METHODS

Mwanza Region, in north west Tanzania, has an estimated population of three million, most children attend primary school, but few go on to secondary school.²⁷ Twenty rural communities, each with four to eight primary schools, were selected from four districts in Mwanza Region.

The baseline survey was carried out between August and December 1998. Teachers, parents, and community leaders were informed of the objectives of the study and all parents were given the opportunity to withhold their child from the study. The school register was used to identify pupils in years 4, 5, and 6, who were encouraged to attend school for the 2 days of the survey. On the first survey day, the study rationale and procedures were explained to all pupils, who were asked for basic demographic information and individual signed consent. Consenting pupils were then photographed, for subsequent preparation of photo-identity cards. Pupils who reported that their birth date was before 1 January 1985 were eligible to join the study.

The second survey day was held between 5 and 6 weeks later. Enrolment procedures were available for eligible pupils who had not been seen on the first survey day. A questionnaire was delivered to all eligible consenting pupils in a confidential interview with a young interviewer of the same sex as the respondent. The questionnaire recorded knowledge of STIs, attitudes towards key sexual health issues, and reported sexual behaviour. After interview, subjects were asked to provide a first void urine sample, and were seen by a clinician, who gave syndromic treatment for STI and offered confidential HIV counselling and testing. Pupils with conditions requiring further management were referred to the nearest health facility. A random sample of 20% of participants were re-interviewed by the clinician for quality control purposes.

One aliquot of urine was stored at +4°C until testing for HIV antibodies using a semiquantitative particle agglutination test (GACPAT-CPHL, London, UK) within 14 days of collection. Specimens that were positive or repeatedly indeterminate on GACPAT were tested with Wellcozyme HIV1+2 GACELISA (Murex Biotech Ltd, Dartford, UK) which provided the definitive result. HIV positive results from this survey were compared to results from blood spots from the same individual in subsequent surveys, and nine results were subsequently reclassified HIV negative.

Further urine aliquots were frozen at −20°C. The second aliquot was sent to the National Laboratory for Sexually Transmitted Diseases, Canada, for testing for CT and NG by pooled polymerase chain reaction (PCR, Roche Diagnostics Systems, USA). The third aliquot was kept as back-up. A fourth aliquot was taken from female pupils only and tested for pregnancy using the IPAS Quickstick HCG test.

Ethics and research clearance for the study was obtained from the Tanzanian National Medical Research coordinating committee and the London School of Hygiene and Tropical Medicine.

Statistical methods

All data were double entered, verified, and cleaned using Dbase IV (Borland) and analysed using Stata Version 6 (Stata Corporation). Sociodemographic factors were examined by school year and sex. Prevalence of HIV and other STI were examined by age and sex. Reported sexual behaviour was analysed separately for males and females, and compared to biological markers for sexual activity. For each sex, life tables were constructed using reported age of sexual debut, to obtain the proportion of males and females who were sexually active at each age. Knowledge, attitudes, and reported sexual behaviour were examined, with the principal comparison between the male and female pupils. Within each sex, the effect of previous sexual and reproductive health education on knowledge, attitudes, and reported sexual behaviour was examined.

The χ² test was used to assess the significance of comparisons, with the χ² test for trend to assess changes in prevalence over age. Logistic regression was used to calculate odds ratios (OR) and 95% confidence intervals (95% CI) as a measure of association between biological markers, reported sexual behaviour, and knowledge and attitudes after adjusting for age and community effects. Logistic regression was also used to compare primary school pupils in this school based survey with those from a previously reported community based survey.⁷ Two tailed significance tests were used throughout.

RESULTS

Study population

In the 121 primary schools in this study, there were 17 084 pupils registered in years 4, 5, and 6 (table 1). Eighteen pupils refused to participate in the study, and 2505 (15%) were absent on both survey days. Basic demographic data on the pupils and their households were obtained from 14 561 (85%) pupils. Further participation in the survey was restricted to the 9970 (68%) pupils who reported that they were born before 1 January 1985. Additional data and a urine sample were collected from 9283 (93%) of these pupils, as the remaining 687 pupils were not present on the second survey day. In each school year and overall, the average reported age of males was almost 12 months older than females. The age reported at enrolment and the age in the school register frequently differed. The reported age was usually higher, and the difference was similar for both males and females, but increased with the age of the pupil, with a mean difference of 3 months (SD 15 months) at age 13 years, rising to 15 months (SD 18 months) at 16 years. Reported age has been used for all subsequent analyses.

