Introduction
Cambodia is an agricultural country located in South-east Asia. It borders with Thailand to the west, Laos and Thailand to the north, the Gulf of Thailand to the south-west and Vietnam to the south-east. It has a total land area of 181,035 square kilometres (National Institute of Statistics, 2011). Although HIV/AIDS in Cambodia is considered to be on the decline, prevalence is still relatively high compared with other countries in Asia (Sopheab et al., Reference Sopheab, Saphonn, Chhea and Fylkesnes2009). Bearing this in mind, two pertinent issues call for attention. The first issue relates to how the prevalence data, which seem to suggest that prevalence is declining, are generated. Based on trend data obtained from the national HIV surveillance system, the estimated prevalence in the general adult population aged 15–49 years appeared to have probably peaked in 1997–1998 at about 3% (Sopheab et al., Reference Sopheab, Neal, Morineau, Saphonn and Vun2006), and diminished to 1.9% in 2003 (Sopheab et al., Reference Sopheab, Saphonn, Chhea and Fylkesnes2009). Nonetheless, the validity of surveillance-based prevalence and trends over time may be controversial due to potential underestimation (Michelo et al., Reference Michelo, Sandoy and Fylkesnes2008). This means there is a need to continue to investigate the factors that will improve understanding of the potential determinants of HIV spread in Cambodia. One area that requires greater attention in Cambodia relates to the function that may enable or disable a woman's ability to negotiate safer sex. Yet a limited number of studies in the burgeoning literature on negotiation for safer sex have relied on population-based data. The second issue relates to the fact that, although there is a decline in HIV/AIDS prevalence among the general Cambodian population, the epidemic is disproportionately high among specific social groups (Gray et al., Reference Gray, Heymer, Hoare, Kwon, Thein and Lote2009). In the case of women in Cambodia, this potentially emanates from a consistent inability on their part to negotiate for safer sex. This is particularly so because Cambodia has reported several forced marriages, and associated mental health feedbacks (Mollica et al., Reference Mollica, Donelan, Tor, Lavelle, Elias, Frankel and Blendon1993; Marshall et al., Reference Marshall, Schell, Elliott, Berthold and Chun2005).
Theoretical context
Broadly, in the literature, psychological, social and cultural factors impinge on women's ability to negotiate for safer sex (Sanders-Philips, Reference Sanders-Phillips2002). Some scholars identify psychological factors that undermine women's capacity to negotiate for safer sex (McNair & Prather, Reference McNair and Prather2004); in particular, experiences in women such as depression (Ickovics et al., Reference Ickovics, Thayaparan, Ethier, Baum, Revenson and Singer2000), trauma history (Wyatt et al., Reference Wyatt, Williams, Loeb, Carmona, Chin and Presley2002) and post-traumatic stress disorder (Myers & Durvasula, Reference Myers and Durvasula1999) have been emphasized. Interestingly, these psychological factors have been linked to the wealth status of women, as very poor women have been found to be excessively impacted. Social issues, such as gender relationships, power dynamics, socioeconomic factors, sex roles and experiences related to ethnicity, have also been identified (Amaro & Raj, Reference Amaro and Raj2000). Furthermore, negotiating safer sex is fundamentally regarded as a gender issue, and most definitely a critical element in the quest to reduce HIV/AIDS prevalence (Kippax et al., Reference Kippax, Noble, Prestage, Crawford, Campbell, Baxter and Cooper1997). Gender and power are fundamental to the analysis of women's HIV/AIDS risks (see Dixon-Mueller, Reference Dixon-Mueller1993; Amaro, Reference Amaro1995; Wingood & DiClemente, Reference Wingood and DiClemente2000; Gupta, Reference Gupta2001). Some researchers have argued that women usually have less power than their male sexual partners, which limits their ability to protect themselves from HIV risk (see Pulerwitz et al., Reference Pulerwitz, Gortmaker and DeJong2000; Exner et al., Reference Exner, Hoffman, Dworkin and Ehrhardt2003; Pulerwitz & Dworkin, Reference Pulerwitz and