Inequalities in medication adherence to statin treatment after stroke: A nationwide observational study

Abstract

Introduction

The objective was to investigate differences in adherence to statins after stroke based on age, sex, socioeconomic status and country of birth.

Patients and methods

Patients with ischemic stroke in 2009–2010 were included from the Swedish stroke register. Adherence to statin treatment was measured over two years as proportion of days covered with 80% as cut-off for adherence. Income, education, and country of birth were obtained from official registers. Factors associated with adherence were controlled for in multivariable logistic regression.

Results

Of 15,192 included patients, 73.9% had an adherence rate ≥80%. The oldest (85+ years) and youngest (18–54 years) had the lowest adherence, and a smaller proportion of women were adherent (odds ratio (OR) 0.84; 95% confidence interval (CI) 0.77–0.92). Adherence was less common in patients born in Nordic countries (OR 0.82; 95% CI (0.68–0.97), Europe (OR 0.78; 95% CI 0.65–0.93), and in non-European countries (OR 0.65; 95% CI 0.50–0.84) compared to Sweden-born. Patients with university education were to a lower extent adherent compared to patients with primary school education (OR 0.81; 95% CI 0.72–0.91). There was no association between adherence and income.

Discussion

The study was based on individual level real-life data with national coverage. Adherence was estimated from data on filled prescriptions, but filled prescription does not mean that drugs are used as intended.

Conclusion

Adherence to statin treatment over two years was suboptimal, and adherence was less common among women, patients born outside of Sweden and patients with university education.

Introduction

Statins have been shown to be effective in preventing ischemic stroke.^1,2 A meta-analysis on data from patients with a history of cerebrovascular disease found a relative risk (RR) of 0.88 (95% confidence interval (CI) 0.78–0.99) for total stroke and 0.80 (95% CI 0.70–0.92) for ischemic stroke when comparing statin treatment with placebo.³ Statins need to be used on a regular basis, and adherence to statin treatment has been shown to decrease over time.⁴ Studies have also shown that higher adherence to statin treatment is associated with lower risk of cardiovascular events.^5,6

The risk of stroke as well as stroke mortality increases with decreasing socioeconomic status,^7,8 and many risk factors have also shown an inverse relationship with socioeconomic status.⁹ We have previously shown that prescribing of statins to Swedish stroke patients is higher in groups with higher socioeconomic status,¹⁰ i.e. groups with lower stroke risk. In Sweden, statins are prescription-only drugs and equity in prescribing is important, but when it comes to medication adherence, the key question is how the patients use their prescribed drugs.

Differences in medication adherence based on socioeconomic status and country of birth have been found previously. A Danish study found that adherence to statin treatment decreased with decreasing income,¹¹ and a study from Israel found poorer persistence to statin treatment among patients with low socioeconomic status and among new immigrants.¹² However, a review on adherence to medications concluded: ‘Race, sex, and socioeconomic status have not been consistently associated with levels of adherence’.¹³ The objective of this study is to investigate differences in adherence to statin treatment after ischemic stroke between patient groups based on age, sex, socioeconomic status, and country of birth in a Swedish population.

Patients and methods

A historical cohort was constructed from national registers and databases, and data were linked through the Swedish personal identification number. Included registers/databases were the national register for stroke care (Riksstroke), the Swedish prescribed drug register (SPDR), the national inpatient register (IPR) and the Longitudinal integration database for health insurance and labour market studies (LISA).^14–17

Riksstroke contains information about stroke events. Data are collected from before the event, from acute care and from follow-up questionnaires. All Swedish hospitals that treat acute stroke report to the register, and the estimated coverage for 2010 was 88.1% according to the Riksstroke annual report. A study has shown high consistency between Riksstroke data and medical records.¹⁸

The SPDR contains individualised information on all prescription drugs sold in all Swedish pharmacies, and coverage is over 99%. The register includes information about item and amount dispensed, date of dispensing and cost. Drugs are classified according to the Anatomical Therapeutic Chemical (ATC) Classification System.

