Pre‐ and post‐stroke oral antithrombotics and mortality in patients with ischaemic stroke

Abstract Background Reducing stroke occurrence requires the effective management of cardiovascular and other stroke risk factors. Purpose To describe pre‐ and post‐stroke medication use, focusing on antithrombotic therapy and mortality risk, in individuals hospitalised for ischaemic stroke (IS) in the United Kingdom. Method Using primary care electronic health records from the United Kingdom, we identified patients hospitalised for IS (July 2016–September 2019) and classed them into three groups: atrial fibrillation (AF) diagnosed pre‐stroke, AF diagnosed post‐stroke, and non‐AF stroke (no AF diagnosed pre‐/post‐stroke). We determined use of cardiovascular medications in the 90 days pre‐ and post‐stroke and calculated mortality rates. Results There were 3201 hospitalised IS cases: 76.2% non‐AF stroke, 15.7% AF pre‐stroke, and 8.1% AF post‐stroke. Oral anticoagulant (OAC) use increased between the pre‐ and post‐stroke periods as follows: 54.3%–78.7% (AF pre‐stroke group), 2.3%–84.8% (AF post‐stroke group), and 3.4%–7.3% (non‐AF stroke group). Corresponding increases in antiplatelet use were 30.8%–35.4% (AF pre‐stroke group) 38.5%–47.5% (AF post‐stroke group), and 37.5%–87.3% (non‐AF stroke group). Among all IS cases, antihypertensive use increased from 66.8% pre‐stroke to 78.8% post‐stroke; statin use increased from 49.6%–85.2%. Mortality rates per 100 person‐years (95% CI) were 17.30 (14.70–20.35) in the AF pre‐stroke group and 9.65 (8.81–10.56) among all other stroke cases. Conclusion Our findings identify areas for improvement in clinical practice, including optimising the level of OAC prescribing to patients with known AF, which could potentially help reduce the future burden of stroke.

• 54.5% of patients with atrial fibrillation (AF) pre-stroke received oral anticoagulant therapy in the 90 days pre-stroke versus 78.7% in the 90 days post-stroke.
• Among all stroke cases, antihypertensive use increased from 66.8% pre-stroke to 78.8% post-stroke; statin use increased from 49.6% to 85.2%.
• Antiplatelet use among the non-AF stroke group increased from 37.5% pre-stroke to 87.3% post-stroke.

Plain Language Summary
The absolute number of people experiencing ischaemic stroke (IS) is increasing, and good management of cardiovascular risk factors is needed to reduce levels of IS risk in the general population. Using primary care electronic health records from the United Kingdom, we calculated mortality rates among patients with IS, and described the use of cardiovascular medications in these patients, both before and after their stroke. We identified 3201 patients who had been hospitalised for IS: 15.7% had a diagnosis of atrial fibrillation (AF) pre-stroke, 8.1% had AF poststroke, and 76.2% did not have AF. Mortality rates were highest in patients with IS and known AF; among these patients, 54.5% received oral anticoagulant (OAC) therapy in the 90 days prestroke compared with 78.7% in the 90 days post-stroke. Antiplatelet use among non-AF stroke patients increased from 37.5% pre-stroke to 87.3% post-stroke; smaller increases in antiplatelet use were seen among other IS cases. Antihypertensives and statins were widely used pre-and post-stroke among all IS cases. Our findings indicate that there is scope for improvement in the level of OAC prescribing to patients with known AF, which could potentially help reduce the future burden of stroke.