Dworkin2006). Socio-cultural factors also play an important role in determining women's ability to negotiate for safer sex. For instance, Logan et al. (Reference Logan, Cole and Leukefeld2002) identify cultural beliefs, values and other practices among factors that are related to women's inability to negotiate sex, thereby reinforcing their risk and vulnerability to HIV. Hence, in the Cambodian context, myths and awareness of HIV/AIDS may be important. On a conciliatory side, cultural values and norms influence, and sometimes define, the behaviour of women in interpersonal relationships in which sexual behaviours take place (Amaro, Reference Amaro1995). This is also true for Cambodian society, where culture and gender characterize women's negotiation for safer sex, especially in situations where women are coerced into early marriage. Although it has been argued that marriages of this nature may constitute a buffer against HIV risks (Bracher et al., Reference Bracher, Santow and Watkins2003; Clark et al., Reference Clark, Bruce and Dude2006), emerging evidence suggests that it may actually exacerbate HIV risks since younger women are, invariably, less experienced and consequently less likely to be involved in decision-making (including negotiating for safer sex) compared with their older counterparts (Dunkle et al., Reference Dunkle, Stephenson, Karita, Chomba, Kayitenkore and Vwalika2008; Tenkorang, Reference Tenkorang2012).
Further, conventional HIV prevention strategies, which encourage women to use condoms, do not recognize the barriers that women encounter in negotiating for safer sex with their male partners (Kippax et al., Reference Kippax, Noble, Prestage, Crawford, Campbell, Baxter and Cooper1997). Kippax et al. (Reference Kippax, Crawford, Waldby and Benton1990) draw attention to the fact that women differ in their ability to adopt safer sexual behaviour since this is based on their level of empowerment in decision-making.
To address many of the issues highlighted above, recent strategies developed to curb the HIV pandemic have underscored the value of communication and development of safer sex negotiation skills to reduce women's risk (see Logan et al., Reference Logan, Cole and Leukefeld2002). A growing body of literature elucidates the types of condom negotiation strategies used, as well as which strategies are more or less successful. In a study by Williams et al. (Reference Williams, Gardos, Ortiz-Torres, Tross and Ehrhardt2001), women proposed different condom negotiation strategies depending on whether the goal was to (a) initially negotiate condom use, (b) manage negative partner reactions, or (c) maintain safer sex over time. The importance of place in a woman's ability to negotiate safer sex has also been noted. For instance, Williams et al. (Reference Williams, Gardos, Ortiz-Torres, Tross and Ehrhardt2001) observed differences in rural and urban women's ability to negotiate for sex.
In many societies, as it is in Cambodia, the empowerment of women is expressed through their involvement in decision-making processes at the household level (Kirshor Reference Kishor, Presser and Sen2000; Kritz et al., Reference Kritz, Makinwa-Adebusoye, Gurak, Presser and Sen2000). For instance, married women's contribution at the household level to decision-making on their own health care, mobility in terms of visits to family and friends, and household expenditure have been used as surrogate measures of empowerment. These empowerment indicators border on women's autonomy, and potentially reflect whether married women are also capable of negotiating with their partners for safer sex or otherwise. Essentially missing from the burgeoning literature on negotiation for safer sex by married women is how these empowerment indicators per se influence their negotiation for safer sex.
This study, therefore, contributes to the extant literature by exploring this gap within the Cambodian context. Specifically, this paper examines how the empowerment of married women, in terms of their involvement in decision-making on health care, mobility and expenditure at the household level, affects their ability to negotiate for safer sex. Some policy implications of the findings and measurement issues pertaining to women's autonomy in Cambodia will be discussed.
