Emma Aarnio

Measurement of statin exposure in the absence of information on prescribed doses

Letter to the editor Pharmacoepidemiological studies on statin use and effectiveness frequently use pharmacy-claims databases as a source of drug exposure data. A variety of methods have been developed to estimate adherence and persistence based on pharmacy-claims databases [1–3]. In each method, theoretical duration of prescriptions is estimated by dividing the quantity dispensed by the prescribed dose. However, many pharmacyclaims databases, including all databases in the Nordic countries, do not include information on days’ supply or prescribed doses in a structured format [4]. Therefore, theoretical duration of statin prescriptions is often estimated based on assumed doses, such as one unit per day [5, 6] or a dose equal to defined daily dose (DDD) per day [7]. In this study, we evaluated the validity of the dosage assumptions of one unit per day and DDD per day for estimating the theoretical duration of statin prescriptions. We identified all statin prescriptions (Anatomical Therapeutic Chemical classification code C10AA [8]) dispensed in May 2007 from the Finnish Prescription Register. We converted free-text dosing instructions into numerical form and analyzed the proportions of prescriptions in which the prescribed daily dose deviated from one unit and DDD, using DDDs available in 2014 [8]. In addition, we described the distribution of prescribed daily doses of statins with means, standard deviations, medians and interquartile ranges. Of all statin prescriptions (n=153,664), 4.2 % were prescribed with a dose other than one unit per day (Table 1). The proportion varied from 0.7 to 12.9 % across individual statins. In contrast, 90.5% of statin prescriptions were prescribed with a dose other than DDD per day. Among individual statins, the proportion varied from 21.8 to 100 %, being 99.2 % for simvastatin. The proportion was similar (99.9 % of simvastatin prescriptions) if the former version of the DDD (15mg) was applied. Our study indicated that the duration of statin prescriptions can be validly estimated by assuming a daily dose of one unit. Our finding is in accordance with two previous studies showing that at least ∼95 % of statin prescriptions are prescribed with a daily dose of one unit [9, 10]. Furthermore, we showed that only one in ten statin prescriptions may be prescribed with a dose equal to DDD per day, indicating that assumed dose of DDD per day would most likely either overor underestimate the duration of statin prescriptions. This would lead to misclassifications of adherence and persistence levels and consequently biased measures of association when studying health outcomes associated with adherence to and discontinuation of statin therapy. Despite the fact that prescribed doses of statins have shown some differences between European countries [11] and seem to have increased over time [11–13], we believe that our finding about lack of validity of the dosage assumption of DDD per day is generalizable to other countries and time periods because statins are unlikely to be prescribed with a dose equal to DDD only. In addition, for statins other than atorvastatin and rosuvastatin, DDD is an average of two T. Romppainen : E. Aarnio School of Pharmacy, University of Eastern Finland, Kuopio, Finland

Register-based predictors of adherence among new statin users in Finland

BACKGROUND Although register-based studies on statin adherence are increasing, for administrative data, little is known about the explanatory power of the predictors that explain adherence. OBJECTIVE The aim was to explore the ability of variables in administrative data to predict statin adherence in an unselected, universally insured population and, especially, to explore dispensation delay (time elapsed between prescription and dispensation) and out-of-pocket costs as explanatory factors. METHODS Statin initiators who were aged 45 to 75 years in 2000-2004 (n = 247, 051) were identified in the Finnish Prescription Register. First-year statin adherence was measured as the proportion of days covered (PDC). The effect of variables related to patient, health care, and payment was assessed with multivariable logistic regression. The C statistic was used to evaluate the explanatory power of different models. RESULTS Overall, 54.6% of the cohort had good adherence (PDC ≥ 80%). The explanatory power of all the models was low (C = 0.666 for the full model). The multivariable models, including only payment variables, had a greater explanatory power (C = 0.627) than models with only patient (C = 0.602) or health care (C = 0.548) variables. A shorter dispensation delay and lower out-of-pocket costs predicted better adherence. Of other patient-related variables, age, presence of acute coronary syndrome, and use of cardiovascular medications were significant predictors of adherence. Type of statin and the prescribers workplace were also significantly associated with adherence. CONCLUSIONS Models based on administrative data do not provide useful prediction of statin adherence. Of the individual predictors, long dispensation delay may serve as a practical tool for identifying patients at risk of poor adherence. Increases in out-of-pocket costs predict nonadherence.

