HIV epidemics in Asia have remained relatively low compared with those in sub-Saharan Africa. However, epidemics differ greatly in magnitude, and they have been mainly driven by either sex work, drug injection or a combination of both [1,2]. The role of demographic, socioeconomic and behavioural features in explaining differential risk of HIV infection has been well documented among defined high-risk groups in Asia [3–5]. Examples in this regard are the association between HIV and mobility, work situation and duration of commercial sex [3,5]. However, there are a limited number of studies from Asia that assess the risk of HIV on the basis of demographic characteristics and socioeconomic status (SES) in the general population .
In Cambodia, substantial transmission of HIV took place among the general population in the 1990s . The 2003 estimates of HIV prevalence by surveillance groups were 20.8% among brothel-based female sex workers (FSWs), 11.8% among nonbrothel-based FSWs, 2.5% among policemen and 2.2% among pregnant women attending antenatal care clinics (ANCs) . It has been suggested that the common transmission of HIV in Cambodia in the 1990s was related to structural factors linked to conflicts and poverty associated with a growing sex industry, migration, dislocation and a high prevalence of sexually transmitted infections (STIs) .
More than 10 years of intensive preventive interventions targeting sex workers and their clients seem to have resulted in a drop in HIV prevalence among the defined high-risk groups (sex workers, military and police) and ANC attendees. The main intervention programmes consisted of STI care services and the 100% condom use programme (CUP), a multisector approach that involved local authorities, health workers, police, brothel owners, sex workers and outreach and peer educators. The programme's aim was to promote condom use in all brothels and sex-related establishments. This decline was convincingly supported by a reduction in HIV-related risk behaviours, STI among high-risk groups and HIV incidence [10,11]. However, the estimated HIV prevalence in Cambodia is still high in Asia.
On the basis of trend data derived from the national HIV surveillance system, the estimated prevalence in the general adult population aged 15–49 years seems likely to have peaked in 1997–1998 at about 3%, and it was reduced to 1.9% in 2003 . However, the validity of surveillance-based prevalence and trends over time may be questionable , and there was an urgent need for validation through the use of population-based data. The first national population-based survey, the Cambodia Demography and Health Survey (CDHS), was conducted in 2005 by including a module of HIV-related questionnaires and an HIV test. We examined HIV prevalence patterns in the general population of men and women aged 15–49 years.
Sampling strategy and HIV testing
CDHS 2005 data were used with permission from the National Institute of Public Health (NIPH). The survey was approved by the National Ethics Committee for Health Research of Cambodia and Institutional Review Boards of the collaborating agencies. The CDHS was a national representative sample of women and men from all 24 provinces, aged 15–49 years. The survey used a two-stage sampling design with urban and rural stratification. In the first stage, 557 villages (clusters) were selected, with a probability proportional to the village size, from the national sampling frame updated by the National Institute of Statistics. The second stage involved the selection of households from the household lists using systematic sampling . All women in the selected households were eligible for interviewing. In the 50% subsample, all women and men eligible for interviews were also eligible for an HIV test after providing their informed consent. Interviews were conducted face-to-face and sex matched. All interviewers were trained for 5 weeks, followed by a full week of field practice. The detail of the interviewing method and data collection was described elsewhere . Blood samples were collected by finger-stick dried blood spot (DBS) cards, and HIV testing was conducted at the NIPH. DBS specimens were tested for HIV antibodies with ELISA 1 (Vironostika HIV Uniform II; Biomerieux SA, Marcy-l'Étoile, France) because of its high sensitivity (100%). All positive samples and 10% of negative samples by ELISA 1 were subsequently tested with ELISA 2 (Murex 1.2.0) because of its high specificity (98.9%). External quality control was performed by the US Centers for Disease Control and Prevention Atlanta laboratory .
We analysed the male and female data sets separately using Stata 10 (Stata Corp, College Station, Texas, USA) taking into account sampling weights, cluster effects and urban–rural stratification in order to assess risk profiles of the population related to HIV infection. The sampling weight is an inverse of sampling probabilities that were calculated separately for each sampling stage and each cluster . Within the scope of this paper, we limited our variables to demographic characteristics, SES and a variable related to risk behaviour, that is, multiple sex partners, comparing those who reported more than one partner in the past year with those who reported only one partner, in order to examine differential risks by sex. Chi-squared and t-tests were used to compare differences between urban and rural populations with respect to demographic characteristics, SES indicators, education level, wealth, risk behaviour and HIV status. Logistic regression was used to analyse factors associated with HIV with adjustment for age group. HIV prevalence estimates were standardized using the national population for 2005 as the standard .
