SEARCH

SEARCH BY CITATION

Keywords:

  • adherence;
  • compliance;
  • discontinuation;
  • dyslipidaemia;
  • HMG-CoA reductase inhibitors;
  • persistence;
  • primary prevention;
  • secondary prevention

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients And Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Objectives: The use of statins in primary prevention of cardiovascular disease is currently under debate. This study characterizes and identifies predictors of the persistence of use of statins in a clinical cohort of subsidized new users of similar age to the WOSCOPS trial subjects.

Methods: Medical, pharmaceutical, and demographic records for the period January 1, 1987 through December 31, 1994 were extracted from the databases of Québec's provincial health plan for a 10% random sample of social assistance recipients. Patients remained eligible for inclusion if they had received a first dispensation of a statin between January 1, 1987 and July 31, 1994. Persistence was defined as the number of days on treatment with a statin while continuing to renew dispensations within a defined time limit.

Results: New users of statins included 983 social assistance recipients who were observed for a total of 2,439,153 person-days. Median persistence on statin treatment was 173 (95% CI = 155, 204) days. Only 13% of patients persisted for 5 years of treatment. A higher index of chronic morbidity, pre-existing cardiovascular disease, and previous use of nicotinic acid were predictive of longer persistence on statin medication. Those patients whose first statin dispensation was for lovastatin discontinued treatment earlier than those whose first dispensation was for pravastatin or simvastatin.

Conclusions: New users showed low persistence on statins in a cohort of socially assisted persons aged 45–64, in spite of the minimal financial cost of the drug for such beneficiaries of Québec's provincial health plan.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients And Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The use of medications can be an effective way to reduce the occurrence of adverse clinical outcomes and the concomitant need for more costly technologic interventions by preventing the development (primary prevention) or progression (secondary prevention) of a number of chronic diseases. Lipid modifying agents have been increasingly prescribed in recent years to reduce the risk of cardiovascular disease (CVD). The United Kingdom, for example, experienced a more than 6-fold increase in the use of cholesterol lowering drugs between 1986 and 1992, and similarly impressive growth in the use of these drugs has been seen in other countries [1]. According to a recent Datamonitor report [2], it is expected that 50% of US residents will suffer from dyslipidaemia by the year 2002. Since life-long treatment is generally necessary in persons with abnormal lipids, the economic consequences of these trends are of great concern. Such concern is heightened by the fact that a substantial proportion of patients who are started on lipid modifying agents stop using them [3–6], thereby eventually losing any health benefit they might have gained from them, as well as wasting the money used to pay for the drugs.

Although it has been argued that statins are generally well tolerated and are thus associated with high patient compliance in comparison with the other classes of medication used for hyperlipidemia [7], few published studies have investigated patient compliance on statins among general clinical populations (Andrade et al. 1995, Simons 1996; Avorn et al. 1998; Eriksson et al. 1998) [3,4,6,8]. Only two of these, Andrade 1995 and Simons 1996, actually estimated the risk of drug discontinuation according to the duration of therapy using survival analysis methods. These studies were only able to provide rates of persistence (survival on therapy) up to one year following treatment initiation, however. Moreover, investigated predictors of discontinuation were limited to subjectively defined reasons for discontinuation. The objective of the current investigation is to study, in a representative sample from a large subsidized population, long-term persistence on statins among new users of these drugs, and to identify and quantify objective predictors of discontinuation.

Patients And Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients And Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Source of Data

The Régie de l'Assurance Maladie du Québec (RAMQ) is the provincial health insurance plan for the province of Québec; during the study period it provided complete drug and medical services coverage for persons receiving social assistance (welfare) and all persons over the age of 65. All claims data are captured by the RAMQ administrative database and encoded to preserve patient and physician anonymity prior to their release for research purposes. The accuracy and comprehensiveness of the RAMQ pharmaceutical data have been validated by Tamblyn et [9].