Socioeconomic Inequalities in Statin Adherence Under Universal Coverage: Does Sex Matter?

Background—Previous research shows that low socioeconomic position (SEP; especially low income) is associated with statin nonadherence. We investigated the relationship between SEP and statin adherence in a country with universal coverage using group-based trajectory modeling in addition to the proportion of days covered. Methods and Results—Using data from Finnish healthcare registers, we identified 116 846 individuals, aged 45 to 75 years, who initiated statin therapy for primary prevention of cardiovascular disease. We measured adherence as proportion of days covered over an 18-month period since initiation and identified different adherence patterns based on monthly adherence with group-based trajectory modeling. When adjusted for age, marital status, residential area, clinical characteristics, and copayment, low SEP was associated with statin nonadherence (proportion of days covered <80%) among men (eg, lowest versus highest income quintile: odds ratio, 1.41; 95% confidence interval, 1.32–1.50; basic versus higher-degree education: odds ratio, 1.18; 95% confidence interval, 1.13–1.24; unemployment versus employment: odds ratio, 1.17; 95% confidence interval, 1.10–1.25). Among women, the corresponding associations were different (P<0.001 for sex-by-income quintile, sex-by-education level, and sex-by-labor market status interactions) and mainly nonsignificant. Results based on adherence trajectories showed that men in low SEP were likely to belong to trajectories presenting a fast decline in adherence. Conclusions—Low SEP was associated with overall and rapidly increasing statin nonadherence among men. Conversely, in women, associations between SEP and nonadherence were weak and inconsistent. Group-based trajectory modeling provided insight into the dynamics of statin adherence and its association with SEP.

Sequential cohort design applying propensity score matching to analyze the comparative effectiveness of atorvastatin and simvastatin in preventing cardiovascular events.

Background Sequential cohort design (SCD) applying matching for propensity scores (PS) in accrual periods has been proposed to mitigate bias caused by channeling when calendar time is a proxy for strong confounders. We studied the channeling of patients according to atorvastatin and simvastatin initiation in Finland, starting from the market introduction of atorvastatin in 1998, and explored the SCD PS approach to analyzing the comparative effectiveness of atorvastatin versus simvastatin in the prevention of cardiovascular events (CVE). Methods Initiators of atorvastatin or simvastatin use in the 45–75-year age range in 1998–2006 were characterized by their propensity of receiving atorvastatin over simvastatin, as estimated for 17 six-month periods. Atorvastatin (10 mg) and simvastatin (20 mg) initiators were matched 1∶1 on the PS, as estimated for the whole cohort and within each period. Cox regression models were fitted conventionally, and also for the PS matched cohort and the periodically PS matched cohort, to estimate the hazard ratios (HR) for CVEs. Findings Atorvastatin (10 mg) was associated with a 11%–12% lower incidence of CVE in comparison with simvastatin (20 mg). The HR estimates were the same for a conventional Cox model (0.88, 95% confidence interval 0.85–0.91), for the analysis in which the PS was used to match across all periods and the Cox model was adjusted for strong confounders (0.89, 0.85–0.92), and for the analysis in which PS matching was applied within sequential periods (0.88, 0.84–0.92). The HR from a traditional PS matched analysis was 0.80 (0.77–0.83). Conclusions The SCD PS approach produced effect estimates similar to those obtained in matching for PS within the whole cohort and adjusting the outcome model for strong confounders, but at the cost of efficiency. A traditional PS matched analysis without further adjustment in the outcome model produced estimates further away from unity.

Drug Adherence with Cardiovascular Medicines: Statins and Aspirin

Statins and aspirin are both widely used low-cost therapies, with known adverse effect profiles, that are prescribed to treat a symptomless condition or to prevent future adverse cardiovascular events. Despite their proven efficacy in the prevention of cardiovascular morbidity and mortality, non-adherence to these medicines is prevalent. Reported rates of non-adherence vary substantially across studies; based on meta-analyses, only every second patient who initiates statin therapy adheres to it, and non-adherence is more common in primary than in secondary prevention. Adherence to aspirin is somewhat better, being comparable to statin adherence in secondary prevention (~70%). However, these rates underestimate non-adherence by missing those who do not initiate medicine use (primary non-adherence). Use of these cardioprotective medicines is a dynamic process where adherence can change over time and many patients who discontinue medicine use reinitiate it. The most important patient-reported reason for non-adherence is adverse effects. Recently, negative media coverage has been implicated as a reason for discontinuation of statin therapy. The multitude of risk factors for non-adherence, in addition to complexity of adherence behavior, is likely to complicate efforts to improve adherence to statins and aspirin.

