Secondary Logo

Journal Logo


Empirical Observations Underestimate the Proportion of Human Immunodeficiency Virus Infections Attributable to Sexually Transmitted Diseases in the Mwanza and Rakai Sexually Transmitted Disease Treatment Trials: Simulation Results

Orroth, Kate K. PhD*; White, Richard G. MSc*; Korenromp, Eline L. PhD; Bakker, Roel PhD; Changalucha, John MSc; Habbema, J Dik F. PhD; Hayes, Richard J. DSc*

Author Information
doi: 10.1097/01.olq.0000204667.11192.71
  • Free

IT IS WELL ESTABLISHED that sexually transmitted diseases (STD) contribute to the spread of human immunodeficiency virus (HIV), especially in populations where STDs are highly prevalent as in sub-Saharan Africa (SSA). Biologic mechanisms for enhancement of HIV acquisition in the susceptible partner and HIV transmission from the infected partner in HIV-discordant couples have been established.1, 2 However, 3 community randomized trials of treatment of STDs as a strategy to prevent HIV transmission showed contrasting results. Improved case management of STDs using the syndromic treatment (ST) approach reduced HIV incidence by about 40% in the general population in Mwanza, Tanzania.3 In contrast, trials of STD mass treatment (MT) in Rakai, Uganda, and of ST combined with an information, education, and communication (IEC) intervention in neighboring Masaka District, Uganda, showed little impact on HIV incidence.4,5

Several hypotheses have been suggested to explain the contrasting results of the Mwanza and Rakai trials.6 One hypothesis was that curable STD played a more important role in HIV transmission in Mwanza than Rakai. In an effort to test this hypothesis, population attributable fractions (PAF) for the effect of STDs on HIV incidence were estimated for the Mwanza and Rakai trial populations based on associations between markers of STD (self-reported symptoms and serology) and HIV incidence.7,8 Results from these analyses showed that the estimated proportion of HIV infections attributable to STDs in Mwanza was higher than that for Rakai. In addition, the PAF for the effect of STDs on HIV spread was higher in the comparison arm of the Mwanza trial (22%) compared to the intervention arm (3%) (Table 1).8 These results supported the interpretation of the trial outcome, that the observed reduction in HIV incidence in the Mwanza ST arm was mainly due to a shortening of the duration of STD infections and the associated cofactor effect on HIV.3 In Rakai, the difference between the intervention and comparison arms in the proportion of HIV infections attributable to STDs (12% in the comparison arm and 7% in the intervention arm) was less than that for Mwanza, and this would explain why STD reductions there did not reduce HIV incidence.7

Selected PAF Results for Mwanza and Rakai Trials From Observational Data

However, the methods of PAF calculations used in the 2 trials differed due to differences in STD measurements, data collection, and follow-up time so that the empirical PAFs were not strictly comparable between the 2 sites (Table 1). Due to limitations in the measurement of STD and HIV coinfections, such association-based PAFs may either over- or underestimate the true effects of STDs on HIV transmission for several reasons. First, confounding due to sexual risk behavior (which cannot be measured accurately) is likely to result in overestimating the PAF. Second, in the Mwanza and Rakai studies, exposure to STDs was based on self-reported, symptomatic STD, which would result in underestimating the overall PAF. Third, in these cohort studies it was only possible to estimate the proportion of HIV infections attributable to STD effects on HIV acquisition because it was not possible to identify those who transmitted HIV due to the presence of an STD (except for stable couples in Rakai, where the latter effect was indirectly estimated, Table 1). This again would result in underestimating the PAF. Finally, statistical associations between STD and HIV cannot capture the indirect effects of cofactor STDs on the onward transmission of HIV in the population, including among people without STDs.9–12

We have previously simulated both trials in a dynamic model that describes the spread of STD and HIV in interaction. The STDSIM model13,14 was quantified to replicate the demographic, behavioral, and epidemiologic characteristics of both trial populations. These simulations confirmed that differences between the populations, notably, higher sexual risk behavior and prevalence of curable STDs in Mwanza15,16 and a more generalized HIV epidemic in Rakai at the time of the trials, could to a large extent explain the larger impact of STD treatment on HIV incidence in Mwanza compared to the Ugandan sites.17,18

As a complement to the simulations of trial intervention impacts, the objective of this paper was to calculate from the same simulations the “true” PAFs of HIV incidence associated with cofactor STDs, irrespective of the fraction of these STD that was actually reached by the interventions. Simulated PAFs, denoted by PAFM, were estimated for different time points into the respective HIV epidemics to capture the dynamic effect of STDs on HIV spread over time. Finally, the effect on HIV incidence of treating each individual STD was investigated for both trial populations to explore how the MT intervention in Rakai and the ST intervention in Mwanza prevented HIV infections.

