Effect of Statin Adherence on Cerebrovascular Disease in Primary Prevention
Article Outline
Abstract
Background
Evidence from meta-analyses shows that statin therapy reduces all-cause mortality and nonhemorrhagic strokes. Nonadherence to statins may reduce this protective effect. The association between statin adherence and incidence of cerebrovascular disease remains unexplored outside the context of clinical trials.
Objective
To evaluate the impact of statin adherence on the occurrence of cerebrovascular disease in a real clinical setting.
Methods
A cohort of 112,092 patients was reconstructed using the Régie d'assurance maladie du Québec and Med-Echo databases. The Régie d'assurance maladie du Québec database contains information from 3 types of health-related data, such as demographic information, medical data, and the prescription claims file. The Med-Echo database contains data on acute care hospitalizations on all Quebec residents. All patients without cardiovascular disease aged 45-85 years who were newly treated with statins between 1999 and 2004 were eligible. A nested case-control design was used to study the occurrence of cerebrovascular disease. Adherence level was reported as a medication possession ratio. Conditional logistic regression models were used to estimate the rate ratio of cerebrovascular disease, adjusting for different covariables.
Results
The mean patient age was 63 years; 49% had hypertension, 21% had diabetes, and 41% were males. Nonadherence was prevalent because only 55% of the patients were exposed to ≥80% of the medication during follow-up. We did not observe any major differences, defined as more than 5%, between the groups, except for the sex, diabetes, and hypertension. High level of adherence to statins was associated with a reduction of cerebrovascular events (rate ratio: 0.74; 95% confidence interval, 0.65-0.84).
Conclusions
Our study suggests a relatively low level of adherence to statins, but more importantly, that adherence is associated with a risk reduction for cerebrovascular disease. Adherence to statin therapy needs to be improved, so that patients can benefit from the full protective effects of statin therapy.
Keywords: Adherence, Cerebrovascular disease, Observational study, Statins
Stroke is one of the leading causes of death in North America, and it also is a leading cause of permanent disability.1, 2 Stroke also is the second largest contributor to hospital care costs for cardiovascular disease.2 Age, sex, smoking, hypertension, diabetes, and atrial fibrillation are well-established risk factors for stroke.1 In addition, dyslipidemia is another documented modifiable risk factor for stroke that is attributable for 10%-15% of ischemic strokes.1
Statins reduce the risk of stroke among patients with coronary artery disease, and those at increased risk for cardiovascular disease in general, although their effect in patients without coronary artery disease has not yet been well established.3, 4, 5, 6, 7, 8, 9 More recently, studies have confirmed that statins reduced the risk of first stroke in patients with coronary artery disease (Treating to New Targets Investigators),10 in other high-risk populations, such as mainly diabetics (Heart Protection Study Collaborative Group4 and Collaborative Atorvastatin Diabetes Study6) and hypertensives (Anglo-Scandinavian Cardiac Outcomes Trial5), even with a normal baseline blood cholesterol level, arguing for a global cardiovascular risk-based treatment strategy. In addition, recent meta-analyses have demonstrated that statins can reduce the risk of ischemic stroke in primary prevention by 14%-23%.11, 12 Additional data from the literature have confirmed a protective effect for ischemic strokes without a definite increase in hemorrhagic strokes; this benefit, however, appears after more than 1 year of treatment.13, 14, 15, 16
The use of statins in the clinical setting is suboptimal. Indeed, after 1 year of treatment, only 40% of patients are still adherent to their statin therapy (ie, take at least 80% of their prescribed dose) and only 65% of patients continue their therapy.17, 18 Nonadherence is likely an important source of preventable cerebrovascular morbidity and mortality. Unfortunately, few data are available on this specific area. The purpose of our study is to evaluate the impact of statin adherence on the rate of cerebrovascular disease among patients 45-85 years of age in a general population setting in individuals without a history of cardiovascular disease.
Methods
Data Sources
This observational study was done using the databases of Med-Echo and the Régie de l'assurance maladie du Québec (RAMQ), which administers the Quebec health insurance as well as the public prescription drug insurance plans. The Med-Echo database contains data on acute care hospitalizations, such as date of admission and length of stay as well as primary and secondary diagnoses, and includes information on all Quebec residents. The RAMQ database contains information from 3 types of health-related data. The demographic information file contains the age, sex, year of death, and postal code for all registered individuals. The medical data file contains all the information relative to medical services received: type, date, and location of medical service (inpatient or ambulatory); as well as the diagnostic codes, which are identified according to the International Classification of Diseases, 9th Revision (ICD-9).19 The procedure codes also are enclosed, and are defined by the Canadian classification of diagnostic, therapeutic, and surgical procedures.20 These procedure codes are carefully audited. The prescription claims file contains data on prescription medications that are dispensed to patients insured by the RAMQ and include common denomination, dose, amount of medication given to the patient, prescription period, and specialty of the prescribing physician.
