Design of a Community-Based Study of Sexually Transmitted Infections/HIV and Infertility in an Urban Area of Northern Tanzania

Abstract

Objective: 

The objective of this study was to describe the design of a community-based study of sexually transmitted infections (STIs)/HIV and infertility in northern Tanzania.

Study Design: 

Households were selected using a 2-stage sampling design. Eligible women and their partners were interviewed before samples were collected for STIs/HIV detection. Posttest counseling and treatment for STIs and infertility were provided.

Results: 

A total of 2019 women and 794 male partners were interviewed. Over 70% of interviewed women and men provided blood and urine samples. Individuals providing blood and urine samples had high-risk profiles for STIs/HIV when compared with others who did not provide these samples. Although the study results may be affected by selection bias, risk factors for STIs/HIV were similar to those in other studies supporting the generalizability of the findings.

Conclusions: 

It is feasible to conduct a community-based survey, including collection of biomarkers and measurement of infertility, in this urban setting.

In Brief

This is a description of a community-based survey, including collection of blood and urine samples for sexually transmitted infections/HIV testing and measurement of infertility in Moshi Urban District in northern Tanzania.

BY THE END OF 2005, MORE than 25 million people were estimated to be living with HIV in sub-Saharan Africa.¹ Until recently, there were little representative HIV data from this region and HIV analyses were typically based on data from blood donors or antenatal clinic attendees. Since 2001, Demographic and Health Surveys (DHS) collected nationally representative HIV data in Mali, Zambia, Ghana, Kenya, Burkina Faso, Tanzania, and Cameroon.² However, there is evidence suggesting that the prevalence and distribution of HIV infection vary significantly from one location to another, making it necessary to conduct local-level studies. For instance, in Tanzania, the prevalence of HIV ranged from 15.2% in the Mbeya region to 2.0% in the Manyara region.³ The DHS include too few cases for HIV risk factor analysis at the regional or district level and they have not yet included information about other sexually transmitted infections (STIs) that act as risk factors for HIV such as herpes simplex virus type 2 (HSV-2). STIs and other genital infections can result in infertility, and determining prevalence and risk factors is important for public health programs. In Tanzania, prevalence of infertility is approximately 20%.⁴ This article describes the design of a study of STIs/HIV and infertility in one local African community, Moshi Urban District (Moshi) in northern Tanzania.

Materials and Methods

The study was carried out by a multidisciplinary team from the Kilimanjaro Christian Medical Centre (KCMC) in Moshi and the National Bureau of Statistics (NBS) in Dar es Salaam, Tanzania, as well as the Centers for Disease Control and Prevention (CDC) in Atlanta, Harvard School of Public Health in Boston, and the University of Maryland in College Park.

Study Site

Moshi had 144,336 inhabitants according to the 2002 census.⁵ Several reasons led us to select this site in the Kilimanjaro region of Tanzania. First, the research team has a long history of collaborating with institutions in this part of Tanzania.^6,7 Second, the project builds on our previous research on STIs/HIV in Moshi.^7,8 Third, it was crucial for assuring high-quality laboratory work that the project was located in an area with available laboratory facilities, and well-trained and experienced personnel. KCMC is one of the 4 referral hospitals in Tanzania with a well-developed laboratory infrastructure to conduct the required tests and store samples before being shipped abroad. The clinical staff at KCMC is able to do the infertility investigations and treatment following the highest standard of care in Tanzania. Finally, medical records suggested that infertility was the most common reason for outpatient care in the Department of Gynecology and Obstetrics at KCMC for the period 1996 to 1999. Thus, the study was expected to be beneficial to the community because community-based analyses of STIs/HIV and infertility had never been done.

Institutional Review Board Approval, and District and Ward Research Permissions

The project was approved by the Harvard School of Public Health IRB (HSC Protocol # 0108ACOM), University of Maryland Institutional Review Board (Protocol #05-0031), Kilimanjaro Christian Medical Centre Ethics Committee, Research & Ethical Clearance of the Tanzanian National Institute for Medical Research, and the Centers for Disease Control and Prevention Institutional Review Board. The municipal office in Moshi granted general research permission, and ward leaders gave ward-specific permissions. Similar permissions were obtained for the pilot study.

Survey Fieldwork

First, 3-week training was conducted by the NBS statisticians for the enumerators. Each enumerator was provided an interviewer manual explaining each question and the entire survey interview process. The enumerators were experienced nurses who knew how to draw blood and interact with patients, although some of the enumerators had no experience dealing with HIV/AIDS issues. Therefore, a senior nurse and STIs/HIV counselor from KCMC joined the training sessions and was responsible for the training in STIs/HIV counseling.

