Table 1. Contributions of proximate causes to the HIV decline in Zimbabwe.

Proximate Cause	Population-Level Effectivenessa	Extent of Changeb	Consistency in Timing of Changec	Major Contribution
Behavioral
Age at first sex - postponement	Low	Low [QN]	Consistent	Unlikely
Partner numbers - reduction	High	High [QN & QL]	Consistent	Likely
Condom use - increase (in non-marital partnerships)	High (if consistent use)	Moderate [P, QN, QL]	Earlier	Plausible
Biological
Transmission probability - reductiond	High	Low [QN & P]	Earlier	Unlikely

Form(s) of evidence supporting conclusion: M, epidemiological modeling; QN, survey data; P, program data; QL, qualitative data. See Text S1 for details.

^aExtent to which changes in the factor concerned can reduce HIV transmission at the population level, as measured and modeled in scientific studies [M & QN].

^bExtent to which changes in the given behavioral or biological determinant (by population sub-group) have occurred as observed in longitudinal surveys and/or program data.

^cExtent to which the changes in risk behavior etc. occurred during the period of most rapid reduction in risk as determined by the epidemiological modeling assessment (i.e., about 1999–2003).

^dTransmission probability could be affected by, for example, levels of blood safety, prevalence of other sexually transmitted infections, HIV medications, or male circumcision.