Menu
Menu
Menu
Menu
INTRODUCTION: In 2018, Mozambique’s Ministry of Health launched a guideline for a nationwide implementation of eight differentiated service delivery models to optimize HIV service delivery and achieve universal coverage of HIV care and treatment. The models were (1) Fast-track, (2) Three-month Antiretrovirals Dispensing, (3) Community Antiretroviral Therapy Groups, (4) Adherence Clubs, (5) Family-approach, and three one-stop shop models for (6) Tuberculosis, (7) Maternal and Child Health, and (8) Adolescent-friendly Health Services. This study identified drivers of implementation success and failure across these differentiated service delivery models. METHODS: Twenty in-depth individual interviews were conducted with managers and providers from the Ministry of Health and implementing partners from all levels of the health system between July and September 2021. National-level participants were based in the capital city of Maputo, and participants at provincial, district and health facility levels were from Sofala province, a purposively selected setting. The Consolidated Framework for Implementation Research (CFIR) guided data collection and thematic analysis. Deductively selected constructs were assessed while allowing for additional themes to emerge inductively. RESULTS: The CFIR constructs of Relative Advantage, Complexity, Patient Needs and Resources, and Reflecting and Evaluating were identified as drivers of implementation, whereas Available Resources and Access to Knowledge and Information were identified as barriers. Fast-track and Three-month Antiretrovirals Dispensing models were deemed easier to implement and more effective in reducing workload. Adherence Clubs and Community Antiretroviral Therapy Groups were believed to be less preferred by clients in urban settings. COVID-19 (an inductive theme) improved acceptance and uptake of individual differentiated service delivery models that reduced client visits, but it temporarily interrupted the implementation of group models. CONCLUSIONS: This study described important determinants to be addressed or leveraged for the successful implementation of differentiated service delivery models in Mozambique. The models were considered advantageous overall for the health system and clients when compared with the standard of care. However, successful implementation requires resources and ongoing training for frontline providers. COVID-19 expedited individual models by loosening the inclusion criteria; this experience can be leveraged to optimize the design and implementation of differentiated service delivery models in Mozambique and other countries.
The study was conducted from July to September 2021. Participants were from MISAU and implementing partners at all levels of Mozambique’s health system—national or central, provincial, district and health facility. National‐level data were collected in Maputo city, the country’s capital. Sofala province, a setting with high HIV prevalence and HIV treatment demand, including a nationally recognized HIV transmission hotspot—the Beira corridor [18], was purposively selected for data collection at subnational levels. In Sofala province, two districts (one rural and one urban) and four health facilities (one small and one large in each district) were selected. Health facilities are defined by the National STI‐HIV/AIDS Programme as small when they have less than 1000 clients enrolled in HIV treatment services, and as large otherwise. The eight models being studied are described in Table 1. Differentiated service delivery models for HIV treatment implemented in Mozambique ‐ Can be implemented in isolation or combined with 3M. ‐ ARV dispensed quarterly when combined with 3M and monthly in health facilities without 3M. Members take turns visiting the health facility for clinical observation; all members must be observed and have lab tests done at least twice a year. ARVs for all group members are dispensed monthly to the group member who visits the health facility. This model requires additional staff for activities coordination and implementation. ARV dispensing depends on the ARV stock in the health facility. This model requires additional staff for activities coordination and implementation. All the appointments of the family members are schedule for the same day. The frequency of visits depends on the existence and age of children and the clinical condition of all members of the family, and can be monthly, quarterly or twice a year. All HIV services are offered in the AFHS sector of the health facility. Clinical observation depends on the client need. ARV dispensing depends on the ARV stock in the health facility. We applied purposive sampling to include at least 9–17 key informants so to satisfy the estimated minimum sample size to achieve code saturation of 90% [19, 20]. The eligibility criteria were involvement on DSD models’ management or implementation at each level of the health system, for both MISAU and implementing partners. Semi‐structured, in‐depth interviews were conducted with selected participants, including HIV programme managers from the national, provincial, district and health facility levels, and providers at the health facility level. The interviews were conducted in Portuguese, using a semi‐structured interview guide that was developed based on purposively selected constructs from the Consolidated Framework for Implementation Research (CFIR) by Damschroder et al. in 2009 [21, 22]. Questions included the perception of barriers and facilitators in general and by selected CFIR constructs, for the intervention overall and for each model individually. Interviews were audio recorded and transcribed verbatim. CFIR is a deterministic framework developed from previous frameworks and relevant theories in various disciplines [23], and is organized into five domains and 39 constructs (including subconstructs) [21]. We chose to use the CFIR given its pragmatic structure to study real‐world implementation and its applicability to guide data collection and analysis, as well as to contextualize the findings [24]. Constructs for this study were selected based on a literature review of known barriers and facilitators for DSD model implementation in sub‐Saharan Africa. Fifteen constructs from all five framework domains were included: (1) Relative Advantage, (2) Adaptability, (3) Complexity, (4) Design Quality and Packaging, (5) Cost, (6) Intervention Source, (7) Client Needs and Resources, (8) Implementation Climate, (9) Readiness for Implementation, (10) Knowledge and Beliefs About the Intervention, (11) Other Personal Attributes, (12) Planning, (13) Engaging, (14) Executing, and (15) Reflecting and Evaluating [10, 14, 15, 16, 17]. We conducted a thematic analysis using an iterative deductive–inductive approach. For the deductive analysis, we used an initial list of codes created based on the 15 pre‐selected CFIR constructs. Emerging themes (both non‐CFIR and CFIR constructs) were added to the initial list and used to code subsequent interviews. Coding was conducted on the original interview transcripts in Portuguese, using ATLAS.ti software, version 9. Two investigators (OU and AD) coded each interview transcript independently using the initial codebook and added new codes to it as they emerged from the data. A third investigator (DMU) reviewed the work of the initial coders and identified new codes and disagreements. To achieve consensus on coding, the three investigators reviewed the disagreements and when consensus was not met, two investigators (SG and CI) acted as tiebreakers. CFIR construct codes were rated as a function of valence and strength. Valence is the directional (positive or negative) influence of the construct on DSD implementation, and is marked by “–” for negative influence (i.e. a barrier), “+” for positive influence (i.e. a facilitator), “X” for mixed negative and positive influence, and “0” for a neutral code. Strength was determined by factors such as the level of agreement across participants, strength of language and use of concrete examples. The number “1” indicated weak influence and “2” strong influence in the intervention. The symbol “*” denoted the level of agreement among participants, meaning that comments were mixed—both positive and negative—and the attributed rating is an aggregated result [25]. Table 2 summarizes the code rating system. Coding rating system This study was approved by the Mozambique National Ethics Committee (634/CNBS/20) and the University of Washington institutional review board (FWA#00006878). Written informed consent for interviewing and recording was obtained from interviewees before all interviews. The names of participants and health facilities are concealed for ethical reasons.
N/A
