Evidence Table

Evidence Table in Systematic Reviews and HTA

An evidence table gives one row or block to each included study, so that a reader can see what was studied, in whom, how well and with what result before any pooling or modelling takes place. It is the visible link between the reports a reviewer read and the numbers that later appear in a meta-analysis, a guideline recommendation or a cost-effectiveness analysis. In health technology assessment the same idea extends to the table of model inputs, where each value is listed with its precision and its source, and NICE asks for both kinds of table. A committee can check a parameter only if it can trace it back through the tables to a named study.

What an evidence table records for each study

The content follows the review question. The appendix to the NICE guidelines manual (PMG20) that gives example evidence tables lists, for intervention studies, the bibliographic reference, study type, number of participants, participant characteristics, intervention, setting, comparison, length of follow-up, outcome measures, effect size with confidence intervals where possible, source of funding, quality assessment and additional comments. NICE says that no fixed template suits all topics and that the items should be agreed at the protocol stage and completed consistently.

The version for economic evaluations adds the type of economic analysis, data sources, time horizon, discount rates, perspective, measures of uncertainty and applicability. A figure worked out by the reviewers, such as a risk ratio, is to be marked as calculated by the review team, and empty headings are marked not reported or not applicable.

Where the evidence table sits among related tables

Several tables are produced in the course of a review, and the names are often used loosely. The data collection form is the working record behind data extraction, and the evidence table is the published product.

Cochrane makes collecting enough detail to populate its characteristics of included studies table mandatory. Its summary of findings tables work at a different level, reporting for up to seven critical or important outcomes the numbers of participants and studies, relative and absolute effects and a GRADE rating of the certainty of the body of evidence. PRISMA 2020 asks for each included study's characteristics, risk of bias and, for every outcome, effect estimate with its precision, ideally in structured tables or plots.

How evidence tables are used to group and appraise studies

The Cochrane framework for synthesis starts by summarising each study in a characteristics table, then compares characteristics across studies, for example in a matrix, to decide which are similar enough to be grouped. Cochrane prefers tables to long text because their structure makes patterns easier to scan and grouping decisions more transparent.

The tables also feed the judgement of certainty. Cochrane notes that a synthesis of study characteristics is needed for GRADE assessments of indirectness and of explanations for inconsistency. The quality column records critical appraisal of each study, while certainty of evidence is rated across studies for each outcome. The NICE economic evidence profile does the same for published cost-effectiveness studies, with columns for limitations, applicability, incremental costs and effects, the ICER and uncertainty.

NICE requirements in technology appraisal

The NICE manual (PMG36) asks that all parameters used to estimate clinical and cost effectiveness be presented in tables with details of data sources, and that summaries of key characteristics and results of included studies follow technical support document 1. For the model, input data should be tabulated with the central value, measures of precision and sources, with details of how bias was assessed and addressed for each source. Each health state utility value should be tabulated with its source and the method used to derive it.

The company evidence submission template (PMG24) suggests tables for these, among them a comparative summary of trial methodology, baseline characteristics by treatment group, quality assessment results, a list of published cost-effectiveness studies and a summary of the variables applied in the economic model. That last table gives each variable's value, its uncertainty and distribution, and the section of the submission that supports it.

Parameter tables, reporting standards and model transparency

A parameter input table is the modelling counterpart of a study-level evidence table. The CHEERS 2022 reporting checklist asks authors to report all analytic inputs, such as values, ranges and references, including uncertainty or distributional assumptions. The ISPOR-SMDM task force on model transparency recommended a non-technical description of each model, covering its inputs and data sources, and technical documentation detailed enough for an expert to evaluate and potentially reproduce it.

Each row of a parameter table should point to something checkable, such as an evidence table row, a named trial publication or a stated calculation. The Cochrane Handbook asks reviewers to keep a clear distinction between data taken from a source document and data obtained through calculation. Carried into a model, that rule shows which values were transcribed and which derived.

Design choices that change what a table shows

Several choices shape an evidence table before any result is pooled, and each can change the conclusions drawn from it. Cochrane expects them to be set in the protocol, with later changes reported as departures from it.

1. The unit of a row. Cochrane treats the study, not the report, as the unit of interest, and links reports of the same study through registration numbers, authors, sponsors, settings, intervention details, participant numbers and dates. PMG20 likewise asks for new studies to be cross-checked against existing ones to avoid double counting.
2. Which result to tabulate. Cochrane notes that results can vary very widely depending on which scales, subgroups and time points are selected, and asks protocols to be as specific as possible about what is collected and which report is used when reports conflict.

