Supporting the development of a health benefits package in Malawi

Malawi, like many low-income and middle-income countries, has used health benefits packages (HBPs) to allocate scarce resources to key healthcare interventions. With no widely accepted method for their development, HBPs often promise more than can be delivered, given available resources. An analytical framework is developed to guide the design of HBPs that can identify the potential value of including and implementing different interventions. It provides a basis for informing meaningful discussions between governments, donors and other stakeholders around the trade-offs implicit in package design. Metrics of value, founded on an understanding of the health opportunity costs of the choices faced, are used to quantify the scale of the potential net health impact (net disability adjusted life years averted) or the amount of additional healthcare resources that would be required to deliver similar net health impacts with existing interventions (the financial value to the healthcare system). The framework can be applied to answer key questions around, for example: the appropriate scale of the HBP; which interventions represent ‘best buys’ and should be prioritised; where investments in scaling up interventions and health system strengthening should be made; whether the package should be expanded; costs of the conditionalities of donor funding and how objectives beyond improving population health can be considered. This is illustrated using data from Malawi. The framework was successfully applied to inform the HBP in Malawi, as a core component of the country’s Health Sector Strategic Plan II 2017–2022.


AbsTrACT
Malawi, like many low-income and middle-income countries, has used health benefits packages (HBPs) to allocate scarce resources to key healthcare interventions. With no widely accepted method for their development, HBPs often promise more than can be delivered, given available resources. An analytical framework is developed to guide the design of HBPs that can identify the potential value of including and implementing different interventions. It provides a basis for informing meaningful discussions between governments, donors and other stakeholders around the trade-offs implicit in package design. Metrics of value, founded on an understanding of the health opportunity costs of the choices faced, are used to quantify the scale of the potential net health impact (net disability adjusted life years averted) or the amount of additional healthcare resources that would be required to deliver similar net health impacts with existing interventions (the financial value to the healthcare system). The framework can be applied to answer key questions around, for example: the appropriate scale of the HBP; which interventions represent 'best buys' and should be prioritised; where investments in scaling up interventions and health system strengthening should be made; whether the package should be expanded; costs of the conditionalities of donor funding and how objectives beyond improving population health can be considered. This is illustrated using data from Malawi. The framework was successfully applied to inform the HBP in Malawi, as a core component of the country's Health Sector Strategic Plan II 2017-2022.

InTroduCTIon
Sustainable Development Goal target 3.8 is to 'Achieve universal health coverage, including financial risk protection, access to quality essential healthcare services and access to safe, effective, quality and affordable essential medicines and vaccines for all' by 2030. 1 However, the resources available for healthcare are limited, so not all services can be provided. Health benefits packages (HBPs) are an increasingly common way of explicitly defining which health services are provided through public expenditure as progress is made towards Universal Health Coverage (UHC). 2-4 At least 64 low-income and middle-income countries (LMICs) defined some form of HBP by 2012. 4 5 However, packages vary widely in terms of how benefits are defined, the cost of the packages, the coverage levels actually achieved and the methods used to inform their design. 3 4

Summary box
What is already known about this topic? ► Health benefits packages (HBPs) are commonly used to set out what should be included in a publicly subsidised package of healthcare interventions to make progress towards the Sustainable Development Goal target 3.8 of Universal Health Coverage (UHC) in low-income and middle-income countries (LMICs). ► HBP design has typically failed to take proper account of all constraints faced (eg, healthcare expenditure, infrastructure and donor restrictions) and has not been informed by explicit analysis that can identify the potential value of including and implementing different interventions; as a result, HBPs are rarely fully implemented and so access to the most valuable interventions is restricted.
What are the new findings?
► The analytic framework is founded on an understanding of the health opportunity costs of the choices faced and so can offer a transparent, principles-based approach to informing the content and scale of a HBP with existing resources, the value of expanding the HBP and the incremental reallocation of resources within the package. ► An assessment of health opportunity costs makes it possible to report the potential net health impact (net disability adjusted life years averted) of including a particular intervention or the amount of additional healthcare resources that would be required to deliver similar net health impacts (financial value to the healthcare system). ► This enables interventions that should be prioritised to be identified and the value of implementation efforts and health system strengthening to be assessed and also indicates the value of expanding the package, the costs of the conditionalities of donor funding and the trade-offs required when considering other objectives.

