Rates, Predictors, and Outcomes of Early and Late Recurrence After Stroke
The North Dublin Population Stroke Study
Background and Purpose—Few recent studies have investigated the rates and predictors of early and late stroke recurrence using prospective population–based methodology. We investigated recurrent stroke at 2 years in the North Dublin Population Stroke Study (NDPSS).
Methods—Patients were ascertained from December 2005 to 2006 from overlapping community and hospital sources using hot and cold pursuit. Stroke recurrence, survival, and functional outcome were ascertained at 72 hours, 7 days, 28 days, 90 days, 1 year, and 2 years.
Results—Of 567 patients, cumulative 2-year stroke recurrence rate was 10.8% and case fatality was 38.6%. Recurrence subtype was associated with initial stroke subtype (P<0.001). On multivariable Cox regression, hyperlipidemia (adjusted hazard ratio, 3.32; P=0.005) and prior stroke (adjusted hazard ratio, 2.92; P=0.01) were independent predictors of 2-year recurrence in 28-day survivors.
Conclusions—Despite rigorous ascertainment, recurrent stroke rates were lower in current study than in earlier studies. Our data suggest that large sample sizes may be needed for future secondary prevention trials in patients treated with modern preventive medications.
Recent literature reports declining incidence and case fatality after stroke, coincident with improved vascular risk factor control.1–4 It is also possible that wider use of antiplatelet, antihypertensive, and statin medications may influence late stroke fatality and functional outcome via reduced recurrent vascular events.4–6 However, few recent population-based data exist on rates and determinants of late stroke recurrence.6
We aimed to quantify rates and predictors of stroke recurrence 2 years after stroke in the era of modern vascular preventive therapy.
The North Dublin Population Stroke Study (NDPSS) is a prospective population-based cohort study of stroke and transient ischemic attack. Methodology and definitions are previously described4 (online-only Data Supplement). We recruited individuals with stroke occurring over a 1-year period in 2006. Qualifying event (QE) was coded as first-ever stroke (QE-FES) or prior stroke (QE-PS) if remote stroke existed before the study onset.
Stroke recurrence, fatality, and functional outcome were assessed at 72-hour, 7-day, 28-day, 90-day, 1-year, and 2-year time interval.
Inclusion criteria were new stroke event (ischemic, intracerebral hemorrhage [ICH], and subarachnoid) during the ascertainment year (2) age >18 years (3) consent available.
A survival-based approach using life-table analysis and Kaplan–Meier curves was used for time-to-event analysis.
Five hundred sixty-seven patients qualified for study entry, 484 (85.4%) with QE-FES and 83 (14.6%) with QE-PS. Overall, 80.1% (454/567) had ischemic stroke, 10.7% (61/567) had ICH, 5.1% (29/567) had subarachnoid hemorrhage, and 4.1% (23/567) were unconfirmed.
Clinical characteristics are in Table 1, risk factors are previously described4 (clinical characteristics analyzed by stroke subtype, Table I in the online-only Data Supplement). Data were available on acute National Institutes of Health Stroke Scale (NIHSS) in 83.6% (474/567), median 5 (interquartile range, 2–11). Data were available on prescribed medications in 92.9% of survivors beyond day 28 (Table II in the online-only Data Supplement).
Recurrent Stroke Rates and Subtypes
At 2 years, follow-up data were complete for stroke recurrence status in 91.4% (518/567). Recurrent stroke occurred in 46 patients. On life-table analysis, the cumulative rate of recurrent stroke was 3.8% at 28 days and 10.8% at 2 years. Two-year recurrence was 11.5% for ischemic stroke and 5.7% for ICH (Table 2). Late recurrence rates were higher for patients, who qualified with QE-PS compared with QE-FES (P=0.004; Table 3; Figure I in the online-only Data Supplement).
Of the 46 recurrent strokes, brain imaging was performed in 93.5% (43/46). Recurrent stroke subtypes were ischemic in 84.8% (39/46), ICH in 8.7% (4/46), and unknown in 6.5% (3/46). Recurrent stroke subtype was strongly associated with initial subtype (P<0.001).
On Cox regression in 28-day survivors, QE-PS (hazard ratio, 3.3; P=0.005) and hyperlipidemia (hazard ratio, 2.9; P=0.01) remained as independent predictors of 2-year recurrence (Table III and Figure II in the online-only Data Supplement).