In April 2002, a 1-week pilot study was conducted in Uchira, a village in Moshi Rural District approximately 20 km away from the border of Moshi, so there was little chance that respondents in the pilot would be sampled for the main survey. The pilot study enabled us to test and revise the protocol. Survey fieldwork was done in the period November 18, 2002, to March 31, 2003.

Sample Size and Power Calculations

The number of women with HIV-1 or infertility needed to detect the minimum specified odds ratios (ORs) with the power of 80% and a 2-sided type I error rate of 5% were estimated using Epi Info Version 6.⁹ Calculations were done for ORs ranging from 1.50 to 2.50 and the prevalence of exposures ranging from 5% to 30%. On this basis, the number of women with HIV-1 or infertility required to detect an OR of 2.0 ranges from 276 for exposure of 5% to 81 for exposure of 30%.¹⁰ We assumed the prevalence of HIV-1 or infertility was approximately 10% and planned to enroll 2000 women and their partners in the survey.^4,8

Sampling

Moshi contains 15 wards and each ward was divided into a number of clusters using 2002 census maps. A total of 150 clusters were selected for interview sampling proportionate to the number of women aged 20 to 44 living in the ward. Approximately 14 households needed to be interviewed in each cluster to obtain a sample of at least 2000 women. The 1988 census suggested there were approximately 0.8 eligible female respondents in each household and therefore 18 households (14/.8 = 17.5) were selected in each cluster. A household listing was developed in June 2002. Households were selected using systematic random sampling within each cluster. In selected households, all women aged 20 to 44 were invited to participate and requested to identify their husbands/partners for an interview. There was no replacement for nonresponse. The sample was not confined to married women because there is evidence suggesting that infertile women are more prone to be divorced or separated. Only women aged 20 to 44 were recruited because it is difficult to distinguish between subfecundity and infertility for teenaged women and infertility from senescence increases rapidly after age 44.

Survey Interviewer Teams

Three survey interviewer teams were formed, each including a supervisor, field editor, and 5 enumerators (4 female and one male). The supervisor was a statistician from the NBS with extensive survey experience. The supervisor was responsible for all aspects of the fieldwork. The field editor was a nurse with prior survey experience who monitored each of the enumerators and assured that all questionnaires were completed satisfactorily. The male enumerators interviewed only male respondents, whereas the female enumerators interviewed both female and male respondents. One local leader joined each interviewer team and assisted with locating selected households and making the first contact.

Informed Consent

Consent for the household interview was generally provided by the head of the household or his wife. If they were not at home, any capable adult household member could give consent and answer the household form. Although almost everybody in Moshi is literate, the enumerator read the consent form. Then, the respondent consented by signing the consent form and the household interview proceeded. If the respondent did not consent, then the enumerator ended the interview and nobody from the household was interviewed. On completion of the household questionnaire, the enumerator identified eligible women and men listed on the questionnaire and approached them to obtain consent for the individual interview. After completing the individual questionnaire, the enumerator offered the respondent STIs/HIV counseling, and a second consent was obtained for collecting blood and/or urine samples. It was the respondent’s choice to be tested and the respondent could choose to have some or all of the tests offered.

Survey Questionnaires

The sampled households were enumerated using a household questionnaire consisting of a cover sheet to identify the household and a form on which de facto and de jure household members were listed. Name, sex, age, name of partner, and relation to household head were recorded for each household member. To assure privacy, interviews were carried out in the presence of only the enumerator. Three attempts were made to obtain, respectively, a household, woman, and husband/partner interview. The woman’s questionnaire collected information on background characteristics, marriage and pregnancy histories, contraception, symptoms of STIs (see Appendix 1 for the questions), reproductive health issues, and healthcare-seeking and treatment for infertility. The husband’s/partner’s questionnaire covered the same topics but was less detailed. After the interview, if the respondent consented, then 5 to 10 mL of blood was collected, and the respondent was given an empty sterile container and requested to fill it with 10 to 15 mL of urine.

Laboratory Methods

Blood and urine samples were transported in cool boxes to the laboratories at KCMC within 3 to 4 hours where blood samples were separated and serum stored in 2 to 8°C, and urine samples were aliquoted and stored at −80°C.

Serology testing was done weekly. HIV-1 infection was determined by using 2 HIV enzyme-linked immunosorbent assay (ELISA) tests. Vironostika HIV Uni-Form II plus O (Organon, Boxtel, The Netherlands) was used for screening and reactive samples were confirmed by using the Wellcozyme HIV-1 ELISA test (Murex 1.2.0; Murex Biotech Ltd., U.K.). Respondents were considered HIV-1-seropositive if both ELISA tests were positive. Samples with discordant ELISA test results were confirmed by Western blot tests (Bio-Rad Laboratories Ltd., Dartford, U.K.). Active or recent syphilis was diagnosed if the serum was reacting on both the rapid plasma reagin (RPR) card test (Macro-Vue; Becton-Dickinson, Cockeysville, MD) and the Treponema pallidum hemagglutination assay (TPHA) (Wecosyph HA; Murex Biotech Ltd., U.K.). Past syphilis infection was detected if the serum was reacting only on the TPHA. A type-specific ELISA test was used to detect HSV-2 (Focus Technologies, Cypress, CA).

