We used data from 216 facilities treating 49,871 HIV-infected patients in Côte d’Ivoire to assess the availability of human resources at HIV prevention, care and treatment facilities and to understand associations between HCW distribution and critical HIV-program outcomes. To our knowledge, this is the first evaluation conducted in SSA to examine the relationship between HCW availability at health facilities and HIV-program outcomes. While more HCWs worked at high ART volume health facilities as compared to ART low volumes facilities, after accounting for patient numbers, we found low volume facilities to have substantially higher staff to patient ratios. This difference was most pronounced in the clinical cadre, where low volume sites had over 2.5-fold higher clinician-to-patient ratios. A previously conducted study in Kenya reported that the maldistribution of clinicians undermined ART service delivery as thousands of HIV-positive individuals especially those in rural areas did not have access to ART [20]. Our analysis demonstrates that policy makers could make data driven decisions regarding the allocation of health workers at ART-facilities based on their patient volume.
Our results show that there were proportionately more lay workers in the low volume health facilities. There has been a rapid emergence of a large number of lay workers across ART programs in SSA [21]. This cadre provides support for HIV testing, initiation on ART, adherence counselling and nutrition support, contacting patients via telephone for appointment reminders, and tracking loss to follow-up patients to bring them back to care [22, 23]. We found this cadre to be associated with a greater number of HIV infected individuals identified, and a higher number of these patients being initiated on ART. Our findings are consistent with previous research which have demonstrated that the use of lay workers increases HIV testing results. A study in Malawi reported that the use of HIV Diagnostic Assistants (a cadre of lay health workers) was associated with a 70% increase (from 28% preintervention to 98% postintervention) in the number of individuals tested for HIV and the identification of HIV-positive cases [24]. Our finding appears to validate the investment in this cadre that has been made by donors, such as PEPFAR in the last decade. Like many countries in SSA, the shortage of clinical HCWs at government sites in Côte d’Ivoire has exerted a huge burden on the remaining HCWs who provide HIV-related services to clients. As a result, lay workers have been recruited and trained to provide these essential services at the facility and community level. Côte d’Ivoire is not alone in this effort, data from Zambia suggested that over 70% of HIV testing services are provided by this cadre[25] and similar findings have also been reported in South Africa and Namibia [26, 27].
Task shifting to lay workers has become a cornerstone of decentralization of ART care [28, 29], and is increasingly recommended as a cost-saving measure to scale-up HIV service delivery in countries across SSA [30]. With support from PEPFAR, many countries have recruited and trained lay workers to initiate patients on ART in the facility or community level, monitor patients and support treatment adherence, thereby achieving better treatment outcomes [31]. In fact, a Cochrane review conducted in 2014 found that shifting the responsibility of HIV care from doctors to nurses or lay workers is safe, and may even improve follow-up of patients after ART initiation [32].
We also found the laboratory cadre to be associated with a higher number of HIV-positive cases identified and a higher number of these cases initiated on ART. Access to laboratory services and the capacity of those services to provide timely results is central to the success of the HIV care cascade [33,34,35]. It seems plausible that investment in onsite laboratory capacity is associated with improvements in the number of HIV cases identified. In addition, with the very high rate of test-positive patients being initiated on ART observed during this evaluation, this relationship was translated to the ART initiation outcome. We did not find the clinical or pharmacy categories to be strongly associated with any of the outcomes. We might have expected these cadres to have greatest impact on the proportion of new cases who were initiated on ART. Despite wide variation in the clinician to patient ratio across sites, we saw universally high rates of ART initiation. This finding suggests that clinicians and pharmacists in the program may have been able to achieve good patient outcomes even when these cadres were relatively understaffed.
We did observe that once initiated on ART, patients at low volume facilities were slightly less likely to achieve viral load suppression. Clinic-level variation in the number of HIV cases identified and the number of patients started on ART was almost entirely explained by the clinics overall patient volume and its human resources. Conversely, there was very little relationship between patient volume and staffing with the proportion of patients achieving viral load suppression. Côte d’Ivoire is a low-prevalence setting, and previous research conducted by HAI found that retention in HIV-care (which is critical in viral load suppression) is determined by stigma; the quality of previous interactions with medical services; and structural barriers, including costs, waiting times, and availability of medicines [36,37,38]. We had expected that a higher number of health workforce reduce barriers to care at a site, and thus translate into improved viral load suppression proportions. It is unclear why none of the HCW categories was associated with viral load suppression in our data. This finding may suggest that patient-level factors, ART and clinical factors as well as social and behavioral factors are more likely to influence adherence to ART and viral suppression than site-level human resources [39,40,41]. Overall, it is important to have multidisciplinary care teams within health facilities providing HIV care and treatment services. These teams made up health professionals (e.g., clinicians, nurses, pharmacists, laboratory technicians, etc.) and other lay cadres (e.g., HIV Diagnostic Assistants, Community Counsellors, etc.) should collaborate in the design and implementation of HIV activities in a health facility as this may contribute to positive results.
The findings of this analysis should be interpreted bearing in mind a few limitations. First, because of the observational design of this evaluation, cause-and-effect relationships cannot be established. Second, human resource categories aggregated multiple types of healthcare workers; therefore, we cannot differentiate specific health personnel (e.g., doctors, midwives, nurses etc.) and their contribution to HIV-program outcomes. Future evaluations should disaggregate the health worker personnel. Third, each category was assessed at a single timepoint, which may misrepresent fluctuations in resources across time. We also cannot account for cadres cross covering each other’s tasks, for example lay workers may shoulder clinical or managerial responsibilities, where there are deficits in those cadres. We aligned our outcome measures with PEPFAR performance metrics critical for achieving the UNAIDs 95-95-95 target. These outcomes cannot be exclusively charactered as either metrics of healthcare access or healthcare quality, but are a product of both access to service and quality of care. Finally, we may have unaccounted for external confounding factors (e.g., incentives, training, quality of services, equipment, medical supplies etc.) which may have contributed to the achievement of the results but were not examined in this analysis. These unmeasured confounders could have attenuated or exaggerated the observed association between HCWs and program outcomes. One important unmeasured confounder is the proportion of HCWs employed by the MOH vs PEPFAR partners, the latter group may be incentivized to focus more on the care of people living with HIV, which may have amplified the association between PEPFAR cadre levels and outcomes.