Table 1 Model Input parameters and distribution types used in uncertainty analysis
Definition of model inputs | Model inputs (boundaries or in italics: 95% CI) | Probability Distribution | Date of behavioural survey and reference for model input values |
|---|---|---|---|
Coverage and Impact | Â | Â | Â |
Overall number of sex workers in Ahmedabad | 4000 (3500–4500) | Triangular | Estimated by staff and peer educators of Jyoti Sangh through snow-balling method. Also personal communication with GJ on 7 July 2005. |
Proportion of CSW recently reached by Jyoti Sangh intervention | 0.586 (0.521–0.669) | Triangular | Average number of CSWs covered by the intervention program. Number covered/Total number = 2342/4000 = 0.5855 Max.: Number covered/min. total number = 2342/4500 = 0.5205 Min: Number covered/max. total number = 2342/3500 = 0.6692 (According to the routine data collected in monthly reports between Jan 2001 and Dec 2003) |
Proportion of CSW reached using STD services per month | 0.028 (0.028–0.041) | Uniform | Average of percentage of STD treatments (Female) in those CSWs covered by the programme who are believed to have STIs (Low estimate: 51.6%; High estimate: 75.5%; figures from [6] calculated from the percentage of CSW who do not have any STIs) |
Average proportion of STDs treated effectively cured | 0.8 (0.7–0.9) | Triangular | No data, treatment is assumed to be fairly effective. |
Proportion reporting using condoms: | Â | Â | Intervention survey data from 1999 and 2003 [4]. 95% CI confidence intervals calculated using STATA. |
Before/Unreached: | Â | Â | Â |
Never | 0.38 (0.33–0.44) | Normal |  |
Sometimes | 0.29 (0.24–0.34) | Normal |  |
Always | 0.33 (0.27–0.37) | Normal |  |
After/Reached: | Â | Â | Â |
Never | 0.047 (0.028–0.073) | Normal |  |
Sometimes | 0.14 (0.11–0.18) | Normal |  |
Always | 0.80 (0.75–0.84) | Normal |  |
Epidemiological | Â | Â | Â |
Initial HIV prevalence amongst sex workers (first survey from 1999) | 11.7% (8.4%–15.9%) | Normal | Survey data from 1999 [4] (Lower and upper estimates are 95% confidence intervals of the point estimates). |
Average duration of generic STD included in model (months) | Â | Â | Varied in order to fit the STD prevalence in 1999 and 2003 surveys [6]. |
CSW | 1 (0.5–1.5) | Uniform |  |
CL | 1 (0.5–1.5) | Uniform |  |
Average duration of initial high infectivity phase (months) | 1.5 (1–2) | Triangular | [10] |
Average median duration between HIV infection and morbidity (months) | 95 | Constant | Median disease progression from HIV infection to AIDS takes 7.9 years in a cohort study in Mumbai, India. This was drawn from a truncated Weibull distribution [36]. |
Behavioural | Â | Â | Â |
Average time span women sell sex (months) | (90–180) | Uniform | In the 2003 survey, average age of CSW = 33 years, average age when became CSW = 25.4 years [6]. Difference was used to estimate the lower bound, and the upper bound was set to twice the lower bound. |
Average time span men buy sex (months) | (60–120) | Uniform | Median age (range) of CL is 28 (18,49); Median age at first sex with any commercial female partner (range) is 23 (10,40) [37]. Difference was used to estimate the lower bound, and the upper bound was set to twice the lower bound |
Average number of clients per month per unreached CSW (using 1999 figure) | 133 (119–157) | N/A | Average number of sexual partners per day during last month [6] and data from routine monthly reports between May 2002 and December 2003 were used to calculate an average. |
Average number of clients per month per reached sex worker (using 2003 figure) | 119 (107–140) | Triangular | Monthly reports of Jyoti Sangh from May 2002 to Dec 2003. |
Average additional number of clients per month per unreached CSW | 14 (11.6–17.0) | Uniform | Difference between the two figures above, calculated using Solver, using percentages from the two surveys [6]. |
Number of sex acts between one client and one sex worker in one encounter | 1 | -- | No data. It is assumed to be near one, and that after discussion with GJ it was decided to be one. |
Average number of CSWs seen by clients per month | (1–8) | Uniform | NACO survey reported median number of commercial female partners seen by a client in the last 3 months is 6 (1,27) [37]. However, high estimates of this parameter result in very high STI prevalences and so a lower range was used for modeling. |
Proportion of time condom used, corresponding to: | Â | Â | Â |
"None of the time" | 0 (0–0.2) | Triangular | Zero was chosen as point estimate as it is the conservative estimation based on the definition "None of the time". |
"Some of the time" | 0.5 (0.2–0.7) | Triangular | Question framed in the questionnaire is 2 or 3 out of 5. So mean of 40% and 60% is 50% [6]. |
"All of the time" | 0.8 (0.7–1) | Triangular | Question framed in the questionnaire is > 3 out of 5. So mean of 60% and 100% is 80% [6]. |
Transmission Probabilities | Â | Â | Â |
Transmission probability of HIV per sex act: | Â | Â | [12, 14, 38] |
Male to Female | 0.002 (0.001–0.003) | Uniform |  |
Female to Male | 0.001 (0.0005–0.003) | N/A |  |
Ratio of transmission probability: (Female to Male)/(Male to Female) | 0.5 (0.5–1) | Triangular |  |
Transmission probability of generic STD per sex act: | Â | Â | [13, 15-20] |
Male to Female | 0.25 (0.1–0.5) | Triangular |  |
Female to Male | 0.25 (0.1–0.5) | Triangular |  |
STD cofactor effect per sex act | 3.1 (1.2–5) | Triangular | [8, 9] |
Multiplicative cofactor during high viraemia phase of HIV infection | 15 (10–20) | Triangular | [21, 22] |
Proportion of time condom used that provides protection | 0.85 (0.8–0.9) | Triangular | [23, 24] |