Materials and Methods

STDSIM Transmission Model and Simulated Populations

The STDSIM transmission model simulates the natural history and transmission of HIV, chancroid (HD), syphilis (TP), herpes simplex virus type-2 (HSV-2), gonorrhea (NG), chlamydia infection (CT) and trichomoniasis (TV).13,19 It is a microsimulation model in which the life histories of individuals and the interactions between them are explicitly modeled. Sexual contacts and relationships form a network through which STD can be transmitted, and the formation and dissolution of partnerships, as well as disease transmission, are modeled as stochastic events.14

STDs are represented in the model as a series of stages through which individuals will pass during an infection. The stages are based on the natural history of infection and vary about susceptibility, infectivity, duration, symptomatology, and associated STD cofactor effects. Individual sex acts are simulated so that transmission probabilities and cofactor effects associated with a given STD disease stage are applied to the specific sexual contacts that a simulated individual has with any partner during that stage.19

The simulations of the HIV epidemics and STD trials in Mwanza and Rakai have been documented in detail previously.17,18 In brief, we modeled higher-risk sexual behavior when HIV was introduced in Rakai (1978) than in Mwanza (where HIV was introduced in 1983). After the end of the Ugandan civil war, in the mid-1980s, we assumed declining sexual risk behavior in Rakai so that at the time of the trials, sexual risk behavior and curable STD prevalences were higher in Mwanza compared to Rakai, in accord with observations.16,20 The biologic parameters concerning STD natural history and transmission were based on literature reviews and were held constant across the 2 model quantifications. These parameters have been discussed in detail elsewhere17 and are summarized in Table 2.

Fitted Values for Representation of Natural History and Transmission of HIV and Sexually Transmitted Diseases for Rakai and Mwanza Scenarios

The STD cofactor effects were also held constant across the 2 model quantifications. STD were assumed to enhance HIV infectivity and susceptibility by the same factor at the per-contact level. For HD and primary HSV-2, the cofactors were the highest, 25, which is at the low end of the range (10–300) estimated from studies of commercial sex workers and clients in Nairobi for genital ulcer disease.21 We selected values at the low end of the range to reflect the impact of residual confounding on these estimates.10 The cofactor effects for the other STDs were less than those for HD and primary HSV-2 (Table 2). Also, if more than 1 cofactor was present, HIV infectivity or susceptibility was only increased by the highest cofactor, not by a combination of the cofactors because it is unclear how cofactors combine to increase transmission.11 Model results are based on the average of 200 simulations per scenario.

Simulated Proportions of HIV Infections Attributable to STDs in the Absence of STD Treatment Interventions

The default scenario included the assumed STD cofactor magnitudes as shown in Table 2. To estimate the proportion of HIV infections attributable to each STD from the STDSIM model (PAFM), we compared the trial simulations with counterfactual scenarios in which the STD cofactor effects for HIV acquisition (i.e., the increased susceptibility of the HIV-negative partner) and for HIV transmission (i.e., the increased infectivity of the HIV-positive partner) were removed. By removing in turn, or in combination, the susceptibility and infectivity cofactor effects, these simulations allowed us to disentangle the relative effects of STDs on HIV acquisition and on HIV transmission. Cofactors were removed starting in 1992 for the Mwanza scenario and 1994 for the Rakai scenario, corresponding to the start of the trials. Scenarios included the removal of the cofactor effect for all STD together, for all curable STDs together (i.e., except HSV-2), and for each STD individually.

The proportion of HIV infections attributable to each STD and the combination were calculated based on HIV incidence reductions using the following formula:

where PAFM is the “true” or modeled proportion of HIV infections attributable to STD cofactor effects, IRnocofactor is the HIV incidence rate in the various counterfactual scenarios with the cofactors removed, and IRdefault is the HIV incidence rate in the default scenario with all cofactor effects included. The incidence rates were calculated as the average adult (15–54 years) incidence over 2 years (1993 and 1994 in Mwanza and 1995 and 1996 in Rakai).

We also investigated how the PAFMs changed over the time course of the respective HIV epidemics in Mwanza and Rakai in the absence of the STD treatment interventions, 2 years (1980 and 1985), 10 years (1988 and 1993), and 20 years (1998 and 2003) into the simulated Rakai and Mwanza HIV epidemics, respectively.

