Opportunity costs, marginal productivity, and cost-effectiveness thresholds: providing useful analysis and communicating results

from The Academic Health Economists’ Blo… at https://bit.ly/3mLX3mn on September 24, 2020 at 07:03AM

In our first post, we explained opportunity cost, marginal productivity (of a healthcare system), net benefit, and cost-effectiveness threshold. In this post, we put forward our views on how to use cost-effectiveness thresholds and the existing estimates of marginal productivity in a cost-effectiveness analysis, and some suggestions about how to explain it to your audience.

We propose that cost-effectiveness analyses should (i) show results compared to the cost-effectiveness thresholds that are relevant to their healthcare system; (ii) note the basis of the cost-effectiveness threshold; and (iii) when feasible, report the expected net benefit using estimates of the marginal productivity of the healthcare system.

Our view on the implications for applied cost-effectiveness analyses

As the purpose of cost-effectiveness analysis is to inform decisions, we propose that cost-effectiveness results reflect the cost-effectiveness threshold set by the decision-making institution (regardless of the approach used to underpin the threshold) in the first instance. This can be reported either in terms of dominance or extended dominance or ICERs compared against the cost-effectiveness threshold, or as net benefits.

However, cost-effectiveness analysis should tell us about the benefits lost due to the opportunity cost as well as those gained. As such, we propose that the benefits gained and lost are reported in terms of net benefit calculated using an estimate of the marginal productivity of the healthcare system, when such an estimate is available.

For cost-effectiveness analyses that include effects outside the health care sector, it is less straight forward to calculate the opportunity cost and the net benefit. This is because less is known about the marginal productivity of other sectors, although the literature is quickly developing. Furthermore, the objectives may differ, and outcomes other than health may be of interest. Therefore, until the methods are well established, we suggest aggregating costs only by sector and/or budget and presenting the losses due to the opportunity cost for the sectors in which it is possible to do so.

Our suggestions for how to explain this approach to your audience

We suggest that analysts are clear about the basis of the cost-effectiveness threshold used in the analysis. 

Doing this can be a challenge. Below, we suggest stock sentences based on our experience as health economists and in consultation with colleagues, which can be used to communicate these concepts to a range of audiences: economics, clinical, or laypeople.

These suggestions have not been formally validated. They are meant to provide a starting point for discourse around how best to communicate these concepts to difference audiences. We welcome feedback on your experiences in explaining cost-effectiveness results and your thoughts on this post and the stock sentences below.

For an audience versed in economics, we suggest stating:

  • This work aims to inform decisions made by << the decision-making institution >>, which employs as a cost-effectiveness threshold ‘XXX/QALY’.  For this reason, we first present results compared to this threshold.
  • We calculate the opportunity cost using the estimate of the marginal productivity of the healthcare system. Estimates of the marginal productivity of the healthcare system reflect the expected number of healthy years gained by an increase in expenditure. It can be used to inform an estimate of how much health would be gained if the money required to fund the option were instead made available within the healthcare system. 
  • We present the expected population net benefit of funding the new option, which estimates the magnitude of the net contribution to population health, given health opportunity costs. Our analysis indicates that << the option >> would have an expected net health benefit of ‘YYY QALYs’ gained. This means that we expect that ‘YYY QALYs’ are gained net of any QALYs forgone (i.e., those that would have been expected to have been gained with the same expenditure elsewhere in the healthcare system).

Our suggestions for a clinical audience are:

  • As this work informs decisions by << the decision-making institution >>, we use ‘XXX/QALY’ as the cost-effectiveness threshold. This is the maximum added cost per QALY gained that << the decision-making institution >> deems acceptable.
  • In principle, the cost-effective option could be considered to be the one with the greatest health benefits net of the health lost due to its costs. Costs result in health lost because funding one option shifts resources from other uses, thereby forgoing their health benefits. The costs can be expressed in terms of health lost using the rate at which the health care service improves health. This was estimated as one QALY gained for every additional ‘XXX’ spent. This is equivalent to using a ‘XXX/QALY’ threshold.
  • We present the cost and health benefits of each option in terms of their net health benefit. This is the health benefits net of the health lost due to the costs, given the rate at which the health care service improves health with more resources.

And lastly, our suggestions in plain English:

  • To choose which option is the best value for money, we used the maximum acceptable cost according to << the decision-making institution >>. This is XXX for one full year in full health (i.e. a QALY). 
  • An option is only good value for money if it improves health per pound at a greater rate than what the healthcare service currently does. Therefore, we also looked at which option is the best value for money given how good the health care service is at improving health.
  • For each option, we show how much health we gain, net of the health lost due to the costs.

Why not use the term ‘willingness to pay for health’?

Our view is that this phrase is ambiguous. It can mean the consumption value of health based on the willingness to pay of individuals (e.g. patients, taxpayers, etc). But it can also mean the willingness to pay of the decision-maker – their cost-effectiveness threshold. This cost-effectiveness threshold can be based on different values. As discussed in our previous post, it may be based on supply-side values, demand-side values, or implied values or norms. Therefore, we prefer to use the term ‘cost-effectiveness threshold’ to refer to the decision-makers’ set value.


We would like to thank Susan Griffin, Richard Cookson and Anthony Culyer for their helpful comments on previous versions of this blog post, and which have greatly improved its content. Any errors or omissions are ours.