Validity of data from prevention of mother to child HIV transmission programme in estimating general population HIV prevallence in Mbeya region, Tanzania 2003-2008

Abstract

Background: Monitoring trends in HIV -1 infection is crucial for planning and evaluation of intervention measures. ANC sentinel surveillance data have been used for that purpose; however, they require significant resources to operate and are biased. Recent literature shows that data from Prevention of Mother to Child HIV Transmission (PMTCT) programmes could be a cheaper and comparable source of data for monitoring HIV -1 trends. However, its validity as compared to ANC sentinel surveillance data has not been well examined. Objective: The objective of this study was to assess the validity of PMTCT programme data as a proxy measure of the general population HIV -1 prevalence estimates compared to the ANC sentinel surveillance data. Materials and Methods: This was a cross-sectional study which involved a review of facility-based data from ANC PMTCT registers. Comparison of age-adjusted HIV-l prevalence estimates were made between ANC sentinel surveillance, PMTCT programme and the Tanzania HIV Indicator survey of 2003-2004 together with the Tanzania HIV and Malaria Indicator survey of 2007-2008. Comparison was made for the same geographical area and time period, i.e 2003-2008. Overall and age-specific HIV -1 prevalence estimates were compared between women in the ANC sentinel surveillance and those who participated in the PMTCT programme, both for the overall study period and separately for each year. Results: In total, 6390 records were retrieved from ANC PMTCT registers in nine PMTCT sites during the study period 2003-2008. Most women in the study population were in the age group 15-24 years (Mean age 25 years, SD: 5.7). Overall HIV-1 PMTCT-based prevalence estimates from 2003-2006 were comparable to HIV -1 prevalence estimates based on ANC sentinel surveillance (p=0.89 and p=0.43 for the year 2003/2004 and 2005/2006 respectively). For 2007/2008, the HIV -1 prevalence estimate from PMTCT programme and that from the ANC sentinel surveillance were significantly higher than the estimate from the population- based survey (p=O.OO for both PMTCT and ANC sentinel surveillance). Women attending clinics at urban and border areas were more likely to be HIV -1 infected than those who attended clinics located at rural areas (OR=1.35, p=O.OO, OR=I.99, p=0.04 respectively). HIV- 1 testing uptake was more than 99%. There was a notable varying data quality across the sites. Conclusions and Recommendations: HIV -1 prevalence estimates from PMTCT programme were comparable to those from the ANC sentinel surveillance as well as to estimates from the general population. However, for the period 2007/2008, estimates from PMTCT programme and the ANC sentinel surveillance significantly overestimated the estimates in the general population. The HIV testing uptake in the PMTCT environment was very high. PMTCT data was of moderate data quality, Behavioral change HIV programmes in Mbeya region should target urban and border areas and male participation in PMTCT programmes should be encouraged. Improvement of data can be done by standardizing data collection tools, regular training and supervision. PMTCT programme has a potential to replace ANC sentinel surveillance for monitoring of HIV epidemic trends in Mbeya region.