The demographic characteristics of the 9283 pupils who participated in the survey are shown in table 1. About two thirds of the pupils were still living in the village in which they had been born, more than three quarters belonged to the Sukuma tribe, and most reported that their parents were farmers. More girls than boys reported that they were Christian.

STI and other biological markers

All recruited subjects gave a urine sample, but HIV results were not obtained for three pupils (two male, one female), CT and NG for nine pupils (four males, five females), and pregnancy for 23 female pupils. Prevalences of biological markers are shown in table 2. Fourteen pupils (0.2%) were HIV positive, and there were no significant differences by age or sex. There were 16 male pupils (0.3%) and 67 female pupils (1.6%, p<0.001) who were positive for CT, and 12 pupils (two males, and 10 females, p<0.01) positive for NG. There was no trend of increasing prevalence of NG or CT with age in either males or females. Thirty two female pupils (0.8%) were found to be pregnant. The prevalence of pregnancy increased from 0.3% in 14 year olds to 2.5% in those aged 17 years and above (χ²_trend = 21.3, p<0.001). In total, there were 136 pupils, 22 males (0.4%) and 114 females (2.8%), with at least one biological marker of sexual activity (table 2).

On examination, 597 (11.7%) males were circumcised, of whom one was HIV positive and five were positive for CT. After adjusting for age and community, biological markers of any STI were significantly more common among circumcised (6/597, 1.0%) than uncircumcised (16/4537, 0.4%) males (OR 3.37, p = 0.02, 95% CI 1.18 to 9.60).

Sexual behaviour

Just over half of the males and one fifth of the females reported that they had ever had sex (table 3). Life tables of reported sexual debut are shown in figure 1 for males and females. However the reported age of sexual debut varied by the age of the respondent. For males, sexual debut by their 14th birthday was reported by 33% overall, but 42% of those aged 14 years old, 35% of those aged 15 years old, 30% of those aged 16 years old, and 21% for those aged 17 years or more (p<0.001, χ² for trend = 129). For females, sexual debut by their 14th birthday was reported by 7% overall, breaking down into 9% of those aged 14 years, 6% of those aged 15 years, 5% of those aged 16 years, and 3% of those aged 17 years and more (p<0.001, χ² for trend = 18.0).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Proportion of boys and girls reporting sexual debut by age.

Among those who reported having sex, over 70% of males, but only 46% of females reported they had had more than one sexual partner. Among the females, 23% reported that they had ever been forced to have sex by a boy or man. The types of sexual partner reported for the first sex act and the most recent sex act were similar, with over 80% of both males and females reporting sex with one of their peers (schoolmate or age mate). Only 4% of the pupils who reported having had sex said they had ever used a condom. There was no association between reported sex and presence of HIV or CT in males, and of the 27 males with a biological marker of sexual activity only 12 (44%) reported ever having sex (table 4). Among the females, there was a significant association between reported sex and the prevalence of CT, NG and pregnancy, but of 114 females with a biological marker of sexual activity only 45 (39%) reported ever having sex (table 4).

Reported sexual and reproductive health knowledge and attitudes

Forty two per cent of males and 48% of females reported receiving some sexual and reproductive health education during the past year, about sex, AIDS, the prevention of pregnancy, or STI (table 5). Knowledge that HIV can be transmitted sexually was high, but overall less than a quarter of the pupils were able to name any other STI. Although knowledge of HIV and STI was associated with recent sexual and reproductive health education, the difference was less than 10% for each of the individual STI. For both males and females, those who had received sexual and reproductive health education were more likely to know that a healthy looking person could have HIV or AIDS than those who had not received sexual and reproductive health education. On the other hand, sexual and reproductive health education was associated with an increased proportion of males and females who had misconceptions about catching HIV through sharing food with an HIV positive person, and catching STI through sharing a towel with someone who has an STI.