The first 2 datasets contain information on all residents covered by the provincial health insurance plan, which is the entire Quebec population. Data from the prescription claims dataset are from beneficiaries of the provincial prescription drug insurance plan, which represents about 40% of the Quebec population aged 45-64 years and about 94% of those 65 years and older.21 The 3 databases are linked by the medical insurance number of the patients. The prescription claims database has been used in pharmacoepidemiological studies and is an accurate mean of determining drugs dispensed to individuals.22 Other studies have assessed the validity of administrative hospital discharge data.23 For ischemic stroke (ICD-9 codes 433, 434, and 436), the sensitivity was 80%, specificity was 96%, and positive predictive value was 91%, using the first 2 discharge diagnoses; for intracerebral hemorrhage (code 431), those values were 85%, 94%, and 83%, respectively.24
Cohort Study
Using the RAMQ databases, we constructed a cohort of patients who initiated a statin treatment (atorvastatin, fluvastatin, lovastatin, pravastatin, rosuvastatin, or simvastatin) between January 1, 1999 and December 31, 2004. The date of the first statin prescription was defined as the cohort entry date, and these patients did not take any lipid-lowering drugs in the 2 years preceding their cohort entry. Patients had to be 45-85 years old at cohort entry and had to be covered by the public prescription drug insurance plan for at least 2 years before cohort entry.
To be eligible, subjects could not have any indication of cardiovascular disease, as evidenced by the absence of a related diagnosis or medical procedure in the last 5 years, and any vascular drug marker in the 2 years before the cohort entry date. Patients had to be free of any marker of cardiovascular disease such as: 1) Coronary artery disease: myocardial infarction, angina, or other forms of ischemic cardiopathies (ICD-9: 410-414); vascular medical procedure such as a vascular endoprosthesis, angioplasty, coronary artery bypass graft surgery; or use of nitrate vasodilators; 2) Cerebrovascular disease (diagnostic codes ICD-9: 430-438): medical procedure such as endartectomy; use of nimodipine; 3) Chronic heart failure (diagnostic code ICD-9: 428): use of furosemide alone or in combination with digoxine, angiotensin-converting enzyme inhibitors, spironolactone, beta-blockers, or carvedilol; 4) Arrhythmias (diagnostic code ICD-9: 427): medical procedures received using a stimulator or an electric fibrillator; use of anti-arrhythmia medication (amiodarone, digoxine, quinidine, disopyramide, flecaidine, mexiletine, procainamide, propafenone, or sotalol); 5) Peripheral artery disease (diagnostic codes ICD-9: 440-447): medical procedure codes of noncoronary revascularization; use of pentoxifylline. The RAMQ drug database also was used to exclude patients who received other drugs such as antiplatelets (excluding low dose of aspirin) or anticoagulants during the 2 years preceding cohort entry.
The study cohort included 112,092 individuals who were followed from cohort entry until their first cerebrovascular disease or until the end of the study period (June 30, 2005). Patients were censored during follow-up when one of the following had occurred: death, end of coverage by the RAMQ drug insurance plan, or a prescription for another class of lipid-lowering drugs (eg, fibrates). Subjects were followed for a minimum of 6 months and up to a maximum of 6.5 years. All cardiovascular events, including all strokes and all-cause mortality rates, were assessed in the cohort. Cardiovascular death was defined as death within 30 days of hospitalization for cardiovascular disease.
Nested Case-Control
A nested case-control design was used to estimate the rate ratio (RR) of cerebrovascular disease in relation to statin adherence level. Fatal and nonfatal cerebrovascular disease was defined as a composite endpoint of: intracerebral hemorrhage (ICD-9: 431), occlusion and stenosis of precerebral arteries (ICD-9: 433), occlusion of cerebral arteries (ICD-9: 434), acute cerebrovascular disease but ill-defined cerebrovascular disease (ICD-9: 436), and other and ill-defined cerebrovascular disease (ICD-9: 437).
All cases of cerebrovascular disease were identified and up to 15 controls were selected at random from the source population (cohort) using density sampling.25, 26 The controls had the same age (± 1 year) at the cohort entry date and the same follow-up period as the cases. A subject selected to be a control was always eligible to become a case, and a control could be selected more than once.
Exposure Assessment
For both cases and controls, adherence was measured using a medication possession ratio (MPR),27 which was obtained by calculating the percentage of days exposed to statins in a given follow-up period. The MPR was calculated using prescription dates, the amount of medication dispensed to the patient, and the prescription period. Adherence for cases was calculated from the cohort entry date to the date of cerebrovascular disease event, while for the controls it was from the cohort entry date to the date of selection. The date of cerebrovascular disease for the cases and the date of selection for the controls were defined as the index date. MPR was evaluated as a categorical variable, that is, <20% (reference group), 20%-39%, 40%-59%, 60%-79%, and ≥80%. Based on literature data,28, 29, 30 equivalency in lowering low-density lipoprotein (LDL)-cholesterol levels was estimated.
Covariables
Sex and social assistance were identified at cohort entry. The occurrence of coronary artery disease, chronic heart failure, peripheral artery disease, or other cardiovascular disease conditions was assessed during follow-up. Diabetes and hypertension were identified in the year before entry in the cohort and during follow-up by ICD-9 code or drug markers. Patients with diabetes or hypertension diagnosed in the year preceding the index date were considered newly diagnosed. As for the other patients, the use of antidiabetic or antihypertensive agents in the year before the index date was dichotomized into 2 levels: high adherence, indicated by having filled at least 80% or more of the prescribed doses, and low adherence <80%. Patients who were diagnosed with diabetes mellitus or hypertension but never treated were defined as such. The reference categories were individuals with no hypertension or diabetes. Dichotomized variables also were defined for the use of antiplatelet drugs (including aspirin in daily doses <650 mg among patients without any cardiovascular conditions). Finally, an updated patient chronic disease score (CDS) was calculated in the year preceding the index date. The CDS is a comorbidity index that uses drugs dispensed as surrogate markers for chronic illness instead of using clinical diagnoses.31 Scores are weighted according to the number of different chronic diseases under treatment.31 The CDS was dichotomized into 2 levels: CDS ≥4 or <4.
