Final report is out! Evaluation of the Quality Improvement Support to Differentiated Care Models for Anti-Retroviral Therapy in Kenya

Jan. 27, 2021

Read the full report here

Since January 2018, iDSI has been working closely with the Global Fund to fight against AIDS, TB and Malaria (henceforth the Global Fund), the Kenyan Ministry of Health (National AIDS and STI Control Program -NASCOP- and the National AIDS Control Council -NACC) to assess the impact of implementation of Quality Improvement (QI) in support to differentiated care (DC) models on the quality of HIV care in Kenya. This was a unique project: the topic was scoped through lengthy consultations and was entirely led by local partners, the Global Fund and iDSI providing support on research methods, analysis and write-up. This modality of collaboration ensured that the study focus was aligned with future programming plans and strategic policy interests, and that the overall research being conducted with local ownership throughout.

Why is this important?

The study looks at the use of Quality Improvement to support the roll out of one of the most important changes in HIV/AIDs clinical guidelines in the country of the recent year. In 2016, the Ministry of Health issued comprehensive guidelines on Antiretroviral Therapy (ART), which introduced differentiated care (DC) pathways also known as “the DC ART model”. The DC ART model separated patients into different clinical groups demanding on their clinical needs, with specific differentiated patient pathways for each group so as to easily address their clinical needs. Prior to the introduction of DC, all patients were cared under one clinical pathway, e.g. whether they were stable or unstable (e.g. not virally suppressed or with a history of poor adherence). With the introduction of DC, stable patients (which represent > 75% of all patients) have fewer clinical visits and can pick up ART refills directly from pharmacies; freeing up resources (e.g., clinical staff time) to more intensive care for unstable patients with poor health outcomes.

There is a lot of interest and enthusiasm about DC globally, and it is seen as a means to achieve global HIV targets, especially in resource-constrained high prevalence settings[1]. However, there are anecdotal reports of barriers and challenges in implementation of DC pathways[2], including in Kenya where our study took place. As a result, the Kenya Ministry of Health and the Global Fund piloted early on a QI Program in support of DC implementation (with a focus on facilities where performance along HIV targets was not satisfactory). The intervention was implemented from December 2017 to May 2019, with 70 sites participating across 7 counties in the country. It followed a ‘Plan-Do-Act-Study’ process, whereby facilities diagnosed implementation barriers and developed with the support of NASCOP and other facilities local tailored solutions, testing change in real time. In addition, additional training and coaching on the new ART guidelines (including DC pathways) was provided in intervention facilities. There was some training to control sites, but it was far less intensive than in facilities receiving QI.

This is to our knowledge, the first comprehensive evaluation of a practical support program to DC pathway implementation. The study yields significant learnings for other countries looking to implement DC in their countries.

How was the study conducted?

The study followed a simple yet robust evaluation method relying on a single end-point evaluation between intervention and control facilities, identified through propensity score matching (PSM). PSM utilized data pre-intervention (June 2016) to ensure that intervention and control facilities were comparable prior to the implementation of QI. 70 intervention sites were matched with 193 sites based on facility level, volume of patients on ART, county epidemiological characteristics (e.g., HIV prevalence, population density) and proportion of patients with a viral load test in the last year (as an indicator of ‘performance’). Only the best 15 matches out of the 70 sites were selected for the study, for a final sample of 30 facilities. In all facilities, three survey instruments were used: (i) patient survey (coupled with chart abstraction for viral load and information on ARV regimens, weight, height etc.), (ii) facility costing tools (including time and motion survey administered to a sub-sample of patients) and (iii) provider survey (to measure satisfaction and knowledge of guidelines). Ethical clearance was granted by the Red Cross Ethical Review Committee.

Descriptive statistics were produced (means comparison between the two groups), and significance of the difference in means was tested using t-tests and Chi2 tests.

What have we learned?

Processes of care varied widely between facilities in both control (facilities implementing DC alone) and intervention (facilities implementing both QI and DC) groups. For instance, some facilities applied changes that improved service delivery, e.g. changes in opening times (e.g. ARV pick up organized from 6am) or appointment days (e.g., pediatric ART versus stable patient days) to implement DC pathways. Those innovations helped staff manage more complex care pathways and keep track of patient needs, although it is not clear whether they supported more effective implementation (not within the scope of this study). Another common theme across all sites was long waiting times for clinical appointments, especially in high volume facilities. Patients spent on average 92.8 minutes in facilities, and only 8.1 minutes were spent actively seeking services (e.g. clinical consultation, nutritional support, pharmacy time). Those results also support the overall rationale for DC: patients who were present for ARV pick up spent less than 25mins in total in the facility. This is in line with the global literature showing that DC models utilise health resources more efficiently, but may also align better with patient preferences given the otherwise long waiting times. It should be highlighted that waiting times in intervention facilities was lower than in control facilities.

Patient satisfaction with HIV services overall was high across all facilities. Patients were most satisfied with drug availability, and the least satisfied with waiting times. Intervention sites scored higher than control sites along the following dimensions: waiting times, convenience of the appointment, time spent with the clinician and observation of privacy. It is worth noting that while the intervention involved coaching on DC guidelines, we found no statistical difference in provider knowledge of DC guidelines between control and intervention groups, and patient satisfaction with provider technical skills. However, 82% of providers in intervention sites reported that the intervention helped them improve their knowledge of DC pathways very much. Patients scored also higher in ‘knowledge questions’ in intervention sites compared to controls: for instance, to the question “what do you do if you have forgotten your medication?”, 70.6% of respondents in intervention sites answered correctly compared to 58.7% in control sites.

There were notable differences in patient outcomes between control and intervention sites. Patients cared for in intervention sites reported a lower incidence of opportunistic infections (9% compared to 12% in control), higher rate of viral suppression (92.7% compared to 89.4%) and higher quality of life as measured by EQ5D 5L instrument. All those differences were significant at the 5% level. This points to the fact that DC+QI did improve processes of care at the facility level, which resulted in greater patient health.

All in all, the costs of implementing DC+QI versus DC alone amounted to 516 KES  (or less than 5 dollars) per patient per year (a health systems perspective was adopted for costing). Overall costs varied significantly across facilities (e.g. based on facility size) and across different patient groups (e.g. between stable and unstable patients). Almost half of the DC+QI implementation costs (43%) were spent on organizing and learning sessions. Another 39% was attributed to administration and overheads.

A research collaboration led by local actors

This research was entirely driven by NASCOP and NACC, and the results of the report were used to discuss funding and modalities for scaling up the intervention in other counties. This is an example of a collaboration with academic partners, donors, led by country partners to ensure direct application into decision-making, as well as building awareness and capacity around project evaluation and costing methods.

Read the full report here

[1] El-Sadr WM, Rabkin M, DeCock KM. Population health and individualized care in the global AIDS response: synergy or conflict? AIDS. 2016;30(14):2145–8.