Assessing overall duration of cardiovascular medicines in veterans with established cardiovascular disease

Abstract

Introduction

Similarly to other developed countries, Australia invests heavily in cardiovascular medicines with the aim of improving cardiovascular outcomes in susceptible individuals. In 2006, government expenditure on cardiovascular medicines was 1.8 billion dollars and accounted for 33% of all subsidized medicine use in Australia [1].

Cardiovascular medicines contribute to improvements in survival and morbidity and are recommended in treatment guidelines [2]. However, adherence to these medicines is essential if the outcomes observed in trials are to be realized in practice, with persons with good adherence having been shown to have a lower risk of mortality [3]. The second Australian National Blood Pressure study found that those who indicated that they sometimes forgot to take their medicine were more likely to experience a cardiovascular event or death (hazard ratio: 1.28, 95% confidence interval: 1.09-1.66) [4].

Despite the potential health gains from cardiovascular medicines, studies have reported variable rates of adherence. An Australian study conducted between 2003 and 2006, which restricted analyses to new users of cardiovascular medicines who had filled a minimum of two prescriptions reported that, at 24 months, 75-79% were persistent with angiotensin-converting enzyme (ACE) inhibitors/angiotensin II receptor blockers (ARBs), 53% with β-blockers, and 76% with antiplatelets other than aspirin [5]. Another Australian study assessed the persistence with antihypertensive medications between 2004 and 2006 and found that median persistence times were 1.9-2.2 years for ACE inhibitors/ARBs and 0.6 year for calcium channel blockers (CCBs) [6]. These results could be interpreted as representing significant under use of cardiovascular medicines. However, the analyses were limited to new users, only a single episode of treatment was examined and date of death was missing from the data set. Preventive therapy may be stopped and then restarted, particularly by new users, as adherence with medications is a dynamic process that is affected by multiple factors [7]. Thus, methods used for determining adherence with chronic therapy need to take into account multiple episodes of use [8]. This is particularly relevant in the Australian context, where there is potential funding under the Pharmaceutical Benefits Scheme for combination products that show improvements in adherence [9]. Internationally, there is also a call for methods for accounting for adherence in economic evaluations [10, 11]. Methods that account for multiple exposures over time are needed when chronic therapies are considered.

In this study, we assess cardiovascular medication persistence and adherence in the veteran population with established cardiovascular disease, including all episodes of use and assess overall time without therapy.

Methods

The Department of Veterans’ Affairs (DVA) Pharmacy Claims database contains details of all prescription medicines dispensed to veterans for which DVA pay a subsidy. The data file contains 112 million records for a treatment population of 345 000 veterans. Details include date of supply, product, strength, and quantity. The DVA maintain a client file, which includes data on sex, date of birth, date of death, and family status. Medicines are coded in the dataset according to the World Health Organization anatomical and therapeutic chemical classification [12]. The DVA also maintain private and public hospital data sets that include admission and discharge dates, primary and secondary diagnoses as well as procedures performed.

A retrospective cohort study was undertaken over a 6.5-year period from 1 January 2001 to 30 June 2007. Veterans were included in the study if they were gold card holders (full access to health services), had a hospitalization with a primary diagnosis of acute myocardial infarction (International Classification of Diseases-10 code I21^∗), angina (I20^∗), or ischemic heart disease (I248^∗, I249^∗, I250^∗, I251^∗, I252^∗, I256^∗, I258^∗, I259^∗) between 1 January 2001 and 31 December 2005, and had been dispensed cardiovascular medicines posthospitalization. Only the initial hospitalization within the dataset was included. Cardiovascular medicines included ACE inhibitors and ARBs, plain or combined with diuretics (anatomical and therapeutic chemical codes C09AA, C09BA, C09CA, C09DA), lipid-lowering therapy (C10), CCBs (C08C, C08D), β-blockers (C07), aspirin (B01AC06), and other antiplatelets (B01AC04, B01AC07, B01AC30, B01AC05). Medicines dispensing in the 12 weeks posthospitalization were included, as this was assumed to indicate therapy prescribed at discharge and thus related to the admission diagnosis. The interval for aspirin was extended to 128 days, as aspirin is most commonly dispensed in quantities that last between 3 and 4 months. Length of follow-up was calculated as time since the first dispensing posthospitalization until study end or death. The date of hospital admission was used as the first dispensing date for those with a prior dispensing still current at admission.

