Pharmacoeconomic impact of non-compliance with statins
- 119 Downloads
There are numerous studies examining the pharmacoeconomic impact of HMG-CoA reductase inhibitor (statin) therapy on healthcare costs and outcomes. A recently published review demonstrated that the cost-benefit of these agents depends primarily on the risk of developing a coronary event. That is, as the risk of a coronary event increases, the cost-effectiveness ratio decreases. The typical cost per life-year saved (LYS) ranged from $US1800 to $US40 000 in patients with pre-existing coronary artery disease (CAD) and from $US15 000 to >$US1 million per LYS in patients without pre-existing CAD.
The literature is sparse on the pharmacoeconomics of medication non-compliance in patients taking statin medications. Data from two studies suggest that >75% compliance results in decreased coronary events such as myocardial infarction. However, retrospective database analyses indicate that the average compliance rate hovers around the 65% mark. Many of the studies discuss medication non-compliance as a factor, but do not independently analyse compliance pharmacoeconomically.
We examined the pharmacoeconomic impact of non-compliance using published studies that contained pharmacoeconomic data and/or compliance data. In general, we used the placebo arm of these published studies as the surrogate marker for complete non-compliance. The results suggest that for almost 100% compliance versus initial non-compliance, the cost effectiveness of statin medications ranges from $US4500 to over $US250 000 per LYS depending on patient age, presence or absence of risk factors and whether the statin is being used for primary or secondary prevention.
Alternate-day or weekly dosing studies were also used to examine the impact of compliance on cost and health outcomes. Alternate-day dosing represented 50% compliance and weekly dosing 29% compliance. Less than full compliance had the expected effect of worse health outcomes and lower drug costs. However, the studies were small and not designed as true pharmacoeconomic studies looking at the relationship between medication compliance and cost.
The results of this review suggest that there needs to be further examination of the relationship between compliance with statins and cost effectiveness, and studies need to include compliance in their data collection and analysis.
KeywordsStatin Lovastatin Fluvastatin Latanoprost Medication Compliance
Cardiovascular disease is the single largest killer of both men and women in the world, and the cost of heart disease and stroke for the year 2003 is projected to be $US351 billion: $US209 billion for healthcare expenditures and $US142 billion for lost productivity from death and disability. Non-compliance with medications contributes to this cost, resulting in 10% of all hospital admissions and 23% of all nursing home admissions. Further, McGhan and Peterson suggest that medication non-compliance results in a cost of approximately $US300 million annually (year 2000 values).
There is a well established relationship between low-density lipoprotein-cholesterol (LDL-C) and cardiovascular events. There are also extensive data supporting the use of HMG-CoA reductase inhibitors (statins) for both primary[5,6] and secondary[7, 8, 9, 10] prevention of cardiovascular events. In the US in 2000, statins represented nearly 1 billion prescription-units sold. Although there are variable reports of non-compliance with these medications, there appears to be substantial reduction in cardiovascular events. Further, the pharmacoeconomic loss of non-compliance with these medications is still uncertain and no one has adequately quantified the pharmacoeconomic impact of statin non-compliance. The purpose of this paper is to discuss the issues of medication compliance with statins and review some of the pharmacoeconomic data on statin non-compliance.
1. Medication Non-Compliance
Medication non-compliance is multifactorial in nature with the interplay of: (i) patient-related factors; (ii) treatment-related factors; and (iii) relationship-related factors contributing to a patient’s lack of ability to comply with drug regimens. Patient-related factors include knowledge of the medication and/or disease, the socioeconomic status of the patient and even the social situation in which the patient exists. The treatment factors include the adverse effects of medication, physical characteristics of the medication like dosage form or taste and dosing frequency. The relationship factors include how the clinician interacts with the patient, the style of communication and the knowledge the clinician gives to the patient.
Medication non-compliance can be considered in a variety of ways. The first type of medication non-compliance is a patient’s failure to initiate therapy, or initial non-compliance. This can be a consequence of failing to present, to a pharmacy, an initial prescription for dispensing, or failing to claim a prescription that has been dispensed. In either case the patient does not initiate therapy at all. The true rate of failure to initiate therapy is not known but estimates of the percent of people failing to present a prescription range from 5.2–18%.[13,15] Further, Ernst and Grizzle report that 10% of prescriptions resulting in treatment failure were never presented to a pharmacy by patients.
