From Access to Adherence: The Challenges of Antiretroviral Treatment - Studies from Botswana, Tanzania and Uganda, 2006
(2006; 320 pages) Voir le document au format PDF
Table des matières
Afficher le documentAcknowledgments
Afficher le documentAcronyms and abbreviations
Afficher le documentForeword
Fermer ce répertoire1. On hunger, transport costs and waiting time: a synthesis of challenges to ARV adherence in three African countries
Afficher le documentIntroduction
Afficher le documentMethods
Afficher le documentDiscussion
Afficher le documentConclusion
Afficher le documentReferences
Ouvrir ce répertoire et afficher son contenu2. Overview of antiretroviral therapy, adherence and drug-resistance
Ouvrir ce répertoire et afficher son contenu3. From training to action: the process of engaging health professionals in operational research on adherence to antiretroviral therapy
Afficher le document4. There's hope - early observations of ARV treatment roll-out in South Africa
Ouvrir ce répertoire et afficher son contenuFactors that facilitate or constrain adherence to antiretroviral therapy among adults at four public health facilities in Botswana: a pre-intervention study
Ouvrir ce répertoire et afficher son contenuA study on antiretroviral adherence in Tanzania: a pre-intervention perspective, 2005
Ouvrir ce répertoire et afficher son contenuFactors that facilitate or constrain adherence to antiretroviral therapy among adults in Uganda: a pre-intervention study
Afficher le documentBack cover


Our pre-intervention country studies were designed to estimate adherence levels in ART programmes in resource-poor settings under routine health care conditions, to identify reasons for sub-optimal adherence from the perspective of both ARV users and front-line health workers, and to recommend context-specific ways of improving adherence support. In this chapter we present an overview of the results of these three-country studies of adherence to ART conducted in Botswana, Tanzania and Uganda, and propose recommendations for action. The countries differ in HIV prevalence, density of ART sites, and in their level of ART coverage by December 2005 (see Table 1).

Table 1: Overview of ART in Botswana, Tanzania and Uganda, 2005


Population size

Estimated percentage adult HIV prevalence

Number of treatment sites

Number of people in need of treatment

Estimated percentage treated as of December 2005*


1.8 million



84 000



37 million



315 000



25 million



148 000



* For this table we refer to the WHO/UNAIDS '3 by 5' Report of 2006 and the UNAIDS 2006 Report on the Global AIDS Epidemic (see references), which have some discrepancies with the data for 2005 reported in the country studies.

Our study used rapid appraisal techniques (Vitolins et al., 2000) for collecting both quantitative estimates of adherence levels and qualitative data as to why sub-optimal adherence occurs. The methods used to collect data were: (i) semi-structured interviews (SSIs) with ARV users, health workers and key informants; (ii) focus group discussions (FGDs) with ARV users and key informants; (iii) adherence interviews with ARV users; and (iv) exit interviews and observations. Rapid appraisals tend to use a mix of methods to increase the validity of results. In this way, the evidence collected with different instruments can be compared and used to validate findings. The strongest evidence is that which emerges from different 'angles', i.e. through 'triangulation'. The qualitative methods (SSIs and FGDs) were mainly used to find out why people do not adhere. The quantitative adherence interviews were used to find out how often they do not adhere. The exit interviews allowed us to interview users about the information flows between health workers and users, as well as the quality of care provided. The observations were used to check on the availability of medicines and laboratory facilities, and to observe interactions between health workers and patients and the quality of care.

Ethical approval for the national studies was provided by national authorities in each of the three countries. Sample sizes are shown in Table 2.

Table 2. Number of respondents by data collection instrument


Number of facilities

SSIswith health staff

SSIs in community

No. of adherence interviews with ARV users

No. of focus group discussions

No. of exit interviews

Botswana (four regions)

4 public






Tanzania (Arusha and Dar es Salaam)

3 public
4 private






Uganda (Jinja only)

1 public
1 private







SSI: semi-structured interview

The adherence measurement tools

Given the lack of a gold standard for measuring adherence (Kent et al., 2003; Vitolins et al., 2000), and the pros and cons of different kinds of adherence measures, the Tanzania and Botswana teams selected three measurement tools for this study: (i) two-day self-report recall (ii) one-month visual analogue and (iii) pharmacy pill counts. The two-day self-report and one-month visual analogue recall methods have been found by Oyugi and colleagues (2004) to be valid instruments for estimating adherence in a recent study in Uganda.

