Arthritis Care & Research Vol. 64, No. 6, June 2012, pp 809 – 816 DOI 10.1002/acr.21643 © 2012, American College of Rheumatology
ORIGINAL ARTICLE
Impact of Statin Discontinuation on Mortality in Patients With Rheumatoid Arthritis: A PopulationBased Study MARY A. DE VERA,1 HYON CHOI,2 MICHAL ABRAHAMOWICZ,3 JACEK KOPEC,4 DIANE LACAILLE4
AND
Objective. To evaluate the impact of statin discontinuation on cardiovascular disease (CVD) mortality and all-cause mortality in a population-based cohort of patients with rheumatoid arthritis (RA). Methods. We conducted a population-based longitudinal study of RA patients with incident statin use followed from 1996 to 2006 using administrative health data. Statin discontinuation was defined as persistent nonuse for >3 months during the therapy course. Primary outcomes were mortality from all CVDs (CVD mortality) and secondary outcomes were deaths from all causes (all-cause mortality). Cox proportional hazards models with statin discontinuation as a time-dependent variable were used and multivariable models were adjusted for age, sex, comorbidities, and risk factors for mortality, and proxy indicators of RA severity. Results. Over 16,144 person-years of followup in the cohort of 4,102 incident statin users, we documented 198 deaths due to CVD and 467 deaths overall. Adjusted hazard ratios for the association of statin discontinuation with death were 1.60 (95% confidence interval [95% CI] 1.15–2.23) for CVD mortality and 1.79 (95% CI 1.46 –2.20) for all-cause mortality. The association between statin discontinuation and mortality outcomes was not modified by timing of the first statin prescription, age, and sex (P values for interaction >0.29). Conclusion. These population-based data indicate that statin discontinuation in patients with RA is associated with an increased risk of death from CVD and all causes. Findings provide support for the importance of compliance with therapy in RA patients who are prescribed statins.
INTRODUCTION Affecting approximately 1% of the population (1), rheumatoid arthritis (RA) is a chronic inflammatory arthritis associated with increased mortality, with death rates in RA 1.5-fold higher than in the general population (2). Cardiovascular disease (CVD) is the leading cause of mortality in RA (3), with a 50% increased risk of CVD mortality in patients with RA compared to individuals in the general population (4). Statins are recommended therapy for management of hypercholesterolemia and ischemic heart disease in RA patients (5,6). However, since statin therapy represents
Supported by an operating grant from the Canadian Institutes of Health Research (MOP-77605). Dr. De Vera received doctoral training support from the Canadian Institutes of Health Research, the Canadian Arthritis Network/ The Arthritis Society of Canada, and the Michael Smith Foundation for Health Research. Dr. Abrahamowicz is a James McGill Professor of Biostatistics at McGill University. Dr. Lacaille holds an Investigator Award from The Arthritis Society of Canada and is the Nancy and Peter Paul Saunders Scholar.
“chronic” or long-term treatment (7) in which therapeutic effects are not readily observed (8), discontinuation is of particular concern (8), as evidenced by a number of previous studies reporting statin discontinuation rates of ⱖ50% in the general population (7,9 –15). Using a populationbased RA cohort, we reported that over a mean followup period of 8 years, 38% of patients with RA discontinue statins permanently (16), and subsequently showed that 1 Mary A. De Vera, PhD: Arthritis Research Centre of Canada, Vancouver, British Columbia, Canada; 2Hyon Choi, MD, DrPH: Arthritis Research Centre of Canada, Vancouver, British Columbia, Canada, and Boston University School of Medicine, Boston, Massachusetts; 3Michal Abrahamowicz, PhD: McGill University, Montreal, Quebec, Canada; 4Jacek Kopec, MD, PhD, Diane Lacaille, MD, MHSc: Arthritis Research Centre of Canada and University of British Columbia, Vancouver, British Columbia, Canada. Address correspondence to Mary A. De Vera, PhD, University of British Columbia School of Population and Public Health, Arthritis Research Centre of Canada, 895 West 10th Avenue, Vancouver, British Columbia, Canada, V5Z 1L7. E-mail:
[email protected]. Submitted for publication July 5, 2011; accepted in revised form February 8, 2012.
809
810
De Vera et al
Significance & Innovations ●
Discontinuation of statin therapy is common and is associated with an increased risk of mortality in the general population, but no corresponding data are available among individuals with rheumatoid arthritis, an inflammatory arthritis associated with increased mortality.
●
By applying time-dependent approaches that accounted for both intermittent and permanent statin discontinuation, we demonstrated the impact of discontinuation of therapy on cardiovascular disease mortality and all-cause mortality in a realworld patient setting.
statin discontinuation any time during the course of therapy was associated with a 67% increased risk for acute myocardial infarction (MI) (17,18), corroborating previous reports of the negative CVD impact of statin discontinuation (19). Statin discontinuation has also been shown to have a negative impact on mortality (20 –22), and therefore our objective was to evaluate the impact of statin discontinuation on the risk of mortality among individuals with RA.