The number of days in inpatient care and recurrent strokes/MI during follow-up was calculated from the IPR, which has information about all admissions to inpatient care in Sweden.

Individual data on highest level of education, income and country of birth were included from the LISA database at Statistics Sweden. LISA covers the total population from 16 years of age and is updated yearly.

Study population

Included patients had had an ischemic stroke and were discharged alive from hospital between 1 January 2009 and 31 December 2010. All patients were, according to Riksstroke, prescribed a statin at discharge. Both new and previous statin users were included. Patients with a first-ever stroke and recurrent strokes were included, but each individual was only included at the first event in the study period. Patients with multi-dose drug dispensing were not included because the “filling of prescriptions” is automatic and thus refill adherence was not considered applicable for this group.

Variables

The main outcome was medication adherence to secondary preventive treatment after stroke measured as the proportion of days covered (PDC) with 80% as the cut-off for adherence.¹⁹ Adherence was calculated for ATC codes C10AA or C10B. All statins, irrespective of substance or dose were included. Adherence was measured over two years from discharge, and only patients who were still alive two years after discharge were included.

All statins that are sold in Swedish pharmacies are included in the reimbursement system if they are registered in the SPDR. Thus, statins bought somewhere else would not have been included. PDC was calculated as the number of filled prescriptions multiplied by 100 and divided by the number of days during the follow-up that the patient was not hospitalised. Assumptions were made that the daily dose was 1 tablet/capsule per day, and that each dispensing would last for 100 days. Dosage instructions in the SPDR are only available as free text. A random sample of 1000 statin prescriptions was chosen for a manual check of the actual daily dose, and the results showed that fewer than 3% had a dose other than 1 tablet/capsule a day. Checking the size of packages showed that 95.4% of all packages contained 98 or 100 doses (68.5% were packages of 100).

Patients’ socioeconomic status was indicated by the highest level of education and family disposable income (Spearman’s rho = 0.26 (p < 0.001) for the correlation between income and education). Education was classified as primary school (≤9 years in compulsory school), secondary school (the upper levels of comprehensive school), and university. Disposable income was individualised from family income in the year before the stroke and adjusted to Consumer Price Index for 2012. Income was categorised as low, medium, or high (tertiles). Patients’ country of birth was grouped into Sweden, Nordic countries (except Sweden), Europe (except Nordic countries) and other countries.

Variables included from Riksstroke were whether living alone, history of stroke, dependent in activities of daily living (ADL), fully conscious at admission to hospital, treated in a stroke unit, statin use before stroke, history of diabetes, antihypertensive medication, atrial fibrillation and smoking, and these variables were all coded as Yes or No. Information about recurrent stroke and/or MI during follow-up was included from the IPR (main diagnosis >28 days after index stroke) and coded as Yes or No. Because adherence is based on number of filled prescriptions, the mean cost per filled prescription had to be calculated. The out-of-pocket cost is available in the SPDR, and patients’ mean out-of-pocket cost per dispensing was calculated.

This study is part of the EqualStroke project that has been approved by the Ethical Review Board in Umeå (2012-321-31M and 2013-176-32M).

Statistical analysis

The distribution of number of filled prescriptions per patient during the follow-up is presented in a figure. The frequencies and proportions of adherent patients were calculated for different patient groups. Factors possibly associated with adherence were controlled for in multivariable logistic regression models, including age, sex, country of birth, highest level of education, disposable income, whether living alone, history of stroke, dependency in ADL, fully conscious at admission to hospital, whether treated in a stroke unit, if statins were used before the stroke, history of diabetes, use of antihypertensive medication, atrial fibrillation, smoking, recurrent stroke and/or MI during follow-up and patient mean out-of-pocket costs per dispensing. The effects of these factors on adherence are presented as odds ratios (OR) with 95% CIs. A two-way interaction term (income-by-mean cost) was included in a multivariable model to test if a possible relation between mean cost and adherence varied with income. The analyses were performed with SPSS 22.0.