BMJ Global Health
Despite the frequent and increasing use of HBPs in LMICs, package design often suffers from a number of common flaws. The process of benefits package design is often non-transparent, non-inclusive and not informed by explicit analysis that makes best use of the often-limited evidence available. Decisions can, therefore, appear ad hoc rather than evidence-based. In particular, analysis rarely reflects the impact of various constraints on intervention provision and uptake. Therefore, the health opportunity cost of decisions is seldom accounted for. These issues are highly context-specific and ultimately affect the scale of the additional benefits and costs of including particular interventions. Attempts have been made to address some of the evidential shortcomings with 'global public goods' (eg, the DCP series). However, they often fail to address local conditions such as constraints on provision and uptake. As a result, packages generally promise more than they can deliver and healthcare is implicitly rationed with the most essential care not necessarily being delivered. 3 If HBPs are to advance UHC goals in a way that makes best use of the resources available for healthcare and informs how additional resources can most productively and equitably be used, an analytic framework is required that exposes the inevitable tradeoffs to assist decision makers in their design. 5 Such a framework was developed in response to a request by the Ministry of Health of Malawi to researchers at the Centre for Health Economics, University of York for an analytic framework to guide resource allocation within the Health Sector Strategic Plan 2017-2022 (HSSP II). The framework needed to inform key questions posed by the Ministry of Health: ► What is the appropriate scale of the HBP? ► Which interventions represent 'best buys' for the healthcare system (HCS) and should be prioritised? ► Where should investments in scaling up interventions and health system strengthening be made? ► Should the package be expanded to include additional interventions?
► What are the costs of the conditionalities of donor funding? ► How can objectives beyond improving population health be considered? The remainder of this paper is organised as follows. The health policy context in Malawi is initially introduced. Then the framework for designing HBPs is described, and an illustrative analysis is presented to answer each of the questions posed by applying the framework to data from Malawi. The application of the framework by the Malawian government to the development of a HBP for the HSSP-II is described, before the applicability of the framework to other settings and suggestions for future work are discussed.

HeAlTH polICy ConTexT In MAlAWI
Malawi introduced its first essential health package (EHP) in 2004 as a means of allocating collectively pooled resources for healthcare in conjunction with the initiation of a health Sector-Wide Approach (SWAp) to funding and resource allocation. 6 As part of the SWAp, donors provided general budget support and resource allocation decisions were made centrally by Government. 7 This replaced a fragmented vertical disease-based approach to funding. 8 The donor share of funding for the SWAp gradually increased from 30% in 2004/2005 to 56% in 2006/2007. 9 However, following the 'cashgate' scandal of 2013, many donors moved away from general budget support. 10 11 In the 2014/2015 financial year, donors contributed only 8% of SWAp pool resources, while the remaining 92% (MK65.8 billion) were raised domestically, 12 with donors instead returning to vertical disease-based funding channels. In FY 2015/2016, on-budget funding (ie, government-raised funds and direct budget support from donors) made up only 32% of total funding while the remainder was mostly off-budget discrete project support. 12 13 Despite the changing fiscal and political landscape, Malawi has continued to use HBPs to prioritise spending from both government and donor partners in the health sector. However, its first two HBPs in 2004 and 2011 6 14 were unsustainable, estimated to cost between 83% and 182% of total health expenditure, of which the package forms only a part. 6 14-16 As is common with packages globally, the HBPs could not be implemented resulting in inequitable variations in access to care and in many circumstances priority 'best buy' interventions were not available. 17 A frAMeWork for desIgnIng Hbps To address the policy questions in Malawi, a general framework was required that enabled the quantification of the health gains that would result from different potential HBPs (ie, with different choices of interventions) and account for actual constraints on implementation, donor restrictions and objectives other than health improvement. Including an intervention in the HBP commits

Summary box
