Secondary Stroke Prevention After Ischemic Stroke in Patients with Alzheimer’s Disease and Other Dementia Disorders

Study population, registries, and variables

We performed an open-cohort study of patients diagnosed and registered with dementia who subsequently had a first IS between 2007 and 2014 in Sweden. At the time of dementia diagnosis, patients were registered in SveDem, the Swedish national dementia registry, together with information on dementia type [13]. Occurrence of IS was identified using Riksstroke, the Swedish national registry for acute stroke with coverage for IS > 90% [14]. Additional information on the registries is available at svedem.se and riksstroke.org.

We used data from additional registries; the Swedish Prescribed Drug Register (PDR) contains data on all prescription medication dispensed at Swedish pharmacies since 2005 with ∼100% coverage [15]. Data on comorbidities before stroke were collected from the Swedish National Inpatient Register (NIR) [16] and used to calculate the Charlson Comorbidity Index (CCI) [17]. All in-hospital and specialist clinic diagnoses were coded according to the ICD-10 and were available from 1998 [16]. Data on death in hospital was obtained from Riksstroke, while data on death after hospital discharge until August 2016 was obtained from the Swedish Cause of Death Register [18]. Data on medication was available until 31 December 2015, consequently, to ensure 3-year follow-up for all patients, patients with IS event after 31 December 2012 were excluded from 3-year and patients with IS after 31 December 2013 were excluded from 2- and 3-year follow-up analyses.

Out of 58,154 patients from SveDem diagnosed with dementia between 2007 and 2014, we identified 1410 patients with dementia and IS. These patients were compared with 7,150 non-dementia IS controls from Riksstroke matched by age (±3 years), sex, year of stroke, and geographic region. Patients with hemorrhagic stroke or prior IS were excluded. Included non-dementia controls never had a SveDem registration, were never diagnosed with dementia or confusional syndrome (ICD-10 codes F00-F09 or G30-G32), and had never been dispensed antidementia medication (donepezil, rivastigmine, galantamine, or memantine).

Variables

Information on dementia type and date was collected from SveDem. Data on subsequent stroke event, demographics, follow-up, and death in hospital was obtained from Riksstroke. Smoking was defined as more than one cigarette per day or if a patient quit less than 3 months ago. Data on medication prior to stroke and at 1st, 2nd, and 3rd year post-stroke was obtained from PDR according to following ATC codes: antiplatelets (platelet aggregation inhibitors – B01AC06, B01AC04 [acetylsalicylic acid, and clopidogrel]), anticoagulants (B01AA, B01AE, B01AF [vitamin K antagonists, direct thrombin, and factor Xa inhibitors]), blood pressure lowering medication – BPLM (C02, C03, C07, C08, C09A-D [diuretics, beta blocking agents, calcium-channel blockers, agents acting on the renin angiotensin system, other]), and statins (C10AA). Data on medication at discharge was obtained from Riksstroke. Similarly, medication at discharge was defined as antiplatelet, anticoagulant, statin therapy, and BPLM.

Medication at discharge is presented as planned initiation of medication (patient was receiving medication or its initiation was planned within 2 weeks of discharge), and for medication prior to stroke or at 1, 2, or 3-year follow-up after stroke, receiving therapy was registered if a patient collected at least one prescription from the pharmacy in the 6 months before the stroke event, 1, 2, or 3-year period after the stroke event. Anticoagulant use is presented only in patients with AF.

Statistical analysis

Data are presented as number of cases and percentages (categorical variables) and as mean (±standard deviation–SD) or median (±interquartile range–IQR) for continuous variables. For calculating significant differences, Chi-square test was used for categorical and Student t-test and Mann-Whitney U-test were used for continuous variables.

To assess the relationship between dementia status and receiving secondary prevention, multivariate logistic regression analyses were used. Initial regression models (Model 1) were adjusted for age and sex. Because of possible confounding, different covariates were added in a stepwise manner. In Model 2, we added nursing home placement after discharge as a surrogate of a worse functional status. Final model (Model 3) was adjusted for possible confounding comorbidities and medication. Adjusted odds ratios (ORs) with 95% confidence intervals (CIs) are presented.

Tests with p value <0.05 (2-tailed) were considered significant. The IBM Statistical Package for Social Sciences (IBM SPSS) for Windows, version 22 (IBM Corporation, Armonk, NY, USA) was used.

