Background

Through cost-effectiveness analysis (CEA), healthcare researchers incorporate health benefits (eg, life-years gained) and costs to evaluate and compare different healthcare programmes and thereby make recommendations to decision-makers on the allocation of health resources.1 The National Institute for Health and Care Excellence (NICE) in the UK published national pharmacoeconomic guidelines to the CEA practice, and so did some other countries (eg, Canada, Australia). China published the first version of pharmacoeconomic guidelines in 2011 and updated it in 2020 to guide the CEAs in China.2 In the updated version, a societal perspective is recommended, especially when the expenditure is publicly funded. The guidelines make it clear that, under the societal perspective, all direct, indirect and intangible costs should be included. Direct costs refer to costs related to the intervention under economic evaluation, whereas indirect costs refer to productivity loss in the guidelines. Intangible costs refer to the physical and psychological discomfort (eg, pain, anxiety, tension) caused by the intervention of interest, being captured by the quality-adjusted life-years (QALYs). The guidelines do not recommend a unified incremental cost-effectiveness ratio (ICER) threshold for economic evaluation. Instead, a range is recommended that lies between one and three times the per capita gross domestic product. Researchers may select a threshold within the range according to the context where the evaluation is conducted.

Although CEA is increasingly used to assess whether healthcare interventions produce value for money, many methodological challenges exist in China that require attention.3 4 One such issue is whether to include future medical costs in CEA and, if so, to what extent they should be included.5 Future medical costs are incurred by future medical consumption during the life-years gained that would not have been consumed without a life-extending intervention. They are typically divided into future ‘related’ and ‘unrelated’ medical costs. The former refers to costs associated with treatments in the extended life that are directly related to the disease being treated by the life-extending intervention. For example, a heart transplant surgery extends a patient’s life, who should thereafter routinely visit a cardiologist to ensure the heart works normally. The costs associated with the visits are future related medical costs. The latter—future unrelated medical costs—refers to all the costs associated with diseases in the extended life that are not related to the disease and the life-extending intervention of interest. Therefore, future unrelated medical costs are the consequences of the life-extending intervention from a societal perspective. In the above-mentioned example, if the patient experiences cancer during the extended life, the costs associated with the cancer treatment belong to future unrelated medical costs.

Future-related medical costs are typically included in current CEA practice. Nevertheless, it is still in a debate whether to include future unrelated medical costs in CEA.6 7 This issue has not been discussed among Chinese health economists, because the latest guidelines do not include future unrelated medical costs in the cost inventory.2 In contrast, this line of research continues in other countries and leads to the recommendation of including such costs.8 Since the Chinese guidelines prioritise the societal perspective, the issue of future unrelated medical costs deserves an extensive discussion. Therefore, we reviewed the theoretical debate surrounding the inclusion of future unrelated medical costs and the recommendations by official pharmacoeconomic guidelines in other jurisdictions and proposed an agenda for future research on this issue in China.

We conducted a scoping review of studies discussing the inclusion of future unrelated medical costs in CEA. The search was conducted in EMBASE and MEDLINE. We used two categories of keywords: (1) terms regarding future unrelated medical cost, including ‘future cost’, ‘future medical cost’, ‘unrelated cost’, ‘unrelated medical cost’ and ‘survivor cost’ and (2) terms regarding economic evaluation, including ‘cost-effectiveness analysis’, ‘cost–utility analysis’, ‘cost–benefit analysis’, ‘health technology assessment’ and ‘economic evaluation’. We conducted the search in May 2021. The inclusion criteria were: (1) peer-reviewed articles discussing the inclusion of future unrelated medical costs in CEA and (2) CEA studies if the analysis was an example to assess the impact of including future costs. In addition, to provide a landscape of official recommendations on the inclusion of future unrelated medical costs, we referred to the Professional Society for Health Economics and Outcomes Research website to search for the pharmacoeconomic guidelines of different jurisdictions.

Two authors (SJ and YW) independently screened the search results to examine their eligibility for inclusion. The references of included studies were also examined to prevent the missing of relevant studies. For the included articles, we extracted their bibliographic and methodological information, opinions and conclusions. For official guidelines, we extracted the bibliographic information (ie, country, year, organisation, title) and their recommendations on whether to include future unrelated medical costs. Patients or the public were not involved in the design, or conduct, or reporting, or dissemination plans to research. We finally included 22 articles (figure 1). Their main conclusions are presented in table 1 and bibliographic information in online supplemental table S1. Through the review of these articles, we summarised their opinions and associated evidence as below.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flow chart for the search of eligible studies (attached as a separate PDF).

Arguments supporting exclusion

First, a widely spread argument is that the new treatments should be evaluated independently.1 It argues that when economic evaluation is being made, there is no commitment to deliver future unrelated care; as such, future unrelated medical care is not a necessary consequence of the life-prolonging intervention and should be excluded. The independence also implies that the future unrelated care could be evaluated separately, apart from the intervention of interest.

