Estimating spatial disease rates from health statistics without geographic identifiers

Effective analysis to support decision-making in public health requires adequate data that can be linked to the sociodemographic characteristics and distribution of communities that would be recipients of health programs. Although official statistics are an important source of data to support decisions on public health strategies, health statistics are of limited use if they do not include categorisations related to the spatial distribution of the population. This study introduces a method to use the frequency of disease reported by health care providers (HCP) to calculate disease rates in geographical areas in urban settings. Specifically, this study uses statistics on acute respiratory infections (ARI) to calculate the rate of these diseases at the census district level in Cúcuta, a major city in Colombia. A Geographic Information System was used to establish the geographical area of influence (spatial scope) of each HCP, according to the distribution of the basic census geographical areas; the Urban Sections (USEC), in Cúcuta. Three levels of increasing spatial aggregation were established considering the characteristics of the population receiving health care in primary, intermediate and tertiary public health services to establish the spatial scope of each HCP. The cases of ARI per USEC were calculated according to the proportion of the population of each USEC in each of the three levels. The spatial rate of ARI per USEC and the hotspots of higher risk of ARI were calculated using an Empirical Bayes method using a geostatistical software. There were 97 HCP, of which, 31 provided health services in USEC in level 1; 20 provided health services in USEC in level 2; and 47 provided health services in USEC in level 3. A higher spatial rate of ARI was found in USEC in central south; central west; north and northwest; northeast; central east; and central regions, compared to the whole of Cúcuta. Hotspots of higher risk were identified in two clusters of USEC in central south and west Cúcuta and three isolated USEC in central and northwest Cúcuta. Health indicators at the census district geographical level could be calculated using basic statistics that do not include geographic identifiers by considering the characteristics of the HCP and the population receiving their services. The methodology of this study can be applied to other socioeconomic contexts where geographic identifiers are not attached to public health statistics to create better public health indicators and support potential health promotion and disease prevention strategie