Sensitivity Analysis

In order to gauge the uncertainty in our model scenarios, we assessed which model parameters most strongly affected the simulated estimates of the proportion of HIV infections attributable to STD. In this analysis, model parameters were varied one at a time, and the resulting PAFM was compared to that for the default scenarios. As each parameter was varied, the HIV transmission probabilities were refitted to maintain the correct observed HIV prevalence. Parameter changes investigated included doubling and halving the magnitude of all cofactor effects, those for ulcerative STDs (HD, HSV-2, and syphilis) alone, inflammatory STDs (NG, CT, and TV) alone, infectivity cofactor effects, and susceptibility cofactor effects. Due to the high PAFMs associated with HD, additional simulations increasing and decreasing the HD prevalence were included by increasing and decreasing the transmission probabilities for HD. The outcomes evaluated were the PAFMs for CT, HD, HSV-2, and all curable STD because these STDs had the greatest impact on HIV spread in the simulated populations.

Impact of STD Treatment on HIV Incidence

For both Mwanza and Rakai, before the trials and in trial comparison arms (the “default scenario”) a low-level of ST was simulated, which resulted in about 5% of symptomatic STD being cured (apart from HSV-2). The Mwanza intervention scenario included increased effectiveness and coverage of ST from 1992 onwards, which resulted in about 25% of symptomatic STD being cured, apart from HSV-2.22,23 For Rakai, 2 rounds of MT took place in 1994 and 1995, in which 70% of adults received a treatment that was 95% efficacious against curable STD.24

In order to estimate what proportion of the HIV incidence reduction was due to each STD treated, simulations were performed in which the improved ST or MT of each curable STD was added individually to the default scenario. The reduction in HIV incidence for each treatment scenario was then compared to that for the intervention scenario, including treatment for all curable STD, to determine the impact of treating each STD on HIV incidence.


Simulated Proportions of HIV Infections Attributable to STD

The simulated proportions of HIV infections attributable to STD cofactor effects for the Mwanza and Rakai baseline scenarios are shown in Figure 1.

Fig. 1
Fig. 1:
Proportion of HIV infections attributable to STD cofactor effects in Mwanza and Rakai trial populations (15–54 years) during the 2 years of each trial. Simulated proportional reductions in HIV incidence when STD cofactors were removed from default scenarios. PAFMs for adults aged 15–54 years. NG = gonorrhea; CT = chlamydia infection; TV = trichomoniasis; TP = syphilis; HD = chancroid. Mwanza left, Rakai right.

For Mwanza, the largest PAFM for the effect of any single STD on HIV incidence was 40% for HD. The PAFM for the individual effects of other curable STD on HIV were lower (5% to 10%), and that for HSV-2 was 12%. The total PAFM was 65% for all curable STD together and 77% for all STD including HSV-2. For the curable STDs combined, the PAFM for acquiring HIV (23%) was lower than for transmitting HIV (36%). This is because STD prevalence is higher among HIV-positive partners than HIV-negative partners (the former group has higher risk behavior). In higher-risk HIV-positive partners, a given single STD episode would contribute to more HIV transmissions than it would in an initially HIV-negative, lower-risk partner. The PAFM for the effect of any curable STD on HIV was similar to the sum of the PAFMs for the effect of individual STDs. The PAFM for overall spread was slightly larger than the sum of the PAFMs for HIV acquisition and transmission (77% compared to 69%). This suggests that (at least in an early epidemic), STD cofactor effects have a dynamic indirect effect of amplifying the overall intensity of onward HIV transmission.

In contrast, in Rakai, the largest PAFM for the effect of any single STD on HIV was 23% due to HSV-2. The proportions of HIV infections attributable to HD and chlamydia were about 6%, and those for NG, TV, and syphilis were less than 5%. Twenty percent of new HIV infections were attributable to cofactor effects of all curable STD combined, about a third of that for Mwanza at the start of the trials. When including the cofactor effect for HSV-2, the proportion of HIV infections attributable to any STD doubled to 42%. The overall PAFMs were somewhat higher than the sum of those for acquisition and transmission (42% compared to 36%).

