Discharge Is a Critical Time to Influence 10-Year Use of Secondary Prevention Therapies for Stroke
Background and Purpose—When optimally managed, patients with stroke are less likely to have further vascular events. We aimed to identify factors associated with optimal use of secondary prevention therapies in long-term survivors of stroke.
Methods—We carefully documented discharge medications at baseline and self-reported use of medications at annual follow-up in the Northeast Melbourne Stroke Incidence Study (NEMESIS). We defined optimal medication use when patients reported taking (1) antihypertensive agents and (2) statin and antithrombotic agents (ischemic stroke only). Logistic regression was used to assess factors associated with optimal medication use between 2 and 10 years after stroke.
Results—We recruited 1241 patients with stroke. Optimal prescription at discharge from hospital was the most important factor associated with optimal medication use at each time point: odds ratio (OR), 32.2 (95% confidence interval [CI], 13.6–76.1) at 2 years; OR, 7.86 (95% CI, 4.48–13.8) at 5 years (425 of 505 survivors); OR, 6.04 (95% CI, 3.18–11.5) at 7 years (326 of 390 survivors); and OR, 2.62 (95% CI, 1.19–5.77) at 10 years (256 of 293 survivors). Associations were similar in men and women. The association between optimal prescription at discharge and optimal medication use at each time point was greater in those who were not disadvantaged, particularly women.
Conclusions—Prescription of medications at hospital discharge was the strongest predictor of ongoing medication use in survivors of stroke, even at 10 years after stroke. Ensuring that patients with stroke are discharged on optimal medications is likely to improve their long-term management, but further strategies might be required among those who are disadvantaged.
Survivors of stroke are at significant risk of further vascular events; the cumulative risk of myocardial infarction, recurrent stroke, or vascular death is as high as 29% at 5 years.1 Established treatments for preventing further vascular events include long-term use of medications2 such as antihypertensive agents in all survivors of stroke3–5 and, in patients with ischemic stroke, antithrombotic therapy and lipid lowering with HMG-CoA reductase inhibitors.5 Individually, these treatments reduce the risk of additional vascular events by 16% to 45%, but combining multiple treatments could reduce the relative risk of more vascular events by as much as 80%.6 Despite this, treatment rates after stroke remain low worldwide.7
Recognition that in-hospital initiation of appropriate medications reduces the likelihood of further vascular events8 has led to the recommendation that therapy should be initiated before discharge from hospital.9 Initiation of therapy can be increased through comprehensive prevention programs.10–12 However, despite evidence for improved treatment rates at 1 year,13 there is little evidence that this improved treatment continues in the long term.
The population-based Northeast Melbourne Stroke Incidence Study (NEMESIS) provided a unique opportunity to assess long-term use of medications. We aimed to identify factors associated with use of medications for secondary prevention in long-term survivors of stroke.
We used baseline and annual follow-up data from NEMESIS. Details of case ascertainment, using hospital and nonhospital sources, have been reported previously.14 Patients were included if they resided in a defined geographic region (population 306 631) in Melbourne, Australia, and if their stroke was medically diagnosed15 <28 days and occurred between May 1, 1997, and April 30, 1999. An expert panel reviewed all potential events before eventual inclusion. Patients with a subarachnoid hemorrhage were not followed up. For additional details, see the online-only Data Supplement.
Research nurses collected baseline demographic, clinical, and medical history data on all patients using medical records.
Survivors were asked to participate in a face-to-face interview 2, 3, 4, 5, 7, and 10 years after their stroke and in a telephone interview 6, 8, and 9 years after stroke. Proxies were used for severely dysphasic or cognitively impaired participants.
Country of birth was used to classify patients as Australian born or other. Body mass index (kg/m2) was calculated using self-reported height and weight. Participants were classified as living in an institution according to their residence at the time of stroke. Level of disadvantage for each participant was defined according to the Index of Relative Socioeconomic Disadvantage for the location of residence at the time of stroke.14 Those living below the median level of socioeconomic status were categorized as disadvantaged.
Ischemic stroke and intracerebral hemorrhage were confirmed by either cerebral imaging or autopsy. An undetermined stroke was categorized when a patient had no imaging <28 days of symptom onset, and no autopsy was performed.
Severity of stroke was assessed using the National Institute of Health Stroke Scale (NIHSS) score16 and, where possible, was completed <7 days of stroke.
Never-smokers were those who never smoked in their lifetime; current smokers were smokers at the time of their stroke; and exsmokers were neither current smokers nor never-smokers.
Regular general practitioner visits were defined from self-reported regularity of visits (≥1 time every 2 months).