Provinces were grouped into three geographical regions: the Thailand border region (Banteay Meanchey, Pursat, Battambang, Pailin, Odor Meanchey, Koh Kong and Sihanouk Ville), the central region (Kampong Chhnang, Kampong Speu, Kampong Thom, Kandal, Phnom Penh and Siem Reap) and the Vietnam border region (Kampong Cham, Kratie, Prey Veng, Svay Rieng, Takeo, Kampot-Kep, Preah Vihear-Steung Treng and Rattanakiri-Mondulkiri). Urban populations were defined as people living in provincial capitals, whereas rural populations were those living elsewhere. The survey did not directly collect data on household income or expenditure. Data were collected on dwelling and household characteristics, goods and assets, which were subsequently used to generate a standardized score called the household wealth index that reflected the household's economic status. It has been suggested that the wealth index is highly comparable to conventionally measured consumption expenditures . The original five wealth quintiles of the households in the data were redistributed into three categories: poorest/poor (low), middle and rich/richest (high).
Overall, 90% of the men (6514/7229) and 95% of the women (8188/8638) agreed to be interviewed and provided blood samples to be tested for HIV. Table 1 illustrates in detail the sociodemographic characteristics of the participants. Urban residents represented about 24% of all participants. The mean age for men was 28.8 years (median 27 years) and for women was 29.9 years (median 29 years). The younger age group (15–24 years) represented about 40% of the participants. Divorced/widow(er) or separated represented less than 3% of the men and 10% of the women. Both educational attainment and employment pattern depended clearly on residence and sex (Table 1). About 80% of urban and 35% of rural respondents were ranked high according to the household wealth index. There was a strong association between wealth index and the highest level of education attained. The rich/richest group was likely to be more educated than the poorest/poor for both men (66.6 vs. 16.5%; P < 0.001) and women (75.4 vs. 21.3%; P < 0.001).
HIV prevalence differentials
Tables 2 and 3 show the distribution of HIV prevalence according to selected factors. Both urban men and women had a significantly higher HIV prevalence than their rural peers. The age-standardized prevalence was 0.61% [95% confidence interval (CI) 0.2–1.8] among men and 0.62% (95% CI 0.3–2.1) among women. HIV infection among urban men was about four times higher than among rural men [odds ratio (OR) 3.8; 95% CI 1.6–9.3]. Similarly, HIV prevalence among urban women was 3.1 times higher than among rural women (95% CI 1.6–5.9).
Tables 2 and 3 revealed the highest prevalence among men aged 30–39 years (1.23%), among women aged 25–29 years (1.28%) and the lowest prevalence in the less than 20-year-old age group. The mean age difference between women and their spouses was about 3–4 years, and 30% of the women were about at least 5 years younger than their husbands. HIV prevalence varied with the age difference between women and their husbands, and was the highest when differed by more than 10 years (OR 5.2; 95% CI 1.6–16.5). The likelihood of being HIV seropositive differed greatly by marital status, that is, HIV transmission had been limited to those having ever been married. In the youngest age group, most men were single (>80%), whereas among women, a significantly higher rate of infection was observed in married compared with single women (0.76 vs. 0.10%; P = 0.006).
Higher educational attainment (secondary/higher) tended to be positively associated with the risk of infection among men (OR 3.7; 95% CI 0.8–17.8), especially among highly educated urban men (Table 2), and tended to be negatively correlated in women. However, among the youth, HIV prevalence was less than 0.1% in the highly educated compared with about 0.8% in noneducated men and women. Men without jobs and those working in agricultural sectors were found to have a very low HIV prevalence (<0.25%) compared with men in professional jobs (1.32%) or labour/service jobs (1.7%). No such differences were observed in women across job categories. The wealth index was found to be significantly correlated with HIV. Men who scored high on the wealth index had the highest HIV prevalence (OR 5.7; 95% CI 2.0–16.4). The pattern was similar among women (OR 2.0; 95% CI 1.0–4.2). Among sexually active people, 9.3% of men reported multiple sex partners, about 65% of whom were sex workers. Urban men reported multiple sex partners at about twice the rate (16.6%; 95% CI 13.0–20.9) of rural men (7.9%; 95% CI 6.6–9.4). In contrast, fewer than 1% of women reported multiple sex partners. Men who reported multiple sex partners were four times more likely to be infected than those with a single partner (95% CI 1.3–12.5) (Table 2).
