most of the relevantdata to a supercomputer at VCU with adequatedata storage capacities. There should be fewerproblems with data analysis in the future. How-ever, the time involved in structuring and pro-gramming data analyses should not change sub-stantially in the future.

The use of Medicaid claims data in this type ofresearch is probably feasible for other states, butthe strengths and limitations of claims data mustbe considered so that appropriate measures arechosen. Each state’s Medicaid claims system isunique. Ideally, data analysis should be per-formed by persons who have extensive experi-ence using the data and have a firm grasp of theidiosyncrasies of the data set. The computersystem capabilities should be assessed for theirsuitability for research data analysis. If it isfinancially feasible, investment in newer hard-ware technologies, such as disk drive systemswith high storage capacity, should be consid-ered if they are not already in place.

SUMMARYThe VHOP effort is designed to improve healthoutcomes among Medicaid patients enrolled inthe Medallion program by improving the com-munication skills of physicians and other healthcare providers and by promoting recently estab-lished, innovative clinical guidelines throughpeer education. The pilot program trained phy-

sicians only on the treatment of asthma in Cen-tral Virginia. The next phase is being imple-mented, in which VHOP will: 1) expand theasthma training for physicians statewide; 2) in-corporate pharmacists as part of the asthmatraining; 3) implement training on congestiveheart failure for physicians statewide; and, 4)develop education materials for diabetes melli-tus and schizophrenia.

The methods of the disease selection process in-volved weaving together empirical data and atheoretical framework to compare relevant dis-ease-state candidates for intervention. Many at-tributes of a disease were considered simulta-neously (e.g., the incidence of poor and costlyoutcomes, the availability of established guide-lines, a structure for enacting the intervention, anda reliable and valid mechanism for evaluation).Therefore, in order to identify the most appropri-ate disease for the pilot program, three procedureswere used: 1) attributes identified in the theoreticalframework were applied to the DMAS data; 2)prospective disease candidates were ranked bytheir ability to meet key attributes; and, 3) opinionsof the Virginia Medicaid DUR Committee mem-bers were determined by survey.

The results of these analyses indicated thatasthma was an appropriate first choice for thepilot intervention. At the time this analyses wascompleted, it was thought that disease stateswarranting consideration for subsequent inter-vention included otitis media and epilepsy.However, subsequent analyses indicate that forthe Virginia Medicaid population, congestiveheart disease, diabetes mellitus, and schizophre-nia are better candidates.

VHOP is a program where the state partnerswith its health professionals. This partnershipnot only builds a sense of trust between theseparties but also relies on the model of profes-sional responsibility for successful implemen-tation. The VHOP researchers and VirginiaMedicaid continue to test this partnership pro-gram and carefully evaluate its effects. Prelimi-nary findings show significant improvementsin relationships between the state and the healthcare providers and cost savings among a tar-geted group of patients to date. This partner-ship model may be useful to other states.

11

cially for some diagnoses that may result inmisclassifications when selecting individualswith particular disease states or identifying clini-cal outcomes. Longitudinal medical historiesare available because of continuous data collec-tion. However, this advantage only applies tothose individuals who remain in the system;patients can be lost to follow-up because ofeligibility changes. DMAS lacks data on vari-ables that may be important mediators of treat-ment effects and potential confounding vari-ables (e.g., smoking status, dietary intake, alco-hol consumption). These missing data indicatethe importance of accessing primary medicalrecords. Some potentially important sources ofmedical information are missing as well, includ-ing inpatient medication use, use of over-the-counter drugs, blood pressure readings, air peakflow readings, and other signs of disease. Nev-ertheless, claims data do provide indirect infor-mation on outcomes through proxy measures.

The Framework for Disease SelectionBased on the DMAS claims data, various at-tributes were developed for: 1) selection of dis-eases; 2) selection of patients for the study group;and, 3) characteristics of the population fromwhich the study group is selected. These at-tributes helped identify the most appropriatediseases.

Disease AttributesSelection of a disease includes consideration ofthe following:

• Existence of treatment guidelines

• Range of treatment modalities

• Consensus about the level of appropriatenessand effectiveness of care for the disorder

• Health care service utilization and factorsaffecting it

• Natural course and progression of the disease

Study Group AttributesThese attributes pertain only to the group ofpatients that will benefit from the intervention,in this case, all patients enrolled in the MedicaidMedallion program. In anticipation of expand-ing the VHOP pilot to the entire Virginia Medi-caid population, attributes of this larger group

(population attributes) also were needed. Studygroup criteria were developed based on 1993–1995 DMAS claims data and enrollment projec-tions. They include the following:

• The study group patients must be homo-genous (e.g., similar diagnoses, similar socio-economic and demographic factors).

• The eligibility requirements must be stable(i.e., no major changes had occurred or wereexpected to occur in the eligibility require-ments for that study group).

• The study group patients must have a signifi-cant claims history before implementation ofintervention.

• The study group patients should remain en-rolled in the program after the interventionhas been implemented so that long-term im-pact can be observed and documented.

Population AttributesThe selection of appropriate target diseases forthe project also depends on attributes of theoverall population (i.e., all Virginia Medicaidpatients with claims). For an effect to be demon-strated in this large, complex social population,the following attributes of the population mustbe present:

• There must be a large number of patients withthe disease, whose therapy can be improved(population characteristics).

• There must be disease outcomes that can besignificantly modified by application of ap-propriate therapy (as outlined in treatmentguidelines) and measured in the population(therapy characteristics).

• There must be no intervening or confoundingfactors that would interfere with interven-tions and/or assessment (outcomes character-istics).

These population attributes are further delin-eated in Table 1.

Other intervening factors critical to the successof a multidisciplinary program aimed to alterhealth care professionals’ behavior are: 1) thecredibility of the effort; and, 2) a consensuswithin the health care community regarding theneed for improvement.

3

plan the research component diligently, espe-cially because the ability to provide tangibleproof of effectiveness can strengthen providers’and legislators’ receptivity.

Nevertheless, research needs can be modified asthe educational intervention is implemented.Any new program designed to improve healthcare services should have a formal evaluation ofits cost-effectiveness periodically.

The research component of VHOP measured theeffectiveness of the program by measuring thehealth status of patients and medical care utili-zation. Different methods were used to measurethe components of effectiveness.

In the pilot program, VHOP used the followingmeasures of effectiveness:

• Patients’ general health status

• Patients’ asthma-specific health status

• Medical care utilization (inpatient hospitaland emergency room) with Medicaidpayments

These measures were assessed at baseline in theintervention and control communities. Healthstatus was measured by household survey withstandardized validated questionnaires(Landgraff, 1994; Ware et al., 1993) administeredvia telephone before the intervention and again6 months after the start of the intervention.Medical care utilization is being measured byusing DMAS claims data (Penberthy et al., 1994).

Measuring Health Status ofPatients Using a Household SurveyThe household survey did not reach as manypatients as desired. In order for the data to bestatistically valid for analysis, a 70% (minimum)response rate was targeted; the actual responserates were 45.7% for the Richmond area and 55.4%for the comparison area. A nonresponse biasanalysis is being conducted to assess the signifi-cance and magnitude of the low response rate.

A variety of factors may have contributed to thepoor household survey response rate, includingthe following:

• Medicaid populations usually are relativelymobile, which hampered attempts to locatepotential respondents.

• Medicaid recipients may have been mistrust-ful of an official survey, fearing that the infor-mation they provided might be used for pur-poses other than the ones stated and mightaffect receipt of their benefits.

• A financial incentive of $5 was offered torespondents for completing the survey, but itwas relatively modest even for a low-incomegroup.

Although increasing the financial incentivesshould increase response rates, the first twofactors may be more important and are not rem-edied easily. An intensive effort to conduct in-person interviews with respondents who do notparticipate in a telephone survey may improvethe overall response rate, but it would dramati-cally increase the costs (completion of an in-person interview costs over $100). If patient dataare going to be used as outcome measures infuture programs, the costs and feasibility ofacquiring those data must be carefully weighedagainst the utility of the information obtainedand alternative means of obtaining information(Fisher et al., 1996).

Measuring Medical CareUtilization by Using Claims DataThe use of claims data to measure outcomes isappealing. There are no extra costs for datacollection, and the data are directly relevant topolicy decisions. In actual practice, the VHOPprogram has encountered some significant diffi-culties in the use of Virginia’s DMAS claimsdata. Among them are the following:

• There are significant delays between the ren-dering of service and the availability of claimsdata.

• DMAS claims information and data set struc-tures are geared toward optimizing reim-bursement efficiency, not toward being usedprimarily as disease management and re-search tools. The storage formats of certainvariables made them more difficult to use in

10

Three Key ProceduresThree procedures were used in a systematiceffort to identify the most appropriate diseasesfor the pilot program:

1. Attributes identified in the theoretical frame-work were applied to the DMAS data.

2. A weighted algorithm was used for rankingthe prospective disease candidates in termsof their overall need for intervention.

3. A qualitative needs assessment (i.e., survey)was completed by the Virginia Medicaid DURCommittee.

These procedures are tools to help identify thepotential disease candidates. They were per-formed sequentially, and each procedure relied,in part, on the results from the previous proce-dure. The results from all three procedures weredrawn upon in ranking diseases for final selec-tion.

Procedure 1: Using Medicaid(DMAS) Data to Identify DiseaseSelection CandidatesThe attributes identified in the theoretical frame-work for disease selection were applied tosorted, ordered, and merged data from DMAS.This process yielded asthma, congestive heartfailure, epilepsy, otitis media, smoking cessa-tion, urinary tract infection, and diabetes asdisease candidates. The results were then pre-sented to the DMAS Prior Authorization/Vir-ginia Health Outcomes Partnership (PA/VHOP) Advisory Committee for considerationand evaluation. This Committee identified anexpanded disease list based on the theoreticalframework and their expert opinion. This listincluded arthritis, dementia, depression,dyslipidemia, hypertension, respiratory ail-ments, and peptic ulcer disease.

When the disease candidate pool was restrictedto the Medallion population (i.e., the studypopulation), asthma, epilepsy, urinary tractinfection, diabetes (during pregnancy), and oti-tis media formed a subset disease candidatepool. These diseases reflected the demograph-ics of Medallion patients (i.e., mainly mothers

with dependent children). In the overall non-Medicare–Medicaid population, additional can-didates included congestive heart failure, res-piratory ailments, dyslipidemia, and smokingcessation.

Several diseases and a risk factor were identi-fied as the final candidates for intervention byVHOP investigators. These included asthma,arthritis, otitis media, epilepsy, smoking cessa-tion, hypertension, and urinary tract infection.Because the Medallion program focuses onwomen and children, diabetes was substitutedfor arthritis. See Appendix for a detailed evalu-ation of specific diseases using actual data inVirginia.