Categorization of Medication Use
Medications prescribed at discharge from hospital were recorded from medical records. At each follow-up interview, patients were asked about all medications currently being taken. Usually patients showed their medications to the research nurse or provided a prescription list. Participants were considered to be taking an antihypertensive agent when they reported taking any diuretic, α- or β-blockers, calcium channel antagonists, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, vasodilators, or α-methyldopa for any reason. Patients were classified according to whether or not they were taking HMG-CoA reductase inhibitors (statins). Antithrombotic agents included antiplatelet agents (aspirin, ticlopidine, dipyridamole, clopidogrel, and aspirin–dipyridamole combinations) and anticoagulants (heparin and heparin salts, dalteparin, and warfarin).
Patients were categorized according to whether their treatment accorded with recent pharmacological recommendations.5 Optimal medication use was defined when patients reported taking (1) an antihypertensive agent (any stroke) and (2) a statin and an antithrombotic agent (ischemic stroke only).
Mantel–Haenszel χ2 tests were used to assess differences in the proportion of patients taking the recommended pharmacological agents at discharge and 10 years poststroke.
Univariable logistic regression was used to identify factors associated with medication use at each follow-up period. The purposeful selection method, as described by Hosmer et al,17 was used to build a parsimonious multivariable model. Those factors identified in univariable analyses as potentially significant (P<0.20) were fitted with serial removal of the least contributing variables until all variables in the model had a P value ≤0.10. In accordance with standard procedures, age, sex, and socioeconomic variables were forced in the model irrespective of whether their P values were ≤0.10. Each excluded variable was then entered separately into the model to determine its contribution to the final model. All possible 2-way interactions were examined during model construction, and a likelihood ratio test was used to assess statistical significance. Because of the large number of potential interactions, and hence the possibility of chance findings, interactions are only reported if the P values were significant (P<0.05) in ≥3 of the 9 time points assessed. People with missing data for any of the variables that comprised part of the multivariable models were excluded.
All analyses were undertaken using Stata version 11.2 (StataCorp) and all P values are 2-sided.
The study was approved by ethics committee at each institution. Informed consent was obtained before any interview was conducted.
In total, 1241 patients with ischemic stroke, intracerebral hemorrhage, and undetermined stroke were recruited (Figure 1). In the 952 patients who survived to 28 days, the mean age was 74.6 years (SD, 13.3), 52.7% were women, 14% were discharged on optimal medications, and 59% were discharged on an antihypertensive agent (Table I in the online-only Data Supplement). Among those with an ischemic stroke, 14% were discharged on a statin and 86% on an antithrombotic agent (Figure I in the online-only Data Supplement). Not all participants were interviewed at each time period because of either gaps in study funding or refusal to participate. Response rates ranged from 72% to 92%.
Use of antihypertensive agents in 10-year survivors was higher at 10 years (74%; 190 of 256) than at hospital discharge (58%; 148 of 256; P<0.001; Figure I in the online-only Data Supplement). In those with an ischemic stroke, use of statins rose from 21% (44 of 212 discharges) to 48% at 10 years (102 of 212; P<0.001), wheres use of antithrombotic agents declined (92% at discharge, 78% at 10 years; P<0.001). Forty-six 10-year survivors (18%) were discharged on optimal medications according to the current recommendations, whereas 88 (34%) used these agents at 10 years poststroke (P<0.001).
Factors independently associated with optimal use of medications 10 years after stroke included age 65 to 74 years (OR, 2.48; 95% CI, 0.93–6.63) and 75 to 84 years (OR, 2.22; 95% CI, 0.85–5.81), history of hypertension at stroke onset (OR, 1.78; 95% CI, 0.93–3.40), greater body mass index (OR, 1.08 per kg/m2; 95% CI, 1.02–1.15), regularly visiting a general practitioner (OR, 1.85; 95% CI, 0.93–3.67), and optimal prescription at hospital discharge (OR, 2.62; 95% CI, 1.19–5.77; see Table).
In multivariable logistic regression, the factor most strongly associated with optimal medication use from 2 to 10 years after stroke was optimal prescription at hospital discharge (see Table). The association between optimal prescription at discharge and optimal prescription at follow-up was greater in those not disadvantaged than those disadvantaged, and this seemed to be limited to women (Table II in the online-only Data Supplement). There also seemed to be an interaction between smoking status and treatment at discharge on optimal use of medications (Table III in the online-only Data Supplement).
In analyses adjusted for only age, sex, and disadvantage (Figure 2; Table IV in the online-only Data Supplement), the association between prescription at discharge and continuing medication use declined over time (for optimal use, antihypertensive agents, and statins). The patterns observed in adjusted analyses were mirrored in univariable analyses (Table V in the online-only Data Supplement) and were similar in men and women (Table VI in the online-only Data Supplement).