Cohort Definition

From the RAMQ database, we retrieved demographic information and all medical services and pharmaceutical records pertaining to the study period January 1, 1987 through December 31, 1994 for a 10% sample of social assistance beneficiaries. Patients remained eligible for inclusion in the cohort of new statin users if they had received a first (index) statin dispensation between January 1, 1987 and July 31, 1994, were between the ages of 45 and 64 at their index statin dispensation, and had remained alive for at least 4 months after the index statin dispensation.

The observation period for each cohort member included: 1) a medical antecedent period which extended from the individual's date of first physician or hospital visit following January 1, 1987, to the index statin dispensation date; 2) a pharmaceutical antecedent period which extended from the individual's date of first reimbursed prescription for any medication following January 1, 1987, to the index statin dispensation date; and 3) a follow-up period which extended from the start of the individual's index dispensation to the date of death or December 31, 1994.

Outcome Definitions

Two types of outcome variables were analyzed: a) the frequency of renewal of the first (index) statin dispensation (a dichotomous variable); and b) the persistence on treatment with any statin, in days from the date of the index dispensation (a continuous variable). As represented diagrammatically in part A of Fig. 1, we determined the frequency of renewal of the first statin dispensation in two ways: 1) the proportion of those in each cohort who ever renewed their initial (index) dispensation for a statin by the end of the follow-up period (December 31, 1994), regardless of any gap between the end of the first dispensation and the beginning of the renewal; and 2) the proportion of those in each cohort who were compliant renewers, defined as those who renewed their initial statin dispensation both within the study period and within the permissible time period (gap) between the prescribed end of the first dispensation and the date of dispensation of the next dispensation. The permissible gap following the first dispensation was defined as a time period equal to one half the duration of the index dispensation, or 7 days, whichever was longer. Patients who switched from one brand of statin to another were considered as continuing on statin therapy.

image

Figure 1. Diagrammatic representation of study outcomes.

Download figure to PowerPoint

As depicted in part B of Figure 1, the persistence on (duration of) statin treatment was defined as the number of days from the date of the start of the index dispensation (index date) to the time of first failure to continue renewals of statin with the permissible gap between dispensations. As above, the permissible gap between the end date of a given statin dispensation and the date of renewal was defined as one half of the duration of the given statin dispensation, or 7 days, whichever was greater. The total duration of treatment included the permissible gap following the final dispensation completed within the study period; for patients who remained on treatment at the end of the study period the duration of treatment was censored at that date. It should be noted that this definition of persistence corresponds to the duration of treatment during which adherence (proportion of pills dispensed which would have been consumed if taken as directed until the renewal date) remained above 66%; patients were considered to have effectively discontinued treatment at the point at which adherence dropped to less than 66%.

Statistical Analysis

Predictors of persistence on statin treatment which were suspected a priori from among the indicators available within the administrative databases of RAMQ included: age at index date; gender; previous use (i.e., during the pharmaceutical antecedent period) of other lipid modifying agents; previous use of antihyperglycemic agents; the type of statin initially prescribed (lovastatin, pravastatin, or simvastatin); the timing (year) of statin therapy initiation; the initial daily dose of statin (in mg per day); the duration of the index statin dispensation; a global index of chronic disease; and the pre-existence of cardiovascular disease. In addition, interactions between gender and two other variables (age and cardiovascular disease) were investigated.

The chronic disease score (CDS) was calculated [10] by assigning scores (0–5) to classes of drugs according to the severity of the disease for which they were prescribed; the CDS was equal to the sum of scores for drugs dispensed to the patients during the index statin dispensation.

Pre-existing cardiovascular disease was indicated by either of the following conditions: i) a diagnosis of ischemic heart disease (including acute myocardial infarction, other acute and subacute forms of ischemic heart disease, angina pectoris, and other forms of chronic ischemic heart disease), heart failure, atherosclerosis, or unspecified cardiovascular disease (ICD-9 codes 410, 411, 412, 413, 428, 440, or 429.2); or ii) any dispensations for drugs commonly prescribed for cardiovascular disease including: nitrate derivatives (used for angina), or digital and furosemide drugs used in combination (for heart failure).

