The American Journal of Managed Care March 2016
LDL Cholesterol Response and Statin Adherence Among High-Risk Patients Initiating Treatment
Objectives: The 2013 American College of Cardiology (ACC)/American Heart Association (AHA) cholesterol treatment guideline recommends monitoring percent reduction in low-density lipoprotein cholesterol (LDL-C) among patients initiating statins as an indication of response and adherence. We examined LDL-C reduction and statin adherence among high-risk patients initiating statins in a real-world setting.
Study Design: Retrospective cohort study.
Methods: The study population included Kaiser Permanente Georgia members (n = 1066) with a history of coronary heart disease or risk equivalent(s) initiating statins in 2011. Percent change in LDL-C was defined using measurements before and 60 to 450 days after statin initiation. Statin adherence was defined by proportion of days covered, categorized as high (≥80%), intermediate (50%-79%), and low (<50%).
Results: Overall, 58.4% of patients failed to achieve a ≥30% LDL-C reduction after statin initiation. The prevalences of high, intermediate, and low statin adherence were 41.3%, 23.2%, and 35.6%, respectively. Of patients with high adherence, 42.3% did not achieve a ≥30% reduction in LDL-C compared with 54.7% and 79.7% of those with intermediate and low statin adherence, respectively. After multivariable adjustment, and compared with those with high adherence, the risk ratios for not achieving a ≥30% LDL-C reduction were 1.31 (95% CI, 1.13-1.52) and 1.88 (95% CI, 1.67-2.11), for those with intermediate and low adherence. Women and African Americans were less likely to have high adherence, whereas having cardiologist visits was associated with high adherence.
Conclusions: In a real-world setting, many patients did not achieve a 30% or larger LDL-C reduction. These data support the ACC/AHA recommendation to monitor LDL-C response among patients initiating statins.
Am J Manag Care. 2016;22(3):e106-e115
- Using real-world clinical data, our study demonstrated that nonadherence to statins was very common (>50%) among high-risk patients initiating statins. Nonadherence was associated with suboptimal low-density lipoprotein cholesterol (LDL-C) reduction.
- Even among patients with high adherence, a small LDL-C response (<30%) to statins was very common.
- LDL-C should be monitored following statin initiation to identify suboptimal LDL-C response and medication nonadherence.
The 2013 American College of Cardiology (ACC)/American Heart Association (AHA) Guideline on the Treatment of Blood Cholesterol to Reduce Atherosclerotic Cardiovascular Risk in Adults3 noted that the magnitude of LDL-C reduction (30%-50% and ≥50% reductions for patients initiating moderate-intensity and high-intensity statins, respectively) should be used as a means to identify patients who may not be sufficiently adherent to their statin. This percent reduction is largely based on findings from meta-analyses of randomized controlled trials of statin therapy.4 However, patients in randomized trials represent select groups and there are limited data describing the reduction in LDL-C observed for patients initiating statins in real-world settings. Additionally, low statin adherence is common, and the degree to which low statin adherence accounts for smaller than expected reductions in LDL-C among patients with high-CHD risk is not well studied.5-7
We conducted a retrospective cohort study of high-risk patients to examine LDL-C response following the initiation of statin treatment. Additionally, we investigated the contribution of statin adherence to achievement of a 30% or larger reduction in LDL-C following statin initiation. As part of these analyses, we determined the factors associated with statin nonadherence and the factors beyond adherence that were associated with failure to achieve a 30% or larger reduction in LDL-C following statin initiation.
The current study was conducted at Kaiser Permanente Georgia (KPGA), an integrated healthcare delivery system serving approximately 235,000 members in the greater metropolitan Atlanta area. KPGA maintains comprehensive electronic medical records (EMRs) and other electronic databases that capture nearly 100% of their members’ health services utilization. Members of KPGA are highly representative of its service areas.8
We included KPGA patients who initiated a statin in 2011 (Figure 1). The date of each patient’s first statin fill in 2011 was defined as their “index date.” The “baseline period” was defined as the 365 days prior to the index date. The “follow-up period” was defined as the time between the index date and a patients’ last available LDL-C measurement on or before March 31, 2012.
To be eligible for this analysis, KPGA patients had to have: 1) been 18 years or older on January 1, 2011; 2) not been pregnant during the time between their index date and March 31, 2012; 3) filled at least 1 statin in 2011; 4) continuous health plan enrollment with drug benefits during the baseline and follow-up periods; 5) an LDL-C measurement performed at least 60 days following the index statin fill (for patients with statin prescriptions containing 30 days of supply [>120 and 180 days following the index date for patients receiving 60 and 90 days of supply, respectively]) but on or before March 31, 2012; 6) at least 1 LDL-C measurement during the baseline period; and 7) data on all National Cholesterol Education Program Adult Treatment Panel III (ATP III) risk factors (ie, age, total cholesterol, high-density lipoprotein cholesterol [HDL-C], systolic blood pressure, antihypertensive medication use, smoking, history of CHD, history of diabetes, peripheral artery disease and abdominal aortic aneurysm) from the baseline period.
