This study suggests that the syndromic STI treatment services in rural KwaZulu-Natal cure around 13% of symptomatic curable STI episodes. This low effectiveness is consistent with the continuing high prevalence of curable STIs in this population. Our results suggest that the effectiveness of curable syndromic STI treatment services could be improved most by increasing rates of treatment seeking and provision of the correct treatment.
Although our estimates were based on the best available data and robustly accounted for statistical uncertainty, some caution is required in the interpretation. Better data on some of the required parameter values would improve our estimates. For example, data on the quality of treatment provision were collected in the late 1990s and provision could have improved or deteriorated since then, which may lead to an underestimate or overestimate of effectiveness, respectively. The available data on STI trends suggest, however, that whatever the trend in overall effectiveness since the late 1990s, there is no evidence this has been accompanied by a marked change in STI rates. We had no information on the total number of syndrome episodes experienced by residents each year, and therefore our estimate of the annual number of symptomatic episodes in residents is likely to be a lower bound. We were also required to use cure rate data on ulcers and urethritis from an urban KwaZulu-Natal population, which may not a bovne be representative of the study area. However, the estimate of overall effectiveness was shown to be robust to increasing the cure rate to 100% and relatively insensitive to a reduction of 25% in this parameter value—a range likely to include the true value. Studies to collect more data on these parameters may be justified and could be used to update these estimates using the supplied WinBUGS code.
We may have overestimated overall effectiveness for at least two other reasons. If we overestimated the proportion of individuals seeking treatment at clinics or their GPs because, for example, we based our estimate on data from survey participants rather than all residents, or because patients actually went to traditional healers, then the overall effectiveness would be lower than estimated. Similarly, if the assumed cure rates were too high because, for example, they were based on data on patients followed-up rather than those lost to follow-up, then overall effectiveness would be lower than estimated. However, this effect was shown to be small and data collected in Tanzania in the mid-1990s showed little difference between cure rates among returners and non-returners.26
Conversely, we may have underestimated the overall effectiveness for at least two reasons. We assumed that the cure rate for incorrect treatment was 0%, but some treatments not in the national guidelines may be partially effective, increasing the effectiveness. Also, with the exception of the treatment for gonorrhoea, only the first-line treatment regimens were considered. Including the second-line treatment regimens would have increased effectiveness. However, as the assumed cure rates were already high, any underestimate will be small.
Our definition of overall effectiveness excluded asymptomatic STI episodes. This definition was chosen because it can be considered to be the upper limit for syndromic STI treatment (barring any indirect treatment via partner notification or cure of asymptomatic co-infections). However, as a large proportion of STI episodes are asymptomatic, particularly in women,27
even if all symptomatic episodes are cured a large proportion of STIs will remain untreated. This study could be extended to include asymptomatic episodes, but this would require currently unavailable aetiology-specific general population data on the frequency of STI episodes and the proportion of episodes that became symptomatic.
Our primary aim was to estimate the effectiveness of syndromic STI treatment services in treating curable STIs. However, the primary cause of genital ulcers in this population is HSV-2.14
HSV-2 infection is incurable and antiviral HSV-2 treatment is not recommended in the national guidelines.6
Our results suggest that if HSV-2 ulcers were included in the effectiveness calculation the proportion of ulcers cured or correctly treated was halved, but this reduction could be entirely countered by including episodic antiviral treatment in the national guidelines. The lack of treatment for HSV-2 ulcers may be an important reason for the low treatment seeking rates in this population, if individuals seeking treatment learn that treatment is ineffective. If true, then adding episodic HSV-2 antiviral treatment to the syndromic STI treatment guidelines may have the additional benefit of increasing the treatment seeking rates for other STIs.
Our STI prevalence trend analysis was limited by changes in diagnostic methods and sample populations over time and, therefore, statistical trend-tests were inappropriate. Molecular diagnostic methods, which have been replacing classical methods over the period under study, tend to have a higher sensitivity than classical methods and therefore it is possible that a decreasing trend in gonorrhoea and chlamydia may have been obscured. The contrasting fall in syphilis prevalence may be due, in part, to syphilis treatment of antenatal clinic attendees.8
Changes in sample populations over time and high rates of migration and urbanisation in the study area may also have led to under or overestimation of effectiveness.
A previous study estimated the proportion of women who were asymptomatic, symptomatic but not seeking care, and symptomatic and seeking care and infected with T vaginalis, N gonorrhoeae, C trachomatis
or T pallidum
on any given day in Hlabisa health district.28
In line with our findings, the authors concluded that STIs remained untreated because of low symptom recognition and treatment seeking rates. Our study updated this analysis, extended the work to include males, incorporated other ulcerative STIs, including HSV-2, and estimated the overall effectiveness of the syndromic STI treatment algorithm.
Another earlier study estimated the proportion of bacterial STI cured by primary healthcare services in Mwanza Region, Tanzania, before and after the introduction of improved STI treatment services, to be <10% and 23–41%, respectively.29
The higher effectiveness estimated for Mwanza than KwaZulu-Natal was primarily due to higher treatment seeking rates and higher rates of correct treatment. The higher rates in Tanzania may have been because community activities were more intensive as part of an intervention trial, or the division of STI treatment services between the private and public sectors in South Africa may make improving services more difficult than in Tanzania.
Only about 13% of symptomatic STI episodes may be cured by syndromic STI treatment services in rural KwaZulu-Natal and there is little evidence to show an improvement in the prevalence of curable STIs. As long as there is no acceptable alternative, syndromic STI treatment remains the only option for the management of STIs in the individual patient. As syndromic STI treatment may be a cost-saving HIV prevention strategy in this high HIV-incidence setting,3
innovative strategies are urgently needed to increase rates of treatment seeking and the provision of the correct treatment.
- Overall effectiveness of syndromic treatment for curable sexually transmitted infections (STIs) in rural KwaZulu-Natal remains low (13.1%, 95% CI 8.9 to 17.8% of symptomatic curable STI episodes cured) and there is little evidence of reduced curable STI prevalences.
- As syndromic treatment is likely to be a cost-saving HIV prevention intervention in South Africa, innovative strategies are urgently needed to increase rates of treatment seeking and correct treatment provision.
- The model used to make this estimate is provided in full in the online material and could be used to (re)estimate effectiveness in this and other settings.