iCCM program description

Community–based care existed in Ethiopia, Malawi and Niger focusing on mother and child and environmental issues, staffed with paid CHWs supervised by health centers staff. iCCM was added to their tasks. Ghana, Mali and Mozambique created or revitalized their CHW cadre. In Ghana iCCM is provided by volunteers while in the other countries CHWs are paid. The program was based on home visits (Ghana and Mozambique), combined with work from health posts (Ethiopia, Malawi, Mali and Niger). Table 1 presents additional country information.

Costs

This analyses combines a budget–holder/program (UNICEF) and health systems perspective. Health systems costs included government provider time, while program costs, included training, bicycles, kits and commodities, allowances for supervisors and review meetings and CHW stipends in Mozambique and Mali. Costing of consumables and supervision was based on clinical protocols and supervision schedules. We used actual data for number of CHWs, size of target population and number of treatments delivered.

Incremental financial and economic costs are calculated. Financial costs reflect additional expenditure incurred for the program by UNICEF. Economic costs covered all resources up to district supervision, including financial costs (although annualisation of capital is calculated differently) and opportunity costs (value of time spent on program). In Mali the stipend was fixed, independent of time spent. In Mozambique 70% of the stipend was costed reflecting the share of Malaria, Diarrhea and Pneumonia (MDP) treatments among under–fives since CHWs also treat adults. For Ghana volunteers, time value was based on a basic agricultural worker’s wage [12].

Budget holder costs were collected retrospectively from UNICEF country offices financial records for the CI/IHSS districts: bicycles/motorbikes, CHW kits and life span per type of equipment, training, salaries/stipend, supervisors allowances, CHW and supervisors attrition rate, malaria positivity rate and unit costs of drugs and rapid diagnostic tests (RDT). All drugs and tests were sourced from UNICEF, apart from malaria supplies in Mozambique, local cost of drugs were provided by UNICEF country offices.

Data were also obtained for the CI regions on the numbers of CHWs trained, frequency and duration of supervision visits, attrition rates, number of treatments per condition, and number of under–fives targeted, from UNICEF and CI/IHSS program documents.

CHW time

CHW time on iCCM was calculated as follows: 1) Average visit duration was set at 30 minutes (based on a previous costing study in Malawi [13]). 2) Travel time was added when iCCM was provided at household level. 3) An additional 20% of visits was added to reflect visits without treatment (eg, Malaria negative tests) while requiring CHW time; drawing on the malaria positivity rate, the share of malaria treatments in case load, and adding a small fraction for other visits without treatment 4) CHW time on supervision or community meetings and visits to the clinic for refilling of kits. 5) The above enabled to calculate the average time per week per CHW. It was assumed that a CHW works an average of 46 weeks per year. Time on iCCM was also used in the sensitivity analysis to assess time impact and feasibility of a 30% increase in demand.

Time line and analysis time horizon

iCCM costs are incurred in 3 overlapping phases. First is the design phase (formative research, design of the intervention, of the training and of materials). These, often substantial, one–off costs are not included because they will not be incurred again if iCCM is scaled–up. Second is the set–up phase (purchase of equipment, training) and the third is the implementation phase. Phases 2 and 3 are the focus of this study. Set–up data were collected for several years, and for 1 year for the implementation phase: the year 2012/13, the only year where iCCM was at scale in the 6 countries ( Table 1 ).

Analysis

Set–up costs were annualized, using a 3% discount rate for economic costs [14] and straight depreciation for financial costs. In the perspective of assessing cost of an on–going program, annualisation of equipment cost was based on the life span of each piece of equipment. Similarly, we annualised training costs, not as per length of intervention, but of the likely life span of training in an on–going program: initial training for CHWs and supervisors was allocated 10 life years when refresher training–mentorship took place, and 5 years in countries without refresher training. CHWs and supervisors attrition rates were applied. Incremental government and budget–holder economic costs are presented separately and combined ( Table 2 ). Costs are presented in US dollars 2015, updating 2013 local currency with local inflation rate [12], then translated into US$, using the 2015 local US$ exchange rate [15].

Fixed costs per CHW, independent of the number of treatments: annualised set–up and 1 year implementation costs were combined to calculate annual fixed costs per CHW: CHW iCCM training, equipment, CHW salaries/stipends and allowances for meetings, supervision and management costs (iCCM training for CHWs’ direct supervisors and district/zonal supervisors, share of supervisors’ salary package, allowances for supervision and meetings, and bicycles or share of motorbike costs), and overheads of 5% of the annualised costs. The annualised fixed costs per CHW divided by the number of treatments per CHW in 2012/13 represent “the consultation cost”.

Variable costs included drugs and RDTs used. Malaria positivity rate was used to calculate the number of RDTs used per child testing positive: with a 40% positivity rate, if 100 children are treated for malaria, 250 children have been tested; with a positivity rate of 60%, 167 children need to be tested. The average cost of consumables by treatment is weighted by the relative share of malaria, diarrhea and pneumonia treatments.

Financial costs per treatment are presented in Figure 1 , highlighting the share of consultation cost per treatment.

Figure 1.  Financial cost per treatment (2015 US$).

Affordability

We used financial costs to calculate the budgetary implications of the program: cost per CHW and per capita total population in study areas, which we then expressed as a percentage of the country 2013 public health expenditure (PHE) per capita (government + donors), as a proxy for affordability of the program. Data on health expenditure and 2013 total and rural population were obtained from the World Bank databank [16]. The number of under–fives in 2013 was extracted from the UNICEF statistical Tables [17].

Sensitivity analysis

We modeled the impact of a 30% increase in demand, keeping the ratio of children per CHW observed in the CI districts and assessed the implications for CHW time, consultation cost, cost per treatment and cost per capita. We then modeled the national program cost if the program was scaled up to all under–fives living in rural areas.

