This analysis aimed to provide input for the revision of Ethiopia’s EHSP, which used seven predefined and pre-agreed criteria, one being the cost effectiveness of interventions [7]. Our analysis encompasses a comprehensive range of health interventions, including preventive, promotive, curative and policy interventions. Of the interventions analysed in this study, a large majority (75%) have ACERs of less than US$1000, and 36% have ACERs below US$100.
Cost-effectiveness analysis is an increasingly important prioritisation tool. The cost-effectiveness evidence for redefining Ethiopia’s EHSP was generated in three ways: by contextualising CEA evidence from other studies using transferability criteria, by using expert opinion for multisectoral interventions and by using the WHO GCEA tool [7]. Using this tool, we provide cost-effectiveness evidence for 159 relevant health interventions for EHSP revision in Ethiopia. We believe that other low-income countries in Africa can also generate these pieces of evidence within a relatively short time and at an affordable cost compared with individual economic evaluation studies.
We provide cost-effectiveness evidence for 77 interventions on RMNCH and infectious disease (e.g., HIV, TB, nutrition, malaria and WASH) and for 82 interventions on NCDs. In general, a majority of the interventions have relatively low ACERs of less than US$1000 per HLY gained. However, when we disaggregate the finding by programme area, the results are mixed. While a vast majority (95%) of RMNCH and infectious disease interventions have an ACER of less than US$1000 per HLY, a substantial proportion (44%) of NCD interventions have an ACER of higher than US$1000 per HLY. In general, findings from our study are consistent with findings of other country specific studies in Ethiopia, Zimbabwe, Mexico [8,9,10, 30,31,32], or other regional and global estimates [33,34,35,36,37,38]. However, head to head comparison of the ACERs and further examination of cost, effectiveness, and its driving factors remain a priority for additional research.
Family planning interventions, for example, are the most cost effective in this study, with ACERs of less than US$1 per HLY gained. This very low ACER may be partly explained by the fact that the model accounts for a reduction in unplanned pregnancies and an associated reduction in maternal mortality. Most of the interventions targeting infectious diseases were cost effective, with an ACER of less than US$500 per HLY. For example, we evaluated four HIV/AIDS interventions, and they all, except the provision of cotrimoxazole for children, have an ACER of less than US$100 per HLY gained. The relatively low ACER in this study may partly reflect the decrement of the price of ART drugs as is shown in several recent studies [39, 40].
Addressing maternal, neonatal and child health issues is a top priority of the Ethiopian Ministry of Health (MoH) [7]. In our study, the majority of the interventions on RMNCH were very cost effective, with an ACER value of less than US$200 per HLY gained. This finding is in line with that of Memirie et al. in a CEA examining the cost effectiveness of 13 maternal and neonatal health (MNH) interventions in Ethiopia. Although not a GCEA and therefore not directly comparable, that study found that 12 of 13 MNH interventions had an incremental cost-effectiveness ratio of less than US$400 per HLY [9].
Most of the preventive NCD policy interventions have a lower cost-effectiveness ratio than the treatment NCD interventions. A substantial proportion (44%) of NCD interventions have an ACER of greater than US$1000 per HLY. This relatively high ACER may reflect the fact that the treatment cost for chronic NCD is higher and the treatment effectiveness lower than for the other interventions. This is particularly consistent with findings from a comprehensive, but relatively old study, examining 101 NCD interventions in Mexico. The study find similar variations among NCD policy interventions and NCD treatment interventions as we do [30].
Strengths and limitations
By applying the GCEA approach, it is possible to evaluate whether the current mix of interventions is efficient and whether proposed new interventions are appropriate. Therefore, GCEA is a more appropriate approach than a marginal analysis for conducting a sector-wide cost-effectiveness analysis of interventions [14, 15]. In this study, which included 159 interventions from diverse programme areas, we conducted a sector-wide cost-effectiveness analysis. Although this study covers a substantial number of crucial interventions, it did not attempt to analyse all interventions in the Ethiopian health sector. We believe, however, that our findings can be used as benchmarks for making better-informed expert judgements on other interventions that could not be analysed in such a standardised way.
WHO-CHOICE GCEA tool is important tool for sector-wide analysis of cost-effectiveness of wider range of interventions for priority setting. A primary advantage of the WHO-CHOICE GCEA tool is the ability to compare many interventions at the same time based on the same assumptions on cost, disease epidemiology and other key health system parameters (e.g., human resource, financing, and infrastructure). When health system plans and strategies are designed, we should evaluate and compare the costs and outcomes of combinations of interventions. However, a barrier to conducting economic evaluation studies is that they are time consuming and demand large amounts of local data and local technical expertise. We believe that this study demonstrates that the existing platform, with a large support team and substantial commitment, makes such an extensive and comprehensive evaluation possible.
Our work has other limitations. First, in this study, we used the health system perspective. In Ethiopia, one-third of the total health care cost is covered by the out-of-pocket expenditure of individuals [19], which can influence individuals’ choices in accessing health care delivery. The choice of perspective should also be taken into consideration when interpreting the results. Second, in this GCEA study, we applied data from diverse sources to model the health impact of interventions and costs. Of course, modelling is inevitably an imperfect representation of reality, and, therefore, robust uncertainty analysis would to some extent alleviate this challenge. However, because of the vast number of interventions included in this analysis, we did not include a sensitivity analysis. Therefore, as the software expands, future GCEA analysis of this kind should integrate a sensitivity analysis of at least some of the critical drivers of costs and health impacts.
A third limitation of this study is the use of DALYs for estimating disease burden and health benefit. Critics of DALY argue that the measure itself has limitations [41, 42]. Using DALYs tends to underrepresent or overestimate the value of interventions (such as palliative care and family planning) with outcomes that are not readily measured in this metric as well as interventions in nutrition for which the outcomes are improved cognition rather than improved health [43]. This is a real limitation that was taken seriously in the revision of Ethiopia’s EHSP. For these interventions, we also relied on the expanded EHSP process with user involvement and expert judgements. Furthermore, criteria other than cost effectiveness, such as equity, financial risk protection, budget impact and public concern are also important for defining the EHSP [3]. A fourth limitation of this study is that the models used do not capture full health benefits. The most striking example is the LiST model which mainly considers mortality outcomes. Future analysis should also account for health benefits from RMNCH interventions that avert non-fatal conditions.
Additionally, there are gaps in the available evidence on the cost of interventions, which can be closed only by conducting substantially more research in developing countries. Therefore, we recommend a concerted effort to establish country-level cost databases. This could be combined with capacity building through the training of researchers to generate such evidence.