Resumen

ALEGRET RODRIGUEZ, Milagros; GRAU ABALO, Ricardo  y  RODRIGUEZ RODRIGUEZ, Mercedes. Multivariate classification of geographic areas as a useful alternative for public health research. Rev Cubana Salud Pública [online]. 2008, vol.34, n.2, pp. 0-0. ISSN 1561-3127.

Introduction Use of hierarchical clustering techniques based on multivariate similarities and its advantages for correlation and explanation of health deterioration statistics was exemplified, as a useful tool for health research involving space. Objectives To identify multivariate contextual constructs in geographical units and relate them to health damage. Methods A multivariate classification of the 169 municipalities by using unsupervised hierarchical cluster techniques. The analytical axes were demographic, economic and living conditions. Correlation and regression techniques with health damage indicators were also used. Results Four multivariate strata were characterized and their effective ordinal character was confirmed in relation to damage response in terms of morbidity and mortality in non communicable chronic diseases (7 out of 11 indicators) and of morbidity from communicable diseases (2 out of 8 indicators). PAMI indicators showed that they did not depend on the constructed strata. Conclusions This methodology allowed identifying a useful contextual structure to explain the distribution of studied damage events and even those events which are not dependent on this contextual structure, since they have closed gaps by means of planned actions. According to this approach, space is addressed in a flexible way, refractory to problems existing in the analysis of small areas, spatial contiguities and changes in the analytical spatial unit. This is due to the creation of a macro space for further analytical purposes and to the structuring of space beyond frontiers, thus setting a relatively homogeneous and stable contextual frame. These tools are essential to plan and evaluate public health interventions, and for the health policy management in general

Palabras clave : Public health; spatial analysis; multivariate analysis.