MATERIALS AND METHODS Data source and study sample. We used data from a previously established population-based RA cohort (n ⫽ 37,151) in British Columbia, Canada (23). Administrative data for the reimbursement of physician visits from the British Columbia Ministry of Health were used to identify adult (age ⱖ18 years) individuals with RA who received care for their RA between January 1996 and March 2006. The definition for RA was the same as previously reported for this cohort (17,23), with individuals included if they had at least 2 physician visits at least 2 months apart with an RA diagnostic code (International Classification of Diseases, Ninth Revision [ICD-9] code 714). To improve the specificity of the definition, the following exclusion criteria were applied: 1) at least 2 visits subsequent to the second RA visit with diagnoses of other inflammatory arthritides (e.g., systemic lupus erythematosus, ankylosing spondylitis), 2) an RA-coded visit by a nonrheumatologist that was not confirmed on a subsequent rheumatologist visit, or 3) no subsequent RA-coded visits during at least 5 years of followup. Using the number of alive prevalent cases in 2006 (n ⫽ 29,417) and census data from Statistics Canada (24), we estimated a prevalence rate of 0.82%, consistent with previous RA prevalence reports (1). From this population-based cohort of RA patients, we identified a cohort of incident statin users, defined as those who had ⱖ1 statin prescriptions between May 1996 and December 2003 and no prior statin prescriptions since January 1996 (earliest date of available medication data). Individuals with a statin prescription in the first 4 months of prescription records were excluded to ensure incident
statin use (17). For included individuals, we obtained information on all provincially funded health services, including physician visits and hospitalizations since 1990, as well as all prescriptions dispensed since 1996, regardless of the funding source. We also obtained information on mortality, including date and primary cause of death, from the Canadian Mortality Database (25). No personal identifying information was available on any individual, and data access procedures complied with the British Columbia Freedom of Information and Protection of Privacy Act. The University of British Columbia Behavioral Research Ethics Board granted ethical approval for this study. Exposure assessment. We established the statin therapy course for each subject in terms of binary (yes/no) indicators of statin use in each month over the entire followup period. Our primary exposure was statin discontinuation, defined as ⱖ3 consecutive months of non–statin use at any time during the followup period, which takes into account a potential delayed effect of statin discontinuation on mortality risk and also avoids classifying periods of low adherence or transient interruptions as therapy discontinuation (17,18). Months with resumed statin prescriptions after a period of discontinuation were classified as “statin use,” allowing capture of both intermittent and permanent discontinuation. Finally, switches between different types of statins were not considered as therapy discontinuation. Outcome assessment. Mortality outcomes were ascertained using vital statistics data in the Canadian Mortality Database, which has a 97.6% ascertainment rate for deaths in Canada (25). The primary study outcome was deaths due to all CVDs (CVD mortality), ascertained using International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes (I00 –I99) for ischemic heart disease and other heart diseases, cerebrovascular disease, hypertension, diseases of the arteries and veins, and other diseases of the circulatory system (26). The secondary outcome was deaths due to all causes (allcause mortality). Assessment of covariates. Potential covariates considered in multivariable regression models included factors known to influence statin discontinuation in our cohort (27) and risk factors for CVD and RA mortality. Fixed-intime binary variables measured over a period of 1 year preceding the start of followup evaluated the presence of chronic comorbid conditions and were based on diagnostic codes for physician visits (ICD-9 codes) or medication use. These included: 1) diabetes mellitus (use of insulin or oral hypoglycemic agents) (18); 2) angina (ICD-9 codes 411 and 413 or use of nitrates) (18); 3) chronic obstructive pulmonary disease (ICD-9 codes 490 – 496 and 505–506) (28); 4) gastrointestinal diseases, including gastric/gastrojejunal ulcer, duodenal ulcer, peptic ulcer, and gastrointestinal hemorrhage (ICD-9 codes 531–534 and 578) (29); 5) renal disease (ICD-9 codes 403.11, 403.91, 404.12, and 585–586) (28); 6) antihypertension medications; 7) congestive heart failure medications; 8) antiarrhythmia medica-
Statin Discontinuation and Mortality in RA
811
Table 1. Characteristics of the RA incident statin users cohort (n ⴝ 4,102)* Value Demographics Age, mean ⫾ SD years Women RA characteristics RA duration at index date, mean ⫾ SD months Rate of RA-related medical visits/person-year followup, mean ⫾ SD RA-related orthopedic procedure (over followup) Use of RA prescription medications (over study followup) Traditional nonsteroidal antiinflammatory drugs Glucocorticoids COX-2 inhibitors Methotrexate DMARD categories† Group 1 Group 2 Group 3 Group 4 Group 5 Comorbid medical conditions and medication use Angina‡ Prior malignancy§ Diabetes mellitus‡ Chronic obstructive pulmonary disease‡ Prior acute myocardial infarction§ Prior infection requiring hospitalization§ Prior cerebrovascular accident§ Gastrointestinal disease‡ Renal disease‡ Use of antihypertension medication‡ Use of congestive heart failure medication‡ Use of anticoagulant‡ Use of antiarrhythmia medication‡ Charlson comorbidity score, mean ⫾ SD‡
66.6 ⫾ 10.4 2,460 (60) 19.0 ⫾ 25.5 2.8 ⫾ 4.3 636 (15.5) 2,364 (57.6) 1,723 (42.0) 1,347 (32.8) 959 (23.4) 2,481 (60.5) 492 (12.0) 825 (20.1) 172 (4.2) 132 (3.2) 903 (22.0) 884 (21.6) 730 (17.8) 508 (12.4) 441 (10.8) 347 (8.5) 117 (2.9) 148 (3.6) 83 (2.0) 2,528 (61.6) 1,344 (32.8) 266 (6.5) 94 (2.3) 1.0 ⫾ 1.3
* Values are the number (percentage) of cases unless otherwise indicated. RA ⫽ rheumatoid arthritis; COX-2 ⫽ cyclooxygenase 2; DMARD ⫽ disease-modifying antirheumatic drug. † DMARD categories: 1 ⫽ no DMARD use; 2 ⫽ antimalarial drugs, sulfasalazine; 3 ⫽ methotrexate, intramuscular gold; 4 ⫽ leflunomide, cyclosporin A, azathioprine, cyclophosphamide, chlorambucil, or mycophenolate mofetil; and 5 ⫽ biologics. ‡ Evaluated over 1 year preceding the index date. § Evaluated prior to the index date since 1990 (earliest available data).
