Objective: To identify determinants of medication complianceand low-density lipoprotein cholesterol goal attainment.
Study Design: This retrospective analysis used claims data froma large, national, employment-based independent practice associationdatabase. Subjects were identified based on the existence of afilled prescription for statin therapy between April 1, 1999, and June30, 2001. Subjects had to be 18 years or older, continuously enrolledin the health plan for 2 years, and new users of statin therapy.
Methods: Multivariate logistic regression models were used toidentify predictors of compliance and goal attainment in high-risksubjects.
Results: As the mean copayment for statins increased, there wasa decrease in the likelihood of compliance. Of the subjects withlaboratory results, 50.7% attained their low-density lipoproteincholesterol goal level established by National CholesterolEducation Program Adult Treatment Panel III guidelines. Olderindividuals and men were more likely to reach their low-densitylipoprotein cholesterol target goal, as were individuals who werecompliant with their statin therapy.
Conclusions: Compliance with statin therapy in the managedcare setting remains poor. Of particular concern is the lower levelof compliance among women and younger high-risk patients,along with patients who have fewer outpatient visits associatedwith hyperlipidemia and lower incidences of cholesterol testing.
(Am J Manag Care. 2005;11:306-312)
Coronary heart disease (CHD) is the single leadingcause of death in the United States, withmore than 650 000 mortalities from CHD-relatedcauses each year.1 Dyslipidemia is acknowledged asa major risk factor for CHD. A large body of evidencefrom epidemiological studies2-5 suggests that lower low-densitylipoprotein cholesterol (LDL-C) levels are associatedwith lower overall risk of CHD morbidity andmortality. Clinical trials have shown that 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductaseinhibitors (statins) can significantly lower cholesterollevels and reduce CHD-related morbidity and mortality.6,7 However, several studies8-10 have demonstratedless than optimal levels of compliance and persistencewith statin regimens in various patient populations,including older persons, managed care enrollees,Medicaid enrollees, and patients with acute coronarysyndrome.
Several studies have attempted to identify correlatesof patient noncompliance with statin therapy. Patientcopayment level, the number of medications taken concurrently,and the days supplied per prescription werefound to be significantly associated with compliancewith statin therapy.11 Increased number of comorbiditieshas also been found to be associated with increasedcompliance with statin therapy,8,10 as has membershipin a primary prevention cohort as opposed to a secondaryprevention cohort.9,12 Increased age, however, hasbeen associated with decreased compliance.10 Mavigliaet al12 identified a relationship between patients' LDL-Clevels and their compliance with statin therapy. Theseauthors report that subjects who were noncompliantwith their statin therapy were least likely to have hadtheir LDL-C levels checked.
To increase compliance rates among statin users, itis crucial to focus interventions that are geared towardincreasing adherence to statin regimens on patientswho are likely to be noncompliant with therapy. Thisstudy examines predictors of patient compliance withstatin therapy and attainment of LDL-C goal levels usinga large retrospective longitudinal administrative claimsdatabase. Using claims data provides a real-world depictionof medication use in terms of the frequency andregularity in which prescriptions are refilled by patientsand their ability to attain LDL-C goal levels.
International Classification of
Diseases, Ninth Revision, Clinical Modification
This study used retrospective longitudinal claimsdata from 23 fee-for-service independent practice associationhealth plans affiliated with a large, national managed care organization. These health plans are locatedin various states throughout the western, midwestern,southern, southeastern, and northeastern UnitedStates, providing coverage for inpatient care, ambulatoryservices, and prescription drugs. The study samplewas selected from the population of commerciallyinsured health plan members 18 years of age or olderwho had pharmacy benefits and who were continuouslyenrolled in the health plan for 2 years. Patients withnewly filled prescriptions for statins were identifiedfrom April 1, 1999, through June 30, 2001 (the indexdate was the date of the first filled statin prescription foreach patient), and medical and pharmacy claims fromApril 1, 1998, through June 30, 2002, were used to capture2 years of data for each patient (12 months beforethe index date and 12 months after the index date).Patients at high risk for CHD or CHD equivalents, asestablished by the National Cholesterol EducationProgram Adult Treatment Panel III guidelines,13 wereidentified using diagnostic codes; 2002 procedural codes; or filledprescriptions in medical and pharmacy claims (Table1). Medical, pharmacy, and laboratory claims data wereused to characterize health status, healthcare use,hyperlipidemia treatment patterns, and LDL-C levels.13
Data on variables related to statin therapy, compliance,and goal attainment were gathered from the medicalclaims database; these included demographicinformation, health conditions, and healthcare resourceuse. Information on demographic variables, includingsex and age, was obtained from enrollment files.