The visual analogue method used in Botswana and Tanzania differed. In Botswana, ARV users were asked to indicate their adherence rate over the past month using a 10-centimetre long 'visual analogue' line. The beginning of the line indicated not taking the medications at all in the past month, while the end meant taking all of them as prescribed. The patient's mark was then measured using a 10 cm ruler and translated into percentages. The Tanzanian team used a glass-full of beads representing the total number of pills that the patient should have taken over the previous month. The researcher then asked ARV clients to pour the beads from one glass into another in order to estimate the number and percentage of pills they had not taken over the past month. The researchers used a centimetre measure to calculate the proportion of beads not 'consumed'. To calculate this, they divided the height of the remaining beads in the glass by the original height of the beads (representing the total number of pills to be taken over one month).

Each of these three measures has strengths and limitations. In order to reduce desirability bias, the trained researchers who conducted the two-day recalls and the one-month visual analogue methods were encouraged to be sympathetic to the problems experienced by respondents. The two-day recall has the advantage of a short time-span, which means that memory of medicine intake is likely to be good. However patients may feel ashamed to report specific instances of non-adherence that occurred in the 48 hours prior to visiting the health facility, especially if they have to specify on the chart exactly when they failed to take a pill and then to explain why. In terms of desirability bias, the one-month visual analogue methods are likely to be better. By estimating the number of pills missed over a one-month period, patients are confronted less with each specific non-adherent event. The pill-counts can be defined as the most 'objective' of the three approaches, measuring the actual number of pills left over since the previous refill. However, patients who fear the possible repercussions of revealing to the dispensing pharmacist that they have not achieved optimal adherence, may present fewer pills to the pharmacist than were actually left over. All three methods are likely to overestimate adherence.

The key to measuring adherence accurately is to ensure that respondents do not feel threatened when reporting in one way or another a non-adherent event. Rather than measuring exact levels of adherence, these measures should be seen as producing 'good enough' estimates of adherence. Given that the optimum level of adherence is at least 95%, the aim of adherence measures should be to determine to what extent such near-complete adherence is actually being achieved. Since lapses in adherence can lead to treatment failure and the emergence of drug-resistant HIV, poor adherence is not only a problem to users but to public health in general. For individual patients, the adherence tools can perhaps best be used as points of reference in counselling sessions on adherence, in order to discuss the reasons for sub-optimal adherence and ways to overcome these constraining factors.

Sampling frame

Since the low budget for these studies did not allow for a representative sample of the countries' ART sites, the sampling frame used was multi-staged. Health facilities were selected purposively, with the aim of including a diversity of facilities as well as limiting travel costs and time. In Tanzania, seven health care facilities were chosen in two cities (Dar es Salaam and Arusha). They included both public facilities and private/faith-based facilities which had been providing ARVs for at least three months. In Uganda, two study sites were selected in Busoga region, a sub-region of eastern Uganda which has both a public health facility and a private facility providing ART, each site relatively research-naive. In Botswana, the study sites were located in three of the country's nine districts: North West (Maun), Central (Serowe and Mahalapye) and Kweneng (Molepolole). Serowe and Maun were among the pilot sites, while Mahalapye and Molepolole were second-generation facilities.

In Tanzania and Botswana, sample-size calculations were conducted to determine the number of users to be interviewed for the purpose of estimating adherence levels at the facilities. In Tanzania, sample size calculation for ARV users using the adherence tool was based on the results of the pilot study (mean overall adherence rate=98% at 95% confidence intervals) which gave 24 per each health facility (estimated total of 168 ARV users for seven facilities). ARV users were randomly chosen using the outpatient attendance register. Inclusion criteria were: adult, aged 18 years or over; willing to participate in the study; and on ART for at least three months. Only 107 ARV users met the inclusion criteria in Tanzania during the days of data collection.