Statistical Analysis
Characteristics of cases and controls were compared using the t test for continuous variables and the chi-squared test for categorical variables. In multivariable analysis, a conditional logistic regression model was constructed to evaluate the association between statin adherence and all fatal and nonfatal cerebrovascular disease. When studying an exposure that varies with time, as was the case in our study (adherence to statins), an additional level of complexity is introduced by the need to account for time-dependent exposure in the analysis; and this can be accomplished by cohort analysis using Cox regression, including time-dependent covariates. Alternatively, a nested case-control approach can be used providing the exposure and covariate information for controls reflects the values corresponding to the time of selection of their respective case. Nested case-control analyses have been found to yield results that were similar to results of Cox regression on the full cohort when studying time-dependent exposures, with the advantage of superior computational efficiency with the conditional logistic regression, given that only a sample of all possible controls are included in the risk set of each case.26, 32
Unselected multivariable models were constructed to maximally adjust for confounding, and included all variables described previously. RRs and 95% confidence intervals were calculated for each independent variable in the multivariable models. The possible effect of time was taken into account by stratifying the analysis by the time of case occurrence (cases occurring in the first year of follow-up and those after 1 year of follow-up).
In the secondary analyses, the association between statins adherence and ischemic stroke (ICD-9: 436, 437) and hemorrhagic stroke (431) was done. Thereafter, the association between statins adherence and fatal and nonfatal cerebrovascular disease also was evaluated across different patient sub-groups, including different age groups and presence of hypertension or diabetes.
To further assess the robustness of our findings, we performed additional analyses using alternative definitions of primary outcomes, such as with the exclusion of other and ill-defined cerebrovascular disease (ICD-9: 437). Finally, to assess the robustness of our results regarding potential biases due to unmeasured confounders, we used the Monte Carlo method proposed by Greenland and Steenland, which considers that an unmeasured risk factor is less frequent among adherent than nonadherent patients.33, 34 This type of analysis was performed because one limitation of administrative databases is that they do not contain information on potential confounders such as smoking or other lifestyle habits. Different scenarios were developed in which we changed the odds ratio between the unmeasured confounder and cerebrovascular disease, as well as changing the prevalence of this confounder between adherent and nonadherent patients. The odds ratios and prevalence of confounders were taken from the literature and surveys. This allowed us to determine how the RR would change when adjusted for the unmeasured confounders.
Residuals from regression models were assessed for violations of the assumptions of multicollinearity or deviance.35, 36 All the analyses used a precision threshold of 5% and were done using SAS version 9.1 (SAS Institute, Cary, NC).
Ethical Considerations
No patient or physician could be identified due to the scrambled identification numbers that were used throughout the study. The University of Montreal's Research and Ethics Committee approved this study.
Results
Characteristics of Patients
The distribution of inclusion and exclusion criteria is shown in the Figure. Of the 112,092 patients, most were prescribed atorvastatin, simvastatin, or pravastatin (Table 1). The mean age of the patients was 63 years (SD
±10), and the mean follow-up time was 2.95 years. Baseline demographic characteristics were comparable across the different statin groups. Table 2 provides demographic and clinical characteristics of the patients at the cohort entry date with respect to the adherence level. We did not observe any major differences, defined as more than 5%, between the groups except for the sex, diabetes, and hypertension.
Figure.
Flow chart of inclusion and exclusion criteria. *Exclusion criteria were assessed in the 2 years preceding cohort entry for medications and in the 5 years preceding cohort entry for hospitalizations, diagnoses, or medical procedures.
