The three main intervention strategies currently available to reduce heterosexual HIV transmission in sub-Saharan Africa involve increasing condom use, reducing the numbers of sexual partners and seeking prompt and appropriate care for sexually transmitted diseases (STDs). These strategies rely mainly on modification of behaviour, a major component of which is the individual‚s belief that he or she is genuinely at risk. This concept of personal risk or vulnerability underlies the theoretical and practical aspects of behaviour change programmes for HIV/AIDS prevention [2,3] but has been little studied[4-6]. This is particularly the case for populations for which HIV prevalence is well characterized. Understanding risk perception is important for designing appropriate behaviour change interventions for HIV prevention. This paper describes the levels and determinants of perceived risk in 15 000 rural adult Ugandans with known HIV-1 serological status.
The population was a subset of participants recruited into a randomized community intervention trial to evaluate the efficacy of a standardized behavioural change programme - Information, Education and Communication (IEC) alone and in combination with improved management of STDs in reducing the incidence of HIV-1 and other STDs. These are largely Baganda subsistence farmers residing in scattered households and small trading centres in Masaka district, south-west Uganda.
As part of the pre-intervention assessment, a census of all household members, an individual interview and a serological survey involving the collection of a blood sample from all consenting adults was carried out. Unique identification numbers assigned to each individual at census enabled questionnaire and serological data to be linked. The individual interview consisted of an interviewer-administered questionnaire that collected descriptive information on socio-demographic factors such as age, sex, religion, marital status, education and duration of residency on compound. Participants who were sexually experienced were asked about STD history, knowledge, attitudes, beliefs about HIV transmission and sexual behaviours relevant to HIV/AIDS and other STDs. The survey was carried out on all adults in the three to five villages closest to the Government Health Unit, selected so as to enrol 700 to 1000 adults (13 years and older), in each parish.
In order to assess risk perception and collect more detailed data on sexual behaviour, a 50% random sample of households was selected. Risk perception was assessed by asking: ‚Do you think you are at risk of becoming infected with HIV?‚ The options of answers were: ‚yes‚, ‚no‚, ‚do not know‚, and ‚already infected‚. For those who answered ‚yes‚ or ‚no‚, the reasons why they thought as they did were explored and answers recorded verbatim and later coded. Sexual behaviour histories were only obtained from those who had ever been married or who were aged 13-24 years and reported having ever had sex. Questions included numbers and type of sexual partners to date, experience of casual sex during the 12 months preceding the interview and use of condoms as a means of risk reduction. Condom use was explored by asking participants whether they (men) or their partners (women) had ever used a condom. Those who reported having had casual sex were asked whether they had used a condom on the last occasion and the main reasons for use or non-use.
HIV-1 sero-status was determined using two independent enzyme-linked immunosorbent assays; Recombigen HIV-1 EIA (Cambridge Biotech, Corporation, Worcester, Massachusetts, USA) and Wellcozyme HIV-1 Recombinant (Wellcome Diagnostics, Dartford, England, UK) following set algorithms  and rigorous quality control procedures.
Data entry was done in duplicate and verified using EPI-INFO. Statistical analyses were performed using STATA. Odds ratios (OR) for risk perception were estimated by maximum likelihood and adjusted where appropriate for HIV infection and socio-demographic factors such as age and marital status using logistic regression. For age adjustment, data were stratified into five groups: 13-19, 20-24, 25-34, 35-44 and 45+ years. Approximate χ2 tests of statistical significance for odds ratios were obtained from log-likelihood ratio tests and 95% confidence intervals (95%CI) were obtained using the Gaussian approximation to the log-likelihood.
A total of 21 313 adults (10 540 males and 10 773 females) resident in sixty-seven villages, of whom 14 818 (70%) had a definitive HIV-1 sero-status, were eligible for study and of these 8142 were randomly selected to respond to the risk perception questions (3670 males and 4472 females) and 7935 (97%) of these had a definitive sero-status. Overall, 51% (95% CI, 50-52) of these respondents perceived themselves at risk of HIV-1 infection, 32% said they were not, 13% did not know, 3% said they were already infected, and 1% refused to answer. Slightly more females than males said they did not know whether or not they were at risk (Table 1). Among those reporting themselves to be at risk of HIV infection, 11% of males and 13% of females were HIV positive. For those reporting themselves not at risk, 3% of males and 4% of females were sero-positive. HIV prevalence rates were 32% for males, and 42% for females, who thought they were already infected. Around 6% of individuals who said they did not know whether they were at risk or not were HIV positive. The remainder of the analyses of risk perception are based on the 6605 adults for whom HIV-1 sero-status was known and perceived themselves to be either at risk or not at risk of HIV infection.