tions; and 9) anticoagulants (preparations of heparin and warfarin). We did not include aspirin because it does not require a prescription and therefore is not accurately captured in our administrative databases. Fixed-in-time binary variables measured from the beginning of data records to the index date evaluated occurrences of the following disease events: acute MI (ICD-9 code 410), cerebrovascular accident (CVA; ICD-9 codes 434 and 436), malignancy (30), and infection requiring hospitalization (31). We also calculated a modified Charlson comorbidity score over the 1-year period preceding the start of study followup using a version adapted for administrative data (32,33). The following variables, evaluated over the study followup period, were considered as proxy indicators of RA severity: disease-modifying antirheumatic drug (DMARD) use, rate of RA-related physician visits, and having received any orthopedic procedure for RA. DMARD use over the followup period was categorized to reflect increasing RA severity: no DMARD use (group 1); antimalarial drugs
or sulfasalazine (group 2); methotrexate or intramuscular gold (group 3); leflunomide, cyclosporin A, azathioprine, cyclophosphamide, chlorambucil, or mycophenolate mofetil (group 4); and biologics (group 5). These categories were mutually exclusive and we used the highest rank ever attained during the followup period. Finally, we determined use of RA medications that could influence mortality risk, including glucocorticoids, traditional nonsteroidal antiinflammatory drugs (NSAIDs), cyclooxygenase 2 selective NSAIDs, and methotrexate, as monthlyupdated time-varying covariates. Statistical analysis. In order to assess the impact of statin discontinuation on mortality in RA, individuals were considered at risk for study outcomes only after they had met both criteria of being diagnosed with RA and initiating statin therapy (18); therefore, we set the index date for individuals whose first statin prescription was dispensed on or after RA diagnosis as the date of the first
812 statin prescription, and for individuals whose first statin was dispensed before RA diagnosis as the date of RA diagnosis. We computed person-time of followup from the index date until date of death, last health care service, or end of the study period (March 31, 2006), whichever came first. We used Cox proportional hazards models with delayed entry (34,35), setting time since statin initiation as the time axis with the origin corresponding to the date of the first statin prescription, and therefore allowing for comparison of subjects’ mortality risk at the same time. This permitted delayed entry into the risk set for subjects who initiated statin therapy before RA diagnosis. Since these subjects met our inclusion criteria only after they were diagnosed with RA, they were not considered at risk for the study outcome until their RA diagnosis; had they entered the study at the time of their first statin prescription, the results could be affected by survival bias (36). We modeled statin discontinuation, the exposure of primary interest, as a binary time-dependent variable, updated monthly. Accordingly, the hazard ratio (HR) represented the independent risk of mortality associated with statin discontinuation in the current month. All multivariable Cox models were adjusted for age at the index date and sex. Fixed-in-time variables representing RA duration at the index date (months), statin duration at the index date (months), and a binary indicator of whether subjects initiated statin therapy before RA diagnosis (yes/no) were forced into final multivariable models. Established risk factors for mortality in RA (e.g., prior acute MI, malignancy) were forced into final multivariable models, and for putative risk factors, a forward selection procedure with a P value less than or equal to 0.20 criterion for entry was employed. We repeated the analyses to model statin discontinuation as a continuous time-dependent variable that increased with every month since discontinuation to account for the possibility that a longer time since statin discontinuation may be associated with an increased mortality risk. We also tested potential interactions of statin discontinuation with the timing of the first statin prescription (before/after RA diagnosis), sex (women/men), age (⬍65/ⱖ65 years), prior acute MI status, and prior CVD status with likelihood ratio tests. Finally, we conducted sensitivity analyses of our definition of statin discontinuation by evaluating the impact of varying the duration of persistent non–statin use (e.g., 1 month, 2 months) required prior to declaring an individual as having discontinued statin therapy. All hypotheses were tested using 2-tailed Wald’s tests at the 0.05 significance level, and analyses were performed using SAS, version 9.1.
