The CHEERS checklist is the reporting standard for health economic evaluations, and the current version is CHEERS 2022, which carries 28 items and replaces the 24-item 2013 statement. Its name stands for Consolidated Health Economic Evaluation Reporting Standards. It applies across the full family of evaluation types: cost-effectiveness analysis, cost-utility analysis, cost-benefit analysis and cost-minimisation analysis, whether trial-based or model-based.
Economic evaluations have an unusual audience problem. A clinical paper is read mainly by researchers and clinicians, but an evaluation is read by a payer deciding whether to fund something, and payers need to know whether your result holds in their jurisdiction, under their cost structure, over the horizon they care about. That is why CHEERS puts so much weight on a small set of framing parameters that a clinical guideline would not mention at all.
A cost-effectiveness result is not a property of an intervention. It is a property of an intervention evaluated from a stated perspective, over a stated time horizon, with a stated discount rate. Change any one and the number changes, sometimes enough to reverse the decision.
Perspective determines which costs count. A healthcare payer perspective counts costs falling on the health system. A societal perspective adds productivity losses, informal care and costs falling on other sectors. An intervention that looks poor value to a hospital can look excellent to society, and neither figure is wrong.
Time horizon determines which consequences count. A one-year horizon on a preventive intervention will capture its costs and almost none of its benefits, which is why horizon should be long enough to capture all differences in costs and outcomes, often a lifetime for chronic conditions.
Discount rate determines how future costs and benefits are weighted against present ones. Rates are usually set by national methods guidance rather than chosen by the analyst, and applying a rate different from the one your target agency mandates is a common reason for a submission to be sent back.
Report all three together, early, in one place. Scattering them through the methods is the most frequent reason a reader cannot reconstruct a published ratio.
The expansion from 24 to 28 items reflects two shifts in what decision-makers now expect, plus consolidation of items that had proved ambiguous.
The first addition is engagement with patients and others affected by the evaluation. The item asks how patients, service users, carers, clinicians and the public were involved in designing the analysis or interpreting its results. The rationale is that the outcomes an evaluation values and the perspective it adopts are choices, and choices made without the people affected tend to miss things that matter to them.
The second is health equity. CHEERS 2022 asks for any effects on equity to be reported, and for distributional effects to be characterised where relevant. An average cost-effectiveness ratio can conceal an intervention that improves aggregate health while widening a gap between groups, and reporting the average alone hides that entirely. There is also a clearer expectation that a health economic analysis plan existed and that the reader is told where to find it, which brings economic evaluation into line with the protocol-first structure that SPIRIT provides for trials.
Described in our own words and grouped for readability, the items ask for: the title identifying the study as an economic evaluation and the abstract summarising it in structured form; the background and objectives, including the broader context and the question being answered; the health economic analysis plan and where it can be accessed; the study population; the setting and location, since costs and practice are jurisdiction-specific; the comparators and the rationale for choosing them; the perspective; the time horizon; the discount rate for both costs and outcomes with justification; the selection, measurement and valuation of outcomes; the measurement and valuation of resources and costs; the currency, price date and any conversion method; the model used with its structure and a rationale, or a statement that none was used; the analytics and assumptions, including methods for adjusting data and extrapolating; the characterisation of heterogeneity across subgroups; distributional effects; the characterisation of uncertainty; the approach to engagement with patients and affected communities; the study parameters with their values, ranges, distributions and sources; the summary of main results including incremental costs, outcomes and ratios; effects on health equity; the discussion of findings, limitations, generalisability and current knowledge; and the source of funding together with conflicts of interest.
The official item wording and the checklist itself sit with the guideline and are indexed on the EQUATOR Network. Take the file from there rather than from a secondary summary, including this one.
The uncertainty item is where submissions most often fall short, and the expected standard has moved. One-way deterministic sensitivity analysis, varying each parameter across a plausible range while holding the others fixed, is useful for identifying which parameters drive the result. It is not sufficient on its own, because it cannot capture the joint effect of several parameters being uncertain at once.
Probabilistic sensitivity analysis, assigning distributions to parameters and sampling from them, is the current expectation for model-based evaluations. The standard outputs are a cost-effectiveness plane showing the scatter of simulated incremental cost and outcome pairs, and a cost-effectiveness acceptability curve showing the probability that the intervention is cost-effective across a range of willingness-to-pay thresholds. Report the distributions you assigned and why, since the curve inherits every assumption in them.
Structural uncertainty, meaning uncertainty about the model's shape rather than its inputs, is not captured by either method and is handled through scenario analysis. Say which structural choices you tested.
The distinction that runs through every reporting guideline applies here too, and in health economics it has a particular sting because a model can be reported in exemplary detail and still be wrong. CHEERS asks whether the evaluation is adequately described. Whether the model is credible, meaning internally consistent, externally validated against observed data, and structurally appropriate to the decision problem, is a separate assessment made with model quality and validation frameworks.
Nor is CHEERS an appraisal tool for use inside a review. If you are systematically reviewing published economic evaluations, you need a reporting judgement and a quality judgement kept apart, exactly as you would for observational studies under STROBE or qualitative studies under COREQ. The evidence feeding the model is a third question again, appraised with instruments such as AMSTAR 2 for the systematic reviews you draw on and rated for certainty using the GRADE framework.
Work through the 28 items against the final manuscript, recording the page and section for each rather than ticking it, and name the version, CHEERS 2022, in your methods. Where an item does not apply, for example model structure in a purely trial-based analysis, write not applicable with a brief reason rather than leaving it blank.
The items most often unanswerable in our experience are the analysis plan and where to access it, the rationale for the comparator, and distributional effects. The first two are quick to fix and the third is a genuine analytical decision, so it is worth reaching that item before the analysis is finished rather than after.