Methods
Participants
Participants for this study were drawn from the 2010 Cambodia Demographic and Health Survey (CDHS), a nationally representative sample survey of 18,754 women and 8239 men aged 15–49. The 2010 CDHS is the third comprehensive survey conducted in Cambodia as part of the worldwide MEASURE DHS project. The CDHS also provides high-quality and reliable quantitative data on basic demographic indices, residential history and awareness of HIV/AIDS, including those related to sexual activity of women in Cambodia. The 2010 CDHS employed a two-stage, stratified sample frame, where systematic sampling with probability proportional to size was applied. Stratification was achieved by separating every reporting domain into urban and rural areas. Thus, the nineteen domains were stratified into a total of 38 sampling strata. Samples were selected independently in each stratum through a two-stage selection process. Implicit stratifications were achieved at each of the lower geographical or administrative levels by sorting the sampling frame according to geographical/administrative order and by using a probability proportional to size selection strategy at the first stage of selection. A total of 16,344 households were selected, of which 15,829 were found to be occupied during data collection. Among these households, 15,667 completed the Household Questionnaire, yielding a response rate of 99%. Only married women whose last sex partner was their husband were selected for this study, resulting in an analytical sample of 11,218. Out of this sample, 57 cases were not used in Model 2 and 3 in the multivariate analyses because of missing values. Married women were focused on because emerging evidence from developing countries indicates that married women face increased risks of contracting HIV compared with their never-married counterparts (Tenkorang, Reference Tenkorang2012). The argument is that, given the higher expectations for childbearing in marriages in developing countries, negotiating for safer sex, especially asking their husbands to use condoms, is difficult. Moreover, condom use within marriage is usually associated with infidelity and lack of trust.
Measures
Two main outcome variables that reflect women's decision-making on condom use and their ability to negotiate for safer sex were employed and analysed. The first variable asked married women whether they can ask their husbands to use condoms during sexual intercourse, coded as no=0 and yes=1; and the second, which asked whether they can refuse sex with their husbands, was also coded as no=0 and yes=1. Missing data were less than 0.05% (Stata automatically excluded these from the analyses). In theory, it is possible that there is a bidirectional (two-way) relationship between negotiating safer sex and autonomy. Therefore, in this study, this two-way relationship was accounted for by making either one of them the dependent variable. From the analysis, however, it emerged that, although autonomy predicts increased negotiation for safer sex, the reverse was not the case.
Predictor variables included in the analysis were organized into three main categories: socioeconomic status, women's autonomy (involvement in decision-making at the household level) and HIV/AIDS awareness and myths measures among married women in Cambodia. The first category (demographic and socio-cultural variables) included the age of the participants measured in 5-year groups, and the type of place of residence (urban=0 and rural=1). The socioeconomic status of participants was captured with their educational background (no education=0, primary education=1, secondary=2 and higher education=3) and also with wealth status, a composite index based on the household's ownership of a number of consumer items, including television and a car, flooring material, drinking water, toilet facilities, etc. (National Institute of Statistics, 2011). The second category (women's autonomy) was based on women's involvement in household-level decision-making (on health care, large household purchases and family visits). Responses to the three types of decision-making were classified into: ‘other than respondent’, ‘solely respondent’ and ‘respondent and husband’. The third category included HIV transmission awareness and HIV transmission myths. HIV transmission awareness was computed from three variables: reducing chance of infecting HIV by using a condom at every sexual encounter, having just one uninfected sex partner and healthy looking person can be HIV positive. In particular, attribution of HIV/AIDS transmission to witchcraft, mosquito bites and sharing of food was aggregated into a composite indicator called ‘transmission myths’. Transmission myths were treated as an underlying unobserved dimension that was estimated using latent variable technique (factor analysis). This was then transformed into a single item.
In addition, participants were asked if they had ever been tested for HIV (no=0 and yes=1) and had access to condoms (no=0, yes=1 and refused to answer=3). The ‘refused to answer’ category for this variable was created due to the lack of a response on this question from about 3000 participants. The fertility preferences of women and their husbands were taken into account. Regarding the former, women who wanted no more children, wanted more children or were undecided were coded 0, 1 and 2, respectively. Whether the couple wanted no more children, both the husband and wife wanted more children, the husband wanted more children, or the husband wanted fewer children were also assessed. The four groups were respectively coded 0, 1, 2 and 3.