RESULTS The cohort of RA patients with incident statin use included 4,102 individuals, altogether contributing 16,144 person-years of followup between May 1996 and March 2006. Table 1 shows the characteristics of this cohort, including demographic characteristics, RA characteristics, comorbid conditions, and medication use. Women comprised 60% of the cohort and at the index date, the mean ⫾
De Vera et al SD age was 66.6 ⫾ 10.4 years and the mean ⫾ SD RA duration was 19.0 ⫾ 25.5 months. Overall, 1,862 individuals (45.4%) in the cohort met the study definition for statin discontinuation at least once during the study followup. Over a mean followup period of 4 years (median 3.6 years), we documented 198 CVD deaths and 467 deaths from all causes. Based on the latter, we calculated a standardized mortality ratio (SMR) of 2.24 for the cohort. Of the 198 CVD deaths, the most common specific causes were acute MI (62 [31.3%]), chronic ischemic heart disease (61 [30.8%]), and CVA (30 [15.2%]). In univariate Cox regression analysis, statin discontinuation was associated with a 41% increased risk of CVD mortality (unadjusted HR 1.41, 95% confidence interval [95% CI] 1.02–1.96) (Table 2). After adjusting for covariates included in the final multivariable Cox model, as shown in Table 2, statin discontinuation was associated with a 60% increased risk in CVD mortality (adjusted HR 1.60, 95% CI 1.15–2.23). Other significant predictors of increased CVD mortality risk included older age, male sex, prior acute MI, use of antihypertension or congestive heart failure medications, as well as current glucocorticoid use, which was associated with a 50% increased risk of CVD mortality (HR 1.51, 95% CI 1.09 –2.09). When we modeled time since statin discontinuation as a continuous timedependent variable, the value of which increased with each additional month since discontinuation, we also found significant associations between statin discontinuation and CVD mortality (adjusted HR 1.004, 95% CI 1.001– 1.016), indicating a 0.4% increase in risk of CVD mortality per month since discontinuation. Of note, when looking at specific causes of CVD mortality, the number of events was small and did not allow meaningful analyses for specific outcomes separately. Unadjusted and adjusted HRs for all-cause mortality are shown in Table 3. Statin discontinuation was associated with an increased risk of all-cause mortality (adjusted HR 1.79, 95% CI 1.46 –2.20). Other significant predictors of increased mortality risk included older age, male sex, higher Charlson comorbidity score, infection requiring hospitalization, and use of antihypertension and congestive heart failure medications. In addition, current use of glucocorticoids was associated with a 2-fold increased allcause mortality risk (adjusted HR 2.28, 95% CI 1.87–2.79), whereas current methotrexate use was associated with an ⬃50% reduction in risk (adjusted HR 0.54, 95% CI 0.39 – 0.74). HRs for acute MI, diabetes mellitus, and cancer did not retain their statistical significance when the Charlson comorbidity score was entered into the model, likely because these comorbidities are components of the score. However, given that they are established risk factors for death in RA, these variables were forced in the final multivariable model. When we modeled time since statin discontinuation as a continuous time-dependent variable, we also found significant associations for statin discontinuation and all-cause mortality (adjusted HR 1.010, 95% CI 1.002–1.015), indicating a 1% increase in risk of death for every month since statin discontinuation. The results of subgroup analyses indicate that the impact of statin discontinuation did not depend on: 1) timing of first statin use (before versus after RA diagnosis), 2) sex, 3) age, and 4) prior acute MI status for either all-cause
Statin Discontinuation and Mortality in RA
813
Table 2. Statin discontinuation and risk of cardiovascular disease mortality*
Primary exposure (time dependent) Statin discontinuation (yes vs. no) Fixed-in-time covariates Age Sex (men vs. women) Prior acute MI (yes vs. no) Prior infection requiring hospitalization (yes vs. no) Diabetes mellitus (yes vs. no) Use of antihypertension medication (yes vs. no) Use of congestive heart failure medication (yes vs. no) Use of anticoagulant (yes vs. no) Charlson comorbidity score Time-dependent covariates Current use of glucocorticoids (yes vs. no) Current use of methotrexate (yes vs. no)
Unadjusted HR (95% CI)
P†
Adjusted HR (95% CI)
P†
1.41 (1.02–1.96)
0.04
1.60 (1.15–2.23)
0.005
1.08 (1.07–1.10) 1.49 (1.12–1.97) 2.94 (2.12–4.06) 2.37 (1.62–3.46) 1.54 (1.10–2.15) 3.36 (2.32–4.86) 3.04 (2.29–4.03) 2.88 (1.94–4.27) 1.37 (1.27–1.47)
⬍ 0.0001 0.006 ⬍ 0.0001 ⬍ 0.0001 0.01 ⬍ 0.0001 ⬍ 0.0001 ⬍ 0.0001 ⬍ 0.0001
1.07 (1.05–1.09) 1.53 (1.15–2.04) 1.47 (1.03–2.09) 1.49 (1.01–2.23) 1.25 (0.88–1.77) 1.87 (1.26–2.78) 1.79 (1.31–2.45) 1.29 (0.85–1.96) 1.15 (1.09–1.26)
⬍ 0.0001 0.004 0.03 0.05 0.2 0.002 0.0002 0.2 0.004
1.98 (1.45–2.72) 0.79 (0.50–0.90)
⬍ 0.0001 0.009
1.51 (1.09–2.09) 0.68 (0.43–1.08)
0.01 0.1
* HR ⫽ hazard ratio; 95% CI ⫽ 95% confidence interval; MI ⫽ myocardial infarction. † P value for 2-tailed Wald’s test.
mortality (all P values for interactions ⱖ0.29) or CVD mortality outcomes (all P values for interactions ⱖ0.61) (Table 4). However, the impact of statin discontinuation varied according to prior CVD for CVD mortality outcomes (P value for interaction ⬍0.01), but not for all-cause mortality outcomes (P value for interaction ⫽ 0.06). Finally, in sensitivity analyses, we found that the association between statin discontinuation and all-cause mortality and CVD mortality remained statistically significant, regardless of whether statin discontinuation was defined as requiring 1 or 2 months of persistent nonuse, rather than 3 months as in the main analyses. Specifically, for CVD mortality and all-cause mortality, adjusted HRs were 1.47 (95% CI 1.22–2.29) and 1.69 (95% CI 1.65–2.43), respec-
tively, when using a definition based on 1 month of nonuse, and 1.57 (95% CI 1.13–2.18) and 1.72 (95% CI 1.58 – 2.35), respectively, when using a definition based on 2 months of persistent nonuse.