HIV-1 prevalence in the total study population was 9.7% (95%CI, 9.2-10.2), 8.6% (95%CI, 7.9-9.2) in males and 10.7% (95%CI, 10.0-11.3) in females. Figure 1 shows the corresponding age-specific rates together with the rates of perceived personal risk within the same age bands for males and females. Although the percentages of those perceiving themselves at risk are considerably higher than HIV sero-prevalence rates in both sexes and for all age-groups (note different y-axis scales), the age and sex-specific patterns are similar. In both males and females, highest rates were found in those aged 20-34 years and lowest in the 13-19 years age-group and in those aged 45 years and over. In other words younger females consider themselves at higher risk than similarly aged men; a finding consistent with HIV-1 infection rates. The reverse is true in older age groups.
Table 2 shows the proportions perceiving themselves at risk by age and sex separately, as well as the corresponding odd ratios for perceived risk for HIV-positive and -negative individuals. The HIV-positive individuals were significantly more likely to perceive themselves at risk compared with those that were HIV negative (OR, 4.40; P < 0.001 for males and OR, 3.28; P < 0.001 for females). Among men, odds ratios increased significantly with age (χ2 for trend = 3.9; P = 0.05). For women, age-specific odds ratios did not increase linearly with age but did differ significantly from each other.
As well as being age-dependent, risk perception was also strongly influenced by marital status (data not shown). For both sexes, the highest levels of risk perception were among the married (66% in males, 77% in females), whereas 52% of single males and 49% of single females perceived themselves at risk. Using single individuals as the base for comparison, the level of perceived risk among divorced or separated males was almost two-fold and statistically significant after adjusting for age (OR, 1.81; 95%CI, 1.12-2.91; P = 0.01). Perceived risk was also higher among married males but this did not achieve statistical significance (age adjusted OR, 1.24; 95%CI, 0.94-1.64; P = 0.13). In comparison with single women, married women had the highest level of perceived risk (age adjusted OR, 2.88; 95%CI, 2.23-3.71; P < 0.001) followed by divorced or separated women (age adjusted OR, 1.72; 95%CI, 1.10-2.70; P = 0.02).
Time spent on compound (a proxy for migration), and education were both significantly associated with perceived risk for males and females (Table 3). Risk perception was highest among those with higher levels of education (primary and above) and also those who had moved into the community in the last 5 years. These relationships remained statistically significant even after adjusting for age and marital status except for time on compound in males.
Table 4 shows percentages of males perceiving themselves at risk according to various risk factors and corresponding odds ratios. Odds ratios are given unadjusted and also after adjusting for age, marital status, HIV infection, education and time spent on compound. The adjusted odds of risk perception increased significantly with the number of lifetime sexual partners (χ2 for trend = 79.4; P < 0.001). Men who reported 10 or more lifetime sexual partners were four times more likely to perceive themselves at risk compared to those reporting only one lifetime partner. Men who had had more than two partners in the 12 months preceding the study were more likely to perceive themselves at risk than those who reported one or no partner. Men who reported ever having had casual sex, ever using condoms, using condoms with last casual partner or having had a genital ulcer (either in the last 12 months or more than 12 months previously) were around twice as likely to perceive themselves at risk in comparison with men without such experiences. Adjusted odds ratios for each of these factors were statistically significant (P < 0.01). Muslim religion, a proxy for male circumcision in this population  did not show an association with risk perception (data not shown). Muslims and non-Muslims showed similar levels of risk perception (71 versus 66%). In females, condom use with the last casual partner was not significantly associated with increased risk perception (Table 5). Otherwise, statistically significant but generally weaker patterns of association to those found in males were observed.
A wide range of reasons was given by respondents as to why they thought or did not think they were at risk of infection. Among women, 54% thought they were at risk because of their partner‚s sexual behaviour whereas 42% of men thought so because of their own sexual behaviour. A total of 8% of males and 7% of females mentioned injections as the reason for being at risk. Other reasons included insect bites (1%), condom non-use (1%), blood transfusion (1%) and sharing sharp instruments (e.g. razor blades/safety pins) (1%). Those who thought they were at no risk also thought so for a variety of reasons; 48% of females and 40% of males mentioned abstinence. Other reasons given were being faithful to a partner (20%), avoiding casual sex (7%) and not sharing sharp-edged instruments (1%). Condom use was mentioned as a reason for not perceiving risk by less than 5% of males and females.
Adults living in sub-Saharan Africa, especially those in countries such as Uganda with mature HIV epidemics, are acutely aware of the high levels of AIDS-related illness and deaths. Anecdotal reports have associated this awareness with widespread perceived risk of HIV infection but supportive data are scarce. In this very large rural population survey, where one in 10 adults are sero-positive, 51% considered themselves to be at risk of infection with HIV. Risk perception showed similar age- and sex-specific patterns, and was significantly associated with HIV-1 sero-positivity. There was also strong correspondence between factors associated with perceived risk and known risk factors for HIV infection[12,13].
