Giovanni Corrao, PhD1; Valentino Conti, MSc1,2; Luca Merlino, MD3; Alberico L. Catapano, PhD4; and Giuseppe Mancia, MD, PhD5
1Department of Statistics, Unit of Biostatistics and Epidemiology, University of Milano-Bicocca, Milan, Italy;
2Regional Centre for Pharmacovigilance, Lombardy Region, Milan, Italy;
3Operative Unit of Territorial Health Services, Lombardy Region, Milan, Italy;
4Department of Pharmacological Sciences, and Centre for Pharmacoepidemiology and Pharmacoutilization, University of Milan, Milan, Italy; and
5Department of Clinical and Preventive Medicine, University of Milano-Bicocca, Milan, Italy
Background: Previous studies have reported that statin use was associated with reductions in cardiovascular morbidity and mortality among patients with dyslipidemia, even without established cardiovascular disease. However, inadequate adherence may reduce statins’ protective effects.
Objective: The aim of this work was to investigate whether an association exists between statin adherence when used as primary prevention and risk of subsequent ischemic heart disease (IHD).
Methods: People aged ≥18 years who were residents of Italy’s Lombardy region and were newly treated with statins in 2002 to 2003 were assessed as part of a retrospective analysis of data from a healthservices database. Patients who were hospitalized for IHD during this period were identified with hospitaldischarge information from a health-services database; IHD-related hospitalizations were identified by International Classification of Diseases, Ninth Revision, Clinical Modification, codes for acute myocardial infarction (410), acute and subacute forms of IHD (411), and/or codes concerning coronary revascularization (36.0–36.9). Four groups of patients were excluded: those with ≥1 lipid-lowering drug within 2 years before the index prescription (to limit the sample to treatment initiators); those who had been hospitalized for cardiovascular disease or had used medications for IHD or heart failure within 2 years before the index date (to limit the study to primary prevention); those who did not have ≥1 year of follow-up; and those who received only 1 dispensation of a statin during the first year after the index prescription. Follow-up continued until hospitalization for IHD or any other cardiovascular cause, death from any cause, emigration, or the end of the study period (June 30, 2007). The proportion of days covered (PDC) by therapy with statins was the exposure variable; it served as a proxy for adherence. PDC (and therefore adherence) was categorized as very low (≤25%), low (26%–50%), intermediate (51%–75%), or high (>75%) coverage. A proportional hazards model was fitted to estimate hazard ratio (HR) and 95% CIs for the association between time-dependent categories of PDC and time of IHD hospitalization, after correcting for covariates.
Results: A group of 90,832 patients was included; during follow-up, 1480 patients experienced a hospitalization for IHD. After the Cox proportional hazards model was adjusted for age, sex, type of statin dispensed at index prescription, current use of other selected drugs (ie, antidiabetics, antihypertensives, digitalis or organic nitrates, or other cardiac medications), Charlson comorbidity index, and whether or not a given patient switched statins, those with low, intermediate, or high statin coverage had HR (95% CI) values of 0.85 (0.72–0.98), 0.82 (0.71–0.95), and 0.81 (0.71–0.94), respectively, compared with patients with very low coverage.
Conclusions: In these Italian subjects without a history of cardiovascular disease, low, intermediate, and high levels of adherence to statin pharmacotherapy were associated with lower risk of nonfatal IHD compared with those who had very low (≤25%) adherence. However, these findings have several limitations, such as the use of database information (rather than medical records), the assumption that PDC accurately represented actual adherence, and confounding (ie, unmeasured factors related to PDC or to adherence may have influenced clinical outcomes). (Clin Ther. 2010;32:300–310) c 2010 Excerpta Medica Inc.
Key words: database analysis, hyperlipidemia, ischemic heart disease, statin, adherence, cardiovascular disease, risk.
Large randomized clinical trials have reported that statins reduced cardiovascular morbidity and mortality among patients with dyslipidemia,1–6 even when established cardiovascular disease was absent.7 However, because published reports suggest that the full therapeutic potential of lipid-lowering pharmacotherapy may be achieved after 1 to 2 years of continuous treatment,8 inadequate adherence to statin therapy (ie, not taking prescribed medication according to the designated schedule) may result in ineffective treatment.9–13 Therefore, the efficacy described in randomized clinical trials may not represent what is observed in usual community care.14,15 A question that remains is whether better adherence to statin therapy might reduce the risk of developing ischemic heart disease (IHD) in a population with a relatively low risk of cardiovascular disease.