Data analysis
Both outcome variables on decision-making regarding sex (refusal of sex and use of condom) are dichotomous. Under the assumption of binary response, there are three potential alternatives: the logit model, probit model and complementary log-log model. Both logit and probit links have the same property, which is link [π(x)]=−link [1−π(x)]. This means that the response curve for π(x) has a symmetric appearance about the point π(x)=0.5 and so π(x) has the same rate for approaching 0 as well as for approaching 1. When the data given are not symmetric in the [0, 1] interval, and increase slowly at small to moderate value but increases sharply near 1, as in the case of the two outcome variables in this study, a complementary log-log link function is appropriate. In this study, ability to refuse sex with husband (‘yes’: 90.06%, ‘no’: 9.94%) and ability to ask husband to use condom (‘yes’: 87.3%, ‘no’: 12.7%) exhibit asymmetry. Therefore, logit and probit models that rely on (50%, 50%) curves are considered inappropriate. The complementary log-log model gives a better representation (Gyimah et al., Reference Gyimah, Tenkorang, Takyi, Adjei and Fosu2010; Tenkorang Reference Tenkorang2012), and was, therefore, used for the analysis. Table 1 shows descriptive and bivariate analysis of the outcome and explanatory variables used in this study.
Table 1. Descriptive and bivariate analysis of ability to refuse husband sex and ask husband to use condoms among married women aged 15–49 in Cambodia, 2010 (N=11,218)
Cross-tabulation of outcome and predictor variables showed statistically significant differences at an α-level of 0.001, except for decision-making on family visit and access to condoms. In this study, ability to refuse sex with husband (90.06%, 9.94%) and ability to ask husband to use condoms (87.3%, 12.7%) exhibit asymmetry.
*p<0.05; **p<0.01; ***p<0.001.
Results
In Table 1, the descriptive statistics for the outcome variables indicate that the majority of married women were able to refuse sexual intercourse with their husbands (90%) and were able to ask their husbands to use condoms (87%). The responses of married women reflect asymmetries in the two outcome variables. Women were distributed evenly across age groups (13–16%), except the 25–29 age group (22%) and the 15–19 age group (3%). Approximately 71% of respondents resided in rural areas. The classification of place of residence into rural and urban in this study needs to be understood in the following context: the National Institute of Statistics (2004) developed a new urban classification using commune-level data rather than district-level data, as was used in previous classifications. Communes were classified as urban if: (a) population density exceeded 200 inhabitants per square kilometre; (b) if the total population of the commune exceeded 2000 inhabitants; and (c) if less than 50% of employed males were employed in agriculture. Close to 53% of married women had attained primary education whereas only about 2% had completed higher education. About one-quarter of married women were in the richest wealth quintile. Joint decision-making of married women with their husband on health care, large household purchases and family visits were 50%, 81% and 74%, respectively. Close to 35% of married women had previously ascertained their HIV status. Approximately 63% of the respondents were aware of modes of HIV transmission whereas 12% believed in HIV transmission myths. Close to 71% of married women reported easy access to condoms, and 26% of the participants refused to answer the question on access to condoms.
