Dorien Brouwer

27 Determinants of intention to change health-related behavior and actual change PART 1 Statistical analysis Statistical analysis were performed with STATA 12.1 statistical package (Statacorp, College Station, Texas). Missing variables of IPAQ-S and FFQ questionnaire were imputed with single imputation. We studied the relation between baseline intention to change and actual health-related behavior change after three months with univariable linear and logistic regression. Second, univariable linear regression analyses were performed to identify sociocognitive determinants of behavioral intention at baseline. The analyses were also conducted for actual change after three months. Determinants with p -value ≤0.2 were further analyzed in multivariable models. Adjustments were made for age, sex, cognitive performance, depression and event type (TIA versus ischemic stroke). Interactions between age, sex, depression, cognitive performance and event type (TIA versus ischemic stroke) on the one hand and sociocognitive determinants on the other were explored as well. Results We included 100 patients between February and October 2012. Mean age was 64 years (SD 12), 60% of the patients were male and 53% had a TIA (Table 1). Follow up was completed in 87 patients: 5 patients refused follow-up, 1 patient was lost to follow-up, 1 patient was excluded because of severe other comorbidity, 1 because of intracerebral hematoma during follow-up, 2 patients because of misdiagnosis, and 3 patients were discharged to another hospital. No significant differences in baseline characteristics were found between included patients and excluded patients (data not shown). Food frequency questionnaire was completed in 62 patients and the physical activity questionnaire IPAQ-S in 70 patients. Median behavioral intention to change was 2 (IQ 2-4) for physical activity, and 1 (IQ 0-1) for dietary behavior and smoking cessation. Thirty-two patients (37%) changed their health- related behavior by improving more than 30 minutes in physical activity a day and 9 patients (31%) stopped smoking. Only one patient (2%) improved in dietary behavior. Forty-four patients (53%) lost weight, on average 1.21 kg (SD 3.43). No changes in waist circumference were found. We found that intention to change was not significantly associated with actual change. However, patients with a higher intention to change tended to change their health-related behavior more often (Table 2). No associations between sociocognitive factors and actual change were found (data not shown).