Survival functions describing persistence on statin treatment were computed using the product-limit method [11]. Patients who remained on treatment at the end of the follow-up period were censored at that time, and contributed information to the survival curve (as cumulative probability of survival on treatment) only until that date.

For each of the two alternative definitions of renewal of the first dispensation and for each of the suspected predictors, crude stratified analyses were conducted by computing the frequency of renewal within strata defined by categories of the predictor. Crude and multivariate analyses to identify independent determinants of persistence on statin treatment were conducted using Cox's proportional hazards model. Potential modifiers of the assumed multiplicative relationship among model covariates were also investigated. These evaluated the potential for variable predictivity of age and cardiovascular disease status by gender. The final analysis included all variables observed to be significant independent determinants of the persistence on treatment. Statistical significance was defined using a liberal alpha level of 0.10.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients And Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Observation Periods

New users of statins included 983 social assistance recipients whose pharmaceutical observation included a total of 2,439,153 person-days (6678 person-years) and a median of 2874 days (8 years). The antecedent period had a median duration of 1761 days (range 79, 2762) in terms of medical services records, and a median duration of 1490 days (range 92, 2762) in respect to pharmaceutical records. The median follow-up (from the date of index dispensation) was 1032 days (range 156, 2191).

Baseline Characteristics

Characteristics of new users of statins are provided in Table 1, both for the entire cohort, and according to compliant (time-limited) renewal (or its complement, nonrenewal) of the index (first) statin dispensation. Of the 983 new statin users who comprised the cohort, 931 (95%) ever renewed their first dispensation of statin and 784 (80%) were compliant renewers who renewed their first dispensation of statin within the defined permissible period. The mean age at the date of the index dispensation was 56 years. Thirty-four percent of the cohort were males; the age distribution did not vary by gender. There was evidence of pre-existing cardiovascular disease (CVD) in 41% of the cohort, as indicated by either ICD-9 diagnoses or medications indicative of CVD in their antecedent medical or pharmaceutical records; 59% of the cohort were thus presumed to have been receiving statin treatment as a primary prevention strategy. Based on previous use of either oral antihyperglycemic agents or insulin, 12% of the cohort were medicated diabetics. Twenty-eight percent of the cohort had received dispensations for other lipid modifying agents (fibrates, resins, or nicotinic acid, in order of prevalence) in the pharmaceutical antecedent period prior to the index dispensation. The majority of index statin dispensations were for lovastatin (59%) with the remainder being for pravastatin (23%) or simvastatin (18%). The mean initial daily dose of statin was 18 mg.

Table 1.  Characteristics of new users of statin according to time-limited* renewal of the first dispensation
 All new users N = 983Non-renewers N = 199Compliant renewers N = 784
  • *

    Time-limited renewal refers to renewal of the first (index) statin dispensation within a permissible gap following the index dispensation, of duration equal to one half of the intended duration of the index dispensation or 7 days, whichever was greater.

  • P-value for chi-squared test <.05.

  • As indicated by previous diagnosis of cardiovascular disease (CVD) or previous use of CVD-related medications.

  • §

    As indicated by previous use of either oral antihyperglycemic agents or insulin.

  • Classes of drugs were assigned scores (0-5) according to the severity of the disease for which they were prescribed; the chronic disease score was the sum of the scores for other drugs dispensed to the patients during the index statin dispensation.