In Botswana, the sample size for the quantitative data required to obtain estimated proportions with 95% probability level was estimated using the CSURVEY design in Epi Info 6, version 3.22 (Centers for Disease Control and Prevention, 2004), based on the assumption that 85% of the patients achieve optimal adherence (i.e. at least 95%). The sample size required was 93, 96, 95 and 93 (total 377) for the four selected facilities (Mahalapye, Serowe, Maun and Molepolole respectively). However, a total of 514 users were interviewed at the four sites. From the determined sample size, the number of patients to be interviewed each day over the five-day period was determined. The number of patients expected on a given day was determined from the consultation appointments at the data clerks' office and from the dispensary appointments. This number was then divided by the number of patients to be interviewed, to determine the Xth patient who was to be picked for the interview. From then on, every Xth patient was picked until the required numbers of patients were interviewed per day. If the patient declined, the immediate next patient was selected.

In Uganda, the study population consisted of patients aged 18 years or over who were receiving treatment at the two study sites. All patients visiting the sites during the study period who were in this age range, on ART, and willing to participate in the study were included. Systematic sampling was used to select the final sample, comprising every third patient visiting the clinic on the day of the fieldwork. Wherever a patient was not interested in being included in the study, the next patient was considered. A registry file from each facility's reception was used as a sampling frame from which ARV users were selected for the study using systematic sampling. In Uganda, a total of 71 ARV users were interviewed for the adherence estimates.

Study period

Data collection took place between May and September 2005. Quantitative analysis involved a standard Access data entry and analysis programme. Qualitative data were analysed thematically, either manually or with NUDIST software. The research and training activities undertaken prior to the study and after data collection was completed are given in Chapter 3.

Quantitative results: adherence estimates

Tables 3 and 4 present the adherence estimates by adherence measurement tool.

Table 3. Average percentage of doses taken at the right time in the study population, by adherence measurement instrument


Two-day recall

One-month visual analogue

One-month pill count


98% (n=508)

92% (n=496)

93%(n = 443)


100% (n=107)

83% (n=107)

98% (n = 107)

The levels of adherence appear high, based on the average percentage of drugs taken at the right time. The estimates are highest for the two-day recall method, suggesting a relatively high desirability bias. For Botswana, the visual analogue and pill count methods produced a remarkably similar average level of adherence. For Tanzania, adherence levels measured by visual analogue method show lower levels of adherence than by pill-counts. In Uganda, a two-day recall was used. No sub-optimally adherent events were reported, suggesting either optimal adherence or a high level of desirability bias. Because only one measure for assessing adherence was used and we doubt the validity of the results, we do not present the Ugandan quantitative data.

Table 4. Percentage of respondents with optimal ARV adherence rates (at least 95%), by adherence measurement instrument


Two-day recall

One-month visual analogue

One-month pill count


96% (n = 508)

60% (n=496)

75% (n = 443)

Table 4 shows the percentage of patients who achieved the optimal level of at least 95% adherence, as calculated in the Botswana study. The estimates based on both the visual analogue recall method and pill counts are low. On the basis of pill counts, an estimated one in four patients in Botswana failed to achieve this level. When measured by visual analogue, the proportion of users who failed to achieve optimal adherence is even higher, 40% in Botswana. The Botswana study also quantified the reasons reported by ARV users for missing a dose of ARVs, see Table 5.

Table 5. Reasons reported for missing medications in Botswana
(N = 514 ARV users)


No. of ARV users reporting this reason

% of ARV users reporting the reason

Simply forgot



Logistics and transport costs



Work or home duties






Lack of care/support



Misunderstood instructions



Lack of food



Distance to the health facility



Being in hospital



Alcohol abuse






Feeling better



Pill burden



Qualitative results: what are the constraints to optimal adherence?

The qualitative findings from these three country studies suggest that although patients are highly motivated to take ARVs as prescribed, constraints such as transport costs, user fees, long waiting times, hunger, stigma, side-effects and lack of appropriate counselling undermine their intentions to do so. Meanwhile, front-line health workers have to contend with heavy workloads, a lack of laboratory facilities and occasional stock-outs of ARVs. However, despite these health system constraints, the perceived quality of care among users is relatively good. In the section below we discuss the most commonly reported challenges to ARV adherence.

