Article Text

Download PDFPDF
National HIV treatment guidelines in Tanzania and Ethiopia: are they legitimate rationing tools?
  1. K A Johansson1,2,
  2. D Jerene2,3,
  3. O F Norheim1,2
  1. 1
    Division of Medical Ethics, Department of Public Health and Primary Health Care, University of Bergen, Bergen, Norway
  2. 2
    Centre for International Health, University of Bergen, Bergen, Norway
  3. 3
    Arba Minch Hospital, Arba Minch, Ethiopia
  1. Kjell Arne Johansson, Division of Medical Ethics, Department of Public Health and Primary Health Care, Centre for International Health, University of Bergen, PB 7804, 5020 Bergen, Norway; Kjell.Johansson{at}


Objective: To provide an ethical analysis of whether the Ethiopian and Tanzanian national HIV/AIDS treatment guidelines can be considered legitimate and fair rationing tools.

Method: Qualitative study and ethical analysis involving guideline documents and interviews with nine key members involved in the development of the guidelines. The analysis followed an editing organising style. The theoretical framework was a guideline-specific framework based on theories of just resource allocation in healthcare and conditions that ensure fair processes in guideline development. According to this framework, legitimate rationing requires reasons for patient selection to be explicit, public and relevant, and decisions must be open to question and revision.

Results: The only explicit rationing criteria that both guidelines recommended were clinical antiretroviral treatment indications. Explicit non-clinical rationing criteria were expressed in a separate Ethiopian implementation guideline. Neither of the guideline development processes fully satisfies minimal requirements of procedural fairness. There is a lack of transparency. The reasons for decisions are rarely given and are not publicly available. This reduces the opportunity for public questioning, debate and revisions. The guidelines were based on expert opinion and consensus. Recommendations from the WHO were copied without much discussion, disagreement or adjustment.

Conclusions: The two national HIV treatment guidelines discussed are de facto mechanisms for rationing but were developed using methods that do not fully satisfy the requirements of fair processes.

Statistics from

Clinical practice guidelines describe the quality and the terms of access for managing a particular disease. Evidence-based guidelines aim to identify patients who will benefit most from an intervention. Effective and fair service provision can be achieved by using accepted evidence-based guidelines in priority-setting processes.1 2

Fair selection of patients for antiretroviral treatment (ART) is important in Ethiopia and Tanzania, as elsewhere. Much is at stake because of extreme resource scarcity and high HIV prevalence. There are 1.3 million patients in Ethiopia and 1.8 million in Tanzania with HIV.3 4 Numbers eligible for ART will vary according to the treatment indications used, but some argue that more than 50% would be eligible for ART if the US guidelines were followed.5 There is not enough ART for everyone in clinical need.6 Up to March 2007, 61 300 patients in Ethiopia and 59 300 patients in Tanzania had received ART.7 National HIV/AIDS guidelines have been implemented in 200 clinics in Tanzania and 106 in Ethiopia, and all the sites were selected to distribute ART by the respective ministries of health.8 9

Rationing in this context refers to the withholding of beneficial ART through national guidelines. This raises a question of legitimacy. When can denial of beneficial healthcare through guidelines be justifiable to the general public and to those who are denied care? The procedural justice framework of accountability for reasonableness, developed by Daniels and Sabin, specifies certain conditions for how decisions should be made so that the development process can be acceptable to everyone. The framework does not seek to solve substantive disagreements on value judgements. Rather it argues that rationing is legitimate only if the reasons behind the decisions are reasonable and transparent. It is not enough just to announce the results of the deliberation.10

To solve controversial issues in the ART roll-out, Daniels has argued for a fair, deliberative process about the facts, reasons and principles that are relevant to the discussion.10 11 The World Health Organization (WHO) and the Joint United Nations Programme on HIV/AIDS (UNAIDS) recommend that ART priorities in resource-scarce settings should be based on principles of efficiency and fairness and that the reasons for treatment indications should be explicit, public and relevant to all affected parties.12 In 2003, the WHO published an ART guideline for resource-limited settings.13 The recommendations in this guideline are followed in most low-income countries.14 The revised edition (2006) included updated evidence but still relied on expert opinion and had no systematic reviews supporting the evidence referred to.15

Policy reviews from Mexico, Senegal, Thailand, Uganda, Zambia and Tanzania found that medical treatment indications, adherence to treatment and ability to pay are explicit rationing criteria in ART programmes.1517 ART indications differ between high- and low-income countries. Rich countries give treatment for asymptomatic patients with a CD4+ count ⩽350 cells/μl.18 Indications for the same patient group have been, until recently, set at CD4+ count <200 cells/μl in resource-limited settings.6 13 14 Historical experience shows that withholding beneficial treatment on clinical criteria alone makes it hard for the public to understand why some patients but not others are denied treatment.19

Our ethical preconceptions draw on a guideline-specific framework based on theories of just resource allocation in healthcare and deliberative democracy.20 The ethical framework, developed by one of us (OFN), specifies conditions of fair processes for guideline development. The reasons for these conditions are published elsewhere.1 20 To the best of our knowledge, there are no ethical analyses of the content and rationale of HIV treatment guidelines in resource-limited settings.