Less than one third of all pupils said there were ways that a woman can use to prevent pregnancy, but when prompted about various methods, many agreed with both correct and incorrect methods for pregnancy prevention (table 5). Significantly more males than females thought that pregnancy could be prevented by using condoms. A high proportion of both males and females agreed with locally common misconceptions, that pregnancy could be prevented by taking chloroquine tablets, and that pregnancy could be prevented through having sex in the standing position. sexual and reproductive health education was associated with a higher agreement with both correct and incorrect methods of pregnancy prevention among both males and females. More males than females felt that a girl could refuse sex with an older man, and sexual and reproductive health education increased this proportion among both males and females. Only one fifth of males and half of females felt that a girl could refuse sex if she had accepted a gift, and this proportion was lower for females who had received sexual and reproductive health education during the past year. Only one quarter of the males and one third of the females felt that a girl could refuse to have sex with her lover, and while this proportion was higher for females who had received sexual and reproductive health education, the proportions were similar among males.

DISCUSSION

As far as we are aware, this is the first reported school based survey of the prevalence of HIV and STI infection in primary school children in rural sub-Saharan Africa. The pupils in this survey formed a cohort that has been followed up for 3 years to assess the impact of a multifaceted sexual and reproductive health intervention targeted to primary school pupils. This survey provides baseline data on their sexual and reproductive health knowledge, reported attitudes and sexual behaviour, and risk of HIV, STI, and pregnancy.

The numbers of pupils registered in years 5 and 6 were substantially lower than those in year 4, 25% lower for boys and 33% lower for girls. These results are consistent with educational statistics in Tanzania, where some pupils drop out of primary school after national examinations at the end of year 4. With intensive community and school mobilisation, the survey team saw 86% of pupils registered in 121 primary schools. Although it was not possible to confirm this conclusively, the survey team had the impression that some school registers included some pupils who had not attended school for a long time.

This study used self reported age for the analysis, so that comparison could be made with other community based studies and to facilitate identification checks at subsequent follow up, when pupils had left school. The 9283 pupils enrolled in our study represent about 81% of all pupils aged 14 years and older in years 4–6 of these primary schools. This school based survey may not be representative of all adolescents in the community, because out of schoolchildren and pupils who do not attend school many have fewer constraints and more time to engage in risky sexual behaviour than their school peers.

HIV infections were present in this school based survey of primary school pupils, although the prevalences were low (0.1% in males and 0.2% in females) and showed little trend with age. In a community based survey in the same communities 1 year earlier, the overall HIV prevalence among adolescents aged 15–19 years was 0.3% in males and 1.0% in females who were still at school, and 0.7% in males and 3.0% in females who had left school.⁷ We compared the results from the two surveys by restricting the analysis to adolescents aged 15–17 years and still at school, and adjusting for year of age. There was a lower HIV prevalence in the school based survey for both males (OR 0.27, 95% CI 0.06 to 1.15, p = 0.08) and females (OR = 0.36, 95% CI 0.15 to 0.88, p = 0.024), although the difference in males was not statistically significant. In the community based survey, detection of HIV was based solely on antibodies in urine. In the school cohort, later follow up surveys, using dried blood spots, confirmed an HIV positive result in only 14 of the 23 subjects initially identified as positive using the urine test. Urine tests for HIV have been shown to have high sensitivity and specificity in another African population.²⁸ However, even with a specificity as high as 99.9%, in a population with a low HIV prevalence, the urine test produced many false positive results.

In this school based survey, the overall prevalence of CT was 0.3% in males and 1.6% in females. In the community based survey 1 year earlier, the prevalence of CT was similar both for adolescents who were still at school (0.5% for males and 2.1% for females) and for those who had left school (1.0% in males and 2.4% in females).⁷ Restricting the analysis to adolescents aged 15–17 years who were still at school, and adjusting for year of age showed no significant difference between the two surveys, in CT prevalence for either males (OR = 0.92, p = 0.85) or females (OR = 1.10, p = 0.74). The prevalence of NG was not measured in the community based survey, but was very low in this survey (0.04% in males and 0.3% in females). Overall, our survey found that 0.5% of male pupils, and 2.1% of female pupils had at least one biological marker of an STI. The tests for CT and NG are based on PCR detection and the pooling methods used in this survey have been shown to have extremely high sensitivity and specificity.²⁹ However the sensitivity of urine based PCR detection of NG has been questioned in other studies.³⁰

Almost 1% of female pupils aged 14 years and above in years 4, 5, and 6, were pregnant, with the prevalence of pregnancy increasing to 2.5% among female pupils aged 17 years and above. Given that girls who become visibly pregnant are usually not allowed to remain in primary school, the point prevalence of pregnancy is likely to have underestimated the burden of pregnancy in terms of early termination of schooling for girls. The pregnancy test used in this survey has been used in the United Kingdom for many years and has been shown to have high sensitivity and specificity (IPAS, NC, USA).

