Sample Sizes, Sampling and Survey Sizes

Sample Size

The sample size was calculated to detect a 25% reduction in the prevalence of syphilis and trichomoniasis, and a 40% reduction in cervicitis (gonorrhoea and/or genital chlamydial infection). The baseline prevalence was estimated to be 40% for syphilis and trichomoniasis and 20% for cervicitis, based on a previous survey in Ndola. The desired change to detect was 10 percentage points for syphilis and trichomoniasis, and 8 for cervicitis. The design effect was estimated at 2. The level of precision was set at 0.05 and the power at 0.80.

Applying the data above for the reduction in syphilis yields a sample size of 558 while applying the formula for reduction in cervicitis yields a sample size of 311. Choosing the larger of the two sample sizes and adjusting for non-response and rounding up, 800 sex workers were to be sampled. The sample was divided proportionally to the populations over the two bilaterally funded sites, Chipata and Livingstone, with 400 female commercial sex workers sampled in each site. An additional 400 female sex workers were sampled in Chirundu, which is a part of the Corridor of Hope Regional Project. Although evaluation criteria for the Corridor of Hope project were not fully developed based on statistical estimations, a sample size of 400 subjects was enough to satisfy the projected needs.

The selection of sites where the 1200 FSWs were selected was made according to city, since the cities included in the study contain different numbers of sites where FSWs may be found.

Working through the National AIDS Control Program (NACP) and relevant Non-Governmental Organisations (NGOs) in the different cities, a list was made of locations where FSWs congregate, including the approximate number of FSWs frequenting each site per day and per night.

Originally, interviewers planned on establishing a sampling frame, using 'time-location' clusters.  However, when the locations where FSWs congregate were listed and the number of FSWs at those locations estimated, it became clear that less than the desired sample size was operating at each site. In the end, interviewers used a 'take-all' approach, visiting all locations and contacting all sex workers who were present on that night, or who entered the site while the interviewer was working, to get consent for an interview.

The interviewers visited the chosen location on the appointed night from 10:00 p.m. onward. The interviewers contacted the women and obtained verbal consent for the interview. Each woman was offered the option of either being interviewed on the spot or making an appointment to meet at another place and time. Incentives were offered. If the woman wished to be interviewed on the spot, the interviewer offered her a drink in return for the time she was taking away from her work.

If the woman chose to meet at another time, the interviewer offered to pay the woman's transport and buy her lunch. Interviewers administered a standardised questionnaire with all consenting women in a private setting followed by witnessed consent for the biological component when biological specimens were obtained. At the conclusion, the subjects received a supply of condoms and a month's supply of vitamins to show appreciation for their participation.

Participation was voluntary and there was witnessed verbal consent given to all participants. Verbal consent was conducted in two phases first for the behavioural interview and again for the biological component. Both phases of consent were administered by the interviewer in a private setting and witnessed by the second study interviewer. The protocol, consent forms and draft questionnaires were submitted for approval to both the Zambian Ministry of Health Ethical Review Committee and the Protection of Human Subjects Committee of Family Health International.

The interviewer used a standardised questionnaire that was based on the BSS prototype. It included questions on socio-demographic characteristics, sexual behaviour and sex work characteristics, and knowledge, perceptions and practices regarding condoms, STDs, HIV, and HIV Voluntary Counseling and Testing (HIV VCT). A copy of the questionnaire is found in attachment 1.

A self-administered vaginal swab was inoculated into the culture medium for T. vaginalis. A second self-administered vaginal swab was directly placed in a cryovial and 10cc of venous blood was collected in a serum separation tube. All tubes were labelled with the subjects study number and collection date.

Women were anonymously linked to their specimen by a number and asked to present at a selected clinic the next day for a physical exam and treatment. All sex workers who presented to the clinic received treatment based on algorithms designed for sex workers. All sex workers received presumptive treatment for gonorrhoea and genital chlamydial infections. Those with vaginal discharge were treated for candida infection and/or trichomoniasis/ bacterial vaginosis, depending on the clinical characteristics of the discharge. Those with confirmed abdominal tenderness and/or cervical motion tenderness were treated for pelvic inflammatory disease.

Genital ulcers found during examination were treated for syphilis and chancroid. All women were asked to return to learn the results of their syphilis serology and Trichomonas vaginalis culture and to be treated accordingly. The choice of all drugs and dosage followed the national STD treatment guidelines. Free condoms were provided and the study staff ensured that participants were counselled regarding their STD and proper use of condoms.

T. Vaginalis: Women inserted a cotton swab into their vagina to collect a vaginal fluid sample. After collection the swab was inoculated into the upper chamber of the pouch of the TV InPouch test. The InPouch was pre-marked with the subjects study number and was put in a separate bag, not in the cool box. The bags were sealed and transported to the laboratory daily. The pouches were examined in the laboratory of the selected clinic on arrival, incubated at 37oC and examined by microscopy at 12, 24 and 48 hours post-incubation for appropriate morphology and motility of the protozoa.

N. gonorrhoeae and C. trachomatis: Study partic ipants collected a second vaginal fluid swab. The lab technician placed it into a 5cc cryovial that was immediately placed in a cold box. The cold box was kept cold with ice packs. At the end of each day, the cryovials were placed in a freezer.

At the end of the survey in each site, all the samples collected in the freezer were re-packed in the cold box with ice packs and brought to Ndola where they were stored at -20C. At the conclusion of the study, all the samples collected were packed in dry ice and shipped to the Institute of Tropical Medicine in Antwerp, Belgium for further analysis. The specimens were tested for N. gonorrhoeae and C. trachomatis using a strand displacement amplification (SDA) test. Positive results were repeated with an in-house polymerase chain reaction (PCR) test and discordant results were confirmed with a third DNA amplification test, the PCR test from Roche.

T. pallidum: The lab technicians collected 10cc of blood in a serum separation tube. The tube was immediately placed in the cold box and brought to the laboratory of the selected clinic at the end of each day. There, the sera were centrifuged at 2000rpm for 10 minutes to separate the sera from the cells. After centrifugation, the blood samples were analysed with a quantitative rapid plasma reagin (RPR) screening test. Positive results were confirmed with a Treponema pallidum hemagglutination assay (TPHA) test to indicate both recent and past infection. The remaining sera were divided over two cryovials of 5cc. Both vials were placed in the freezer. At the end of the survey in each site, all the samples collected in the freezer were re-packed in the cold box with ice packs and brought to Ndola and stored at -20C. Twenty percent of the total serum collected was submitted to ITM for quality control and assurance. Discordant results were retested at ITM.

Neisseria gonorrhoeae SDA and 1 of 2 PCR positive Vaginal swab

Chlamydia trachomatis SDA and 1 of 2 PCR positive Vaginal swab

Trichomonas vaginalis InPouch Culture positive Vaginal swab

Syphilis RPR and TPHA positive Sera

The field supervisors checked all questionnaires and clinical data forms to be sure they were complete, then transported them with the specimens and stored them at the TDRC office. Data entry was done at the TDRC office using Epi-Info software, with data entered twice to ensure accuracy. Data analysis was performed using Epi-Info 6.04.

Data were analysed in a descriptive way only, by site. Descriptive measures, such as simple proportions, means and medians were calculated to determine the prevalence of relevant variables by site and for the total sample.