At the completion of the survey, urine samples were shipped to the CDC. Urine samples were tested for chlamydia, gonorrhea, trichomoniasis, and Mycoplasma genitalium by using a real-time multiplex polymerase chain reaction (M-PCR) assay. Initially, all urine specimens were pooled in groups of 3 and DNA was extracted using the Qiagen Viral RNA kit (Qiagen), which is recommended for extraction of DNA from urine specimens. DNA was then extracted from all specimens falling within positive pools and tested by the M-PCR assay. Sequence-specific detection of M-PCR amplification products was based on TaqMan technology and was performed using the Rotor-Gene 3,000 instrument (Corbett Research, Australia). Amplification targets were cryptic plasmid for chlamydia, cytosine-specific DNA methyltransferase for gonorrhea, MgPa for M. genitalium, and repeat DNA fragment for trichomoniasis. The real-time M-PCR assay allows detection over a broad range of target concentrations with analytical sensitivities of 1 to 10 genomic copies for M. genitalium and gonorrhea, 0.01 to 0.1 genomic copies for trichomoniasis, and the equivalent of 0.01 to 0.1 Inclusion Forming Unit for chlamydia.

Measures of Infertility

Six definitions of infertility were examined.¹¹ It is desirable to have similar infertility definitions for research and clinical practice and therefore we recommended that studies of infertility use the question “How long have you tried to get pregnant?” to estimate the rate of “tried to conceive for at least 2 years” following the World Health Organization recommendation.^11,12 In clinical work, it might be warranted to initiate treatment after only 12 months or an even shortened period.¹³ We referred couples for an infertility examination at KCMC after 12 months of trial. That is, based on survey questionnaire responses of the woman, a couple was considered to experience infertility if the woman had not conceived in a period of at least 12 months, she wanted to get pregnant, she did not use contraception, neither partner was sterilized, and she was regularly sexually active (engaged in sexual intercourse within the last month of the survey date).

Posttest Results and Sexually Transmitted Infection Treatment

Posttest counseling for STIs/HIV, and treatment for syphilis and self-reported STIs symptoms were provided weekly at a dispensary, health center, government school, or other local institution in the ward where the survey interviews had taken place. All study subjects were encouraged to collect their test results. To maintain confidentiality, posttest results and counseling were given individually to each respondent by the enumerator who gave pretest counseling and collected the specimens. We offered couple counseling when both the wife and husband/partner had been interviewed, but in some cases, only the woman came for posttest counseling and results. This was an issue when the couple was discordant for an STI or HIV-1. There were instances in which one spouse (usually the woman) returned for posttest results and medication and requested medication for the absent partner. However, medication was provided only to individuals who had given specimens for testing of STIs or answered the survey questions about STI symptoms.

Posttest results and medication for STIs were not given in the individual homes because many respondents chose not to receive the results, and it was perceived to be less invasive on people’s privacy if these meetings were held at public institutions. We attempted to register the date and identification number of people who participated in the posttest counseling sessions and collected test results. Unfortunately, these efforts were less successful because the enumerators were unable to keep complete records partly because some participants came for posttest results several weeks after they were interviewed. In general, it took substantial time and resources to return test results, to maintain confidentiality, and to keep complete and accurate records. The proportion receiving posttest results was low.

Respondents reporting STIs symptoms in the survey questionnaires and those with laboratory-confirmed STIs received medication following the guidelines of the Tanzania Ministry of Health. All medications prescribed were paid by the study and respondents were reimbursed for public transportation costs that they incurred during follow-up visits. At regular time intervals, all interviewed households were visited by one of the enumerators to thank people who participated in the survey and encourage individuals to collect STIs/HIV test results and possible medication. At these return visits, we made an effort to visit all participating households, irrespective of the test results, to assure confidentiality.

Urine results were available approximately 1 year after the survey was completed and they were returned in the same manner as the serology test results. It was a concern that the urine tests could not be done simultaneously with the survey interviews, leaving respondents a long time between providing specimens and receiving test results. We attempted to make up for this shortcoming by providing immediate treatment for STI symptoms during the survey, as described previously. Less than 2% of respondents were positive for chlamydia and gonorrhea infections. Respondents positive for chlamydia or gonorrhea infection had been treated for these infections based on symptoms by the time test results were available.