Procedure 2: Weighted Scoring ofDiseasesAfter candidate diseases were identified by us-ing Procedure 1, the disease selection attributeswere simplified and weighted by the DeggeGroup, Ltd. The attributes were then analyzedin a matrix form (Table 2). Expert cliniciansproviding services to Medallion patients wereasked to rate the final disease candidates forspecific attributes using the following scoringsystem:

Scoring values:

4 Meets attribute easily

3 Meets attribute with some difficulty

2 Meets attribute with great difficulty

1 Unable to meet attribute

A sample matrix is presented in Table 2. In thisexample, pediatric asthma is the disease most inneed of intervention.

Procedure 3: Survey of MedicaidDUR Committee MembersVirginia Medicaid’s DUR Committee members(n=13) were surveyed regarding their opinionson the disease selection candidates. They wereasked to answer the survey questions as theypertained to the study population (i.e., the Me-dallion population).

4

tices and that they would recommend the pro-gram to a friend. Participants expressed interestin training in additional SPOKEs.

Insights on the Recruitment ProcessParticipants received a follow-up contact letterthat identified Medallion asthma patients in theirpractices who had emergency room visits orhospitalizations. There was a 1- to 2-monthinterval between class attendance and the con-tact letter. Finding ways to shorten this intervalwould benefit the participants.

The overall response indicates that there is sig-nificant provider interest in these types of classes.The extra efforts to recruit physicians with highernumbers of asthma patients did have some suc-cess.

If this program is to be funded as standardpolicy for a state, it would be desirable to maxi-mize the percentage of eligible physicians par-ticipating in sessions. A number of potentialapproaches can be used to increase participationrates. These include the following:

Recruitment• More aggressive recruiting of physicians (e.g.,

making in-person visits to practices, multipleletters from different sources per mailing)

• Increasing the number of opportunities totake the class, being careful to avoid schedul-ing conflicts with professional society meet-ings

Logistics• Offering classes at more attractive locations

(such as local resort areas although these maybe difficult to budget for a Medicaid agency)

• Having site-based classes for academic phy-sicians

• Altering the timing of the class (e.g., makingit shorter or making it into an overnight activ-ity and including activities for families)

Incentive• Incorporating a refundable registration de-

posit to increase physicians’ commitment toattend the class

The effort to recruit additional health care pro-fessionals to teach the class was of limited suc-

cess. Discomfort with the HUB communicationskill curriculum appeared to be a major barrierto participation in teaching. For a statewideprogram requiring multiple faculty, it would beuseful to consider having communication pro-fessionals teach the HUB skills, while reservingthe participation of health professionals for theSPOKE skills.

Insights on MultidisciplinaryTrainingIn this pilot intervention, training of only onegroup of health care providers (physicians) wasattempted. There are credible arguments forand against training multiple types of healthcare providers to achieve cost savings. A posi-tive aspect of simultaneous multidisciplinarytraining is that it makes the system more fail-safe. In other words, if the pharmacist, thenurse, or the physician alters his or her behavior,the patient may change his or her behavior. Italso reduces the likelihood of different groupsworking at cross-purposes (e.g., a pharmacistand a physician giving a patient conflicting in-formation).

Negative aspects of simultaneous multidisci-plinary training include creating a potentiallyunwieldy administrative structure and in-creased costs. An initial commitment to train-ing multiple groups of professionals also makesit difficult or impossible to evaluate the mar-ginal impact of training any particular group.

THE RESEARCHAlthough mandatory prior authorization pro-grams are in widespread use without proof oftheir cost-effectiveness (Soumerai et al., 1987), itis hoped that voluntary prior authorization pro-grams such as VHOP will be able to produceevidence of decreased costs and/or improvedoutcomes.

If this program is to be replicated in other states,the extent of the research component should bedetermined early in the program. If the researchcomponent grows unexpectedly, the costs andcomplexity also will increase, and the originalresearch design may impose restrictions onimplementation of the program. It is vital to

9

Table 1.Attributes Necessary for Intervention Effects to be Demonstrated in a Patient Population

Source: The Degge Group, Ltd.

ATTRIBUTE VALUE OF ATTRIBUTE LIMITATION OF ATTRIBUTE

POPULATION CHARACTERISTICS

Prevalence of disease Identifies population with Does not necessarily identify a homogeneoussufficient numbers for study population with respect to either therapy or

interventions

Prevalence of drug use Same as above High drug use may not always implyfeasibility of education or other interventions,or any need for changes

Population is a high-risk Potential for demonstrated Populations within many health care systemsgroup with measurable, effects is higher are smallercommon, serious outcomes

Receipt of other health May ensure likelihood of Utilization, outcomes are not alwayscare benefits from outside more complete care and detectable unless data are available from bothprograms better outcomes Medicaid and Medicare

THERAPY CHARACTERISTICS

Presence of generally Represents a useful basis for Some guidelines or treatment protocols haveagreed upon treatment interventions not been validated as to their positive effectprotocols or guidelines health outcomes in large populations

Presence of generally Helps with general Nonerecognized problems in acceptance by partnerstherapy, well documentedin the medical literature

Self-authorization of Represents a useful attribute May not be essentialguidelines for prior authorization

program initiatives

OUTCOMES CHARACTERISTICS

Measurability: Comes closest to measuring Health outcome may be affected by factors Morbidity (health) achievement of program goal other than changes in drug use

Measurability: Represents a tool to Some interventions may increase some types Utilization characterize both type and of utilization and/or not correlate well with

cost factors in outcomes medical outcomes

Outcomes due to nonoptimal Increases the likelihood of At present, there are limited data on whichtherapy usually lead to measuring changes outcomes meet this attributemeasurable and problematicoutcomes (e.g., interventionhas potential for high yield)

Outcomes should be Would make effort more Data are limited at presentpreventable in a sufficient generally acceptednumber of cases if therapyis improved

OTHER INTERVENING FACTORS

Therapeutic area is the Positive environment may assist Conflicted environment may result intarget of other intervening in the acceptability of the outcomes poorly correlated withfactors in the program and/ program to all partners interventionsor in the political environment

5

courses and how they would respond to theproposed VHOP program (Alan Newman Re-search, 1995). The participants were commu-nity-practice, Medallion primary care providersfrom the Richmond area. Important findingsfrom the physician focus groups included thefollowing:

Positive Motivational Factors for CMEParticipation• Schedule outside of normal office hours for

one-day courses, preferably on Saturdays

• Free courses

• CME Category I credits

• Convenient locations with plentiful parking

• Materials to take back to the physicians’ prac-tices

• Identities and credentials of presenters

Reactions to Proposed VHOP Intervention• Distrust and other negative attitudes towards

Medicaid officials

• Negative perceptions of the intellect and per-sonalities of Medicaid patients

• No perceived need to improve communica-tion skills

• Suspicion of “cookbook medicine”

The information from these sessions assisted theVHOP researchers in recruiting physicians forthe educational program. The focus group re-sults also were useful in helping tailor the pro-gram to the individual needs of physicians, al-though the needs related to participation in CMEwere easier to address than the deeper concernsabout Medicaid and about intervention validity.Some of the reported findings probably wouldbe replicated in other samples of physiciansnationwide, but there are unique needs andperspectives that would be best determined byquerying members of each target practitionerpopulation.

Recruiting PhysiciansRecruitment was targeted to Medallion primarycare physicians who had at least one Medallionpatient with asthma and preferably at least one

patient with an emergency room visit for asthma.A multiple-contact recruitment strategy wasused, as follows:

• The first contact was a letter endorsed by oneof the principal investigators and a primarycare physician with an established practice inthe recruitment community.

• After the first letter, a glossy 4-page colorbrochure promoting the program was mailed.

• After the brochure, a third letter was mailedthat contained copies of program endorse-ment letters from professional societies.

• The final recruitment contacts were telephonecalls and faxes for nonresponders who had ahigh volume of patients who visited emer-gency rooms for asthma treatment.

The course was 6 hours long and was offered on6 occasions. The average lead time for recruit-ment was 3 to 4 weeks. The participation incen-tives offered to physicians were free CME Cat-egory I credits, free asthma education teachingkits and materials, and mandatory prior autho-rization waivers.

The response rate was 17% for the physicianswho had patients with emergency room visitsfor asthma, and 15% of the physicians whoseasthma patients did not have emergency roomvisits also participated in the program. Physi-cians with a higher volume of patients whomade emergency room visits were somewhatmore likely to participate in the program. Theresponse was much better for community physi-cians than for academic physicians. Approxi-mately 20% of the community participants re-sponded, but only 3% of the academic physi-cians responded.

Evaluation forms were provided to participantsimmediately after the classes. There were ques-tions on the factual knowledge of the HUB andSPOKE curriculum, on global ratings of the effec-tiveness and probable impact of the workshop,and on specific components of the classes. Re-sponses were obtained from approximately 85%of the participants. Analysis of the data revealedthat the respondents had acquired an excellentknowledge of the factual content of the curricula.Most of the respondents predicted that the pro-gram would have a positive impact on their prac-

8

The survey (a qualitative analytic instrument)asked for information relating to the followingfour parameters:

• Prevalence of disease

• Prevalence of drug use

• Incidence of the disease within thepopulation

• Availability of a viable intervention

Otitis media was the disease most often citedby the committee members, followed byasthma and epilepsy, then hypertension andurinary tract infection. Most comments by therespondents were associated with otitis me-dia. Four reviewers pointed to the high occur-rence of otitis media in the Medallion popula-tion.

Asthma was the second most cited disease andalso was noted in terms of the frequency ofoccurrence, particularly the increase in emer-gency room visits and hospitalizations. It alsomet many of the other attributes identified in thetheoretical framework. During the initiation ofthe pilot study, summary data from the VirginiaMedicaid program led to the conclusion thatasthma was the most useful disease to serve as amodel for the pilot development. Asthma isincreasing in frequency, has been the subject ofNIH and other guidelines, and accounted for$6.9 million of the $159 million total VirginiaMedicaid expenditures.

THE INTERVENTION: HUBAND SPOKE MODELThe VHOP model is designed around the HUBand SPOKE model; the HUB consists of basiccommunication skills, and the SPOKE consistsof a clinical treatment plan for a specific disease.VHOP was developed as an educational inter-vention that would improve both HUB andSPOKE skills in health care providers. Thistraining is intended to decrease inappropriateprescribing patterns and to increase patient ad-herence to the treatment plan, both of whichcontribute to improved patient outcomes(Kaplan, Greenfield & Ware, 1989; Soumerai &Avorn, 1990).

