Gesundheitswesen 2017; 79(08/09): 656-804
DOI: 10.1055/s-0037-1605859
Vorträge
Georg Thieme Verlag KG Stuttgart · New York

Impact of Baseline Factors Influencing Poor Outcome after Ischemic Stroke – The PROSpective Cohort with Incident Stroke Berlin (PROSCIS)

C Malsch
1   Universität Würzburg, Institut für klinische Epidemiologie und Biometrie, Würzburg
,
T Liman
2   Charité Universitätsmedizin, Centrum für Schlaganfallforschung BerlinBerlin, Berlin
,
S Wiedmann
1   Universität Würzburg, Institut für klinische Epidemiologie und Biometrie, Würzburg
,
B Siegerink
2   Charité Universitätsmedizin, Centrum für Schlaganfallforschung BerlinBerlin, Berlin
,
M Endres
2   Charité Universitätsmedizin, Centrum für Schlaganfallforschung BerlinBerlin, Berlin
,
PU Heuschmann
1   Universität Würzburg, Institut für klinische Epidemiologie und Biometrie, Würzburg
› Author Affiliations
Further Information

Publication History

Publication Date:
01 September 2017 (online)

Also available at
 

Introduction:

Long-term dependency affects patients suffering from stroke, but few studies quantify the contribution of risk factors to poor outcome. We aimed to identify baseline factors influencing poor outcome one year after first-ever stroke and to quantify the respective population attributable risks [PARs].

Methods:

We used data from the PROSCIS study (a longitudinal study on prediction of vascular risk after stroke; NCT01364168) to identify factors influencing poor outcome one year after first-ever ischemic stroke including sociodemographic and clinical characteristics and stroke comorbidities. Poor outcome was defined as death or dependency (modified Rankin Scale > 3 or Barthel Index < 60). For multivariable analysis, binary logistic regression was used. We calculated the proportion of poor outcome due to the relevant risk factors using average PARs and conducted as sensitivity analysis Firth“s likelihood penalization to reduce small sample bias.

Results:

499 patients with ischemic stroke were included. Mean age was 66 years, 39% were females, median NIHSS was 2. 11.4% suffered poor outcome one year after stroke. Variables independently associated with poor outcome included age (> 65 years), stroke severity (NIHSS on admission> 2) and following comorbidities pre stroke: diabetes mellitus, atrial fibrillation, myocardial infarction/angina pectoris and transitory ischemic attack. The average PARs were 30.8%, 17.4%, 9.7%, 14.6%, 8.4% and 3.3%, respectively. Firth“s maximum likelihood penalization confirmed odds ratios gained from binary logistic regression; confidence interval estimation reached higher precision: confidence interval width decreased in 80.9% of all performed estimations and could be decreased by up to 36.9%.

Discussion:

Pre-stroke cardiovascular comorbidities contributed substantially to poor outcome and explain about one third of its occurrence in our model. Firth“s likelihood penalization is a suitable method to control small sample bias.