This survey revealed the HIV prevalence to be relatively low and substantially below previous estimates based on the surveillance system among ANC attendees, that is, 0.6 vs. 1.9%. The prevalence did not differ by sex but was substantially higher in urban than in rural settings and among married individuals than among singles. Furthermore, a high age difference between spouses substantially increased the risk among women of being HIV infected. The two indicators used for SES revealed somewhat different association between sex and infection. The prevalence among men increased in correlation with both household wealth and education level, whereas among women, infection increased with household wealth but tended to decrease with education level. Women did not report nonmarital sex, whereas 10% of the men reported more than one sex partner in the previous year. The majority of these men reported paid sex. This finding, together with the observed sociodemographic pattern of HIV, suggests that most women were infected by their spouses.
Our study indicated a positive relationship between the household wealth index and HIV infection. A number of studies [18,19] showed that wealth, rather than poverty, has been positively related to HIV prevalence in the past. However, a possible interpretation is that the rates may differ by sex. In our setting, in which the HIV epidemic is mainly driven by sex work and male clients, paid and extramarital sex were likely to be strongly associated with higher SES among men. In contrast, the higher HIV prevalence among women in the wealthier quintiles is likely to be linked to their husbands' behaviours rather than the women themselves. Hargreaves and Glynn  have previously suggested this precise relationship. This seems to be supported by our finding of higher education to increase the risk of HIV among men but not among women. Studies [20–23] in the earlier stages of HIV epidemics in Africa reported a positive association between educational attainment and HIV. However, this pattern later reversed to a situation in which HIV became negatively associated with education level [23–26]. In Cambodia, where the magnitude of the epidemic is still relatively low, educated men may feel freer to engage in risky behaviours associated with different types of sex work. On the contrary, educated women may be more independent in terms of economic and negotiation power with their partners regarding safer sex behaviour. There is a need for intensive interventions aimed at curbing the transmission between spouses while sustaining the preventive programme efforts focused on commercial sex [8,27].
We found the likelihood of HIV infection among women to increase with the age difference between spouses. Furthermore, the age-specific HIV prevalence among women peaked about 5 years earlier than among men, and among spouses, the women were on average 4 years younger than men. This has also been reported in African settings [28,29]. As older men were more likely to be infected with HIV, the age mixing pattern seems to play a major role in the transmission from the older partners to the younger women [2,29]. In Africa, this type of sexual mixing pattern (young women forming sexual relationships with older partners) is observed in nonmarital sex, and there have been suggestions that efforts to reduce it could be an important HIV preventive strategy .
Surveys are faced with a number of potential biases. In our survey, the nonresponse was low and is not likely to have substantially affected the results. The second concern is that survival among HIV-infected individuals may be associated with SES , that is, longer survival among the better off than among the poor due to better access to care and treatment. However, this type of bias is not likely to be substantial as the scaling up of antiretroviral treatment began in 2005. Social desirability bias is well known in surveys and is especially true when asking women about past sexual practices . However, a study  among female factory workers in Cambodia found that fewer than 4% of women reporting never having had sex tested positive for sexually acquired herpes simplex virus. This study  confirmed a high reliability of behavioural data that were consistent with biological data. A final concern is the possible bias that may be due to the undersampling of some unreachable groups (i.e., high-risk groups, migrant workers). However, the Cambodia's Consensus Group found little impact on HIV estimates when taking into account these nonhousehold populations . In the result update of HIV estimates for Cambodia, both the trends of surveillance data and CDHS 2005 data were taken into account to get a better HIV prevalence estimate. The previous estimate that was purely based on ANC surveillance data is likely to overestimate the true prevalence .
In summary, this first national population survey revealed the HIV prevalence to be relatively low and substantially below previous estimates. The 100% CUP has been shown to be highly effective in Cambodia and should therefore remain as the core preventive strategy. Furthermore, there is an obvious need to develop interventions to reduce the HIV transmission from men to women. In this regard, efforts to empower women for better access to information, education and care seem critically important.