Transport costs and user fees

Although participants in all three studies received medicines free of charge, transport costs are an important reason why ARV users fail to visit the health facility for follow-up and refill. These accounts from semi-structured interviews illustrate the problems:

A minibus delivers researchers for fieldwork in South Africa, but for many ARV users the cost of transport to health facilities and the general lack of transport in rural areas are serious problems.

"I came from very far, over 50 kilometres from here. Before I come to the hospital I have to plan the money for a journey fare to the clinic. In fact my extra drugs got finished yesterday." (Male ARV user, Uganda)

For others, the lack of a means of transport - especially from remote areas - can be an additional challenge:

"I once missed my appointment for refill because there were no vehicles coming here. I was in the stop from early morning and by noon I went back home. Fortunately I still had some medications." (Male ARV user, Botswana)

An ARV user in Uganda suggested opening more treatment centres that were closer to home:

"I have very many people in the village, they are dying because they don't have money to transport themselves to the hospital. You need to have this money monthly. Like me, from the village where I come from, getting up to this place costs 15 000 Shillings (US$ 8.50). To and fro is 30 000 Shillings (US$ 17.00), which is a lot of money. And getting that money is a problem. So maybe, like people in Kyoga, if they can send that drug up to Kyoga, I think that could be good. Right now only Lira Referral Hospital gives ARVs, and that is 130 kilometres from our place (Kyoga). Very far !"

However, problems can persist even when treatment centres are more locally available, as explained by a respondent in Tanzania, who complained that he was denied access to a centre that had opened close to his village:

"I was registered to start ART in Kilimanjaro Christian Medical Centre (KCMC) in Moshi a year ago. At that time there was no ART clinic near my village. Now there is a clinic near my home but I am denied transfer from KCMC to my home clinic. KCMC is very far from here, about 170 km away. Some times I do not have the fare to travel to KCMC, hence I miss my doses." (Male ARV user, Tanzania)

Health workers in all three countries were well aware that transport costs impede adherence. As one health worker in Uganda reported:

"Some people have failed to report to the clinic on time because they failed to get transport to reach the clinic. Some people come from the islands, and they will tell you that they did not get money to cross the waters and that is why they did not come on time. And when you are told that, you cannot do much but to hope that when the next visit comes, he can afford to come on time."

In addition to recurrent transport costs, patients have to pay registration and user fees in private facilities in Uganda and Tanzania. In Uganda, a fee of US$ 3.00 was charged in the private facility at each visit. In Tanzania, the private facilities charge user fees with a range of US$ 1.50 to US$ 3.00 per visit, and an additional US$ 15.00 for laboratory investigations.

Waiting times

In all three studies, the problem of long waiting times was cited as a major challenge to adherence. In Tanzania, the mean time spent at the clinic was six hours. About half (12/28) of the health workers interviewed in Tanzania identified long waiting times as a problem. In Botswana, most respondents reported that they spent around four hours at the clinic. Nearly half of the respondents spent even more than that, with the longest wait being 12 hours, as shown in Figure 1.

Figure 1. Distribution of the time spent by Botswana participants at the clinic (N = 128 exit interviews with ARV users)

In Uganda, the average waiting time for ARV users was five hours in the public facility and one hour in the private facility. The findings suggest that ARV users may miss one working day per month in order to get ARV refills. This can be a problem for some ARV users whose employers do not know that they are HIV-positive or do not support their need for care. One ARV user in Botswana said:

"I resorted to asking my relative to pick up my medications, because my employer refuses to release me to go and pick up my medications."


ARV users in all three countries complained about hunger during the initial stages of treatment, when the body needs extra nutrition as it regains strength and weight. They said they could not afford the amount of food needed to satisfy their increased appetites. The following quotes illustrate this.

Carrying home food parcels from a health care facility in South Africa - but not everyone is so fortunate. Some participants in the studies reported that hunger had an adverse effect on adherence, particularly in the initial stages of ART.

"The problem I have with ARVs is related to food. I have no money and ARVs increase appetite. I am not capable of buying food.

(Male ARV user, Tanzania)


"I want to eat all the time and fear the hunger will eat into my stomach, since I have ulcers already. Sometimes I wake up in the night to eat food. This is a difficult situation for me."