Table 1 provides bivariate relationships of outcome variables (‘refuse husband sexual intercourse’ and ‘ask husband to use condoms during sexual intercourse’) with socio-demographic, autonomy and HIV attitude and awareness variables. Compared with those in the 15–19 age group, married women in all other age groups were more likely to report being able to refuse sexual intercourse with their husbands. However, only married women aged 20–24 were more likely to be able to ask their partners to use condoms (OR=1.16, p<0.05). Married women residing in rural areas were less likely to be able to refuse sex (OR=0.78, p<0.001) and less likely to able to ask their partners to use condoms (OR=0.75, p<0.001). Compared with women with no education, those who had attained primary, secondary and higher education were more likely to be able to refuse sex and were more likely to be able to ask their partners to use condoms. Wealthier women were more likely to be able to refuse sex and were more likely to be able to ask their partners to use condoms. Married women who made decisions on health care alone and those who made joint decisions on health care with their husbands were more likely (OR=1.23, p<0.001 and OR=1.29, p<0.001) to be able to refuse sex with their husbands. Those who made decisions on large household purchases alone were 16% less likely to be able to refuse sex with their partners (OR=0.84, p<0.05). Those who made joint decisions on family visits with their husbands were less likely (OR=0.87, p<0.01) to be able to ask their partners to use condoms. Those who were aware of HIV transmission modes were 14% and 16% more likely to refuse sex and ask their partners to use condoms, respectively. Married women who believed in myths about HIV transmission were 14% and 18% less likely to refuse sex and ask their partners to use condoms, respectively. Married women who reported that they had previously ascertained their HIV status were more likely to be able to refuse sex (OR=1.31, p<0.001) and more likely to ask partners to use condoms (OR=1.38, p<0.001). Those who reported easy access to condoms were more likely to be able to ask partners to use condoms (OR=1.28, p<0.001). However, compared with participants who answered ‘no’, those in the ‘refused to answer’ category were less likely to refuse sex (OR=0.72, p<0.001). Additionally, the participants in the ‘refused to answer’ category were also less likely to ask their partners to use condoms (OR=0.73, p<0.001) compared with their counterparts who answered ‘no’. At the bivariate level, women's fertility preference was not a significant predictor of their ability to refuse their husbands sex, although it was a significant predictor of their ability to ask their husbands to use condoms. Surprisingly, women who wanted more children were more likely to be able to ask their husbands to use condoms. Conversely, undecided women were less likely to be able to ask their husbands to use condoms. None of the husband's fertility preferences was a significant predictor of ability of women to refuse sex, except joint desire of husband and wife for more children. Desire to have at least one or more children by either husband or wife, or jointly, was associated with higher odds of ability to ask husbands to use condoms.
Tables 2 and 3 provide the multivariate relationships between the predictor and outcome variables. Model 1 includes control variables (5-year age groups, type of residence, education, wealth index). Model 2 includes decision-making variables (decision-making on health care, large household purchases and family visiting) and Model 3 includes HIV and fertility preferences (HIV transmission awareness, HIV transmission myths, ever had HIV test, access to condoms, women's fertility preferences and husband's fertility preferences).
Table 2. Complementary log-log models of ability to refuse husband sex among married women aged 15–49 in Cambodia, 2010
*p<0.05; **p<0.01; ***p<0.001.
Table 3. Complementary log-log models of ability to ask husband to use condoms among married women aged 15–49 in Cambodia, 2010
*p<0.05; **p<0.01; ***p<0.001.
In Model 1, after controlling for socio-demographic and economic variables, married women in all age groups were more likely to be able to refuse sex (Table 2); however, only those in the 20–24, 25–29 and 30–34 age groups were more likely to be able to ask their partners to use condoms compared with the 15–19 reference age group (Table 3). Those who had attained primary, secondary or higher education were more likely to be able to both refuse sex and ask their partner to use condoms, compared with those who had no education. Compared with the poorest quintile, married women who were in the poorer, middle, richer or the richest wealth quintile were all more likely to be able to refuse sexual intercourse and ask their partner to use condoms.
In Model 2, after introducing autonomy variables (namely, decision-making on health care, large household purchases and family visits), age, education and wealth index remained significant predictors of married women's ability to negotiate safer sexual intercourse. Compared with women whose health care decisions were made by others, those who made such decisions by themselves (OR=1.22, p<0.001) or jointly with their husbands (OR=1.26, p<0.001) were more likely to be able to refuse sexual intercourse with their partners. Those who made decisions on large household purchases by themselves were less likely (OR=0.78, p<0.01) to be able to refuse sex compared with married women who were not involved in such decisions (Table 2). Similarly, those who made joint decisions on family visits with husbands were less likely (OR=0.86, p<0.05) to be able to ask their partners to use condoms compared with those who were not involved in this type of decision-making (Table 3).
