This paper has provided a descriptive analysis of HR integration and staff workload across 40 facilities in Kenya and Swaziland. Our results indicate that HR integration was more likely to be improved in facilities which also improved other elements of integration, such as availability of services within the MCH unit and integrated use of physical space. Furthermore, while we there was no overall relationship between integration and workload at the facility level, certain services had a significantly lower workload ratio in more integrated facilities. These results suggest that HIV/SRH integration may be most influential on staff workload for PITC, PNC and STI services - particularly where HIV care and treatment services are being supported with extra staffing. Our findings therefore suggest that there may be potential for improving use of excess capacity through integration, but that overall the pace of increase in productivity is slow.
Our findings should be viewed in the context of a number of studies that have highlighted the urgent need for improving staffing and skill mix, especially in light of the demands on health services in the context of the HIV epidemic [14, 20, 28]. However, our findings are in line with excess capacity found in a number of settings [29–32]. The differences we found in workload between different services within facilities imply that this under-utilization of staff is not solely a site location-related issue, and can be improved through re-allocation of staff duties across services within sites. Similar imbalances in allocation of staff duties within facilities have been found in Uganda  and South Africa .
However facility-level workload ratios also varied substantially between facilities, from 1.18 to 28.05 at baseline and from 1.07 to 26.08 at endline. No facilities were found to be ‘overworked’ at baseline or endline given our assumptions and calculations. This was observed despite an overall increase in the number of HIV/SRH outpatient visits for most services over the study period, as in both countries there was a corresponding increase in staff time available. Some of these increases, in particular increased staffing of HIV-related services, may have come at the cost of reductions of staff available for other services such as PNC, and lead to greater imbalances in staff workload within a facility.
Although integration may be a route to streamlining HR use on the whole, integration was not achieved uniformly across facilities in either country, suggesting implementation challenges. HR integration was positively associated with other aspects of integration, including service availability and physical integration of services, suggesting that various aspects of integration go hand-in-hand. For example, investments in physical infrastructure and drug/supply availability are perhaps required before proceeding with HR integration. We therefore recommend that efforts to integrate should be preceded by some investigation at the facility level regarding capacity to integrate services. Moreover, planners should be flexible in design, as there is no ‘one-size fits all’ solution to integrating.
There are some indications from qualitative work that integration can lead to an improvement in quality of care, as providers may be able to deliver a more well-rounded service to patients . However, in other cases, high workloads can negatively impact service quality . Although our workload estimates indicate excess capacity this does not mean that staff are not under stress at all periods in the day. This is especially true in settings where client load is higher in the mornings, with demand falling in the afternoon. In this context, we therefore recommend careful monitoring and evaluation of workloads as integration progresses on a site-by-site basis.
It also should be noted that integration of HIV services into primary health care or other services typically delivered by lower-paid cadres (for example, nurses) has been proposed by some as a vehicle to reduce the reliance on higher-paid members of staff (such as clinical officers or doctors), enabling them to deal with more complicated cases [35–37]. However, care has to be taken, as at some lower-level health facilities there may not be sufficient demand for these more complex services. For example, in many cases, we observed the workload for SRH services to be much higher than that of HIV care and treatment, or other outpatient services typically delivered by a clinical officer or other higher-paid staff members. There is a risk that integration could lead to further overworking those lower-paid staff members, suggesting that integration efforts should be placed within the broader picture of HR planning and management. This may mean that additional staff training is required at the facility level in order to ensure that integration through task shifting does not result in a reduction of service quality.
Finally, the above findings and interpretation need to be seen in the context of a number of limitations. Although this is the largest study to date evaluating integration, we were limited in any multivariate analysis by the relatively small sample size. Furthermore, the nature of time use data is such that there is a risk of bias in all staff time observations . Workload ratio estimates are relatively sensitive to the staff time observed per service. We tried to minimize bias through triangulation of several data collection approaches; however, it is possible that we have under- or over-estimated the time taken to deliver services. Similarly, as our utilization data was collected retrospectively from routine data registers, it is possible that services were delivered but not recorded in some facilities; this would under-estimate utilization, and thus workload. This under-reporting may not be consistent across service types. Of particular concern is STI utilization data, which was not consistently recorded across facilities. In addition, all staff time observations used in our calculations were recorded at the endline of the Integra study. Although staff were observed at baseline, observations were not systematically recorded at each of the study facilities, largely because few facilities consistently provided integrated services at baseline. We also did not account for the time differences between first visits and revisits for services such as FP or PNC, where a first visit can take much more time than a follow-up visit. For these particular services, we observed a large number of consultations and, therefore, capture some of this timing variability within our calculations. Finally, our recommendations are made from the provider perspective; and we take little account of how patients may value different visit times and experience facilities with relatively high workloads .