Agreement between information provided by stroke patients and their relatives on psychophysical and vascular risk factors

Abstract

Background: In studies examining vascular risk factors and the effects of stress in stroke patients, information sometimes has to be provided by a close relative.

Objectives: The present study was designed to assess agreement between the information provided by a stroke patient and his/her next-of-kin or other close relative on prior vascular risk factors and situations of psychophysical stress based on a standardized interview within 72 h after stroke.

Design: Case-control observational study.

Participants and settings: All patients with incident stroke aged 18-65 years admitted to our centre were enrolled as cases, and their coinhabiting relatives as controls. The study was conducted from January to June 2008.

Methods: A structured questionnaire about vascular risk factors and psychophysical stress factors were obtained from the participants within 72 h after stroke. Subjects were required to grade themselves or their relatives using validated Spanish versions of the Life Events Stress Scale of Holmes and Rahe, Goldberg's General Health Questionnaire GHQ-28 and the SF-12 Health Survey, as well as the Spanish Type A Behaviour Scale. The questionnaires were self-administered. Cases and controls completed the same questionnaires and which were administered with a time difference of no longer than 24 h between the two groups of subjects. Risk factors were analyzed by determining kappa statistics and interclass correlation coefficients (ICC). Bland-Altman plots were used to examine the scores obtained in graded scales.

Results: 25 patients and 25 relatives were recruited. Agreement between cases and controls was good for reported atrial fibrillation, diabetes, alcohol consumption and smoking (range 0.83-1.0). ICC were low for the questionnaires Holmes-Rahe Life Events (0.26; 95%CI: 0.08-0.57), General Health GHQ-28 (0.39; 95%CI: 0.03-0.67) and SF12 Health Survey (0.52; 95%CI: 0.16-0.76 and 0.35; 95%CI: 0.06-0.66), and good for the Type A Behaviour Scale ERCTA (0.62; 95%CI: 0.32-0.81).

Conclusions: The information obtained from family members on patient risk factors could be considered reliable, while that related to psychophysical aspects was not sufficiently reliable for research purposes. Agreement assessments could be useful to avoid misclassification biases.