Table 1. Characteristics of Patients Starting a New Statin Treatment in the RAMQ in 1999-2004
| Atorvastatin | Fluvastatin | Lovastatin | Pravastatin | Rosuvastatin | Simvastatin | |
|---|---|---|---|---|---|---|
| No. patients | 70,968 | 2117 | 750 | 13,559 | 7927 | 16,771 |
| Mean age (continuous)⁎ | 63(±10) | 63(±10) | 63(±10) | 64(±10) | 63(±10) | 64(±10) |
| Mean dose (mg) | 13(±6) | 26(±10) | 21(±6) | 20(±8) | 11(±4) | 18(±10) |
| Follow-up time (days) | 1063(±601) | 1531(±598) | 1485(±649) | 1324(±632) | 430(±134) | 1167(±596) |
| Sex (% male) | 42 | 35 | 35 | 37 | 46 | 41 |
| Social assistance (%) | 16 | 18 | 19 | 15 | 14 | 14 |
| Diabetes mellitus (%)† | 27 | 20 | 19 | 24 | 26 | 19 |
| Hypertension (%)† | 54 | 52 | 47 | 53 | 57 | 56 |
| Antiplatelet use (%) | 32 | 28 | 26 | 33 | 32 | 37 |
| Dose distribution (mg) (%)⁎ | ||||||
| 5 | 1 | 0 | 0 | 0 | 1 | 6 |
| 10 | 78 | 0 | 5 | 21 | 96 | 42 |
| 20 | 20 | 60 | 89 | 71 | 3 | 43 |
| 40 | 1 | 40 | 6 | 8 | 1 | 9 |
| 80 | 0 | 0 | 0 | 0 | 0 | 0 |
⁎At treatment initiation. |
†ICD-9 or pharmacologic treatment in the year before cohort entry. |
Table 2. Characteristics of Patients Who Were Adherents and Nonadherents
| Low Adherence <80% | High Adherence ≥80% | |
|---|---|---|
| Number of subjects | 26,585 | 28,549 |
| Age (mean±SD)⁎ | 62.5±10 | 63.5±9 |
| Statins at entry date (%) | ||
| Atorvastatin | 60.7% | 61.2% |
| Fluvastatin | 3.1% | 2.2% |
| Lovastatin | 1.2% | 0.9% |
| Pravastatin | 15.8% | 15.2% |
| Simvastatin | 17.2% | 17.0% |
| Rosuvastatin | 2.0% | 3.5% |
| Male | 40% | 39% |
| Social aid (yes/no)⁎ | 16% | 18% |
| Use of antiplatelets† | 12.4% | 14.9% |
| Having a coronary artery disease during follow-up‡ | 5.7% | 4.9% |
| Having a chronic heart failure during follow-up§ | 1.3% | 1.2% |
| Having a peripheral artery disease during follow-up∥ | 1.9% | 1.3% |
| Having other cardiovascular conditions during follow-up¶ | 4.0% | 3.3% |
| Having ≥2 cardiovascular conditions during follow-up | 9.6% | 8.3% |
| Diabetes⁎⁎ | 27.2% | 32.9% |
| Hypertension†† | 64.3% | 70.9% |
| Chronic disease score¶‡‡ (≥4) (%) | 18.4% | 19.8% |
⁎At treatment initiation. |
†Antiplatelet users: use of dipyridamole, sulfinpyrazone, ticlopidine, dipyridamole+ AAS, clopidogrel, aspirin in doses ranging from 80 mg to 650 mg, and that without coronary artery disease, chronic heart failure, peripheral artery disease or other cardiovascular disease. |
‡Diagnosis of coronary artery disease: MI or angina (ICD-9: 410-414), a medical procedure, ie, coronary artery bypass grafting, angiography, or angioplasty, use of nitrate, including nitroglycerin. |
§Diagnosis of a chronic heart failure: diagnosis (ICD-9: 428) or drug markers. |
∥Diagnosis of peripheral artery disease: diagnosis (ICD-9: 440-447), medical procedure of noncoronary angioplasty, or use of pentoxifylline). |
¶Diagnosis of other cardiovascular disease such as arrhythmia: diagnosis (ICD-9: 427), a medical procedure using a pacemaker and the use of drugs for cardiac arrhythmias (amiodarone, digoxin, quinidine, disopyramide, flecainamide, mexiletine, procainamide, propafenone, or sotalol); or patients who received anticoagulants. |
⁎⁎ICD-9 or pharmacologic treatment of diagnosed diabetes detected in the year preceding the cohort entry or during the follow-up.Hypertension as followed: |
††ICD-9 or pharmacologic treatment of diagnosed hypertension detected in the year preceding the cohort entry or during follow-up. |
‡‡In the year preceding the index date. |
In the cohort, 41% of patients were men, 16% were welfare recipients, 26% had diabetes, and 54% had hypertension. Of the 3962 cerebrovascular events identified in our study, 68% were ischemic strokes, 7% were intracerebral hemorrhages, and 25% were classified as carotid atherosclerosis of precerebral or cerebral arteries.
Among the 112,092 patients in the cohort, 3.5% had a cerebrovascular event (1.2 per 100 person-years), 12.5% had a coronary artery disease event (4.2 per 100 person-years), 4.0% had chronic heart failure (1.4 per 100 person-years), 3.6% had a peripheral artery disease condition (1.2 per 100 person-years), 10.6% had another cardiovascular diagnosis (3.6 per 100 person-years), and 32.9% took antiplatelet drugs (11.1 per 100 person-years) without any diagnosed cardiovascular disease. The percentages of cardiovascular mortality and all-cause mortality were 0.4% and 2.9%, respectively.
The proportion of men, welfare recipients, patients with diabetes or hypertension, those who developed a cardiovascular condition during follow-up, and patients with a higher chronic disease score was significantly higher among cases (Table 3).