Unclaimed prescriptions, those presented to a pharmacy for dispensing but never picked up by patients, represent a more easily quantifiable aspect of non-compliance. Secnik and colleagues collated data from ten studies showing the unclaimed prescription rate ranged from as low as 9% to as high as 79%. Other estimates of the percent of unclaimed prescriptions ranged from 0.8% to 33% and varied depending on the age, gender of the patient and the delivery mechanism. However, the overall rate of unclaimed prescriptions appeared to be in the low- to mid-teens (12–16%). Together, the combination of never presented and never claimed prescriptions may result in an initial non-compliance rate of approximately 13–25%.
The next type of non-compliance occurs when a patient has filled the initial prescription, but does not consume the medication as scheduled. This could range from forgetting (or intentionally skipping) a dose or multiple doses, to delays in refilling prescriptions. The clinical consequences of this type of non-compliance vary depending on the disease and the relative importance of the drug therapy to the disease. For example, erratic compliance to anticonvulsant agents can result in lower plasma drug concentrations and an increased risk of a seizure. Another example is the potential for the development of HIV resistance due to erratically consumed protease inhibitors. Paterson et al. showed that 80–90% compliance resulted in a 63% virological failure rate versus a 21% virological failure rate seen in patients 95% compliant or better. This demonstrates that with these agents there needs to be a highly consistent level of medication to maintain a therapeutic coverage. However, some drugs are more forgiving. Missed doses of aspirin (acetylsalicylic acid) for the prevention of clot development or ill-timed doses of latanoprost for glaucoma have little impact on clinical outcome since the duration of drug activity in the body far exceeds the dosing interval and therefore more appropriate therapeutic coverage is maintained.
For the statin medications, it appears that there is a minimum level of therapeutic coverage which needs to be maintained for a positive clinical benefit. A subgroup analysis of WOSCOPS (West of Scotland Coronary Prevention Study) showed that individuals taking 75% or more of their medications had a one-third greater reduction in mortality than those taking <75%. While an argument can be made for the possibility of a chance finding in subgroup analyses, Wei and colleagues found that in >5500 patients who experienced a myocardial infarction (MI) there was a lower risk of MI recurrence when patients were >80% compliant with statin medications.
The last type of non-compliance can be considered the premature discontinuation of a medication. For some medications, the result of this premature discontinuation may be worse than initial non-compliance. This is because treatment discontinuation carries with it additional clinical consequences such as antimicrobial resistance or a worsened condition (e.g. rebound hypertension with β-adrenoceptor antagonist withdrawal). The rates of discontinuation of lipid-lowering drugs prescribed in primary care settings are substantial and have generally been much higher than those reported in randomised clinical trials. For drugs used to treat chronic conditions such as hypercholesterolaemia, drug discontinuation can be useful when evaluating patients’ reasons for failing to adhere to drug therapy. For example, in the community setting, discontinuation of therapy has been found to signal patient intolerance, drug ineffectiveness or patient doubt about the necessity of treatment.
2. Medication Compliance and Statins
The latest report from the of the NCEP-III (National Cholesterol Education Program) urges healthcare professionals to embark on an intensive LDL-C lowering programme for patients at high risk of developing, or patients with existing, cardiovascular disease. The recommendations from that report support statins as first-line agents for reducing LDL-C levels, with these drugs on average yielding an 18–55% dose-dependent reduction in LDL-C levels. All of this is predicated on good compliance with diet, exercise and medication consumption.
As noted in section 2, the WOSCOPS study indicated that a 75% or better compliance rate leads to a one-third reduction in coronary events. To determine what the natural rate of compliance is in an HMO setting, Andrade and associates estimated the 1-year probability of discontinuation of lipid-lowering drugs at two HMOs. Using the computerised HMO pharmacy files and medical charts of 2369 new users of these agents during the period 1988–90, the investigators reported the 1-year discontinuation rate was 41% for bile acid sequestrants, 46% for niacin, 15% for lovastatin and 37% for gemfibrozil. The predominant reasons documented for the discontinuation were adverse effects (56%) and therapeutic ineffectiveness (31%). In this study, the persistence (100% minus the discontinuation rate) for the statins was 85%, exceeding the 75% needed for a reduction in coronary events. However, the study only examined data for 1-year follow-up; not long enough to determine the true impact of chronic therapy on long-term persistence and clinical outcomes.