The curriculum was designed to be offered as acourse suitable for CME Category I accredita-tion; course accreditation was provided by theMedical College of Virginia. Both the HUB andthe SPOKE materials were developed with theassistance of COMSORT, an independent com-pany specializing in educational program de-velopment for health care professionals.

Having curricula of sufficient rigor for CMEcertification is a critical incentive to encourageproviders’ receptivity and participation. Be-cause it is not feasible for a state-level agency tobecome accredited to provide CME, it is neces-sary for an accredited university or an educa-tional organization to be incorporated into theprogram and to be responsible for developingthe content of the intervention.

The expertise required to develop the HUB cur-ricula is not likely to be found in state govern-ment agencies. It is appropriate to outsourcethis part of the program. Many private organi-zations and university programs have devel-oped curricula that can be used.

However, some agencies in state government(e.g., health departments) can develop SPOKEcurricula. In some cases, depending on theworkload of the department and the ease ofarranging interagency contracting agreements,this may not be feasible. It also is appropriate tooutsource this part of the program. In eithercase, the input of both the primary care and thespecialty care medical communities is impor-tant when developing guideline materials. Phar-maceutical companies also may be willing toprovide demonstration materials, such as asthmainhalers and spacers. Written materials frompharmaceutical companies also can be helpfulwhen developing SPOKE curricula; however,care must be taken to avoid using marketingmaterials for specific products.

Using Physician Focus Groups forFeedback on the InterventionAfter the pilot intervention component was de-veloped and before its implementation, physi-cian focus groups were conducted. The purposeof these groups was to determine what factorswere important to physicians in choosing CME

6

Table 2.Sample Matrix of Weighted Population Attributes Applied to Virginia Medicaid Data on Selected Diseases

1. The total weight assigned to each attribute in each of the disease examples is multiplied by the scoring value (1–4) assignedby the respondent.

DISEASE PEDIATRIC ASTHMA ADULT ASTHMA ARTHRITIS

Total Scoring Weighted Scoring Weighted Scoring Weighted Weight1 Value Score Value Score Value Score

POPULATION CHARACTERISITICS

Prevalence of disease 10 4 40 3 30 3 30

Prevalence of drug use 10 4 40 4 40 4 40

Population is a high-risk group withmeasurable, common, serious outcomes 20 4 80 4 80 2 40

Population receives other healthcare benefits (e.g., Medicare) -10 0 0 0 0 4 -40

THERAPY AREA CHARACTERISTICS

Presence of generally agreed upontreatment protocols or guidelines 10 4 40 4 40 2 20

Presence of generally recognizedproblems in therapy, welldocumented in the literature 15 4 60 4 60 4 60

Potential for self-authorization ofguidelines 8 3 24 3 24 3 24

Measurability: morbidity (health) 10 4 40 4 40 4 40

OUTCOMES CHARACTERISTICS

Measurability: utilization of services 8 4 32 4 32 4 32

Outcomes due to nonoptimal therapyusually lead to measurable andproblematic outcomes (e.g.,intervention has potential for high yield) 20 3 60 3 60 2 40

Outcomes should be preventable in asufficient number of cases if therapyis improved 15 4 60 4 60 2 30

OTHER INTERVENING FACTORS

Therapeutic area is the target of otherintervening factors in the programand/or in the political environment -10 0 0 0 0 0 0

OVERALL SCORE 476 468 316

7

The survey (a qualitative analytic instrument)asked for information relating to the followingfour parameters:

• Prevalence of disease

• Prevalence of drug use

• Incidence of the disease within thepopulation

• Availability of a viable intervention

Otitis media was the disease most often citedby the committee members, followed byasthma and epilepsy, then hypertension andurinary tract infection. Most comments by therespondents were associated with otitis me-dia. Four reviewers pointed to the high occur-rence of otitis media in the Medallion popula-tion.

Asthma was the second most cited disease andalso was noted in terms of the frequency ofoccurrence, particularly the increase in emer-gency room visits and hospitalizations. It alsomet many of the other attributes identified in thetheoretical framework. During the initiation ofthe pilot study, summary data from the VirginiaMedicaid program led to the conclusion thatasthma was the most useful disease to serve as amodel for the pilot development. Asthma isincreasing in frequency, has been the subject ofNIH and other guidelines, and accounted for$6.9 million of the $159 million total VirginiaMedicaid expenditures.

THE INTERVENTION: HUBAND SPOKE MODELThe VHOP model is designed around the HUBand SPOKE model; the HUB consists of basiccommunication skills, and the SPOKE consistsof a clinical treatment plan for a specific disease.VHOP was developed as an educational inter-vention that would improve both HUB andSPOKE skills in health care providers. Thistraining is intended to decrease inappropriateprescribing patterns and to increase patient ad-herence to the treatment plan, both of whichcontribute to improved patient outcomes(Kaplan, Greenfield & Ware, 1989; Soumerai &Avorn, 1990).

The curriculum was designed to be offered as acourse suitable for CME Category I accredita-tion; course accreditation was provided by theMedical College of Virginia. Both the HUB andthe SPOKE materials were developed with theassistance of COMSORT, an independent com-pany specializing in educational program de-velopment for health care professionals.

Having curricula of sufficient rigor for CMEcertification is a critical incentive to encourageproviders’ receptivity and participation. Be-cause it is not feasible for a state-level agency tobecome accredited to provide CME, it is neces-sary for an accredited university or an educa-tional organization to be incorporated into theprogram and to be responsible for developingthe content of the intervention.

The expertise required to develop the HUB cur-ricula is not likely to be found in state govern-ment agencies. It is appropriate to outsourcethis part of the program. Many private organi-zations and university programs have devel-oped curricula that can be used.

However, some agencies in state government(e.g., health departments) can develop SPOKEcurricula. In some cases, depending on theworkload of the department and the ease ofarranging interagency contracting agreements,this may not be feasible. It also is appropriate tooutsource this part of the program. In eithercase, the input of both the primary care and thespecialty care medical communities is impor-tant when developing guideline materials. Phar-maceutical companies also may be willing toprovide demonstration materials, such as asthmainhalers and spacers. Written materials frompharmaceutical companies also can be helpfulwhen developing SPOKE curricula; however,care must be taken to avoid using marketingmaterials for specific products.

Using Physician Focus Groups forFeedback on the InterventionAfter the pilot intervention component was de-veloped and before its implementation, physi-cian focus groups were conducted. The purposeof these groups was to determine what factorswere important to physicians in choosing CME

6

Table 2.Sample Matrix of Weighted Population Attributes Applied to Virginia Medicaid Data on Selected Diseases

1. The total weight assigned to each attribute in each of the disease examples is multiplied by the scoring value (1–4) assignedby the respondent.

DISEASE PEDIATRIC ASTHMA ADULT ASTHMA ARTHRITIS

Total Scoring Weighted Scoring Weighted Scoring Weighted Weight1 Value Score Value Score Value Score

POPULATION CHARACTERISITICS

Prevalence of disease 10 4 40 3 30 3 30

Prevalence of drug use 10 4 40 4 40 4 40

Population is a high-risk group withmeasurable, common, serious outcomes 20 4 80 4 80 2 40

Population receives other healthcare benefits (e.g., Medicare) -10 0 0 0 0 4 -40

THERAPY AREA CHARACTERISTICS

Presence of generally agreed upontreatment protocols or guidelines 10 4 40 4 40 2 20

Presence of generally recognizedproblems in therapy, welldocumented in the literature 15 4 60 4 60 4 60

Potential for self-authorization ofguidelines 8 3 24 3 24 3 24

Measurability: morbidity (health) 10 4 40 4 40 4 40

OUTCOMES CHARACTERISTICS

Measurability: utilization of services 8 4 32 4 32 4 32

Outcomes due to nonoptimal therapyusually lead to measurable andproblematic outcomes (e.g.,intervention has potential for high yield) 20 3 60 3 60 2 40

Outcomes should be preventable in asufficient number of cases if therapyis improved 15 4 60 4 60 2 30

OTHER INTERVENING FACTORS

Therapeutic area is the target of otherintervening factors in the programand/or in the political environment -10 0 0 0 0 0 0

OVERALL SCORE 476 468 316

7

Table 1.Attributes Necessary for Intervention Effects to be Demonstrated in a Patient Population

Source: The Degge Group, Ltd.

ATTRIBUTE VALUE OF ATTRIBUTE LIMITATION OF ATTRIBUTE

POPULATION CHARACTERISTICS

Prevalence of disease Identifies population with Does not necessarily identify a homogeneoussufficient numbers for study population with respect to either therapy or

interventions

Prevalence of drug use Same as above High drug use may not always implyfeasibility of education or other interventions,or any need for changes

Population is a high-risk Potential for demonstrated Populations within many health care systemsgroup with measurable, effects is higher are smallercommon, serious outcomes

Receipt of other health May ensure likelihood of Utilization, outcomes are not alwayscare benefits from outside more complete care and detectable unless data are available from bothprograms better outcomes Medicaid and Medicare

THERAPY CHARACTERISTICS

Presence of generally Represents a useful basis for Some guidelines or treatment protocols haveagreed upon treatment interventions not been validated as to their positive effectprotocols or guidelines health outcomes in large populations

Presence of generally Helps with general Nonerecognized problems in acceptance by partnerstherapy, well documentedin the medical literature

Self-authorization of Represents a useful attribute May not be essentialguidelines for prior authorization

program initiatives

OUTCOMES CHARACTERISTICS

Measurability: Comes closest to measuring Health outcome may be affected by factors Morbidity (health) achievement of program goal other than changes in drug use

Measurability: Represents a tool to Some interventions may increase some types Utilization characterize both type and of utilization and/or not correlate well with

cost factors in outcomes medical outcomes

Outcomes due to nonoptimal Increases the likelihood of At present, there are limited data on whichtherapy usually lead to measuring changes outcomes meet this attributemeasurable and problematicoutcomes (e.g., interventionhas potential for high yield)

Outcomes should be Would make effort more Data are limited at presentpreventable in a sufficient generally acceptednumber of cases if therapyis improved

OTHER INTERVENING FACTORS

Therapeutic area is the Positive environment may assist Conflicted environment may result intarget of other intervening in the acceptability of the outcomes poorly correlated withfactors in the program and/ program to all partners interventionsor in the political environment

5

courses and how they would respond to theproposed VHOP program (Alan Newman Re-search, 1995). The participants were commu-nity-practice, Medallion primary care providersfrom the Richmond area. Important findingsfrom the physician focus groups included thefollowing:

Positive Motivational Factors for CMEParticipation• Schedule outside of normal office hours for

one-day courses, preferably on Saturdays

• Free courses

• CME Category I credits

• Convenient locations with plentiful parking

• Materials to take back to the physicians’ prac-tices

• Identities and credentials of presenters

Reactions to Proposed VHOP Intervention• Distrust and other negative attitudes towards

Medicaid officials

• Negative perceptions of the intellect and per-sonalities of Medicaid patients

• No perceived need to improve communica-tion skills

• Suspicion of “cookbook medicine”

The information from these sessions assisted theVHOP researchers in recruiting physicians forthe educational program. The focus group re-sults also were useful in helping tailor the pro-gram to the individual needs of physicians, al-though the needs related to participation in CMEwere easier to address than the deeper concernsabout Medicaid and about intervention validity.Some of the reported findings probably wouldbe replicated in other samples of physiciansnationwide, but there are unique needs andperspectives that would be best determined byquerying members of each target practitionerpopulation.