DISCUSSION We conducted a population-based study of an RA cohort of incident statin users followed for up to 10 years to determine the impact of statin discontinuation on risk of CVD mortality and all-cause mortality. We found that statin discontinuation was associated with a 60% increased risk of cardiovascular mortality and a 79% increased risk of
Table 3. Statin discontinuation and risk of all-cause mortality*
Primary exposure (time dependent) Statin discontinuation (yes vs. no) Fixed-in-time covariates Age Sex (men vs. women) Prior acute MI (yes vs. no) Prior malignancy (yes vs. no) Prior infection requiring hospitalization (yes vs. no) Diabetes mellitus (yes vs. no) Use of antihypertension medication (yes vs. no) Use of congestive heart failure medication (yes vs. no) Charlson comorbidity score Time-dependent covariates Current use of glucocorticoids (yes vs. no) Current use of methotrexate (yes vs. no)
Unadjusted HR (95% CI)
P†
Adjusted HR (95% CI)
P†
1.70 (1.39–2.08)
⬍ 0.0001
1.79 (1.46–2.20)
⬍ 0.0001
1.08 (1.06–1.09) 1.30 (1.08–1.56) 2.21 (1.76–2.79) 1.66 (1.36–2.04) 2.08 (1.60–2.69) 1.43 (1.15–1.78) 2.60 (2.08–3.24) 2.51 (2.09–3.01) 1.36 (1.29–1.43)
⬍ 0.0001 0.0048 ⬍ 0.0001 ⬍ 0.0001 ⬍ 0.0001 0.002 ⬍ 0.0001 ⬍ 0.0001 ⬍ 0.0001
1.07 (1.05–1.08) 1.40 (1.17–1.69) 1.16 (0.91–1.49) 1.07 (0.86–1.34) 1.36 (1.03–1.78) 1.23 (0.97–1.54) 1.64 (1.29–2.08) 1.59 (1.31–1.94) 1.18 (1.10–1.25)
⬍ 0.0001 0.0004 0.2 0.5 0.02 0.08 ⬍ 0.0001 ⬍ 0.0001 ⬍ 0.0001
2.81 (2.32–3.40) 0.66 (0.49–0.90)
⬍ 0.0001 0.009
2.28 (1.87–2.79) 0.54 (0.39–0.74)
⬍ 0.0001 0.0001
* HR ⫽ hazard ratio; 95% CI ⫽ 95% confidence interval; MI ⫽ myocardial infarction. † P value for 2-tailed Wald’s test.
814
De Vera et al
Table 4. Multivariable HRs for statin discontinuation and risk of all-cause and CVD mortality stratified by initiation of first statin before RA, sex, age group, prior acute MI, or prior CVD* Subgroup
No. of deaths
Adjusted HR (95% CI)
58 140 101 97 37 161 49 149 141 57
1.69 (1.02–3.11) 1.53 (1.01–2.28) 1.43 (1.04–2.26) 1.74 (1.06–2.85) 1.29 (0.59–2.82) 1.63 (1.12–2.35) 1.24 (0.55–2.79) 1.66 (1.15–2.39) 1.50 (1.01–2.25) 1.31 (0.73–2.35)
127 340 252 215 94 373 91 376 297 170
1.67 (1.10–2.53) 1.84 (1.45–2.34) 1.67 (1.27–2.20) 2.02 (1.48–2.77) 1.70 (1.09–2.67) 1.84 (1.46–2.32) 1.40 (0.79–2.45) 1.91 (1.53–2.39) 1.49 (1.14–1.95) 2.11 (1.47–2.77)
CVD mortality First statin before RA diagnosis First statin after RA diagnosis Women Men Age ⬍65 years Age ⱖ65 years Prior acute MI No prior acute MI Prior CVD‡ No prior CVD All-cause mortality First statin before RA diagnosis First statin after RA diagnosis Women Men Age ⬍65 years Age ⱖ65 years Prior acute MI No prior acute MI Prior CVD‡ No prior CVD
P for interaction†
0.65 0.61 0.82 0.64 ⬍ 0.01
0.93 0.29 0.29 0.31 0.06
* HR ⫽ hazard ratio; CVD ⫽ cardiovascular disease; RA ⫽ rheumatoid arthritis; MI ⫽ myocardial infarction; 95% CI ⫽ 95% confidence interval. † P value for the 2-tailed Wald’s test of the interaction between a given stratification variable (shown in the left-hand column) and statin discontinuation. ‡ Prior CVD defined using International Classification of Diseases, Ninth Revision codes 390 – 459 (26).