The proportion of girls who reported ever having sex was lower than among boys, and lower than that reported in similar community based surveys.⁷ This may have been due partly to the reluctance of schoolgirls to admit sexual debut, as the survey was held in school classrooms, although no teachers or adults were present, and young interviewers of the same sex as the survey subjects were trained in non-judgmental interview techniques. Other methods of collecting sexual behaviour data from the same adolescents, including self completion questionnaires administered in schools, have shown a higher proportion reporting sex.³¹ This indicates that data on self reported sexual behaviour obtained from adolescents in relatively brief face to face interviews must be treated with considerable caution. Other studies have used biological markers, including herpes simplex virus type 2 (HSV-2), as a measure of sexual behaviour.^32,33 Our survey measured HIV, CT, NG, and pregnancy, and almost two thirds of girls who were positive for any of these biological markers denied ever having sex. Given the low prevalence of these biological markers in this adolescent population, some of the biological results may have been false positives, but the discrepancy between reported sexual behaviour and the biological markers highlight the need for more accurate measures of sexual behaviour.²²

Among pupils who reported having had sex, over 80% of both boys and girls reported that their sexual partners were school pupils or adolescents of their own age. The relatively small number of reported sexual partners, may indicate that the sexual networks of the great majority of school pupils are at relatively low risk for HIV and other STI. However, the risk of acquiring HIV and STI after leaving school is considerable, as demonstrated by the prevalence of HIV and CT in older adolescents in the same communities.⁷

In our survey, 23% of girls reported that they had been forced to have sex by a man or boy, which is consistent with other studies in this region that have reported forced sex to be common.²⁴ In these communities, forced sex does not necessarily imply rape, but may include coercion, bribes, and threats against adolescent girls.³⁴ Further evidence that schoolgirls have little power to make decisions about who they can have sex with was provided by the responses to some of the questions regarding sexual attitudes. These showed that very few of the pupils believed that a girl can refuse sex with a man if she had accepted a gift from him, or has consented to sex with him on previous occasions. Measures are needed to improve girls’ ability to refuse unwanted sexual advances and to negotiate safe sex, and to change attitudes of the community towards sexual abuse and coercion.

Almost half of pupils reported having received some form of sexual and reproductive health education in the past year, although we did not ask about its content, or where and by whom it was delivered. Formal sexual and reproductive health education in Tanzanian primary schools has focused largely on knowledge of STI using didactic teaching methods and most sexual and reproductive health teaching in churches and mosques is limited to simple messages promoting sexual abstinence. Both correct and incorrect beliefs about pregnancy prevention and STI were more likely to be held by pupils who claimed to have received sexual and reproductive health education before the survey. Our survey shows that sexual and reproductive health education did not seem to have changed the reported attitudes of pupils to a girl’s ability to refuse sex under different hypothetical circumstances. This is consistent with other studies on the effectiveness of sexual and reproductive health education.^18,19

In conclusion, this survey of rural primary school pupils has shown relatively low prevalences of HIV and STI, but high rates of sexual activity, inadequate knowledge of sexual and reproductive health, and sexual attitudes that are likely to put adolescents, and particularly girls, at high risk of STI and unwanted pregnancy. With HIV prevalence rising to 5% by age 19, data from the community based survey⁷ suggest that girls are at high risk shortly after leaving school. These findings highlight the need for high quality, innovative sexual and reproductive health education for adolescents that includes efforts to change attitudes relating to sexual decision making. This should include skills based approaches, opportunities for self reflection and internalisation of decisions taken by participants, and be linked to youth friendly sexual and reproductive health services and initiatives within the wider community.²⁶ The vision of the 1994 International Conference on Population and Development will only be fully realised when young people, especially young women, are empowered to avoid unwanted or risky sex and to make informed decisions about their own sexual behaviour within a supportive community environment.