Background
In light of rapidly ageing populations, disease prevention and health promotion (DPHP) is attracting increasing attention from health policy makers all around the world [
1]. According to the World Health Organization (WHO), disease prevention aims to minimize the burden of specific diseases and their associated risk factors by covering population-based and individually focused interventions. Slightly different, health promotion complements these efforts by empowering people to increase their control over determinants of health and developing supportive environments [
2].
In 2015, across the member states of the Organisation for Economic Co-operation and Development (OECD), the financial resources spent on DPHP were on average less than 3% of all health spending [
1]. An increase of investments in DPHP programs would require conclusive and valid information about the costs and benefits of DPHP-measures. Usually, this information is provided to funding agencies by health economic studies such as cost-effectiveness analyses (CEA, where benefits are expressed in natural effects or physical units), cost utility analyses (CUA, with benefits as health state preference scores), cost-benefit analyses (CBA, monetary terms), cost-minimization analyses (CMA, where only costs are compared) and cost-consequences analyses (CCA with an array of output measures as benefit) [
3]. In addition, the social return on investment (SROI) methodology targets broader socio-economic outcomes by analysing views of multiple stakeholders and expressing these in a singular monetary ratio [
4].
Although economic evidence for DPHP is increasing [
5], health economic studies for DPHP can be affected by significant methodological challenges. Motivated by the Wanless report [
6] that described factors likely to have an impact on the resources required to deliver a high-quality health service, the Public Health Research Consortium (PHRC) from the United Kingdom (UK) identified four key elements of economic evaluation in DPHP [
7].
(i) the attribution of intervention effects should ensure an adequate reflection of a complex DPHP-measure; since trial-based designs such as randomised controlled trials (RCTs) are not always feasible, economic modelling offers a flexible approach, as it uses multiple data sources [
8]. Similarly, the applied data on the consideration of (ii) outcomes and (iii) inter-sectoral costs and consequences (i.e., costs and monetary benefits which spread to other sectors) should reflect the specific context of DPHP. Furthermore, (iv) health equity is one of the main objectives stated in public health policy worldwide [
9].
To achieve the social goal of allocating funding more efficiently between health care and public health, it is vital that valid and comparable analytic methods are used for both [
6]. Therefore, the objective of this overview was to summarize the above listed key elements for methodological rigor in economic evaluations of DPHP-measures. The evidence was obtained from systematic and scoping reviews of a) previous health economic analyses over a broad range of DPHP areas/measures with b) a methodological focus on the attribution of effects, outcomes, inter-sectoral costs and consequences, or equity. We also discuss the reported challenges in view of published recommendations for economic evaluations of DPHP interventions obtained from various methodological papers.
Discussion
Our comprehensive overview summarizes how economic evaluations of DPHP-measures considered methodological challenges of economic evaluation in DPHP. According to eleven reviews (494 analyses), there were methodological inconsistencies over a broad range of DPHP measures for all predefined dimensions (attribution of effects, quantification of outcomes and costs, consideration of equity). Whereas current theoretical debates (e.g., [
9,
14,
18,
32,
33]) of specific challenges in economic evaluations of DPHP are rarely reflected in health evaluation practice, non-compliance with well-established general standards of economic evaluation [
34] was also often observed. As a result, the information obtained from many health economic analyses of DPHP measures may be limited in value for decision maker. Encouragingly, recent reviews [
23,
24,
30] indicate a tendency to address the specific challenges of DPHP more appropriately (e.g., inter-sectoral costs and consequences, equity).
To account for the
attribution of effects, the increased number of relationships and interactions in DPHP-models (e.g., between behavior changes, biomedical health indicators, patients, care giver) requires the application of models with more flexibility than conventional Markov-models (e.g., [
5,
25,
28,
31]). Because the cost-effectiveness of a DPHP measure is predominantly determined by individual behavior, some authors proposed more sophisticated modelling techniques such as econometric modeling, individual-level Markov models, discrete event simulation (DES), social network analysis or agent-based simulation (ABS) [
5,
19].