Information regarding comorbid conditions presentduring the study was obtained from the medical andpharmacy claims database using codes, codes, and prescription drug fills (Table 2). Drug use andprescription fill dates were obtained from the pharmacyclaims database using national drug codes. Healthcareuse during the baseline period—including measures ofambulatory visits associated with hyperlipidemia(defined using codes 272.0x-272.2x, 272.4x,or 272.9x), any visits to an emergency department, anyinpatient hospital visits, any cardiovascular procedures(defined using codes 33503-33506, 33510-33516,33517-33530, 33536, 92980, 92982, 92984, 92997, or92998 or procedure codes 36.00-36.03,36.05, or 36.09), and any diagnoses of ischemic heartdisease (defined using codes 410.xx-414.xx orprescription fills for nitrates)—was measured using medicaland pharmacy claims data. The number of cholesteroltests conducted in the baseline period wasidentified in medical claims using the following codes: 80061, 82465, 83715, 83716, and 83721.Healthcare use in the baseline period and baselinehealthcare costs were used as measures of health anddisease severity.
The mean statin prescription copayment was calculatedusing pharmacy claims data for all patients duringthe 12-month follow-up. The ratio of the number offilled statin prescriptions by mail order to the numberfilled by retail pharmacies was constructed for the follow-up period using pharmacy claims data.
Pharmacy claims data were used to determine patternsof statin therapy during the study. Informationwas extracted regarding the number of filled prescriptionsand days' supply of filled prescriptions for statinmedications during follow-up. An adherence ratio wascalculated to measure compliance by dividing the totaldays' supply of statins by the length of possible treatment(365 days). Compliance was defined as an adherenceratio of at least 80% or a daily dose of medicationavailable for at least 80% of the days in the period.14 Theratio incorporates a measure of persistency as wellbecause it includes all statin users, including those witha single filled prescription.
A total of 21 239 new statin users were identified asbeing at high risk for CHD and met all the inclusion criteriafor the analysis. Table 3 provides descriptive characteristics,as well as statistical analyses of means andproportions, for the compliant and noncompliantcohorts.
Compliance With Statin Therapy.
A logistic regressionmodel was used to determine the predictors ofcompliance with statin therapy in the first 365 days oftreatment after the index date. Covariates in themodel included age, sex, number of comorbidities, presenceof ischemic heart disease, number of hyperlipidemiaambulatory visits, number of cholesterol tests,occurrence of a cardiovascular procedure, occurrenceof an inpatient hospitalization,any visits to an emergencydepartment, total healthcarecosts, the mean statin copayment,and the ratio of filledstatin prescriptions via mailorder vs retail store. Predictorswere selected based on informationfrom the medical literatureregarding compliance with medicationsand hypothesized relationshipsbetween the covariatesand compliance.8-12,15,16
According to the results ofthe multiple logistic regressionanalysis for high-risk patients,older subjects and men were significantlymore compliant withstatin therapy, as were patientswith more baseline outpatientvisits associated with hyperlipidemia(Table 4). Patients withan occurrence of a cardiovascularprocedure or inpatient hospitalizationduring the baselineperiod were more compliantwith statin therapy. However,patients who had at least 1 visitto an emergency departmentduring the baseline period wereless likely to be compliant withstatin therapy than patients with no visits. Patientswho had filled their statin prescriptions via mail orderwere more likely to be compliant with therapy comparedwith those who filled more of their statin prescriptionsat retail pharmacy stores. Furthermore, asthe mean copayment for statins increased, there was adecrease in the likelihood of compliance.