Table 3. Characteristics of Patients with Cerebrovascular Disease and Matched Controls
| Cases Occurring in the First Year of Follow-up and Their Controls | Cases Occurring after the First Year of Follow-up and Their Controls | |||||
|---|---|---|---|---|---|---|
| Cases | Controls | P Value⁎ | Cases | Controls | P Value⁎ | |
| Number | 1366 | 20,425 | 2593 | 38,547 | ||
| Age (years)† | 68(±9) | 68(±9) | .9842 | 67(±9) | 67(±9) | .7346 |
| Mean equivalent dose (mg)‡ | 23(±13) | 23(±13) | .1081 | 23(±14) | 22(±13) | .001 |
| Statin adherence (%)§ | ||||||
| <20% | 4 | 5 | .52 | 16 | 14 | .005 |
| 20-39% | 6 | 6 | 10 | 9 | ||
| 40-59% | 7 | 6 | 10 | 10 | ||
| 60-79% | 9 | 9 | 13 | 13 | ||
| ≥80% | 74 | 75 | 51 | 55 | ||
| Sex (% male) | 45 | 39 | <.0001 | 43 | 37 | <.0001 |
| Social assistance (%) | 12 | 10 | .0033 | 14 | 11 | <.0001 |
| Use of antiplatelets (%)∥ | 15 | 12 | .0008 | 15 | 15 | .7952 |
| Having a coronary artery disease during follow-up (%)¶ | 4 | 2 | <.0001 | 9 | 7 | <.0001 |
| Having a chronic heart failure during follow-up (%)⁎⁎ | 1 | 0.4 | <.0001 | 3 | 2 | <.0001 |
| Having a peripheral arterial disease during follow-up (%)†† | 5 | 1 | <.0001 | 5 | 2 | <.0001 |
| Having other CVD condition during follow-up (%)‡‡ | 4 | 1 | <.0001 | 7 | 5 | <.0001 |
| Having ≥2 CVD conditions during follow-up (%) | 11 | 2 | <.0001 | 26 | 11 | <.0001 |
| Diabetes (%) | 31 | 30 | .0356 | 39 | 32 | <.0001 |
| Diagnosed and untreated diabetes (%)§§ | 5 | 6 | .012 | 7 | 7 | <.0001 |
| Newly diagnosed (%)§§ | 7 | 7 | 2 | 2 | ||
| Antidiabetic agents adherence <80% (%)∥∥ | 5 | 4 | 10 | 6 | ||
| Antidiabetic agents adherence ≥80% (%)∥∥ | 14 | 13 | 20 | 17 | ||
| Hypertension (%) | 71 | 59 | <.0001 | 80 | 70 | <.0001 |
| Diagnosed and untreated hypertension (%)§§ | 4 | 5 | <.0001 | 6 | 6 | <.0001 |
| Newly diagnosed (%)§§ | 25 | 14 | 7 | 4 | ||
| Antihypertensive agents adherence <80% (%)∥∥ | 8 | 6 | 17 | 13 | ||
| Antihypertensive agents adherence <80% (%)∥∥ | 35 | 34 | 50 | 47 | ||
| Chronic disease score (≥4) (%) | 22 | 16 | <.0001 | 29 | 20 | <.0001 |
⁎P values are related to analyses made to compare cases with controls. |
†At treatment initiation. |
‡Statins equivalence in simvastatin dose.34, 35 Simvastatin 10 mg |
§Proportion of days covered (%). |
∥Antiplatelet users: use of dipyridamole, sulfinpyrazone, ticlopidine, dipyridamole+ AAS, clopidogrel, aspirin in doses ranging from 80 mg to 650 mg, and that without coronary artery disease, peripheral artery disease, chronic heart failure or other CVD conditions. |
¶Diagnosis of coronary artery disease: MI or angina (ICD-9: 410-414), a medical procedure, ie, coronary artery bypass grafting, angiography, or angioplasty, use of nitrate, including nitroglycerin. |
⁎⁎Diagnosis of chronic heart failure: diagnosis (ICD-9: 428) or drug markers. |
††Diagnosis of peripheral artery disease: diagnosis (ICD-9: 440-447), medical procedure of noncoronary angioplasty, or use of pentoxifylline). |
‡‡Diagnosis of other cardiovascular disease: arrhythmia (ICD-9: 427), a medical procedure using a pacemaker, or use of drugs for cardiac arrhythmias (amiodarone, digoxin, quinidine, disopyramide, flecainamide, mexiletine, procainamide, propafenone, or sotalol); or valvular heart disease; or anticoagulants. |
§§ICD-9 or pharmacologic treatment; newly diagnosed diabetes or hypertension was detected in the year preceding the index date. |
∥∥Proportion of days covered (%) in the year preceding the index date. |
Impact of Adherence on Cerebrovascular Disease and Risk Factors
The mean high adherence level to statins was close to 98% during the first year and 95% after 1 year of follow-up; and those values were at 13% and 9%, respectively, for the adherence level lower than 20%. Multivariate analysis revealed that the benefits of statin adherence on cerebrovascular disease are only apparent after at least 1 year of follow-up. An adherence level of ≥80% reduced the RR of cerebrovascular disease by 26% (RR 0.74, 95% confidence interval [CI], 0.65-0.84) compared with an adherence of <20% (Table 4). Finally, compared with atorvastatin, there was no significant difference with the other statin agents (data not shown).