In the absence of dosage information in the dataset, the duration of each prescription was defined as the number of days in which 75% of people returned for a prescription refill. For all new subsidized medicines, pack sizes are generally sufficient for a duration of 1 month. Calculated refill rates were slightly longer; 35 days for ACE inhibitors/ARBs, 36 days for CCBs, 35 days for lipid-lowering therapy, 34 days for clopidogrel, 41 days for dipyridamole, and 50 days for β-blockers. Aspirin is supplied in pack sizes sufficient for 3 months supply and 128 days was the refill rate calculated. The treatment was considered to be discontinued if the gap between two dispensings was more than three times the duration of use. This definition was chosen to avoid overestimating the cessation tally due to Australia's safety net system. The system results in a phenomenon described as ‘stockpiling’, where increased dispensings are observed in November and December each year with a subsequent decrease in January and February [13]. Each episode of use was calculated as the number of days from the date of the first prescription to the date coinciding with the last prescription and its duration of use before medicine being considered as ceased. The overall treatment duration was calculated as the sum of treatment days for all episodes. Time without treatment was calculated as the sum of the number of days of all gaps in treatment. Adherence was assessed as the proportion of participants who had sufficient medicine dispensed, as measured by pack size to cover from 80 to 120% of the treatment duration.

For each of the medication classes, Kaplan-Meier analyses were undertaken for the first treatment episode (time to first cessation event), overall treatment duration, and time without treatment. Participants were censored if an episode of use was terminated by death or by study end where an episode crossed the study end date. Log-rank tests were used to compare differences between medication classes. All analyses were performed using SAS v9.1. (SAS Institute Inc., Cary, North Carolina, USA). This research was approved by the University of South Australia Human Research Ethics Committee and the Department of Veterans’ Affairs Ethics Committee.

Results

There were 9635 veterans of which 74.3% were men. The mean age at study entry was 77.2 (SD 7.9) years for men and 78.1 (SD 5.9) years for women. On average, they were dispensed four of the cardiovascular medicines studied. Sixty-nine percent of veterans were dispensed ACE inhibitors/ARBs, 80% lipid-lowering therapy, 44% CCBs, 57% β-blockers, and 73% antiplatelets. The majority of veterans were existing users of therapy, with new users accounting for 16% of participants receiving ACE inhibitors/ARBs, 20% of participants receiving lipid-lowering therapy, 16% of participants receiving CCBs, 22% of participants receiving β-blockers, and 25% of participants receiving antiplatelets.

Median duration of treatment for first episode post-hospitalization, overall duration, time without treatment, and the proportion adherent are summarized in Table 1. Duration analyses for first episode of use, overall duration, and time without treatment are shown in Figs 1, 2 and 3. The median duration of overall treatment episodes was the longest for the lipid-lowering therapy, followed by ACE inhibitors/ARBs, with shorter rates observed for asprin, other antiplatelets, and CCBs (P < 0.0001). Lipid-lowering therapy and ACE inhibitors/ARBs also had the shortest times without treatment. The biggest drop in duration occurred just after initiation of therapy, with an average 8.2% (SD 4.1%) of participants across the medicine classes not receiving a second prescription.

Fig. 1

Duration times of first treatment episodes of cardiovascular medicines posthospitalization. ACEIs, angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers; AP, antiplatelets; BBs, β-blockers; CCBs, calcium channel blockers; LL, lipid-lowering therapy.

Open in new tabDownload slide

Fig. 2

Duration times of overall treatment of cardiovascular medicines posthospitalization. ACEIs, angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers; AP, antiplatelets; BBs, β-blockers; CCBs, calcium channel blockers; LL, lipid-lowering therapy.

Open in new tabDownload slide

Fig. 3

Time without treatment of cardiovascular medicines posthospitalization. ACEIs, angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers; AP, antiplatelets; BBs, β-blockers; CCBs, calcium channel blockers; LL, lipid-lowering therapy.

Open in new tabDownload slide

Discussion

This study found high rates of persistence and adherence with lipid-lowering therapies, ACE inhibitors/ARBs, and antiplatelets in veterans with established cardiovascular disease, with median durations over 5 years, median times without therapy 4.5 months or less, and adherence during treatment 75% for ACE inhibitors/ARBs and 84% for nonaspirin antiplatelets and lipid-lowering therapy. Lower rates of persistence were observed with CCBs and β-blockers. However, disease progression may have contributed to the latter effect. Although diagnosis at the time of hospitalization was available, we had no means of accounting for disease progression or the development of comorbidity or adverse reactions over time, and it may be that some of the observed cessation for these medicines was appropriate.

