Perspectives of key stakeholders regarding task shifting of care for HIV patients in Mozambique: A qualitative interview-based study with Ministry of Health leaders, clinicians, and donors

Background: Task shifting is a common strategy to deliver antiretroviral therapy (ART) in resource-limited settings and is safe and effective if implemented appropriately. Consensus among stakeholders is necessary to formulate clear national policies that maintain high-quality care. We sought to understand key stakeholders’ opinions regarding task shifting of HIV care in Mozambique and to characterize which specific tasks stakeholders considered appropriate for specific cadres of health workers. Methods: National and provincial Ministry of Health leaders, representatives from donor and non-governmental organizations (NGOs), and clinicians providing HIV care were intentionally selected to represent diverse viewpoints. Using open- and closed-ended questions, interviewees were asked about their general support of task shifting, its potential advantages and disadvantages, and whether each of seven cadres of non-physician health workers should perform each of eight tasks related to ART provision. Responses were tallied overall and stratified by current job category. Interviews were conducted between November 2007 and June 2008. Results: Of 62 stakeholders interviewed, 44% held leadership positions in the Ministry of Health, 44% were clinicians providing HIV care, and 13% were donors or employed by NGOs; 89% held a medical degree. Stakeholders were highly supportive of physician assistants performing simple ART-related tasks and unanimous in opposing community health workers providing any ART-related services. The most commonly cited motives to implement task shifting were to increase ART access, decrease physician workload, and decrease patient wait time, whereas chief concerns included reduced quality of care and poor training and supervision. Support for task shifting was higher among clinicians than policy and programme leaders for three specific task/cadre combinations: general mid-level nurses to initiate ART in adults (supported by 75% of clinicians vs. 41% of non-clinicians) and in pregnant women (75% vs. 34%, respectively) and physician assistants to change ART regimens in adults (43% vs. 24%, respectively). Conclusions: Stakeholders agreed on some ART-related task delegation to lower health worker cadres. Clinicians were more likely to support task shifting than policy and programme leaders, perhaps motivated by their front-line experiences. Harmonizing policy and programme managers’ views with those of clinicians will be important to formulate and implement clear policy. Data were collected in semi-structured, one-on-one interviews that were conducted with individuals intentionally selected by two individuals (K.S. and a senior Ministry of Health manager) to represent the key stakeholders in the development of MoH policies for the provision of HIV care in Mozambique, including (1) MoH leadership at the national and provincial levels, (2) health sector representatives of donors and NGOs at the national level, and (3) HIV clinicians providing HIV care and treatment in six public hospitals that had pioneered ART delivery in Mozambique leading up to, and during, the initiation of the national HIV Care and Treatment Program (between 2002 and 2004). For the MoH leadership at a national level, we mapped out the key individuals based on leadership positions and/or programmatic relevance (i.e., heads of programmes germane to each topic). All but two of the recruited individuals consented to be interviewed and were interviewed. For MoH staff at a provincial level, we selected provinces from the north, south, and central regions (two provinces per region) and targeted the provincial chief medical officer and head of the HIV programme; those present on the day of the interviews were interviewed. For hospital staff, we targeted MoH-run health facilities providing ART and selected purposively to overlap with the provinces we were visiting to interview MoH provincial staff. We attempted to interview clinical leadership and a non-random sample of clinical staff in these facilities based on availability, until saturation was reached. For leaders of NGOs and donors, we listed the key donors and NGOs in Mozambique at the time and reached out to the individuals involved in policy development, support for public HIV care, or provision of HIV care (one individual per organization). There were no exclusion criteria for any of the groups. The study instrument (Additional file 1) was developed to include both closed- and open-ended questions, with the intent of characterizing respondents’ perceptions regarding task shifting in general and regarding the shifting of specific ART-related tasks to lower-level health cadres. Interview questions were tested for understandability and contextual appropriateness in a small number of individuals (representing national and provincial leadership, NGO/donors, and clinicians, reflecting similar backgrounds to respondents) before interviewing respondents. All interview questions were developed in Portuguese, all respondents spoke Portuguese, and all interviews were conducted in Portuguese. Interviews were not recorded. All interviews were conducted in the private offices of interviewees or in private spaces in HIV clinics. Notes were taken in Portuguese by the data collection team. The data collectors recorded responses to open-ended questions as summary statements. If the question or answer was not clear, the interviewers repeated the question, or repeated their interpretation of the respondent’s answer, to ensure that the message was mutually understood. Interviewers collected data on interviewees’ demographics, education, and employment using closed-ended questions. Attitudes about the delegation of tasks to different levels of health-care workers were ascertained using both closed- and open-ended questions. For each of seven cadres of health workers in Mozambique (Table 1), interviewees were asked whether the given cadre should perform each of eight tasks related to the provision of ART (yes/no): initiate ART in adults, pregnant women, adults with tuberculosis, or children; follow adults or children after ART initiation; change ART regimen in adults or children. These tasks were chosen based on our experience, because they would be more representative of key steps to expanding and sustaining ART coverage. In presenting results, we ordered tasks from simple (e.g., following patients in whom ART has already been initiated) to complex (e.g., changing ART regimen in a child), based on the proportion of participants who thought that a given task would be appropriate to delegate to lower cadre health workers. Finally, interviewees were asked to give up to three benefits and up to three risks of delegating ART tasks to physician assistants and up to three suggestions to ensure the quality of services provided by non-physician health workers. All interviews were conducted by one of three people and took place in person between November 2007 and June 2008. Relevant cadres of non-physician health workers in Mozambique, by training level Abbreviations: SMI, saúde maternal e infantil; MCH, maternal and child health. aInformal cadre. The frequencies of responses to closed-ended questions were tabulated overall and by employment of the respondent (policy-maker/clinician/donor or NGO worker) using simple percentages. Tables presenting the percentage of respondents who thought that a given health worker should perform a given task were colour-coded to highlight the direction and magnitude of agreement among respondents. Ranges for the colour-coding were chosen to highlight near-universal support (≥90% in darkest green) or opposition (<10% in darkest red) of the use of specific cadres to perform specific tasks; the width of the ranges was slightly larger for middle-level agreement (e.g. 40% to <60% was colour-coded white). Open-ended answers were analysed thematically to identify the most commonly cited responses. This content analysis focused on the most common answers, though all responses were tabulated and included in the results. We chose not to report p values because the goal of this study was not to test a hypothesis or to draw inferences beyond the study sample, but rather to describe major themes in the attitudes of stakeholders. All analyses were conducted using Stata, version 13.1 (College Station, TX). This study was approved by the Institutional Review Boards at the University of Washington and the Ministry of Health of Mozambique. All respondents provided written consent to study participation prior to being interviewed.