Changes in the Proportion of HIV Infections Attributable to STDs Over the Course of the HIV Epidemic

Figure 2A shows how the simulated proportion of new HIV infections attributable to STD changed over the time course of the Mwanza and Rakai HIV epidemics. Results are included for chlamydia, HD, and HSV-2. The results reflect the effects of STD on both HIV acquisition and transmission (overall effect). In both populations, PAFMs of curable STD decreased over the time course of the HIV epidemic (Fig. 2B), as expected because HIV spreads out of the high-STD core groups into the general population, and HIV-attributable mortality among high-risk individuals reduces population prevalence of STDs.25,26

Fig. 2
Fig. 2:
Simulated proportions of HIV infections attributable to STDs (PAFMs) over the course of the Mwanza and Rakai HIV epidemics. Simulated proportional reductions in adult (15–54 years) HIV incidence when cofactors were removed 2, 10, and 20 years into the epidemics. A, PAFMs for individual STD. B, PAFMs for HIV acquisition, transmission, and overall spread due to curable STD. CT = chlamydia infection; HD = chancroid; HSV-2 = herpes simplex virus type-2. Mwanza left, Rakai right. Vertical dashed lines represent start of the trials.

The reduction in the PAFM was steepest for HD. This is because the reproductive number for HD is lower than for other STDs,27 so small reductions of risk behavior (due to risk reduction or selective HIV mortality) cause HD prevalence to fall considerably. For Rakai, the prevalence of HD initially increases as a result of the modeled increase in partner change rates during Uganda's civil war between 1979 and 1986 but then falls while HIV spreads widely during the late 1980s (Fig. 2A). For Mwanza, the PAFM for HD decreases over time because HD infection remains largely restricted to high-risk core groups in which, as HIV spread progresses, fewer new HIV infections take place due to saturation (Fig. 2A). Comparing the 2 populations, early in the epidemics the PAFMs of curable STD were equally high in Mwanza and Rakai (Fig. 2B). In Rakai, PAFMs for curable STDs fell slightly faster and farther (especially for HD), due to the behavioral risk reduction and AIDS-related mortality in the second decade of the epidemic.

The proportion of HIV infections attributable to HSV-2 was relatively constant or increased somewhat over time but remained low in both sites. The prevalence of HSV-2 is high even among low-risk individuals (about 50% in adults for both Mwanza and Rakai), i.e., the reproductive number is high,27 but the duration of the cofactor effect for primary HSV-2 is relatively short (3 weeks for primary herpetic ulcers compared to 10 weeks for HD, Table 2) and HSV-2 recurrences are even shorter, with a smaller cofactor effect. As HIV spreads from the high-risk groups to the general population, the importance of HSV-2 for HIV spread increases because of its high prevalence; this effect is especially apparent in Rakai, where the HIV epidemic was more generalized compared to Mwanza (Fig. 2A).

Similar patterns of decline in PAFM during the HIV epidemic were seen for the separate STD cofactor effects on HIV acquisition and on transmission (Fig. 2B). In Mwanza, curable STDs continued to contribute more to HIV transmission than acquisition at any time during the HIV epidemic. In Rakai, 20 years into the HIV epidemic, the contribution of curable STD to HIV spread was very low for both acquisition and transmission due to the low-risk behavior in the Rakai population at that time.

Sensitivity Analysis

Results from the sensitivity analysis are shown in Table 3.

Sensitivity Analysis for Mwanza and Rakai Scenarios

The magnitude of inflammatory and infectivity cofactor effects and the prevalence of HD had the largest impact on the proportion of HIV infections attributable to curable STD in Mwanza and Rakai. If inflammatory cofactor effects were doubled, the proportion of HIV infections attributable to CT in Mwanza increased from 7% to 15% and that for curable STDs, from 65% to 71%. In Rakai, the PAFM for curable STDs increased from 21% to 32%. If we assumed the ulcerative cofactors were halved (12.5 for HD and primary HSV-2 instead of 25), the PAFM for the effect of HD in Mwanza decreased from 39% to 30%, and that for HSV-2, from 11.9% to 6.6%. Reducing the HD prevalence (from 1% to 0.2% in Mwanza and 0.3% to 0.16% in Rakai) reduced the PAFM for curable STD from 65% to 52% in Mwanza and from 21% to 18% in Rakai. A similar reduction was observed when halving the infectivity cofactor effects.

When a low cofactor effect for HSV-2 infection was included for latent infection to simulate potential effects of HSV-2 on HIV transmission during the periods between ulcerative recurrences,28 the PAFM of the effect of HSV-2 on HIV was increased from 12% to 32% in Mwanza and from 23% to 53% in Rakai.