Limitations of cross-sectional surveys for epidemiological research are well known. The sensitive nature of data on risk behaviours and practices such as number of sexual partners, past STDs and condom use introduce further concerns. In this study, a confidential interview was administered by a well-trained multi-disciplinary team with long experience in these and related methods. There are reports of the validation of data from similar surveys in other rural Ugandan settings. Schopper et al.  examined the validity of survey information on AIDS-related sexual behaviours gathered during a Knowledge, Attitudes and Practices (KAP) survey in a rural district in Northern Uganda. They found consistency in reported and expected patterns of sexual behaviour, in age trends, and within-couple responses. They found however that women tended to under-report casual partnerships. This study is vulnerable to these and other similar shortcomings which should be borne in mind when interpreting the results.
Surveys of risk perception pose additional problems particularly in the African context. Separating personal risk from general risk is difficult. For example, in Luganda, questions of the form ‚do you think you are at risk‚ can be misinterpreted as ‚does one feel at risk‚. Conveying the exact meaning of risk and distinguishing between risk for the individual and the community risk is consequently difficult. Also, ‚risk‚ can often be interpreted more loosely as ‚chance‚ and seen as a possibility no matter how small. This disparity between lay and expert understanding of risk may be one explanation for the high levels of perceived risk found in this population. In this district, however, half of all deaths in adults and 80% of deaths in those aged 13-44 years have been found to be due to HIV/AIDS[18,19]. In fact 80% of adults reported having known a relative or friend, very much like themselves, who had suffered from AIDS. Against this background, the ‚chance‚ interpretation of risk is therefore not unreasonable. The high level of reported risk perception might also indicate awareness that risky practices are prevalent. This is perhaps not surprising given the high levels of knowledge about these factors in these communities. This also may explain why there was correspondence between the factors associated with perceived risk and risk factors for HIV infection. This correspondence, which was stronger in men, is encouraging and further supports the notion that high levels of knowledge of risk behaviours exists. Earlier in the epidemic, there were reports that HIV/AIDS was a punishment from God for wrong-doing or the activity of witchcraft. These findings indicate that people now recognize the important role of sexual behaviour in their risk of HIV infection.
Of concern are the generally weaker associations observed for women with regard to condom use with last casual partner although it comes as no surprise in the absence of female-controlled preventive methods. Reported genital ulcer symptoms in the last 12 months were also weakly associated with risk perception in women perhaps due to lack of awareness of symptoms. An alternative explanation for these two observations could be gender differentials in reporting of sensitive information rather than a real difference in risk factor knowledge between men and women.
It seems counter intuitive that in this population married women and divorced or separated women and men feel at ‚higher‚ risk than those who are single. For married women, this finding is perhaps less surprising as condom use is unacceptable within marriage whereas singles might have the option of condom use if they feel the need for it. The widowed and divorced are likely to feel more at risk because their ex-partners have either died or may have been promiscuous in the past, which may have been the cause of the marital breakdown.
A counselling and HIV-testing service was offered to the population under study but only 5% of the population have undergone an HIV test through this service and are aware of their HIV status. The majority of those who thought they were already infected must have based their judgement on suspicion rather than knowledge of HIV status. This is reflected in the finding that close to 60% of those perceiving themselves to be already infected were in fact HIV negative.
Risk perception was assessed by asking: ‚Do you think you are at risk of becoming infected with HIV?‚ The options of answers were: ‚yes‚, ‚no‚, ‚do not know‚, and ‚already infected‚. These categories may not be sensitive enough and particularly in the context of HIV infection where present, past and future risk can be aggregated. Irrespective of this limitation, 83% of those interviewed were able to give a ‚yes‚ or ‚no‚ response. More research is needed on how best to measure and quantify risk perception in large-scale surveys.
Abstinence from sex was much more commonly mentioned as a protective practice than condom use despite earlier observations that abstinence is not a feasible control measure. We have reported previously that in these populations abstinence may include ‚not having sex at the moment‚ or ‚having less sex than before‚. The low reported risk reduction role of condom use (5%) contrasts with the relatively higher levels of ‚ever use‚ of condoms in the population (22% males, 12% females) for which HIV prevention was given as the main reason by both sexes and all age groups. Sixty-eight per cent of males who reported ever using a condom did so to prevent HIV transmission. Pregnancy prevention was a recognized reason for condom use by teenage females; 39% of users reporting it as the main reason for use. In contrast, only 15% of male teenagers who had ever used a condom did so to prevent pregnancy.
The differences between men and women in the main reasons given for risk perception are notable. In general women felt vulnerable because of their partner‚s sexual behaviour whereas men recognized that their own behaviour was the main determinant of their perceived risk. These data were however collected in the context of a large survey and could not elicit detailed reasons for risk perception. Such questions are better addressed by in-depth qualitative studies such as those which are currently underway in this population. Nevertheless, the relative powerlessness of women in sexual decisions and the absence of female-controlled preventive methods, emphasizes the importance of targeting AIDS prevention interventions to men as well.