Mansur and colleagues studied the adherence behaviour of 207 consecutive patients with coronary artery disease (CAD) and dyslipidaemia. Of these, 68 (33%) patients were not prescribed statins, 54 (26%) took them irregularly and 85 (41%) took them regularly. Of the 54 patients taking statins irregularly, 36 (67%) identified cost as the major barrier to adherence, 17 (31%) identified lack of proper instruction as the barrier and only one patient identified adverse effects as the barrier. The authors also determined the clinical effect of adherence on cholesterol levels. The results showed that patients taking statins regularly had significant improvements in total LDL-C and high-density lipoprotein-cholesterol (HDL-C) levels, whereas nonadherent patients did not have significant changes. Unfortunately, the authors did not report the percentage of medications consumed that constituted ‘adherent’. Their definition included inconsistent medication consumption as well as drug discontinuation. Further, the authors did not report on morbidity and mortality as it related to adherence.
Avorn and colleagues reported that adherence (as measured by pharmacy refills) among patients taking lipid-lowering agents, averaged only 59%. This was a retrospective database analysis of two separate datasets, one from New Jersey (US) and one from Canada. The New Jersey dataset was from the state Medicaid programme and enrollees had incomes ≤$US19 250 (1990 values). The Canadian dataset consisted of a 10% random sample of the administrative files of the universal healthcare system of Quebec. Both datasets used patients enrolled during the 1 January 1990 to 1 July 1991 timeframe. Both populations had no co-payments or other co-insurances required at that time, removing the cost barrier from the compliance equation.
There were 5611 New Jersey patients and 1676 Canadian patients meeting the entry criteria of ≥65 years of age with prescription of at least one lipid-lowering medication during the time frame. In both the US and Canada, the highest compliance was associated with the use of statins (64.3% ± 29.8%), while the lowest was associated with the use of cholestyramine (36.6% ± 29.1%). These results are in some conflict with those of Andrade et al. suggesting a lower than expected compliance rate with statins. The reason for this may be in the study design by Andrade which only used data from two local health maintenance organisations (HMOs). Further, the earlier time frame (1988–90) may have played a role since the agents used in the Andrade study were more novel and perhaps with fewer perceived adverse effects than commonly used agents, therefore resulting in comparatively higher compliance rates. Regardless, the 64.3% compliance in the Avorn study is still lower than the 75% needed to achieve the clinical outcome seen in WOSCOPS.
3. Pharmacoeconomics and Statins
There are a multitude of studies examining the pharmacoeconomic impact of statin therapy on healthcare costs and outcomes. When reading the literature one will surely notice the complexity and varied sophistication used by many analysts in determining the cost effectiveness of this class of medication.
There are several types of cost-effectiveness ratios used in the statin pharmacoeconomic literature. These include cost per reduction in intermediate outcome measures (e.g. LDL-C levels, total cholesterol levels or the LDL/HDL ratio). Other measures include cost per cardiovascular event avoided, cost per life-year saved (LYS), cost per death prevented and cost per quality-adjusted life-year (QALY) gained. Further, there is a myriad of approaches to timeframe selection (e.g. 5 years vs lifetime) on which the analysis is based. This is coupled with a variety of approaches to discounting future costs, a variety of monetary units describing the costs with an equal variety of cost factors taken into consideration.
While many of these factors can limit the generalisability of the studies, the results of the Morrison and Glassberg review demonstrates a remarkable similarity. Their review demonstrates that the benefit from these agents depends primarily on the risk of developing a coronary event. That is, as the risk of a coronary event increases, the cost-effectiveness ratio decreases. The typical cost ranges from $US1800 to $US40 000 per LYS in patients with pre-existing CAD to estimates ranging from $US15 000 to more than $US1 million per LYS in patients without pre-existing CAD. For the most part, cost-effectiveness ratios for statins are typically ≤$US50 000 per LYS for patients with pre-existing CAD, which we deem to be an acceptable threshold for a cost-effective intervention; however, the jury is still out on the cost effectiveness of statins in patients without CAD. Within this population the number of risk factors, as well as the age of the individual, play a significant role in the development of a coronary event; therefore, pharmacoeconomic analyses in this population are sensitive to these factors.