More specifically, a conceptual modelling approach as recommended by Squires et al. 2016 can support the development of model structures that reflect the dynamic and complex nature of DPHP interventions [
32]. Conceptual modelling allows to consider data on the uptake of an intervention, unforeseen participant responses (e.g., non-participation), or variations in the provision of measures (e.g., frequency or the care-giver involved). To reflect a complex and dynamic DPHP-measure, conceptual modeling framework refers to a holistic way of thinking about the interactions between parts within a system and with its environment [
35]. By defining multiple system levels which are subjectively defined, higher level systems correspond to lower level systems reflecting more detailed aspects [
32].
Moreover, conceptual models are assumed to allow the consideration of various determinants of health [
36] (e.g., the social, economic, and physical environment, as well as a person’s individual characteristics). By including broader determinants of health, conceptual models can also be used to facilitate identification of non-health costs and outcomes associated with the DPHP-measures [
32].
When preferring a trial-based approach, cluster-randomised community trials or registry-based RCTs (i.e., pragmatic trials that use registries for data collection, randomisation, and follow-up) may prove an appropriate tool for comparative effectiveness in real-world settings [
37]. However, the majority of trial-based evaluations are based on short time horizons (e.g. [
23]), indicating that decision analytic modelling should be used more often (to date: 10%) for extrapolating the findings of a study beyond the period of observation (e.g., [
5,
23,
26]).
With regard to the
outcomes used for evaluations of DPHP, authors of the reviews reported both a lack of usage of non-health outcomes and a lack of valuing outcomes. According to the reviews, intermediate outcomes were predominantly used in health economic analyses of DPHP interventions (56%), while non-health outcomes and preference-based outcomes are rarely applied (8 and 36%, respectively), resulting in a broad consensus for increasing their consideration [
5,
25‐
27]. Based on current methodological guidance, some researchers argue for the inclusion of non-health outcomes in a tool [
14,
38,
39] but disagree on what is more promising: the development of a new tool, or the use of established questionnaires with a theoretical founding (e.g., QALYs).
Over 60% of the studies in the included reviews did not attempt any outcome valuation, i.e. they provide a CCA (15%) or CEA (42%). In contrast, 20% of the studies were preference-based CUAs, with valuations restricted to health outcomes expressed in QALYs or DALYs. A CUA is grounded in a non-welfarist approach, while other studies with outcome valuation were based on a CBA (6%), a welfarism-approach with outcomes valuated in monetary terms using willingness-to-pay estimates or SROI. A CBA might be useful for adapting the analysis to the relevant outcomes beyond health [
31]; however, practical issues associated with monetary valuation remain unresolved [
32]. Given the ongoing discussion on how to measure and value non-health outcomes, a CCA appears to be useful for providing a disaggregated overview of DPHP interventions [
19,
32].
For the future, available methods should increasingly be used for the valuation of non-health outcomes. These include ‘contingent valuation’ [
5], ‘willingness to pay’ for eliciting a benefit of an intervention [
5,
19], and ‘discrete choice techniques’ [
40]. In addition, outcomes caused by externalities (i.e. outcomes for individuals who are not directly targeted by the intervention) should be included more often [
5].
Because
inter-sectoral costs and consequences were excluded in economic studies of DPHP for a long time [
5,
19,
28,
29], several researchers highlighted the need to consider these costs [
5,
14,
19,
32,
41]. However, the tools developed were heterogeneous and showed limited evidence on validity and reliability [
42].
In 2013, Drost et al. developed and applied several approaches for the inclusion of impacts for the education and criminal justice sectors [
18], resulting in a slight increase of reporting inter-sectoral costs and consequences in later reviews. However, a prerequisite is that data valuation can be based on proxy unit prices, on self-constructed unit prices or on labor costs [
18].
The consideration of labor costs, even in evaluations adopting a societal perspective, is controversial for different reasons. First, the inclusion of labor costs is contentious and, there is no consensus or guidance on how to measure and value productivity appropriately, in particular with regard to unpaid work. Second, productivity costs are often disregarded because they are assumed to be negligible. However, the exclusion of productivity costs may result in underestimating the cost-effectiveness of DPHP measures [
25].