This paper aims to provide an ethical analysis of whether the Ethiopian and Tanzanian national HIV/AIDS treatment guidelines can be considered legitimate and fair rationing tools.


We provide a qualitative study and ethical analysis. The material was drawn from national HIV guidelines in Ethiopia21 and Tanzania,22 supplemented by audio taped interviews of members from the core guideline development groups. Data were collected between April 2005 and May 2006.

Both countries had technical working groups with 10–11 members, which consulted stakeholders through workshops. Both countries had smaller clinical review groups responsible for the clinical recommendations.

Sampling of key informants was purposeful—that is, we wanted to explore the perceptions of people with different backgrounds and perspectives. Representatives were selected from the ART guideline acknowledgement list. We identified key informants from each technical working group and clinical review group, six from each country. Five Tanzanian and four Ethiopian members were interviewed. One member of the Tanzanian development group did not want to participate and two members in the Ethiopian group were not accessible. All interviewees were experienced physicians from internal medicine, paediatrics or public health. Experiences ranged from treating HIV patients to clinical HIV research to administration of large public health initiatives. The individuals represented national governmental hospitals, the National Aids Control Programme, WHO country offices or Family Health International. Representatives from the general public were not interviewed, since they were not included in any of the technical working groups.

The interview guide was developed using nine key questions from the predefined normative framework.20 Each question represents a criterion that is considered important for the legitimacy of guidelines (table 1). Four criteria help to decide what information is needed to see and evaluate the consequences of applying the guidelines. Five criteria are important for assessing the actual process of developing the guideline.

Table 1 Acceptability of the information and development process for each country

We used an editing organising style to analyse the transcribed interviews and guideline documents.23 First, we identified meaningful units or segments of text in the transcribed interviews and the guideline documents that stood on their own or related to this study. We tried to code these units without a template to see if the interviewees or guideline documents incorporated valuable information that was not included in the normative framework. Next, codes with similar contents or themes were put into broader categories. Finally, these categories were integrated into the structure of the predefined normative framework.

In addition to this qualitative analysis, all authors graded acceptability for each of the nine criteria, using a scoring system with values ranging from 1 to 4, where 1 is not acceptable and 4 is fully acceptable. Scores were consensus based and there were no major disagreements. All authors participated collaboratively in the analysis.


Development of the two national ART guidelines was initiated in 2004 by the two ministries of health and their cooperating institutions. Both guidelines are part of large national ART roll-out programmes that were initiated when the WHO launched the 3 by 5 programme.13 This programme uses initialed in 2001, with a global aim of treating three million HIV patients by 2005. The ministries of health in Ethiopia and Tanzania have developed national health sector and multisectoral HIV strategies and policies (box 1).9 24 Health ministries in each country are supported by two separate national governmental institutions, coordinating health sector responses or multisectoral responses.8 22 25 26

Box 1: Key HIV documents in Tanzania and Ethiopia


  • Guidelines for use of antiretroviral drugs in Ethiopia (2005)21

  • Guidelines for use of antiretroviral drugs in Ethiopia (2003)

  • Guideline for implementation of antiretroviral therapy in Ethiopia (2005)26

  • Ethiopian strategic plan for intensifying multi-sectoral HIV/AIDS response—2004–2008 (2004)

  • National anti-retroviral therapy (ART) strategic communication framework (2005)

  • Accelerating access to HIV/AIDS treatment in Ethiopia—road map for 2004–2006

  • AIDS in Ethiopia report: 6th edition (2006)

  • Health sector strategic plan - 2005-2009 (2005)

  • Policy on HIV/AIDS of the Federal Democratic Republic of Ethiopia (1998)

  • Antiretroviral therapy info kit (2005)


  • National guidelines for the clinical management of HIV and AIDS (2005)22

  • HIV/AIDS care and treatment plan 2003–2008 (2003)

  • National multi-sectoral strategic framework on HIV/AIDS 2003–2007 (2003)

  • National policy on HIV AIDS (2001)

  • Health sector HIV/AIDS strategy for Tanzania 2003–2006 (2003)

Guideline documents and answers from the guideline developers, shown in detail below, showed that three important ethical choices were made in both sets of ART guidelines. The first concerned medical indications for ART, the second concerned non-medical indications for ART and the third concerned the choice of drug regimes. Value judgements and methodological uncertainty surround all these issues.

