Long-term Evaluation
Interpretation of the 2007-2008 Survey Results
NEW! Presentation of results of long-term evaluation (link to external website)
The 2007-8 survey investigated whether the lack of any significant beneficial impact of the intervention on the key biological outcomes seen in the 2001-2 survey was because such interventions needed more time to work. The results showed that, even when the intervention had been implemented for over 8 years, and 67% of the young people surveyed had received 3 years of the in-school intervention, there was no significant impact on either HIV or HSV-2, nor any consistent or clearly significant impact on other STIs, nor on reported pregnancy rates. However, the 2007-8 survey showed that the interventions had had a sustained beneficial impact on knowledge related to sexual health, although the impacts were less substantial in 2007-8 than in 2001-2.
MEMA kwa Vijana results policy brief
MEMA kwa Vijana results technical brief
The MkV Further Survey aimed to see whether the MEMA kwa Vijana 1 intervention programme led to a decrease in HIV and other STIs in the longer-term (i.e. approximately 8 years after participation in the intervention programme). It surveyed more than 13,500 young people between the ages of 17-26, including people who participated in the intervention programme whilst at school.
The 2007-8 trial survey evaluated the long-term impact of the intervention, as by that time nine consecutive school year groups had participated in the in-school component of the intervention, and the health services intervention had also been in place for 8-9 years. External evaluations conducted in 1999-2002 had found that the multi-sectoral intervention was well implemented, and achieved high coverage. Though the intervention was maintained in the 10 intervention communities after 2002 with continued support from AMREF, no such external evaluations were carried out in the trial communities after 2002. The long-term evaluation survey in 2007-8 was therefore restricted to young people who had attended at least one year of school Years 5-7 within the period from 1999-2002 inclusive.
In 2007-8, 13,814 young people were surveyed, almost twice as many as were seen in the 2001-2 survey (7,040). The median ages of the young men and women included in the 2007-8 survey ere higher than for those in the 2001-2 survey, so their HIV and HSV-2 prevalences were also higher. The combination of these two factors meant that the 2007-8 survey had much greater power to detect any true differences in the two primary trial outcomes. Furthermore, approximately 40% of the trial cohort included in the 2001-2 survey had only received one year of the in-school component of the intervention, with approximately 30% each having received 2 or 3 years, respectively. However, in the 2007-8 survey, 67% had received 3 years of the inschool component, 17% 2 years, and only 16% had received 1 year.
Results of the 2007-8 long-term evaluation survey
15,707 young people attended the survey, 13,814 of whom met the eligibility criteria for inclusion; 7,083 (51%) from intervention communities and 6,731 (49%) from comparison communities.
Knowledge and reported attitudes
Correct knowledge and desirable attitude reports were higher in intervention communities than comparison communities, and these differences were all statistically significant or borderline significant, except for the “attitudes to sex” score in females. However, these differences were not as
great as those observed in the 2001-2002 survey.
Reported sexual behaviours
Sexual debut and numbers of partners: Males in intervention communities reported sexual debut before the age of 16 years less often than males in comparison communities. Male and female intervention participants also tended to report fewer lifetime sexual partners. These differences were of borderline statistical significance, and were consistent with what had been observed in 2001-2.
Condom use: The absolute proportions of respondents who reported using condoms at last sex within the past 12 months were relatively low in both intervention and comparison communities, even when this was restricted to sex with a nonregular partner. Reported condom use was higher in intervention communities, although this difference was only statistically significant for reported condom use with a non-regular partner among young women.
Contraceptive use: Reported use of modern contraceptives (condoms, oral contraceptive pills, injectable contraceptives) was only assessed among young women in the 2007-8 survey. Modern contraceptive use ever and specifically at last sex were both reported more frequently by sexually active young women in intervention communities, but neither difference was statistically significant.
Concurrency: The period prevalence of reported concurrency of sexual partnerships was assessed in two ways within the 2007-8 survey (more than one partner within the same time period in the past 12 months, and more than one partner within the past 4 weeks). Both outcomes were reported less frequently by young men in the intervention communities, but neither difference was statistically significant. The two outcomes were inconsistent and not statistically significant in young women.
Use of health services for suspected STIs: For participants who reported STI symptoms within the past 12 months, there were no statistically significantly differences by intervention status in who reported attending a health facility for treatment.
Reported clinical/biological outcomes
Current genital discharge was only reported by 8% of young men and 5% of young women in the 2007-8 survey. Although reported less frequently by both young men and women in the intervention communities, these differences were not significant.
Current genital ulcers were also reported less frequently by both young men and women in the intervention communities, and both differences were of borderline statistical significance.
No consistent or statistically significant differences were seen in the various measures of the frequency of reported pregnancies.
Primary biological outcomes
The predefined primary outcomes of the trial were both based on biological outcomes, measured using laboratory tests on serum: HIV prevalence, and HSV-2 prevalence.
HIV: The participants’ median age of 22 years in males and 21 years in females, and the large sample size meant that a substantial number of HIV cases were identified in the 2007-8 survey (133 in males, 262 in females). However, the HIV prevalence was very similar in intervention and comparison
communities for both males (RR=0.91, 95%CI 0.50,1.65) and females (RR=1.07, 95%CI 0.68,1.67). The lower limit of the 95% confidence interval shows that it is extremely unlikely that the true impact of the intervention on HIV prevalence could have been greater than 50% in males or greater than 32% in females.
HSV-2: There was no evidence of any impact on HSV-2 prevalence in either direction, with tight confidence intervals (Males: RR=0.94, 95%CI 0.77,1.15; Females: RR=0.96, 95%CI 0.87,1.06).
Secondary biological outcomes
Syphilis, chlamydia, gonorrhoea: There was no evidence of any impact in either direction on syphilis. The prevalence of CT had been slightly higher in the intervention communities than in the comparison communities at baseline in 1998, and this difference persisted among both males and females in both 2001-2 and 2007-8, though it was not statistically significant on either occasion. The prevalence of gonorrhoea was not
measured at the baseline survey in 1998. The prevalence of (unconfirmed, threshold OD>1.0)) gonorrhoea was also higher, in both males and females, in intervention communities than in comparison communities in 2001-2. In 2007-8, the prevalence of gonorrhoea (unconfirmed, but threshold OD>2.0) was higher in males and slightly lower in females, but these differences were not statistically significant.
- Screening Trial Participants for HIV and Eligibility for Antiretroviral Therapy (346kb)
- Using Personal Digital Assistants (PDAs) for data collecti on in rural Mwanza, Tanzania (229kb)