Article Text

The feasibility of integrated STI prevalence and behaviour surveys in developing countries
1. E W MacLachlan1,
2. E Baganizi1,
3. F Bougoudogo2,
4. S Castle3,
5. Z Mint-Youbba4,
6. P Gorbach5,
7. K Parker1,
8. C A Ryan1
1. 1Centers for Disease Control and Prevention, Atlanta, Georgia, USA
2. 2Mali National Institute for Public Health Research, Bamako, Mali
3. 3London School of Tropical Medicine and Hygiene, London, UK
4. 4Mali National AIDS Control Program, Bamako, Mali
5. 5University of California at Los Angeles, Los Angeles, California, USA
1. Correspondence to:  Ellen W MacLachlan, International Activities Unit, Division of STD Prevention, Centers for Disease Control and Prevention, 1600 Clifton Road, NE, MS E-04, Atlanta, GA 30333, USA;  eew7{at}cdc.gov

## Abstract

Background: In countries where STI/HIV prevalence data and behavioural data are scarce UNAIDS second generation HIV surveillance guidelines recommend measuring STI/HIV prevalence and risk behaviours in vulnerable populations but do not recommend conducting these surveys concurrently because of concerns about participation rates, cost, and provision of services.

Objectives: To assess the feasibility of conducting a national combined STD prevalence and behaviour survey in Mali among vulnerable populations with the intention of institutionalisation.

Methods: From March to June 2000 an integrated STI prevalence and behaviour survey was conducted using cluster sampling among five risk groups in four sites in Mali, west Africa. 2229 individuals in non-traditional settings such as taxi/bus stations, market areas, households, and brothels participated in any one or all components of the study: (1) behavioural questionnaire, (2) urine sample for Neisseria gonorrhoeae (GC)/Chlamydia trachomatis (CT) testing, (3) a fingerstick drop of blood for syphilis, and/or (4) HIV testing.

Results: High participation rates of 84%–100% were achieved despite specimen collection and HIV testing. Rates fell only slightly when participants were asked to provide biological samples and participants were more likely to provide urine than blood. Rates among the different groups for HIV and syphilis testing are similar and suggest that refusal was most probably because of a reluctance to give blood rather than because of HIV testing. The cost of the biological component added approximately $30 per participant. Included in the$30 are the costs of training, participant services, laboratory personnel and supplies, STI drugs, and STI testing costs. The total cost of the survey was $154 905. Biomarkers aided in validation of answers to behavioural questions. Consenting individuals received HIV pretest and post-test counselling and referral to a trained health provider for treatment of STI and the provision of services provided the framework for interventions in the groups following the survey. Conclusion: This represents an effective methodology for collecting risk behaviour and STI/HIV prevalence information concurrently and should be considered by countries expanding STI/HIV surveillance as part of UNAIDS second generation HIV surveillance. • core groups • HIV surveillance • linked surveys ## Statistics from Altmetric.com Lessons learned from the epidemiology of HIV thus far indicate that governments need to act quickly once HIV has entered a population to ensure prevention of infection among those populations most likely to contract and spread HIV.1 To accomplish this requires reliable information about the risk behaviours and the level of infection with HIV and other sexually transmitted infections (STI) in the general population and in these high risk “core” groups.2 Behavioural and biological surveys in these groups provide this information and can be repeated over time in order to follow trends in the evolution of an HIV epidemic and assess where intervention is most likely to have an impact. Surveys in core groups are an important part of a population based perspective in the control of STIs and HIV.3,4 In countries where such data are scarce it is imperative to identify practical methods for measuring STI/HIV prevalence and risk behaviours in high risk groups and to institutionalise those methods within the public sector so they can be repeated. The combined importance of behavioural surveys, biological surveys, and HIV sentinel surveillance has been highlighted in UNAIDS recommendations for second generation surveillance for HIV.5 Although UNAIDS recommends conducting both behavioural and biological surveys it does not recommend conducting them concurrently. We believe this is because of concerns that collecting biological specimens will negatively impact participation rates for behavioural surveys, creating bias, increasing the survey cost, and complicating them with the difficulty of providing results and services to participants; in spite of evidence that integrated surveys provide much rich information.6 Several innovative surveys that link behavioural information with STI or HIV biomarkers have been successfully implemented, though mostly on a small scale or in a research setting with no intention of institutionalisation.7–9 In effect, the full integration of such surveys into public sector planning has not yet been accomplished. In this case study in Mali a linked survey was implemented at the national level and we show how it has provided the government with a feasible and practical tool for following trends in behaviours and STI/HIV rates in at-risk populations. A consideration for using this tool in other settings will be the stage of the HIV epidemic. In low HIV prevalence settings subgroup or site specific analysis can be a problem and STI/HIV testing is more prone to misclassification from false positive results. In high prevalence settings youth or general population samples should be considered. In Mali there has been no HIV sentinel surveillance since 1995, the same year that the last cross sectional study of STI/HIV prevalence that included core groups was conducted. Few recent studies have investigated specific STI/HIV risk behaviours in vulnerable populations. To better understand the factors involved in the HIV epidemic and to test the feasibility of a combined survey we conducted the ISBS (Integrated STI Prevalence and Behavior Survey) in Mali from March to June of 2000 among a sample size of 2229 people in four urban sites. The survey was meant to complement reinvigoration of HIV sentinel surveillance in antenatal settings and thereby assist the national AIDS control programme to not only follow HIV trends in the general population, with pregnant women as a proxy, but to also collect STI/HIV and behavioural data in vulnerable populations as is recommended by second generation HIV surveillance. The survey collected biological specimens and behavioural information concurrently in non-traditional settings such as markets and taxi/bus stations. Sentinel groups were chosen following formative research that identified two high risk “core” groups (prostitutes and truck drivers) and three potential “bridging groups.”10–12 The bridging groups were (1) female ambulatory vendors working in taxi/bus stations, (2) “ticket touts” or young men who work in taxi/bus stations finding clients, and (3) young women who leave rural Mali to work in urban areas as maids. Random one stage cluster sampling was used. Participants could take part in any one or all four components of the study: (1) behaviour questionnaire, (2) a urine sample for GC and CT testing, (3) a fingerstick drop of blood for syphilis testing, and/or (4) HIV testing. For GC and CT testing PCR was used and for syphilis and HIV testing RPR/TPHA and Immunocomb/Genie II were used. Consenting individuals received HIV pretest and post-test counselling and referral to a trained health provider for treatment of STI. High participation rates were achieved for all components of the study, as shown in table 1. Rates fell only slightly when participants were asked to provide biological samples and participants were more likely to provide urine than blood. Rates among the different groups for HIV and syphilis testing are similar and suggest that refusal was most probably because of a reluctance to give blood rather than because of HIV testing. Those who refused biological testing were similar to consenting individuals except in the following: maids who refused blood testing were more likely to have said that they had an STI in past 6 months (43.8% v 17.8%, p=0.02); ticket touts who refused blood testing were more likely to have said that they had been tested for HIV (16.1% v 7.6%, p=0.05); and truckers who refused urine testing were less likely to have had sex with a sex worker in the past 6 months (17.7% v 5.3%, p=0.06). With regard to cost, the total cost of the behavioural component of the survey was approximately$38 per participant and the cost of the biological component added approximately $30 per participant. Included in the$30 are the costs of training, participant services, laboratory personnel and supplies, STI drugs, and STI testing costs. The total cost of the survey was \$154 905.

