Cost differences between program approaches
The most important finding of this study is that the unit cost differences between the horizontal and diagonal program approaches are modest, $38.62 for APHIA II and $44.62 for NRHS. Ninety percent of the adult MCs conducted during the study period were through the NRHS diagonal approach and 10% through the APHIA II horizontal approach. NRHS’ greater share of the total is due to the use of dedicated (full-time) MC teams that contrasts with non-dedicated (part-time) MoH teams providing MCs during only 12–38% of their time in the APHIA II approach. NRHS also deployed 2–9 dedicated teams per district, compared with 2–3 non-dedicated teams deployed by the MoH with APHIA II support. When the surgery sessions are underway, the teams work at capacity, but staff wait times, particularly before the start of the surgical sessions, are substantial. The unit cost figures we report include the cost of all MC-related staff time including wait time, and thus capture these inefficiencies.
The larger NRHS service volume may suggest that the diagonal NRHS approach can be scaled up more quickly in the short term and can increase service volumes over time. However, long-term operational challenges may emerge as it seeks to achieve fuller integration with existing MoH adult MC services. The horizontal APHIA II approach, while producing consistently lower service volumes, has also demonstrated rising service volume. The NRHS strategy may be more suited for rapid clearing of the unmet demand for MC, whereas the APHIA II strategy, since it has been more thoroughly integrated with the Kenyan health care system, may be more suited to serving the smaller volume of ongoing new cases. Of Kenya’s provinces, Nyanza experienced the highest increase in male circumcision rates, between the 2007 and 2012 Kenya AIDS Indicator Surveys from 48% in 2007 to 66% in 2012, and thus may approach this more routine caseload demand over the next few years.
Given the shortage of health workers in Nyanza across all cadres [
19], shifting resources in the direction of either approach in the short term is no substitute for long-term investments to increase the health care workforce. Hiring should be done in a way that ensures the right balance of non-dedicated and dedicated MC workers. Disparities in health worker wages can potentially contribute to inequities through internal migration of health workers to MC from other important health services and from one geographic area to another. Disparities in wages may also contribute to a lack of motivation among existing MoH staff in the absence of MC-specific incentives. Decisions regarding which MC program approaches to emphasize should consider human resource shortages, cost, and long-term sustainability within the aims of national health policies and strategies.
Unit costs by service delivery mode
In assessing unit costs by mode, a more complex picture emerges. Unit costs for adult MCs delivered at base facilities are very similar, $38.33 and $39.58 at APHIA II and NRHS, respectively. This is likely due to the fact that the majority of NRHS-supported base sites are MoH sites. The difference between outreach at APHIA II and combined outreach/mobile at NRHS sites is also modest, $40.51 vs. $46.20, a difference of 14.1%. However, the difference in unit cost between mobile MCs supported by APHIA II and outreach/mobile services supported by NRHS is more substantial, 36.5% higher in the latter. Part of this difference can be explained simply: Compensation for direct service providers at the APHIA II-supported sites is 45% of the level of equivalent staff compensation at NRHS. If the cost of direct service personnel at NRHS were reduced accordingly, the difference in unit costs for this portion of field activities drops to 22.1%, and the direction reverses: $40.51 at APHIA II–supported sites versus $37.61 at NRHS sites. Thus, it is hard to explain differences in unit costs by factors that are inherent in the relative virtues of a horizontal versus diagonal approach. On balance, we believe that higher efficiencies are more likely to be attained by adjusting the way MC activities are implemented within service delivery modes in either approach, than by attempting to select one broad approach as generally more efficient than the other.
Placing the results reported here in a broader context, the unit costs are of the same general magnitude as those reported elsewhere in the MC cost literature. The Futures Group has empirically estimated the unit cost of MCs in various African settings at $35–$50 [
5]. Other modeling of adult MC scale-up in 16 geographic areas estimated an average of $168 per HIA and 5.6 MCs per HIA, thus implicitly $30 per MC [
13]. Finally, a study of high-volume circumcision surgery at a fixed facility in Orange Farm, South Africa found that the procedure could be performed for an average of US$40, and that the procedure required 20 minutes (versus 29.5 for combined APHIA and NRHS at base facilities) including 7.5 minutes of the surgeon’s time (versus 16.3 for combined APHIA and NRHS). The reported cost was similar to the $38.33 and $39.58 per procedure we found for APHIA-II and NRHS, respectively [
14]. However, these costs are not directly comparable since the Orange Farm estimate includes direct services and the rental and maintenance of the surgical space only. It excludes the cost of training, outreach and of overhead and administration. On the other hand, medical personnel salaries are higher in South Africa than in Kenya, and adjusting for this difference would move the unit cost estimates toward convergence. The shorter periods of total time for the procedure and for the surgeon’s time in the Orange Farm facility may be due to the different organization of the surgery including the use of disposable kits, and electrocautery which saves suturing time, the single most time-intensive part of the procedure. Another important study of MC costs in Zambia [
20] was used as the basis of an MC scale-up modeling exercise carried out by the Futures Institute [
4]. The data from Zambia suggests a unit cost of US$46.82, somewhat higher than the findings we report for Nyanza.