H.S., V.S. and C.C. conceived and designed the study. H.S. conducted the statistical analyses, interpreted the data and wrote the first draft of the paper. V.S. and C.C. were involved in the data interpretation. K.F. assisted and guided H.S. in the statistical analysis, data interpretation and provided conceptual ideas in drafting the paper. H.S. and all authors contributed to the final draft. All authors read and approved the final manuscript. We would like to thank the Cambodia's National Institute of Public Health for its permission for the authors to use the CDHS data.
H.S. has received A scholarship from the Quota Programme funded by the Norwegian government for his PhD programme at the University of Bergen. There are no conflicts of interest.
1. Ruxrungtham K, Brown T, Phanuphak P. HIV/AIDS in Asia. Lancet 2004; 364:69–82.
2. AIDS in Asia: face the facts – a comprehensive analysis of the AIDS epidemics in Asia; 2004. www.who.int/hiv/pub/epidemiology/aids_in_asia.htm
3. Ohshige K, Morio S, Mizushima S, Kitamura K, Tajima K, Suyama A, et al
. Behavioural and serological human immunodeficiency virus risk factors among female commercial sex workers in Cambodia. Int J Epidemiol 2000; 29:344–354.
4. van Griensven GJ, Limanonda B, Chongwatana N, Tirasawat P, Coutinho RA. Socio-economic and demographic characteristics and HIV-1 infection among female commercial sex workers in Thailand. AIDS Care 1995; 7:557–565.
5. Entz AT, Ruffolo VP, Chinveschakitvanich V, Soskolne V, van Griensven GJ. HIV-1 prevalence, HIV-1 subtypes and risk factors among fishermen in the Gulf of Thailand and the Andaman Sea. AIDS 2000; 14:1027–1034.
6. Dandona L, Lakshmi V, Sudha T, Kumar GA, Dandona R. A population-based study of human immunodeficiency virus in south India reveals major differences from sentinel surveillance-based estimates. BMC Med 2006; 4:31.
7. Phalla T, Leng HB, Mills S, Bennett A, Wienrawee P, Gorbach P, Chin J. HIV and STD epidemiology, risk behaviors, and prevention and care response in Cambodia. AIDS 1998; 12:S11–S18.
8. Sopheab H, Neal JJ, Morineau G, Saphonn V, Vun MC. Report on HIV Sentinel Surveillance in Cambodia, 2003
. Phnom Penh, Cambodia: National Center for HIV/AIDS, Dermatology and STDs; 2006. www.nchads.org
9. Expanding the response to HIV and AIDS in the Kingdom of Cambodia: AIDS Impact Model, 2000
. National AIDS Authority, NCHADS, Ministry of Health, The Futures Group International, UNFPA; 2000.
10. Saphonn V, Hor LB, Ly SP, Chhuon S, Saidel T, Detels R. How well do antenatal clinic (ANC) attendees represent the general population? A comparison of HIV prevalence from ANC sentinel surveillance sites with a population-based survey of women aged 15–49 in Cambodia. Int J Epidemiol 2002; 31:449–455.
11. Saphonn V, Parekh BS, Dobbs T, Mean C, Bun LH, Ly SP, et al
. Trends of HIV-1 seroincidence among HIV-1 sentinel surveillance groups in Cambodia, 1999–2002. J Acquir Immune Defic Syndr 2005; 39:587–592.
12. Michelo C, Sandoy I, Fylkesnes K. Antenatal clinic HIV data found to underestimate actual prevalence declines: evidence from Zambia. Trop Med Int Health 2008; 13:171–179.
13. Cambodia Demographic and Health Survey: chapter 1 – Introduction
. Phnom Penh, Cambodia: National Institute of Public Health, National Institute of Statistics and Calverton, Maryland: ORC Macro; 2005. http://www.measuredhs.com/pubs/pdf/FR185/FR185%5BNov-11-2008%5D.pdf
14. Cambodia Demographic and Health Survey: chapter 18 – HIV prevalence and associated factors
. Phnom Penh, Cambodia: National Institute of Public Health, National Institute of Statistics, and Calverton, Maryland: ORC Macro; 2005. pp. 239–241.
