Economic evaluation of health interventions

Because clinical efficacy is necessary but not sufficient for an intervention to be funded: HTA agencies, international donors and health service managers usually require economic evidence to make their decision. This applies when a therapy, prevention programme, nutrition intervention or device already works and the next step is to show that it is worth the investment compared with the alternative. An economic evaluation of health interventions quantifies whether the health benefit an intervention delivers justifies its cost, integrating clinical and economic data into a decision model that estimates the incremental cost-effectiveness ratio and its associated uncertainty. Showing that something works and arguing that its health benefit justifies its cost are two different questions, and the second is the one a funding decision-maker has to answer. Without the economic component, a dossier with strong clinical evidence can lack the argument the payer needs to decide.

The ICER (incremental cost-effectiveness ratio) is the difference in costs divided by the difference in health effects between an intervention and the relevant alternative (ΔC/ΔE): the additional cost of gaining one extra unit of health, for example the cost per QALY gained. It is interpreted within the decision-maker's framework, against the willingness-to-pay threshold they use, which varies by country and funder: the same ICER may be acceptable in one setting and not in another. It is also an estimate with uncertainty, not a verdict: it depends on the comparator, perspective, time horizon and model assumptions. We therefore report the probabilistic sensitivity analysis (PSA) and the probability that the intervention is cost-effective at different thresholds. We also identify dominance, where one option is both more effective and less costly than the other. The ICER summarises the trade-off between cost and health; it informs the decision but does not replace the decision-maker's budgetary, clinical and ethical judgement.

The difference lies in the time horizon and the structure of the problem: a decision tree suits comparisons with a short horizon and a defined sequence of events; a Markov model suits cases where you need to represent health states that evolve over the medium or long term. This applies at the model-structuring stage, where the choice depends on the disease and the intervention, not on a default preference. A decision tree works well for a decision whose outcomes play out within a limited period. A Markov model, with its health states, transition probabilities and time cycles, represents chronic or recurrent processes in which patients move between states over years. The structure is chosen according to the complexity of the problem and the horizon relevant to the decision. Choosing the wrong structure can bias the result, so that choice is justified explicitly, not assumed.

Because the model parameters, such as costs, utilities or transition probabilities, are not known with certainty, and PSA is what carries that uncertainty through to the result instead of hiding it. It applies especially when the decision depends on parameter uncertainty and the result has to be defended before an HTA agency or an assessor. PSA simulates over the parameter distributions using Monte Carlo methods and generates cost-effectiveness planes and acceptability curves that show the probability of the intervention being cost-effective at each willingness-to-pay threshold. Deterministic analysis, with tornado plots and scenario and threshold analyses, complements it by identifying which parameters carry most weight, but it does not replace the probabilistic analysis, which is what characterises overall uncertainty. A cost-effectiveness result presented as a single figure, without its uncertainty, is open to methodological objections. Showing the probability of cost-effectiveness is more honest and more defensible than asserting a point ICER.

QALYs (quality-adjusted life years) and DALYs (disability-adjusted life years) are metrics that express health impact in units that can be compared across different interventions. They apply when you want to prioritise between options whose benefits differ in nature, by reducing those benefits to a common unit. QALYs gained, calculated from utilities such as EQ-5D or SF-6D and from quality-adjusted survival, are common in evaluations for HTA agencies, especially in high-income countries. DALYs averted, which combine years lived with disability and years of life lost, are frequent in evaluating public health and nutrition interventions, and in international donor settings. The choice of metric depends on the type of intervention, the decision-maker and the context; the two are not simply interchangeable. Each metric carries its own assumptions about how health is valued, and those assumptions are made explicit in the analysis.

Yes. The analytical perspective, the discount rate, the willingness-to-pay threshold and the utility sources are adapted to the client's decision context. This applies because the same analysis presented to different decision-makers can be irrelevant if it does not speak the language of their evaluation framework. A model for a European HTA agency, with explicit thresholds and a health system perspective, is not the same as a model for an international donor assessing a nutrition intervention against a threshold set by the funder, by the country or by a framework linked to GDP per capita, possibly from a societal perspective. These structural decisions, such as which costs are included, at what rate future effects are discounted or which threshold the result is compared against, change the result and how it is read. That is why they are set explicitly, with a justification, at the outset, in line with whoever will make the decision. A model that is technically correct but foreign to the decision-maker's framework does not inform their decision.

It compares interventions. Economic evaluation of health interventions (HEOR) fits when the question is whether an intervention, technology or treatment offers value compared with its alternative, using cost-effectiveness methods such as Markov models, QALYs and ICERs. Analysing an organisation's internal spending is the other question, and it is answered by hospital cost and operational efficiency analysis: how much each episode costs, where the variation lies and which units have room for improvement. They do not answer the same question: HEOR compares the value of different interventions to inform a funding or adoption decision; operational cost-efficiency analysis looks inwards at the process to improve the efficiency of what is already being done. They can complement each other within the same organisation, but one informs a choice between options and the other optimises operations. If the question is "do we fund this intervention rather than the other one?", it is HEOR; if it is "where are we overspending in this process?", it is cost-efficiency analysis.

The service prepares the economic component of the value dossier, that is, the cost-effectiveness model, the PSA, the planes and the acceptability curves, not the management of market access or the price negotiation itself. This limit is worth stating because a "value dossier" covers regulatory, commercial and negotiation elements that go beyond the economic analysis. What the service delivers is the economic evidence behind that dossier: a documented, defensible model, with its uncertainty characterised and its code documented and reproducible so that it can be updated when new data become available. Market access strategy, dealings with the agency and negotiation with the payer are separate areas of expertise, led by the client or their access advisers, drawing on the analysis produced here. The service models, simulates and interprets value; the decision and the negotiation belong to whoever leads them. What you receive is the economic argument, not the whole access process.