Abstract
Cambodia has one of the highest prevalence rates of HIV in Asia and is scaling up HIV testing. We conducted a cross-sectional survey with 358 health care providers in Phnom Penh, Cambodia to assess readiness for voluntary testing and counselling for HIV. We measured HIV knowledge and attitudes, and predictors of intentions to take a sexual history using the Theory of Planned Behaviour. Over 90% of health care providers correctly answered knowledge questions about HIV transmission, but their attitudes were often not positive towards people living with HIV. The Theory of Planned Behaviour constructs explained 56% of the variance in intention to take a sexual history: the control providers perceive they have over taking a sexual history was the strongest contributor (51%), while social pressure explained a further 3%. Attitudes about taking a sexual history did not contribute to intention. Interventions with Cambodian health care providers should focus on improving skills in sexual history-taking.
The prevalence of HIV in Cambodia is among the highest in Asia; it approached 2% in the late 1990s and has now levelled at just under 1%. 1 Sexual transmission of HIV through unprotected heterosexual intercourse remains the most common route for acquiring the virus in Cambodia 2 and the Cambodian government is scaling up voluntary counselling and testing (VCT) services for HIV. 3 VCT requires health-care providers (HCPs) to have knowledge about HIV transmission, to have positive attitudes about HIV, and to be willing to provide sexual risk counselling, including questioning about sexual behaviour. 4 VCT programmes in developing countries have been shown to significantly reduce the rates of unprotected sex. 5 The purpose of this study was to assess HIV knowledge and attitudes, and intentions to take a sexual history among Cambodian HCPs who work with migrant garment factory workers. Constructs of the Theory of Planned Behaviour (TPB) were used to predict intentions to take a sexual history.
LITERATURE REVIEW
The core premise of the TPB is that behavioural intention is explained by three constructs: the individual's attitudes towards the behaviour, how much social pressure is felt to perform the behaviour (‘subjective norms’) and the control the individual perceives they have over the behaviour (‘perceived behavioural control’). There have been several extensive reviews of TPB research. In an early review of 56 TPB studies of health behaviours, Godin and Kok 6 determined that the theory was effective at predicting behavioural intention with an average R 2 of 0.41. In their review of 185 studies, Armitage and Connor 7 found that the TPB constructs accounted for 39% of the variance in behavioural intention. Two reviews of HCP intentions and behaviour led by Perkins et al. 8 and Godin et al. 9 confirmed that TPB is useful in predicting HCP behaviour, although the efficacy of prediction varies with the behaviour and type of HCP studied. The TPB has also proven useful in HIV prevention research in predicting both the intention and the actual use of condoms by clients in diverse settings; 10 however, we found no studies of HCPs' intentions or behaviour related to HIV prevention using the TPB.
METHODS
In an earlier qualitative phase of this study, we found that migrant garment factory workers obtained reproductive health care from a variety of private and public HCPs (including local pharmacists), close to their place of employment. The purpose of the present survey was to assess these HCPs' knowledge and attitudes about HIV, and factors influencing their intentions to take a sexual history from a migrant garment factory worker presenting with a vaginal discharge. The clinical scenario used to assess intentions was a client presenting with a vaginal discharge (as this is a common complaint of migrant workers), and when women present with this symptom, HCPs should assess sexual risks in order to decide on appropriate treatment. In addition, sexual history-taking is a key component of the World Health Organization's guidelines on HIV testing. 4
The survey of HCPs covered demographics, TPB constructs on intention to take a sexual history, and their knowledge and attitudes about HIV. The series of questions related to intention to take a sexual history were developed based on the TPB Manual developed by Francis et al. 11 and consultation with an experienced TPB researcher. Only direct measures of intention to take a sexual history were used. The HIV knowledge and attitudes components of the survey drew questions modified from two surveys of Chinese HCPs. 12,13 The English version of the survey was translated into Khmer, back-translated into English and pilot tested to assess clarity of the items.
We obtained ethics approval from the Cambodian National Ethics Committee and the University of Ottawa Health and Social Sciences Research Ethics Board. We trained 10 Cambodian research assistants to administer the survey over a three-week period in November and December 2006. The research assistants followed standardized distribution and administration procedures of the survey. Data were collected in the four regions of Phnom Penh where the garment factories are concentrated. Inclusion criteria included all government health centres and non-government organization (NGO) clinics in these four regions of the city, plus two government hospitals in Phnom Penh where garment factory workers are known to go, and the private clinics and pharmacies that are within a 1-km radius of the garment factories. All HCPs working on the day of the survey and who were available to participate were recruited. Eligible HCPs included physicians, midwives, nurses and pharmacists.