The role of adherence in the onset of coronary artery disease was recently investigated in a case–control study that was part of a larger cohort of new users of statins.16 However, very high levels of medication adherence characterized the investigated population, thereby limiting new understanding about the impact of adherence with lipid-lowering therapy on primary prevention of IHD in a real-life setting. We conducted a large, population-based cohort study to assess whether an association exists between statin adherence in primary prevention and the risk of subsequent IHD in Italian subjects.
The data used for this study were retrieved from the health service database of Lombardy, a region of Italy that accounts for ~16% (9 million people) of the country’s population. In Italy, the National Health Service (NHS) provides universal coverage for many health care services, including drug treatment for hyperlipidemia.
Since 1997, Lombardy has used an automated database system to collect detailed information about the use of health services, including data regarding the following: residents who received health assistance from the NHS (representing most of the region’s residents); demographic and administrative data; all hospitalizations in the public and private hospitals of Lombardy; outpatient drug prescriptions that were reimbursable by the NHS and dispensed by a pharmacy in Lombardy. These data can be linked for each person using a unique identification code.
At the time of this study, formal approval of study protocol from an ethical committee or institutional review board was not required in Italy. However, to preserve patients’ privacy, the investigators decided to deidentify patient codes from the database by converting each one to an anonymous code; the table used for these conversions was ultimately destroyed to ensure that subjects could not later be identified. Full details about this procedure have been described previously.17
Study Cohort and Follow-Up
The target population was all Lombardy residents aged ≥18 years who were beneficiaries of the NHS. According to the 2001 Italian census, this population comprised 7,606,219 individuals.18 Those who received ≥1 statin prescription anytime from 2002 to 2003 were identified, and the first prescription dispensation was defined as the index prescription. To make the data as relevant as possible to the study objective, 4 categories of patients were excluded: patients to whom ≥1 lipid-lowering drug (ie, statin, fibrate, resin, or niacin) was dispensed during the 2 years before the index prescription (to limit the included subjects to newly treated individuals); patients who had been hospitalized for cardiovascular disease, or to whom drugs used for IHD or heart failure (eg, digitalis, organic nitrates) had been prescribed during the 2 years before the index prescription (to focus the data on primary cardiovascular prevention); patients who did not reach ≥1 year of follow-up (to ensure a sufficiently prolonged exposure to therapy with statins); and patients who, during the first year after the date of the index prescription, received only 1 dispensation of statin (to ensure that continuous drug treatment was indicated for included subjects). Each member of the selected cohort accumulated person-years of follow-up from the date of the index prescription until the earliest of the following end dates: hospital admission for IHD (outcome onset) or any other cardiovascular cause; death from any cause; emigration; or end of follow-up (June 30, 2007).
Because we had no information about drug prescriptions for inpatients, the observation period was temporarily censored at the date of hospital admission for a cause that differed from cardiovascular disease, and reestablished 10 days after hospital discharge.
Assessment of Exposure to Statins
Drug types, dosages, and number of canisters dispensed to each subject during follow-up were retrieved from the outpatient drug prescription database and used to measure exposure to the drug of interest. The period covered by a prescription was calculated by the number of tablets in the dispensed canisters, assuming a treatment schedule of 1 tablet per day.19,20 For overlapping prescriptions, the individual was assumed to have refilled the first one early and completed it before starting the second.
Adherence to therapy with statins was assessed by the ratio between the cumulative number of days during which the medication was available and the whole number of days of follow-up (proportion of days covered [PDC]).21 This indicator was categorized into 4 groups: very low coverage (≤25%); low coverage (26%–50%); intermediate coverage (51%–75%); and high coverage (>75%). These PDC categories served as a proxy for adherence level.