Given the low levels of ever-use of condoms, and the failure to recognize condom non-use as a reason for risk, intensive condom promotion and distribution programmes are vital for this population. Female teenagers clearly recognize avoidance of pregnancy as an important reason for condom use but may not always be in a position to negotiate their use.
Summarizing our findings, half of the adults we have studied already see infection with HIV as a real possibility in their lives and are aware of HIV risk behaviours. More efforts should be made to implement sustainable control measures rather than simply raising awareness. In addition to recommending abstinence, these include mutual faithfulness, condom use and better treatment for STDs.
The authors are grateful to the following: the population of the study area, the field and laboratory staff, the Director of the Uganda Virus Research Institute and the Director of Medical Services, Ministry of Health, Uganda. They are also grateful to Gillian Maude for statistical input to some of the earlier analyses.
1. Kengeya-Kayondo JF. Strategies for HIV/AIDS control in Adults in Sub-Saharan Africa. J Epidemiol
2. Leviton L. Theoretical Foundations of AIDS-Prevention Programs.
In Preventing AIDS: The Design of Effective Programs.
Edited by O.valdiserri, R. New Brunswick: Rutgers University Press; 1989, 42-90.
3. Bandura A. Social Cognitive Theory and exercise of control over HIV infection.
In Preventing AIDS: Theories and Methods of Behavioural Interventions
. Edited by DiClemente RJ, Peterson JL. New York: Plenum Press; 1994, 25-59.
4. Cleland J. Risk perception and behavioural change.
In Sexual Behaviour and AIDS in the Developing World.
Edited by Cleland J, Ferry B. London: Taylor and Francis Ltd; 1995, 157-192.
5. Bunnell R. Promoting or Paralysing Behavior Change: Understanding Gender and High Levels of Perceived Risk of HIV infection in Southwest Uganda.
Unpublished doctoral dissertation. Boston: Harvard School of Public Health; 1996.
6. Lindan C, Allen S, Carael M, et al
. Knowledge, attitudes, and perceived risk of AIDS among Rwandan women: relationship to HIV infection and behaviour change. AIDS
7. Nunn AJ, Biryahwaho B, Downing RG, et al
. Algorithms for detecting antibodies to HIV-1: results from a rural Ugandan cohort. AIDS
8. Nunn AJ, Biryahwaho B, Downing RG, Ojwiya A, Mulder DW. Computer-assisted quality assurance in an HIV serology laboratory. Meth Inform Med
9. Dean AG, Dean JA, Burton AH, Dicker RC. Epi Info, Version 5: a Word Processing, Database, and Statistics Programme for Epidemiology on Microcomputers.
Stone Mountain, Georgia: USD, Incorporate; 1990.
10. StataCorp. Stata Statistical Software: Release 4.0
. College Station, Texas: Stata Corporation; 1995.
11. Clayton D, Hills M. Statistical Models in Epidemiology
. Oxford: Oxford University Press; 1993.
12. Malamba SS, Wagner HU, Maude G, et al
. Risk factors for HIV-1 infection in adults in a rural Ugandan community: a case-control study. AIDS
13. Nunn AJ, Kengeya-Kayondo JF, Malamba S, Seely JA, Mulder DW. Risk factors for HIV-1 infection in adults in a rural Ugandan community: a population study. AIDS
14. Hennekens C, Buring J. Epidemiology in Medicine
. Lippincott-Raven, 1988.
15. Ankrah M. AIDS: methodological problems in studying its prevention and spread. Soc Sci Med
16. Huygens P, Kajura E, Seeley JA, Barton T. Rethinking methods for the study of sexual behaviour. Soc Sci Med
17. Schopper D, Doussantousse S, Orav J. Sexual behaviour relevant to HIV transmission in a rural African population. How much can a KAP tell us? Soc Sci Med
18. Mulder DW, Nunn AJ, Kamali A, Nakiyingi J, Wagner HU, Kengeya-Kayondo JF. Two year HIV-1-associated mortality in a Ugandan rural population. Lancet
19. Nunn AJ, Mulder DW, Kamali A, Ruberantwari A, Kengeya-Kayondo JF, Whitworth JA. Mortality associated with HIV-1 infection over 5 years in a rural Ugandan population: cohort study. BMJ
20. Konde-Lule JK, Musagara M, Musgrave S. Focus group discussions about AIDS in Rakai district of Uganda. Soc Sci Med
21. Hooper E. AIDS in Uganda. Africa Affairs
, 1978, 86
Keywords:© 1999 Lippincott Williams & Wilkins, Inc.
HIV-1 prevalence; rural population; sub-Saharan Africa; risk perception; risk factors