After data input by a senior research assistant, 20% of the data were re-entered by the first author. The initial data entry error rate was calculated to be less than 0.6%.
RESULTS
Administrators from all of the government health centres, NGO clinics and the two hospitals gave permission for their respective institutions to participate (100% organizational participation). Over 75% of the private clinics and pharmacies participated. Within these settings, the total number of HCPs eligible to complete the survey was not known, but a total of 358 surveys were completed.
The demographic characteristics of the participants are reported in Table 1. As almost 25% of HCPs reported more than one profession, their ‘primary profession’ was recorded as that which required the most training. Bivariate analyses using chi-square testing (or Fisher's exact test, when appropriate) assessed differences in HIV knowledge and attitudes between the sexes and among the primary professions, presented in Table 2. Generally, the HCPs showed good knowledge of HIV transmission, with over 90% of the respondents answering most questions about modes of transmission correctly. Knowledge of perinatal risks of HIV transmission through the placenta and by breastfeeding was generally not as high, and attitudes towards people with HIV were not consistently supportive. Overall, the HCPs agreed that HIV testing should be available and most were in favour of mandatory HIV testing of pregnant women.
Demographic characteristics of Cambodian HCPs surveyed (n = 358)
HCP = Health-care providers; NGO = Non-government organization
*Percentage exceeds 100% as several HCPs reported more than one location of employment
HIV knowledge and attitudes of Cambodian HCPs: bivariate analysis by gender and primary profession
HCP = health-care providers
*Note that n for individual items may be less than total n for the group due to missing data
†Using chi-square or Fisher's exact test, as appropriate
‡For the HIV knowledge questions, the percentage of correct answers was calculated by dividing total correct answers by total responses, counting a ‘don't know’ response as incorrect. Missing data were excluded. For the HIV attitudes questions, the percentage of correct answers was calculated using only those responses portraying positive attitudes. ‘Don't know’ responses were recoded and excluded as they indicated neither positive nor negative attitudes
All TPB constructs (generalized intention, attitude, subjective norms and perceived behavioural control) had a Cronbach's alpha value of 0.6 or higher, which is considered acceptable for new scales. 14 Imputation was used with one item when there were more than 25% missing values. Pearson's correlation coefficients between the four TPB constructs ranged from 0.39 to 0.71.
Each of the four TPB constructs was standardized and a hierarchical multiple linear regression model was fit using generalized intention as the dependent variable and attitudes, perceived behavioural control and subjective norms (in that order) as predictors, using the procedure described by Amireault et al. 15 Two-way interaction terms were tested for gender with each TPB construct. Dummy variables were created for primary profession (physician was the reference group) to assess the two-way interaction between profession and each construct. Due to singularity (all midwives were women), three-way interactions could not be tested.
The overall R 2 for the multiple linear regression was 0.56. Only two TPB constructs made significant contributions to the final regression model, after testing for the interaction terms. Perceived behavioural control (PBC) explained 51% of the variance (P < 0.001) while 3% was explained by subjective norms (P < 0.001) and 1% by the profession × PBC interaction (P = 0.031). Profession alone was insignificant (P = 0.066). The only significant profession × PBC interaction was pharmacist × PBC (P = 0.003), suggesting that perceived behavioural control is a stronger predictor of intention to take a sexual history for pharmacists than for other HCPs.
DISCUSSION
Most of the Cambodian HCPs had accurate knowledge of HIV transmission, comparable to that of Chinese providers. 12,13 Like Chinese HCPs, these Cambodian HCPs were not supportive of people living with HIV marrying or having children, and they generally approved of compulsory testing for HIV for pregnant women. 13 Such attitudes to people living with HIV and compulsory HIV testing have implications for the success of a VCT programme. The World Health Organization has opposed compulsory HIV testing because it infringes on a person's human rights. 4 Hence, there is a need for more education, and normalization of HIV testing, in order to reduce stigma and promote the voluntary testing of pregnant women and others at risk.
This study demonstrated that the TPB constructs were useful in explaining intention with an R 2 of 0.56. This is higher than the average variance explained by the TPB constructs in the health behaviour literature, 6,8 and similar to the variance for TPB studies of HCP intention. 9 As found in these reviews, perceived behavioural control was the most important predictor of intention in this study, while subjective norms played a minor role. However, in our analysis, the attitude construct was not a significant contributor to intention when the interaction of profession with PBC was included in the regression model. The interaction term indicated that for pharmacists, belief in their control over taking a sexual history was a stronger predictor of intentions to take a sexual history than for the other HCPs. This is not surprising, for pharmacists often worked independently, whereas the physicians, nurses and midwives typically worked in a clinic or hospital setting under supervision.