Second, some researchers admitted that the incurrence of future unrelated medical costs was a consequence of life-prolonging intervention, but the exclusion of these costs would not affect the ranking of the cost-effectiveness ratios of alternative interventions.9–11 For example, Garber and Phelps argued that, based on their theoretical model, including the future costs would only add a constant figure to all the ratios and excluding these costs would simplify the calculation.9 Their conclusion was based on the assumption that patient’s consumption plus medical expenditures were equal to earnings in each period.12

Third, some researchers emphasised the difficulties in estimating the future unrelated medical costs as an argument not to include these costs.7 13 Indeed, taking a bottom-up approach and predicting the risk of unrelated diseases in the future seems very difficult, if not impossible.14 Another concern is the difficulties when disentangling the ‘related’ and ‘unrelated’ costs.15 In practice, it is impossible to identify a ‘clear-cut’ between the two categories because we rarely know the covariance between future diseases. The two difficulties may place an extreme analytic burden on researchers.

Fourth, the ethical concern is another reason supporting the exclusion.1 16 17 Some researchers were concerned that the inclusion might cause the removal of some interventions previously funded by public insurance and the patients’ lives would be threatened.17 A frequently mentioned example is dialysis.17 When assessing a life-extending intervention in patients receiving dialysis, if the costs of dialysis as the future unrelated medical care are included, the value of life-years gained by the intervention may not outweigh the cost of dialysis. In that case, the intervention will not be funded even if it is provided at zero cost.

Arguments supporting inclusion

All arguments supporting exclusion have received rebuttals. In contrast to the first argument supporting the exclusion, some researchers argued that future unrelated medical care would usually be funded and cause real opportunity costs in the future.15 Because the healthcare authority functions intertemporally and future medical costs are part of opportunity cost considered by the authority. Therefore, it makes no sense to exclude future unrelated costs just because they are independent of the intervention being evaluated.

Second, unlike Garber and Phelps’s model, Meltzer developed a generalised model, proving that lifetime utility maximisation would only be achieved by accounting for future related and unrelated medical costs.18 He appraised the models by Garber and Phelps and by Lee and criticised that the assumptions of their models were based on unrealistic assumptions that were difficult to justify.12 In his model, Meltzer relaxed the assumptions and assumed that future medical costs vary under different conditions. Meltzer demonstrated that the inclusion of future unrelated medical costs could be large enough to alter the rank of cost-effectiveness ratios of different interventions. Therefore, to ensure that the health resources are distributed unbiasedly, including future unrelated medical costs is necessary.

Third, the concerns regarding the practical difficulties have been refuted.19 Although the difficulties exist, the estimation of ageing effect on healthcare expenditure shed light on how we could predict the future unrelated medical costs.17 Instead of predicting future risk of unrelated diseases, we could apply a top-down approach and use the healthcare expenditure per capita by age and sex to estimate the average future medical costs.14 In the Netherlands, this approach has been applied by a group of health economists to develop a future cost estimation tool, known as Practical Application to Include Future Disease Costs (PAID).14 They linked the age-specific and sex-specific per-capita medical costs to the survival curves and then add these costs to the cost inventory of the intervention in research to approximate the total future medical costs. This approach has made the estimation feasible and straightforward. Moreover, the inclusion of future unrelated medical costs would reduce the analytic burden. In contrast to the exclusion approach, which requires researchers to disentangle ‘related’ and ‘unrelated’ costs, the inclusion approach makes it unnecessary to disentangle the two types but consider them all, which is a relief to CEA analysts.

Fourth, the ethical concerns have been refuted.5 17 Since the future unrelated medical costs are real opportunity costs, indicating that people in the future may use the money for health gains, the inclusion has ethical implications. If we ignore these costs, are we willing to sacrifice the health gains of patients in the future? As van Baal et al said, ‘Pretending that future unrelated medical costs do not exist is equivalent to deeming the lives of some future patients worthless, which is neither fair nor ethically acceptable.’17 The ignorance of these costs is always inappropriate in dealing with ethical questions, threatening the credibility of CEAs. If we care about the ethical issues, we should address them through a formal process to assess the ethical impact of the inclusion and evaluate whether the budget has been optimally allocated.

In addition to the rebuttals, researchers provided other arguments supporting the inclusion, one of which is about the internal consistency of CEA.5 20 21 In the current practice, the improved quality of life and prolonged life-years are projected in the analysis. Implicitly, the health benefits require the use of unrelated medical care in the future. It is internally inconsistent to include the future benefits derived from unrelated medical care but exclude the associated costs. To ensure internal consistency, Nyman developed rules and required that all costs producing projected health benefits be included.22

Another argument is that excluding future unrelated medical costs would underestimate the opportunity cost of life-prolonging interventions.15 This argument is connected to the rebuttal to the ethical concerns. Under a fixed intertemporal healthcare budget, life-prolonging interventions will consume some budgets in the future. The consumption will lead to a decrease in the per-capita healthcare budget in the future. In this sense, the life-prolonging interventions cause opportunity costs for others in the future. Thus, intervention with very high opportunity costs may simply mean it is unattractive and should not be invested.

External consistency is another argument.6 20 Some researchers argued that the CEAs should be most informative for decision-makers. Thus, excluding future unrelated medical costs would decrease the external validity of CEA, as it reduces the amount of useful information for decision making.

