Table 2.

Description of sampling methods used for the modes of transmission revised model sensitivity analysis, including data sources and rationale.

Parameter	Method of sampling uncertainty
Population sizes	Population size estimates were sourced from the National HIV/AIDS and Reproductive Health Survey Plus (NARHS+) surveys for general population subgroups. Details on high-risk group population estimates and additional calculations are contained in appendix D, with most of these estimates conserved from the original model. Owing to the high levels of uncertainty in the size of the population estimates we sampled each parameter relatively ±50%.
HIV prevalence	For all HIV prevalence values estimated from survey data for Cross River state or regional estimates from NARHS+ 2007, 95% confidence interval (CI) intervals were generated and applied in the sensitivity analysis. These included all risk-groups with the exception of women engaging in transactional sex and their partners, clients of brothel-based and non-brothel-based female sex workers (FSW) and their regular female partners. Because of the greater level of uncertainty in these estimated values each parameter was varied ±50% relative to its original value.
STI prevalence	All estimates were generated from survey data for which 95% CI were available. For general population groups, the estimate was taken from the 2003 sentinel survey for individuals tested for syphilis. From this, 95% CI were derived based on the sample size for the south-south geopolitical zone. Sexually transmitted infection (STI) prevalence estimates for high-risk groups and clients were generated from the Integrated Biological and Behavioural Surveillance Survey (IBBSS) 2010 for reported genital ulcer/sores. This estimate was selected because no serological data were available and as it was considered that genital ulcers/sores present as more visible forms of STIs and are more likely to reflect the presence of syphilis. 95% CI were generated from the data to reflect the uncertainty in these estimates.
Number of partners and number of sex acts with individual partners per year	To allow for the uncertainty in the number of partners and number of sex acts with each partner for individual subgroups, all parameter estimates were varied ±50%. This was to reflect the uncertainty in these estimates and the fact that some estimates were generated through triangulation methods based on population size estimates, whereas others were based on assumptions from the original MoT model.
Condom usage	Estimates for condom use for clients of brothel-based and non-brothel-based FSW come from the DHS for men who report ‘payment for sex’. The same value was assumed for brothel-based and non-brothel-based FSW, as FSW in the Integrated biological and behavioural surveillance survey (IBBSS) survey report 100% condom use and this estimate was deemed to be unrealistic. Estimates for condom use in men who have sex with men (MSM) and sterile needle use for injecting drug users (IDUs) are taken from the IBBSS 2007 and all other subgroup estimates were generated using appropriate estimates from the NARHS 2007+, taking 95% CI on each occasion to generate uncertainty ranges for data sampling.