Attainment of Low-
Density Lipoprotein Cholesterol
A model of goal attainment was developed todetermine whether compliance with statin therapy wasassociated with health outcomes after controlling forother covariates. Approximately 4219 individuals fromthe entire cohort had at least 1 LDL-C laboratory resultfrom the 60 days after the index date to the end of thestudy. The subcohort with LDL-C results is similar tothe larger cohort in terms of age, sex, compliance, baselinecosts, and health status measures. Of these, 2141(50.7%) attained their LDL-C goal level, according toNational Cholesterol Education Program AdultTreatment Panel III guidelines,13 during the relevantperiod. The mean + SD LDL-C laboratory value forthose who attained their goal level was 80 + 15 mg/dL(2.07 + 0.39 mmol/L). The mean + SD number of daysto attain the LDL-C goal level after initiation of statintherapy was 189 + 90 days. A mean + SD of 5 + 3 statinprescriptions were filled before attainment of the LDLCgoal level, with a mean + SD days' supply of 34.75 +14.68 days.
Maximum likelihood estimationwas used to predict whether individualswith laboratory resultsattained their LDL-C goal level. Alogistic regression model was usedwith a dichotomous dependent variablerepresenting whether individualsattained their LDL-C goal withinat least 60 days after their indexdate. The results (Table 5) suggestthat older individuals at high riskfor CHD were more likely to attaintheir LDL-C goal level comparedwith younger individuals (adding10 years results in a 12.7% higherodds of attaining their goal). Menhad a 27.4% higher odds of attainingtheir goal compared withwomen. Individuals with morebaseline comorbidities had a 6.8%higher odds of attaining their goal.Individuals with 2 additional cholesteroltests had a 22.1% higherodds of attaining their goal. Individualswho underwent a cardiovascularprocedure during thebaseline period had a 50.4% higherodds of attaining their goal. Individualswith a higher proportionof their statin prescriptions filledvia mail order had a higher odds ofattaining their goal. Finally, individualswho were compliant with statintherapy in the first 90 days had a 5times higher odds of attaining theirgoal compared with individuals whowere not compliant.
A second logistic regression modelwas estimated using a subcohort of1464 patients with LDL-C laboratoryresults in the follow-up period, inaddition to baseline LDL-C testresults. The results from this model(Table 6) suggest that individualswith higher baseline LDL-C valueswere less likely to attain their LDL-Cgoal level.
The objective of this study was to identify predictorsof compliance with statin therapy and LDL-C goalattainment among patients at high risk for CHD. Thisstudy demonstrated that, among high-riskusers of statin therapy, youngersubjects and women were significantlyless compliant with statin therapy, aswere patients with fewer baseline outpatientvisits associated with hyperlipidemia.The mean statin copaymenthad a large effect on compliance. If themean copayment increases by $15, theprobability of compliance decreases by10 percentage points; if it increases by$50, the probability of compliancedecreases by 34 percentage points.High-risk patients were less likely toattain their LDL-C goal if they wereyounger, were female, were noncompliantwith statin therapy, or had higherbaseline LDL-C results. Interventionprograms should target these patientsto improve their health outcomes.
There are several limitations whenusing secondary administrative healthcareclaims data. First, the studyresults may not be generalizable to theentire US population, as the data source consisted ofemployer-based managed care enrollees. However, theresults can be generalized to other employer-basedhealth plans with enrollees of similar age and diagnosis.Second, an adherence ratio was used as a proxy forcompliance. These ratios may significantly underestimatecompliance because of different factors that mayprolong the supply of medication; these include but arenot limited to changes in prescription by a physician,hospitalization, or provision of medication samples. Onthe other hand, they may overestimate compliance ifpatients fill their prescriptions but do not take them.
Despite the limitations of retrospective claims data,the present study allowed for evaluation of predictors ofcompliance with statin therapy in a real-life setting.Clinical trials of statin therapy often exclude patientswith comorbidities and generally are not representativeof all patients who subsequently receive treatment. Thedata for the present study came from diverse regions ofthe country and thus may be considered representativeof health plan enrollees from various geographic areasin the United States. In addition, the claims data examinedare current and likely representative of currentpatterns of care.