Impact of Each STD Treated on HIV Incidence

The majority of the simulated impact of ST in the Mwanza scenario was due to the treatment of HD (Fig. 3), which alone reduced HIV incidence by 26% (95% of the impact of treating all STDs). Treatment for NG contributed about 5% to the HIV incidence reduction, and that for chlamydia and syphilis, about 10% each. The absolute impact of treatment for NG in Mwanza was less than that in Rakai. In Rakai, HD and TV treatment contributed the most to the overall reduction in HIV incidence, even though the absolute reduction in incidence was extremely low (about 4%) for both STDs.

Fig. 3
Fig. 3:
Simulated contribution of treatment of each STD on overall HIV incidence impact for Mwanza and Rakai intervention scenarios. Treatment for each STD, at the coverage achieved in the actual trial,17 was added one at a time to the default scenario of no intervention and HIV incidence compared to intervention scenario including treatment for all STD. NG = gonorrhea; CT = chlamydia infection; TV = trichomoniasis; TP = syphilis; HD = chancroid.


This analysis showed that the PAFM for the effect of curable STD on HIV incidence was much higher in the Mwanza scenario (65%) than the Rakai scenario (20%) (Fig. 1). These results were broadly consistent with earlier PAF estimates based on observed associations between STD and HIV incidence, which estimated the PAF for the effect of symptomatic STD on acquiring incident HIV as 22% in Mwanza and 12% in Rakai.7,8 These empirical PAFs are much lower than our simulated estimates because they captured only the effects of STDs on increasing HIV acquisition and not on HIV infectivity and the onward dynamic effect in the STD-unexposed population. The PAFMs for the effect of any STD (including HSV-2) on HIV acquisition were 29% in Mwanza and 17% in Rakai, closer to the empirical PAF estimates.

The higher PAFM for curable STDs in Mwanza than in Rakai is consistent with a higher observed4,29 and simulated17,18 impact of STD treatment on HIV incidence in Mwanza compared to Rakai. This comparison shows clearly that the simulated impact of STD treatment in these trials was much less than the total simulated PAF (28% compared to 65% in Mwanza and 10% compared to 20% in Rakai). Neither periodic MT of all STDs in Rakai nor ST of symptomatic STD in Mwanza eliminated the effects of cofactor STDs on HIV spread. For Mwanza, the discrepancy relates to the large proportion of STDs that were asymptomatic (but were included in the PAFM estimations). For Rakai, the discrepancy relates to the proportion of STDs that periodic MT at 10 month intervals would not capture.

Another major difference between the Mwanza and Rakai simulations was the contribution of the different STD to the overall HIV impact achieved in each site. At the time of the trials, HD was the most important STD in Mwanza, but in Rakai HSV-2 was most important. This is because HD and HIV were clustered in high-risk groups in the Mwanza epidemic when the trial took place 9 years into the epidemic. On the other hand, in Rakai, due to reductions in risky behavior and the later stage of the epidemic, treatable STDs were less prevalent, and HSV-2 was the most important STD at the time of the trial 16 years into the epidemic.

This finding of a high proportion of HIV infections attributable to HD in Mwanza should be interpreted with caution because empirical data on the prevalence or treatment of HD were not available. HD was shown to be a significant cause of genital ulcer disease (GUD) in STD clinic attendees in Mwanza Town at the time of the trial30 and in the empirical PAF estimates from the trial were highest for genital ulcer disease (among men).8 The simulated prevalence of HD in Mwanza (1%) was consistent with data on the incidence of ulcers and was projected by fitting sexual risk behavior to the observed prevalence of other STDs.17 This value remains uncertain, and yet it is the single most critical contributor to the overall PAF of curable STDs in Mwanza. Given this uncertainty, our qualitative conclusions were relatively robust to this unknown. If we reduced HD prevalence to an extremely low level (0.2%) in the Mwanza simulation, the PAFM of HD was reduced considerably from 39% to 14% and that for all curable STD combined from 65% to 52% (Table 3), so the PAFM for Mwanza would still be considerably higher than that for Rakai, where HIV was less clustered among high-risk individuals and HD prevalence was assumed to be 0.3% at the time of the trial.17