| INTRODUCTION
Age-specific incidence and mortality rates of stroke have been declining in all European countries, yet the number of people experiencing stroke is rising due to the ageing population and the strong age relationship with stroke risk. 1 In the United Kingdom, stroke is the fourth largest cause of death, and more than 1 in 8 people admitted to hospital with stroke in England and Wales die within 30 days. 2 Good management of cardiovascular and other stroke risk factors is needed to reduce the occurrence of stroke. Eighty-five percent of strokes are ischaemic, 3 and up to a third of these are caused by cardiac thromboembolism attributed to atrial fibrillation (AF). 4,5 Cardioembolic strokes are often severe, leading to substantial disability or death, 6 yet up to two-thirds could be preventable with use of oral anticoagulant (OAC) therapy in patients with AF. 7 However, underuse of OAC therapy in patients with atrial fibrillation (AF) remains common. [8][9][10][11] Improving stroke prevention and outcomes is a key priority in the UK National Health Service (NHS) Long Term Plan. 12 Understanding the pattern of medication use in patients at risk of stroke would help target efforts to improve stroke prevention. Using population-based primary care data from 2016 to 2018, we aimed to describe pre-and post-stroke cardiovascular medication use (with a focus on antithrombotic therapy), and 1-year mortality in individuals hospitalised for IS in the United Kingdom: those with AF diagnosed pre-stroke, those diagnosed with AF post-stroke, and those without a diagnosis of AF.

| Data source
We used the IQVIA Medical Research Data-UK (IMRD-UK) primary care database, previously The Health Improvement Network. The database contains de-identified longitudinal patient data entered by primary care practitioners (PCPs) during routine patient care. Clinical data are entered using Read codes, the standard clinical codes used in the NHS. Prescriptions issued by the PCP are recorded automatically, and information from secondary care is entered retrospectively. In the United Kingdom, chronic disease is managed in primary care, making the IMRD-UK suitable for this study. Additionally, in another UK primary database similar to IMRD-UK, cerebrovascular disease Read codes have a 93% positive predictive value (PPV). 13 The IMRD-UK covers around 6% of the UK population 14 and is generalisable to the country as a whole. 15

| Source population and identification of ischaemic stroke cases
The source population included individuals aged 20-89 years between 1 July 2016 and 30 June 2018, permanently registered with their PCP and with at least 3 years' registration after their first recorded prescription. We followed individuals from the date they met the study entry criteria (start of follow-up), until the first coded IS entry, death, or the end of the observation period (September 2019), whichever came first. We included only hospitalised cases, identified as having a record of hospitalisation between 15 days pre-stroke and 30 days post-stroke. This was determined by manually reviewing patient records as described previously. 16 The date of hospitalisation for the stroke was the index date, and we retained only cases aged ≥55 years. As members of the source population may have experienced a stroke before the start date, cases could have been either first or recurrent events.

| Ischaemic stroke cases with/without AF
We divided IS cases into three groups: AF diagnosed pre-stroke ('AF pre-stroke group'), patients newly diagnosed with AF either on the date of stroke or in the year after ('AF post-stroke group'), and patients without AF diagnosed pre-stroke and without AF diagnosed in the year post-stroke ('non-AF stroke group'). Atrial fibrillation was ascertained by the presence of an AF code. We excluded patients with a code for mitral stenosis or valve replacement surgery before the date of AF diagnosis or in the 2 weeks after.

| Comorbidities and other patient variables
We obtained information on patient demographics (age and sex) and lifestyle factors (body mass index, smoking and alcohol intake) any time before the index date using the most recent recorded value/ status as appropriate. Healthcare use was measured by the number of PCP visits, referrals and hospitalisations in the year before the index date. Information on comorbidities (including cardiovascular disease and risk factors, renal function and frailty) were determined any time before the index date; for a history of cerebrovascular disease, we looked at the time period before the start of follow-up. Further details on the ascertainment of co-variates have been published previously. 16

| Antithrombotics and other cardiovascular medications
We determined pre-and post-stroke exposure to antithrombotics [OACs and antiplatelets (low-dose aspirin and clopidogrel)] and other medications (including antihypertensives, diuretics, beta-blockers, angiotensin-converting enzyme inhibitors, angiotensin II receptor blockers, statins, digoxin, and antiarrhythmics) among stroke cases still alive 30 days post-stroke, based on prescription records. Pre-stroke medication exposure was defined as a prescription in the 90 days before the index date, and post-stroke medication exposure as a prescription on the index date or in the 90 days after.

| Mortality rates
To determine deaths following stroke, we followed IS cases from the index date until the date of death or 31 September 2019 (the last available data collection date), whichever came first.