Table 4. Rate Ratio (RR) of Cerebrovascular Disease
| RR (95% CI) | ||||
|---|---|---|---|---|
| Cases Occurring in the First Year of Follow-up and Their Controls | Cases Occurring after 1 year of Follow-up and Their Controls | |||
| Crude | Adjusted | Crude | Adjusted | |
| Statin adherence (%)⁎ | ||||
| <20% | Reference | Reference | Reference | Reference |
| 20%-39% | 1.17(0.82-1.67) | 1.06(0.74-1.53) | 1.01(0.86-1.19) | 0.97(0.83-1.15) |
| 40%-59% | 1.34(0.95-1.90) | 1.24(0.87-1.77) | 0.90(0.76-1.06) | 0.83(0.70-0.98) |
| 60%-79% | 1.20(0.86-1.67) | 1.09(0.77-1.53) | 0.93(0.80-1.07) | 0.82(0.71-0.96) |
| ≥80% | 1.14(0.86-1.53) | 1.03(0.76-1.38) | 0.83(0.74-0.93) | 0.74(0.65-0.84) |
⁎Proportion of days covered (%). |
Subgroups Analysis
A subgroup analysis by the type of cerebrovascular disease revealed that an adherence of ≥80% compared with <20% reduced the RR of ischemic stroke (RR 0.67, 95% CI, 0.58-0.77), while no difference was seen for hemorrhagic stroke (RR 0.97, 95% CI, 0.59-1.58). Subgroup analysis by age showed that the risk reduction for cerebrovascular disease was nonsignificant for patients younger than 65 years (RR 0.80, 95% CI, 0.65-1.00); while a significant 28% reduction was found for patients 65 years and older (RR 0.72, 95% CI, 0.63-0.84). Among high-risk patients; that is, those with hypertension or diabetes, an adherence level of ≥80% reduced the risk of cerebrovascular disease by 29% (RR 0.71, 95% CI, 0.60-0.83), which is almost equivalent to the risk reduction found (RR 0.72, 95% CI, 0.58-0.89) for lower-risk patients (those without hypertension or diabetes).
Sensitivity Analyses
For the sensitivity analyses, we estimated the overall effect (RR 0.80, 95% CI, 0.71-0.89). When other and ill-defined cerebrovascular disease (ICD-9: 437) was excluded from the outcome, the point estimate in the RR was only slightly modified (RR 0.77, 95% CI, 0.67-0.88). As shown in Table 5, we considered that an unmeasured risk factor was less frequent among subjects with a high adherence level compared with those with a low adherence level. This analysis revealed that our conclusions might be overturned only when a confounder (eg, smoking) had an odds ratio for cerebrovascular disease of 3.0 or 4.0 for current smokers and 2.0 for former smokers, in addition to a prevalence in the high adherence group of 15% smokers (8% current and 7% former smokers) as compared with 25% (18% current and 7% former smokers) in the low adherence group (scenarios 5 and 6).
Table 5. Change in RR of Cerebrovascular Disease after Adjustment for Unmeasured Confounders
| Prevalence of Risk Factor⁎ (High Risk; Medium Risk)† | OR‡ (95% CI) | Estimated RR (95% CI) of CD | |||
|---|---|---|---|---|---|
| High Adherence Group | Low Adherence Group | High Risk† | Medium Risk† | ||
| Scenarios 1 | 15%(8%;7%) | 19%(12%;7%) | 1.8(1.2-2.3) | 1.3(1.1-1.8) | 0.77(0.69-0.87) |
| Scenarios 2 | 15%(8%;7%) | 19%(12%;7%) | 2.5(1.6-3.5) | 2.0(1.2-3.0) | 0.79(0.70-0.89) |
| Scenarios 3 | 15%(8%;7%) | 25%(12%;13%) | 1.8(1.2-2.3) | 1.3(1.1-1.8) | 0.79(0.70-0.89) |
| Scenarios 4 | 15%(8%;7%) | 25%(18%;7%) | 3.0(1.2-4.0) | 2.0(1.1-3.0) | 0.87(0.75-1.04) |
| Scenarios 5 | 15%(8%;7%) | 25%(18%;7%) | 4.0(1.2-5.0) | 2.0(1.1-3.0) | 0.92(0.77-1.12) |
⁎Proportion at high risk and medium risk among high-adherence and low-adherence groups. |
†High risk; medium risk are defined as smokers (current smokers or former smokers) or obesity (severe or moderated). |
‡Risk factor between the confounder and cerebrovascular disease. |
Discussion
The present study was designed to assess the use and effectiveness of statin therapy in a real-life setting and, in particular, to focus on the potential impact of statin adherence on the occurrence of cerebrovascular disease. The assessment of effectiveness based on administrative databases offers several advantages, such as the ability to access a large and representative study population and the capacity to determine medication use in a real-world setting.
Our results indicate that higher adherence levels were associated with greater reductions in cerebrovascular disease occurrence. The analyses revealed that an adherence level of ≥80% reduced the risk of cerebrovascular disease by 26%, compared with an adherence level of <20%. This result is similar to the 27% risk reduction found in the primary prevention Anglo-Scandinavian Cardiac Outcomes Trial–Lipid Lowering Arm trial, and to the 19%-23% risk reduction found in recent meta-analyses of primary prevention studies.11, 13 The decrease in cerebrovascular disease risk found in our study also is comparable with other studies that examined the relationship of LDL cholesterol as a marker for stroke;13, 16 all strokes were reduced by 21%-27%, ischemic stroke by 22%, and there was no apparent effect on hemorrhagic stroke. The possible benefits of statins on cerebrovascular disease occurrence seem to be mediated through several mechanisms, such as LDL-cholesterol reduction,16 anti-inflammatory effect37, 38 on biological markers of plaque instability,39 or the development of atherosclerosis.40, 41, 42
Our study design took into account the possibility of some methodological limitations. To avoid selection bias, we used all incident statin users. In addition, given that the patients studied were all receiving statins, the likelihood of confounding by indication is reduced. However, we could not control for all patient characteristics that may influence the physician's choice. Unmeasured comorbidities as well as unavailable clinical data for cholesterol levels could lead to residual confounding; yet there is no reason to believe that the prescription of different statins would be strongly influenced by cholesterol values. The analysis of available baseline characteristics did not suggest preferential prescribing of a particular statin to patients at higher vascular risk.