Recruiting PhysiciansRecruitment was targeted to Medallion primarycare physicians who had at least one Medallionpatient with asthma and preferably at least one

patient with an emergency room visit for asthma.A multiple-contact recruitment strategy wasused, as follows:

• The first contact was a letter endorsed by oneof the principal investigators and a primarycare physician with an established practice inthe recruitment community.

• After the first letter, a glossy 4-page colorbrochure promoting the program was mailed.

• After the brochure, a third letter was mailedthat contained copies of program endorse-ment letters from professional societies.

• The final recruitment contacts were telephonecalls and faxes for nonresponders who had ahigh volume of patients who visited emer-gency rooms for asthma treatment.

The course was 6 hours long and was offered on6 occasions. The average lead time for recruit-ment was 3 to 4 weeks. The participation incen-tives offered to physicians were free CME Cat-egory I credits, free asthma education teachingkits and materials, and mandatory prior autho-rization waivers.

The response rate was 17% for the physicianswho had patients with emergency room visitsfor asthma, and 15% of the physicians whoseasthma patients did not have emergency roomvisits also participated in the program. Physi-cians with a higher volume of patients whomade emergency room visits were somewhatmore likely to participate in the program. Theresponse was much better for community physi-cians than for academic physicians. Approxi-mately 20% of the community participants re-sponded, but only 3% of the academic physi-cians responded.

Evaluation forms were provided to participantsimmediately after the classes. There were ques-tions on the factual knowledge of the HUB andSPOKE curriculum, on global ratings of the effec-tiveness and probable impact of the workshop,and on specific components of the classes. Re-sponses were obtained from approximately 85%of the participants. Analysis of the data revealedthat the respondents had acquired an excellentknowledge of the factual content of the curricula.Most of the respondents predicted that the pro-gram would have a positive impact on their prac-

8

Three Key ProceduresThree procedures were used in a systematiceffort to identify the most appropriate diseasesfor the pilot program:

1. Attributes identified in the theoretical frame-work were applied to the DMAS data.

2. A weighted algorithm was used for rankingthe prospective disease candidates in termsof their overall need for intervention.

3. A qualitative needs assessment (i.e., survey)was completed by the Virginia Medicaid DURCommittee.

These procedures are tools to help identify thepotential disease candidates. They were per-formed sequentially, and each procedure relied,in part, on the results from the previous proce-dure. The results from all three procedures weredrawn upon in ranking diseases for final selec-tion.

Procedure 1: Using Medicaid(DMAS) Data to Identify DiseaseSelection CandidatesThe attributes identified in the theoretical frame-work for disease selection were applied tosorted, ordered, and merged data from DMAS.This process yielded asthma, congestive heartfailure, epilepsy, otitis media, smoking cessa-tion, urinary tract infection, and diabetes asdisease candidates. The results were then pre-sented to the DMAS Prior Authorization/Vir-ginia Health Outcomes Partnership (PA/VHOP) Advisory Committee for considerationand evaluation. This Committee identified anexpanded disease list based on the theoreticalframework and their expert opinion. This listincluded arthritis, dementia, depression,dyslipidemia, hypertension, respiratory ail-ments, and peptic ulcer disease.

When the disease candidate pool was restrictedto the Medallion population (i.e., the studypopulation), asthma, epilepsy, urinary tractinfection, diabetes (during pregnancy), and oti-tis media formed a subset disease candidatepool. These diseases reflected the demograph-ics of Medallion patients (i.e., mainly mothers

with dependent children). In the overall non-Medicare–Medicaid population, additional can-didates included congestive heart failure, res-piratory ailments, dyslipidemia, and smokingcessation.

Several diseases and a risk factor were identi-fied as the final candidates for intervention byVHOP investigators. These included asthma,arthritis, otitis media, epilepsy, smoking cessa-tion, hypertension, and urinary tract infection.Because the Medallion program focuses onwomen and children, diabetes was substitutedfor arthritis. See Appendix for a detailed evalu-ation of specific diseases using actual data inVirginia.

Procedure 2: Weighted Scoring ofDiseasesAfter candidate diseases were identified by us-ing Procedure 1, the disease selection attributeswere simplified and weighted by the DeggeGroup, Ltd. The attributes were then analyzedin a matrix form (Table 2). Expert cliniciansproviding services to Medallion patients wereasked to rate the final disease candidates forspecific attributes using the following scoringsystem:

Scoring values:

4 Meets attribute easily

3 Meets attribute with some difficulty

2 Meets attribute with great difficulty

1 Unable to meet attribute

A sample matrix is presented in Table 2. In thisexample, pediatric asthma is the disease most inneed of intervention.

Procedure 3: Survey of MedicaidDUR Committee MembersVirginia Medicaid’s DUR Committee members(n=13) were surveyed regarding their opinionson the disease selection candidates. They wereasked to answer the survey questions as theypertained to the study population (i.e., the Me-dallion population).

4

tices and that they would recommend the pro-gram to a friend. Participants expressed interestin training in additional SPOKEs.

Insights on the Recruitment ProcessParticipants received a follow-up contact letterthat identified Medallion asthma patients in theirpractices who had emergency room visits orhospitalizations. There was a 1- to 2-monthinterval between class attendance and the con-tact letter. Finding ways to shorten this intervalwould benefit the participants.

The overall response indicates that there is sig-nificant provider interest in these types of classes.The extra efforts to recruit physicians with highernumbers of asthma patients did have some suc-cess.

If this program is to be funded as standardpolicy for a state, it would be desirable to maxi-mize the percentage of eligible physicians par-ticipating in sessions. A number of potentialapproaches can be used to increase participationrates. These include the following:

Recruitment• More aggressive recruiting of physicians (e.g.,

making in-person visits to practices, multipleletters from different sources per mailing)

• Increasing the number of opportunities totake the class, being careful to avoid schedul-ing conflicts with professional society meet-ings

Logistics• Offering classes at more attractive locations

(such as local resort areas although these maybe difficult to budget for a Medicaid agency)

• Having site-based classes for academic phy-sicians

• Altering the timing of the class (e.g., makingit shorter or making it into an overnight activ-ity and including activities for families)

Incentive• Incorporating a refundable registration de-

posit to increase physicians’ commitment toattend the class

The effort to recruit additional health care pro-fessionals to teach the class was of limited suc-

cess. Discomfort with the HUB communicationskill curriculum appeared to be a major barrierto participation in teaching. For a statewideprogram requiring multiple faculty, it would beuseful to consider having communication pro-fessionals teach the HUB skills, while reservingthe participation of health professionals for theSPOKE skills.

Insights on MultidisciplinaryTrainingIn this pilot intervention, training of only onegroup of health care providers (physicians) wasattempted. There are credible arguments forand against training multiple types of healthcare providers to achieve cost savings. A posi-tive aspect of simultaneous multidisciplinarytraining is that it makes the system more fail-safe. In other words, if the pharmacist, thenurse, or the physician alters his or her behavior,the patient may change his or her behavior. Italso reduces the likelihood of different groupsworking at cross-purposes (e.g., a pharmacistand a physician giving a patient conflicting in-formation).

Negative aspects of simultaneous multidisci-plinary training include creating a potentiallyunwieldy administrative structure and in-creased costs. An initial commitment to train-ing multiple groups of professionals also makesit difficult or impossible to evaluate the mar-ginal impact of training any particular group.

THE RESEARCHAlthough mandatory prior authorization pro-grams are in widespread use without proof oftheir cost-effectiveness (Soumerai et al., 1987), itis hoped that voluntary prior authorization pro-grams such as VHOP will be able to produceevidence of decreased costs and/or improvedoutcomes.

If this program is to be replicated in other states,the extent of the research component should bedetermined early in the program. If the researchcomponent grows unexpectedly, the costs andcomplexity also will increase, and the originalresearch design may impose restrictions onimplementation of the program. It is vital to

9

cially for some diagnoses that may result inmisclassifications when selecting individualswith particular disease states or identifying clini-cal outcomes. Longitudinal medical historiesare available because of continuous data collec-tion. However, this advantage only applies tothose individuals who remain in the system;patients can be lost to follow-up because ofeligibility changes. DMAS lacks data on vari-ables that may be important mediators of treat-ment effects and potential confounding vari-ables (e.g., smoking status, dietary intake, alco-hol consumption). These missing data indicatethe importance of accessing primary medicalrecords. Some potentially important sources ofmedical information are missing as well, includ-ing inpatient medication use, use of over-the-counter drugs, blood pressure readings, air peakflow readings, and other signs of disease. Nev-ertheless, claims data do provide indirect infor-mation on outcomes through proxy measures.

The Framework for Disease SelectionBased on the DMAS claims data, various at-tributes were developed for: 1) selection of dis-eases; 2) selection of patients for the study group;and, 3) characteristics of the population fromwhich the study group is selected. These at-tributes helped identify the most appropriatediseases.

Disease AttributesSelection of a disease includes consideration ofthe following:

• Existence of treatment guidelines

• Range of treatment modalities

• Consensus about the level of appropriatenessand effectiveness of care for the disorder

• Health care service utilization and factorsaffecting it

• Natural course and progression of the disease

Study Group AttributesThese attributes pertain only to the group ofpatients that will benefit from the intervention,in this case, all patients enrolled in the MedicaidMedallion program. In anticipation of expand-ing the VHOP pilot to the entire Virginia Medi-caid population, attributes of this larger group

(population attributes) also were needed. Studygroup criteria were developed based on 1993–1995 DMAS claims data and enrollment projec-tions. They include the following:

• The study group patients must be homo-genous (e.g., similar diagnoses, similar socio-economic and demographic factors).

