Study setting

The study was conducted as part of the Ethiopian HEP’s national assessment of the relevance of the HEP packages and the provision of services [9]. Both primary and secondary data were used. Primary data were collected from 300 health posts, 54 health centers and 57 district health offices from June 7 to July 1, 2019.

Interventions

The following 21 interventions were selected based on the availability of cost data, effectiveness measures, and interventions, mostly implemented by HEWs (Table 1).

Table 1

Selected HEP interventions from the family health services, disease prevention and control, hygiene and environmental sanitation subdomains.

Family health services (preventive)

1.1 Maternal and child health

Antenatal care

Tetanus toxoid vaccinations

Iron folate supplementation

Family planning: oral contraceptive pills (OCP), condoms, injectables and implants

1.2 Expanded program of immunization (EPI)

Pentavalent vaccination

Measles vaccination

Pneumococcal vaccination

Disease prevention and control (curative services)

2.1 Human immunodeficiency virus (HIV) testing and counselling

HIV testing and counselling

2.2 TB prevention and control

Directly observed treatment short course therapy (DOTS)

2.3 Malaria prevention and control

Insecticide treated bed nets (ITN)

Indoor residual spray (IRS)

Malaria treatment

2.4 Diarrheal disease management (oral rehydration salt (ORS) and Zinc)

2.5 Pneumonia treatment (i.e. cotrimoxazole, amoxicillin, gentamycin)

Hygiene and environmental sanitation

3.1 Improved water source

3.2 Hand washing with soup

3.3 Hygienic disposal of children’s stool

3.4 Latrine use

Costing approaches

Costing was conducted from a provider perspective, where the costs incurred by the government were included [10]. The bottom-up ingredient and top-down costing methods was applied. The bottom-up ingredient costing method specifies activities under each intervention, measures the resources used for the activities (i.e., expenditure), and assigns prices to the resources. The top-down costing method considers overall spending at a central level to allocate some of the costs to each intervention [11]. All cost inputs from both methods were crosschecked and accounted for to avoid double-counting. The cost items that last for more than one year were treated as capital and those used within a year were considered recurrent. The capital inputs include pre-service and in-service training, the building of health posts, and equipment. The recurrent inputs include supplies, personnel, supervision, and review meetings.

Cost inputs and analysis

The cost components include personnel, medicine, supplies, infrastructure, capacity building, and equipment (Table 2). To estimate personnel costs, the number and salaries of HEWs with level 3 and 4 career certifications were obtained [12]. The average HEW salary per minute was then calculated based on 8 working hours per day and 22 working days per month. The average HEW salary per minute was multiplied by the average working time HEWs spent on each service (i.e., a time motion study) to obtain personnel costs per service [13]. The total supply costs per service include the cost of items, the pharmaceutical management supply cost, and vaccination wastage. The cost of supervision and review meetings at the national and sub-national levels was estimated per health post per year. Health posts are made from hollow brick, stone, and wood, with an average service life of 30, 20, and 15 years, respectively. The average cost of the three types of buildings was calculated and subsequently annualized. Similarly, 5 life-years for equipment, 10 for pre-service training, and 5 for in-service training were considered. The capital and recurrent costs (except personnel) were distributed to each service using the total number of people served per health post as an allocation base. The cost per intervention was then estimated by adding the unit costs to the recurrent and capital costs.

Table 2

Description of input costs and source of data.

Costs inputs	Description	Source of Data
Building	Average cost of health post made of hollow block, wood/mud and stone.	Primary data
Equipment	Number of functional equipment per health post.	Primary data
Pre-service training	Cost of pre-service training per year.	Federal Ministry of Health (FMOH) report
In-service training	Cost of providing integrated refresher training.	FMOH report
Personnel	Personnel time per client served estimated by time motion study and total client served.	Primary data [12, 13]
Supplies*	Supplies used for provision of selected service.	Primary data
	Number of supervisions conducted at all level per year.	Primary data, FMOH report, and expert opinion.
Supervision
Review meeting	Number of review meetings held by health centers and health offices with the HEWs per year.	Primary data

* Supplies includes medicines, consumables, and stationery.