Results

In total, 15,192 patients were discharged from a Swedish hospital in 2009–2010 after an ischemic stroke, were ≥18 years of age and alive two years after discharge, were prescribed statins at discharge and did not use multi-dose drug dispensing. Of these patients, 41.1% were women, 37.0% were over 75 years old, 85.2% were born in Sweden and for 39.5% primary school was the highest level of education. The patients filled a mean of 6.47 (standard deviation = 2.71) prescriptions during the follow-up. The distribution of the number of filled prescriptions is presented in Figure 1.

Figure 1.

The distribution of number of filled prescriptions per patient during two-year follow-up.

Over the two-year follow-up, 11,228 patients (73.9%) were classified as adherent to statin treatment. The proportion of adherent patients was significantly lower in the oldest age group (85+ years) and significantly higher in ages 55–64 and 65–74 years compared to the youngest (18–54 years) (Table 1). A smaller proportion of women than men were classified as adherent (OR 0.84; 95% CI 0.77–0.92).

Table 1.

Number of valid observations, frequencies, and proportions for statin adherence presented in subgroups of ischemic stroke patients with a statin prescribed at discharge from hospital.

	Valid observations	Adherence (PDC > 80%) frequency	Proportion (%)	Age-adjusted odds ratio (95% CI)	Multiple logistic regressiona odds ratio (95% CI)
Sex
Men	8950	6822	76.2	1	1
Women	6242	4410	70.7	0.78 (0.72–0.84)	0.84 (0.77–0.92)
Age group
18–54	1329	906	68.2		1
55–64	3054	2339	76.6		1.39 (1.18–1.64)
65–74	5188	3995	77.0		1.34 (1.15–1.56)
75–84	4556	3370	74.0		1.03 (0.88–1.21)
85+	1065	622	58.4		0.62 (0.50–0.77)
Country of birth
Sweden	12939	9778	75.6	1	1
Nordic countriesb	857	599	69.9	0.73 (0.62–0.85)	0.82 (0.68–0.97)
Europec	825	569	69.0	0.70 (0.60–0.82)	0.78 (0.65–0.93)
Other countries	394	269	68.3	0.72 (0.58–0.90)	0.65 (0.50–0.84)
Education
Primary school	5996	4534	75.6	1	1
Secondary school	5978	4477	74.9	0.93 (0.85–1.01)	0.91 (0.82–1.00)
University	2821	2064	73.2	0.84 (0.76–0.93)	0.81 (0.72–0.91)
Income
Low	5005	3612	72.2	1	1
Medium	5005	3758	75.1	1.16 (1.06–1.27)	1.08 (0.98–1.20)
High	5005	3845	76.8	1.20 (1.09–1.32)	1.09 (0.98–1.22)
Living alone
No	9841	7478	76.0	1	1
Yes	5296	3712	70.1	0.79 (0.73–0.85)	0.86 (0.79–0.94)
History of stroke
No	12576	9333	74.2	1	1
Yes	2538	1840	72.5	0.92 (0.84–1.02)	0.79 (0.70–0.89)
Recurrent stroke/MI during follow-up
No	13264	9848	74.2	1	1
Yes	1928	1384	71.8	0.90 (0.81–1.01)	0.82 (0.72–0.92)
Dependent in ADL
No	14911	11046	74.1	1	1
Yes	246	161	65.4	0.71 (0.54–0.93)	0.94 (0.65–1.35)
Fully conscious at admission
No	497	355	71.4	1	1
Yes	14641	10836	74.0	1.16 (0.95–1.41)	1.02 (0.80–1.30)
Stroke unit
No	1178	857	72.8	1	1
Yes	14014	10375	74.0	1.05 (0.92–1.21)	1.07 (0.92–1.26)
Statin use before stroke
No	9659	6905	71.5	1	1
Yes	5460	4280	78.4	1.46 (1.35–1.58)	1.75 (1.58–1.93)
History of diabetes
No	12229	8985	73.5	1	1
Yes	2911	2207	75.8	1.10 (1.00–1.21)	1.03 (0.92–1.15)
Antihypertensive medication
No	6509	4756	73.1	1	1
Yes	8609	6420	74.6	1.08 (1.01–1.17)	0.94 (0.86–1.03)
Atrial fibrillation
No	12533	9294	74.2	1	1
Yes	2560	1860	72.7	0.97 (0.88–1.07)	0.92 (0.82–1.03)
Smoking
No	11504	8600	74.8	1	1
Yes	2893	2073	71.7	0.78 (0.71–0.86)	0.80 (0.72–0.89)
Mean out-of-pocket cost per dispensing
4th quartile (highest cost)	3656	2513	68.7	1	1
3rd quartile	3659	2873	78.5	1.72 (1.54–1.91)	1.79 (1.60–2.00)
2nd quartile	3650	3034	83.1	2.35 (2.10–2.63)	2.38 (2.11–2.69)
1st quartile (lowest cost)	3658	2812	76.9	1.58 (1.42–1.76)	1.58 (1.41–1.77)