• The eligibility requirements must be stable(i.e., no major changes had occurred or wereexpected to occur in the eligibility require-ments for that study group).

• The study group patients must have a signifi-cant claims history before implementation ofintervention.

• The study group patients should remain en-rolled in the program after the interventionhas been implemented so that long-term im-pact can be observed and documented.

Population AttributesThe selection of appropriate target diseases forthe project also depends on attributes of theoverall population (i.e., all Virginia Medicaidpatients with claims). For an effect to be demon-strated in this large, complex social population,the following attributes of the population mustbe present:

• There must be a large number of patients withthe disease, whose therapy can be improved(population characteristics).

• There must be disease outcomes that can besignificantly modified by application of ap-propriate therapy (as outlined in treatmentguidelines) and measured in the population(therapy characteristics).

• There must be no intervening or confoundingfactors that would interfere with interven-tions and/or assessment (outcomes character-istics).

These population attributes are further delin-eated in Table 1.

Other intervening factors critical to the successof a multidisciplinary program aimed to alterhealth care professionals’ behavior are: 1) thecredibility of the effort; and, 2) a consensuswithin the health care community regarding theneed for improvement.

3

plan the research component diligently, espe-cially because the ability to provide tangibleproof of effectiveness can strengthen providers’and legislators’ receptivity.

Nevertheless, research needs can be modified asthe educational intervention is implemented.Any new program designed to improve healthcare services should have a formal evaluation ofits cost-effectiveness periodically.

The research component of VHOP measured theeffectiveness of the program by measuring thehealth status of patients and medical care utili-zation. Different methods were used to measurethe components of effectiveness.

In the pilot program, VHOP used the followingmeasures of effectiveness:

• Patients’ general health status

• Patients’ asthma-specific health status

• Medical care utilization (inpatient hospitaland emergency room) with Medicaidpayments

These measures were assessed at baseline in theintervention and control communities. Healthstatus was measured by household survey withstandardized validated questionnaires(Landgraff, 1994; Ware et al., 1993) administeredvia telephone before the intervention and again6 months after the start of the intervention.Medical care utilization is being measured byusing DMAS claims data (Penberthy et al., 1994).

Measuring Health Status ofPatients Using a Household SurveyThe household survey did not reach as manypatients as desired. In order for the data to bestatistically valid for analysis, a 70% (minimum)response rate was targeted; the actual responserates were 45.7% for the Richmond area and 55.4%for the comparison area. A nonresponse biasanalysis is being conducted to assess the signifi-cance and magnitude of the low response rate.

A variety of factors may have contributed to thepoor household survey response rate, includingthe following:

• Medicaid populations usually are relativelymobile, which hampered attempts to locatepotential respondents.

• Medicaid recipients may have been mistrust-ful of an official survey, fearing that the infor-mation they provided might be used for pur-poses other than the ones stated and mightaffect receipt of their benefits.

• A financial incentive of $5 was offered torespondents for completing the survey, but itwas relatively modest even for a low-incomegroup.

Although increasing the financial incentivesshould increase response rates, the first twofactors may be more important and are not rem-edied easily. An intensive effort to conduct in-person interviews with respondents who do notparticipate in a telephone survey may improvethe overall response rate, but it would dramati-cally increase the costs (completion of an in-person interview costs over $100). If patient dataare going to be used as outcome measures infuture programs, the costs and feasibility ofacquiring those data must be carefully weighedagainst the utility of the information obtainedand alternative means of obtaining information(Fisher et al., 1996).

Measuring Medical CareUtilization by Using Claims DataThe use of claims data to measure outcomes isappealing. There are no extra costs for datacollection, and the data are directly relevant topolicy decisions. In actual practice, the VHOPprogram has encountered some significant diffi-culties in the use of Virginia’s DMAS claimsdata. Among them are the following:

• There are significant delays between the ren-dering of service and the availability of claimsdata.

• DMAS claims information and data set struc-tures are geared toward optimizing reim-bursement efficiency, not toward being usedprimarily as disease management and re-search tools. The storage formats of certainvariables made them more difficult to use in

10

VHOP PROGRAMSTRUCTUREThe VHOP program has three main compo-nents: the disease selection process, the interven-tion, and the research to determine the program’seffectiveness. First a process was developed toidentify the most appropriate diseases for theprogram. Once a disease is selected, an inter-vention is developed. The intervention is de-signed around the HUB and SPOKE model ofhealth care provider-patient communications.The HUB consists of basic communication skills,and the SPOKE consists of treatment guidelinesfor a specific disease. The research componentevaluates the economic, service utilization, pa-tient and provider satisfaction, and outcomes ofthe intervention. Each of these three compo-nents is discussed in this publication.

THE DISEASE SELECTIONPROCESSA key element of VHOP was to develop a sys-tematic process to identify the most appropriatediseases for the program. The process had toutilize both empirical data from the MedicaidManagement Information System (i.e., DMAS’sclaims data) and clinical knowledge about thepatterns of disease and treatment (i.e., a theoreti-cal framework) to compare attributes of diseasecandidates for intervention.

Disease candidates were chosen by identifyingsources of information about them and then devel-oping standard methods for using this informa-tion. Primary source data were provided by DMAS,and the patient-specific information came directlyfrom claims records and enrollment files. Informa-tion describing medical providers came fromDMAS provider record files. All of this informa-tion was confidential and used for research pur-poses only. This highly specific information wasthe basis for most of the decisions related to selec-tion of a disease SPOKE. When appropriate, theprimary data were supplemented with data fromother sources (e.g., other studies and surveys).

Information about disease treatment was obtainedfrom national consensus-based sources (e.g., NIH,AHCPR). Reviews of treatment guidelines weresought at the state and local levels to ensure

applicability to community practice. The expertopinions of the clinicians providing services tothe Medallion patient population were used toassess community perceptions of the areas ofneed, problems, practical barriers, and potentialfor improvement.

Final disease selection was made by a grouprepresenting all parties working on the imple-mentation of VHOP: the Williamson Institute,the Schools of Pharmacy, Medicine, and Nurs-ing of the Medical College of Virginia, DMAS,the Degge Group, Ltd., and the National Phar-maceutical Council. The decision was based onthe results of the many analyses described in thissection; key factors were the prevalence and costof the disease to Virginia Medicaid, and thepotential for improving patient outcomes inMedicaid managed care programs. All repre-sentatives in the group agreed to the selection ofasthma as the first disease in the pilot program.

The Value of Empirical Data toMeasure OutcomesIdeally, successful outcomes management re-quires detailed population-based data, data onhealth care service use, and clinical outcomes.Service use and cost data are typically availablefrom medical claims information traditionallycollected in Medicaid populations for the past 20years or more. For each Medicaid recipient inMedallion, billing data include age, sex, race,county, inpatient hospitalization and outpatientdiagnoses, outpatient drugs and procedures (e.g.,laboratory, radiologic), and number and costs ofphysician office and emergency department vis-its. Pharmacy records—including prescriptionname, amount, dose dispensed, and estimateddays of therapy—also are collected. In an inter-nal analysis of pharmacy usage among Medal-lion recipients (n=2,399), 93% used only one ortwo pharmacies, whereas only 7% used three ormore pharmacies (Miller, 1994). Matching pa-tient data is facilitated by the lack of variance inthe number of pharmacies, so assessing compli-ance with certain drug regimens is possible.

As is true for all claims data, these data arecollected for billing and so are free of somebiases (e.g., interview, recall, reporting). How-ever, the data are of uncertain validity, espe-

2

research, although the formats were cost-ef-fective for storing billing data.

• Medicaid claims data sets are very large, anddata analysis strained the technical capacitiesof the DMAS computer systems. DMAS storesmost of its data on tapes, which results inslower processing.

• Researchers had limited access to the datadue to competing demands for administra-tive and other research purposes. For ex-ample, access to DMAS storage tapes waslimited to overnight hours.

• There has been no independent verificationof the validity of the data.

VHOP is developing a comprehensive data dic-tionary and has migrated most of the relevantdata to a supercomputer at VCU with adequatedata storage capacities. There should be fewerproblems with data analysis in the future. How-ever, the time involved in structuring and pro-gramming data analyses should not change sub-stantially in the future.

The use of Medicaid claims data in this type ofresearch is probably feasible for other states, butthe strengths and limitations of claims data mustbe considered so that appropriate measures arechosen. Each state’s Medicaid claims system isunique. Ideally, data analysis should be per-formed by persons who have extensive experi-ence using the data and have a firm grasp of theidiosyncrasies of the data set. The computersystem capabilities should be assessed for theirsuitability for research data analysis. If it isfinancially feasible, investment in newer hard-ware technologies, such as disk drive systemswith high storage capacity, should be consid-ered if they are not already in place.

SUMMARYThe VHOP effort is designed to improve healthoutcomes among Medicaid patients enrolled inthe Medallion program by improving the com-munication skills of physicians and other healthcare providers and by promoting recently estab-lished, innovative clinical guidelines throughpeer education. The pilot program trained phy-

sicians only on the treatment of asthma in Cen-tral Virginia. The next phase is being imple-mented, in which VHOP will: 1) expand theasthma training for physicians statewide; 2) in-corporate pharmacists as part of the asthmatraining; 3) implement training on congestiveheart failure for physicians statewide; and, 4)develop education materials for diabetes melli-tus and schizophrenia.

The methods of the disease selection process in-volved weaving together empirical data and atheoretical framework to compare relevant dis-ease-state candidates for intervention. Many at-tributes of a disease were considered simulta-neously (e.g., the incidence of poor and costlyoutcomes, the availability of established guide-lines, a structure for enacting the intervention, anda reliable and valid mechanism for evaluation).Therefore, in order to identify the most appropri-ate disease for the pilot program, three procedureswere used: 1) attributes identified in the theoreticalframework were applied to the DMAS data; 2)prospective disease candidates were ranked bytheir ability to meet key attributes; and, 3) opinionsof the Virginia Medicaid DUR Committee mem-bers were determined by survey.

The results of these analyses indicated thatasthma was an appropriate first choice for thepilot intervention. At the time this analyses wascompleted, it was thought that disease stateswarranting consideration for subsequent inter-vention included otitis media and epilepsy.However, subsequent analyses indicate that forthe Virginia Medicaid population, congestiveheart disease, diabetes mellitus, and schizophre-nia are better candidates.

VHOP is a program where the state partnerswith its health professionals. This partnershipnot only builds a sense of trust between theseparties but also relies on the model of profes-sional responsibility for successful implemen-tation. The VHOP researchers and VirginiaMedicaid continue to test this partnership pro-gram and carefully evaluate its effects. Prelimi-nary findings show significant improvementsin relationships between the state and the healthcare providers and cost savings among a tar-geted group of patients to date. This partner-ship model may be useful to other states.