What are the recommendations for policy and practice? ► The purpose of this analytic framework is not to prescribe a particular package or what health expenditure ought to be, rather it shows how evidence, such as it is, can be marshalled and analysis presented in a way that can empower Ministries of Health (MoH) as they engage with a range of stakeholders in making explicit, accountable and evidence-based decisions. ► The framework can contribute to advancing UHC goals in a way that makes best use of the resources available and shows the value of committing additional resources for healthcare, addressing common challenges and trade-offs faced by diverse healthcare systems in LMICs. BMJ Global Health resources that could otherwise have funded other interventions that also improve health. These forgone interventions and their associated health improvements represent the health opportunity cost of including a particular intervention in the HBP. An explicit and evidence-based assessment of health opportunity costs enables metrics of value to be reported. These indicate the scale of the potential health impact of including an intervention in the HBP net of associated health opportunity costs and of ensuring it is fully implemented. This information can be reported in health or monetary terms, which in turn can inform the value of committing resources to implementation efforts. These metrics of value (see box 1) inform prioritisation decisions more directly than other measures that have been used previously. For example, estimates of burden of disease or cost-effectiveness ratios do not indicate the scale of population health benefits offered by providing interventions to defined populations. 18 19 The illustrative analysis that follows relies on an empirical estimate of health opportunity costs (see box 2) and uses estimates of the costs and health effects of interventions from the Tufts Global Health Cost-Effectiveness Registry and WHO CHOosing Interventions that are Cost-Effective (WHO-CHOICE) analyses. Budgetary analysis to determine the total cost of the package uses drug and supply costs from a 2014 costing mid-term review of the previous HSSP made available by in-country partners Palladium and the Clinton Health Access Initiative (CHAI). The size of eligible patient populations for each intervention and an assessment of the levels to which interventions were actually implemented in Malawi in 2014 use bottleneck analysis and data from CHAI. Therefore, the data requirements do not extend far beyond the data collected regularly in many LMIC health systems. Shortfalls in data availability are inevitable, in any environment, but the framework allows the best use to be made of routinely collected local data, which complements relevant and available globally available data, within decision-making processes.
InforMIng key quesTIons In Hbp desIgn What is the appropriate scale of the Hbp? Figure 1 shows the interventions for which all required estimates were available, ordered and numbered from the lowest (left) to highest (right) ratio of cost per disability adjusted life year (DALY) averted. The height of each bar represents the intervention's effectiveness-cost ratio, and the width of each bar represents the intervention's total cost. 20 The latter is a function of the number of patients that require it and the cost per patient of delivering it, assuming each intervention is fully implemented. If Malawi can currently afford to pay up to $61 to avert one DALY (ie, 16 DALYs averted per $1000, see box 2), interventions 1-48 would be included in the HBP resulting in a budget of $265 million (shown as vertical dashed line A).
The estimate of $61 per DALY might be regarded as too low if policy makers felt able to commit more funding to healthcare given the size of other budgets and overall public resources. Higher estimates imply an expanded EHP with a larger budget. Once the health that is likely to be delivered by greater healthcare expenditure is set out, it then becomes possible to have a more meaningful deliberation about how Malawi's public finance resources might be allocated between competing claims (health education, infrastructure and so on) and/or how increases in public finance to accommodate increased health expenditure might be achieved.
For example, if the widely cited norms of 1 or 3 GDP per capita are adopted, the 'threshold' would be $372 or $1116 per DALY averted, with interventions 1-60 or 1-65 included, implying a budget of $362 or $380 million, respectively, as indicated by dashed lines B and C. 21 It should be noted that in this illustrative example only those interventions where estimates of cost, health benefit, eligible population and level of implementation were available were Box 1 Metrics of value ► Net disability adjusted life years (DALYs) averted represent the net health impact of an intervention on population health. It is the difference between the DALYs averted by an intervention and DALYs that could have been averted if the money required to deliver it had been spent on other interventions. If the intervention saves resources, it is the DALYs averted by the intervention plus the DALYs that can also be averted by including other interventions with the cost savings offered. ► The financial value to the healthcare system (the value of the intervention expressed in monetary terms) is the amount of additional healthcare resources that would be required to deliver the equivalent net DALYs averted with other interventions.