This study showed that the ISBS design is feasible and can be a powerful tool for monitoring an HIV epidemic. We found that high participation rates can be obtained even when biomarkers are added in non-traditional settings and groups. Participation in ISBS was most probably high because of the services provided or could have been because of the relatively low prevalence of HIV—lower societal awareness and experience with HIV/AIDS may have resulted in fewer participants feeling fear and trepidation about their HIV test results. We believe, however, that comprehensive preparation in the field and the provision of high quality services will most likely result in high participation rates in other settings. In addition, linked surveys such as ISBS may be more ethical in an era when more and more HIV related services are becoming available and blinded serosurveys are becoming controversial. Behavioural surveys with biomarkers also offer the important opportunity to validate individual answers to behavioural questions. For example, several maids who reported to be virgins tested positive for an STI. Finally, though we do not present data comparing the cost of a combined survey to separate behaviour and biological surveys in the same population, it seems clear that combining surveys can save on the costs incurred when the two surveys are conducted separately. The added cost of biomarkers can, in most cases, be justified in terms of the information provided—though rapid tests could play an important part in decreasing the cost.

Table 1

Participation rates for 2000 Mali integrated STI prevalence and behaviour survey

## Acknowledgments

The authors thank Drs Sevgi Aral, Peter Ghys, Rebecca Martin, and Chris Murrill for their helpful comments. We are also indebted to Shirley Davis, Sharon Cassell, Mahamet Abdouramen, Alhousseni Sangare, Nene Aicha Traore, Saran Sidibe, Adema Sangare, and Paul Sangala for their assistance with this study. We also thank USAID-Mali, especially Ursula Nadolny and Aida Lo, for their support of this study.

Funding source: United States Agency for International Development.

CONTRIBUTORS  EM, protocol development, ethical review, field implementation, analysis, writing; EB, protocol development, ethical review, field implementation, analysis, review; FB, protocol development, ethical review, laboratory support, field implementation, review; ZM-Y, protocol development, field implementation, review; PG, protocol development, review; SC, protocol development, formative research, review; KP, protocol development, ethical review, analysis, writing; CR, protocol development, ethical review, field implementation, analysis, writing.

View Abstract

## Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.