11

INTRODUCTIONIn early 1996, approximately 7.8 million peopleacross the country (23% of all Medicaid recipi-ents) were enrolled in managed care programs(Physician Payment Review Commission, 1996).Nearly 3 million people were enrolled in pri-mary care case management systems. In mostof these systems, physicians are paid case man-agement fees (typically $3 per recipient permonth) in addition to their regular fee-for-ser-vice payments for the primary care servicesthey provide.

Virginia established the Medallion program, itsprimary care case management program, in theearly 1990s. This program was implementedunder a waiver granted by HCFA in an effort tocontrol costs. By 1996, nearly 500,000 recipientswere enrolled with primary care physicians ascase managers in Virginia Medicaid statewide.As the primary care case management programwas being implemented in the early 1990s, thestate also was looking for ways to control risingdrug costs in the Medicaid program.

Although a number of mandatory prior autho-rization programs were already in place in otherstates, no voluntary prior authorization pro-gram had been implemented. Mandatory priorauthorization programs have been shown to beexpensive to administer and have not proven toresult in overall Medicaid cost savings. Vir-ginia had some success with a voluntary priorauthorization program in conjunction with theVirginia Pharmacy Association to reduce theutilization of H2-antagonists and save costs(American Pharmacy, 1994).

VHOP is a voluntary prior authorization pro-gram in which the physician self-authorizesprescriptions. This program has helped Vir-ginia Medicaid decrease service utilization, aswell as maintain and invest in its relationshipwith Medicaid health care providers.

VHOP PROGRAM HISTORYIn May 1993, VHOP was proposed by Virginia’sDepartment of Medical Assistance (DMAS),also commonly known as Virginia Medicaid.The project was directed at health care provid-

ers and patients enrolled in Virginia Medicaid’sMedallion program.

The Medallion program was developed to offera primary care physician to each enrollee in thehopes of improving access while lowering Med-icaid costs (Physician Payment Review Com-mission, 1996; Wilson, 1995). The primary careprovider agrees to become the case manager orgatekeeper for the enrollee. Included in thisresponsibility is the obligation to be available toprovide and/or authorize all nonemergency carefor the enrollee.

The primary goals of the VHOP–Medallion col-laboration are:

• to improve patient health outcomes by im-proving the general and disease-specific com-munication skills of physicians and otherhealth care providers; and

• to increase physicians’ use of established prac-tice guidelines and appropriate pharmaco-therapy for specific disease states.

In 1994, the Williamson Institute for Health Stud-ies of VCU was chosen by DMAS to administerthe VHOP program. A working coalition wasformed among DMAS; the Schools of Medicine,Pharmacy, and Nursing of the Medical Collegeof Virginia; and the National PharmaceuticalCouncil. Early support was obtained from pro-fessional societies including the Virginia Acad-emy of Family Physicians, the Medical Society ofVirginia, the Old Dominion PharmaceuticalAssociation, and the DMAS–DUR Committee.Many of the major health professional societiesin Virginia have expressed support for VHOP.

Potential disease candidates were systematicallyevaluated, and asthma was chosen as the firstdisease for the pilot program. The pilot inter-vention and research designs were completed inDecember 1994. The asthma intervention andresearch program were implemented in October1995. The intervention for asthma was com-pleted in Central Virginia in September 1996,and the data are currently being analyzed, com-paring the intervention to the usual care in popu-lation centers outside of Central Virginia. Theasthma intervention is expected to be imple-mented throughout Virginia by September 1997.

1

APPENDIXEvaluation of Specific Diseases Using Actual Data in Virginia

AsthmaPrevalence and Associated Drug UseNationally, pediatric asthma appears to be onthe rise, particularly among minority popula-tions. Childhood asthma (i.e., asthma in indi-viduals <21 years of age) in Virginia had a preva-lence of approximately 5% in 1993 (Maish &Sagraves, 1993). For adult and childhood asthma,approximately 4.25% of Virginia’s budget wasdirected to payment of pharmacy claims. Basedon drug use, DMAS reimbursed $6,899,213 dur-ing FY 1993 for 294,142 claims made by 63,115individuals (who had to have at least two claimsfor asthmatic drugs to be eligible). However,these drugs also are used in the treatment ofCOPD. In addition, these claims apply through-out the Medicaid system rather than being re-stricted to the Medallion program. Thus, thereare problems in interpretation. More reliable isa diagnosis based on the ICD-9-CM codes ap-pearing on the physician invoice. According tothis criterion, 21,711 individuals were diagnosedwith asthma. This discrepancy between asth-matic patients identified by drug use versusphysician forms involves approximately 66% ofthe patients and points to the need for multiplecriteria to ensure the valid selection of patients.

Asthma in the Medallion PopulationThe statewide population of asthmatic individu-als in Medallion in FY 1994 was 6,405. Therewere 729 hospitalizations for asthma represent-ing 630 individuals. The average payment perhospitalization was $1,964 (ranging from $394 to$12,409). The total amount paid during the yearfor hospital claims with a diagnosis of asthmawas $1,431,802. There were 4,422 claims foroutpatient asthma clinic services during FY 1994for 2,851 Medallion patients. The total cost foroutpatient visits was $582,194, which averagesout to $132 per visit. These outpatient clinicservices are not necessarily services rendered byemergency departments. There were 5,784 Me-dallion recipients with a diagnosis of asthmalisted on physician claim forms, for a total of12,737 claims. The total cost for these visits was$579,789, an average cost of $46 per office visit.Among the asthma patients, 2,782 had claims forlaboratory services, for a total of 15,600 claims

(about 5.6 claims per patient). The total cost was$179,062, an average cost of $64 per laboratoryclaim (DMAS, 1995).

Feasibility of Voluntary AuthorizationGuidelinesPart of the problem in treating asthma is relatedto a shift in the perceived cause of asthma byspecialists. The conventional view was thatasthmatic episodes were caused by broncho-spasm (airway constriction) that progresses toairway obstruction. However, the recent ap-proach is to view asthma as an inflammatorydisease as well, in that inflammation is critical toairway hyper-responsiveness (Djukanovic et al.,1990). Therefore, anti-inflammatory agents arebelieved to contribute to the optimal care of suchpatients (Sheffer & Taggart, 1993); this view isreflected in current NIH and other guidelines.

Specific Measures for Assessing AsthmaInterventionsTwo categories of measures are available:

• Assessing quality of care (e.g., medical recordsabstraction for peak air flow readings or tele-phone interview assessing whether and towhat extent patients were counseled and edu-cated by physicians and pharmacists)

• Assessing actual clinical outcomes (e.g., emer-gency room visits or hospitalizations and/orlength of stay)

Diabetes MellitusPrevalence and Associated Drug UseDiabetes mellitus is the leading cause of periph-eral neuropathy in developed nations. Periph-eral neuropathy affects at least 15% of diabeticpatients overall (Dyck & O’Brien, 1989) and 37%of persons 18 years and older with insulin-de-pendent diabetes mellitus (Maser et al., 1989).

Diabetes in the Medallion PopulationBased on limited data in 1994, only 871 patientswere diagnosed with diabetes statewide. Amongthese patients, there were 65 hospitalizations(involving 61 unique individuals), with an aver-age payment per hospitalization of $3,151. Therewere 324 claims for outpatient hospital clinic

12

ABBREVIATIONS

AHCPR Agency for Health Care Policy and Research

CME Continuing Medical Education

COPD Chronic Obstructive Pulmonary Disease

DMAS Department of Medical Assistance Services (i.e., Virginia Medicaid)

DUR Drug Utilization Review

FY Fiscal Year

HCFA Health Care Financing Administration

NIH National Institutes of Health

PA/VHOP Prior Authorization/Virginia Health Outcomes Partnership(i.e., Virginia Medicaid’s Prior Authorization/Virginia Health OutcomesPartnership Committee)

UTI Urinary Tract Infection

VCU Virginia Commonwealth University

VHOP Virginia Health Outcomes Partnership

ii

services representing 178 individuals for an av-erage cost of $133. Among Medallion patients,803 had at least one physician office visit with adiagnosis of diabetes appearing on the physi-cian claim. The average cost of these visits was$38. Of all patients diagnosed with diabetes, 662had claims for laboratory services for an averagelaboratory claim charge of $108.

Feasibility of Voluntary AuthorizationGuidelinesIntensive diabetes therapy is designed to achievenormal blood glucose values through the use ofthree or more insulin injections per day or insulinadministration with an external pump. Thistherapy prevented development and retardedprogression of diabetic retinopathy and neuropa-thy, albeit at the expense of a 3-fold increase in thenumber of severe hypoglycemic events (DiabetesControl and Complications Trial Research Group,1993). Intensive therapy was guided by frequentself-monitoring of blood glucose levels, leadingto systematic insulin dose adjustments (DiabetesControl and Complications Trial Research Group,1993).

Specific Measures for Assessing DiabetesInterventions• Rates of diabetic retinopathy

• Rates of cardiovascular disease

Otitis MediaPrevalence and Associated Drug UseOtitis media (inflammation of the middle ear) isthe most frequent primary diagnosis for chil-dren less than 15 years of age based on physicianoffice recordings. Almost all children experi-ence one or more episodes of otitis media by age6 (Stool et al., 1994). The incidence of otitismedia is highest in children between 6 monthsand 3 years of age. A smaller peak incidenceoccurs between 4 and 7 years. Infections areuncommon after age 8. It has been estimatedthat 76% to 95% of all children will suffer at leastone episode during their childhood. Approxi-mately 50% of children who have experiencedotitis media have had 3 or more episodes, while25% have had 6 or more episodes. Nationally,otitis media is estimated to be responsible formore than 30 million outpatient visits per year ata cost that exceeds $1 billion. Another $1 to $2

billion is spent annually on surgical treatmentssuch as tube replacement (Kligman, 1992).

In Virginia, based on the ICD-9-CM codes (codes381 and 382) for otitis appearing on the physi-cian invoice, 73,608 separate individuals pre-sented with otitis during FY 1993. The annualnonduplicated number of Medicaid recipientsunder the age of 21 during FY 1993 was 279,531.However, the number of claims and total cost ofthese claims were not available.

Otitis Media in the Medallion PopulationThe total number of otitis media patients in theMedallion population in FY 1994 was 18,469.Only 32 of these patients were hospitalized withthis diagnosis at an average cost per hospitaliza-tion of $1,505. Claims for outpatient services(n=11,714) representing 7,969 individuals in-volved an average payment of $74. The averagephysician office visit cost $36, but with 31,940visits from 16,434 Medallion recipients the totalamount paid was more than $1.1 million.