Standard protocol approvals and patient consent

This study was approved by the regional ethical review board in Stockholm, Sweden (dnr 2015/743-31/4) and it complies with the Declaration of Helsinki. At the time of diagnosis, patients and relatives were informed of inclusion in SveDem and Riksstroke and could decline participation or withdraw consent. Data were de-identified before the analysis and no connection could be made to an individual.

Anticoagulation

Oral anticoagulants (OACs) reduce the risk for stroke in patients with AF from 4.5% to 1.4% per year [19]. Long-term anticoagulation is warranted in patients with history of stroke and AF, as this places the patient at a high risk for systemic cardioembolism, with CHADS₂ score of 2 or higher [20].

Our results are in line with previous studies, which reported underutilization of anticoagulation in patients with dementia and AF [9–11, 21]. In one study, patients with dementia were less likely to receive any antithrombotic medication (aspirin or warfarin) [9], while in another, dementia was a significant predictor of not receiving secondary prevention medication, including anticoagulants and statins, but only in a univariate and not in a multivariate analysis [12].

Multiple reasons exist why OACs are not initiated in patients with dementia. Patients with AF are undertreated with OACs; between 2005–2007 only 45% of patients with AF and IS or transitory ischemic attack received OACs [10]. This especially holds true for older patients, which have 1) a greater prevalence of amyloid-β deposition in the vasculature predisposing them to dementia and intracranial hemorrhage, 2) are taking more medication that may interact with warfarin, and 3) have a greater tendency to fall [8]. However, to offset the stroke reduction benefits of warfarin, it has been estimated that an older patient would need to fall 295 times in 1 year [22] and according to some studies benefit of oral anticoagulation might even increase with age [23, 24]. Older age, female sex, worsening disability status, and dementia, decreased the chances of receiving OACs [10]. Non-initiation of OACs during hospitalization despite recommendations was also associated with a lower adherence [25]. On the other hand, treatment in a specialized neurological department facilitates OAC initiation in secondary prevention [10].

Dementia is not a contraindication for OACs and clinicians should not withhold effective therapy in patients with cognitive decline if other contraindications are not present. Low utilization of OACs and high antiplatelet use in patients with AF could partly be due to our study’s time frame. Each year, OACs utilization in patients with AF increased substantially; e.g., between 2011–2015 in Stockholm, Sweden, OAC utilization climbed from 50% to 70%, which was accompanied by a concomitant decrease in use of aspirin in AF from 30% to 15% [26]. This could be partly explained by the introduction of non-vitamin K oral anticoagulants (NOACs), which were infrequent at the time of our study. In Sweden, the use of OAC in the AF-population overall has increased significantly after the introduction of NOAC [27, 28]. As newer anticoagulants require less monitoring, patients with dementia might benefit more from NOACs, and this might facilitate higher anticoagulant rates than those seen in our study—however, data on present anticoagulation rates in Sweden in patients with dementia is lacking.

Antiplatelets

Antiplatelets offer an absolute risk reduction of 2% in vascular events per year [2] and have probably a larger population impact on acute stroke treatment than thrombolysis [29]. To the best of the authors’ knowledge, there are no randomized placebo-controlled studies on the benefit of secondary prevention with antiplatelets versus increased bleeding risk specifically in patients with dementia. Thus, in all patients with a noncardioembolic ischemic stroke, antiplatelet agents are recommended to reduce the risk of recurrent stroke either as aspirin monotherapy, the combination of aspirin and extended-release dipyridamole, or clopidogrel monotherapy [19].

Patients with dementia were more likely to receive antiplatelet therapy at discharge and at first year after IS. In contrast to our findings, patients with dementia were less likely to receive aspirin or warfarin in an older study [9], while Eissa and colleagues did not find dementia was a significant predictor for receiving antiplatelets [12].

Even though antiplatelet therapy does not protect from embolic stroke, the increase in prescription of antiplatelets in patients with dementia was probably due to use of antiplatelet instead of OAC in patients with dementia and AF. This became apparent in a stratified analysis for AF. In the presence of AF, patients with dementia were more likely to receive antiplatelets, while in the non-AF group, there were no differences between dementia and nondementia patients. In fully adjusted models, this association disappeared or became less evident in the 2nd and 3rd year post-stroke. It seems other comorbidities play a more important role than dementia and AF when considering antiplatelet treatment in patients ≥2 years after stroke.