The observed variation in unit costs for these MC programs in Nyanza must also be considered in the context of far wider unit cost variation observed previously in HIV prevention programs. In our five-country study of 215 HIV prevention programs (the Prevent AIDS Network for Cost-Effectiveness Analysis [PANCEA] Project), we found variations in unit cost of 10- to 100-fold within a range of prevention strategies and countries [
21]. These differences mainly represented variations in the number of delivered units of service, accompanied by some variation in the intensity of service per client, with relatively fixed personnel and other input costs. By comparison, the differences between adult MC approaches in Nyanza are small and are largely explained by variations in salaries. The roughly similar cost may reflect multiple homogenizing factors, including standardization of the service content; communication and coordination among the MC partners; and similar motivations by both APHIA II and NRHS program managers to try to optimize performance.
Further refinements in the staffing and logistical organization of the adult MC procedure itself may yield only modest gains in efficiency. This is because the marginal cost of supplies and personnel for each procedure is a small portion of the total unit cost. Yet, a dollar deducted from costs represents resources that can be freed to expand services, whatever the source of that savings. We therefore support further operations research into the possibility of streamlining the surgical procedure and immediately proximate activities. However, our data suggest that once programs have trained lower-cost surgical staff, large further reductions in costs must be sought elsewhere. The areas described in the accompanying Additional file
5 (Five Areas of Possible Efficiency Improvements for MC Delivery) include reduced staff wait times, trimmed overhead and more efficient scheduling of surgery days. Of these, scheduling and administrative efficiencies appeared most likely to yield a substantial reduction in cost per MC and thus per HIA. Operational efficiency (reducing start-up time on MC days) appears to offer smaller gains. Gains in technical efficiency through the Shang Ring and electro-cautery appear unavailable given their current costs and the relatively low cost of the labor and supplies they displace.
The measure of increased efficiency should be placed in this broader context of cost-effectiveness, using the cost per HIA metric, as it takes into account the possible trade-offs between potential economies and the number of MCs delivered.
Study limitations
This report is limited to retrospective data from 222 MC service delivery sites over an 18-month period and prospective T&M data from 246 procedures. While these are sufficient to document unit costs and their variations, they are insufficient to support a robust multivariate analytic approach that might more definitively identifies the correlates of efficiency. Further, NRHS costing data for outreach and mobile services were unavailable in a disaggregated form, making only rough comparisons between the NRHS and APHIA approaches for outreach and mobile service delivery modes possible. The results might be quite different if mobile and outreach modes were separated for the NRHS approach. The T&M data were limited to a two-month period, and it is possible that seasonal variations in caseload or other factors captured by longer data collection period could affect the estimates of incremental per-case costs. The unit costs we reported are associated with the caseload shown in Figure
5. Since HIV intervention programs display economies of scale [
21‐
23], if demand declines, unit costs could rise. This could occur if, for example, the most willing clients having already been served, and more money is needed for outreach to maintain the caseload size. Finally, we did not have data on the incidence of adverse events, though the cost of treating adverse events is implicitly captured in the personnel time and cost calculations.
The observed unit costs for adult MC programs in Nyanza, while likely to be similar in other provinces in Kenya, must also be considered in the context of far wider unit cost variation observed previously in the region. In particular the reported personnel costs are most relevant to other countries in which there is a clinical officer cadre. Personnel costs will be different in countries in which medical officers are required to be part of MC surgery. Moreover, MC programs using techniques other than the forceps-guided method may have different unit costs. The findings of this study, while not directly generalizable to other countries where MC programs are being implemented, offer insights into expected efficiencies and cost-effectiveness where similar types of program approaches and service delivery modes are being implemented.