Feasibility of Voluntary AuthorizationGuidelinesPneumatic otoscopy is recommended for assess-ment of the middle ear because it combines thevisualization of the tympanic membrane with atest of membrane mobility and results in anaccurate diagnosis of otitis media in up to ap-proximately 75% of patients.

Longitudinal studies of otitis media with effu-sion demonstrate high rates (>50%) of spontane-ous resolution within 3 months of symptomonset. Surgery is recommended only for thechild who has had bilateral effusion for 3 monthsand who has a bilateral hearing deficiency (de-fined as a 20-decibel hearing threshold level orworse in the better-hearing ear). In this case,bilateral myringotomy with tube insertion be-comes an additional treatment option. Place-ment of tympanotomy tubes is recommendedafter a total of 4–6 months of bilateral effusionwith a bilateral hearing deficit.

Antibiotic therapy resulted in a 14% improve-ment in clearance of effusion at 1 month. How-ever, nausea, vomiting, and diarrhea are sideeffects 2% to 32% of the time, depending on theantibiotic type and dosage. Also, there is thepotential for developing resistant strains of bac-teria with this treatment.

13

Louis F. Rossiter, Ph.D.Co-Principal InvestigatorThe Williamson Institute for Health StudiesMedical College of VirginiaVirginia Commonwealth UniversityP.O. Box 980203Richmond, VA 23298-0203(804) 828-5223

Michelle Y. Whitehurst-Cook, M.D.Co-Principal InvestigatorSchool of MedicineMedical College of VirginiaVirginia Commonwealth UniversityP.O. Box 980216Richmond, VA 23298-0216(804) 828-9626

Ralph E. Small, Pharm.D.Co-Principal InvestigatorSchool of PharmacyMedical College of VirginiaVirginia Commonwealth UniversityP.O. Box 980533Richmond, VA 23298-0533(804) 828-6333

Charles A. Shasky, R.Ph., M.B.A.Project Manager, VHOPThe Williamson Institute for Health StudiesMedical College of VirginiaVirginia Commonwealth UniversityP.O. Box 980203Richmond, VA 23298-0203(804) 828-0172

Dave Shepherd, R.Ph.Pharmacy SupervisorProgram OperationsVirginia Department of Medical Assistance Services600 E. Broad St., Suite 1300Richmond, VA 23219(804) 225-2773

Judith K. Jones, M.D., Ph.D.PresidentThe Degge Group, Ltd.1616 North Fort Myer DriveSuite 1430Arlington, VA 22209-3109(703) 276-0067

John W. HawksPresidentCOMSORT2300 North Charles StreetBaltimore, MD 21218(410) 467-1100

Richard A. Levy, Ph.D.Vice President, Scientific AffairsNational Pharmaceutical Council1894 Preston White DriveReston, VA 20191-5433(703) 620-6390

VHOP PROGRAM CONTACTS

Jeann Lee Gillespie, Pharm.D., M.S.Director, Scientific AffairsNational Pharmaceutical Council1894 Preston White DriveReston, VA 20191-5433(703) 620-6390

i

Specific Measures for Assessing Otitis MediaInterventions

The following are measures available to assessinterventions in otitis media patients:

• Percent utilization of first- versus second- orthird-generation of antibiotics in patients withno prior history of otitis (no diagnosis of otitisin the previous 6 months)

• Percentage of patients placed on prophylac-tic therapy after a diagnosis of 3 episodes ofacute otitis media within a 6-month period or4 episodes within a 12-month period

• Rates of follow-up office visits after treatmentand relapse rates

• Rates and costs of tympanotomy tube re-placement overall and among those treated

prophylactically; significant medical com-plications and persistent symptoms of dis-equilibrium (vertigo or ataxia), validated bymedical records; include documented de-velopmental delays due to language diffi-culties

• Rates of medical complications: hearing loss,perforation of the eardrum, cholesteatoma(often requiring multiple surgical procedures),acute mastoiditis, atelectasis of the eardrum,retraction pockets, and ossicular discontinu-ity and fixation; other behavioral manifesta-tions (e.g., irritability, disturbed sleep, de-creased responsiveness, social withdrawal)

• Rates of referral for speech and languagedevelopmental lags

• Rates of side effects from medications

14

REFERENCES

Alan Newman Research. Final report of findings from focus groups. Richmond, VA:Virginia Health Outcomes Partnership Project; December 5, 1995.

American Pharmacy, NS34, 12; 1994. Virginia uses pharmacists in compliance program.

Anderson GM, Spitzer WO, Weinstein MC, Wang E, Blackburn JL, Bergman U.Benefits, risks, and costs of prescription drugs: a scientific basis for evaluating policyoptions. Clin Pharmacol Ther 1990;48:111–9.

Berman S. Using clinical practice improvement: An interview with Susan Horn. J QualImprove 1995;21:301–308.

Bloom BS, Jacobs J. Cost effects of restricting cost-effective therapy. Med Care1985;23:872–80.

Brook RH, Lohr KN. Efficacy, effectiveness, variations, and quality. Boundary-crossingresearch. Med Care 1985;23:710–22.

The Degge Group, Ltd. Needs Assessment: Selection of Disease States for the VirginiaHealth Outcomes Partnership (VHOP) Project. Richmond, VA: Virginia Health Out-comes Partnership Project; October 15, 1995.

The Diabetes Control and Complications Trial Research Group. The effect of intensivetreatment of diabetes on the development and progression of long-term complicationsin insulin-dependent diabetes mellitus. N Engl J Med 1993; 329:977–86.

Djukanovic R, Roche WR, Wilson JW, et al. Mucosal inflammation in asthma. Am RevRespir Dis 1990; 142:434.

Donabedian A. Explorations in Quality Assessment and Monitoring. The Definition ofQuality and Approaches to its Assessment. Ann Arbor, MI: Health AdministrationPress; 1980:1–31.

Dranove D. Medicaid drug formulary restrictions. J Law Econ 1989;32:143–162.

Dyck PJ, O’Brien PC. Meaningful degrees of prevention or improvement of nerveconduction in controlled clinical trials of diabetic neuropathy. Diabetes Care 1989;12:649–52.

Ellwood PM. Shattuck lecture–outcomes management. A technology of patientexperiences. N Engl J Med 1988;318:1549–56.

Fisher KH, Harris RH, Smith W, Cotter JJ. Use of a Toll-Free Phone Number andIncentives to Improve Survey Response Rates Among Vulnerable Populations. Manu-script, January 1996.

Gergen PJ, Weiss KB. Changing patterns of asthma hospitalization among children:1979 to 1987. J Am Med Assoc 1990; 264:1689–93.

Grumback K, Bodenheimer T. Reins or fences: a physician’s view of cost containment.Health Aff (Millwood) 1990;9(4):120–6.

Guyatt GH, Sackett DL, Cook DJ. User’s guides to the medical literature. How to use anarticle about therapy or prevention. What were the results and will they help me in caringfor my patients? Evidence-Based Medicine Working Group. JAMA 1994;271:59–63.

Harr RE, LoGerfo JP. Impact of a formulary change restricting minor tranquilizers: TheWashington State Medicaid experience. Contemp Drug Probl 1977;6:515–531.

Hawks JW, Levy R, Hass SL. The Virginia Health Outcomes Project: A UniqueApproach to Lowering Medicaid Costs and Improving Health Outcomes. Am J ManCare 1995; 1:151–154.

Horn SD, Sharkey PD, Tracy D, et al. Intended and unintended consequences of HMOcost containment strategies: Results from the Managed Care Outcomes Project. Am JMan Care 1996;2:253–265.

15

TABLE OF CONTENTS

VHOP Program Contacts ........................................................................................... i

Abbreviations .............................................................................................................. ii

Introduction ..................................................................................................................1

VHOP Program History .............................................................................................1

VHOP Program Structure...........................................................................................2

The Disease Selection Process ....................................................................................2The Value of Empirical Data to Measure Outcomes ......................................................... 2The Framework for Disease Selection ................................................................................. 3Three Key Procedures ............................................................................................................ 4Procedure 1: Using Medicaid (DMAS) Data to Identify

Disease Selection Candidates ................................................................................... 4Procedure 2: Weighted Scoring of Diseases ....................................................................... 4Procedure 3: Survey of Medicaid DUR Committee Members ......................................... 4

The Intervention: HUB and SPOKE Model .............................................................6Using Physician Focus Groups for Feedback on the Intervention .................................. 6Recruiting Physicians............................................................................................................. 8Insights on the Recruitment Process .................................................................................... 9Insights on Multidisciplinary Training ............................................................................... 9

The Research.................................................................................................................9Measuring Health Status of Patients Using a Household Survey ..................................10Measuring Medical Care Utilization by Using Claims Data ...........................................10

Summary ..................................................................................................................... 11

Appendix: Evaluation of Specific Diseases Using Actual Data In Virginia ...... 12Asthma ....................................................................................................................................12Diabetes Mellitus ...................................................................................................................12Otitis Media ............................................................................................................................13

Hurley RE, Rossiter LF. Final Report of the Evaluation of the Medallion Program.Richmond, VA. Virginia Commonwealth University; April 1993.

Jang R. Medicaid formularies: A critical review of the literature. J Pharm MarketingManage 1988;2:39–61.

Kaplan SH, Greenfield S, Ware JE. Assessing the Effects of Physician-Patient Interac-tions on the Outcomes of Chronic Disease. Med Care 1989; 27:S110–127.

Kligman EW. Treatment of otitis media. Am Fam Phys 1992; 45:242–50.

Kozma CM, Reeder CE, Lingle EW. Expanding Medicaid drug formulary coverage:Effects on utilization of related services. Med Care 1990;28:963–77.

Krakauer H, Bailey RC. Epidemiological oversight of the medical care provided toMedicare beneficiaries. Stat Med 1991;10:521–40.

Lipton HL, Bird JA. Drug utilization review in ambulatory settings: state of the scienceand directions for outcomes research. Med Care 1993;31:1069–82.

Lipton HL, Bird JA. Drug utilization review: state of the art from an academicperspective. Clin Pharmacol Ther 1991;50:616–9.

Maish M, Sagraves R. Childhood asthma. US Pharmacist 1993; Jan:36–110.

Maklan CW, Greene R, Cummings MA. Methodological challenges and innovations inpatient outcomes research. Med Care 1994;32:JS13–21.

Malveaus FJ, Diamond E. Deaths from asthma by race, sex and age, 1979-1983. J AllergyClin Immunol 1987; 79:183.