Table 1 shows the numeric acceptability grades the authors gave for all nine criteria. The theoretical framework originally assesses legitimacy using only the guideline documents. However, we had to supplement our information with interviews, because both guideline documents had scarce information on some important issues. The acceptability scores are therefore divided in two—one based on the information in the guideline documents and one based on information that came from the interviews. The reasons for each score are given in detail below. Quotations are attributed by a number assigned to each informant.

Acceptability of information needed for guideline appraisal

The first criterion is transparency. Indications for ART are stated explicitly (box 2), but no reasons for their choice are given in the guideline documents. Developers in both countries decided to provide ART to asymptomatic patients with a CD4+ count <200 cells/μl, to patients in WHO stage 3 with a CD4+ cell count <350 cells/μl and patients in WHO stage 4 regardless of CD4+ count. Recommendations were in concordance with the 2003 WHO guideline.

Box 2: Explicit ART indications in the Ethiopian21 and Tanzanian22 HIV guidelines

If CD4+ count is available (recommendations present in both guidelines):

  • CD4+ count <200 cells/μl (irrespective of WHO stage)

  • WHO stage 3 and CD 4 count <350 cells/μl

  • WHO stage 4 (irrespective of CD4+ count)

If CD4+ count is not available (recommendations present only in the Ethiopian guideline):

  • WHO stage 4 irrespective of total lymphocyte count (TLC)

  • WHO stage 3

  • WHO stage 2 and total lymphocyte count ⩽1200 cells/μl

The second and third criteria focus on the rationale behind decisions. The reasons behind the treatment indications were not stated explicitly in the final national guideline documents. The WHO guideline had some references to medical evidence. The lack of good evidence on treatment effect was the main reason why asymptomatic patients with a CD4+ count >200 cells/μl were not considered eligible for treatment. Treatment indications were justified only with reference to medical evidence. The strict cut-off was not explicitly justified by reference to resource scarcity, but is rather framed as a purely medical judgement. However, according to one of the developers, cost played a role in setting the cut-off:

What I’m saying is to start ART at 350, rather than 200, means that you are going to include a large number on initiating treatment. This means you are going to use more money. (TZ 1)

Indications for ART were not discussed explicitly or justified with reference to non-medical characteristics of patients. Non-medical characteristics were considered important for adherence to treatment. Active alcohol or other substance abuse, lack of openness about HIV status, not understanding the importance of adherence to treatment, untreated active depression, and previous non-compliance at HIV clinics were perceived as non-adherence risk factors in both guideline documents. If any of these risk factors were present, then both guidelines suggested that health workers should try to increase patient adherence. It was not stated clearly whether any of the risk factors mentioned were clear contraindications for treatment. One of the guideline developers said:

For people who can not adhere for some reason or another, like drug abusers, like people who drink too much alcohol and people who chew khat … so these kind of people are very careless and they do not adhere to treatment. So if there is anything which may compromise, then the treatment should be different or delayed until we correct this problem. (E5)

In the interviews, it became clear that people in the development groups disagreed about whether poor adherence to treatment should be a treatment exclusion criterion. Some wanted it to be a clear condition for exclusion, to ensure maximum health effect and to prevent drug resistance. They feared that too many patients would need expensive second-line treatment if patients did not take the drugs properly. Others feared that low adherence-assessment skills among health workers would lead to discrimination.

There were various explanations for the fact that treatment indications were only medical. The major impression was that a clinical guideline should focus only on clinical criteria and medical aspects. In Tanzania, some individuals said that inclusion of non-medical criteria in the ART guideline was not even discussed during the development process. Others had the impression that a majority of patients were not coming to clinics for treatment and therefore they saw no urgent need to set further priorities in the guideline.

According to some guideline developers, long explanations of reasons and large systematic reviews could hinder effective guideline implementation. Short, simple, easy-to-read and standardised documents with recommendations that are easy to find were regarded as best for busy clinicians.