| Statistical analysis
The study was designed to be predominantly descriptive in nature with no formal statistical hypotheses. For each stroke group, preand post-stroke cardiovascular medication use, as well as other patient characteristics, were described using frequency counts and percentages for categorical data and means with standard deviation (SD) for continuous data. We calculated mortality rates per 100 person-years with 95% confidence intervals (CIs) by dividing the total number of deaths during follow-up by the person-years at risk. We compared mortality rates between the AF pre-stroke group and non-AF group, but not the AF post-stroke group due to the potential of immortal time bias among these cases from the need to have survived long enough to have the post-stroke workup for AF diagnosis. We performed stratified analyses by sex and age, and a sensitivity analysis where we re-analysed medication use among the AF post-stroke group changing the index date to the date of AF diagnosis. Analyses were undertaken using Stata version 12 (StataCorp. 2017).

| Characteristics of the IS patients
We identified 3201 hospitalised stroke cases. The majority (76.2%, 2441/3201), were non-AF stroke, 15.7% (n = 501) had AF pre-stroke, and 8.1% (n = 259) were AF post-stroke. In the latter group, 61.0% (n = 158) had AF diagnosed within 30 days of the stroke date, and 39.0% (n = 101) had AF diagnosed between 31 days and 12 months after the stroke date. Characteristics of stroke cases are shown in Table 1 (see also Table S1 for distribution of CHA 2 DS 2 -VASc score).
Males accounted for slightly more than half of each group. The non-AF stroke group were youngest on average and had better renal function. The AF pre-stroke group had the highest prevalence of moderate/severe frailty, cerebrovascular disease, cardiovascular disease and reduced renal function, while the AF post-stroke group were more overweight/obese.

| Pre-and post-stroke OAC use
In each group, OAC use increased in the post-stroke period ( Figure 1;

| Pre-and post-stroke antiplatelet use
The proportion of stroke cases prescribed an antiplatelet pre-stroke was broadly similar between groups with notable differences poststroke: 30.8% (pre-stroke) to 35.4% (post-stroke) in the AF pre-stroke group, 38.5% (pre-stroke) to 54.9% (post-stroke) in the AF post-stroke group (47.5% when using date of AF diagnosis as the index date), and 37.5% (pre-stroke) to 87.3% (post-stroke) in the non-AF stroke group (Table 2; Figure S1a and Table S3a). Antiplatelet use according to whether this was prescribed with/without OAC therapy is shown in Table S4.

| Pre-and post-stroke use of other medications
Generally, use of all cardiovascular medications increased between the pre-and post-stroke periods (    Figure 2B).

| Summary
Among 3201 patients hospitalised with IS in this study, close to one in six had a pre-stroke diagnosis of AF. These patients were, on average, older, frailer, and they more commonly received preventative cardiovascular medication pre-stroke, and had a higher mortality rate than stroke cases without AF. Among the pre-stroke AF group, 54.5% received OAC therapy in the 90 days pre-stroke, compared with 78.7% in the 90 days post-stroke. The majority (87.3%) of stroke cases without AF received post-stroke antiplatelet therapy, which was a much higher level of use than among AF stroke cases, and, in general, use of other cardiovascular medications were seen to increase between the pre-and post-stroke periods.

| Strengths and limitations
Strengths of the study include the representative sample of the UK population, making the results generalisable to the country as a whole, the suitability of the database to capture long-term use of cardiovascular medications, and the high 93% PPV of coded cerebrovascular entries. 13 We did not evaluate whether OAC therapy was adequately dosed as this is challenging to establish for VKAs and was beyond the scope of the study. Our main analyses were not restricted to high-risk patients, however, additional analysis by CHA 2 DS 2 VASc score in AF T A B L E 2 OAC and other medication use in the 90 days pre-stroke or in the 90 days post-stroke among patients still alive 30 days after their stroke (n = 3119), according to AF group  Soriano has no potential competing interests.

ETHICS STATEMENT
The study protocol was approved by the Independent Scientific Research Committee for IMRD-UK (reference number 19THIN057).