CI: confidence interval; MI: myocardial infarction; ADL: activities of daily living.

^aAdjusted for all variables in Table 1.

^bExcept Sweden.

^cExcept the Nordic countries.

Adherence was less common in patients born in the Nordic countries (OR 0.82; 95% CI 0.68–0.97), in Europe (OR 0.78; 95% CI 0.65–0.93), and in non-European countries (OR 0.65; 95% CI 0.50–0.84) compared to Swedish-born patients. Patients with university education were less adherent compared to patients with primary school education (OR 0.81; 95% CI 0.72–0.91). There were larger proportions of adherent patients among patients with higher income, but the associations between income and adherence were not statistically significant in the multivariable model (Table 1). The two-way interaction term (income-by-mean cost) that was included in the multivariable model was also not significant (p = 0.44).

Patients who had been using statins before the stroke were more likely to be adherent compared with new users (OR 1.75; 95% CI 1.58–1.93). Lower patient out-of-pocket mean cost per dispensing was also associated with higher adherence (Table 1). Factors associated with lower adherence were living alone, a history of previous stroke, recurrent stroke or MI during follow-up and smoking (Table 1).

The variables with most missing cases were smoking (5.2%), education (2.6%), country of birth (1.2%) and income (1.2%).

Discussion

In this study, 73.9% of the included stroke patients were classified as adherent to statin treatment over the two-year follow-up after hospital discharge. Medication adherence was less common among women, in the oldest and youngest age groups, in patients born outside of Sweden, and in patients with high level of education.

The overall adherence level was similar^11,20 or higher²¹ compared with previous studies. Other studies have also found lower adherence to statin treatment in younger patients and in women.^12,21,22 One explanation might be a lower perceived risk of cardiovascular disease in these groups and the belief that such ailments mostly affect men and older people. However, in our sample, adherence was lowest in the oldest age group (85+ years). Questions about the risks and benefits of statin treatment in the oldest age groups might account for this, especially in cases with many other drugs or if experiencing side effects.

The proportion of patients classified as adherent decreased with higher levels of education and increased with higher income, but the only association in the multivariable model was that patients with university education were less adherent compared to patients with primary school education. A Danish study on adherence to statins found that when controlling for both income and education, the OR for PDC <80% was lower in higher income groups and higher among women with higher education.¹¹ This is similar to the results of the present study. Previous research have shown that people with higher education more often want to be involved in decisions about treatment.²³ If they feel not involved or have doubts about treatment, they might to a higher degree end treatment. Another Danish study found poorer refill persistence to statins in low income patients, but better persistence to beta-blockers after MI.²⁴ This adds to the results of a review that concludes that associations between adherence and socioeconomic status have so far been inconsistent.¹³

Studies have found poorer adherence among immigrant patients similar to this study. A study from Denmark found that immigrants from Turkey and Pakistan more often discontinued treatment with beta-blockers but not statins after MI compared with Danish-born patients.²⁵ In a cohort from Israel, persistence to statins was found to be poorer in new immigrants compared to persons who had been resident in Israel for a longer time.¹² This may reflect communication problems and language barriers. Time since immigration was not considered in this analysis, but language and communication are likely a bigger problem in new immigrants.