To estimate costs per client for services requiring more than one visit, the number of visits was multiplied by the cost per service. Similarly, the cost per couple year of protection (CYP) was estimated for family-planning services. We then estimated the total annual cost of delivering selected HEP interventions at health posts by multiplying the unit cost of selected HEP interventions by the number of clients served. All costs were collected in Ethiopian birr and converted to United States dollars (US$) using the average exchange rate for 2018 (US$1 = 27.67 birr). Costs were inflated to 2018 rates using the Ethiopian consumer price index. A discount rate of 3% was applied [14, 15]. The cost analysis was carried out on an Excel spreadsheet [16].

Health outcome measures

Health outcomes were measured in LYG and estimated using Spectrum software. Three steps were used to calculate the number of LYG due to the HEP to estimate the number of lives saved by an intervention at a given coverage. First, the current health services coverage, including the HEP, was extracted from the 2016 Ethiopian Demographic and Health Survey (EDHS; i.e., the health system with HEP). Second, by using scientific literature on impact studies, we created counterfactual scenarios of what would have happened to the health system coverage in 2016 were there is no HEP (i.e., a health system without the HEP). Baseline coverage was presumed to be similar for both steps and was extracted from the 2005 and 2011 EDHS [17, 18]; linear interpolation was then made between baseline and current coverage for the two steps (Table 3). Spectrum software [19] (lives saved tool [LiST], FamPlan, and TB impact and modelling estimate [TIME]) then generated the number of lives saved for both steps. As a third step, the difference in the number of lives saved between the health system coverage (Health system with HEP) and the counterfactual (health system without the HEP) was calculated as the number of lives saved due to the HEP.

Table 3

Coverage inputs for effectiveness measures.

Intervention	Target population	Baseline coverage (%) (2005)	With HEP coverage (%) (2016)	Without HEP coverage (%) (2016)	Source
Ante-natal care (ANC):provision of iron, pregnant mother malaria treatment and tetanus toxoid (TT) vaccination	Total number of estimated pregnancies/births	28	62	42	[17, 22, 23]
Percentage receiving two or more TT injections during the pregnancy for the most recent live birth.	Total number of estimated pregnancies/births	32	41	34	[8, 17, 22]
Iron supplementation	Total number of estimated pregnancies/births	10	42	24	[17, 22, 23]
Family planning	Non-pregnant	24	35	24	[17, 22]
EPI: Pentavalent 3	Estimated live births	29	53	30	[7, 17, 22]
EPI: Measles	Estimated live births	28.5	54	33	[7, 17, 22]
TB: DOTs*	Number of populations in need of TB treatment	84	89	83	[24–26]
Malaria: ITN	Population at risk (60%) of malaria	33	40	32	[6, 27, 28]
Malaria: IRS	Population residing in highland fringe/epidemic-prone areas	14.2	29	21	[6, 27, 28]
Malaria-diagnosis and treatment*	Prevalence of malaria in all age group	24.2	38	18	[18, 22, 27]
Diarrheal treatment (zinc)*	Diarrhoea prevalence among children <5 years	0.3	33	28	[6, 18, 22], primary data
Diarrheal treatment (ORS)*		40	46	40	[6, 18, 22]
Acute respiratory tract infection and Pneumonia treatment*¥	Pneumonia Prevalence among children <5 years	13	31	22	[18, 22, 29]
Improved water source	Households	51	62	58.9	[17, 22]
Hand washing with soap	Households	50	60	57.2	[17, 22]

*Baseline year is 2010.

¥ Treatment includes the provision of cotrimoxazole, amoxicillin, and gentamycin.