Maser RD, Steenkiste AR, Dorman JS, et al. Epidemiologic correlates of diabeticneuropathy: Report from Pittsburgh Epidemiology of Diabetes Complications Study.Diabetes 1989; 8:1456–61.

Moore WJ. Medicaid drug utilization review: a critical appraisal. Med Care Rev1994;51:3–37.

Moore WJ, Newman RJ. Drug formulary restrictions as a cost-containment policy inMedicaid programs. J Law Econ 1993;36:79–97.

Moore WJ, Newman RJ. US Medicaid drug formularies: Do they work?Pharmacoeconomics 1992;1(suppl1):28–31.

Oxman Ad, Sackett DL, Guyatt GH. User’s guides to the medical literature. How to getstarted. The Evidence-Based Medicine Working Group. JAMA 1993;270:2093–5.

Pearce MJ, Begg EJ. A review of limited lists and formularies: Are they cost-effective?PharmacoEconomics 1992;1:191–202.

Penberthy L, Rossiter L, Whitehurst-Cook M, et al. Research Design Report. Richmond,VA: Virginia Health Outcomes Partnership Project; December 22, 1994.

Physician Payment Review Commission, Annual Report to Congress, Washington,D.C.: U. S. Congress, 1996

Roper WL, Winkenwerder W, Hackbarth GM, Krakauer H. Effectiveness in health care:An initiative to evaluate and improve medical practice. N Engl J Med 1988;319:1197–202.

Sheffer AL, Taggart VS. National Asthma Education Program: Expert Panel and ReportGuidelines for the Diagnosis and Management of Asthma. Med Care 1993; 31:MS20–8.

Sloan FA, Gordon GS, Cocks DL. Hospital drug formularies and use of hospitalservices. Med Care 1993;31:851–867.

Smalley WE, Griffin MR, Fought RL, Sullivan L, Ray WA. Effect of a prior-authoriza-tion requirement on the use of nonsteroidal anti-inflammatory drugs by Medicaidpatients. N Engl J Med 1995;332:1612–7.

16

Soumerai SB, Avorn J. Principles of educational outreach (‘academic detailing’) toimprove clinical decision making. JAMA 1990;263:549–56.

Soumerai SB, Avorn J, Ross-Degnan D, Gortmaker S. Payment restrictions for prescrip-tion drugs under Medicaid: effects on therapy, cost, and equity. N Engl J Med1987;317:550–6.

Soumerai SB, McLaughlin TJ, Ross-Degnan D, et al. Effects of limiting Medicaid drug-reimbursement benefits on the use of psychotropic agents and acute mental healthservices by patients with schizophrenia. N Engl J Med 1994;331:650–655.

Soumerai SB, McLaughlin TJ, Avorn J. Improving drug prescribing in primary care: acritical analysis of the experimental literature. Milbank Q 1989;67(2):268–317.

Soumerai SB, Ross-Degnan D. Experience of state drug benefit programs. Health Aff(Millwood) 1990;9(3):36–54.

Soumerai SB, Ross-Degnan D, Avorn J, McLaughlin TJ, Choodnovskiy I. Effects ofMedicaid drug-payment limits on admission to hospitals and nursing homes. N EnglJ Med 1991;325:1072–7.

Soumerai SB, Ross-Degnan D, Fortess EE, Abelson J. A critical analysis of studies ofstate drug reimbursement policies: research in need of discipline. Milbank Q1993;71(2):217–52.

Steinwachs DM, Wu AW, Skinner EA. How will outcomes management work? HealthAff (Millwood) 1994;13:153–62.

Stool SE, Berg AO, Berman S, et al. Managing otitis media with effusion in youngchildren: Quick reference guide for clinicians. AHCPR Publication No. 94-0623,Rockville, MD, July 1994.

Tengs TO, Adams ME, Pliskin JS, Safran DG, Siegel JE, Weinstein MC, et al. Five-hundred life-saving interventions and their cost-effectiveness. Anal Intern Med1995;15:369–89.

Ware JE, Snow KK, Kosinski M, Gandek B. SF-36 Health Survey: Manual & Interpre-tation Guide. Boston: The Health Institute, New England Medical Center, 1993.

Whitehurst-Cook M, Roberts M, Nelson M, et al. Educational Planning Design Report.Richmond, VA: Virginia Health Outcomes Partnership Project; December 22, 1994.

Wickizer TM. The effect of utilization review on hospital use and expenditures: areview of the literature and an update on recent findings. Med Care Rev 1990;47:327–63.

Wilson CN. Medicaid moving to managed care. Hospital Pharm 1995; 30;214–20.

17

EXECUTIVE SUMMARY

Virginia Health Outcomes Partnership (VHOP) is a model program in whichthe state partnered with its health professionals to improve health outcomes forVirginia Medicaid patients. VHOP was implemented as part of VirginiaMedicaid’s Medallion program, a fee-for-service, primary care case manage-ment program; the enrollees consisted mostly of children and women of child-bearing age. The Medallion program was implemented under a waiver grantedby HCFA.

Cost containment of pharmaceutical expenditures usually involves variousefforts to control drug utilization. These efforts typically are aimed at physicianproviders, pharmacist providers, or the pharmaceutical industry. They haveincluded the establishment of drug formularies and prior authorization pro-grams. Federal legislation has had an important role in shaping these efforts.

A relatively new approach to cost containment is disease management. Thegoal of disease management programs is to optimize therapy. Optimizingtherapy should improve outcomes and decrease overall expenditures associ-ated with a disease.

The VHOP approach involves a patient-centered disease management strat-egy. With this approach, better disease management results from patients’optimal adherence to prescribed therapies. Increasing providers’ communica-tions skills, as well as their disease-specific knowledge, can result in betterchoice of treatments and improved patient compliance, leading to improvedoutcomes and lower costs.

This publication provides general background information on: 1) the methodsused in the disease selection process; 2) the choice of asthma as the first diseaseto be studied; and, 3) implementation of the intervention and research compo-nents of VHOP. VHOP did not have an a priori disease selected for intervention.One challenge for the investigators was determining what diseases to selectinitially for intervention since the selection would shape the content of themedical treatment guidelines taught to health care providers.

The first disease selected was asthma, and its selection was based primarily onthe prevalence and cost of the disease to Virginia Medicaid, but also on thepotential for improving patient outcomes in Medicaid managed care programs.The disease selection process incorporated empirical claims data from VirginiaMedicaid and a theoretical framework to identify relevant disease candidates.Additionally, multiple procedures were used to ensure that the most appropri-ate diseases were selected for the program.

This publication describes the disease selection process and the implementa-tion of the pilot component of VHOP in general terms. A manual outlining howto implement a project like VHOP in a state will be available in late 1997.

18

THE VIRGINIA HEALTH

OUTCOMES

PARTNERSHIP

Reducing the use of health care services and their associated costsby educating pharmacists, physicians, and nurses on communication

techniques and new advances in clinical practice.

Participants:

The Williamson InstituteThe Schools of Pharmacy, Medicine, and Nursing

of theMedical College of Virginia

Virginia Commonwealth UniversityRichmond, Virginia

The Virginia Department of Medical Assistance ServiceRichmond, Virginia

The Degge Group, Ltd.Arlington, Virginia

COMSORTBaltimore, Maryland

Sponsored by:The National Pharmaceutical Council

Reston, Virginia

REFERENCES

Alan Newman Research. Final report of findings from focus groups. Richmond, VA:Virginia Health Outcomes Partnership Project; December 5, 1995.

American Pharmacy, NS34, 12; 1994. Virginia uses pharmacists in compliance program.

Anderson GM, Spitzer WO, Weinstein MC, Wang E, Blackburn JL, Bergman U.Benefits, risks, and costs of prescription drugs: a scientific basis for evaluating policyoptions. Clin Pharmacol Ther 1990;48:111–9.

Berman S. Using clinical practice improvement: An interview with Susan Horn. J QualImprove 1995;21:301–308.

Bloom BS, Jacobs J. Cost effects of restricting cost-effective therapy. Med Care1985;23:872–80.

Brook RH, Lohr KN. Efficacy, effectiveness, variations, and quality. Boundary-crossingresearch. Med Care 1985;23:710–22.

The Degge Group, Ltd. Needs Assessment: Selection of Disease States for the VirginiaHealth Outcomes Partnership (VHOP) Project. Richmond, VA: Virginia Health Out-comes Partnership Project; October 15, 1995.

The Diabetes Control and Complications Trial Research Group. The effect of intensivetreatment of diabetes on the development and progression of long-term complicationsin insulin-dependent diabetes mellitus. N Engl J Med 1993; 329:977–86.

Djukanovic R, Roche WR, Wilson JW, et al. Mucosal inflammation in asthma. Am RevRespir Dis 1990; 142:434.

Donabedian A. Explorations in Quality Assessment and Monitoring. The Definition ofQuality and Approaches to its Assessment. Ann Arbor, MI: Health AdministrationPress; 1980:1–31.

Dranove D. Medicaid drug formulary restrictions. J Law Econ 1989;32:143–162.

Dyck PJ, O’Brien PC. Meaningful degrees of prevention or improvement of nerveconduction in controlled clinical trials of diabetic neuropathy. Diabetes Care 1989;12:649–52.

Ellwood PM. Shattuck lecture–outcomes management. A technology of patientexperiences. N Engl J Med 1988;318:1549–56.

Fisher KH, Harris RH, Smith W, Cotter JJ. Use of a Toll-Free Phone Number andIncentives to Improve Survey Response Rates Among Vulnerable Populations. Manu-script, January 1996.

Gergen PJ, Weiss KB. Changing patterns of asthma hospitalization among children:1979 to 1987. J Am Med Assoc 1990; 264:1689–93.

Grumback K, Bodenheimer T. Reins or fences: a physician’s view of cost containment.Health Aff (Millwood) 1990;9(4):120–6.

Guyatt GH, Sackett DL, Cook DJ. User’s guides to the medical literature. How to use anarticle about therapy or prevention. What were the results and will they help me in caringfor my patients? Evidence-Based Medicine Working Group. JAMA 1994;271:59–63.

Harr RE, LoGerfo JP. Impact of a formulary change restricting minor tranquilizers: TheWashington State Medicaid experience. Contemp Drug Probl 1977;6:515–531.

Hawks JW, Levy R, Hass SL. The Virginia Health Outcomes Project: A UniqueApproach to Lowering Medicaid Costs and Improving Health Outcomes. Am J ManCare 1995; 1:151–154.

Horn SD, Sharkey PD, Tracy D, et al. Intended and unintended consequences of HMOcost containment strategies: Results from the Managed Care Outcomes Project. Am JMan Care 1996;2:253–265.

15