Continuous Variables   
Age at index dispensation (yrs)56.2 ± 5.255.5 ± 5.256.3 ± 5.1
Mean daily statin dose (mg)18.4 ± 5.918.9 ± 6.918.3 ± 5.7
Categorical Variables–no. of subjects (%)   
Male gender337 (34.3)61 (30.6)276 (35.2)
Pre-existing cardiovascular disease405 (41.2)66 (16.3)339 (83.7)
Medicated diabetic§122 (12.4)26 (13.1)96 (12.2)
Chronic Disease ScoreA   
0403 (41.0)106 (53.3)297 (37.9)
1–3360 (36.6)58 (29.2)302 (38.5)
geqslant R: gt-or-equal, slanted4220 (22.4)35 (17.6)185 (23.6)
Previous Use of Other Antihyperlipidemic Agent   
None705 (71.7)150 (75.4)555 (70.8)
Fibrates209 (21.3)34 (17.1)175 (22.3)
Resins50  (5.1)11  (5.5)39  (5.0)
Nicotinic Acid14  (1.4)4  (2.0)10  (1.3)
Other5  (0.5)0  (0.0)5  (0.6)
Index Statin   
Lovastastin578 (58.8)120 (60.3)458 (58.4)
Pravastatin231 (23.5)45 (22.6)186 (23.7)
Simvastatin174 (17.7)34 (17.1)140 (17.9)

As can be seen from Table 1, of the potential determinants studied, only two factors, chronic disease score and pre-existing CVD (as indicated by previous diagnoses, or medications for, CVD), were significant crude predictors of compliant renewal of the index statin dispensation. With respect to the criteria for defining pre-existing CVD, when analyzed separately, both previous CVD diagnoses (P = .019) and previous CVD medications (P = .009) were significant predictors of compliant renewal. The substantial agreement (83% overall) in patient classifications on CVD status by these criteria are evident from Table 2. The kappa statistic was equal to 0.61, indicating good agreement between the two methods [11] of CVD classification.

Table 2.  Cardiovascular disease status as indicated by medications or diagnoses in the relevant antecedent period
Antecedent medication for cardiovascular diseaseAntecendent diagnosis of cardiovascular disease*
 Number of persons (%)
 NoYesTotal
  • *

    As indicated by diagnoses of ischemic heart disease, including acute myocardial infarction, other acute and subacute forms of ischemic heart disease, angina pectoris, heart failure, atherosclerosis, or unspecified cardiovascular disease (ICD-9 codes 410, 411, 412, 413, 428, or 440), in the medical antecedent period; the median duration of the latter period was of 1761 days (range 79, 2762).

  • As indicated by dispensed medications for nitrate derivatives used for angina, or digital and furosemide drugs used in combination for heart failure in the pharmaceutical antecendent period; the median duration of the latter period was 1490 days (range 92, 2762).

No578 (59)90  (9)668  (68)
Yes77  (8)238 (24)315  (32)
Total655 (67)328 (33)983 (100)

Persistence on Statin Treatment

Figure 2 shows the survival curve depicting persistence on statin treatment among the entire cohort of new users. A steep drop (of 20%) in the number of patients remaining on treatment was observed 45 days after the index date, which corresponds to the end of the permissible renewal interval following a typical 30-day dispensation. Thereafter the curve continued to decline, though at a decreasing rate. Persistence at one year was only 33% (95% CI = 30%, 36%). Just over a third these patients (13%, 95% CI = 10%, 16%) persisted on treatment after 5 years. Median survival on statin treatment was 173 days (95% CI = 155, 204) or approximately six months.

image

Figure 2. Persistence on statin treatment among new users in a subsidized clinical cohort.

Download figure to PowerPoint

Determinants of Persistence on Statin Treatment

Table 3 provides crude and adjusted rate ratios and 95% CIs for the determinants of persistence on statin which were observed to be important within the cohort of new statin users based on Cox proportional hazards regression. As is evident from Table 3, for those factors that were significant independent predictors of persistence in multivariate analyses, only marginal differences were observed between crude and adjusted rate ratios. Figure 3 graphically depicts the adjusted rate ratios and 95% CIs. Figure 4 provides survival curves depicting persistence on treatment for categories defined by the predictors.