Assessment of Outcomes
The hospital discharge database was used to identify cohort members who experienced ≥1 ospitalization for IHD during the follow-up period. The World Health Organization Multinational Monitoring of Trends and Determinants in Cardiovascular Disease criteria for diagnosing IHD were followed.22 Codes included acute myocardial infarction (International Classification of Diseases, Ninth Revision, Clinical Modification, code 410), acute and subacute forms of IHD (code 411), and/or codes concerning coronary revascularizations (codes 36.0–36.9). The earliest hospitalization date for any of these causes was considered to be the time of outcome onset.
Cox proportional hazards regression was used to estimate the hazard ratio (HR) and 95% CIs for the association between exposure to statins and time of outcome onset. The predictor variables of interest were the factors constructed according to the categories of PDC, using the first category (ie, very low coverage) as reference. Because drug exposure may vary over time, assessment of its effect requires proper accounting for the cumulative and varying nature of the measure. This was done by fitting the Cox model to include time-dependent variables. With this approach, each patient’s adherence is recalculated from the start of follow-up to the time of each outcome. Thus, the HR associated with a category of PDC is derived using exposure information concurrent to the observed outcomes, rather than the cumulative exposure profile over the full length of follow-up.23 The effect of adherence to statin treatment on IHD risk was measured by unadjusted, age- and sex-adjusted, and fully adjusted estimates of HRs. Full adjustment included age, sex, type of statin dispensed in the index prescription, and the current use of other drugs, such as medications commonly used for treatment of hypertension, heart failure, coronary heart disease, or diabetes mellitus. In addition, the Charlson comorbidity index score was calculated for each patient using diagnostic information from inpatient encounters in the 2 years leading up to and 1 year after the index date.24 The score was included as a covariate in the fully adjusted model. Finally, an indicator to assess whether switching among statins was experienced by each cohort member was included as a time-dependent factor in the fully adjusted model. Tests for linear trend were based on the statistical significance of the regression coefficient of PDC obtained by scoring the corresponding categories. To evaluate the joint effect of adherence on statin treatment and covariates, interaction terms were included in the model; their significance was tested using the likelihood ratio test.25
To verify the robustness of our findings, 2 sensitivity analyses were performed to assess the effect of alternative categorizations of adherence and of censoring criteria on the IHD risk. The SAS statistical package (SAS Institute Inc., Cary, North Carolina), version 9.1, was used for the analyses. For all hypotheses, P < 0.05 (2 tailed) was considered to be significant. Statistical tests are reported in the Results where appropriate (eg, test for linear trend).
The distribution of the exclusion criteria is shown in the figure. The 90,832 patients included in the study accumulated 4,638,161 person-months of follow-up (a mean of 51 months per patient) and 2,040,480 personmonths of statin therapy (a mean of 44% of follow-up time was covered by statin therapy). The cohort generated 1480 IHD events, with an incidence rate of 38 cases per 10,000 person-years at risk.
Table I summarizes selected characteristics of the whole cohort and of the categories of varying levels of adherence to statin therapy. Mean age at entry was 61.8 years, and 59.3% of patients (53,901/90,832) were women. Most patients were initially treated with simvastatin or atorvastatin and concomitantly received antihypertensive drug therapy. Overall, 59.5% of patients (54,081/90,832) had very low or low adherence to statin therapy (ie, <50% of days of followup were covered by statins), and only 19.6% (17,833/90,832) had high adherence (ie, >75% of days of follow-up were covered by statins). Compared with those who had high adherence, patients who had very low or low adherence were slightly younger, and had a better profile of cotreatments and comorbidities. The proportion of patients who started with simvastatin decreased along the categories of adherence (from 38.3% [12,064/31,537] for very low adherence to 29.4% [5249/17,833] for high adherence), whereas the opposite happened for patients who started with fluvastatin (from 9.6% [3027/31,537] among those with very low adherence to 19.8% [3524/17,833] among those with high adherence). Switching between statins was experienced by 6.3% of patients (1995/31,537) with very low adherence, compared with the other categories, among whom the proportion of switchers ranged from a low of 2.3% (416/17,833) for high adherence to 3.1% (695/22,544) for low adherence.
