Do patients who take part in stroke research differ from non-participants? Implications for generalizability of results.

BACKGROUND: Stroke is one of the most disabling neurological conditions. Clinical research is vital for expanding knowledge of treatment effectiveness among stroke patients. However, evidence begins to accumulate that stroke patients who take part in research represent only a small proportion of all stroke patients. Research participants may also differ from the broader patient population in ways that could potentially distort treatment effects reported in therapeutic trials. The aims of this study were to estimate the proportion of stroke patients who take part in clinical research studies and to compare demographic and clinical profiles of research participants and non-participants.
METHODS: 5,235 consecutive patients admitted to the Stroke Care Unit of the Royal Melbourne Hospital, Melbourne, Australia, for stroke or transient ischaemic attack between January 2004 and December 2011 were studied. The study used cross-sectional design. Information was collected on patients' demographic and socio-economic characteristics, risk factors, and comorbidities. Associations between research participation and patient characteristics were initially assessed using χ(2) or Mann-Whitney tests, followed by a multivariable logistic regression analysis. The logistic regression analysis was carried out using generalised estimating equations approach, to account for patient readmissions during the study period.
RESULTS: 558 Stroke Care Unit patients (10.7%) took part in at least one of the 33 clinical research studies during the study period. Transfer from another hospital (OR = 0.35, 95% CI 0.22-0.55), worse premorbid function (OR = 0.61, 95% CI 0.54-0.70), being single (OR = 0.61, 95% CI 0.44-0.84) or widowed (OR = 0.77, 95% CI 0.60-0.99), non-English language (OR = 0.67, 95% CI 0.53-0.85), high socio-economic status (OR = 0.74, 95% CI 0.59-0.93), residence outside Melbourne (OR = 0.75, 95% CI 0.60-0.95), weekend admission (OR = 0.78, 95% CI 0.64-0.94), and a history of atrial fibrillation (OR = 0.79, 95% CI 0.63-0.99) were associated with lower odds of research participation. A history of hypertension (OR = 1.50, 95% CI 1.08-2.07) and current smoking (OR = 1.23, 95% CI 1.01-1.50) on the other hand were associated with higher odds of research participation.
CONCLUSIONS: The results of this study indicate that stroke patients who take part in clinical research do not represent 'typical' patient admitted to a stroke unit. The imbalance of prognostic factors between stroke participants and non-participants has serious implications for interpretation of research findings reported in stroke literature. This study provides insights into clinical, demographic, and socio-economic characteristics of stroke patients that could potentially be targeted to enhance generalizability of stroke research studies. Given the imbalance of prognostic factors between research participants and non-participants, future studies need to examine differences in stroke outcomes of these groups of patients.