Box 2 estimating health opportunity costs ► Recent research, although in high-income countries, has demonstrated that an empirical assessment of health opportunity costs is possible based on estimates of the health effects of changes in healthcare expenditure. 28-30 ► Some of these estimates have been used to infer possible health opportunity costs in low-income and middle-income countries (LMICs). 31 ► Published estimates of the effect of changes in health expenditure on mortality using country-level data, including LMICs, can also be used to estimate health opportunity costs (cost per disability adjusted life year (DALY) averted) for particular healthcare system, reflecting their demography, epidemiology, healthcare expenditure, income and other characteristics. 32 ► The results of this type of empirical estimation suggest that the GDP per capita-based 'thresholds' that have been widely used to judge cost-effectiveness in LMICs are likely to be significantly higher than an assessment of health opportunity costs. ► In Malawi, the range of estimates available suggests that $61 spent on healthcare at the margin would be expected to avert one DALY. 31 32 BMJ Global Health included. Since other possible interventions are missing from figure 1, the difference in total budget for increases in the 'threshold' will tend to be underestimated, especially if high-cost interventions are under-represented. This also illustrates the experience of previous EHPs in Malawi and many other LMICs, where adopting 'threshold' norms that exceed the reality of health opportunity costs results in the inclusion of more in the package than can actually be funded. This leads to arbitrary and inequitable rationing, reduced health impact of the more limited resources that are actually available. 5 Which interventions represent 'best buys' for the healthcare system and should be prioritised?
While figure 1 provides a useful way to visualise the budget implications of using a higher or lower 'threshold' value, cost per DALY averted ratios are not useful for prioritising interventions because they do not indicate the scale of the potential health impact. Table 1 ranks interventions according to the net DALYs averted that they achieve, again initially assuming they are fully implemented. Ranking interventions by the net DALYs they avert results in a different ordering than ranking by ratios because the net DALYs averted reflects the size of the patient population as well as the individual health effect and costs. For example, management of obstructed labour, which is ranked 30th (ie, intervention 30) by cost-effectiveness ratios, is ranked second by net benefit because it generates a large health impact and remains higher than other interventions even when health opportunity costs are considered. The intervention ranked first GIT, gastrointestinal; ITN, insecticide-treated bed net; IPT, intermittent preventive therapy; LLIN, long-lasting insecticidal net; ORS, oral rehydration salts; PMTCT, prevention of mother to child transmission; pPRoM, preterm premature rupture of membranes. Table 1 Prioritising interventions in terms of impact on overall population health (net DALYs averted) Intervention

BMJ Global Health
by cost-effectiveness ratios (ie, intervention 1-cotrimoxazole prophylaxis for children) averts fewer net DALYs than other interventions that impose costs on the system, despite being cost saving. Those interventions to the right of the dashed line 'A' in figure 1 would result in negative overall population health impacts (ie, negative net DALYs averted) if they had been included in the package, as shown in table 1. This is because the cost associated with those interventions could be used elsewhere to better effect (ie, the resources would generate higher DALYs averted than if used for these particular interventions).
Interventions that represent 'best buys' for the HCS and should be prioritised are those that generate the most net health. These include HIV prevention strategies (including prevention, testing and treatment strategies); treatment for tuberculosis; maternal and child health interventions (such as management of pre-eclampsia, caesarean section and labour and delivery management) and prevention of and treatment for malaria.
Where should investments in scaling up interventions and health system strengthening be made? In Malawi, the mean actual implementation level in 2014 among all interventions included in the analysis is 46%, with a range of 1%-100%. Constraints to implementation include, on the demand side, individuals' lack of perceived benefits of care and difficulty in getting to clinics due to poor road infrastructure and, on the supply side, lack of equipment, lack of trained staff, supply chain bottlenecks, lack of beds, water and electricity shortages. 22 As a result, less money is spent delivering interventions and fewer DALYs are averted (see Columns 7 and 9 in table 2, respectively.) This results in a gap between current and potential spend of $198 million. One possible way of investing this spending gap is on policies to improve implementation levels, for specific interventions or across the HCS. Which interventions to invest in depends on the health gains that could be achieved by such investments. Table 2 ranks interventions by financial value to the HCS (Column 12).