We also assessed the possibility that patients with more comorbidities may have had more motivation to be adherent to prescribed statin regimens. In addition, these patients may be more likely to have a higher cardiovascular risk. We adjusted for several determinants with proxy comorbidity status such as diabetes, hypertension, and occurrence of cardiovascular conditions after the initiation of statin agents. Our multivariate analysis should have minimized the influence of confounding factors. Nevertheless, residual confounding effects due to incomplete or inaccurate measurement of covariates or unmeasured confounders cannot be excluded. For instance, patients who are nonadherent may have other traits that contribute to worse outcomes, including factors such as depression, lower socioeconomic status, and associated adverse health behaviors,43 although we were able to adjust in part for these factors. The link between nonadherence and adverse vascular outcomes should be further investigated in future observational studies to better understand factors associated with medication nonadherence.
Our study had other limitations. First, administrative databases do not allow adjustment for clinical severity of diseases. We thus did not have access to—and so cannot adjust for—cholesterol values before and after treatment. To assess this possible bias, we evaluated the rate of changing to different doses of statins and found that most patients (84%) remained with the same dose, while the rate of switching to other statins was only 9%. Secondly, in order to reduce the likelihood of confounding by dose, we evaluated the doses of the different statins compared with an equivalency in lowering LDL-cholesterol levels and found that the doses were comparable among cases and controls. Thirdly, we also could not adjust for blood pressure, one of the most important modifiable risk factors for stroke.44 However, if individuals were using medication for hypertension or diabetes, we included these variables in the statistical model.
In addition, residual confounding by unmeasured factors is always possible. The RAMQ databases also do not contain any information about the patients' lifestyle habits. Therefore, it was not possible to adjust for smoking, obesity, lack of exercise, and poor diet, which are associated with stroke risk.1 These factors may introduce a bias in our results because they are well-documented risk factors for stroke and they are more likely to be present among patients who are not adherent to their medication.1, 45 Another limitation is that some patients included in our study may have had a previous cardiovascular condition, which we could not identify in our databases. However, the probability of this misclassification is minimal because we had access to pharmaceutical and medical data for 2 and 5 years, respectively, before cohort entry. In addition, because adherence was calculated using prescription claims, it is not possible to know whether or not patients took their medication. However, because patients pay a portion of the drug's cost, they may be more likely to take it. Consequently, the possibility of bias is reduced. Finally, while the sensitivity and specificity of the databases used were reasonable, the validity of the hospital administrative data is far from complete.
Although nonadherence to statins is prevalent, our study indicates that better adherence to these agents is associated with a risk reduction for cerebrovascular disease in a real-world setting. The assessment of medication adherence should be incorporated into routine clinical practice. Interventions in this area are essential so that the therapeutic benefits translate into clinical practice.
Acknowledgments
The Canadian Institutes of Health Research (CIHR) supported this work. Sylvie Perreault, Lyne Lalonde, Lucie Blais, and Anick Bérard are research scholars who receive financial support from the Fonds de recherche en santé du Québec.
References
- Primary prevention of ischemic stroke: a guideline from the American Heart Association/American Stroke Association Stroke Council: cosponsored by the Atherosclerotic Peripheral Vascular Disease Interdisciplinary Working Group; Cardiovascular Nursing Council; Clinical Cardiology Council; Nutrition, Physical Activity, and Metabolism Council; and the Quality of Care and Outcomes Research Interdisciplinary Working Group: the American Academy of Neurology affirms the value of this guideline. Stroke. 2006;37:1583–1633
- . The Growing Burden of Heart Disease and Stroke in Canada, 2003. Ottawa, Canada: Heart and Stroke Foundation of Canada; 2003;
- Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet. 1994;344:1383–1389
- . MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20,536 high-risk individuals: a randomised placebo-controlled trial. Lancet. 2002;360:7–22
- . Prevention of coronary and stroke events with atorvastatin in hypertensive patients who have average or lower-than-average cholesterol concentrations, in the Anglo-Scandinavian Cardiac Outcomes Trial—Lipid Lowering Arm (ASCOT-LLA): a multicentre randomised controlled trial. Lancet. 2003;361:1149–1158
- Primary prevention of cardiovascular disease with atorvastatin in type 2 diabetes in the Collaborative Atorvastatin Diabetes Study (CARDS): multicentre randomised placebo-controlled trial. Lancet. 2004;364:685–696
- Pravastatin in elderly individuals at risk of vascular disease (PROSPER): a randomised controlled trial. Lancet. 2002;360:1623–1630