Blood pressure lowering medication (BPLM)

On a population level, targeting hypertension carries the highest benefit in reducing stroke burden. Newest guidelines define hypertension as blood pressure (BP) higher than 130/80 mmHg [30], but BPLM as stroke secondary prevention should not be reserved for only those with history of hypertension and should only be withdrawn in cases where hypotension could potentially be more harmful [2]. BP reduction with 10/5 mmHg reduces chances for recurrent stroke by about a third [1]; however, ACE inhibitors and diuretics reduce stroke risk regardless of BP levels and history of hypertension [8]. Moreover, not only the BP level, but also BP variability increases cerebrovascular risk, and diuretics and calcium antagonists have been shown to reduce BP fluctuations [31].

In our study, patients with dementia had a lower probability of receiving BPLM. No previous studies have addressed this topic. The reasons behind non-initiation and discontinuation could be poor adherence by the patient, the presence of potentially detrimental hypotension and greater prevalence of orthostatic hypotension, lack of guideline adherence by physicians or regular medical follow-up, or reduced need for lowering BP since it decreases after dementia diagnosis [32, 33].

Statins

Reduction in LDL cholesterol concentration has been shown to be associated with a 21% stroke risk reduction [3], as well as with reduced risk of cognitive impairment in stroke patients [6]. Even though statins are associated with numerous side effects [34], their withdrawal after stroke significantly worsens the outcome at 3 months [8]. The target LDL concentration is <2 mmol/L [1]. In our study, patients with dementia were half as likely to receive statins in years after stroke compared to patients without dementia. Reasons for non-initiation of lipid lowering therapies in patients with dementia could be shorter life expectancy.

Adherence and plausibility

The therapeutic benefit of secondary prevention medication is dependent on adherence. By 2 years, approximately 20% of patients discontinue aspirin or warfarin according to prior studies [35]. Adherence is even poorer in statins as only 43% of patients were taking them at 6 months [35]. The introduction of medication is time and place dependent; not being started on therapy in hospital or within 4 weeks of discharge were risk factors for poor adherence to therapy [11]. Evidence for the association between cognitive impairment and non-adherence is inconsistent. In one pooled analysis, there was no evidence of an association between cognitive impairment and non-adherence [5]. It may, however, be important to distinguish between degrees of cognitive impairment, as individuals with more severe impairments and dementia may rely on caregivers to administer medication, leading to increased adherence. The association between cognitive impairment and adherence, may in fact be U-shaped [5].

After IS, patients with dementia receive less specific diagnostic tests and assessments by the interdisciplinary stroke team [36] and have worse functional status [37, 38]. It is important to weigh possible therapeutic benefits after severe IS against a particular patient’s limited life expectancy. Most studies demonstrated a reduction in secondary events only after ≥2 years of therapy, implying some patients may not benefit from secondary prevention [8]. Biological age does not always match the chronological age and when planning treatment and rehabilitation plans, functional status, co-morbidities, and the patient’s goals of care should take precedence over age [8].

Limitations

Firstly, we used data on medication dispensation and not actual medication consumption, which may overestimate medication use. Whether there was a difference in primary non-compliance between dementia and non-dementia patients is outside the scope of this study. Because of the time period of the study, NOACs were not widely used and we were not able to assess their use in dementia and AF separately. Secondly, assessment of baseline functional status was limited to living situation and data on cognitive status (MMSE) were obtained with a median 1.5 years before the stroke at the time of dementia diagnosis. However, it is impossible to determine dementia severity at the time of stroke. Thirdly, the proportion between dementia and non-dementia patients at 3rd year post-stroke does not match the baseline ratio, since a greater proportion of dementia patients died. This may lead to some bias, as the two groups were matched at the time of stroke event only. Lastly, diabetes is an important comorbidity in stroke, but because of the absence of laboratory data which may influence the use and choice of antidiabetic medication, we did not include these in our study.

Despite these limitations, this study provides important insights into utilization of secondary stroke prevention in dementia.

Summary

Dementia was a predictor of lower anticoagulant, statin, and blood pressure lowering medication use for secondary stroke prevention; these represent key target areas for quality improvement initiatives.