The mathematical formula to estimate the number of lives saved due to an intervention is:

The number of lives saved was then multiplied by remaining life expectancy retrieved from Ethiopia’s life tables [20] to obtain the number of LYG [21]. Remaining life expectancy was calculated by weighing the number of male and female children below five years old and averaging the remaining years of life. Likewise, the weighted average life expectancy is calculated for women of reproductive age.

The life years gained were calculated as:

The Lives Saved Tool is a mathematical modelling tool used to estimate the effect of changes in coverage on mortality in low-and middle-income countries. The FamPlan model estimates the number of lives saved by considering family planning needs, the mix of methods, the contraceptive prevalence rate, and other criteria. TIME is an epidemiological compartmental transmission model that projects the drug-susceptible and multidrug-resistant TB burden.

Cost-effectiveness analysis

Since the founding of the HEP in 2004, the health system has incurred additional costs of health services. This additional cost to the health system will bring additional benefits to the community. We calculated the total incremental cost of a health service intervention due to the HEP by multiplying the additional number of the population covered due to the HEP and the unit cost of the selected intervention. The incremental cost-effectiveness ratio (ICER) is calculated as the ratio of the additional costs of intervention divided by the additional benefit of the intervention due to the HEP. The ICER is calculated for each intervention as well as for the whole selected HEP packages.

A one-year time horizon was applied. All cost and health outcome measures were discounted at 3%. The WHO’s CHOICE recommendation based on the GDP per capita for the year 2018 (i.e., US$852.80) was used to determine cost-effectiveness [27].

Sensitivity analysis

One-way sensitivity analysis was performed by varying the unit costs, the discount rate, the lifetime duration of equipment, buildings, pre-service and in-service training, and the salary of HEWs and others to check for the robustness of the findings when those parameters are changed.

Cost analysis

The unit cost was US$4.90 for family health service, US$7.40 for DPC, and US$0.70 for the hygiene and sanitation component of the HEP. The cost of delivering family health service by HEWs ranged from US$0.30 (iron folate supplementation) to US$9.90 (family planning: implant). Likewise, it ranged from US$2.0 for diarrheal disease management to US$43.10 for TB diagnosis, a service that is part of the DPC component. The unit costs of hygiene and environmental sanitation–which include improved water sources, hand-washing with soap, the hygienic disposal of children’s stool, and latrine use–range from US$0.60 to US$0.80 (Table 4). In addition, the average annual cost of delivering the selected HEP interventions per health post is US$9,897.

Table 4

Unit cost of selected HEP interventions by service delivery modalities in Ethiopia, 2018.

Package	Service Type	unit cost (US $)
Hygiene and environmental sanitation	Improved water source	0.6
	Hand washing with soap	0.7
	Hygienic disposal of children’s stool	0.7
	Use of latrine	0.8
	Average cost	0.7
Family health service	ANC	1.88
	Family planning: OCP	1.2
	Condom	1.8
	Injectable	3.2
	Family planning -Implant	10.5
	Pentavalent vaccination	15.2
	Measles vaccination	2.5
	Tetanus toxoid vaccination	3.0
	Iron folate supplementation	0.7
	Pneumococcal vaccination	9.4
	Average cost	4.9
DPC	HIV testing and counselling	3.7
	TB treatment (DOTS)	43.1
	Malaria case management	1.8
	Malaria prevention: ITN	2.1
	IRS	3.4
	Diarrheal disease	2.0
	Pneumonia treatment	2.6
	Average cost	7.4

Unit cost disaggregated by HEP components

The major cost drivers of the cost of family health services were drugs and supplies, which accounted for 68% of the total, followed by equipment (15%), capacity-building (8%), construction (8%), and personnel (4%) (Fig 1).

Fig 1

Percentage distribution of family health services unit costs by ingredients.

The study also showed that drugs and supplies were the leading cost drivers of DPC interventions (53%) (Fig 2). The cost of personnel (15%) followed by building (14%), equipment (6%), and capacity building (12%) also contributed to the unit cost of DPC interventions.