In Model 3 of Tables 2 and 3, behavioural and HIV awareness, and fertility preference variables were introduced. In Table 2, after controlling for these predictors, age and wealth variables remained significant predictors of married women's ability to refuse sex, whereas in Table 3, education and wealth variables remained significant predictors of married women's ability to ask their partner to use condoms. Compared with those who were not involved in decision-making on health care, those who made decisions on their own and those who made joint decisions with their husbands were 16% and 24%, respectively, more likely to be able to refuse sexual intercourse (Table 2). Women who made decisions on large household purchases on their own were less likely (OR=0.77, p<0.01) to refuse sexual intercourse compared with those who were not involved in decision-making. Participants who made joint decisions on family visits were less likely (OR=0.93, p<0.05) to refuse sexual intercourse compared with those who were not involved (Table 2). Also, those who made joint decisions on family visits with husbands were less likely (OR=0.87, p<0.05) to be able to ask their partners to use condoms (Table 3).
Relative to participants who had no awareness on HIV transmission, those who were aware of HIV transmission were more likely (OR=1.08, p<0.001) to be able to refuse sexual intercourse (Table 2) and were also more likely (OR=1.07, p<0.001) to be able to ask their partner to use condoms (Table 3). Those who believed in HIV transmission myths were less likely to be able to refuse sexual intercourse and less likely to be able to ask their partner to use condoms (OR=0.94, p<0.001 and OR=0.91, p<0.001), respectively. Compared with those who had not previously ascertained their HIV status, those who had discovered their status were more likely to be able to refuse sexual intercourse (OR=1.10, p<0.01) and more likely to be able to ask their partners to use condoms (OR=1.10, p<0.01) (Tables 2 and 3, respectively). Those who had easy access to condoms were more likely (OR=1.22, p<0.01) than those with no access to ask their partners to use condoms (Table 3). A two-way connection between negotiating safer sex and women's autonomy was accounted for. A two-way connection also existed between refusal of sex and decision-making on own health care, and between condom use negotiation and decision-making on family visits. Participants in the ‘refused to answer’ category were less likely (OR=0.85, p<0.05) to be able to refuse sexual intercourse and were also less likely (OR=0.89, p<0.05) to be able to ask their partners to use condoms relative to those who had no access to condoms. It was unclear as to the reasons for the behaviour of this category (i.e. ‘refuse to answer’) in relation to their ability to refuse sex and ability to ask their partners to use condoms. The characteristics of this group were therefore further probed using cross-tabulation. It emerged that there was a consistent increase of refusal to answer the question on ‘access to condoms’ with increasing age. Additionally, there was a consistent decrease in the proportion of ‘refusal to answer’ the question with increasing wealth. Also, more than half of the respondents who refused to answer the question on ‘access to condoms’ had primary education. This suggests that older, less educated and poorer women were less likely to belong to the ‘refuse to answer’ category on the question of access to condoms.
Women's fertility preference was not a significant predictor of their ability to refuse sex. Husband's fertility preference was not a significant predictor of women's ability to refuse sex, except when husbands and wives simultaneously wanted more children, as well as when the husband alone wanted more children. Therefore, wives who jointly wanted more children with their husbands were more likely to refuse sex compared with their counterparts who did not want children (OR=1.05, p<0.01). However, women whose husbands alone wanted more children were less likely (OR=0.90, p<0.05) to refuse sexual intercourse.
Also, married women's fertility preference was not a significant predictor of women's ability to ask husbands to use condoms, except when married women were undecided about their fertility preference. Within this context, undecided women were less likely (OR=0.79, p<0.01) to be able to ask their partners to use condoms compared with their counterparts who did not want children. Similarly, husband's fertility preference was not a significant predictor of married women's ability to ask husbands to use condoms, except when both wanted more children. In that case, wives who jointly wanted more children with their husbands were more likely to be able to ask their partners to use condoms (OR=1.21, p<0.001).