Adherence was associated with lower mean out-of-pocket cost, a result consistent with other studies.^26,27 In the Swedish Pharmaceutical Benefits Scheme, out-of-pocket costs are based on the price of the drug and on different levels of co-payment based on accumulated out-of-pocket costs. Every citizen is insured irrespective of income. In another study, out-of-pocket costs were shown to be associated with adherence in patients with lower income but not in patients with higher income.²⁸ In this study, there was no significant interaction between adherence, income and out-of-pocket cost.

In randomised trials, statins have proven effective in preventing ischemic stroke,^1,2 but due to e.g. poorer adherence the same effect cannot always be expected in clinical practice. However, adherent patients were in this study less likely to have a follow-up event, and this replicates the results of clinical trials. Stroke is a common condition that affects patients with lower socioeconomic status to a higher extent. It is therefore important to investigate how stroke preventive treatments are prescribed and used by patients. A previous study showed that statins are prescribed less often to low socioeconomic groups.¹⁰ The present study showed that patients’ use of statins did not increase the disparities, at least not for differences based on income and education. In patients with high education, poor adherence may cancel out the previously observed higher prescribing at hospital discharge.

Strengths and limitations

Strengths of this study include the national coverage of registers and databases and the individual level real-life data. The use of register-based data overcomes the problem with patients not remembering what drugs they have used. Included patients were prescribed a statin at discharge from hospital, and this made it possible to include patients who never filled any prescription. The proportion of missing values was very low for most variables.

A filled prescription does not necessarily mean that the drugs are used as intended, but data on filled prescriptions are often used to estimate adherence. Many different methods to calculate refill adherence have been suggested²⁹ and in this study, PDC was calculated for both new and current users. We have not taken into account if a prescription was filled in the months before the index stroke, nor excluded a prescription that was filled just before the end of follow-up. The objective of this study was to compare adherence in different patient groups, and we find it unlikely that the proportion of prescriptions filled just before the stroke or at the very end of follow-up would differ substantially between patient groups.

Adherence has only been calculated for patients with equally long follow-up. The reasons is a possible association between adherence and follow-up time – it is easier to be adherent for a shorter time – and because case-fatality has been found to vary with socioeconomic status.³⁰ Some non-adherent patients could have had fatal stroke or MI, hence excluding patients who die early might introduce bias and affect the generalisability of the results. Patients with multi-dose drug dispensing were not included, and this selection might also affect the generalisability of the results because larger proportions of older and sicker patients are prescribed multi-dose drug dispensing.

Only statins sold in Swedish pharmacies were included, but statins bought somewhere else would not have been subsidised. If patients switched from one statin to another, this could have created an oversupply (wastage) not related to the actual use.

Several variables related to adherence have been controlled for. Others, such as patients’ blood lipid levels and concurrent medications, were not available but could possibly have partially explained adherence to statins. Level of consciousness was used as a proxy for stroke severity.

Conclusion

Adherence to statin treatment over a two-year follow-up was suboptimal in this sample of Swedish stroke patients who were prescribed a statin at discharge from hospital. Differences in adherence were found between patients groups. Adherence was less common among the oldest and youngest patients, in women, in patients born outside of Sweden, and in patients with highest level of education. To strive for optimal and equal stroke care, special attention must be given to reasons for poor adherence in patients groups with lower medication adherence after stroke.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The work was supported by grants from the Swedish Research Council for Health, Working Life and Welfare (2011-0657) and the Swedish Research Council (2011–2395).

Informed consent

Patients are informed about registration in Riksstroke, that data may be used for research purposes, and their rights to deny participation (opt-out consent). Consent is not collected for specific research projects. Data were de-identified prior to being received by the researchers.

Trial registration

Not applicable.

Ethical approval

This study is part of the EqualStroke project that has been approved by the Ethical Review Board in Umeå (2012-321-31M and 2013-176-32M).

Guarantor

MS

Contributorship

All authors are responsible for the study concept and design. MS analysed the data and prepared the manuscript. All authors carried out a critical revision of the manuscript, contributed with comments and approved the final version.