"That is why I stopped the ART": Patients' & providers' perspectives on barriers to and enablers of HIV treatment adherence in a South African workplace programme

The number of HIV-infected individuals receiving ART in Sub-Saharan Africa increased from 100,000 (an estimated 2% of those requiring it) in 2003 to 1.3 million (24–33% of those requiring it) in 2006 [1]. This has given people living with HIV/AIDS greater hope as treatment has led to improved survival and decreased morbidity [2, 3].

Though the definition of the "sufficient" level of adherence is still evolving, close adherence to ART remains important to achieving viral suppression, reduced risk of disease progression and better survival [4‐6]. Early reports from pilot ART programmes in African settings show high levels of adherence to treatment [7]. However, as ART delivery programmes grow beyond pilot phases, a focus on maintaining these levels is essential. Given that adherence is the strongest predictor of treatment success, a clearer understanding of adherence barriers in low-income countries providing ART is needed [8‐11]. Due to the relatively limited experience of ART use in low-income countries, information about barriers to adherence is limited. Even more limited are qualitative studies looking not only at who is having difficulty in adhering but also at why. A recent systematic review of adherence reported on only two such studies [12].

Within the existing body of literature from low-income countries a few factors have more frequently been reported to be associated with poor adherence. These include structural barriers such as high cost and unavailability of drugs [13‐15], and individual factors such as lack of self efficacy for medication taking [16, 17], lack of social or family support [16, 18], high anxiety, perception that ART has a negative effect [16, 17], and fear of stigma and discrimination through disclosure of HIV status [19, 20]. However, lower level of education has not consistently been reported to be associated with poor adherence [17‐19, 21].

This study was conducted in a large workplace ART programme providing treatment in South Africa since 2002. All services and medications were provided free of charge to employees with HIV infection. Patients received routine care based on a standardized protocol. All patients underwent intensive adherence counselling aimed at preparing them to start taking ART. Counselling was also offered and provided at all subsequent programme visits. This study mirrored many public sector ART programmes in that it offered ART to all who fulfil medical criteria without other pre-requisites. The purpose of this study was to qualitatively explore patient and health provider perspectives of barriers to, and enablers of, adherence in a workplace setting.

The interviews were conducted from June to July 2005 with 12 participants: six ART patients, five health service providers (including one traditional healer), and one company human resources manager. Four of the health care providers were female. The six patients were all male and they had been on ART for more than 8 weeks. Four were classified as poor adherers.

The effect of traditional medicine use on adherence has not been widely reported. In this setting, traditional medicine use appeared to affect adherence negatively. Patients reported that both clinic providers and traditional healers advised them not to "mix" ART with traditional medicines, leading to ART interruptions. Given that traditional medicine use is common in this setting [23], a deeper understanding and clarification to providers of how to counsel patients on the use of traditional medicines while on ART is critical.

A key and, to our knowledge, novel finding of our study is that doubt about the existence of HIV disease, own HIV-infected status, or both were reported to inhibit adherence. In contrast, an earlier study in an urban setting in South Africa indicated that a high proportion of participants had correct knowledge of HIV [12]. This difference in findings perhaps reflects that the earlier study was conducted in an urban setting where traditional beliefs concerning disease causation may be less common.

Results from a pilot quantitative study conducted among 69 patients in this clinic subsequently confirmed that lack of belief in HIV disease or in own status was prevalent. The study showed that 19% patients were uncertain or did not believe that HIV existed, while 24% were uncertain or did not believe that they were HIV-infected [24]. This is despite patients receiving at least three intensive counselling sessions on HIV and ART prior to starting treatment. To what degree this factor truly affects adherence to ART is not yet clear and is being further investigated. Meanwhile, there is a need to find effective ways to support adherence to ART even if the individual does not accept biomedical concepts of HIV disease.

No consistent association has been found in low income countries between level of education and poor adherence [8, 25‐27]. However, in this setting providers reported that lower levels of education were strongly related to poor adherence, a view not shared by patients interviewed. In a multilingual setting it may be possible that having less education is related to speaking a different language from that of providers, a factor which has previously been found to be related to poor adherence in South Africa [28]. The extent to which education affects adherence in this population, if at all, is not yet clear. However, providers' perceptions that illiterate patients have a poorer ability to adhere may have important implications for how they interact with and provide counselling to patients with less education. This may also be important in other ethnically and linguistically diverse African countries where communication barriers between patients and providers may negatively affect adherence.

Few studies in the existing literature have examined the relationship between adherence and health system factors. Among those that did, cost was found to be the foremost reason for poor adherence [13, 14]. This was not a barrier in this setting as a stable and free supply of treatment was available to all patients. However, long waiting times for clinic services were reported as an important barrier to a patient's ability to return to the clinic for service and to pick up medication. This might be especially stigmatizing in this working population as spending additional time in the clinic may require additional time off work. Further research is warranted into health system factors potentially affecting adherence.

Finally, consistent with previously published research in low-income settings, drug side-effects, alcohol use, being away from usual place of residence and fear of stigma were thought to contribute to poor adherence in this setting [19, 20]. In contrast, having social support, a strong belief in the value of treatment and the ability to fit ART into daily routines were reported as facilitators of adherence [16, 18]. Interestingly, most patients, even those who were not completely adherent, reported having some social support and/or belief in the value of treatment. However, these alone did not seem to ensure complete adherence. This suggests that in this setting careful consideration of multifaceted interventions to promoting adherence is important.

In our study we purposively sampled a relatively small number of individuals who were not intended to be representative of the clinic population, and our results may not be generalisable. The study was designed to identify potential barriers to adherence relevant to this setting, using in-depth interviews and taking various perspectives into account. Specifically, by complementing patient information with provider information we were able to describe a more complete picture of adherence issues. There was great congruency between factors mentioned by the patients and by the providers, to the extent that after interviewing 12 participants, we felt we had reached saturation of findings. We are currently conducting a much larger quantitative study that investigates the prevalence of these barriers in the population, and their association with poor adherence. However further work is needed to establish whether these factors are also associated with poor adherence in other populations The study setting, among mostly male participants involved in a workplace ART programme, may also limit generalisability. However issues related to uncertainty about western biomedical concepts may arise in other populations, particularly those where traditional, non-western concepts concerning illness are prevalent.

Given the above limitations, our data highlight novel barriers to adherence that merit further study in populations where traditional beliefs around disease causation are common. Our data also illustrate the value of formative qualitative research in identifying culturally-relevant barriers to adherence, particularly given the paucity of such studies in low-income countries [12].

Uncertainty about the existence of HIV disease and use of traditional medicines while on ART was reported despite a programme emphasising ART counselling. Thus there is a need to find effective ways to support adherence to ART even if the individual does not accept biomedical concepts of HIV disease or decides to use traditional medicines. Additionally, providers should identify ways to minimize barriers in communication with patients who speak different languages. Finally, dissatisfaction with clinical services, namely due to long waiting times, should be addressed.