Table 1 Summary of evidence for changes in behaviour
Strength of evidenceConclusions
Modelling of HIV incidence and prevalenceModerately strong for timing

  • Suggest incidence peaked about 1987/8 and began to decline

  • Suggest incidence declined more rapidly about 1993

Reports of behaviour change in newspaper articlesStrong for timing; very weak for representativeness

  • Indicate behaviour change began in 1987 in Kampala and some other places in Uganda

  • Suggest initial primary behaviour change was greater faithfulness (fewer casual partners) and not greater condom use

DHS and GPA surveys with questions about personal behaviour changeStrong for representativeness; weak for validity

  • Indicate large percentage decrease in sex before or outside marriage

  • Indicate small percentage began using condoms

DHS survey dataVery strong for representativeness

  • Demonstrate small increase between 1988/9 and 1995 in all women who abstained from sex during the previous year

  • Demonstrate large increase between 1988/9 and 1995 in young single women who abstained from sex during previous year

  • Demonstrate large increase in condom use during sex with non-marital and non-cohabiting partners by 1995, especially in urban areas

GPA survey dataModerately strong for representativeness

  • Among all women, suggest large decrease between 1989 and 1995 in percentage who had sex with non-marital or non-cohabiting partners

  • Among married women, suggest very low and stable percentage who had extramarital sex

  • Among young women, suggest large decrease between 1989 and 1995 in percentage who had premarital sex

  • Among single women, suggest decrease between 1989 and 1995 in percentage who had two or more partners

  • Among all women, demonstrate large increase in condom use during sex with non-marital and non-cohabiting partners by 1995, especially in urban areas

  • Among single men, suggest large decrease between 1989 and 1995 in percentage who had premarital sex

  • Among all men, suggest large decrease between 1989 and 1995 in percentage who had sex with non-marital or non-cohabiting partners

  • Among married men, suggest large decrease between 1989 and 1995 in percentage who had extramarital sex

  • Among young men, suggest decrease between 1989 and 1995 in percentage who had premarital sex

  • Among single men, suggest decrease between 1989 and 1995 in percentage who had two or more partners

  • Among all men, demonstrate large increase in condom use during sex with non-marital and non-cohabiting partners by 1995, especially in urban areas

Other surveys of sexual behaviourWeak for representativeness

  • Suggest decrease in the number of sexual partners

  • Suggest delays in the initiation of sex

  • Suggest increase in the use of condoms

  • Suggest moderately high level of condom use during casual sex in Kampala and other selected places by 1993

Reports of shipments of condoms to UgandaStrong for timing; strong for validity of receipt; weak for actual condom use

  • Suggest relatively few condoms in Uganda before 1989

  • Demonstrate that the number of condoms received in Uganda grew roughly exponentially

  • Demonstrate that there were substantial numbers of condoms in Uganda by 1993

Historical documents describing programmatic efforts to address AIDSStrong for timing; strong for validity

  • Demonstrate that, beginning about 1986, programmatic efforts focused primarily on being faithful and partner reduction

  • Demonstrate that, beginning in the early 1990s, condom promotion and provision encouraged condom use

  • DHS, Demographic and Health Survey; GPA, the World Health Organization’s Global Program on AIDS.