Table II shows associations between adherence and IHD risk. Crude and partially adjusted estimates did not generally offer evidence that higher levels of adherence were associated with better outcomes (trend test, P = NS), although the low adherence group had less risk than the very low adherence group (HR = 0.86 [95% CI, 0.75–0.99] in the crude analysis; HR = 0.86 [0.75–0.99] in the partially adjusted analysis). However, fully adjusted estimates indicated that, compared with very low adherence, IHD risk was reduced among those with low (HR = 0.85 [0.72–0.98]), intermediate (HR = 0.82 [0.71–0.95]), or high adherence (HR = 0.81 [0.71–0.94]). The trend test indicated a statistically significant effect in the fully adjusted analysis (P = 0.014). Based on the fully adjusted model, the risk of IHD increased with age (HR = 1.03 [1.02–1.03] per 1-year increase), and was higher in men (HR = 2.92 [2.62–3.25]) and among patients who were concomitantly treated with antidiabetic drugs (HR = 1.69 [1.51–1.89]), antihypertensive drugs (HR = 1.79 [1.54–2.09]), or drugs used for heart failure (ie, digitalis glycoside or organic nitrate) (HR = 3.19 [2.81–3.62]). Those with a Charlson comorbidity index score ≥2 also had higher IHD risk, compared with those with a score of 0 (HR = 1.45 [1.18–1.78]), although there was no significant difference between those who had a score of 1 and those who had a score of 0. Finally, there was no evidence that the type of statin administered as initial therapy, or the experience of switching among statins, was significantly associated with changes in IHD risk. However, P values for our findings did not offer statistical evidence that the effect of adherence to statin therapy on the risk of IHD was modified by the concomitant use of drugs for diabetes mellitus or hypertension, or of other cardiac medications. Similarly, using this criterion, Charlson comorbidity score ≥1 was not associated with a statistically significant change in IHD risk.
These findings were supported by varying the adherence categorization; that is, patients with PDC >90% had a 21% reduction in IHD risk with respect to those with PDC <20%. Moreover, with respect to patients with PDC <25%, those with PDC >75% had 17% and 14% reductions in IHD risk when the observations were censored at the date of hospital admission for any cause and at the date of hospital admission for cardiovascular disease, respectively.
The results of this large, population-based observational cohort study suggest that adherence to therapy with statins is generally poor in daily clinical practice in Italy and indicates that nearly 1 in 5 ischemic nonfatal events may be avoided by optimizing adherence to therapy with statins (ie, 0.19 HR reduction, as shown in Table II). Our study indicated that the initial prescription was not followed by any other statin dispensation in 47.5% of newly treated patients [82,145/172,977], who were therefore excluded from the investigated cohort. Furthermore, only 44% of the follow-up period was covered by therapy with statins. Finally, 59.5% of patients used statins for <50% of the follow-up period. These findings support the results of previously published reports that initial therapy with a statin was frequently abandoned just a few months after treatment began.9–13,19,20,26,27 A recent metaanalysis suggested that as many as 6 out of 10 patients may stop taking statins during the first 6 months of treatment.28 The reduction in nonfatal ischemic events observed in the current study is comparable with that reported in a recent observational investigation of primary cardiovascular prevention.16 In that study, compared with patients with <20% adherence, those with adherence ≥80% had an 18% reduction in the risk of coronary artery disease. Statin use has been associated with reductions in the incidence of recurrent acute myocardial infarction and death among patients who have had a myocardial infarction,1,3,29,30 as well as reduced mortality among patients with no previous indication of cardiovascular disease.29 Other trials have enrolled subjects with and without previous cardiovascular diseases, but who were at high risk of future cardiovascular events, and yielded similar results.31,32 Small relative reductions in IHD risk have been found by summarizing the evidence from randomized trials and cohort studies.8
Compared with patients who had very low adherence (≤25%), IHD risk was significantly lower in those with low adherence (26%–50%); those with intermediate or high levels of adherence experienced only a modest additional reduction. This implies that, versus negligible adherence, far-from-perfect adherence to a prescribed treatment regimen of statins may offer considerable advantages. We suspect, however, that this unexpected finding might be partially explained by confounding. It is unclear whether the effect size observed across trials varied because of the characteristics of the trials, or because the statin agents had different effects. The Pravastatin or Atorvastatin Evaluation and Infection Therapy-Thrombolysis in Myocardial Infarction 22 trial reported that, compared with patients treated with pravastatin, those who received atorvastatin had a 28% lower risk of death, myocardial infarction, or rehospitalization.33 In our population-based study, which was designed to represent regular clinical practices in Italy, the 4 statins considered in the analysis appeared to be associated with similar changes in cardiac risk, based on the fully adjusted risk model comparing atorvastatin, fluvastatin, pravastatin, and simvastatin as initial pharmacotherapy. Comparable findings have been reported for secondary prevention of myocardial infarction in elderly patients.34 However, we also observed that patients initially treated with pravastatin or simvastatin had lower adherence, suggesting the possibility that there may be a greater incidence of adverse events that influence adherence when an older agent is used.35 Because outcome data for the present study were drawn from the records of hospitalized patients, our findings refer only to nonfatal outcomes. However, there is no evidence that statin therapy has different effects on nonfatal versus fatal coronary event risk36; therefore, the selection of only nonfatal events should not substantially affect the validity of our estimates. This study was unique in several respects. First, it was based on a uniformly organized health care system that allowed a large-scale, population-based design.