Table 3.  Crude and adjusted rate ratios from Cox Regression analyses predicting failure to persist on statin treatment
 CrudeAdjusted
Model covariateNEstimated Rate Ratio [95% CI]P ValueEstimated Rate Ratio [95% CI]P Value
  • *

    Classes of drugs were assigned scores (0–5) according to the severity of the disease for which they were prescribed; the chronic disease score was the sum of the scores for other drugs dispensed to the patients during the index statin dispensation.

  • As indicated by previous diagnosis of cardiovascular disease (CVD) or previous use of CVD-related medications.

Previous nicotinic acid use     
Yes14 0.59 [0.31, 1.41]0.1169 0.53 [0.27, 1.02]0.0562
No (Reference)9691.00 1.00 
Chronic disease score*     
4–14220 0.51 [0.42, 0.62]0.0001 0.60 [0.48, 0.74]0.0001
1–3360 0.71 [0.61, 0.83]0.0001 0.74 [0.63, 0.87]0.0003
0 (Reference)4031.00 1.00 
Pre-Existing Cardiovascular Disease     
Yes578 0.60 [0.51, 0.69]0.0001 0.71 [0.60, 0.84]0.0001
No (Reference)4051.00 1.00 
Index Statin     
Lovastatin405 1.17 [1.01, 1.35]0.0375 1.20 [1.04, 1.39]0.0001
Pravastatin or Simvastatin (Reference)5781.00 1.00 
image

Figure 3. Determinants of persistence on statin treatment.

Download figure to PowerPoint

image

Figure 4. Persistence on statin treatment by significant predictors.

Download figure to PowerPoint

In order of relative size, from those most strongly favoring longer persistence to those most predictive of earlier discontinuation (i.e., from smallest to largest adjusted RR point estimate in a model predicting failure to persist), independent predictors of persistence included previous use of the lipid modifying agent nicotinic acid, a relatively higher index of chronic disease, the pre-existence of cardiovascular disease, and type of statin initially dispensed.

Although few (14 of 983) of the new statin users had received nicotinic acid in the pharmaceutical antecedent period prior to their initial statin dispensation, these patients were (1–0.53) 47% less likely to discontinue treatment, or (1/0.53) 1.9 (95% CI = 0.98, 3.70) times more likely to persist with treatment, at any given time postindex (conditional on survival until that time) than those patients who had not (Table 3 and Figs 3 and 4a).

Those new statin users with greater levels of chronic disease also showed higher rates of persistence on statin treatment than those for whom the statin drug was the only drug dispensed during the index statin dispensation (Table 3 and Figs 3 and 4b). Individuals with a chronic disease score between 1 and 3 were 36% less likely to discontinue treatment, or 1.4 (95% CI = 1.15, 1.59) times more likely to persist with treatment, than those patients with a chronic disease score of 0. Cohort members with a chronic disease score of more than 4 were 40% less likely to discontinue treatment, or 1.7 (95% CI = 0.35, 2.08) times more likely to persist with treatment, than those patients with a chronic disease score of 0.

New statin users who were presumed, on the basis of previous CVD-related diagnoses or medications, to have pre-existing CVD at the time of their first statin dispensation, were 39% less likely to discontinue treatment, or 1.4 (95% CI = 1.19, 1.67) times more likely to persist with treatment than those patients who did not (Table 3 and Figs 3 and 4c).