all-cause mortality. These associations were independent of age, sex, comorbid medical conditions, and prescription medication use. Findings also persisted across different subgroups evaluated based on sex, age, timing of first statin use, and prior acute MI status. Of note, we found a modifying effect of prior CVD status on the impact of statin discontinuation on CVD mortality outcomes. The mortality risk associated with RA has been evaluated across different settings, and studies have consistently shown that RA is associated with increased mortality, with death rates in RA 1.5- to 1.6-fold higher than in the general population (2). CVD is the main cause of excess of mortality in RA; a recent meta-analysis showed a 50% increased risk of CVD death in RA patients (pooled SMR 1.50, 95% CI 1.39 –1.69) (4). Given its burden on RA mortality, care for CVD is important in the management of RA, and clinical practice guidelines recommend regular assessment of cardiovascular risk factors, including lipid tests and treatment with statins, for hyperlipidemic or high-risk patients (5,6). To our knowledge, this is the first study to evaluate the association of statin discontinuation with mortality outcomes specifically in RA, a patient population in whom it is particularly relevant given their increased risk of CVD death (4). By demonstrating the impact of statin discontinuation for specific CVD mortality and all-cause mortality outcomes, our study suggests the
importance of ensuring compliance with statin treatment among RA patients prescribed statins. Indeed, the problem of statin discontinuation has become well recognized over the last decade, with earlier studies demonstrating that in real-life settings, discontinuation of statins is more frequent than in randomized clinical trials (7,37) and subsequent studies reporting high statin discontinuation rates (7,9 –15). Equally important to quantifying the magnitude of the problem is evaluating associated adverse outcomes. However, the literature on the impact of statin discontinuation on health outcomes is still limited, and there is a need for studies across different patient populations. Nonetheless, our results corroborate previously reported findings. Specifically, statin discontinuation was associated with an increased risk of 1-year all-cause mortality in post–acute MI patients using data from the UK General Research Practice Database (adjusted relative risk [RR] 1.88, 95% CI 1.13–3.07) (20) and a US managed care organization (HR 2.86, 95% CI 1.47–5.55) (22). Potential biologic mechanisms that may explain the association between statin discontinuation and mortality outcomes observed in this study include the loss of cholesterol-lowering effects of statins as well as loss of antiinflammatory and immunomodulatory effects, which have been postulated to mediate beneficial effects of statins on
Statin Discontinuation and Mortality in RA CVD in inflammatory diseases such as RA. However, a potential behavioral factor may be the “healthy adherer” effect, whereby individuals who tend to closely follow prescribed medication regimens are also those who tend to exhibit healthier behaviors such as better diet, more exercise, and less smoking (38). However, a recent study of the association between adherence to statins, aspirin, or both, and CVD outcomes showed that in subjects taking both drugs, adherence to statins but not aspirin was associated with a lower risk of CVD recurrence (RR 0.64, 95% CI 0.49 – 0.82), but the same was not observed with adherence to aspirin but not statins (RR 0.91, 95% CI 0.72–1.15). These findings suggest that the healthy adherer effect may not fully explain adverse outcomes in poorly adherent patients (39). Although evaluating the impact of RA medications on mortality was not the primary intent of this study, it is nonetheless noteworthy that current use of glucocorticoids was associated with an increased risk of mortality and current use of methotrexate was associated with a decreased mortality risk. These findings confirm previous reports of the beneficial effects of methotrexate on RA mortality (40) and add to the literature cautioning clinicians of the risks associated with glucocorticoid use (41,42). Strengths and limitations of our study deserve comment. Use of a population-based cohort of patients with RA diminishes selection bias. Universal data on all prescriptions, health services, and vital statistics for the cohort ensured complete capture of statin discontinuation and death outcomes over a long period of observation (up to 10 years of followup). Treatment characteristics of the RA population studied, particularly low baseline DMARD (40.1%) and biologics (3.1%) use, may limit the generalizability of findings to other RA populations. However, inappropriately low DMARD use was the topic of the original study of this cohort (23), and these findings have been replicated by other researchers in Canada in the provinces of Ontario (43) and Quebec (44). The low rate of biologics used may be explained by the time period evaluated (1996 –2006), and the rate of use has increased over time, now reaching over 10%. Nonetheless, the RA population studied is representative of the RA population in British Columbia and in most provinces in Canada. Our study is also subject to limitations inherent to administrative data and observational studies. For example, because we used an algorithm based on diagnostic codes for physician visits to define RA, there is potential for misclassification of diagnosis. However, inclusion of non-RA cases would likely bias results toward the null, given the increased risk of mortality in RA. Furthermore, we used the strictest published case definition for RA (23), which was validated against self-report of a diagnosis of RA, yielding a positive predictive value of 0.92 (45,46), and improved specificity with additional exclusions. The use of administrative pharmacy records is well established in pharmacoepidemiologic studies (47), and our data have the advantage of including all medications dispensed to the entire RA population in British Columbia. However, we did not have information on whether pills were actually taken or reasons for statin discontinuation. For example,
815 patients may discontinue due to unfavorable side effects, which may have an impact on patients’ willingness to continue with therapy (48). Despite this, it has been reported that evaluation of prescription fill patterns represents the most accurate way of estimating actual medication use in large populations (49). Although we adjusted for all known risk factors for discontinuation (exposure) and mortality (outcome) available in our data, unmeasured or unknown confounders could still affect the results. Finally, we did not have access to data that would allow comparison of the effect of statin discontinuation with the general population. However, a number of previous reports have described the problem of statin discontinuation in the general population and its impact on mortality outcomes (20 –22). In conclusion, our population-based study of RA patients with incident statin use demonstrated an increased risk of cardiovascular and all-cause mortality associated with statin discontinuation. Given the increased mortality risk in RA and contribution of CVD to this risk, these findings provide support for raising awareness among health professionals and people with RA of the importance of therapy compliance among patients prescribed statins. AUTHOR CONTRIBUTIONS All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. De Vera had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design. De Vera, Choi, Kopec, Lacaille. Acquisition of data. Choi, Lacaille. Analysis and interpretation of data. De Vera, Choi, Abrahamowicz, Lacaille.
