Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Public Health

First Advisor

Dr. Richard Rothenberg

Second Advisor

Dr. Heather Bradley

Third Advisor

Dr. Duncan MacKellar


Mozambique ranks third in the number of persons living with HIV (PLHIV) worldwide, with 2.1 million cases. HIV prevalence is higher in southern Mozambique and highest in Gaza Province, at 24.4% among adults. In 2010, the Chókwè Health Research and Training Centre established the Chókwè Health and Demographic Surveillance System (CHDSS) with ongoing health surveillance coverage of a catchment area including Chókwè city and seven surrounding villages in rural Gaza Province.

From 2014 to 2019, the CHDSS conducted five consecutive CHDSS-wide testing campaigns (rounds, R) consisting of home-based rapid HIV testing and counseling (HTC) offered to all residents and an in-depth HIV Prevention Survey additionally offered to residents in 14% (R1-R2) to 23% (R3-R5) random samples of households.

Using these data, this dissertation aims to address the following three research questions:

1: Compare methods of estimating HIV incidence, including: i) LAg-Avidity assay-based; ii) UNAIDS Goals model; iii) prospective cohort of home-based HTC repeat participants; iv) retrospective cohort of persons with self-reported HIV negative test history; v) time since age of sexual debut; and, vi) self-reported testing history model describing time from infection to diagnosis.

2: Demonstrate increased viral load suppression (VLS) and decreased HIV incidence after five rounds of home-based HTC.

3: Characterize misclassification of HIV outcomes and demonstrate misclassification’s impact on estimates of HIV outcomes.

Declining point estimates were observed for all six methods and were significant for all methods except one. Three methods yielded estimates that were consistently similar in magnitude and trends, suggesting their aptness for incidence estimation in the CHDSS. After five rounds of home-based HTC with intensive linkage services, significant and substantial increases in VLS and a 69% reduction in HIV incidence were observed. And, despite exhaustive efforts to identify and remedy misclassification, the volume of misclassification in this study was relatively low. Estimates using reclassified data were similar to original estimates.

These analyses demonstrate the complementary use of multiple incidence estimation methods, suggest the potential of home-based HTC and linkage to reduce incidence, and highlight minimal misclassification and robust study outcomes in the CHDSS.


File Upload Confirmation


Available for download on Tuesday, December 10, 2024