Infertility Investigations

Sixty-six women (54% of the 123 couples referred) came to KCMC for infertility examination and treatment, which were provided at no cost. However, only 36 men agreed to a semen analysis, so only 29% of couples referred completed the infertility investigations. Hence, the clinical findings may not be representative of all couples with infertility in the study area. Based on the clinicians’ assessment, some infertile women were offered tubal surgery and others were given hormonal medication such as clomiphene, whereas there was no treatment offered for men with abnormal semen. The infertility clinic started by this research project demonstrated a demand for infertility services and KCMC continued operating the infertility clinic.

Results

Of 2824 households sampled, 2523 (89.3%) households were interviewed (Fig. 1). These households had 2203 eligible women of which 2019 (91.6%) were interviewed. As many as 27.9% of women interviewed reported that they had no husband/partner, resulting in 1456 (72.1%) eligible men, of which 1233 (84.7%) were located and 794 (54.5%) interviewed. Women and men were about equally likely to consent to STIs/HIV testing and more than 70% gave blood and urine samples, although the latter accounted for only 64.3% of eligible women and 38.9% of eligible men. Individuals consenting to STI/HIV testing had characteristics suggesting they were more likely to have an STI/HIV infection as compared with those who did not consent.¹⁴ However, risk factors for STIs/HIV were similar to those in other studies supporting the generalizability of the findings.

Fig. 1: Response rates in the 2002–2003 Moshi infertility survey.

Discussion

This study succeeded in conducting face-to-face interviews with a community-based sample of women and 91.6% of eligible women participated in these interviews. In addition, more than 70% of interviewed women provided samples for laboratory testing of STIs/HIV (64.3% of eligible women), which is considered high for an urban African population.³ In contrast, men’s participation rate in the interviews was relatively low at 54.5%. In Moshi, a sizable number of men engaged in occupations away from home, which is probably the main reason that 15.3% were not located at the time of interview (we cannot exclude that some men passed the information that they were away to avoid being interviewed). Men’s refusals included 30.3%. The men were suspicious about the intentions of the study, questioning the use of participating, opposed to give information about their sexual practices, and in general averse to participate in the interviews. Men, as compared with women, were less likely to collect their posttest STIs/HIV results or to participate in the clinical examination for infertility. However, the men who did participate provided as complete questionnaire information as the women and they were as likely to give blood and urine samples. Finally, 29% of the couples referred for infertility completed the examinations at KCMC, suggesting that the clinical infertility sample may not be representative.

We held regular sessions with the ward and street leaders providing information about the study goals and answering questions arising in the community about this project. These sessions prevented negative community responses or opposition, but they did not seem to influence the male participation rate. Although an appointment for the husband/partner interview would be set for an evening or weekend after the woman’s interview, often the husband/partner was not present at the time of the appointment or refused to participate. In January 2003, we started to contact the husbands/partners at their workplace and this helped to increase the male participation rate. Previous surveys in Tanzania such as the DHS found also that women had higher participation rates than men and participation rates were lower in urban compared with rural areas (our NBS collaborators worked on the DHS).

One major goal of this study was to provide data for analyses of infertility. Hence, the National Institutes of Health funding was limited to collecting information about women and their partners. It was not anticipated that 27.9% of interviewed women had no husband/partner. During the survey period, it became obvious that women only reported a husband/partner when they were married or cohabiting, suggesting that women in less stable relationships did not report their husbands/partners. This factor reduced the number of eligible men and limited the sample to men in unions, whereas single men and men in tenuous relationships were excluded from the study. As a consequence, the women data are representative of the community, whereas the men data only represent men in union.

At the time of survey interviews in 2002–2003, HIV voluntary testing centers were not yet established in Moshi, the majority of respondents did not know their HIV status, and antiretroviral therapy for HIV was not available. Hence, this study, including the STI/HIV counseling, was crucial in raising awareness about HIV, other STIs, and infertility. For instance, very few people had heard about the virus HSV-2, and knowledge about the role of HSV-2 as a risk factor for HIV transmission was limited. Thus, this 2002–2003 study and other more recent studies in Moshi have raised the community’s understanding of the value of research on STIs/HIV, and enhanced laboratory facilities and technical expertise in STI/HIV testing. For example, PCR testing can now be done at KCMC.

Approximately 85% of the interviewed women reported that they anticipated to continue living in Moshi in the next 5 years, suggesting that most women subjects who participated in this study could be reached in the future as part of a follow-up study. Some exploratory work investigating ways to obtain higher male participation rates are required and it would be desirable to interview all men, say age 18 to 59, instead of only men in union. This type of data collection would complement the nationally representative Tanzania HIV/AIDS Indicator Survey 2003–2004³ and contribute to estimating STIs/HIV district-level incidence, trend, and risk factors. Such information would be valuable in developing and testing new interventions aimed at reducing STIs/HIV infections in this population.