Case Report

Case Report in Health Technology Assessment and Drug Safety

A case report tells the story of one patient in enough clinical detail for others to recognise the same pattern, and is often the first way a new disease, an unexpected harm or an unusual response to treatment reaches the literature. It shows that something can happen, but it cannot say how often it happens or how much a treatment changed the outcome, because there is neither a denominator nor a comparison group. In health economics that limit matters most in rare and ultra-rare conditions, where published cases may be a large share of the evidence a health technology assessment (HTA) committee sees, and in drug safety, where single reports of suspected reactions feed the signals that later change model inputs.

What a case report contains and how it is reported

Murad and colleagues describe a case report as the clinical course of one individual, which may include exposures, symptoms, signs, interventions or outcomes, and call it the smallest publishable unit in the literature. A case series aggregates individual cases in one publication. The boundary is loose in practice, since the authors report a median of four patients in articles indexed with "case report" as the publication type, and the CARE group's own definition covers a medical problem experienced by one or several patients.

The CARE (CAse REport) guidelines published by Gagnier and colleagues in 2013 set the reporting standard. A three-phase consensus process involving 27 participants produced a 13-item checklist running from title and abstract through patient information, clinical findings, a timeline, diagnostic assessment, therapeutic interventions, follow-up and outcomes, and discussion, to the patient's perspective and informed consent. The authors acknowledged that judging causality remains difficult for case reports even when the guidelines are followed, and stated that CARE concentrates on the quality of information, independent of any causality assessment.

Place at the base of the evidence hierarchy

Case reports sit at or near the bottom of conventional evidence hierarchies because they are uncontrolled. Murad and colleagues describe case reports and case series as uncontrolled designs known for increased risk of bias, and note that a single-patient case report does not allow the estimation of an effect size, only descriptive or narrative results. Under GRADE, the certainty of evidence drawn from case reports and series is commonly very low.

Very low certainty does not mean the evidence is unusable. Because GRADE separates the certainty of evidence from the strength of a recommendation, Murad and colleagues show that strong recommendations can rest on case-based evidence, as with preferring paracetamol (acetaminophen) to aspirin in febrile children because of the association with Reye's syndrome, or hyperbaric oxygen for purpura fulminans, which is based only on case reports. The authors note that guideline developers sometimes ignore such evidence and focus guidelines on areas with higher-quality evidence.

What single cases have revealed

The CARE authors list the identification of adverse and beneficial effects, the recognition of new diseases and unusual forms of common ones, and the presentation of rare diseases as areas where case reports have proved helpful. They cite the link between thalidomide and congenital abnormalities, and the use of propranolol for infantile haemangiomas, as discoveries that began with case reports. Murad and colleagues add James Herrick's 1910 report of sickle-shaped red cells, after which sickle cell disease became a diagnosis.

The history also shows the limits. Nissen and Wynn record that the first published concern about thalidomide was a letter to the Lancet on 16 December 1961, two weeks after the manufacturer had announced withdrawal of the drug, and stress that unusual events in some exposed children could not by themselves prove the drug was the cause.

Case reports in pharmacovigilance and signal detection

In drug safety the single case becomes a regulatory record. The ICH E2D(R1) guideline, adopted in the EU on 18 September 2025 and in effect from 18 March 2026, defines an individual case safety report (ICSR) as a description of an adverse event or adverse drug reaction, or other observation, in an individual patient at a specific point in time. To qualify for submission, a report needs at least one event or observation, one suspect or interacting medicinal product, an identifiable patient and an identifiable reporter.

Published case reports enter this system through literature screening. E2D(R1) encourages, and in some regions requires, marketing authorisation holders to monitor the worldwide scientific literature for safety information on their products. It warns that literature cases may differ from other sources concerning causality, since authors may mention many events and many products without suspecting each product of causing each event. Spontaneous reports from health professionals and consumers reach regulators through adverse event reporting systems.