4. Pharmacoeconomics and Compliance with Statins
The pharmacoeconomic impact of non-compliance can be considered from a variety of perspectives. The patient perspective may be short-term in nature, with decreased compliance resulting in an obvious decreased out-of-pocket expenditure for the medication. The long-term impact may not be financially felt by the patient because if a major consequence occurs as a result of the non-compliance, the major medical expenses are often absorbed by a third-party payer (e.g. insurance company or government). The payer perspective is hopefully long-term; managed care organisations (MCOs) have a short-term increase in cost outlay for the pharmaceutical in the hope that there will be a greater reduction in medical expenses at some point in the not too distant future. The economic dilemma occurs in that the benefit of many long-term therapies such as lipid-lowering therapy may not be seen until after years of treatment (e.g. 3–5 years) but the average person stays with a US health plan for 2–3 years. From the societal perspective, adherence to statin medications can be cost effective in the long run.
In a retrospective analysis of a comprehensive US database, Peterson and colleagues determined the short-term (≤1 year) impact of non-compliance to statin medications. Their analysis showed that there is little correlation between non-compliance and overall total medical and drug costs. In fact, the variable of compliance is only slightly correlated to total medical costs (r = 0.114, p < 0.0001). Further, when drug costs are removed from the dependent variable the correlation is even less (r = 0.05, p = 0.01). In this analysis the authors only tested for a linear relationship. It is conceivable that a higher-order relationship (e.g. exponential) between compliance and medical costs exists.
In a separate study, Peterson and McGhan also attempted to demonstrate the economic impact of a medication compliance intervention programme. Taking the MCO perspective, they determined that any intervention programme will cost the MCO $US154.12–$US279.12 more per person in the first year of the programme (actual costs for 1999–2000 were used). Because of lack of data extending beyond 1 year, they were unable to project the 3- to 5-year cost effectiveness.
However, one can consider that the impact of initial non-compliance can be estimated through the pharmacoeconomic analyses of randomised, placebo-controlled studies. The placebo arm represents the initial non-compliant patients. The economic impact of erratic medication-taking behaviour will be estimated using studies looking at alternate-day dosing or other studies attempting to demonstrate the clinical and economic impact of less frequent dosing than is typical.
5. Economic Impact of Non-Compliance
5.1 Initial Non-Compliance
Only selected studies will be presented as representative of the body of pharmacoeconomic studies looking at non-compliance. The studies selected have a placebo arm and are based on larger clinical trials. This will help us determine the economic impact of initial non-compliance.
5.1.1 Primary Prevention Studies
The study by Martens and Guibert examined the cost effectiveness of four statins (fluvastatin, lovastatin, pravastatin and simvastatin). The perspective used in this study was the payer. This study used data from the Canadian Heart Health Survey as well as the Framingham study for CAD risk estimates. Costs were estimated from pooled costs across three Canadian provinces and were indexed to 1993 Canadian dollars ($Can) and estimates of lifetime costs were discounted at a rate of 5% per annum. The estimates of treatment effects were calculated from data pooled from 40 published, randomised, controlled clinical trials and discounted at the same rate as costs (5%). The authors used the most commonly reported dosages, or dosage equivalents, of the statins. The results indicate that for a 45-year-old male smoker, with a cholesterol level of 4.5 mmol/L, the discounted LYS ranged from 0.174 (fluvastatin 40 mg/day) to 0.215 (simvastatin 10 mg/day). The discounted cost per LYS ranged from $Can38 800 for fluvastatin 40 mg/day to $Can56 200 for pravastatin 20 mg/day (year 1993 values). The risk is only estimated for 45-year-old men with average risk of CAD. This limits the applicability of these data to other patient populations such as females, older patients or those with a greater independent risk for CAD. Further, the authors assumed 100% compliance with the medications over the patient’s lifetime. There was no sensitivity analysis performed adjusting for the change in risk due to changes in medication compliance.