Our study suggests a longer duration of therapy than previously reported and highlights the importance of including multiple episodes of use in persistence and adherence analyses. The durations extended 1.5-2 fold when multiple episodes of use occur. This suggests that people commonly reinitiate therapy and persistence studies that only account for first episode of use may underestimate true population persistence. Although there are few comparable studies, an Italian study that assessed persistence in new and existing users of chronic cardiovascular drug treatments [14] found that duration of use among existing users at 1 year was almost complete, whereas over 40% of new users withdrew in 1 year.

The impact of disease severity seems uncertain. A Canadian study examining persistence and adherence rates for statins in persons who had an acute myocardial infarction and were dispensed a statin within 30 days of discharge found a 1-year duration rate of 90% [15]. Another study examined adherence to recommended cardiovascular therapies 6 months after discharge for acute myocardial infarction or unstable angina and found adherence rates between 80 and 92% 6 months post-discharge [16]. By comparison, one study assessing persistence to ACE inhibitors in the primary and secondary prevention populations found little difference between the populations, and 1-year persistence rates less than 80% [17]. Another assessing statin use, reported slightly longer persistence rates in the secondary prevention population, however, median persistence rates were less than 1 year [18]. Both studies only considered single episodes of use, again reflecting the potential underestimate of duration.

People with higher chronic disease scores have shown greater adherence rates to statins, β-blockers, and ACE inhibitors [19]. Multiple medicines have also been associated with greater rates of adherence [14, 20], as have higher numbers of tablets per day [21]. Notably, in our study, participants were on an average of four cardiovascular medicines, which supports this association by demonstrating that multiple medicines do not necessarily lead to poor adherence or duration.

It is unknown whether the veteran population is likely to be more compliant with medicines than the general Australian population. The veteran population have slightly more general practice visits (rate ratio 1.17; P < 0.05) and hospitalizations (rate ratio 1.21; P < 0.05) per year than other Australians aged 40 years and above [22]. Veterans with no service-related disability have similar levels of use [22]. Similar numbers of prescriptions per general practitioner visit are observed between the veteran population and the Australian population; however, because of the higher rate of general practitioner visits, veterans receive slightly more prescriptions annually than other Australians (rate ratio 1.13; P < 0.05) [22]. This suggests our study results are likely to be similar to the Australian population.

Our results have implications for economic evaluations that include compliance assessments [11]. In Australia, changes to the Pharmaceutical Benefits Scheme in 2007 include possibility for products that demonstrate ‘significant improvement in patient compliance’ to receive higher benefits [23]. However, there is no consensus on the best methods to assess adherence and what could constitute a ‘significant’ improvement. The methods used to determine adherence and characteristics of the population studied in terms of disease severity, number of chronic diseases, number of medicines, and whether new or existing users are included, all impact on duration and adherence rates, and could impact on cost-effectiveness assessments [19–21, 24]. The method used in this study could assist in economic models, enabling gaps in treatment and durations across subsequent episodes to be more accurately modeled when considering transition probabilities in economic models [11].

The major limitation of our study is lack of dosage information in the dataset, hence dosage duration was inferred from refill rates. The refill rates were generally, of slightly longer duration than the pack sizes supplied, suggesting refill rates are an appropriate proxy where dosage information is missing. This method has been employed by other researchers [25], but more certain measures of adherence would have been possible if the dosage information had been included. For this reason, the adherence measure for aspirin should be interpreted cautiously. The large pack sizes for aspirin as well as the fact that it can be bought without prescription may have contributed to the result. Despite this, the method should have no impact on the comparative results across medicine classes for duration as the same method was applied each time and for most classes, similar refill rates applied. We were also limited by diagnostic data only available at study entry. These types of studies would be strengthened if reason for cessation was available. Although we were able to attribute cessation due to death, inappropriate versus appropriate cessation for clinical reasons could not be determined. Finally, the study is based on dispensing data, not consumption data. It is assumed that the two are likely to be similar, but not known.

Conclusion

The analysis presented in this study of Australian veterans with established cardiovascular disease suggests that a significant proportion persist with their cardiovascular therapy over several years with high rates of adherence, particularly for lipid-lowering therapy, ACE inhibitors/ARBs, and antiplatelets. However, 15 to 25% do not receive sufficient supply of cardiovascular medication to be considered fully adherent during therapy. Further understanding of this at-risk group would assist appropriately targeted quality use of medicine interventions.

Acknowledgements

This work was supported by funding from a National Health and Medical Research Council/Australian Research Council Ageing Well Ageing Productively (AWAP) Program grant. Authors acknowledge the contribution of all the Chief Investigators who assisted with the preparation of this manuscript: Assistant Professor Robyn McDermott, Professor Adrian Esterman and Professor Mary Luszcz.

Conflicts of interest: none.

References