15. Cambodia Demographic and Health Survey: appendix A – Sample implementation
. Phnom Penh, Cambodia: National Institute of Public Health, National Institute of Statistics and Calverton, Maryland: ORC Macro; 2005. pp. 305–306.
16. Population projections for Cambodia 1998–2020, June 2004
. Phnom Penh: National Institute of Statistics, Ministry of Planning Center for Population Studies, Royal University of Phnom Penh; 2004.
17. Cambodia Demographic and Health Survey: chapter 2 – Household population and housing characteristics, household wealth
. Phnom Penh, Cambodia: National Institute of Public Health, National Institute of Statistics and Calverton, Maryland: ORC Macro; 2005. pp. 22–23.
18. Shelton JD, Cassell MM, Adetunji J. Is poverty or wealth at the root of HIV? Lancet 2005; 366:1057–1058.
19. Piot P, Greener R, Russell S. Squaring the circle: AIDS, poverty, and human development. PLoS Med 2007; 4:1571–1575.
20. Hargreaves JR, Glynn JR. Educational attainment and HIV-1 infection in developing countries: a systematic review. Trop Med Int Health 2002; 7:489–498.
21. Fylkesnes K, Musonda RM, Kasumba K, Ndhlovu Z, Mluanda F, Kaetano L, Chipaila CC. The HIV epidemic in Zambia: socio-demographic prevalence patterns and indications of trends among childbearing women. AIDS 1997; 11:339–345.
22. Fylkesnes K, Ndhlovu Z, Kasumba K, Mubanga Musonda R, Sichone M. Studying dynamics of the HIV epidemic: population-based data compared with sentinel surveillance in Zambia. AIDS 1998; 12:1227–1234.
23. Fylkesnes K, Musonda RM, Sichone M, Ndhlovu Z, Tembo F, Monze M. Declining HIV prevalence and risk behaviours in Zambia: evidence from surveillance and population-based surveys. AIDS 2001; 15:907–916.
24. Wojcicki JM. Socioeconomic
status as a risk factor for HIV infection in women in East, Central and Southern Africa: a systematic review. J Biosoc Sci 2005; 37:1–36.
25. Michelo C, Sandoy IF, Fylkesnes K. Marked HIV prevalence declines in higher educated young people: evidence from population-based surveys (1995–2003) in Zambia. AIDS 2006; 20:1031–1038.
26. Hargreaves JR, Bonell CP, Boler T, Boccia D, Birdthistle I, Fletcher A, et al
. Systematic review exploring time trends in the association between educational attainment and risk of HIV infection in sub-Saharan Africa. AIDS 2008; 22:403–414.
27. Pisani E, Garnett GP, Grassly NC, Brown T, Stover J, Hankins C, et al
. Back to basics in HIV prevention: focus on exposure. BMJ 2003; 326:1384–1387.
28. Gregson S, Nyamukapa CA, Garnett GP, Mason PR, Zhuwau T, Carael M, et al
. Sexual mixing patterns and sex-differentials in teenage exposure to HIV infection in rural Zimbabwe. Lancet 2002; 359:1896–1903.
29. Hallett TB, Gregson S, Lewis JJ, Lopman BA, Garnett GP. Behaviour change in generalised HIV epidemics: impact of reducing cross-generational sex and delaying age at sexual debut. Sex Transm Infect 2007; 83(Suppl 1):i50–i54.
30. Guest G, Bunce A, Johnson L, Akumatey B, Adeokun L. Fear, hope and social desirability bias among women at high risk for HIV in West Africa. J Fam Plann Reprod Health Care 2005; 31:285–287.
31. Sopheab H, Gorbach P, Detels R, Harwell H, Pugatch D. Sexual risk and HIV/STD in vulnerable Cambodian females: the Cambodian young women's cohort – factory workers
. Los Angeles: National Center for HIV/AIDS, Dermatology and STDs, University of California; 2006. pp. 32. http://www.nchads.org/Publication/Others/Factory%20Workers.pdf
32. Surveillance unit: report of a consensus workshop on HIV estimates and projections for Cambodia
. Phnom Penh, Cambodia: National Center for HIV/AIDS, Dermatology and STD, Ministry of Health; 2007. http://www.nchads.org/Events/28062007/report.pdf