Second, basing determination of adherence to pharmacotherapy on a drug-prescription database allowed the use of high-quality data. In fact, prescription databases have been repeatedly checked and found to be accurate, likely because of their importance in the reimbursement of the pharmacies by the public health care system, and because the filing of an incorrect claim about drugs dispensed has legal consequences.37
Third, patients were identified from the point of the initial dispensation of statins; the complete sequence of subsequent lipid-lowering prescription drug use was known. Fourth, we were able to identify patients whose prescription history did not reveal signs of previous cardiovascular disease, so our findings represent statin use as part of primary cardiovascular prevention. Fifth, patients followed for <1 year were excluded because the efficacy of statin agents in cardiovascular prevention may take some time before becoming apparent.16,38 Finally, sensitivity analyses indicated the robustness of our results by accounting for alternative categorizations of adherence and censoring criteria. Nevertheless, some sources of systematic uncertainty, such as possible misclassification and confounding, should be considered in the interpretation of our findings. Because of privacy regulations, medical records were not available for analysis. A previous study by our group found that hospitalization for IHD has a positive predictive value of 88% when IHD diagnoses from hospital discharge information and a local population-based registry were compared.39 Outcome misclassification, however, can be expected to occur independently of adherence status, so this source of uncertainty was likely to affect our estimates to favor the null hypothesis, rather than to favor evidence of an effect of adherence on IHD risk.40 Evaluation of statin adherence in the present study was based on pharmacy-dispensing information. This method, however, assumes that the proportion of days covered by a prescription corresponds to the proportion of days of medication use. Small, insignificant differences between the assessed number of dispensed pills and the actual pill count were reported in a recent study investigating adherence to statin therapy over 12 months.41 Furthermore, data on dispensing history have shown considerable consistency with other compliance measures, drug serum levels, and clinical drug effects.42 Nevertheless, the use of medication dispensing as a measure of adherence remains a main source of uncertainty in our estimates. The means of assessing adherence as PDC used in this study may also serve as a limitation, but there was a rationale for establishing ranges of PDC values to categorize adherence, rather than including the PDC in the analyses as a continuous variable. Doing so would have assumed a linear relationship. Attempts to represent nonlinearity could be made by means of polynomial models (eg, quadratics), but this approach may be unfeasible when the study design includes the exposure variable as a time-dependent variable. Categorical methods are the most commonly used ways to analyze the relationship between adherence to pharmacotherapy and disease risk. To determine whether the category assignments we created might have influenced our findings, we performed a sensitivity analysis. As noted, this analysis confirmed the robustness of our results. The allocation of statin therapy was not randomized in our study; therefore, the results may be affected by confounding. That is, the observed reduction in IHD risk associated with increased adherence to statin therapy might have been generated by patients’clinical characteristics, such as the severity of hyperlipidemia, the rate of lipid control, the presence of comorbidities, and cardiovascular risk factors, as well as their correction by concomitant treatments. We attempted to limit confounding by adjusting for several demographic, therapeutic, and clinical characteristics, but greater adherence to statins could still be a surrogate for other, unmeasured characteristics. Because the included patients were all receiving statins, there is a lower likelihood of confounding by indication than if users and nonusers were compared.43 Confounding by indication could have occurred if patients with better prognosis were more likely to be treated more intensively. Conversely, patients