We excluded patients with any statin fills during the baseline period and restricted our analysis to high-risk patients (ie, those with a history of CHD or a CHD risk equivalent). CHD risk equivalents included diabetes, history of stroke, a 10-year CHD risk greater than 20% based on the Framingham Risk Score, and other forms of symptomatic atherosclerotic disease including peripheral arterial disease, abdominal aortic aneurysm, and carotid artery disease.1
Statin adherence. Statin adherence was defined using the proportion of days covered (PDC). We calculated the PDC as the cumulative number of days for which the patient had a statin available to take between their first fill in 2011 and their last LDL-C measurement during follow-up divided by the total number of days in this interval. The days of supply for statins that each patient had was a cumulative sum of days’ supply from all statin medications regardless of whether the patient changed statin dose or type. In several prior studies of medication adherence using pharmacy fill data, patients with medications available to take on 80% or more days have been categorized as adherent.9-11 Adherence based on this cut point for cardiovascular disease–related conditions has been associated with improved outcomes.12-14 Additionally, the 80% threshold for defining high adherence is recommended by CMS, the Pharmacy Quality Alliance, and the National Quality Forum.15-18 We categorized adherence as high (PDC ≥80%), intermediate (PDC 50%-79%), or low (PDC <50%).12,19,20 We used PDC to define medication adherence because it provides more conservative estimates than the medication possession ratio.21
LDL-C levels. Total cholesterol, HDL-C, and triglycerides were measured at KPGA laboratories as part of a lipid panel following standard measurement procedures. The LDL-C measures were direct measures or calculated using the Friedewald equation.22 Percent change in LDL-C was calculated as the difference between baseline and follow-up LDL-C divided by baseline LDL-C (defined using measurements before and 60 to 450 days after statin initiation). Our primary outcome was having a small reduction in LDL-C, defined as a change in LDL-C less than 30%. A secondary outcome—uncontrolled LDL-C—was defined as an LDL-C 100 mg/dL or greater at the last LDL-C measured on or before March 31, 2012. It is important to note that although the LDL-C target of less than 100 mg/dL is no longer recommended by the 2013 ACC/AHA cholesterol treatment guideline,3 it is included in this analysis as a secondary outcome for comparison purposes.
Study variables were chosen according to a conceptual framework23 describing how factors interact to influence medication adherence.
Patient factors. Demographic information (ie, age, sex, self-reported race) was obtained from Kaiser Permanente’s EMR database. Area-level income was determined by matching patients’ geocoded addresses to 2010 Census data at the census tract level. Smoking status was obtained from the EMR as a self-reported response to whether patients currently smoke cigarettes. History of CHD, diabetes, stroke/carotid disease, and peripheral artery disease (PAD) or abdominal aneurysm were defined by International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic codes obtained from KPGA’s claims database. In addition, hospitalizations during the baseline period were enumerated as a measure of patient health. Medication characteristics (ie, number of medications dispensed, type of statin, statin dose titration, number of statin refills, and use of a high-dose statin) were obtained from the KPGA pharmacy database. Statin type was defined based on the fill most recent to, and preceding, the last LDL-C measurement during follow-up. Statin titration was defined as an upward change in statin dose equivalents (eAppendix Table 1 [eAppendices available at www.ajmc.com]).24-26 High-intensity statins were defined as 80 mg of simvastatin, 40 or 80 mg of atorvastatin, or 20 or 40 mg of rosuvastatin, according to the 2013 ACC/AHA cholesterol treatment guideline.3
The presence of CHD or CHD risk equivalents were defined by ICD-9-CM diagnostic codes during the baseline period and published algorithms (eAppendix Table 2). For patients without CHD or risk equivalents, the Framingham CHD risk score was calculated using the ATP III point scoring system and measurements from the patients’ EMRs from the baseline period.1
Provider factor. Patients’ cardiologist visits were assessed during the follow-up period.
Patient characteristics and percent change in LDL-C were calculated overall and by level of statin adherence. Risk ratios for a reduction in LDL-C <30% and uncontrolled LDL-C associated with low and intermediate versus high statin adherence, were calculated separately using log-binomial regression models with 3 levels of adjustment. An initial model adjusted for age, race, and sex. A second model included additional adjustment for area-level income, smoking, diabetes, history of stroke/carotid disease, Framingham 10-year CHD risk score of less than 20%, history of PAD or abdominal aneurysm, type of statin, number of medications, use of a high-dose statin, titration of statin dose, and cardiologist care. A final model additionally included adjustment for baseline LDL-C.
We conducted sensitivity analyses to examine the robustness of the findings. We examined the adjusted relative risk of a small LDL-C reduction (<30%) associated with statin adherence, restricting the cohort to: 1) patients with a 30-day supply for their initial statin fill, and 2) restricting the cohort to patients with at least 2 measurements of LDL-C measures during the follow-up period (and defining a small reduction in LDL-C as less than 30% on each of the last 2 measurements). Additionally, we calculated adjusted risk ratios for having intermediate/low statin adherence (PDC <80% vs ≥80%). Statistical analyses were conducted using SAS version 9.1.3 (SAS Institute, Cary, North Carolina).