In Mwanza, the estimated empirical PAF for HSV-2 infection (74% in males and 22% in females)31 was much higher than that simulated (12% in both males and females). There are two possible explanations for this large discrepancy between the empirical and simulated estimates. One is that the empirical estimates overestimate the effect of HSV-2 on HIV spread. This could be due to confounding by sexual behavior. It is plausible that this confounding in the observational estimates for Mwanza was more important for HSV-2 than for curable STD because HSV-2 is of long duration similar to HIV so that their risk factors may more strongly overlap. However, the 2 infections have very different transmission dynamics because HSV-2 is so widespread in the population and is much less likely to be associated with high-risk sexual behavior compared to HIV. Data from the study indicated very little confounding between HSV-2 and sexual behavior.31

A second explanation is that the model may have underestimated the cofactor effect of HSV-2. We investigated an underestimation of the cofactor during the asymptomatic stages of HSV-2 infection, for which there is some, albeit equivocal, evidence.28,32,33 Assuming an additional cofactor effect in between ulcers increased the PAFM of HSV-2% to 32% (Table 3). This latter scenario may overestimate the “asymptomatic” HSV-2 cofactor effect because ulcerative episodes that may be unrecognized by patients were simulated in the default scenario, and these would contribute to the “asymptomatic” HSV-2 cofactor effect.34 We did not investigate the impact of increasing HSV-2 cofactors during ulcers on the spread of HIV, which may also result in an increased PAFM due to HSV-2. Considering the empirical estimate only accounts for increased susceptibility due to HSV-2, the large discrepancy between the model estimates and the empirical estimate requires further research.

In both the Mwanza and Rakai scenarios, the proportion of HIV infections due to treatable STDs decreased over the time course of the epidemic. This finding is consistent with previous investigations of the role of risk factors for HIV transmission over time.9,12,26 However, it should be noted that as more interventions become available in these settings and incidence declines, the proportion of new HIV infections attributable to the core group may increase again, similar to the dynamic seen among men who have sex with men in industrialized countries. Our findings are also consistent with the recent STD treatment trial conducted among sex workers in Nairobi.35 In that study, before treatment, no ulcers cultured positive for H ducreyi (even among this high-risk group) and overall incidence of ulcers was extremely low, characteristics of a late-stage epidemic and/or a setting where years of interventions may have reduced the prevalence of HD. The authors suggested the lack of impact of STD treatment on HIV incidence was potentially due to STDs playing a more important role in HIV transmission compared to acquisition, as was shown to be the case in our study.

Several limitations to the current analysis should be kept in mind. Most important, not only for HD but for any STD, the magnitudes of cofactor effects assumed in the model are uncertain.10 However, our qualitative findings were robust to halving or doubling the magnitude of all cofactors. In these scenarios, at the time of the trials the PAFM for curable STD was in the range of 52%–72% in Mwanza, much higher than that in Rakai, 9%–30% (Table 3).

Second, it was assumed that infectivity and susceptibility cofactors were equal. This may not be the case,11,36 so results regarding the relative PAFM for the effect of STD on HIV transmission and acquisition should be interpreted with caution. Results showed STD had a larger impact on HIV transmission than acquisition in both Mwanza and Rakai. Results from the sensitivity analysis indicated lowering infectivity cofactors would reduce the PAFM for curable STD by about 7% (Table 3), but lowering susceptibility cofactors reduced it by 2%. Therefore, if infectivity cofactors are truly less than those for susceptibility, the PAFMs for the effect of STD on HIV spread would be lower than estimated here.

Finally, if a simulated individual had multiple STDs at the time of HIV transmission, the single largest cofactor was applied. It is likely that the presence of multiple STDs would result in increased HIV transmission compared to that for a single STD.11 This would result in an underestimation of the PAFM for Mwanza, where STDs were generally clustered in the same individuals.

In conclusion, observational PAF estimates from the trials underestimated the role of curable STDs in both trials but especially in Mwanza. This principally occurred because the effects of STDs on HIV transmission, which according to simulations accounted for over half of the total PAFM, were not fully captured in the empirical analyses. The simulated proportion of HIV infections attributable to curable STD was much larger in Mwanza than in Rakai, in line with the larger trial impact in Mwanza, and this related mostly to the behavioral risk reduction in Rakai, which reduced the prevalences of curable STDs before the onset of the trial.

Importantly, the model projections imply that HD (eminently controllable with simple, low-cost interventions)37 may play a very important role in some HIV epidemics, emphasizing the need for more population-based data on the incidence and prevalence of this pathogen and trends over time. This may be especially important in early epidemics where HIV remains concentrated among high-risk groups. On the other hand, in later-stage epidemics, where HSV-2 may have a larger impact on HIV spread than curable STD, HSV-2 interventions may help reduce HIV incidence. Several studies are currently under way to evaluate this hypothesis.