Our major finding is that prescription of optimal medications at hospital discharge was the factor most strongly associated with optimal medication use from 2 to 10 years after stroke, being ≈32-fold at 2 years after stroke. This association was particularly strong in women who were not disadvantaged. These findings were robust, being consistent across univariable analyses, in multivariable analyses adjusting for age, sex, and disadvantage, and in fully adjusted models. Remarkably, the association between prescription of these agents at discharge from hospital and subsequent use remained evident 10 years after stroke. Other authors have shown, using univariable analysis only, that 3-month persistence of medication use was associated with persistence of medication use at 12 months.18 However, there has been no previous analysis of factors associated with continued persistence beyond 2 years poststroke.18,19 Our findings demonstrate, for the first time to our knowledge, the critical impact of good hospital management on long-term medication use in survivors of stroke.
Unsurprisingly, the strength of the association between prescription of medications at discharge and medication use seemed to decline with each successive year, although there was some apparent variability, largely attributable to fewer observations in some years (eg, year 6). Despite this decline, the relationship was still strong at 10 years after stroke.
It is encouraging that use of antihypertensive agents and statins increased over time. This contrasts with previous findings of declining use of medications over time.19,20 However, in many previous studies, only patients who were prescribed secondary prevention agents at hospital discharge were followed up,19 thereby eliminating those who were newly prescribed these agents. Furthermore, the publication of landmark trials of secondary prevention in stroke using antihypertensive and lipid-lowering therapies in the intervening years3,21 has likely led to an uptake in treatment.
Although other authors have found improvements in medication use <12 months of discharge,13,22 sustained improvement in use >2 years has not been previously shown. Indeed, the continued improvement in use of antihypertensive agents and statins for such an extended period, to 10 years, has not been previously observed, except in patients after cardiopulmonary bypass surgery.23 Collectively, these observations suggest that local medical practitioners have heeded the results of the major clinical trials on secondary stroke prevention.
Despite improvement in medication use over time, there were still large numbers of patients not taking these agents. A range of factors, including hypotension, off-target side effects, and adverse drug reactions, may have contributed to the relatively low (70%) use of antihypertensive therapy at 10 years. However, it is unlikely that 30% would be ineligible for treatment. The fact that antihypertensive therapy has been associated with up to a 25% reduced rate of recurrent stroke24 provides strong impetus to ensure that patients are treated optimally.
Our findings are particularly important given the concern raised by the results of the Scandinavian Candesartan Acute Stroke Trial (SCAST) in which blood pressure–lowering agents were found to be harmful when administered for a 7-day period during the acute phase of stroke.25 If clinicians are not prescribing blood pressure–lowering drugs within the first week of stroke, some patients may be discharged before therapy could be initiated. This would be a lost opportunity to initiate and potentially influence long-term therapy. Therefore, there is a need for follow-up appointments to be made closer to the time of discharge, thereby ensuring that patients are prescribed appropriate medications early after stroke.
Two factors might explain why many patients (51.9%) with ischemic stroke were not treated with statins in this cohort. First, before 2006, statins were not covered by the Australian Pharmaceutical Benefits Scheme for patients at risk of a major cardiovascular event, unless the patient’s blood cholesterol concentration met a certain threshold.26 Therefore, a large proportion of patients were, until then, required to pay the full price for their statin. Second, the groundbreaking research demonstrating the benefits of statins in secondary prevention of stroke and transient ischaemic attack was published at a time when our participants were between 7 and 9 years after stroke.21
In contrast to Lummis et al,27 we found that current smokers at the time of stroke were more often taking optimal medications compared with never-smokers, although the findings of the interaction analyses did not always support this. We speculate that local medical practitioners might have adopted the absolute risk approach to patient management and so are more likely to prescribe secondary prevention therapies in those with multiple risk factors than in those with fewer risk factors.