Fig 2

Percentage distribution of DPC unit costs by ingredient.

Personnel and capacity-building costs were the only cost centers for the hygiene and environmental sanitation package. Personnel costs accounted for 18%, 24%, 24%, and 32% of the costs of improved water sources, hand washing with soap, the hygienic disposal of children’s stools, and the use of latrines, respectively. The remaining share of the expenditure was contributed by capacity building, including start-up and monitoring and evaluation.

Cost-effectiveness analysis

The cost-effectiveness results, ranked from the most cost-effective intervention to the least are presented in Fig 3. The cost-effectiveness ratios of improved water sources, measles vaccination, hand-washing with soap, TT vaccination for pregnant women, antenatal care and iron supplementation for pregnant women, pentavalent vaccination, and oral antibiotics for pneumonia treatment were below the overall average and range between US$21.60 and US$67.20 per LYG. This suggests that providing the above interventions through the HEP would incur an additional US$21.60 to US$67.20 to the health care system for each additional LYG. Similarly, the provision of zinc and ORS for diarrheal case management, malaria case management, pneumococcal vaccination, TB treatment follow-up (i.e., DOTs), long-lasting insecticide-treated net, and family planning services had ICERs that ranged between US$78.10 and US$295.40 per LYG. Overall, the incremental costs of the HEP to the health system were US$24,886,899, with an additional benefit of 321,463.0 LYG over 15 years. In summary, for each additional LYG, the cost-effectiveness ratio of the HEP is US$ 77.40.

Fig 3

Incremental cost-effectiveness ratio of selected HEP interventions in Ethiopia, 2018.

All the interventions are considered very cos-effective, as the cost per LYG is less than 1 times Ethiopian GDP per capita. Among the interventions, the ICER for improved water source, measles vaccination, hand-washing with soap, TT vaccination for pregnant women, antenatal care, iron supplementation for pregnant women pentavalent vaccination, and oral antibiotics for pneumonia treatment less than 10% of the country’s GDP per capita for each additional LYG. Interventions like ORS for diarrheal disease control, zinc, and ORS for diarrheal case management and the treatment of malaria, pneumococcal vaccination, TB treatment follow up (DOTs), LLIN use, and the provision of family planning service have costs ranging from 10% to 34% of the country’s GDP per capita for each LYG (Fig 3).

Sensitivity analysis

Figs 4 and 5 present the results of the sensitivity analysis, which was conducted by varying the various parameters, such as unit cost, discounting LYG, life-years of capital items, pre-service and in-service training, and salary scale. Among all the interventions, discounting the LYG had the strong effect impact on the base case ICER, which changed slightly, when the discount rate for LYG was altered from no discounting to discounting by 6%. In almost all interventions, age discounting increased the base case ICER by a factor of two when changed from 3% to 6%. Conversely, when it was changed from a 3% discount to no discounting, the base case ICER fell by half. Life-years of equipment had the strongest effect on ICER after discounting the LYG. For example, when equipment life-years was changed from five to three, the ICER of the TT vaccination increased by 30% (Fig 4). Conversely, when equipment life-years were increased from five to seven, the ICER fell by 10%. The change in the ICER of the TT vaccination is a good example of the effect of poor-quality equipment and furniture, which incur a 30% additional cost to gain one more life year due to their decreased number of service years. Other variables, such as unit costs, service coverage, discounting costs, the total number of service users, building service years, building costs, HEWs’ salaries, and pre-service and in-service training years of service, have a minimal effect on the baseline ICER. Although changes in the rate used for age discounting do affect the ICER for the lower (19.2) and upper bounds (73.3), the range still lies within 1 times GDP per capita of the country, and the overall finding is robust with the base case results.

Fig 4

One-way sensitivity analysis showing cost-effectiveness ratio of TT injections over a range of key parameters.

Fig 5

One-way sensitivity analysis showing cost-effectiveness ratio of pneumonia treatment over a range of key parameters.