The generally larger role of cofactor STDs in HIV transmission than in HIV acquisition underscores the notions that STD treatment among core groups in early-stage epidemics38 and among HIV-infected individuals39 may be an important HIV prevention strategy. The impact of the introduction of antiretroviral therapy in resource-poor settings on the interactions between HIV and cofactor STDs should be monitored to inform HIV prevention. As voluntary counseling and testing becomes more widespread in southern Africa, education regarding STD symptom recognition and the importance of prompt treatment could be an important aspect of posttest counseling.


1. Wasserheit JN. Epidemiological synergy: interrelationships between human immunodeficiency virus infection and other sexually transmitted diseases. Sex Transm Dis 1992; 19:61–77.
2. Cohen MS. Sexually transmitted diseases enhance HIV transmission: no longer a hypothesis. Lancet 1998; 351(suppl III):5–7.
3. Grosskurth H, Mosha F, Todd J, et al. Impact of improved treatment of sexually transmitted diseases on HIV infection in rural Tanzania: randomized controlled trial. Lancet 1995; 346:530–536.
4. Wawer MJ, Sewankambo NK, Serwadda D, et al. Control of sexually transmitted diseases for AIDS prevention in Uganda: a randomised community trial. Lancet 1999; 353:525–535.
5. Kamali A, Quigley M, Nakiyingi J, et al. Syndromic management of sexually transmitted infections and behaviour change interventions on transmission of HIV-1 in rural Uganda: a community randomised trial. Lancet 2003; 361:645–652.
6. Grosskurth H, Gray RH, Hayes R, et al. Control of sexually transmitted diseases for HIV-1 prevention: understanding the implications of the Mwanza and Rakai trials. Lancet 2000; 355:1981–1987.
7. Gray RH, Wawer MJ, Sewankambo NK, et al. Relative risks and population attributable fraction of incident HIV associated with STD symptoms and treatable STDs in Rakai District, Uganda. AIDS 1999; 13:2113–2123.
8. Orroth KK, Gavyole A, Todd J, et al. Syndromic treatment of sexually transmitted diseases reduces the proportion of incident HIV infections attributable to these diseases in rural Tanzania. AIDS 2000; 14:1429–1437.
9. Boily M-C, Anderson RM. Human immunodeficiency virus transmission and the role of other sexually transmitted diseases: measures of association and study design. Sex Transm Dis 1996; 23:312–332.
10. Korenromp EL, de Vlas SJ, Nagelkerke NJD, et al. Estimating the magnitude of STD cofactor effects on HIV transmission: how well can it be done? Sex Transm Dis 2001; 28:613–621.
11. Rottingen JA, Cameron DW, Garnett GP. A systematic review of the epidemiologic interactions between classic sexually transmitted diseases and HIV: how much really is known? Sex Transm Dis 2001; 28:579–597.
12. Koopman JS, Longini IM, Jacquez JA, et al. Assessing risk factors for transmission of infection. Am J Epidemiol 1991; 133:1199–1209.
13. Van der Ploeg CBP, Van Vliet C, De Vlas SJ, et al. STDSIM: a microsimulation model for decision support in STD control. Interfaces 1998; 28:84–100.
14. Korenromp E, Vliet CV, Bakker R, et al. HIV spread and partnership reduction for different patterns of sexual behaviour: a study with the microsimulation model STDSIM. Math Popul Studies 2000; 8:135–173.
15. Orroth KK, Korenromp EL, White RG, et al. Comparison of STD prevalences in the Mwanza, Rakai and Masaka trial populations: the role of selection bias and diagnostic errors. Sex Transm Inf 2003; 79:98–105.
16. Orroth KK, Korenromp EL, White RG, et al. Higher risk behaviour and STD rates in Mwanza compared to Rakai and Masaka: a possible explanation for outcomes of HIV prevention trials? AIDS 2003; 17:2653–2660.
17. White RG, Orroth KK, Korenromp EL, et al. Can population differences explain the contrasting results of the Mwanza, Rakai and Masaka HIV/STD intervention trials? a modelling study. J Acquir Immun Defic Syndr 2004; 37:1500–1513.
18. Korenromp EL, White RG, Orroth KK, et al. Determinants of the impact of STD treatment on HIV prevention: a synthesis of evidence from the Mwanza, Rakai and Masaka intervention trials. J Infect Dis 2005; 191(suppl 1):S168–S175.