A landscape of current guidelines in other jurisdictions

A total of 43 guidelines from 41 jurisdictions are summarised in table 2 with references presented in online supplemental table S2. Of the 43 guidelines, most (60%) do not specify the inclusion or exclusion of future unrelated medical costs in CEA. Three guidelines (7%) from the USA and the Netherlands recommend an inclusion approach. The second panel guidelines from the USA indicate that both the healthcare sector perspective and societal perspective need to include the future related and unrelated medical costs. The Institute for Clinical and Economic Review (ICER) guidelines from the US also recommend the inclusion approach and presenting a separate scenario analysis by excluding future unrelated medical costs when an intervention that increases QALYs is found to be cost-ineffective. The Netherlands guidelines indicate that if the intervention could increase the life expectancy, both related and unrelated future medicals costs should be considered. This is the first national guideline that make the inclusion approach mandatory. It is worth noting that the Swedish guidelines experienced a back-and-forth process on the inclusion of future unrelated medical costs.23 In 2013, the guidelines prescribed an inclusion approach. In 2015, the guidelines were changed by excluding future unrelated medical costs.

Contrarily, 11 guidelines (26%) explicitly recommend an exclusion approach, while most of them do not provide justifications. The NICE guidelines currently exclude future unrelated medical costs from the analysis.24 However, it is reviewing the inclusion in its guidance.25 Three guidelines (7%) from Hungary, Belgium and Germany recommend excluding the future unrelated medical costs in the base case and considering them in the separate analysis if they were deemed essential for some interventions.

A research agenda to inform Chinese guidelines

By summarising the arguments supporting the exclusion and inclusion of future unrelated medical costs, we found a growing consensus that the inclusion has a solid theoretical basis. While practical difficulties exist, researchers have made progress in addressing these issues. However, we found that most pharmacoeconomic guidelines do not yet respond to the methodological progress on this issue, whereas only a few guidelines recognise the importance of this issue and recommend inclusion. Since Chinese pharmacoeconomic guidelines prescribe a societal perspective, this issue is worthy of more attention than other methodological and ethical challenges that persist in China.4 26 We acknowledge that recent CEAs have made efforts in defining a clear cost inventory for the Chinese population;27–29 however, we deem that more research is needed to reduce methodological difficulties concerning future unrelated medical costs.

First, due to the establishment of province-level electronic reimbursement systems in recent years, most Chinese provinces have accumulated individual-level reimbursement records. Though the variables are limited (eg, age, sex, health condition and reimbursed payment), these data may play as a starting point in estimating the future unrelated medical costs, as what has been done in PAID.21 For example, researchers may use the data to establish regression models to predict the age-specific, sex-specific and condition-specific costs paid by the reimbursement system. Since there is no provincial administrative data in most Chinese provinces, the numbers derived from the regression models may provide a justifiable estimate of future unrelated medical costs for patient lives. The regression results could also be used to calculate condition-specific total costs because the condition-specific reimbursement rate is transparent.

Second, the end-of-life costs are unclear in China. Since the healthcare expenditure is usually concentrated in the last stage of life,21 30 it is essential to figure out how much has been spent at the end-of-life stage by disease-related groups (DRGs). A possible approach is to work with provincial top-tier hospitals and use the hospital administrative data to establish regression models for end-of-life cost estimation by DRGs. As a result, given that the intervention of interest belongs to one DRG, the end-of-life costs of other DRGs contribute largely to the future unrelated medical costs.

Third, the literature discussing future unrelated medical costs implicitly assumes that the standard of care for the unrelated future diseases and associated expenditure remains the same in the projected life-years gained.31 This is an unrealistic assumption because healthcare technologies advance over time, and the standard of care and associated costs would change accordingly. However, it would be extremely difficult to predict how we may deal with health conditions in the future and how the technology progress may alter the costs. As a starting point, we may assume that healthcare technology develops constantly in the future and the associated costs decrease at a corresponding rate. With this assumption, we could establish functions to predict the future unrelated medical costs, based on the amount derived from the provincial reimbursement data.

Fourth, the Chinese Health Technology Assessment (HTA) agencies and health economists should establish a formal process to evaluate the ethical and economic impact of including future unrelated medical costs in CEAs. If the HTA agency determines to include future unrelated medical costs, the ICERs may be significantly impacted, as indicated by previous research.15 16 20 32–35 In this case, the ICER threshold recommended the current guidelines would be inappropriate and HTA agencies need to establish an explicit and systematic mechanism to adjust the ICER thresholds.

In addition to the research on future unrelated medical costs, it would be valuable to investigate how patient preference could be engaged36 37 and whether to consider future non-medical costs,18 38 as required by the societal perspective Future non-medical costs refer to the consumption during the life-years gained if the patient receives a life-extending intervention. There is no consensus on this issue. Nyman argued that future non-medical costs should be excluded because the benefits of future non-medical consumption are not systematically included in the health outcomes.22 Other health economists disagree and persist that only inclusion of these costs could achieve utility maximisation.18 39 40 Chinese health economists should pay attention to it to participate in the discussion on this topic.