Median deterioration in Rankin score was greater at 2 years in patients with stroke recurrence (P=0.03). Two-year case fatality was 38.6%. Stroke recurrence was not associated with higher 2-year fatality (online-only Data Supplement).
We report rates and predictors of early and late recurrent stroke in a recent population-based study with gold-standard rigorous ascertainment methods, high rates of modern stroke preventive medication utilization, and high follow-up rates. Despite rigorous follow-up for recurrent events and inclusion of very early recurrences, the rate of stroke recurrence at 2 years in our study was 10.8%. Direct comparison of studies, which report stroke recurrence, is difficult because of variation in factors, such as study setting, methodology, case mix, and study era.7 Allowing for these caveats, the recurrence rates observed in FES patients in our study (7.4% at 1 and 8.7% at 2 years) are at the lower limits of the ranges reported at these time intervals from developed countries in the past decade.5
We suspect that our relatively low recurrence rate may relate to the high rates of preventive medication use in our cohort. One systematic review found a temporal reduction of 5-year stroke recurrence rates in recent epidemiological studies,5 whereas another review found that recurrent stroke rates had decreased substantially since 1960, associated with increased antiplatelet medication use and blood pressure control.8 Our findings are also consistent with data from the Vitamin Intervention for Stroke Trial (VISP) trial, which reported 2-year recurrent stroke rates of 7.3% in patients appropriately prescribed secondary prevention medications compared with 15.9% in those who were not.3 Although we were unable to detect an association between preventive medication use and recurrence, this cannot be fully excluded, given that our data set may have lacked statistical power to detect a modest association.
We found that PS and hyperlipidemia were independent predictors of late recurrent stroke. Although we observed high rates of statin prescribing, our study predated the Stroke Prevention by Aggressive Reduction of Cholesterol Levels (SPARCL) trial and subsequent updated lipid-lowering guidelines.3 Our findings emphasize the high future risk of patients with 2 or more cerebrovascular events, which persists despite modern preventive therapies. The relationship of lipids with recurrent stroke in earlier studies is inconsistent, possibly reflecting subtype-specific risk or heterogeneity of methods. A recent large study reported that baseline dyslipidemia was independently associated with increased risk of late stroke recurrence.9
The strengths of our study include its prospective population-based design with rigorous ascertainment, high rates of late follow-up, in-person physician verification and subtyping of recurrences, and inclusion of ICH. We performed detailed analyses, including modern medication data, and included first-ever and PS cases.
Our main limitation is that our sample size may have been insufficient for detection of modest associations between candidate predictor variables and recurrent stroke.
For investigators considering new randomized trials, our data provide further evidence that low recurrence rates are to be anticipated, leading to a requirement for very large sample sizes to demonstrate efficacy of new candidate agents for secondary prevention after stroke.
Sources of Funding
Drs Kelly, Kyne, Sheehan, Merwick, and Hannon received support from Health Research Board and Irish Heart Foundation. Dr Marnane received unrestricted grant from Pfizer. Dr Hayden, Newman Fellowship, funded by Bayer. Additional funding was from National Lottery of Ireland, Servier.
Dr Kelly received uncompensated speaking engagements with Bayer, Boehriger-Ingelheim, and Daiichi Sanko. The other authors report no conflicts.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.115.011248/-/DC1.
- Received August 27, 2015.
- Revision received August 27, 2015.
- Accepted September 28, 2015.
- © 2015 American Heart Association, Inc.
- Rothwell PM,
- Giles MF,
- Chandratheva A,
- Marquardt L,
- Geraghty O,
- Redgrave JN,
- et al
- Ní Chróinín D,
- Asplund K,
- Åsberg S,
- Callaly E,
- Cuadrado-Godia E,
- Díez-Tejedor E,
- et al
- Kelly PJ,
- Crispino G,
- Sheehan O,
- Kelly L,
- Marnane M,
- Merwick A,
- et al
- Mohan KM,
- Wolfe CD,
- Rudd AG,
- Heuschmann PU,
- Kolominsky-Rabas PL,
- Grieve AP.
- Hankey GJ,
- Jamrozik K,
- Broadhurst RJ,
- Forbes S,
- Burvill PW,
- Anderson CS,
- et al
- Hong KS,
- Yegiaian S,
- Lee M,
- Lee J,
- Saver JL.
- Kumral E,
- Evyapan D,
- Gökçay F,
- Karaman B,
- Orman M.