19. Korenromp EL, Vliet CV, Grosskurth H, et al. Model-based evaluation of single round mass STD treatment for HIV control in a rural African population. AIDS 2000; 14:573–593.
20. Orroth AK. Investigations of the Proportion of HIV Infections Attributable to Sexually Transmitted Diseases in Sub-Saharan Africa Based on Data from the Mwanza and Rakai Trials. London: University of London, 2003.
21. Hayes RJ, Plummer FA. The cofactor effect of genital ulcers on the per-exposure risk of HIV transmission in sub-Saharan Africa. J Trop Med Hyg 1995; 98:1–8.
22. Grosskurth H, Mwijarubi E, Todd J, et al. Operational performance of an STD control programme in Mwanza Region, Tanzania. Sex Transm Inf 2000; 76:426–436.
23. Buve A, Mayaud P, Gavyole A, et al. How many patients with a sexually transmitted infection are cured by health services? a study from Mwanza Region, Tanzania. Trop Med Int Health 2001; 6:971–979.
24. Wawer MJ, Gray RH, Sewankambo NK, et al. A randomized, community trial of intensive sexually transmitted disease control for AIDS prevention, Rakai, Uganda. AIDS 1998; 12:1211–1225.
25. Robinson NJ, Mulder DW, Auvert B, et al. Proportion of HIV infections attributable to other sexually transmitted diseases in a rural Ugandan population: simulation model estimates. Int J Epidemiol 1997; 26:180–189.
26. Korenromp EL, Bakker R, De Vlas SJ, et al. HIV dynamics and behaviour change as determinants of the impact of STD treatment on HIV transmission in the context of the Rakai trial. AIDS 2002; 16:2209–2218.
27. Brunham RC, Plummer FA. A general model of sexually transmitted disease epidemiology and its implications for control. Med Clin North Am 1990; 74:1339–1352.
28. McClelland RS, Wang CS, Overbaugh J, et al. Association between cervical shedding of herpes simplex virus and HIV-1. AIDS 2002; 16:2425–2430.
29. Hayes R, Grosskurth H, Ka-Gina G. Impact of improved treatment of sexually transmitted disease on HIV infection: authors' reply. Lancet 1995; 346:1159–1160.
30. Grosskurth H. The Impact of Improved STD Case Management on HIV Infection and STD in Mwanza Region, Tanzania. London: London School of Hygiene and Tropical Medicine, 2000.
31. Rodriguez MMP, Obasi A, Mosha F, et al. Herpes simplex virus type 2 infection increases HIV incidence: a prospective study in rural Tanzania. AIDS 2002; 16:451–462.
32. Mbopi-Keou F, Grensenguet G, Mayaud P, et al. Interactions between herpes simplex virus type 2 and human immunodeficiency virus type 1 infection in African women: opportunities for intervention. J Infect Dis 2000; 182:1090–1096.
33. Corey L, Wald A, Celum CL, et al. The effects of herpes simplex virus-2 on HIV-1 acquisition and transmission: a review of two overlapping epidemics. J AIDS 2004; 35:435–445.
34. Korenromp EL, Bakker R, De Vlas SJ, et al. Can behaviour change explain increases in the proportion of genital ulcers attributable to herpes in sub-Saharan Africa? a simulation modelling study. Sex Transm Dis 2002; 29:228–238.
35. Kaul R, Kimani J, Nagelkerke NJ, et al. Monthly antibiotic chemoprophylaxis and incidence of sexually transmitted infections and HIV-1 infection in Kenyan sex workers: a randomized controlled trial. JAMA 2004; 291:2555–2562.
36. Fleming DT, Wasserheit JN. From epidemiological synergy to public health policy and practice: the contribution of other sexually transmitted diseases to sexual transmission of HIV infection. Sex Transm Infect 1999; 75:3–17.
37. Steen R. Eradicating chancroid. Bull WHO 2001; 79:818–826.
38. Alary M, Mukenge-Tshibaka L, Bernier F, et al. Decline in the prevalence of HIV and sexually transmitted diseases among female sex workers in Contonou, Benin, 1993–1999. AIDS 2002; 16:463–470.
39. Serwadda D, Gray RH, Sewankambo NK, et al. Human immunodeficiency virus acquisition associated with genital ulcer disease and herpes simplex virus type 2 infection: a nested case-control study in Rakai, Uganda. J Infect Dis 2003; 188:1492–1497.
© Copyright 2006 American Sexually Transmitted Diseases Association