In this enumeration, a two-sample capture-recapture calculation was used to estimate that 6,904 sex workers were operating in the Nairobi CBD in April 2009. This is among the first robust enumerations of this important most-at-risk population (MARP) in Nairobi, which has important implications for planning purposes. Moreover, geographic data demonstrate close proximity between our sex worker outreach program and sex work hot spots, which themselves cluster in a relatively small area. Location of HIV/STI prevention and care services near sex workers' places of business is likely an important variable for them to access these services regularly, as has been shown for MARPs on long-distance truck routes in East Africa 
Two samples of sex workers operating at 137 mapped hotspots were drawn during peak sex work; considering the large number of enumerators, and the size of teams ranging from three to five members, assigned at most three hotspots that were adjacent to each other, we predict that hotspot coverage was adequate. This arrangement aimed to minimize the chances of missing the target population working in the CBD at the time of enumeration. Furthermore, we have scaled up this exercise for a recent World Bank-funded initiative to enumerate sex workers across Kenya, and found this to be very feasible, even at a large scale, at relatively low cost (unpublished).
For capture-recapture estimates to be valid, certain conditions need to be met: the target population must be closed; the two samples must be independent; recaptures must be correctly identified; the probability of being captured during both rounds equal; and that the people captured must belong to the target population. Although the target population in this study is mobile, little change was anticipated given the short interval between samples. The sample locations did not change, and therefore the samples were independent. Ensuring correct identification of recaptures prevented overstatement of the actual number of recaptures that would have led to an underestimate of the target population. SW counted the first day may have been more likely to be included on the second day, due to familiarity with enumerators. An increase in the recaptures and subsequently a reduction of the estimate may have been balanced by SW who refused or were too busy to receive enumeration cards on the first day, but included on the second day. Double counting on both days was minimized by not giving incentives that would have facilitated movements of SW from one hotspot to another, and by establishing whether a card had been received from another enumerator that evening. Even though peer leaders ensured that the people counted were SW, a few individuals might have received the enumeration cards by mistake, possibly because they expected to gain in return.
By the end of 2011, we recruited and enrolled 6,572 female sex workers to the SWOP-City clinic, in the same area where this enumeration took place (). Based on this enumeration in 2009, it appears nearly 95% of sex workers in this area are being reached. However, this clinic draws attendees from many areas of Nairobi. SW who access this clinic receive free education and risk reduction counseling, condoms, and HIV/STI testing and care. Therefore, although coverage of the Nairobi CBD is dramatically improving, it is likely that there are still more sex workers who need to be reached. Furthermore follow-up of enrolled SW is important to maximize the impact of HIV/STI prevention messages. To ensure access to STI/HIV services has been improved for sex workers in Nairobi County, six dedicated clinics were opened in 2011.
Cumulative enrollment of female sex workers at the SWOP-City Clinic in the Nairobi Central Business District, Aug 2008 until Oct 2011.
The current study did not use any male sex worker peer leaders to identify male SW. This probably explains why only five percent of the total estimated SW population is male, despite belief that there are many male sex workers in the CBD. This also implies that even though reaching out to SW is generally difficult, male SW may be even more difficult to reach. Through a more targeted approach, we have enrolled 463 male SW in SWOP-City clinic by the end of 2011. The proportion of the actual Nairobi CBD male SW population this represents is difficult to ascertain. Another limitation is that sex workers are a mobile population, and the methodologies used here might not capture this challenging aspect of population estimation. Future studies should devise better strategies to take this important (and potential risk) factor into account.
Finally, it should also be appreciated from a map of Nairobi that the area covered represents a small proportion of the total metropolitan area, and that there are at least half a dozen other districts around town that contain a large number of entertainment bars and clubs, or where sex workers are known to work in large numbers. Indeed, preliminary data from an ongoing mapping exercise suggests that the total number of SW in greater Nairobi is several times that in the CBD. However, the methodology utilized for this study seemed to work well, and could be easily replicated in other locations in Nairobi or elsewhere. There is a need to accurately estimate the size and location of this high-risk population since it has important implications for the design of successful HIV/STI prevention programs.