Discussion
Existing literature has shown a decline in HIV prevalence within the Cambodian general population but an increasing prevalence among vulnerable and high-risk social groups (Sopheab et al., Reference Sopheab, Saphonn, Chhea and Fylkesnes2009). One of these groups includes married women, which may seem counterintuitive since initial evidence seemed to suggest that marriage acts as a shield to HIV infection (Tenkorang, Reference Tenkorang2012). Nonetheless, a plethora of recent literature has indicated otherwise (Clark, Reference Clark2004; Cohen Reference Cohen2004; Clark et al., Reference Clark, Bruce and Dude2006; Dunkle et al., Reference Dunkle, Stephenson, Karita, Chomba, Kayitenkore and Vwalika2008; Tenkorang, Reference Tenkorang2012). This study sought to add to this literature by examining factors that enable negotiation of safer sex in marriage, with a central focus on women's autonomy. The results provide potentially new directions for policy interventions with the goal of ensuring a decreasing rate of HIV transmission among married women in Cambodia.
The results here revealed that married women aged 20 and above were more likely to be able to refuse sexual intercourse with their husbands and ask for use of condoms compared with their counterparts aged 15–19. However, likelihood of ability to refuse sex generally declined with increasing age. This makes older married women more prone to HIV infection if their husbands are infected. In this context, involvement in decision-making may have translated into trust and risk compensation. This result supports the findings of Bracher et al. (Reference Bracher, Santow and Watkins2003), that women who marry early may be shielded against HIV risks. It is, however, inconsistent with the findings of Clark (Reference Clark2004), Clark et al. (Reference Clark, Bruce and Dude2006), Dunkle et al. (Reference Dunkle, Stephenson, Karita, Chomba, Kayitenkore and Vwalika2008) and Tenkorang (Reference Tenkorang2012), which suggest that younger married women are less likely to be able to negotiate safer sex with their husbands.
The proportion of highly educated women was relatively higher for individuals below 40 years. In this context, it appears that attainment of higher education makes women more autonomous. This signals the possible contributory effects of female autonomy in the negotiation for safer sex. While some studies have suggested that education and wealth serve as risk factors for HIV and AIDS within the developing world (Hargreaves & Glynn, Reference Hargreaves and Glynn2002; Abebe et al., Reference Abebe, Schaap, Mamo, Negussie, Darimo, Wolday and Sanders2003; Cogneau & Grimm, Reference Cogneau and Grimm2006), our findings showed that educated and wealthier married women were more likely to negotiate safe sexual intercourse than uneducated and poorer women in Cambodia. This is consistent with the findings of Shannon et al. (Reference Shannon, Kerr, Allinott, Chettiar, Shoveller and Tyndall2008) and Tenkorang (Reference Tenkorang2012) on the argument that economic and social empowerment enhances women's ability to negotiate safer sexual relations. It can, therefore, be argued that educated and wealthier women often have autonomy and are less dependent on their husbands; hence, they have the agency to say no to unsafe sexual intercourse. Further, there is no fear of future consequences of the women's actions since their well-being does not solely depend on their husbands' work. Invariably, while the effect of education and wealth may be beneficial to married women in urban areas, this is not the case for their counterparts in rural Cambodia. Married women in the rural areas of Cambodia are less likely to be able to say no to sexual intercourse when their husbands ask for it and less likely to ask their husbands to use condoms during sexual intercourse. This correlates with the poor economic background of such women, the predominance of traditional norms, which lay emphasis on high fertility in marriage (Al-Riyami et al., Reference Al Riyami, Afifi and Mabry2004), and their lack of any safety net or autonomy. The attenuation of the effect of education on the refusal of sex when controlling for knowledge of HIV and modes of transmission points to the importance of self-efficacy in explaining safer sex negotiation. It is worth noting that a small proportion of those who ascribe to HIV transmission myths still remain. Interventions will need to be targeted at this group in order to stem the potential spread of HIV as these groups remain less likely to initiate preventive measures against HIV transmission.