For example, schistosomiasis mass drug administration is only available to 13% of the eligible patient population. If it were fully implemented, it would avert 23 754 DALYs (vs only 3088 at actual implementation levels). Table 3 presents the calculations underlying the values reported in table 2. Using the $61 per DALY averted estimate of health opportunity costs, if fully implemented, schistosomiasis mass drug administration would have a net effect of 12 562 DALYs averted (vs 1633 at actual implementation). As such, scaling up from actual levels of implementation to 100% would result in an additional 10 929 net DALYs averted (the difference between net DALYs averted at full and actual implementation (Column 3), equivalent to a $670 393 value to the HCS (Column 4). This means that, at most, $670 393 could be spent on removing the constraints to implementing schistosomiasis mass drug administration for that to remain a cost-effective use of resources.
Aggregating the total DALYs averted at 100% implementation across the interventions in the package (49.5 million) and subtracting the total DALYs averted at actual implementation (11.4 million) gives the maximum health gains that system strengthening could achieve (38.0 million DALYs averted). This suggests that there are potentially substantial gains from investing in policies which reduce or remove constraints to implementation at the intervention level and across the HCS as a whole.
should the package be expanded to include additional interventions? The Ministry of Health could accept existing constraints and instead use the budget spending gap resulting from constraints on full implementation to fund the inclusion of additional interventions not included in the initial package (ie, any intervention with cost per DALY averted estimates greater than $61). Whether this should be judged as a good use of money depends on the DALYs that can be averted by the additional interventions at actual implementation levels.
Using the spending gap ($198 million) to include interventions 49-67 would avert 2.7 million additional DALYs, resulting in a total of 14.2 million DALYs averted. This is 35.3 million fewer DALYs than could potentially be averted by investing in policies to improve implementation of already included interventions. This suggests that investing in implementation efforts should be prioritised if there are effective ways to relax the constraints. Although the effectiveness of such policies is often unknown, understanding the scale of the potential benefits can support informed judgements by decision makers. For example, even if only 14% of the potential health gains of implementation efforts were achieved using the spending gap it would be preferable to package expansion with that money. Furthermore, expanding the package may adversely impact the implementation of higher priority interventions so the additional DALYs that could be averted by including additional interventions probably overestimates the health benefits of expanding the package.
What are the costs of the conditionalities of donor funding? Donors, who fund approximately 70% of the HCS in Malawi, may also impose constraints through their funding arrangements. 23 Analyses comparing the health benefits of the donor's offers of assistance with the health opportunity cost can inform a discussion with donors about the need to impose constraints on their funding and can engage stakeholders in understanding the implications of particular policy options regarding donor offers. Such options may include accepting the donor proposal but being clear about the health opportunity cost of doing so or rejecting offers of matched funding for interventions that do not offer net health benefits. Proposals that might make the implementation of high BMJ Global Health

BMJ Global Health
priority interventions more difficult might be mitigated by other policies (eg, use of user fees to deter uptake of the imposed intervention). The framework provides estimates of the health opportunity cost of the constraints that a donor proposal might impose, which provide a valuable basis for explaining decisions to stakeholders. When an intervention that is not cost-effective is included it always reduces the total health generated by the package. The difference in the health gains associated with a health maximising package that uses all available resources, including those provided by donors, and a package where the donor specifies that particular interventions be included as a condition of the resources provided, indicates the minimum health opportunity cost of these restrictions. For example, requiring that first-line treatment for cervical cancer (intervention #65 in figure 1) is included in the package as a condition of existing levels of assistance will not increase the budget, so the health opportunity cost of this requirement is the health that would have been gained by the interventions that must be removed to accommodate it. The health opportunity cost of these types of conditions will be higher if it is not the least cost-effective interventions that are displaced and/or if they make other higher priority interventions more difficult to implement. Other examples of how different types of restrictions on assistance can be assessed are illustrated in online supplementary file 1. Evidence of the scale of the health opportunity costs associated with restrictions and conditions on donor assistance enables a more informed and accountable negotiation between stakeholders including careful examination of the reasons for restrictions.
