Application of Weighted Time-Series to Address Bias in Evaluation of Clinic- and Community-Level Research

This study will use simulation to develop, test and apply new analytic methods (weighted time-series) for evaluation of community-level interventions. It will then compare results using weighted time-series and conventional methods within the context of a clinic-level intervention to provide family-centered HIV care, voluntary counseling and testing (VCT) and prevention services at Family AIDS Care and Education Services (FACES), a community-based organization in Kenya. Because FACES includes observational data on virtually all patients in care at participating clinics, it provides an excellent platform to evaluate the effectiveness of this intervention using both cohort and time-series methods. The results of this study will be used to seek funding to test the broader application of these methods in both community- and clinic-level interventions. The specific aims of the proposed project are:

• To provide the rationale and framework for applying weighted time-series to serial cross-sectional data.
• To use simulation (created data) to apply and test the use of weighted time-series in a setting where the distribution of demographic characteristics and the health status of the population changes over time.
• To use existing clinical data to compare the effect of introducing family-centered HIV care, VCT and prevention services on the transmission of HIV among the families served by participating clinics using cohort analysis, time-series analysis and weighted time-series.