A comprehensive study of ARV adherence in Tanzania has not yet been attempted. This study has been conducted in order to identify factors which constrain or facilitate adherence to ART and to suggest possible ways to improve adherence. Reports from other countries have emphasized that sub-optimal adherence is the main cause for the failure of ARV therapy. The unforgiving nature of HIV requires that levels of adherence be higher (>95%) and more sustained than in most other areas of medicine (Garcia et al., 2003; (Paterson, 2000).

The present study examined a range of factors that can have a negative impact on adherence, including: stigma, poor social support, mistaken beliefs, lack of food, side-effects, inadequate counselling, long waiting times at treatment facilities, transport-related and other costs, and long distances to the health facility. Factors which can help facilitate adherence which were investigated included: adherence counselling, disclosure of HIV status to family members, social support, a reliable medicine supply chain, information and education.

Adherence measurement

Three different adherence measurements were used in this study. Two-day recall was found not to be discriminatory. Use of tablet counts and 28-day recall using a visual device (beads) were combined to produce a composite adherence measure. However, there were substantial differences between the results of these two measures. As there is no gold standard for adherence measures, further studies are needed to validate these different measures by comparing the different results with viral load counts, as has recently been undertaken by Carrieri et al. in Malawi.

The results of this survey indicate a composite one-month average adherence rate of 90% for ARV users at the seven facilities involved. This meant that 90% of all the pills that should have been taken over the previous month by those interviewed were taken. However, only 21% of ARV users interviewed reported achieving the optimal level of adherence (over 95%) as measured by the composite adherence rate. The remaining 79% self-reported to achieve only moderate adherence (85%-95% adherence rate) and are therefore at risk of treatment failure and the development of drug-resistant forms of the virus.

There was little difference in the composite adherence rate among ARV users interviewed in Arusha and Dar es Salaam, and there was no difference in adherence rates between males and females. In addition, there was no significant difference in adherence rates among ARV users with different levels of education. These findings were similar to the observation by DiMatteo in 2004 that adherence was generally unrelated to variables such as gender, education or socioeconomic status.

In a study conducted in Cameroon, Akam et al. (2004) found an overall adherence rate of 68%. In that study, the constraining factors were found to be the cost of medication and some side-effects. Elsewhere, a study conducted in Botswana by Weiser et al. (2003), reported that between 54% and 56% of patients were achieving the optimal adherence rate of at least 95%. In the Weiser study, financial constraints were identified as one of the principal barriers to adherence. However, it was estimated that if cost was eliminated as a barrier, then the adherence rate would increase to 74%.

Our study would appear to confirm Weiser's findings, which suggest that at population level there are many patients who do not achieve optimal (95%) adherence and who are therefore at risk of treatment failure and the development of drug-resistant virus. We remain interested in investigating further the differences in our various measurements of adherence - ideally comparing these with viral loads, as was recently undertaken in Malawi (Carrieri et al., 2001).

Structural issues

Administering an effective ARV programme is a daunting task. In the present study, there was no uniformity among the treatment facilities in terms of follow-up monitoring and checking of drugs. Some facilities, especially the private ones, did not have procedures for asking patients to come with their medicines for pill count monitoring and to check on patients' adherence to medication. Some health care facilities did not have proper medicine storage facilities, which may compromise the quality of the medicines.

Some of the health care facilities studied did not ensure adequate confidentiality for patients. The worst example of this was a situation in which three doctors were sharing one consultation room and consulting with three different patients at the same time. This can inhibit some patients from attending consultations or from communicating openly, and so impair adherence.

In this study, a major issue for patients was the lengthy waiting times. ARV users spent an average of five hours waiting and being attended to at a facility. Lengthy waiting times could have a negative impact on both attendance at clinics and adherence. Patients complained that staff working at ART clinics were overworked, and in some health care facilities they started the clinic late. Some patients had to travel long distances, and then had to wait for a long time at the facility, sometimes for 10 hours. In addition to the impact of this on adherence, the long waiting times were said to strain the relationship between staff and patients. Food was also a related issue since patients had to buy food while waiting for treatment, thereby incurring additional costs.

In administering ARVs, efforts should be made to minimize waiting times. In many settings it may be possible to provide patients with appointment times, so they do not have to wait all day. Another improvement that could be considered is an increase in the provision of laboratory services at the treatment facilities. At present, many ARV users continue taking medicines without regular checks to monitor their CD4 counts and liver and kidney function. The availability of laboratory services could also help motivate adherence since patients would know through the laboratory results that they were improving. This is an area that needs to be strengthened.