REFERENCES 1. Gabriel S, Michaud K. Epidemiological studies in incidence, prevalence, mortality, and comorbidity of the rheumatic diseases. Arthritis Res Ther 2009;19:229 – 44. 2. Sokka T, Abelson B, Pincus T. Mortality in rheumatoid arthritis: 2008 update. Clin Exp Rheumatol 2008;26 Suppl: S35– 61. 3. Van Doornum S, McColl G, Wicks IP. Accelerated atherosclerosis: an extraarticular feature of rheumatoid arthritis? [review]. Arthritis Rheum 2002;46:862–73. 4. Avina-Zubieta JA, Choi HK, Sadatsafavi M, Etminan M, Esdaile JM, Lacaille D. Risk of cardiovascular mortality in patients with rheumatoid arthritis: a meta-analysis of observational studies. Arthritis Rheum 2008;59:1690 –7. 5. Peters MJ, Symmons DP, McCarey D, Dijkmans BA, Nicola P, Kvien TK, et al. EULAR evidence-based recommendations for cardiovascular risk management in patients with rheumatoid arthritis and other forms of inflammatory arthritis. Ann Rheum Dis 2010;69:325–31. 6. Pham T, Gossec L, Constantin A, Pavy S, Bruckert E, Cantagrel A, et al. Cardiovascular risk and rheumatoid arthritis: clinical practice guidelines based on published evidence and expert opinion. Joint Bone Spine 2006;73:379 – 87. 7. Andrade SE, Walker AM, Gottlieb LK, Hollenberg NK, Testa MA, Saperia GM, et al. Discontinuation of antihyperlipidemic drugs: do rates reported in clinical trials reflect rates in primary care settings? N Engl J Med 1995;332:1125–31. 8. Tsoneva J, Shaw J. Understanding patients’ beliefs and goals in medicine-taking. Prof Nurse 2004;19:466 – 8. 9. Chodick G, Shalev V, Gerber Y, Heymann AD, Silber H, Simah
816
10.
11.
12.
13. 14. 15. 16. 17.
18.
19.
20.
21. 22.
23. 24. 25. 26.
27.
28. 29.
De Vera et al V, et al. Long-term persistence with statin treatment in a not-for-profit health maintenance organization: a populationbased retrospective cohort study in Israel. Clin Ther 2008;30: 2167–79. Ellis JJ, Erickson S, Stevenson J, Bernstein S, Stiles R, Fendrick M. Suboptimal statin adherence and discontinuation in primary and secondary prevention populations. J Gen Intern Med 2004;19:638 – 45. Deambrosis P, Saramin C, Terrazzani G, Scaldaferri L, Debetto P, Giusti P, et al. Evaluation of the prescription and utilization patterns of statins in an Italian local health unit during the period 1994-2003. Eur J Clin Pharmacol 2007;63:197–203. Foody JM, Joyce AT, Rudolph AE, Liu LZ, Benner JS. Persistence of atorvastatin and simvastatin among patients with and without prior cardiovascular diseases: a US managed care study. Curr Med Res Opin 2008;24:1987–2000. Howell N, Trotter R, Mottram D, Rowe P. Compliance with statins in primary care. Pharm J 2004;272:23– 6. Simons LA, Levis G, Simons J. Apparent discontinuation rates in patients prescribed lipid-lowering drugs. Med J Aust 1996;164:208 –11. Vinker S, Shani M, Baevsky T, Elhayany A. Adherence with statins over 8 years in a usual care setting. Am J Manag Care 2008;14:388 –92. Goycochea-Robles M, Rahman MM, Choi H, Esdaile J, Lacaille D. Compliance with statin therapy in rheumatoid arthritis patients [abstract]. Arthritis Rheum 2008;58 Suppl:S877– 8. De Vera M, Choi H, Abrahamowicz M, Rahman MM, Goycochea M, Lacaille D. Non-compliance with statin therapy and risk of acute myocardial infarction in patients with rheumatoid arthritis. Vancouver: Canadian Arthritis Network Annual Scientific Conference; 2009. De Vera M, Choi H, Abrahamowicz M, Kopec J, Goycochea M, Lacaille D. Statin discontinuation and risk of acute myocardial infarction in patients with rheumatoid arthritis: a population-based cohort study. Ann Rheum Dis 2011;70:1020 – 4. Penning-van Beest FJ, Termorshuizen F, Goettsch WG, Klungel OH, Kastelein JJ, Herings RM. Adherence to evidencebased statin guidelines reduces the risk of hospitalizations for acute myocardial infarction by 40%: a cohort study. Eur Heart J 2007;28:154 –9. Daskalopoulou S, Delaney J, Filion K, Brophy J, Mayo N, Suissa S. Discontinuation of statin therapy following an acute myocardial infarction: a population-based study. Eur Heart J 2008;29:2083–91. Colivicchi F, Bassi A, Santini M, Caltagirone C. Discontinuation of statin therapy and clinical outcome after ischemic stroke. Stroke 2007;38:2652–7. Ho PM, Spertus JA, Masoudi FA, Reid KJ, Peterson ED, Magid DJ, et al. Impact of medication therapy discontinuation on mortality after myocardial infarction. Arch Intern Med 2006; 166:1842–7. Lacaille D, Anis AH, Guh DP, Esdaile JM. Gaps in care for rheumatoid arthritis: a population study. Arthritis Rheum 2005;53:241– 8. Province of British Columbia. The 2006 Census of Canada in BC stats. 2006. URL: http://www.bcstats.gov.bc.ca/census. asp. Schnatter R, Acquavella J, Thompson F. An analysis of death ascertainment and follow-up through Canada’s mortality database system. Can J Public Health 1990;81:60 –5. Tu JV, Nardi L, Fang J, Liu J, Khalid L, Johansen H, et al. National trends in rates of death and hospital admissions related to acute myocardial infarction, heart failure and stroke, 1994-2004. CMAJ 2009;180:E118 –25. Goycochea-Robles M, Rahman MM, Choi H, Esdaile JM, Lacaille D. Compliance with statin therapy and predictors of compliance in rheumatoid arthritis (RA) [abstract]. J Rheumatol 2009;36:2575. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care 1998;36: 8 –27. Andrade SE, Gurwitz JH, Chan KA, Donahue JG, Beck A,
30.