A female's ability to control and make decisions without the influence of any man has been used to explain the condition of a woman's self-determination and autonomy (Abadian, Reference Abadian1996). This is reiterated in the work of Kabeer (Reference Kabeer2005) as the ability of women to have agency and participate in the decision-making process. In examining the variables that were used to investigate women's involvement in decision-making, those who made decisions on their own health care were more likely to refuse sexual intercourse than those who were not involved, yet they were less likely to be able to ask their husbands to use condoms relative to those who were not involved in the decision-making process. Subsequently, women who had the autonomy to make large household purchases and who were involved in decision-making on family visits were less likely to be able to refuse their husbands sexual intercourse and enforce the use of condoms. This finding is inconsistent with the suggestion by Amaro & Raj (Reference Amaro and Raj2000), that power dynamics and sex roles influence women's ability to negotiate safer sex. Thus, in this paper, even though they have the power, they are still unable to enforce the use of condoms by their husbands. This then brings to the fore the argument that, while autonomy may reflect a woman's power, power may not always reflect a woman's autonomy.
In conclusion, a married woman's wealth, education and knowledge of the modes of HIV transmission and serostatus then become central to her autonomy and ability to negotiate safer sexual relations. However, this study is not without its limitations. First, it is noteworthy that women's autonomy is used in a limited sense in this work, although it has much broader connotations beyond socioeconomic and household autonomy. Secondly, negotiation for safer sexual intercourse may be represented by several surrogate variables, of which only two were used in this study. These limitations notwithstanding, the findings of this study have several policy implications.
Policy implications
This paper examines how autonomy influences the capacity of married women to negotiate safer sex in Cambodia. Fundamentally, inequities in education, wealth and autonomy between married women younger than 35 years and their older counterparts exist. Also, palpable differences in women's ability to refuse their husband sex and to ask their husband to use condoms exist between rural and urban dwellers. This underscores the importance of place in appreciating safer sex negotiation. A nuanced understanding of the potential mechanisms that influence safer sex negotiation and the relative contribution of the economic, social and educational dimensions of HIV intervention are of central policy relevance in Cambodia and elsewhere. Broadly, addressing the gender dimensions of the HIV/AIDS epidemic – and the implications for policymakers and practitioners – requires a far deeper understanding of how to support families and communities as they mediate the epidemic's repercussions for household restructuring, gender and intergenerational relations, reproductive decision-making, livelihood choices, education planning, economic status and civic participation. Equally urgent is the need to develop the knowledge necessary to strengthen national response capacities so that those most affected by HIV and AIDS do not also have to shoulder its associated burdens.
Given that wealth and education inequities exist among married women in Cambodian society, structural interventions may be required to elicit better health policy outcomes. Specifically, structural interventions that focus on three areas of change are imperative. First, social change: these approaches should focus on factors affecting multiple groups, such as legal reform, stigma reduction and efforts to cultivate strong leadership on AIDS (Adimora & Auerbach, Reference Adimora and Auerbach2010). Second, change within specific groups: these approaches should address social structures that create vulnerability among specific populations, such as forced-marriage victims, young women or poor women (Auerbach, Reference Auerbach2009). Examples include efforts to organize micro-finance programmes for poor women, and interventions to change harmful male norms. Finally, harm reduction or health-seeking behaviour change: these approaches should aim to make harm-reduction technologies available to those in need, and to change rules, services and attitudes about these technologies (Auerbach et al., Reference Auerbach, Cáceres and Parkhurst2011). A good example is the ‘100% Condom Use Programme’ (CUP) campaign. The 100% CUP has been nationally implemented in Cambodia since 2001, after a successful pilot in 1998 in Phres Sihanok province (DHAPP, 2010; National AIDS Authority, 2010). The 100% CUP's main principle is to encourage ‘No Condom–No Sex’ in all types of sexual intercourse (Rojanapithayakorn, Reference Rojanapithayakorn2006).