(Male ARV user, Uganda)


"Majority of people say the ARV treatment makes them to eat a lot. They go to an extent of begging for old age pension from their grandparents. Others quit the treatment because they complain about the lack of food."

(FGD participant, Botswana)


"Some patients have expressed lack of food as a reason for not wanting to swallow the life-saving drugs. In fact we have one woman who has declined her life-saving drugs because she does not have enough food to feed herself." (Doctor, Uganda)

In Tanzania, a female participant of a FGD reported that, because some ARVs have to be taken with food, some patients take their medicines only once a day in the evening (instead of twice daily), because that is the only time they have food.

In Uganda, some patients receive food support (soya flour, cooking oil, rice, sugar and maize flour) from TASO, the nongovernmental AIDS Support Organisation. In Botswana, the Government provides a food basket for ARV users who have been assessed by social workers and found to meet certain criteria.


All three country studies reported on ARV users' experiences of stigmatization and discrimination and lack of social support. Some ARV users reported that after disclosing their HIV-positive status they had lost their job (Tanzania); were abandoned or badly treated by their partners (Botswana); or were isolated by community members (Uganda). Fearing such stigmatization, ARV users often decide to hide their HIV status, from colleagues, friends and others.

If ARV users do not disclose their HIV-positive status it may affect adherence in different ways. Firstly, non-disclosure may lead to patients taking their ARV medicines secretly and irregularly, as illustrated by quotes from ARV users in Uganda and Botswana:

"I cannot take my drugs when people are seeing. I always go and hide when I take them. Otherwise, people start whispering about you all the time." (ARV user, female FGD, Uganda)

"I usually miss my medications when I visit friends because I have not told them about my HIV and so I do not want them to see my medications." (Male ARV user, Botswana)

Moreover, when ARV users do not disclose their HIV-positive status to others they will not receive adequate social support and encouragement to take their drugs regularly and on time. The Tanzania study found that of 30 ARV users interviewed, 94% had disclosed their HIV-positive status and 83% received various forms of help from family and friends on the use of drugs (e.g. transport support, food, reminding them to take drugs).

In all three countries, children were found to be important sources of support for ARV users, as illustrated here:

"My children after seeing the state I was in and after getting ARVs, I called them and told them about my state. They got encouraged and as a result they buy me passion fruits and sugar because they know the drugs I am taking are so strong. I even wrote my file number in TASO on the wall and told them that just in case I am badly off they can go to TASO and get me help. My children know very well that because of my drugs I have to drink enough and to eat on time. One thing that motivated me to tell them is because I thought I could be so weak to collect my refill of the drugs (ARVs). They even know the name of my counsellor." (ARV user, mother of five, Uganda)


The side-effects most frequently mentioned by ARV users were: body rash, swollen legs, nausea, headache, increased heart rate, diarrhoea and vomiting. In Tanzania and Uganda, the occurrence of side-effects was mentioned as an important reason for skipping doses.

"I had side-effects and decided to take medication only once per day." (Male FGD, Tanzania)

"Feeling a lot of heat in the body, especially after taking the drug and excess sweating makes one embarrassed in public. So, you feel like postponing the drug to a later time when you are not relating with people." (Male ARV user, Uganda)

In most cases, side-effects disappear over time. However, ARV users in Tanzania and Uganda have not always been given this important information. By contrast, in Botswana, where side-effects are discussed extensively in pre-treatment counselling, of the 58% of ARV users (n=514) who reported that they had experienced side-effects, only 8% cited them as one of the reasons for missing their medication, suggesting that effective counselling increases tolerance.

Lack of counselling

Counselling is a key requirement for successful ARV adherence. However, the frequency and quality of counselling was found to differ greatly both between the different countries and among the different facilities within each country. In Botswana, well-trained counsellors (nurses, social workers and lay counsellors) are available in all health facilities providing ARVs. In the public facility in Uganda, the counselling was done by nurses, who were not well trained because the public health facility could not afford to pay for good quality training courses for counsellors. ARV users in Uganda valued support from the community-based volunteers of TASO, many of whom are HIV-positive themselves. In Tanzania, the quality of counselling was found to be different in Dar es Salaam and Arusha. While patients in Dar es Salaam appreciated the quality of the counselling received, several ARV users in Arusha complained about the quality of their counselling due to the lack of trained counsellors. The exit interviews confirm that only a small proportion of ARV users see a counsellor (21%), while, in contrast, almost all (97%) see a doctor.