The fourth criterion concerns the availability and accessibility of the information. The Tanzanian ART guideline is publicly available9 but the Ethiopian one was available only at the clinics implementing it. ART indications were written in technical language and were comprehensible to key stakeholders with prior knowledge about HIV, CD4+ counts and WHO clinical staging. The empirical rationale behind ART indications was only indirectly accessible through the 2003 WHO guideline. Since the 2003 WHO guideline was based on expert opinion and consensus and did not itself follow the development methods recommended by evidence-based medicine, the evidence was not easily accessible. The best evidence referred to was the 2002 International AIDS Society–USA Panel review consensus report.27

Acceptability of the development procedure

The fifth criterion underlines the need for information about how the guidelines were developed. There were no separate Methods sections in any of the guideline documents. Nothing was written about the process of arriving at recommendations, the composition of the development group or whether the guidelines were subject to consultation among a broad range of stakeholders. There was no information about how the development groups resolved disagreements or how the national guideline development groups reviewed international recommendations about eligibility.

Better information about the development process was obtained through the interviews. The ministries of health in Ethiopia and Tanzania initiated development of the national ART guidelines. The National AIDS Control Programme was responsible for development in Tanzania and the HIV/AIDS Prevention and Control Office and the Drug Administration and Control Authority were responsible in Ethiopia. Experienced HIV clinicians and representatives of internal medicine and paediatrics first developed guideline drafts in small groups. Previous national guidelines were revised and new drafts were presented to large workshops in their respective countries.

Criteria 6 to 8 consider representation of different interests and public accountability. We heard similar stories about involvement of stakeholders, key disciplines and patients. A broad range of stakeholders were consulted in both countries. Health providers, health planners, representatives of various international agencies and several key disciplines (psychiatry, paediatrics, internal medicine, nutritional medicine and laboratory medicine) were present at the national workshops. Few were consulted about the medical ART indications. Patient representatives were included at the workshop in Tanzania but not in Ethiopia. Participants from various groups of stakeholders were not given an opportunity for input and disagreement on important decisions in these workshops. Those invited were mainly told what the experts had decided. No other efforts were made to include the views of patients or public in either Tanzania or Ethiopia. One of the informants said:

… the attempt was to include all stakeholders. But as you go down to the guidelines, I think the speed at which the document has needed to come out excluded some people as part of the process. (TZ5)

The final criterion recognises the value of explicitness about economic or political decisions. Several donor agencies were involved in the ART roll-out programmes. How the views or interests of the funding bodies, resource scarcity or political decisions influenced the final guideline recommendations was not stated explicitly. The Ministry of Health in Tanzania decided before guideline development that funding from NORAD (the Norwegian Agency for Development Cooperation), the Canadian International Development Agency (CIDA) and the government itself should go to generic first-line drugs, and funding from PEPFAR (the US President’s Emergency Plan for AIDS Relief) should go to second-line drugs. This political decision had a direct influence on the decision to provide both first- and second-line drugs.

The WHO recommendations had a major influence in both countries on the indications for ART. Most members of the development group felt there was no real alternative and claimed that low-income countries were dependent on the WHO recommendations for two reasons: first, the costs of just copying the WHO indications were lower; second, the WHO recommendations were seen as state of the art. One member said:

There were no alternatives to consider, so WHO recommendations were taken at face value because people trust WHO very much. And that was how it was developed. (E1)

Before the ART roll-out started in Ethiopia, there was fear about a lack of drugs relative to the number of patients coming to the clinics. As a result, a separate ART implementation guideline and a rather complex cost-sharing scheme was developed. The sickest patients with ART-reversible conditions, children, pregnant women, health workers exposed in the line of duty, rape victims and gender equity were given explicit priority in the implementation guideline (box 3). Responsibility for the poorest was the guiding principle in the cost-sharing reform, and concrete economic eligibility requirements were suggested. Just before the cost-sharing scheme was put into practice (October 2005) the Ministry of Health decided that it would provide free ART in public hospitals. The cost sharing scheme was therefore dropped, but the implementation guideline, including non-medical priority criteria, was implemented at HIV clinics distributing ART.

Box 3: Recommendations in the Guideline for implementation of antiretroviral therapy in Ethiopia26

Priority should be based on

  • Clinical criteria:

    • ART guideline treatment indications

    • The sickest with ART-reversible conditions

  • Non-clinical criteria:

    • The most vulnerable groups (defined as children, pregnant women, health workers exposed in the line of duty and rape victims)

    • Gender equity

  • Facility-level HIV/AIDS committees:

    • recommended to serve as ART access committees


In sum, neither guideline document gives enough information about the rationale behind the recommendations or the development process for them to be considered fair tools for patient selection. The medical criteria in the guidelines are explicit and were made public, but their scientific and ethical rationale is not publicly accessible. Explicit non-clinical rationing criteria were expressed only in a separate Ethiopian implementation guideline. Several stakeholders were involved and workshops functioned as open hearings during the development. However, information about this involvement is scarce in both guideline documents. The guidelines were based on expert opinion and consensus rather than evidence-based development methods. Recommendations from the WHO were copied without much discussion, disagreement or adjustment.