The initial type of statin dispensed also influenced persistence on the drug class. Those new statin users for whom lovastatin was the index statin had a 20% higher rate of discontinuation than those whose first statin dispensation was for pravastatin or lovastatin (Table 3 and Figs 3 and 4d). The timing of statin therapy initiation (i.e., calendar year of the index dispensation) did not account for this trend.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients And Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

A subsidized clinical cohort of new statin users aged 45–64, whose first statin dispensation occurred between January 1, 1987 and July 31, 1994, demonstrated remarkably low persistence with these lipid modifying agents. The proportions of the cohort remaining on treatment at 1, 2, 3, 4, and 5 years following initiation, respectively, were 33% (95% CI = 27, 35), 24%, 17%, 14%, and 13%. Only two other published studies to date have investigated persistence on lipid modifying agents among new users using time-to-event (survival analysis) methods; both were limited in duration to one year. Among 138 participating pharmacies in Sydney, Australia, Simons [4] observed a 1-year rate of persistence of 43% (95% CI = 38, 48) on simvastatin, and 44% (95% CI = 37, 51) on pravastatin (the only two statins studied) between 1994 and 1995. The latter 1-year rates are consistent with those observed for these two agents in the present study (36%, 95% CI = 31, 41). In two Massachusetts HMOs between 1988 and 1990, Andrade et al. [3] observed a significantly higher 1-year persistence probability of 85% (95% CI = 71, 81) for lovastatin (the only statin studied) compared with that observed herein for this agent (31%, 95% CI = 27, 35).

Substantial under-ascertainment of discontinuation is likely in Andrade's study in which discontinuation was identified by one of three conditions: a switch to another type of lipid modifying agent; a 6-month gap between the last refill and the end of the study period or plan termination (regardless of the duration of the last refill); or omitted or inactive lipid modifying therapies flagged in clinical encounter files. In the present study, discontinuation was presumed on the basis of the failure of the patient to renew a completed dispensation of any statin agent within a permissible gap whose length was calibrated to the duration of the previous dispensation (being the maximum of seven days, or half of the intended duration of the previous prescription). The use of an arbitrary fixed gap by both Andrade et al. [3] and Simons et al. [4] introduced a degree of systematic error to the ascertainment of discontinuation that is consistent with the observed degree of interstudy discrepancy in estimated 1-year persistence rates. Since the median and mean pharmaceutical dispensation duration (at least in Québec) is 30 days [9], such error would have been much greater in the case of Andrade's study, which used a 6-month gap, than was the case in the study by Simons, in which the effective lag period varied between four and eight weeks.

In addition to the type of statin prescribed at the index dispensation, we identified three other independent predictors of persistence on statin treatment: previous use of nicotinic acid; comorbidity as indicated by the chronic disease score; and pre-existing cardiovascular disease (as indicated by prior medical diagnoses and CVD-related medication dispensations). After accounting for these factors, age at index date, gender, the timing (year) of statin therapy initiation, the initial daily dose of statin (in mg per day), duration of the index statin dispensation, and interactions between gender and two other variables (age and cardiovascular disease) were not predictive of statin persistence. It should be noted that conversions from one statin to another could not have affected our results, since a patient switching from one brand to another was considered as persisting on statin therapy.

The higher persistence observed among previous nicotinic acid users may be a selective phenomenon. The proportion of patients who never filled a second prescription or switched to another class of antilipid agent was observed by Avorn et al. [6] to be greater for nicotinic acid (89.1%) compared with cholestyramine, gemfibrozil, probucol or lovastatin (range 20–45%). The minority of patients who successfully persist with this relatively unfavourable agent are therefore more likely to persist with statin treatment when it is prescribed.

Our observation that persistence is better among patients with pre-existing cardiovascular disease confirmed similar findings by Simons [4] and Avorn et al. [6] In the latter study, Avorn et al. observed that the proportion of days covered by a dispensation (a measure which we term adherence) over one year was associated with coronary artery disease, hypertension, and diabetes. As did Avorn et al. we observed an association between persistence on statins and comorbidity (as indicated by medications received for other conditions including hypertension and diabetes). The importance of comorbidity may simply reflect behavioural habit; patients who are already consuming pharmaceuticals on a regular basis more readily persist with the addition of another pill to their existing drug-taking regimen than do patients who must establish a drug-taking regimen, particularly in the absence of a disease diagnosis or symptoms.