Although great care was taken in the design of this study, there are some limitations. First, the use of self-report to elicit information on medication use, height, and weight may have introduced some bias. In particular, the use of medications was not evaluated independently by pill counts. However, given the reported excellent agreement between self-report and pharmaceutical claims data, at least for antihypertensive agents (>0.70), this bias was likely to be minimal.28 In addition, nurses were usually shown medications patients were taking, although this would not help if there was a disparity between prescription and actual use by the patient. Furthermore, prescription of medication or its use cannot help determine whether control of risk factors was achieved. It is also likely that some patients among those not interviewed have been less compliant, and so this may have resulted in an overestimate of overall medication use. In addition, some patients may have contraindications to medications, but because we did not collect such information, we were unable to exclude these patients from the analyses. Contraindications to secondary prevention therapies have been reported in 3% to 13% of patients.10 This is significantly lower than the proportion of patients in our study not prescribed medications at discharge (41% of eligible patients were not prescribed an antihypertensive and 86% who were not prescribed statins). Thus, it is unlikely that the inclusion of these patients greatly affected the interpretation of our findings. In fact, the inclusion of patients with contraindication to treatment has likely resulted in an underestimation of effect because these patients could only contribute to the reference category. Finally, it would have been useful to separately analyze the impact of antiplatelet and anticoagulant use. However, as anticoagulant use is recommended to those with ischemic stroke and atrial fibrillation, and we did not obtain information on atrial fibrillation at follow-up, this could have led to inaccurate classification of optimal medication use within these groups.
There are several strengths to the study. First, this investigation was undertaken within the framework of an incidence study of stroke, and so the findings are relevant to a generalizable patient cohort. A unique aspect of our data set is the availability of robust data from 2, 3, 4, 5, 6, 7, 8, 9, and 10 years after stroke, enabling the identification of a major factor in best practice management. Another major strength is our careful assessment of medication use.
In conclusion, we provide compelling evidence of the importance of initiating secondary prevention therapies before patients leave hospital and its potential to influence long-term use of medications. As shown by the Preventing Recurrence of Thromboembolic Events through Coordinated Treatment (PROTECT) program, it is possible to improve appropriate prescription at discharge from hospital.10 Therefore, educating both medical staff and patients about the importance of secondary prevention therapies will likely result in improved uptake of these life-saving therapies. Because there is evidence that treatment at hospital discharge has a major impact on long-term use of secondary prevention medications, the improved prescription of optimal medications at discharge is likely to have major long-term benefits, including a reduction in further vascular events after stroke.
The contribution of the following research nurses is acknowledged: Dennis Young, Sue Mosley, and Mary Staios. Li Chun Quang provided assistance with data management. The National Death Registry was used to follow up deaths in the cohort.
Sources of Funding
The Northeast Melbourne Stroke Incidence Study was supported by grants from the National Health and Medical Research Council (154600, 307900, 526601), VicHealth, the Foundation for High Blood Pressure Research, and the National Stroke Foundation. J. Kim was supported by a National Heart Foundation of Australia (co-funded with National Stroke Foundation) postgraduate scholarship (PP 10M 5505). Dr Thrift was supported by a Senior Research Fellowship from the NHMRC (1042600) and Dr Gall was supported by a National Heart Foundation of Australia Public Health Post Doctoral Fellowship (PH 11H 6047).
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.113.003368/-/DC1.
- Received August 30, 2013.
- Revision received October 21, 2013.
- Accepted November 6, 2013.
- © 2013 American Heart Association, Inc.
- Dhamoon MS,
- Tai W,
- Boden-Albala B,
- Rundek T,
- Paik MC,
- Sacco RL,
- et al
- Furie KL,
- Kasner SE,
- Adams RJ,
- Albers GW,
- Bush RL,
- Fagan SC,
- et al
- 5.↵National Stroke Foundation. Clinical Guidelines for Stroke Management. 2010. http://www.strokefoundation.com.au/clinical-guidelines. Accessed June 2, 2013.
- Hackam DG,
- Spence JD
- Yusuf S,
- Islam S,
- Chow CK,
- Rangarajan S,
- Dagenais G,
- Diaz R,
- et al
- Kirshner HS,
- Biller J,
- Callahan AS III.
- Fonarow GC,
- Reeves MJ,
- Smith EE,
- Saver JL,
- Zhao X,
- Olson DW,
- et al
- Mouradian MS,
- Majumdar SR,
- Senthilselvan A,
- Khan K,
- Shuaib A
- Thrift AG,
- Dewey HM,
- Sturm JW,
- Paul SL,
- Gilligan AK,
- Srikanth VK,
- et al
- Brott T,
- Adams HP Jr.,
- Olinger CP,
- Marler JR,
- Barsan WG,
- Biller J,
- et al
- Hosmer DW Jr.,
- Lemeshow S,
- May S
- Glader EL,
- Sjölander M,
- Eriksson M,
- Lundberg M
- Irving RJ,
- Oram SH,
- Boyd J,
- Rutledge P,
- McRae F,
- Bloomfield P
- Rashid P,
- Leonardi-Bee J,
- Bath P
- 26.↵Rational Assessment of Drugs and Research. Revised PBS criteria for lipid-modifying drugs (October 2006). 2007. http://www.nps.org.au/publications/health-professional/nps-radar/2007/february-2007/pbs-lmd-criteria. Accessed August 28, 2013.