The EMA's good pharmacovigilance practices module on signal management (GVP Module IX, Rev 1) defines a signal as information from one or more sources suggesting a new potentially causal association, or a new aspect of a known association, judged likely enough to justify verificatory action. Signal detection may involve reviewing individual case safety reports, statistical analysis, or both, depending on the size of the data set. A single well-documented case can therefore start a process whose purpose is to test the association.

Causal reasoning about a single case relies on features inside the report. Murad and colleagues single out alternative causes, challenge and rechallenge, and dose-response as the appraisal questions most relevant to adverse drug events. A reaction that resolves when the drug is stopped and returns when it is restarted gives a stronger inference than a temporal association alone.

Rare diseases and highly specialised technologies

Nissen and Wynn note that randomised controlled trials can be impossible in rare disorders for lack of patients, so research on treatments must rely partly on less rigorous designs, including case series and case reports. In this setting case reports can describe the natural history of an ultra-rare condition, show which complications and harms a model structure should include, and supply narrative detail on outcomes that matter to patients. The relative effect that drives incremental QALYs has to come from trials, indirect comparisons or comparative real-world studies.

The NICE manual for technology appraisal and highly specialised technologies (PMG36) states a strong preference for high-quality randomised controlled trials for relative treatment effects, with non-randomised studies complementing them or forming the primary source when there is no randomised evidence (section 3.3.2). It allows clinical and patient experts to give opinions that supplement, support or refute observed data from trials or non-randomised studies, among which it lists case studies (section 3.3.22). For rare diseases, NICE says committees may accept a higher degree of uncertainty, after considering how the nature of the condition affects the ability to generate high-quality evidence (section 6.2.33). For highly specialised technologies, the committee also weighs the robustness of the current evidence and the contribution the guidance might make to strengthen it (section 6.2.7). Below a most plausible ICER of GBP 100,000 per QALY gained for highly specialised technologies, the decision to recommend is normally based on the cost-effectiveness estimate and the acceptability of the technology as an effective use of NHS resources (section 6.3.4).

Why a case report cannot supply a model parameter

A decision model needs two kinds of number that a case report cannot give. A treatment effect needs a comparison between what happened with the technology and what would have happened without it, and a case report observes one patient under one course of care, so the counterfactual is missing, which is the basic problem of causal inference. An event probability needs a count of events over a population at risk.

$$p = \frac{e}{N}$$

where $p$ is the probability of the event over the model cycle or period, $e$ is the number of events and $N$ is the number of patients at risk over the same period. Case reports supply some of the events in $e$ and nothing about $N$. Nissen and Wynn put it as a numerator with no denominator, so case reports cannot generate rates, incidences or prevalences.

The numerator is also incomplete. Hazell and Shakir's systematic review of 37 studies found a median under-reporting rate of 94% (interquartile range 82% to 98%) for adverse drug reactions in spontaneous reporting systems, and 85% even for specific serious or severe reaction and drug combinations.

Worked example of a reported harm in an ultra-rare disease model

The following figures are illustrative only. A new therapy for an ultra-rare disease has an incremental cost of GBP 300,000 and an incremental gain of 4.0 QALYs per patient against best supportive care, so the base-case ICER is GBP 300,000 / 4.0 = GBP 75,000 per QALY gained. At 10 or fewer incremental QALYs, PMG36 applies a size-of-benefit weight of 1 (section 6.2.24). Three case reports then describe acute liver failure in treated patients. The modeller adds a liver failure event to the treatment arm, costing GBP 30,000 and causing a loss of 5 QALYs per episode.

If 6,000 patients had been treated over the period, $p$ = 3 / 6,000 = 0.0005. The added cost per patient is 0.0005 x 30,000 = GBP 15 and the QALY loss is 0.0005 x 5 = 0.0025, giving an ICER of 300,015 / 3.9975 = GBP 75,051 per QALY.

If only 600 patients had been treated, $p$ = 3 / 600 = 0.005, the added cost is GBP 150 and the QALY loss 0.025, so the ICER is 300,150 / 3.975 = GBP 75,509 per QALY.