Andrade et al's [3] finding of a higher rate of discontinuation among women was observed only in our crude Cox regression analysis, but gender appeared to be a confounding factor associated with pre-existing CVD. Andrade et al. evidently did not investigate in multivariate analyses this or other possible factors that accounted for crude gender differences. The survival analyses by Andrade stratified only by subjectively inferred reasons for discontinuation (based on review of medical charts) categorized as either adverse effects, perceived drug ineffectiveness and noncompliance. Consistent with our negative findings with respect to gender, both Simons [4] and Avorn et al. [6] ruled out gender as an independent determinant of persistence on lipid modifying agents.

Advancing age is another potentially confounding factor that has been reported in other studies [3,4] to be predictive of persistence. As noted by Avorn et al. [6] old age is not a risk factor per se for reduced persistence, but rather is associated with chronic disease, which we accounted for in our analyses. We also controlled for confounding by age by restricting our cohort to those patients who were 45–64 years of age at the index dispensation.

The particular strengths of the present investigation include: a relatively long-term follow-up (up to 6 years postindex with a median of 3 years); a clear and consistent definition of persistence (equivalent to a minimum 66% adherence to any single dispensation) with a permissible gap calibrated to the duration of the previous dispensation; the investigation of a large set of objectively defined potential determinants of treatment persistence; control of potential confounders such as gender (through multivariate analysis) and age (through cohort restriction); and the investigation of potential modifying factors (through interaction terms defined a priori).

Previous analyses by our pharmacoepidemiology unit (for example, Desgagné and LeLorier [12]) have routinely used the cut-off value of 66%, permitting us to make consistent comparisons between drugs and classes of drugs. While this value (of 66%) may seem overly precise, it simply results from (admittedly arbitrarily) allowing a permissible time equal to one and one half times the intended duration of a prescription (or 7 days, whichever is greater) for a person to renew a dispensation. In these natural units, allowing a gap of one half the intended dispensation duration seemed reasonable. Concerned, however, that the above might be viewed as unduly strict, we undertook a sensitivity analysis which relaxed this definition such that a period double that intended by the prescriber was allowed to consider a patient compliant without substantive impact on the analytic results. Although median persistence was increased (from 170 to 244 days, or 0.47–0.67 years), it remained low, and the same set of predictors were identified as important determinants of persistence on statins.

This study of new statin users is subject to a number of limitations, however, the most prominent of which are discussed below. The first of limitations concerns the source of the data; we were limited herein to information routinely collected by the provincial health plan's administrative databases. The available data indicates only drugs dispensed, rather than drugs consumed. We could therefore infer persistence only from the fact of continued (renewed) dispensations. We may thus have overestimated persistence. Also unavailable to us was qualitative information such as one might ascertain from patient interview, e.g., subjective reasons for discontinuation of the drug such as a belief that the treatment is ineffective, or that the adverse effects were not balanced by enhanced feelings of wellness.

Another limitation related to the source of data is the fact that hospitals and other in-patient institutions do not submit drug claims data to RAMQ, and hence such prescription data is missing from the database. Some cohort members could have been admitted to institutions for a sufficient duration during the follow-up period to have been presumed discontinued from therapy. This would have led to an underestimate of persistence. Similar undetected losses could have occurred if members became employed during the period of follow-up. Unfortunately, no data are available with regard to this possibility.

The results of this study of a social assisted clinical population may be of limited generalizability to other socio-economic classes, regardless of subsidization of drug costs. According to Avorn et al. [6] persistence on lipid modifying agents over one year was related to socio-economic class, but not to drug costs among subsidized patients; Simons [4] also observed medication costs to be unrelated to persistence. The present study may therefore somewhat under-estimate persistence in a broader clinical population.