The results indicate that compliance with statintherapy in the managed care setting remains poor. Ofparticular concern is the lower level of complianceamong female and younger high-risk patients, alongwith patients who have fewer outpatient visits associatedwith hyperlipidemia and lower incidences of cholesteroltesting. Together, these results suggest thatmanaged care intervention programs would benefitpatients by targeting services toward groups at risk oflower compliance, while at the same time increasingprovider contact related to dyslipidemia and the availabilityof cholesterol testing. Intervention programscould consist of initiating information campaigns toproviders, establishing and enforcing guidelines forcholesterol testing, monitoring the volume of physiciancontact with patients who are not filling their statinmedications, sending out reminder cards to targetedpatients for physician checkups, offering patient incentivesto increase use of mail order statin prescriptions,and providing incentives for lowering LDL-C levels.Managed care providers interested in improving compliancemight also consider addressing the barriersimposed by higher copayment costs.
From the Department of Economics, University of Minnesota, Duluth (JSS);AstraZeneca LP, Wilmington, Del (JCO, KLM, RS); and Ingenix Inc, Eden Prairie, Minn (JM).
Funding for this study was provided by AstraZeneca LP.
Address correspondence to: Jennifer S. Schultz, PhD, Department of Economics,University of Minnesota, 165 SBE, 412 Library Drive, Duluth, MN 55812. E-mail:email@example.com.
Heart Disease and Stroke Statistics—2005
1. American Heart Association. Dallas, Tex: American Heart Association. Available at: http://www.americanheart.org/downloadable/heart/1105390918119HDSStats2005Update.pdf.Accessed March 11, 2005.
2. Scandinavian Simvastatin Survival Study Group. Randomised trial of cholesterollowering in 4444 patients with coronary heart disease: the ScandinavianSimvastatin Survival Study (4S). 1994;344:1383-1389.
3. West of Scotland Coronary Prevention Study Group. Influence of pravastatinand plasma levels on clinical events in the West of Scotland Coronary PreventionStudy (WOSCOPS). 1998;97:1440-1445.
4. Downs JR, Clearfield M, Weis S, et al. Primary prevention of acute coronaryevents with lovastatin in men and women with average cholesterol levels: resultsof AFCAPS/TexCAPS: Air Force/Texas Coronary Atherosclerosis Prevention Study.1998;279:1615-1622.
5. Castelli WP, Garrison RJ, Wilson PW, et al. Incidence of coronary heart diseaseand lipoprotein cholesterol levels: the Framingham Study. 1986;256:2835-2838.
6. LaRosa JC, He J, Vupputuri S. Effect of statins on risk of coronary disease:a meta-analysis of randomized controlled trials. 1999;282:2340-2346.
Arch Intern Med.
7. Ross SD, Allen IE, Connelly JE, et al. Clinical outcomes in statin treatment trials:a meta-analysis. 1999;159:1793-1802.
8. Avorn J, Monette J, Lacour A, et al. Persistence of use of lipid-lowering medications:a cross-national study. 1998;279:1458-1462.
9. Jackevicius CA, Mamdani M, Tu JV. Adherence with statin therapy in elderlypatients with and without acute coronary syndrome. 2002;288:462-467.
10. Benner JS, Glynn RJ, Mogun H, et al. Long-term persistence in use of statintherapy in elderly patients. 2002;288:455-461.
Manag Care Interface.
11. Coombs JH, Cornish L, Hiller P, Smith DG. Compliance and refill patternbehavior with HMG-CoA reductase inhibitors after acute myocardial infarction.2002;15:54-58, 60.
J Gen Intern Med.
12. Maviglia SM, Teich JM, Fiskio J, Bates DW. Using an electronic medicalrecord to identify opportunities to improve compliance with cholesterol guidelines.2001;16:531-537.
13. Expert Panel on Detection, Evaluation, and Treatment of High BloodCholesterol in Adults. Executive summary of the Third Report of the NationalCholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, andTreatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). 2001;285:2486-2497.
14. Rizzo JA, Simons WR. Variations in compliance among hypertensive patientsby drug class: implications for health care costs. 1997;19:1446-1457.
Eur Heart J.
15. Strandberg TE, Lehto S, Pyorala K, Kesaniemi A, Oksa H. Cholesterol loweringafter participation in the Scandinavian Simvastatin Survival Study (4S) inFinland. 1997;18:1725-1727.
Br J Clin Pharmacol.
16. Larsen J, Andersen M, Kragstrup J, Gram LF. High persistence of statin use ina Danish population: compliance study 1993-1998. 2002;53:375-378.