Several authors have been critical of the ‘intention-behaviour gap’ of the TPB. 7,8,10 They maintain that the TPB does not determine the difference between those individuals who state their intention to act but then fail to perform the behaviour (‘non-performers’) and those who do complete the behaviour (‘performers’). Factors such as habits and automatic processes, behavioural skills and cues, and environmental obstacles may interrupt the translation of intentions into actual behaviour. 8 These barriers may certainly be an issue for Cambodian HCPs providing VCT. Environmental obstacles include lack of private space to take a sexual history, while the gender norms of Cambodian society may also form a barrier to sexual history-taking for Cambodian HCPs. The social pressure against premarital sex for women in Cambodia makes sexual history-taking of young migrant garment factory workers a challenge for HCPs, as single women are expected to be naïve about sexual matters, 16 and thus may be unwilling to discuss these issues with HCPs. The cultural barriers for male HCPs holding such a discussion with young women are particularly high as young women would not generally discuss such intimate matters with men. Future TPB research could incorporate a new construct into the model to assess the impact of societal gender norms. Assessing these beliefs for Cambodian HCPs would involve questioning their underlying gender stereotypes (particularly with respect to the sexual activity of young women) and their understanding of their own role as an HCP in asking about sexual activity, including with patients of the opposite sex.
The relatively strong contribution of perceived behavioural controls in this regression model further suggests that future training for HCPs should address how to improve their skills in sexual history-taking. Interventions for Cambodian HCPs need to focus on making them feel more capable and confident when asking about sexual histories, in order to increase their perception of behavioural control, and hence their intention to take a sexual history.
Study limitations
A major limitation of this study is self-report bias: HCPs may have overestimated their intentions to take a sexual history. There is evidence that HCPs overestimate their adherence to guidelines by as much as 27%. 17 Such over-reporting may have been uneven between professional groups, thus biasing the results. A second limitation results from sampling: we sampled available eligible HCPs on day shifts during the data collection period.
Finally, we chose not to include indirect measures of the TPB constructs in the questionnaire (i.e. asking about beliefs and then combining this score with the individual's directly measured items to determine whether the attitude was favourable or unfavourable). The indirect measures were excluded due to the complexity of translating complex concepts. While we believe this pragmatic consideration was appropriate, inclusion of indirect measures, if found to be significant, might have altered the findings. 11
CONCLUSIONS
Preventing HIV among female Cambodian garment factory workers, and indeed in the general population, will require several different strategies. HCPs need further education about the risks of perinatal transmission and interventions to address some negative attitudes towards people with HIV. Furthermore, education about the problems of mandatory testing is needed for Cambodian HCPs, because voluntary testing and counselling, an important component of HIV prevention, requires positive attitudes about people with HIV and about voluntary testing. 4 Interventions to improve skills in sexual history-taking are needed, and programmes to target pharmacists may be particularly important as most lack formal training, yet pharmacies are where many migrant garment factory workers access reproductive healthcare. The TPB is a useful framework for studying intention to take a sexual history in this population. Further research should focus on barriers to sexual history-taking for Cambodian HCPs, including the gendered implications of this behaviour.
Cambodia has been very successful in slowing the HIV epidemic, in part through promotion of condom use by sex workers and their clients. As the government continues to scale up VCT programmes and to further expand HIV prevention activities, there is a need to also focus on Cambodian HCPs. These providers play an important role in HIV prevention for migrant garment factory workers as well as other Cambodian citizens.
Footnotes
ACKNOWLEDGEMENTS
We acknowledge the support and assistance of the following organizations in this research: International Organization for Migration Cambodia, CARE Cambodia, the Reproductive Health Association of Cambodia and Marie Stopes Cambodia. We especially thank the health-care providers who participated in the research and the Cambodian research assistants who administered the survey. We also thank Lynne Leonard for conceptual guidance, and Jacqueline Tetroe for assistance with survey development. Financial support for the study was provided for Gail Webber through the Ontario Women's Health Council Scholar's Doctoral Award, and from the Department of Family Medicine, University of Ottawa. Nancy Edwards holds a Nursing Chair funded by the Canadian Health Services Research Foundation, the Canadian Institutes of Health Research and the Government of Ontario.