INTRODUCTION A population health profile is a cumulative product of socioenvironmental and political factors that create the contexts in which health problems arise, as well as opportunities and barriers to addressing them. Research on context has focused on demonstrating its effects, direct or indirect, on health indicators, but has made few incursions into assessing its role as a mediator of other factors. While individual risk factors for chronic diseases are well known, the same cannot be said for the complex of contextual factors operating at various levels and over the lifespan.

OBJECTIVES Estimate relative influences by contextual versus individual factors as determinants of diabetes type 2 and heart disease.

METHODS A cross-sectional study was carried out in populations served by 12 family doctor-and-nurse practices in Havana, in 840 families selected by simple random sampling, 70 per practice. Principal components analysis was used, as well as contextual logistic regression models with a nested model strategy, whose fit was meant to estimate the relative contributions of contextual compared to individual risk factors for diabetes and heart disease. Context was described and analyzed at two levels: that of the family or household and that of the catchment area served by a family doctor-and-nurse practice (geographically defined as a neighborhood).

RESULTS For diabetes, the contextual effect of neighborhood was modified when household effect was removed; that is, the effect of neighborhood was indirect and mediated by household. Individual coefficients were practically invariant; the principal effect of household changed noticeably on removal of individual effects, while age maintained its effect without variation.

For heart disease, the effect of neighborhood was slightly modified when household effect was controlled for. Individual coefficients showed little change. There was an important direct effect of household on risk of heart disease. Age and high blood pressure coefficients hardly varied.

CONCLUSIONS We confirmed interactions between individual and contextual (neighborhood and household) factors, whose effects on individual health are not entirely mediated by individual factors. Research needs to pay more attention to context beyond its direct effect on individual risk factors.

KEYWORDS Risk factors, risk reduction, community, neighborhood, type 2 diabetes mellitus, heart disease, logistic models, Cuba