DPHP interventions can be evaluated from a number of different perspectives, e.g., the health sector perspective, the public sector perspective or the perspective of particular agencies involved in the system [
32]. The failure to consistently adopt a societal perspective (46% and, in studies with a societal perspective, the omission of certain relevant costs such as those relating to productivity losses and participants’ time, may underestimate the cost-effectiveness of DPHP-measures [
43] and often precludes a deeper understanding of the monetary consequences of DPHP.
Although validated and well-accepted tools for the inclusion of these inter-sectoral costs are lacking, the adoption and consistent application of a societal perspective would stimulate efforts to include costs and effects beyond the health sector [
42].
Although many authors have called for the incorporation of
equity concerns in economic evaluations in DPHP, existing methods [
9,
38,
44] are neither common practice nor included in guidelines for economic evaluations. As a result, most evaluations in DPHP fail to consider equity. At least, the most recent review observed an increased consideration of equity aspects by conducting subgroup analyses or targeting a population deemed in need of intervention [
23]. This might indicate growing awareness of this issue.
In general, equity considerations might be qualitatively examined by providing background information on aspects of fairness [
5,
19,
45]. A more extensive approach aims to calculate cost-effectiveness for different ‘equity-relevant’ subgroups [
9] characterized by socio-economic status, geographical location or ethnicity [
5,
19,
23,
45]. A third approach aims to estimate the opportunity cost of a particular equity consideration in terms of population health sacrifice. For example, the life years lost by pursuing a more equitable option can be opposed to a more health maximizing option. The resulting opportunity cost is an approximate for the monetary value of considering equity [
9,
20].
The equity weighting analysis approach requires more information. This method aims to explicitly value the reduction of health inequality [
19,
45]. This analysis quantifies the trade-offs between improving total health and other equity objectives. The idea of this approach is to weight health gains (e.g., life years) with different equity-relevant characteristics. The weights are based on values elicited from specific stakeholders such as the general public or/and policy makers. The weights can be elicited using common health outcome valuation techniques (e.g., discrete choice experiments) [
9,
20] Weatherly et al. (2009) emphasize that more research is needed on equity-weighting issues for an individual’s social-economic status, personal responsibility for health risks, and the preference toward treating current illnesses versus preventing future health risk [
19]. Alternatively, equity might be considered by Sen’s capability theory [
46] which accounts for the distribution of capability across society.
Strengths and limitations of this overview
By taking up the quality criteria for economic evaluations identified by the PHRC, specific methodological challenges could be observed for predefined key elements of DPHP-interventions. Based on these categories, the results of our overview provide a comprehensive picture of the challenges for economic evaluations in DPHP. Thus, this analysis may provide a starting point for developing structured guidance for conducting economic evaluations in DPHP.
Although the methods used for this analysis were in line with the principles of good systematic reviewing [
10,
12,
13] some limitations are inherent: first, the overview was limited to studies focusing on methodological aspects. As a result, we did not include methodological challenges from systematic or scoping reviews of economic studies without this focus.
Second, in the reviews methodological aspects were heterogeneously reported. Furthermore, data on sub-criteria to be extracted was not always provided (or extractable), resulting in only approximate estimates and the impossibility to evaluate interrelationships between these sub-criteria (e.g., to calculate the proportion of studies with a societal perspective that included productivity costs). Therefore, this overview presents a trend in the modus operandi for DPHP evaluations.
Third, there may be other dimensions and sub-dimensions reflecting relevant aspects of the validity of economic DPHP evaluations which were not addressed in the PHCR-criteria (and thus not in our overview). Among others, these may include dealing with uncertainty, the choice of discount rates, and the inclusion of costs due to screening or costs resulting from added life years. However, by relying on the PHCR-criteria, we focused on dimensions of quality that were considered to be essential in the context of DPHP [
7].
Fourth, the specific relevance of each of the challenges (i) to (iv) in an economic study depend on several factors such as the specific interventions under evaluation, the target group, the objective, the chosen study type or the perspective. As a result, the need to make particular efforts in the methodological key elements may have differed between the studies. Because we included evaluations of any DPHP-measure, the differences requiring to address the specific challenges more or less were not considered in this overview.
Finally, some economic evaluations might have been included in different reviews. Since not all reviews reported the individual studies they included, we cannot rule out that double-counting occurred (Fig.
2). However, we believe that the main conclusions drawn from this analysis would not alter.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.