at high cardiovascular risk (eg, those with severe hyperlipidemia and those with worse clinical profiles) have been described as more adherent to treatment than patients at lower cardiovascular risk.26 This suggests that the observed inverse relationship between adherence to statin therapy and IHD risk might be even stronger than described in our study. The modest additional reduction in IHD risk noted among patients with intermediate or high adherence, compared with those with low adherence, might also be explained by this source of confounding. The decision to exclude patients with only 1 dispensation of a statin did not skew the observed relationship between adherence and IHD risk in favor of statin therapy. In fact, patients who were excluded from the analysis because they had only 1 statin prescription had lower rates of IHD events than those with ≥2 prescriptions (25 and 38 cases per 10,000 personyears at risk, respectively), suggesting that patients who continued therapy may have been in worse health, and therefore at higher baseline risk of IHD, than those who did not. If the study had been designed to include patients with only 1 statin dispensation, confounding by indication might have been greater. Nevertheless, unmeasured residual confounders (which may overinflate the supposed protective effect of adherence to statin therapy) cannot be excluded. For instance, patients who do not adhere to therapy may have other traits that can be associated with worsened outcomes, such as lower cognitive function or depression and associated adverse health behaviors,44,45 as well as lifestyle habits, including smoking, lack of exercise, and alcohol abuse.16,26
These factors may introduce a bias to our results because they are well-documented risk factors for IHD and they are more likely to be present among patients who are not adherent to their medications.46 It has been recently reported that adherence to statin therapy was related closely to the attainment of an LDL-C goal, and that the rate of attaining such a goal increased substantially when adherence was >80%.47 Nevertheless, because adherence to therapy is not a surrogate for hyperlipidemia control, the lack of information about cholesterol values before and after treatment is a limitation of our study.
In these Italian subjects without a history of cardiovascular disease, low, intermediate, and high levels of adherence to statin pharmacotherapy were associated with lower risk of nonfatal IHD compared with those who had very low (≤25%) adherence. However, these findings have several limitations, such as the use of database information (rather than medical records), the assumption that PDC accurately represented actual adherence, and confounding (ie, unmeasured factors related to PDC or to adherence may have influenced clinical outcomes).
This study was funded by grants from the Italian Ministry for University and Research. The authors have indicated that they have no conflicts of interest regarding the content of this article.
1. Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: The Scandinavian Simvastatin Survival Study (4S). Lancet. 1994;344:1383–1389.
2. West of Scotland Coronary Prevention Study: Identification of high-risk groups and comparison with other cardiovascular intervention trials. Lancet. 1996;348:1339–1342.
3. The Long-Term Intervention with Pravastatin in Ischaemic Disease (LIPID) Study Group. Prevention of cardiovascular events and death with pravastatin in patients with coronary heart disease and a broad range of initial cholesterol levels. N Engl J Med. 1998;339:1349–1357.
4. Downs JR, Clearfield M, Weis S, et al. 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.
5. Farmer JA, Gotto AM Jr. The Heart Protection Study: Expanding the boundaries for high-risk coronary disease prevention. Am J Cardiol. 2003;92:3i–9i.
6. Schwartz GG, Olsson AG, Ezekowitz MD, et al, for the Myocardial Ischemia Reduction with Aggressive Cholesterol Lowering (MIRACL) Study Investigators. Effects of atorvastatin on early recurrent ischemic events in acute coronary syndromes: The MIRACL study: A randomized controlled trial. JAMA. 2001;285:1711–1718.
7. Brugts JJ, Yetgin T, Hoeks SE, et al. The benefits of statins in people without established cardiovascular disease butwith cardiovascular risk factors: Meta-analysis of randomized controlled trials. BMJ. 2009;338:b2376.