A recent WHO study of equity and fairness in the scale-up of HIV/AIDS services in Tanzania supported our findings that there had been broad consultation processes and that WHO medical eligibility criteria were adhered to, but there was no discussion of patient selection beyond medical eligibility criteria.15 However, the WHO study concluded that many decisions, especially clinical recommendations, rested on rational arguments and evidence. We found it difficult to assess the reasonableness behind recommendations, because many reasons were not openly shared.

Guideline developers in Ethiopia and Tanzania did not see any alternative other than to copy WHO guidelines. Many countries follow WHO recommendations.14 This raises the issue of legitimacy at different levels of guideline development. Legitimate international guidelines will increase the likelihood of legitimate national guidelines, because transparency facilitates public questioning and debate and local revisions. However, guideline developers at all levels must be held accountable for the limitations they put on healthcare and for ensuring that they follow the underlying conditions needed for a fair process. The WHO guidelines themselves fall short in this respect.

The revised 2006 WHO guideline acknowledges that there is a potential medical benefit from treating asymptomatic patients in resource-scarce settings at CD4+ counts >200 cells/μl and <350 cells/μl, and recommends that these patients should be included or at least considered eligible.6 If guideline developers in Ethiopia and Tanzania in further revisions decide to follow the WHO early-treatment recommendations, they will need to be aware that at least three times as many patients will then be eligible for treatment.5

Could the decisions in these guidelines be made more legitimate and fair? The recommendations about a treatment cut-off of CD4+ count <200 cells/μl is probably acceptable to many, but it should be acknowledged openly that this is based on both medical and ethical judgements. The choice was made on evidence available before 2003. In 2002, King and colleagues modelled the survival of patients receiving HAART (highly active antiretroviral therapy) in high-income countries. Patients had a relatively consistent 4–6-year increased survival benefit, regardless of when treatment was initiated below a CD4+ threshold of 350 cells/μl.28 Based on this model, one could argue that severity of disease guided the choice of cut-off among patients, with similar expected gain from treatment. Recent evidence on treatment effect indicates that timing of ART initiation affects treatment benefit more than was previously believed. Early start of ART reduces mortality more than late start, but it costs more.27 2932 The choice of cut-off is therefore clearly based on ethical judgements and this should be stated and explained.

Non-clinical rationing criteria were not included in the guidelines because the developers had the impression that patients did not come to the clinics in large numbers and they saw no need to narrow the range of treatment indications any further. Priority based on a first-come-first-served criterion is in itself an ethical choice. The lack of explicit decisions and rationales, both clinical and non-clinical indications for patient selection, opens the process to arbitrary decisions and fails to make reasons for decisions publicly accessible. The opportunity of using open discussion to make difficult choices, where critique and revisions are possible, is lost.

Some interviewees did not want to provide full rationales because they felt this would confuse users with unnecessary complexity. This could be solved by developing an additional short version of the guideline with clear recommendations.

This study is, to our knowledge, the first to provide a critical ethical analysis of the guidelines that determine who should live and who should die in this context of extreme resource scarcity. Even if all of us regard it as a gross injustice that there is such a resource scarcity, we also believe that allocating the scarce drugs that are available at any given time should be done in a way that is considered legitimate and fair.

Our findings are based on document reviews and interviews with a strategic sample of key informants. Although we believe we identified people with a range of different backgrounds and perspectives, the views they expressed are not necessarily representative of the whole group. One may also argue that focusing on guidelines is inappropriate and that what actually goes on in the clinics is more important. Although we appreciate this objection, national policies as expressed through national guidelines should, we believe, be the norm for assessing actual practice. It is therefore appropriate to analyse them as a first step.


The two National HIV treatment guidelines discussed are de facto mechanisms for rationing but were developed with a method that does not fully satisfy the requirements of fair process.


Thanks to all informants for giving valuable information. Thanks to Ingrid Miljeteig, Neema Sofaer and the qualitative methodology research group at the Department of Public Health and Primary Health Care, University of Bergen, for comments on earlier drafts.



  • Funding: Johansson and Jerene received funding from the University of Bergen, Norway, and Norheim from a Young Investigator Award, Norwegian Research Council, Norway. The funding sources had no influence on the study design.

  • Competing interests: None.

  • Ethics approval: This study involved interviews with guideline developers and official guideline documents and did not involve patients or sensitive information about patients and therefore did not require ethics approval. Oral informed consent was collected and the interviewees were informed about the purpose of the study. Confidentiality and anonymity were assured in analysis and reporting of statements.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Other content recommended for you