RESUMO
Hospitalizations for primary care sensitive conditions (HPCSC) have been used as a measure of the effectiveness of primary care. Few studies, however, have measured the quality of primary health care (PHC) and how it impacts HPCSC trends. We employed previously calculated quality scores of PHC attributes (access, continuous/ongoing care, coordinated care and comprehensive care) assigned to health centres and investigated the correlation between these quality scores and HPCSC time trends in public hospitals in the city of Belo Horizonte, Brazil, adjusted by demographic and socioeconomic factors between 2010 and 2013. Socioeconomic risk was determined using the Health Vulnerability Index, a composite indicator including sanitation, habitation, education, income and health variables. Between 2010 and 2013, 447 500 hospitalizations, excluding births, were recorded in Belo Horizonte. Of these, 85 211 were HPCSC (19% of total admissions). Of 145 health centres, 133 were analysed in this study, with 75 059 HPCSC in the 4-year period. In the initial univariate model, only the PCH attribute 'access' and 'social vulnerability' had an effect on hospitalization rates. In the multilevel analysis, socioeconomic vulnerability became the only predictor of HPCSC rates. A 10% increase in high socioeconomic risk population meant an increase of 5 HPCSC/10 000 in women and 4/10 000 in men for each individual primary care unit, while a 10% increase in low-risk population meant a reduction in HPCSC of 6/10 000 in women and 5/10 000 in men. The results support the importance of using HPCSC as a measure of PHC effectiveness, considering individual, health service-related and socioeconomic characteristics; however, additional measures of effectiveness of care provided by family health teams seem necessary to assess PHC performance.