P1-528 Statistical models in longitudinal evaluation of changes in health behaviours

Introduction During long-lasting intervention program, the socio-demographic structure of the target population may change and influence the results of the overall evaluation. The Primary Prevention Program of Neural Tube Defects in Poland which was carried out in 1997–2007 in the whole country, gives the opportunity to discuss application of different statistical models in this context. Methods Three surveys were conducted on representative samples of women aged 18–35 years during the Program activity. The samples were drawn from patients of the same randomly selected primary health centres in five chosen main administrative regions in 2001 (n=775), 2003 (n=756) and 2007 (n=756). Relative changes in women's knowledge and behaviour concerning folic acid were analysed by Poisson regression and generalised estimating equation models. Results The proportion of women taking folic acid during the pregnancy increased from 52% in 2001 to 68% in 2003 and 86% in 2007 and the proportion of women beginning supplementation before the pregnancy increased from 11% to 15% and 29%, respectively. The behaviours towards folic acid strongly differentiated according to socio-demographic characteristic, which was also not stable during the period of program implementation. For example, the proportion of post-secondary educated women increased from 20.4% in 2001 to 23.1% in 2003 and 41.7% in 2007. These changes influenced slightly, although not significantly and depending on statistical approach chosen, the overall magnitude of effects. Conclusions Time-depending socio-demographic structure of the target population has to be considered in evaluation of health promotion programs

Context (archaeology) · Environmental health · folic acid · Poisson regression · Population · Pregnancy · Demography · Internal Medicine · Medicine · Nutritional Studies and Diet · Gerontology · Pediatrics