How can objectives beyond improving population health be considered? Inevitably, the Ministry of Health and stakeholders may want to consider a range of objectives in addition to gains in population health when making decisions about what interventions to include in the EHP. These might include, for example, using interventions to promote financial protection or to reduce health inequalities and recognising the impact of interventions on wider social objectives such as productivity. In principle, it is possible to extend the measures of benefit and opportunity cost to include these other considerations. 24 25 In practice, this may be challenging based on available evidence, in which case it is possible to inform decisions about relevant trade-offs based on changes in population health. 26 The health losses associated with including an intervention that would not be included on the basis of net benefit alone can be quantified in the same way as the health losses associated with conditions on donor funding. These can be weighed against the gains in other objectives that result from the inclusion of the intervention. This quantification provides policy makers with a basis to understand whether the trade-offs are worth making and a means of communicating their ultimate decisions to stakeholders.

ApplICATIon of THe frAMeWork To THe developMenT of A revIsed eHp In MAlAWI
This analysis is intended to provide an analytic framework which can be used to support rather than prescribe decisions. The framework and data supporting the initial analysis were shared with the Ministry of Health in Malawi, which mandated an already existing EHP Technical Working Group (TWG) to conduct the EHP revision process (including the Ministry of Health Heads of Departments and Programme Managers, technical partners such as the local WHO office, donors, academic institutions and other key national health stakeholders). The TWG added other criteria to health maximisation including: equity (whether an intervention targeted at risk or marginalised groups); continuum of care (where interventions are linked, eg, screening and treating); complementarities (whether interventions are part of package) and exceptional donor funded interventions (donor funding for interventions that were expected to remain largely stable in the medium term). The framework was used to quantify the health gains that would result from different choices of interventions that met the agreed criteria to varying degrees, enabling explicit consideration of the necessary trade-offs between maximising health and other objectives. The EHP TWG presented their draft package to District Health Officers and then the Ministry of Health management for approval. The whole process was facilitated by Ministry of Health economists.
The final agreed package costed $247 million per year and was predicted to avert 41.5 million DALYs if fully implemented. Like both previous packages, the cost of this package is more than the resources budgeted for it. However, it costs 31% less than the 2011 package ($362 million) and averts 92% as many DALYs. As such, it offers better value for money overall than its predecessor, implying significant progress towards a package that is more realistic and less aspirational. 17 This also highlights that there are valuable health gains from expanding the budget for the package to its full cost. An example of the deliberative process undertaken by the Ministry of Health, alongside the analysis, was the decision not to

BMJ Global Health
include male circumcision in the final EHP despite the analysis showing it to be a 'best buy' intervention. It was judged that the type of demand-side constraints which would need to be overcome to increase the implementation levels would be too great and render the intervention not cost-effective. The conditionalities of donors were considered in the process, particularly with respect to funding from the Global Fund and GAVI towards HIV and immunisation, respectively. After deliberation, the decision taken was to include many of the interventions funded by these organisations in the package, regardless of their cost-effectiveness, reflecting a lack of flexibility in health financing in Malawi and in the role of donors. The framework, however, provided a means to initiate conversation about the impact of a high proportion of earmarked funding within the health sector and the subsequent effect on population health.
The framework was augmented through further data collection on the additional criteria deemed important in package design within Malawi. Data in these fields were largely populated through expert elicitation. A benefit of the framework is its adaptable use in the policy-making environment. Additional data can be combined with the framework to the extent desired and possible. Within Malawi, quantitative data on other criteria considered for inclusion in the decision-making process (eg, financial risk protection) proved scarce, leading to the decision to focus primarily on health maximisation with other criteria for which data were elicitable from expert judgement considered within the deliberative process.
There were a number of limitations and challenges in using the analytic framework to revise the Malawi EHP. Initially there was limited understanding of opportunity cost, cost-effectiveness and budget constraint principles by some stakeholders. While the EHP TWG agreed on inclusion criteria, adhering to the implications of these choices was difficult in practice. In part, this was due to low total health expenditure per capita, $39, which suggested a much more restricted package than previous unaffordable packages. The historical vertical funding arrangements also meant that there was limited willingness by Heads of Departments and Programme Managers to consider disinvestment in their own interventions.
