Linking public health surveillance systems to program effectiveness evaluation

The formal theory, methodological approach, and results of the effectiveness evaluation of a public health intervention is presented using information from public health surveillance systems and utilizing examples from interventions seeking to prevent and control chronic non-transmittable diseases (CNTD). Objective: The effectiveness of the «Cali in Motion» program was evaluated; the objective was to reduce the prevalence of cardiovascular disease in the city of Cali, Colombia. Effectiveness was assessed by the percentage of reduction in the prevalence of physical inactivity as a risk factor for cardiovascular disease. Additionally, other criteria were considered as intermediate indicators of the success of the intervention: building institutional capacity to produce and use information in action planning across disciplines concerning CNTD. The attribution of changes to the intervention was established by taking into account criteria such as: adherence to protocol, the bases for the changes found, contextual variables that influenced both outcomes and the implementation of the intervention, and the adequacy of time for the expected changes to become evident. Method: The evaluation model combines both qualitative and quantitative information to judge the effectiveness of the intervention on a scale constructed from four variables related to quality of design and implementation of the intervention, statistical significance of changes in the prevalence of risk factors, and attribution of the changes to the intervention according to the previously noted variables. Data on the changes in the prevalence of physical inactivity were obtained from the Surveillance System of Risk Factors Associated with CNTD (SIFREC) during the period of 2005-2007, and the qualitative information was obtained by systematizing the interventions. The study population was 1,385,470 individuals over 18 years of age; after correcting for design effect and taking into account an estimated prevalence of the event, p=50%; maximum sampling error of 4.4% (e); 95% confidence level (Z=1.96); and non-response rate was estimated at 1,080 subjects sampled, but the completed surveys totaled 927. For the processing and analysis of quantitative information, the EpiInfo program (version 3.4) was used, along with the module for population surveys by Stata (version 9) and Answer Tree by SPSS® (version 15). The qualitative data analysis was performed by using the contrasting results technique for the var