8. Law MR, Wald NJ, Rudnicka AR. Quantifying effect of statins on low density lipoprotein cholesterol, ischaemic heart disease, and stroke: Systematic review and meta-analysis. BMJ. 2003;326:1423.
9. Meyer JW, Schultz JS, O’Donnell JC, et al. Patterns and effectiveness of lipid-lowering therapies in a managed care environment. Value Health. 2005;8:601–612.
10. Kamal-Bahl SJ, Burke T, Watson D, Wentworth C. Discontinuation of lipid modifying drugs among commercially insured United States patients in recent clinical practice. Am J Cardiol. 2007;99:530–534.
11. Caspard H, Chan AK, Walker AM. Compliance with a statin treatment in a usual care setting: Retrospective database analysis over 3 years after treatment initiation in health maintenance organization enrollees with dyslipidemia. Clin Ther. 2005;27: 1639–1646.
12. Schultz JS, O’Donnell JC, Mc-Donough KL, et al. Determinants of compliance with statin therapy and low-density lipoprotein cholesterol goal attainment in a managed care population. Am J Manag Care. 2005; 11:306–312.
13. Walley T, Folino-Gallo P, Stephens P, Van Ganse E. Trends in prescribing and utilization of statins and other lipid lowering drugs across Europe 1997–2003. Br J Clin Pharmacol. 2005;60:543–551.
14. Jackevicius CA, Mamdani M, Tu JV. Adherence with statin therapy in elderly patients with and without acute coronary syndromes. JAMA. 2002;288:462–467.
15. Benner JS, Glynn RJ, Mogun H, et al. Long-term persistence in use of statin therapy in elderly patients. JAMA. 2002;288:455–461.
16. Perreault S, Dragomir A, Blais L, et al. Impact of better adherence to statin agents in the primary prevention of coronary artery disease. Eur J Clin Pharmacol. 2009;65:1013–1024.
17. Corrao G, Cesana G, Merlino L. Pharmacoepidemiological research and the linking of electronic healthcare databases available in the Italian region of Lombardy. BioMed Stat Clin Epidemiol. 2008;2:117–125.
18. 2001 Census of the population and homes [in Italian]. http://www. istat.it/censimenti/popolazione. Accessed January 14, 2010.
19. Poluzzi E, Strahinja P, Lanzoni M, et al. Adherence to statin therapy and patients’ cardiovascular risk: A pharmacoepidemiological study in Italy. Eur J Clin Pharmacol. 2008;64: 425–432.
20. Helin-Salmivaara A, Lavikainen P, Korhonen MJ, et al. Long-term persistence with statin therapy: A nationwide register study in Finland. Clin Ther. 2008;30:2228–2240.
21. Rasmussen JN, Chong A, Alter DA. Relationship between adherence to evidence-based pharmacotherapy and long-term mortality after acute myocardial infarction. JAMA. 2007; 297:177–186.
22. Tunstall-Pedoe H, Kuulasmaa K, Amouyel P, et al. Myocardial infarction and coronary deaths in the World Health Organization MONICA Project: Registration procedures, event rates, and case-fatality rates in 38 populations from 21 countries in four continents. Circulation. 1994;90: 583–612.
23. Marubini E, Valsecchi MG. Analysing Survival Data from Clinical Trials and Observational Studies. Chichester, UK: Wiley; 1995.
24. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613–639.
25. Woodward M. Epidemiology: Study Design and Data Analysis. 2nd ed. Boca Raton, Fla: Chapman & Hall; 2005.
26. Perreault S, Blais L, Lamarre D, et al. Persistence and determinants of statin therapy among middle-aged patients for primary and secondary prevention. Br J Clin Pharmacol. 2005;59:564–573.
27. Vinker S, Shani M, Baevsky T, Elhayany A. Adherence with statins over 8 years in a usual care setting. Am J Manag Care. 2008;14:388–392.
28. Liberopoulos EN, Florentin M, Mikhailidis DP, Elisaf MS. Compliance with lipid-lowering therapy and its impact on cardiovascular morbidity and mortality. Expert Opin Drug Saf. 2008;7:717–725.
29. Shalev V, Chodick G, Silber H, et al. Continuation of statin treatment and all-cause mortality. A population based cohort study. Arch Intern Med. 2009;169:260–268.