By applying the framework to data from Malawi, this study illustrates how metrics of value that reflect health opportunity costs can provide a principled and evidencebased support to decision-making processes. Specifically, they can quantify the health opportunity costs of constraints that inhibit delivering interventions fully; donor constraints on how funding is spent and the inclusion of objectives additional to improving population health.
Such analysis forms a critical part of package design. However, it also emphasises the important role of the decision-making process and how it interacts with analysis. As evidenced in the framework's application in Malawi, that process needs to, for example, define the objectives of the package, deliberate on the relevance of the evidence provided by analysis and to make final decisions around what should (or should not) be included in the package. 27 To ensure that it can be implemented, the package should also inform other health systems inputs and standards, such as treatment guidelines, essential medicines lists and payment or reimbursement mechanisms, which currently are not typically informed by such economic criteria. This can also inform broader questions such as the benefits of moving to a whole system approach to funding. For example, where funding is vertical and tied to one specific disease as is commonly the case, the health opportunity cost of this type of planning as opposed to a whole-system approach can be identified. The analysis also provides quantification of the health benefit of expanding the health sector budget and, therefore, clarifies trade-offs with other claims on public finance.
Inevitably the evidence available to conduct this analysis was limited in a number of respects. The interventions included in the analysis in this paper are those for which data were readily available on costs, health effects, the size of the patient population and actual levels of implementation. There were a number of interventions where some but not all of these data were available or were reported in ways that were not useful. There are also likely to be complementarities and interactions between interventions for which there is little evidence and have not been addressed, although the analysis can be extended to consider the cost and effects of different combinations of interventions. The analysis suggests that there are potentially substantial gains from investing in policies which reduce or remove constraints to implementation at the intervention level and across the HCS as a whole. However, additional evidence is needed about the cost and effects of specific policies and projects that could improve the implementation of high priority interventions.

ConClusIon
This study has illustrated the value of an analytic framework, founded on an understanding of the health opportunity costs of funding choices. It offers a transparent, principles-based approach to informing the content and scale of a HBP with existing resources, the value of expanding the HBP and the incremental reallocation of resources within the package. The paper has further shown that, even in the most evidence sparse of environments, available information can be marshalled and analysis presented in a way that empowers policy-makers and facilitates engagement of stakeholders in making explicit, accountable and evidence-based decisions on how limited resources can best be employed to improve population health. The Malawi case shows that the analytic framework is not prescriptive but rather a tool to guide decision-making that reflects the context in which BMJ Global Health they are made and which can be adapted and applied to different settings.
Contributors All authors contributed to the concept and design of this paper. JO led the manuscript drafting with PR, MS and KC. It was edited by all authors with substantial contributions from GM and FM and approved by all authors. This paper draws on a visit by JO to the Malawi Ministry of Health hosted by GM, DN and FM and a workshop on health economics in Lilongwe, Malawi organised by PR and AR, in which all authors participated and which contributed to the conceptual design of this research.
funding This study was funded by ESRC Impact Acceleration Account, York External Engagement Award, Bill and Melinda Gates Foundation (grant number: OPP1165566).
Competing interests None declared. patient consent Not required.
provenance and peer review Not commissioned; externally peer reviewed.
data sharing statement In addition to published empirical estimates of the costs and effects of health interventions from the Tufts Global Health Cost-Effectiveness Registry and World Health Organization CHOosing Interventions that are Cost-Effective (WHO-CHOICE) analyses, this article relied on unpublished data on drug and supply costs, the size of the eligible patient population and the levels to which interventions were actually implemented in Malawi. Data on drug and supply costs were made available from a2014 costing mid-term review of the previous HSSP made available by in-country partners Palladium and the Clinton Health Access Initiative (CHAI) and an assessment of the size of the eligible patient population for each intervention and the levels to which interventions were actually implemented in Malawi in2014 use bottleneck analysis and data from CHAI.
open Access This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http:// creativecommons. org/ licenses/ by/ 4. 0/