Counselling

Although counselling is a key requirement for successful adherence to ART, the importance of the need for regular ongoing counselling is not always recognized. This was confirmed in our study by the fact that only 21% of patients saw a counsellor on the day of their exit interview. Patients are counselled intensively prior to treatment and at the time they start treatment. However, once on treatment there is very little counselling unless they have a particular problem. Yet it is well recognized by Horne and others that adherence rates decline over time (Horne et al., 2001). Patients need to be counselled whenever they come for refills and if they are achieving at least 95% adherence, they should be congratulated and encouraged. If they are not, the counsellor should be able to spend time with patients and suggest adherence strategies. If a patient forgets, possible solutions are the use of alarm clocks or reminders from a treatment "buddy." ARV users also mentioned using calls to prayer at the mosque as well as reminders from family members. If the patient has not disclosed that they are on treatment, the counsellor could suggest approaches to disclosing or finding a treatment supporter.

Costs

Poverty is a serious problem in Tanzania, which can have a negative impact on adherence. Both key informants and ARV users said that many patients complained that these medicines increase the appetite and cause hunger. Since lack of food has been described as a problem for most people on ARVs, food shortages among the general population in Tanzania are a serious concern. It was reported that some patients were taking their ARVs only once a day, in the evening, because that was the time when they had food, and that some patients were selling their ARVs in order to buy food. This implies that food scarcity can be a drawback to adherence.

Although ARV users received medicines free of charge, the additional costs incurred were cited as an important reason for not visiting the health facility for follow-up and medicine refill. These costs included travel costs, registration or consultation fees at health facilities, and money spent on food while attending the treatment facilities. In addition, some patients had to travel long distances to the facilities and, in some cases, had to stay overnight near the clinic, thereby incurring additional accommodation expenses. Costs were seen as a constraint to adherence. Some ARV users suggested that they should be given loans to run small-scale businesses to help them cope with the additional costs incurred through being on ART.

Knowledge and beliefs

To be successful, an ART programme depends on a certain level of knowledge and awareness among ARV users. However, this study found wide variation in the level of knowledge among ARV users about HIV, AIDS and ART. While knowledge about HIV and AIDS is generally good, beliefs that people have been bewitched, had a spell cast on them or been afflicted by an AIDS devil/spirit (a 'jini') are commonplace and inhibit adherence to ART. Greater efforts are needed to educate both the community and ARV users about HIV in an effort to dispel beliefs about witchcraft.

Stigma

According to the testimonies of some ARV users, stigma remains a major problem. Moreover, disclosure and stigma seem to be different sides of the same coin. On the one hand, disclosure may cost the individual their job, social support and their family. On the other hand, a substantial number of ARV users who had disclosed were receiving support from family members. This support included financial assistance for travel costs and food, as well as reminders to ensure that they take their medicines on time. Although disclosure can have both negative and positive effects on adherence, it was more generally linked to better adherence, since 82% of those who had disclosed received various forms of help on the use of medicines. Efforts to ensure that the community is better educated about HIV and treatment would go a long way towards reducing stigma and encouraging disclosure.

Side-effects

A study conducted by Weiser et al. (2003) in Botswana indicated that side-effects did not pose a major barrier to adherence. The study found that while 51% of respondents experienced some side-effects, less than 10% of them reported side-effects as a significant barrier to treatment adherence. This was also noted by Akam et al. (2003) who found that very few side-affects were noted or cited as a cause of poor adherence (5%). In the present study, very few ARV users cited side-effects as a constraint to adherence. While some ARV users who participated in FGDs mentioned side-effects as a cause for skipping doses or taking medication only once per day, in most cases the side-effects were reported to disappear over time. However, this important information was not always communicated to ARV users in advance. In order to promote adherence, ARV users should have access to adequate education about potential side-effects and their likely duration.

Study limitations

There were some limitations to this study. First, the very few clients who refused to be interviewed or to turn up for FGDs may have significant information which has not been captured by this study. Second, the budget and time for this study were limited. Third, the method of determining adherence rates included self-assessment by ARV users and the literature suggests that patients tend to overestimate adherence (Chesney, 2000). Fourth, we were unable to relate the obtained adherence rate to viral loads and CD4 cell responses since this was not in our original plan, due to financial and logistical barriers to frequent laboratory monitoring in this setting. However, the combination of different approaches and respondents permitted extensive triangulation and gave us a comprehensive set of results in spite of the various problems faced.