If the three reports were only 6% of true cases, borrowing for illustration the median under-reporting that Hazell and Shakir found in spontaneous reporting systems, there would be 3 / 0.06 = 50 cases. With 600 treated patients, $p$ = 50 / 600 = 0.0833, the added cost is GBP 2,500 and the QALY loss 0.4167, giving 302,500 / 3.5833 = GBP 84,419 per QALY.

A threshold analysis shows how far the harm would have to go to change the decision against GBP 100,000 per QALY. Setting (300,000 + 30,000p) divided by (4.0 minus 5p) equal to 100,000 gives 530,000p = 100,000, so p = 0.189, about 113 events among 600 patients. The case reports establish that the event belongs in the model, but its probability spans a range of more than a hundredfold across these assumptions, and the decision is robust only because the threshold probability lies far above every scenario shown. A real submission would take the probability from trial safety data or registries and present the case-based range in scenario analysis.

Case report, case series and N-of-1 trial

A case series groups several patients, which allows proportions and event rates within the series but still has no comparison group. Nissen and Wynn note that a cluster of similar cases can form the case group for a later case-control study, which adds controls and can estimate an odds ratio. A cohort design samples on exposure and follows people forward, so it can give absolute risks.

An N-of-1 trial also involves one patient but is experimental. The CENT 2015 statement describes N-of-1 trials as using key methodological elements of group clinical trials to evaluate treatment effectiveness in a single patient, which an observational case report cannot do.

Common misreadings of case reports

A common error is the anecdotal fallacy, in which a vivid case is taken as evidence of a general effect. Nissen and Wynn describe overinterpretation as the temptation to generalise without justification, encouraged by the emotional appeal of a patient's story.

Published cases are also a biased sample. Nissen and Wynn cite a study of reports in the Lancet in which only 5% of case reports and 10% of case series described treatment failures, and another in which 99.1% of 435 dental case reports had a clearly positive outcome.

The absence of case reports does not show the absence of harm, given the scale of under-reporting to spontaneous reporting systems found by Hazell and Shakir. A published report of a suspected reaction is not proof of causation either, since E2D(R1) notes that the author may not suspect every product mentioned of causing the event described.

Sources

• European Medicines Agency. Guideline on good pharmacovigilance practices (GVP) Module IX, Signal management (Rev 1). EMA/827661/2011 Rev 1. 9 October 2017. https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-good-pharmacovigilance-practices-gvp-module-ix-signal-management-rev-1_en.pdf
• European Medicines Agency, Committee for Medicinal Products for Human Use. ICH E2D(R1) Guideline on post-approval safety data: definitions and standards for management and reporting of individual case safety reports. EMA/CHMP/ICH/59123/2024. Adopted 18 September 2025, effective 18 March 2026. https://www.ema.europa.eu/en/ich-e2d-post-approval-safety-data-management-scientific-guideline
• Gagnier JJ, Kienle G, Altman DG, Moher D, Sox H, Riley D; CARE Group. The CARE guidelines: consensus-based clinical case reporting guideline development. Journal of Medical Case Reports. 2013;7:223. https://doi.org/10.1186/1752-1947-7-223
• Hazell L, Shakir SA. Under-reporting of adverse drug reactions: a systematic review. Drug Safety. 2006;29(5):385-396. https://doi.org/10.2165/00002018-200629050-00003
• Murad MH, Sultan S, Haffar S, Bazerbachi F. Methodological quality and synthesis of case series and case reports. BMJ Evidence-Based Medicine. 2018;23(2):60-63. https://doi.org/10.1136/bmjebm-2017-110853
• 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
• Nissen T, Wynn R. The clinical case report: a review of its merits and limitations. BMC Research Notes. 2014;7:264. https://doi.org/10.1186/1756-0500-7-264
• Vohra S, Shamseer L, Sampson M, et al. CONSORT extension for reporting N-of-1 trials (CENT) 2015 Statement. Journal of Clinical Epidemiology. 2016;76:9-17. https://doi.org/10.1016/j.jclinepi.2015.05.004