Country contexts

In Niger over 80% of the trained CHWs operated in 2013, in Mali the majority of CHWs were trained in 2011 and in Ethiopia, Ghana, Malawi and Mozambique the CHWs had only been functioning at that level for about a year ( Table 1 ).

Annual number of iCCM treatments per CHW ranged from 10 in Ghana to 603 in Niger, influenced by the number of under–fives per CHW, from 72 in Ghana to 632 in Malawi.

Time on iCCM was estimated between 3 hours a week or less per CHW in Ethiopia, Ghana, Mali and Mozambique to around 1 full day in Malawi and Niger ( Table 1 ).

Fixed costs per CHW and consultation cost

Annualised financial fixed costs per CHW ranged from US$ 811 in Mali to US$ 55 in Malawi, lower costs reflecting the pre–existence of community–based care with CHWs and supervisors salaries already paid by the state. Financial cost per consultation ranged from US$ 0.1 in Malawi to US$ 9 in Ghana. Annualised economic fixed costs per CHW ( Table 2 ) ranged from US$ 128 in Ghana to US$ 870 in Mali ( Figure 1 ), with an economic cost per consultation between under US$ 1 in Malawi and Niger with high utilization, and over US$ 11 in Ethiopia and Ghana.

Training costs are higher in Mali and Mozambique, with longer training (15 and 23 days respectively) compared to 6 days in other countries. Annualised equipment costs: bicycles, if relevant, and kits (excluding consumables) ranged from US$ 88 in Mali to US$ 13 in Ethiopia. Management and supervision cost represented a significant share of CHW fixed cost, from US$ 324 in Mozambique to US$ 51 in Ghana, largely reflecting the ratio of CHWs per supervisor ( Table 1 ) and supervisors’ daily allowance.

Cost per treatment

Cost of consumables per condition varied, partly due to differences in protocols, with use of additional drugs (Paracetamol), or in Niger, using the more expensive drug ASAQ for malaria treatment. Positivity rates for malaria varied from 40% to 67%. Drug pricing systems also differed: in Mali, drugs distributed from clinics are 50% more expensive than at national level (distribution costs and revenue for the clinic) ( Table 3 ).

Economic cost per malaria treatment ranges from US$ 3 in Malawi to US$ 14 in Ghana (financial cost US$ 2.4 to US$ 10.5 respectively), for diarrhea from US$ 1.3 to $12.9 (financial cost US$ 0.7 to US$ 9.3). User fees are implemented in two countries. In Mali patients pay US$ 0.2 per consultation and a weighted average of US$ 0.11 for drugs (excluding for malaria). In Ghana, patients contribute for drugs only, at a weighted average of US$ 0.34 per treatment ( Figure 1 ).

Affordability

Government paid from 5% of economic costs in Mali to under 20% in Mozambique and Niger, between 25% and 30% in Ghana and Malawi and 36% in Ethiopia ( Table 2 ), while 100% of financial costs were paid by UNICEF. Financial costs range from US$ 0.06 per capita total population in Ethiopia to US$ 0.74 in Malawi and Niger, representing under 0.6% of the 2013 PHE per capita in Ethiopia, Ghana and Mozambique, but 1.8% in Malawi, 2.7% in Mali and 7.4% in Niger ( Table 4 ), program cost per CHW ranges from US$ 101 in Ghana to US$ 2 047 in Niger ( Table 4 ).

Implications of increased utilization

With 30% higher utilization, time on iCCM would increase by 10% to just over 1 hour a week per CHW in Ghana, but in Niger would be 28% higher to 11 hours a week. A 30% increase in utilization could be absorbed by the existing CHWs. In Ethiopia, Ghana, Mali, and Mozambique economic and financial costs per treatment would be about 20% lower. In Malawi and Niger, where consultation cost represents a small share of cost per treatment, the decrease is small (2 to 4%). Total costs of the program increase, however, due to additional consumables ( Table 5 ). Financial costs of the program with higher utilization would remain under US$ 1 per capita.

Affordability of scale–up

If the program with 30% increased utilization was scaled–up to cover all the country under–fives living in rural areas, the cost per CHW would remain the same, iCCM would represent between 0.4% and 7.7% of the country PHE.

Limitations

This economic analysis has several limitations: first, in its scope: it focuses on resource implications of iCCM and affordability and does not include an analysis of value for money, a pre–requisite, but the programs were mostly too recent to assess effectiveness. Additionally, CHWs work was supported by multi–programs volunteers whose cost has not been included. Second, in its measurement methods: implementation costs were measured for 1 year, recent guidelines suggest that a wider window of implementation costs should be measured [27], however in this evaluation there was only 1 common year of implementation at scale. In addition, the normative approach used for calculating commodities and supervision costs has the benefit of estimating costs as per protocol, but does not reflect variations in actual implementation. CHW time on the program was based on the same assumptions of length of visit and meetings for all countries rather than on observation. However informal observation during country visits showed that our assumption of 30 minutes per visit is unlikely to be an underestimation. Another limitation is that the modeling of extending the program to all rural areas was made by extrapolating CI districts data and did not reflect other iCCM districts with potentially different cost profiles. Additionally affordability is based on per capita spending and does not include possible savings/costs to other levels of the health system as a consequence of the program in a way a Budget Impact Analysis would do [28].

A strength of this analysis is the separation of economic from financial costs to avoid double–counting in the calculation of budgetary implications. Second, rather than separating set–up costs from recurrent costs, annualised set–up and implementation costs are combined since over time equipment has to be replaced and training redone, and cannot be considered as one–off costs.