Important factors affecting the choice of an oral anticoagulant may be missed in database studies

We read with great interest the article by Gundlund et al. ‘Initiation of anticoagulation in atrial fibrillation: which factors are associated with choice of anticoagulant?’ [1]. The authors characterized a cohort of more than 50 000 patients with atrial fibrillation (AF) who initiated warfarin or a nonvitamin K antagonist oral anticoagulant (NOAC; dabigatran, rivaroxaban or apixaban) during the period 2011–2016 using comprehensive Danish health databases. Characteristics of interest included patients’ gender, age, CHA2DS2-VASc and HAS-BLED scores, comorbidities (e.g. stroke, myocardial infarction, chronic kidney disease, history of bleeding, alcohol abuse, hypertension) and concomitant medication use. Consistent with findings from other database studies [2–4], differences in the estimated risk of stroke were observed between initiators of various oral anticoagulants [1]. In their discussion, the authors acknowledge that they did not have access to many important factors of potential influence in the choice of an oral anticoagulant, such as race, body mass index, smoking, alcohol use, creatinine clearance and socio-economic status, and call for studies to investigate how physicians make choices between different oral anticoagulants.

Health-related quality of life in relation to symptomatic and radiographic definitions of knee osteoarthritis: data from Osteoarthritis Initiative (OAI) 4-year follow-up study

BackgroundThe purpose was to quantify the decrement in health utility (referred as disutility) associated with knee osteoarthritis (OA) and different symptomatic and radiographic uni- and bilateral definitions of knee OA in a repeated measures design of persons with knee OA or at increased risk of developing knee OA.MethodsData were obtained from the Osteoarthritis Initiative database. SF-12 health-related quality of life was converted into SF-6D utilities, and were then handled as the health utility loss by subtracting 1.000 from the utility score, yielding a negative value (disutility). Symptomatic OA was defined by radiographic findings (Kellgren-Lawrence, K-L, grade ≥ 2) and frequent knee pain in the same knee. Radiographic OA was defined by five different definitions (K-L ≥ 2 unilaterally / bilaterally, or the highest / mean / combination of K-L grades of both knees). Repeated measures generalized estimating equation (GEE) models were used to investigate disutility in relation to these different definitions.ResultsUtility decreased with worsening of symptomatic or radiographic status of knee OA. The participants with bilateral and unilateral symptomatic knee OA had 0.03 (p < 0.001) and 0.02 (p < 0.001) points lower utility scores, respectively, compared with the reference group. The radiographic K-L grade 4 defined as the mean or the highest grade of both knees was related to a decrease of 0.04 (p < 0.001) and 0.03 (p < 0.001) points in utility scores, respectively, compared to the reference group.ConclusionsKnee OA is associated with diminished health-related quality of life. Health utility can be quantified in relation to both symptomatic and radiographic uni- and bilateral definitions of knee OA, and these definitions are associated with differing disutilities. The performance of symptomatic definition was better, indicating that pain experience is an important factor in knee OA related quality of life.

Physicians’ views on patient participation in choice of oral anticoagulants in atrial fibrillation - a qualitative study

Direct oral anticoagulants provide an alternative to vitamin K antagonists for the anticoagulation therapy in atrial fibrillation (AF). The availability of several treatment options with different attributes makes shared decision‐making appropriate for the choice of anticoagulation therapy. The aim of this study was to understand how physicians choose an oral anticoagulant (OAC) for patients with AF and how physicians view patients’ participation in this decision. Semi‐structured interviews with 17 Finnish physicians (eight general practitioners and nine specialists) working in the public sector were conducted. An interview guide on experience, prescribing and opinions about oral anticoagulants was developed based on previous literature. The data were thematically analysed using deductive and inductive approaches. Based on the interviews, patients opinion was the most influential factor in decision‐making when there were no clinical factors limiting the choice between OACs. Of patients preferences, the most important was the attitude towards co‐payments of OACs. Patients’ opinions on monitoring of treatment, dosing and antidote availability were also mentioned by the interviewees. The choice of an OAC in AF was patient‐centred as all interviewees expressed that patients opinion affects the choice.