31.
32.
33. 34. 35. 36.
37. 38.
39.
40. 41. 42. 43.
44.
45. 46.
47. 48. 49.
Boles M, et al. Validation of diagnoses of peptic ulcers and bleeding from administrative databases: a multi-health maintenance organization study. J Clin Epidemiol 2002;55:310 –3. Smitten A, Simon T, Qi K, Suissa S, Askling J, Franklin J, et al. Malignances in the Abatacept RA Clinical Development Program: an updated epidemiological assessment with ⬎ 10,000 person-years of exposure [abstract]. Arthritis Rheum 2008;58 Suppl:S787. Lacaille D, Guh DP, Abrahamowicz M, Anis AH, Esdaile JM. Use of nonbiologic disease-modifying antirheumatic drugs and risk of infection in patients with rheumatoid arthritis. Arthritis Rheum 2008;59:1074 – 81. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40: 373– 83. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613–9. Allison P. Survival analysis using the SAS system: a practical guide. Cary (NC): SAS Institute; 1995. Abrahamowicz M, Tamblyn R. Drug utilization patterns: encyclopedia of biostatistics. Hoboken (NJ): John Wiley & Sons; 2005. Zhou Z, Rahme E, Abrahamowicz M, Pilote L. Survival bias associated with time-to-treatment initiation in drug effectiveness evaluation: a comparison of methods. Am J Epidemiol 2005;162:1016 –23. Insull W. The problem of compliance to cholesterol altering therapy. J Intern Med 1997;241:317–25. The Coronary Drug Project Research Group. Influence of adherence to treatment and response of cholesterol on mortality in the Coronary Drug Project. N Engl J Med 1980;303:1038 – 41. Wei L, Fahey T, MacDonald TM. Adherence to statin or aspirin or both in patients with established cardiovascular disease: exploring healthy behaviour vs. drug effects and 10year follow-up of outcome. Br J Clin Pharmacol 2008;66: 110 – 6. Choi H, Herman M, Seeger J, Robins J, Wolfe F. Methotrexate and mortality in patients with rheumatoid arthritis: a prospective study. Lancet 2002;359:1173–7. Caplan L, Russell AS, Wolfe F. Steroids for rheumatoid arthritis: the honeymoon revisited (once again). J Rheumatol 2005;32:1863–5. Ramos-Remus C, Sibley J, Russell AS. Steroids in rheumatoid arthritis: the honeymoon revisited. J Rheumatol 1992;19:667– 70. Shipton D, Glazier RH, Guan J, Badley EM. Effects of use of specialty services on disease-modifying antirheumatic drug use in the treatment of rheumatoid arthritis in an insured elderly population. Med Care 2004;42:907–13. Roussy JP, Bessette L, Bernatsky S, Rahme E, Legare J, Lachaine J. Use of disease-modifying anti-rheumatic drugs for rheumatoid arthritis in Quebec, Canada [abstract]. Arthritis Rheum 2011;63 Suppl:S855. MacLean C, Louie R, Leake B, McCaffrey D, Paulus D, Brook R, et al. Quality of care for patients with rheumatoid arthritis. JAMA 2000;287:984 –92. MacLean CH, Park GS, Traina SB, Liu HH, Hahn BH, Paulus HE, et al. Positive predictive value (PPV) of an administrative data-based algorithm for the identification of patients with rheumatoid arthritis (RA) [abstract]. Arthritis Rheum 2001;44 Suppl:S106. Suissa S, Garbe E. Primer: administrative health databases in observational studies of drug effects—advantages and disadvantages. Nat Clin Pract Rheumatol 2007;3:725–32. Daskalopoulou S. When statin therapy stops: implications for the patient. Curr Opin Cardiol 2009;24:454 – 60. Avorn J, Monette J, Lacour A, Bohn RL, Monane M, Mogun H, et al. Persistence of use of lipid-lowering medications: a crossnational study. JAMA 1998;279:1458 – 62.