This analysis has revealed low levels of persistence on statins among social assistance recipients aged 45–64 in the province of Québec. To the extent that these rates are generalizable to clinical populations at large, the potential impacts both clinical and economic, of such low persistence, are alarming. Large long-term (5 years follow-up) clinical trials have demonstrated the efficacy of statins for both primary and secondary prevention of CVD in men of similar age with hyperlidaemia; in terms of the most conservative outcome, all-cause mortality, the WOSCOPS trial of pravastatin in 6595 subjects without CVD resulted in a 22% (95% CI = 0, 40; P = .051) greater survival probability among those treated compared to those on placebo [13]; the 4S trial of simvastatin in 4444 subjects with CVD [14] resulted in a 30% (95% CI = 14, 42; P = .0003) reduction in relative risk. In both trials, the statins were well tolerated, with similar frequencies of adverse events and study withdrawals in the treated and placebo groups. Because statins require sustained use for maintenance of their therapeutic action, however, the premature loss of patients from long-term treatment in general clinical practice constitutes a waste of health care dollars spent on medications that may ultimately benefit a fraction of those to whom they are dispensed.

Efforts must be made to educate patients of the importance to drug effectiveness of persistence; physicians considering prescribing statins for primary prevention may be justified in reserving their prescription to those patients who, once informed, express an intention to persist with treatment once it is prescribed.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients And Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The authors gratefully acknowledge helpful comments from Drs. L. Blais, A. Lacour, and E. Rahme, as well as those of Ms. O. Sheehy, and the administrative support of Anita Massicotte.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients And Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  • 1
    McKenney JM. The cost of treating dyslipidaemia using National Cholesterol Education Program (NCEP) guidelines. Pharmacoeconomics 1998;14: 1928.
  • 2
    Datamonitor. Treatment Algorithm: Improving outcomes in hyperlipidaemia? Media Release; Pharmacoeconomics Outcomes News, 1998;188.
  • 3
    Andrade SE, Walker AM & Gottleib LK, et al. Discontinuation of antihyperlipidemic drugs—do rates reported in clinical trials reflect rates in primary care settings? N Eng J Med 1995;332:112531.
  • 4
    Simons LA, Levis G & Simons J. Apparent discontinuation rates in patients prescribed lipid-lowering drugs. Med J Aust 1996;164: 20811.
  • 5
    Lacour A, Derderian F & LeLorier J. Comparison of efficacy and cost among lipid–lowering agents in patients with primary hypercholesterolemia. Can J Cardiol 1998;14:35561.
  • 6
    Avorn J, Monette J & Lacour A, et al. Persistence of use of lipid–lowering medications. JAMA 1998; 279:145862.
  • 7
    Czirak M. Clinical positioning of HMG-CoA reductase inhibitors in lipid management protocols. PharmacoEconomics 1998;14(Suppl. 3):2938.
  • 8
    Eriksson M, Hadell K, Holme G, Walldius G & Kjellstrom T. Compliance with and efficacy of treatment with prevastatin and cholestryamine: a randomized study on lipid-lowering in primary care. J Int Med 1998;243:37380.
  • 9
    Tamblyn R, Lavoie G, Petrella L & Monette J. The use of prescription claims databases in pharmacoepidemiological research: The accuracy and comprehensiveness of the prescription claims database in Quebec. J Clin Epidemiol 1995;48:9991009.DOI: 10.1016/0895-4356(94)00234-h
  • 10
    Von Korff M, Wagner EH & Saunders K. A chronic disease score from automated pharmacy data. J Clin Epidemiol 1992;45:197203.
  • 11
    Armitage P & Berry G. Statistical Methods in Medical Research. Oxford: Blackwell Scientific Publications, 1994.
  • 12
    Desgagné A & LeLorier J. Incontinence drug utilization patterns in Québec, Canada. Value Health 2000;2:4528.
  • 13
    Shepherd J, Cohen DJ & Ford I, et al. Prevention of coronary artery disease with pravastatin in men with hypercholesterolemia. N Engl J Med 1995; 333:13017.
  • 14
    Scandinavian Simvastatin Survival Study Group. Randomized trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet 1994; 344:13839.