Article type
Year
Abstract
Background:
Public health interventions target a diverse population, composed of multiple components, mostly context-dependent and, therefore, they are often regarded as complex. The complex nature of public health interventions discourages the use of conventional meta-analytic techniques and acts as a hurdle to combine the findings of studies conducted across different locations. Measuring complexity of different studies and adjusting for these complexities during meta-analysis would be an appropriate method to address this problem and obtain the pooled estimate.
Objectives:
Development of a tool for quantifying the complexities in public health intervention studies.
Methods:
With the help of a large number of published public health interventions, a check-list was prepared for compiling the instrument. The check-list was obtained for all the domains, namely, population, intervention, context and outcome. This was further debated and cross-checked with experts to form the final list. Items in the final list were appropriately scored for each domain based on long discussions, debates and expert meetings and were incorporated into the tool. The reliability of the tool was assessed by two raters independently scoring the same studies with the developed tool and computing the Intra-class correlation coefficient (ICC). Furthermore, the tool was applied to a number of studies to measure domain-specific and overall complexities.
Results:
The ICC for population, intervention and outcome was found to be good (0.855, 0.859 and 0.916 respectively) and ICC for context was found to be low (0.381). The tool was used to measure the complexities of population, intervention, context and outcome of 26 public health intervention publications.
Conclusions:
The tool has a tremendous scope for application in meta-analysis in public health systematic reviews. Future researchers need to consider more relevant items and may update the scoring pattern adopted.
Public health interventions target a diverse population, composed of multiple components, mostly context-dependent and, therefore, they are often regarded as complex. The complex nature of public health interventions discourages the use of conventional meta-analytic techniques and acts as a hurdle to combine the findings of studies conducted across different locations. Measuring complexity of different studies and adjusting for these complexities during meta-analysis would be an appropriate method to address this problem and obtain the pooled estimate.
Objectives:
Development of a tool for quantifying the complexities in public health intervention studies.
Methods:
With the help of a large number of published public health interventions, a check-list was prepared for compiling the instrument. The check-list was obtained for all the domains, namely, population, intervention, context and outcome. This was further debated and cross-checked with experts to form the final list. Items in the final list were appropriately scored for each domain based on long discussions, debates and expert meetings and were incorporated into the tool. The reliability of the tool was assessed by two raters independently scoring the same studies with the developed tool and computing the Intra-class correlation coefficient (ICC). Furthermore, the tool was applied to a number of studies to measure domain-specific and overall complexities.
Results:
The ICC for population, intervention and outcome was found to be good (0.855, 0.859 and 0.916 respectively) and ICC for context was found to be low (0.381). The tool was used to measure the complexities of population, intervention, context and outcome of 26 public health intervention publications.
Conclusions:
The tool has a tremendous scope for application in meta-analysis in public health systematic reviews. Future researchers need to consider more relevant items and may update the scoring pattern adopted.