Berto and colleagues conducted a pharmacoeconomic analysis using a Markov model for a cohort of men and women with varying risk factors such as age, percent smokers, blood pressure and body mass index. The cost estimates used were calculated from both the Italian National Health Service (NHS) and the societal perspective using the 1998 Italian lire (L). Estimates of the risk of CAD were taken from the Framingham study and life-expectancy estimates were taken from the published annual statistics from Italy. The efficacy of drug therapy was estimated from Bayer Pharmaceuticals (for cerivastatin) and published data for the two other statins (simvastatin and pravastatin). Table I shows the results for the 50- to 54-year-old cohort for the various dosages of the three drugs studied from the NHS perspective using a 5% per annum discounting estimate. The cost per LYS ranged from L39 986 000 for cerivastatin 0.2 mg/day to L68 814 000 for simvastatin 40 mg/day. The strength of this study was that varied age ranges as well as the varied risk factors were entered into the Markov model. Further, the authors employed a sensitivity analysis to the data, varying not only these factors, but the discount rates. However, similar to the Martens and Guibert study, the authors assumed 100% compliance with the medications.
Hamilton and colleagues conducted a similar analysis to the Berto et al. study with the CAD prevention model based on Framingham data. Costs and outcomes were discounted at the 5% per annum rate from the societal perspective using 1992–3 Canadian dollars. The baseline lipid levels were derived from the US National Health and Nutrition Examination Survey. The effect of the lipid modification was derived from data published in the EXCEL (Expanded Clinical Evaluation of Lovastatin) study and therefore limited the results to the effect of lovastatin. The cohort consisted of males and females aged 30–70 years of age with varying levels of risk for CAD. The investigators reported the costs taking into consideration CHD and non-CHD lifetime costs for patients at low risk of CAD (non-smokers with diastolic blood pressure of 80mm Hg) through to those patients at high risk (smokers with a diastolic blood pressure of 100mm Hg). Not taking into consideration the benefit of elevated HDL-C levels, the investigators found a range of cost-effectiveness ratios, discounted as to costs and outcomes. The cost per LYS ranged from $Can30 366 for 50-year-old high-risk men to $Can257 170 in 30-year-old low-risk women. Again, the authors assumed all members of the cohort were 100% compliant with the medications.
5.1.2 Secondary Prevention Studies
6. Erratic Compliance — Economic Impact of Alternative Dosing Schedules
Metz and Lucas analysed the literature on alternate-day dosing of statin medications and the impact on this strategy on clinical outcomes. Although their review of the literature concluded that there are insufficient data to recommend alternate-day dosing routinely, the data presented suggest that there may be a clinical benefit.
Iliff in a 24-patient study, demonstrated beneficial lipid-lowering effects with atorvastatin 20mg once weekly, albeit not as great as those achieved with atorvastatin 10mg daily. It was concluded that the 20mg weekly dosage would be comparable to a patient taking 10mg twice weekly; in other words, 20mg once weekly would be the equivalent of a patient being 29% compliant with 10mg once-daily regimen (20mg/70mg).
Matalka and colleagues performed a randomised, placebo-controlled trial comparing atorvastatin 10mg every other day (alternate-day dosing group [ADDG]) and 10mg every day (daily dosing group [DDG]). This would be the equivalent of a person being 50% compliant.
Of the original 35 patients, 26 completed the 12-week study (14 in ADDG vs 12 in DDG). The authors did not specify the disposition of the nine dropouts, only that four dropped out from the ADDG and five from DDG. There were no significant differences between the groups. Dosage increases at the 6-week point had to be made in 79% of the ADDG versus 17% in the DDG. The average dosage at the end of the study was 18 mg/day in the ADDG group and 12 mg/day in the DDG. The end of study results (12 weeks) showed a 35% decrease in LDL-C levels in the ADDG group versus a 38% decrease in the DDG for a non-statistically significant difference of 3%. However, the authors did not provide an estimate of the statistical power related to this conclusion. The results also indicate that 43% of the ADDG achieved the pre-specified target LDL-C level goal versus 75% in the DDG, though this was not statistically tested. Using the average wholesale price, the authors concluded that ADD would result in a $US269 average annual savings (table III) [year of costing not reported].