- Prevention of coronary heart disease with pravastatin in men with hypercholesterolemia (West of Scotland Coronary Prevention Study Group). N Engl J Med. 1995;333:1301–1308
- 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;279:1615–1622
- Intensive lipid lowering with atorvastatin in patients with stable coronary disease. N Engl J Med. 2005;352:1425–1435
- . Effects of statins on stroke prevention in patients with and without coronary heart disease: a meta-analysis of randomized controlled trials. Am J Med. 2004;117:596–606
- . Primary prevention of cardiovascular diseases with statin therapy: a meta-analysis of randomized controlled trials. Arch Intern Med. 2006;166:2307–2313
- . Statin therapy in stroke prevention: a meta-analysis involving 121,000 patients. Am J Med. 2008;121:24–33
- . Efficacy and safety of cholesterol-lowering treatment: prospective meta-analysis of data from 90,056 participants in 14 randomized trials of statin. Lancet. 2005;366:1267–1278
- Effects of cholesterol-lowering with simvastatin on stroke and other major vascular events in 20,536 people with cerebrovascular disease or other high-risk conditions. Lancet. 2004;363:757–767
- . Statin in stroke prevention and carotic atherosclerosis (Systematic review and up-to-date meta-analysis). Lancet. 2004;356:2902–2909
- Long-term persistence in use of statin therapy in elderly patients. JAMA. 2002;288:455–461
- Persistence and determinants of statin therapy among middle-aged patients for primary and secondary prevention. Br J Clin Pharmacol. 2005;59:564–573
- International Classification of Diseases. Manual of the International Statistical Classification of Diseases, Injuries, and Cause of Death. 9th edn.. Geneva, Switzerland: World Health Organization; 1977;Publication no PHS 80-1260
- . Canadian Classification of Diagnostic, Therapeutic, and Surgical Procedures. 2nd edn.. Ottawa, Ontario, Canada: Statistics Canada Health Division; 1986;
- . Principales variables selon la catégorie de personnes assures [French]. Studies and Statistics 2005 http://www.ramq.gouv.qc.caAccessed September 20, 2006
- . The use of prescription claims databases in pharmacoepidemiological research: the accuracy and comprehensiveness of the prescription claims database in Quebec. J Clin Epidemiol. 1995;48:999–1009
- . Validation of diagnostic codes within medical services claims. J Clin Epidemiol. 2004;57:131–141
- . Validating administrative data in stroke research. Stroke. 2002;33:2465–2470
- . Nested case-control studies. Prev Med. 1994;23:587–590
- . The nested case-control study in cardiology. Am Heart J. 2003;146:581–590
- . Estimating medication persistency using administrative claims data. Am J Manag Care. 2005;11:449–457
- . Effect of statins risk of coronary disease: a meta-analysis of randomized controlled trials. JAMA. 1999;282:2340–2346
- . Approximate equivalent rosuvastatin doses for temporary statin interchange programs. Ann Pharmacother. 2004;38:1286–1292
- . Comparative dose efficacy study of atorvastatin versus simvastatin, pravastatin, lovastatin, and fluvastatin in patients with hypercholesterolemia. Am J Cardiol. 1998;81:582–587
- . A chronic disease score from automated pharmacy data. J Clin Epidemiol. 1992;45:197–203
- . Comparison of nested case-control and survival analysis methodologies for analysis of time-dependent exposure. BMC Med Res Method. 2005;5:5–11
- . Sensitivity analysis, Monte Carlo risk analysis, and Bayesian uncertainty assessment. Risk Anal. 2001;21:579–583
- . Monte Carlo sensitivity analysis and Bayesian analysis of smoking as an unmeasured confounder in a study of silica and lung cancer. Am J Epidemiol. 2004;160:384–392
- . Econometric Analysis. 3rd edn.. Upper Saddle River, NJ: Prentice-Hall; 1997;
- . Regression Diagnostics: Identifying Influential Data and Sources of Collinearity. New York: John Wiley & Sons; 1981;
- Inflammation, pravastatin, and the risk of coronary events after myocardial infarction in patients with average cholesterol levels (Cholesterol and Recurrent Events (CARE) Investigators). Circulation. 1998;98:839–844
- Measurement of C-reactive protein for the targeting of statin therapy in the primary prevention of acute coronary events. N Engl J Med. 2001;344:1959–1965
- Pravastatin treatment increases collagen content and decreases lipid content, inflammation, metalloproteinases, and cell death in human carotid plaques: implications for plaque stabilization. Circulation. 2001;103:926–933
- . Cholesterol reduction in cardiovascular disease: clinical benefits and possible mechanisms. N Engl J Med. 1995;332:512–521
- . Antiatherothrombotic properties of statins: implications for cardiovascular event reduction. JAMA. 1998;279:1643–1650
- . Pleiotrophic effects of 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors. Arterioscler Thromb Vasc Biol. 2001;21:1712–1719
- . Adherence to medication. N Engl J Med. 2005;353:487–497
- Risk factors. Stroke. 1997;28:1507–1517
- Determinants of non-compliance with lipid-lowering therapy in hyperlipidemic patients. Pharmacoepidemiol Drug Saf. 2002;11:593–600
Funding: Financial support was received from the Canadian Institutes of Health Research.
Conflict of Interest: None.
Authorship: All coauthors have read and approved the manuscript, which represents, to the best of our knowledge, honest and accurate work.
PII: S0002-9343(09)00286-1
doi:10.1016/j.amjmed.2009.01.032
© 2009 Elsevier Inc. All rights reserved.