30. Sacks FM, Pfeffer MA, Moye LA, et al, for the Cholesterol and Recurrent Events Trial Investigators. The effect of pravastatin on coronary events after myocardial infarction in patients with average cholesterol levels. N Engl J Med. 1996;335:1001–1009.
31. Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20,536 high-risk individuals: A randomized placebo-controlled trial. Lancet. 2002;360:7–22.
32. Shepherd J, Blauw GJ, Murphy MB, et al, for the PROSPER Study Group. Pravastatin in elderly individuals at risk of vascular disease (PROSPER): A randomised controlled trial. PROspective Study of Pravastatin in the Elderly at Risk. Lancet. 2002;360: 1623–1630.
33. Ray KK, Cannon CP, McCabe CH, et al, for the PROVE IT-TIMI 22 Investigators. Early and late benefits of high-dose atorvastatin in patients with acute coronary syndromes: Results from the PROVE IT-TIMI 22 trial. J Am Coll Cardiol. 2005;46:1405–1410.
34. Zhou Z, Rahme E, Abrahamowicz M, et al. Effectiveness of statins for secondary prevention in elderly patients after acute myocardial infarction: An evaluation of class effect. CMAJ. 2005;172:1187–1194.
35. MRC/BHF Heart Protection Study Collaborative Group. Effects of simvastatin 40 mg daily on muscle and liver adverse effects in a 5-year randomized placebo-controlled trial in 20,536 high-risk people. BMC Clin Pharmacol. 2009;9:6.
36. LaRosa JC, He J, Vupputuri S. Effect of statins on risk of coronary disease. A meta-analysis of randomized controlled trials. JAMA. 1999;282: 2340–2346.
37. Strom BL. Overview of automated databases in pharmacoepidemiology. In: Strom BL, ed. Pharmacoepidemiology. 3rd ed. Chichester, UK: Wiley, 2000:219–222.
38. Murphy MJ, Wei L, Watson AD, MacDonald TM. ‘Real-life’ reduction in cholesterol with statins, 1993 to 2002. Br J Clin-Pharmacol. 2008;65:587–592.
39. Corrao G, Zambon A, Nicotra F, et al. Persistence with oral and transdermal hormone replacement therapy and hospitalisation for cardiovascular outcomes. Maturitas. 2007;57:315–324.
40. Copeland KT, Checkoway H, Mc-Michael AJ, Holbrook RH. Bias due to misclassification in the estimation of relative risk. Am J Epidemiol. 1977;105:488–495.
41. Lee JK, Grace KA, Foster TG, et al. How should we measure medication adherence in clinical trials and practice? Ther Clin Risk Manag. 2007; 3:685–690.
42. Steiner JF, Prochazka AV. The assessment of refill compliance using pharmacy records: Methods, validity, and applications. J Clin Epidemiol. 1997;50:105–116.
43. Salas M, Hofman A, Stricker BH. Confounding by indication: An example of variation in the use of epidemiology terminology. Am J Epidemiol. 1999;149:981–983.
44. Stilley CS, Sereika S, Muldoon MF, et al. Psychological and cognitive function: Predictors of adherence with cholesterol lowering treatment. Ann Behav Med. 2004;27:117–124.
45. Osterberg L, Blaschke T. Adherence to medications. N Engl J Med. 2005;353:487–497.
46. Kim YS, Sunwoo S, Lee HR, et al, for the Korea Post-Marketing Surveil- lance Research Group. Determinants of non-compliance with lipidlowering therapy in hyperlipidemic patients. Pharmacoepidemiol Drug Saf. 2002;11:593–600.
47. Parris ES, Lawrence DB, Mohn LA, Long LB. Adherence to statin therapy and LDL cholesterol goal attainment by patients with diabetes and dyslipidemia. Diabetes Care. 2005;28: 595–599.
Address correspondence to: Prof. Giovanni Corrao, Dipartimento di Statistica, Universita degli Studi di Milano-Bicocca, Via Bicocca degli Arcimboldi, 8, Edificio U7, 20126 Milano, Italy. E-mail: firstname.lastname@example.org