These results must be interpreted with caution, since the sample size was small and the impact on clinical outcomes was not studied. Further, as noted by Morrison and Glassberg, the cost effectiveness of statins is primarily sensitive to the risk of developing CAD, not the cost of the drug itself. Also, the 32% lower achievement of desired LDL-C in the ADDG suggests that this strategy may not be as effective as needed in real life.
There exist few data on the relationship between the cost effectiveness of statin medications and medication compliance. The existing data suggest that for near 100% compliance versus initial non-compliance, the cost effectiveness of statin medications ranges from $US4500 to >$US250 000 per LYS depending on patient age, presence or absence of risk factors and whether the statin is being used for primary or secondary prevention. As for the erratic dosing behaviour, the existing data suggest, albeit weakly, that the more compliant patients achieve desired LDL-C levels more consistently, but at a higher medication cost. One study indicates that there is a statistically significant, but likely clinically unimportant, relationship between medication compliance and health outcome costs. However, the data were limited by the timeframe (≤1 year) and may not truly reflect the long-term benefit of compliance. More study needs to be conducted examining the relationship between medication compliance and cost effectiveness of statins in the primary and secondary prevention of CAD.
The authors have no conflicts of interest directly relevant to the content of this review. Each author has contributed equally to the development and writing of this review. No sources of funding were used in the preparation of this manuscript; however, Dr Petersen did receive a grant from Merck US Human Health to conduct pharmacoeconomic research in medication compliance, with a component of that research on statins.
- 1.National Center for Chronic Disease Prevention and Health Promotion. Preventing heart disease and stroke [online]. Available from URL: http://www.ede.gov/needphp/bbheartdisease/index.htm [Accessed 2004 May 1]
- 2.Task Force for Compliance. Noncompliance with medication regimens: an economic tragedy: emerging issues in pharmaceutical cost containing. Washington, DC: National Pharmaceutical Council, 1994: 1–32Google Scholar
- 3.McGhan WF, Peterson AM. Pharmacoeconomic impact of noncompliance. US Pharmacist Impact 2001, 3–13Google Scholar
- 5.Downs JR, Clearfield M, Weis S, et al. Primary prevention of acute coronary events with lovastatin in men and women with average cholesterol levels: results of AFCAPS/TexCAPS. Air Force/Texas Coronary Atherosclerosis Prevention Study. JAMA 1998; 279 (20): 1615–22Google Scholar
- 7.Scandinavia Simvastatin Survival Study group. Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet 1994; 344 (8934): 1383–9Google Scholar
- 11.Top 200 brand-name drugs by units in 2002. Drug Topics 2003; 60–2Google Scholar
- 13.Flanagan A, Gartenmann T, Lovich D, et al. The hidden epidemic: finding a cure for unfilled prescriptions and missed doses [online]. Available from URL: http://www.beg.com/publications/files/TheHiddenEpidemic_Rpt HCDec03.pdf [Accessed 2003 Dee 18]Google Scholar
- 16.Ernst FR, Grizzle AJ. Drug-related morbidity and mortality: updating the cost-of-illness model. J Am Pharm Assoc (Wash) 2001; 41: 192–9Google Scholar
- 17.Secnik K, Pathak DS, Cohen JM. Postcard and telephone reminders for unclaimed prescriptions: a comparative evaluation using survival analysis. J Am Pharm Assoc (Wash) 2000; 40: 243–51Google Scholar
- 24.Executive Summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA 2001; 285 (19): 2486–97Google Scholar
- 25.Mansur AP, Mattar APL, Tsubo CE, et al. Prescription and adherence to statins of patients with coronary artery disease and hypercholesterolemia. Arq Bras Cardiol 2001; 76: 115–8Google Scholar
- 42.Matalka MS, Ravnan MC, Deedwania PC. Is alternate daily dose of atorvastatin effective in treating patients with hyperlipidemia? The Alternate Day Versus Daily Dosing of Atorvastatin Study (ADDAS). Am Heart 2002; 144: 674–7Google Scholar