Table 6: Categories of health staff seen by patients in Tanzania
(N=70 exit interviews with ARV users)

Cadre of staff

No. of patients who consulted with this category of health worker





General nurse









As one FGD participant commented:

"You find 25 patients and only one person attending all these patients and he just tells you to go and collect your medication." (Male FGD participant, Tanzania)

Heavy workloads

At the public facilities in both Uganda and Tanzania, the scaling up of ART had occurred without any increase in personnel to cater for the increasing numbers. As a result, health workers were visibly overworked as they struggled to cope with the large number of patients on clinic days. One health worker in Uganda said:

"You overwork like this without even a break because there are too many people all coming one day and yet you are very few." (Health care worker FGD, Uganda)

Table 7 shows the main challenges reported by 28 health workers in Tanzania.

Table 7: Challenges most frequently mentioned by health workers in Tanzania (N=28)


Number of respondents


Low motivation



Heavy workload



Inadequate training



Long waiting hours for patients



Too few staff



Work fatigue



Being faced with difficult or non-compliant ARV users



Commenting on the large number of clients attending ART clinics, a female participant of a FGD in Tanzania said:

"If the situation remains like this, doctors will be tired and the last patient will not be attended (to) properly."

It is remarkable that, despite these heavy workloads, patients in all three countries expressed satisfaction with the quality of care at the facilities. In Botswana, 99% of ARV users interviewed in exit interviews at the facilities said that they felt listened to. In Tanzania, 80% of the ARV users at the facilities reported that they had been asked about their experiences in taking ARVs. In Uganda, out of 26 users in the public facility, 16 said the services offered were good, 8 rated them as fair and only two said they were poor. In contrast, all 20 participants interviewed at the private facility said the services were very good. With the exception of two users at the public facility, participants at both the public and private facility felt that the service providers listened to them. They trusted the health workers.

Lack of space for confidential consultations

The findings suggest that space for consultations is generally adequate in private facilities, but can be lacking in the public sector. In Uganda, the public facility lacked enough room to accommodate patients for counselling and to discuss personal issues. As a result, the nurses took them to any free space available, thereby compromising confidentiality. In contrast, the private facility in Uganda was very well organized, with comfortable seating, less waiting time and everybody was attended to. In Tanzania, space for the ART clinics was generally adequate in both Dar es Salaam and at the faith-based organizations in Arusha. However, at one public hospital in Arusha there was no separate room for consultation and thus no confidentiality. At the time of the study, three doctors at this facility were sharing a single room and consulting with three different ARV users at the same time. In Botswana, the space for consultation was found to be inadequate in two of the four public health facilities.

Lack of CD4 machines and ARV stock-outs

In Botswana and Tanzania, health facilities were found to lack laboratory facilities to conduct CD4 counts, which are needed to monitor the efficacy of the treatment and also to decide when to initiate treatment. In Uganda, the private facility conducted all recommended tests for HIV management within the centre while the public facility had to send out some of their samples for CD4 and viral load testing. This implied that people who were very sick had to wait at least two weeks before they could be put on therapy.

In Tanzania, four of the seven facilities studied lacked a CD4 machine. In Botswana and Uganda, both ARV users and health workers complained about inoperative testing facilities. Even where CD4 machines were available, reagents were sometimes out of stock. Moreover, two of the seven facilities also reported stock-outs of ARVs, which is a major challenge to adherence. Although the supply of ARVs was reliable in most of the facilities at the time of the study, some patients expressed concern about possible future stock-outs, as one respondent from Uganda explained:

"We are grateful for the government for bringing medicine to the people, but we hear it is only for five years. Whenever I take these drugs, I am wondering whether in the next five years I will still have them free. Actually I get disturbed by that." (Male ARV user, FGD, Uganda)


vers la section précédente
vers la section suivante
Le Portail d'information - Médicaments essentiels et produits de santé a été conçu et est maintenu par l'ONG Human Info. Dernière mise à jour: le 29 octobre 2018