Worked example where linking reports changes the uncertainty in an ICER

All numbers in this example are illustrative. A review for a model of a preventive treatment finds three reports. Report A is a trial with a risk ratio of 0.75 and a standard error of 0.12 on the log scale, report B is a second trial with 0.90 and 0.14, and report C gives 0.78 and 0.12. An evidence table that records trial registration numbers shows that report C is a later follow-up of the trial in report A, so that trial contributes once, using the report the protocol specifies, assumed here to be report A. A fixed-effect model pools the log risk ratios by inverse-variance weighting, as in the Cochrane Handbook.

$$\ln \hat{RR} = \frac{\sum_i w_i \ln RR_i}{\sum_i w_i}, \qquad w_i = \frac{1}{SE_i^{2}}, \qquad SE_{pooled} = \frac{1}{\sqrt{\sum_i w_i}}$$

where $RR_i$ is the risk ratio in study $i$, $SE_i$ is the standard error of its log risk ratio, $w_i$ is its weight and $\hat{RR}$ is the pooled risk ratio. With A and B the weights are 69.4 and 51.0, a total of 120.5, the pooled standard error is 0.091 and the pooled risk ratio is 0.81 with a 95% confidence interval of 0.68 to 0.97. If report C is wrongly entered as a third study, the total weight rises to 189.9, the standard error falls to 0.073 and the result becomes 0.80 with an interval of 0.69 to 0.92.

A one-year model applies the risk ratio to a baseline event risk of 0.20, with a treatment cost of GBP 600, an event cost of GBP 5,000 and a QALY loss of 0.5 per event.

$$\Delta C = c_T - p_0 (1 - RR) c_E, \qquad \Delta E = p_0 (1 - RR) q_E, \qquad ICER = \frac{\Delta C}{\Delta E}$$

where $\Delta C$ is the incremental cost in GBP, $\Delta E$ the incremental QALYs, $c_T$ the treatment cost, $p_0$ the baseline event risk, $RR$ the risk ratio, $c_E$ the cost of an event and $q_E$ the QALY loss from an event. The table uses the rounded risk ratios.

The point estimate barely moves, but double counting narrows the interval, so the pessimistic end of the plausible range falls from GBP 190,000 to GBP 65,000 per QALY. A probabilistic analysis built on the narrower interval would understate parameter uncertainty and the value of further research. The error cannot be seen in the pooled figure, only in a study-level table that records which reports belong to which trial.

Limitations and common misreadings

An evidence table describes studies and does not synthesise them. Counting rows with favourable results as if the count were a finding is a form of vote counting, which the Cochrane Handbook treats as unacceptable when based on statistical significance. A good quality rating for each study does not make the body of evidence certain, since imprecision, inconsistency, indirectness and publication bias are judged across studies.

The table is only as accurate as the extraction behind it, and the Cochrane Handbook notes that extraction errors are rarely detected by peer reviewers, editors or users of reviews.

Tables also simplify, and Cochrane observes that the many subjective decisions about similarity behind them are rarely described. A parameter table that cites another model as its source moves the question back one step, and the original evidence for that value still has to be found and appraised.

Sources

• Eddy DM, Hollingworth W, Caro JJ, Tsevat J, McDonald KM, Wong JB. Model transparency and validation: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force-7. Value in Health. 2012;15(6):843-850. https://doi.org/10.1016/j.jval.2012.04.012
• Husereau D, Drummond M, Augustovski F, et al. Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS 2022) statement: updated reporting guidance for health economic evaluations. BMC Medicine. 2022;20:23. https://doi.org/10.1186/s12916-021-02204-0
• Li T, Higgins JPT, Deeks JJ. Chapter 5: Collecting data (last updated October 2019). In: Higgins JPT, Thomas J, et al, editors. Cochrane Handbook for Systematic Reviews of Interventions, version 6.5. Cochrane; 2024. https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-05
• McKenzie JE, Brennan SE, Ryan RE, Thomson HJ, Johnston RV. Chapter 9: Summarizing study characteristics and preparing for synthesis (last updated October 2019). In: Higgins JPT, Thomas J, et al, editors. Cochrane Handbook for Systematic Reviews of Interventions, version 6.5. Cochrane; 2024. https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-09
• Deeks JJ, Higgins JPT, Altman DG, McKenzie JE, Veroniki AA, editors. Chapter 10: Analysing data and undertaking meta-analyses (last updated November 2024). In: Higgins JPT, Thomas J, et al, editors. Cochrane Handbook for Systematic Reviews of Interventions, version 6.5. Cochrane; 2024. https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-10
• McKenzie JE, Brennan SE. Chapter 12: Synthesizing and presenting findings using other methods (last updated October 2019). In: Higgins JPT, Thomas J, et al, editors. Cochrane Handbook for Systematic Reviews of Interventions, version 6.5. Cochrane; 2024. https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-12
• Schünemann HJ, Higgins JPT, Vist GE, Glasziou P, Akl EA, Skoetz N, Guyatt GH. Chapter 14: Completing 'Summary of findings' tables and grading the certainty of the evidence (last updated August 2023). In: Higgins JPT, Thomas J, et al, editors. Cochrane Handbook for Systematic Reviews of Interventions, version 6.5. Cochrane; 2024. https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-14
• National Institute for Health and Care Excellence. Developing NICE guidelines: the manual (PMG20), chapter 6 and appendix H. London: NICE; 2014, last updated 23 October 2025. https://www.nice.org.uk/process/pmg20
• National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). London: NICE; 2022, last updated 13 October 2026. https://www.nice.org.uk/process/pmg36
• National Institute for Health and Care Excellence. Single technology appraisal and highly specialised technologies evaluation: user guide for company evidence submission template (PMG24). London: NICE; 2015, last updated 31 March 2026. https://www.nice.org.uk/process/pmg24
• Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. https://doi.org/10.1136/bmj.n71