Real-world evidence in drug safety
Real-world data fill in the safety picture a trial leaves incomplete: they keep a medicine under surveillance after approval, follow patients for as long as late effects take to appear, reach the people trials exclude, and have the scale to find rare adverse events.
Where trials fall short on safety
Trials often fail to capture the full range of risks and benefits met in real-world practice, and they are often restricted to a select patient population under controlled conditions. They have long been regarded as the gold standard for assessing the safety and efficacy of novel treatments, and they give valuable insight into a drug's performance under precisely defined conditions. Real-world data, drawn from routine clinical practice, give a more extensive and representative view of a drug's safety and performance, and they have become increasingly vital for monitoring the safety of medical interventions, including pharmaceutical treatments, once they are approved and used in broader populations. Real-world evidence uses these data to assess how a treatment performs in the patient population at large, as opposed to the often highly selective cohort of a randomised controlled trial. Safety data within real-world evidence provide a more diverse and comprehensive dataset than traditional trials. Real-world evidence also supplies epidemiological information about adverse events and atypical treatment reactions, and it constitutes the basis for safety signalling.
Set side by side, a randomised controlled trial has standardised treatment conditions, subgroups too small for analysis and limited follow-up, while real-world data come from routine clinical practice, include patients not eligible for trials, make subgroup analysis possible, detect new adverse effects and have unlimited follow-up.
Surveillance after approval
Once a drug or device is approved, continuous monitoring of its safety and effectiveness in real-world conditions is necessary. Real-world evidence, drawing on data from a broader and more varied population, can identify adverse events or rare side effects that may not have been detected during the trial phase. Real-world safety data can also uncover potential drug–drug interactions that were not identified in controlled trial settings. Real-world data provide an effective mechanism for ongoing surveillance, letting healthcare providers, patients and regulatory authorities track adverse events and safety signals as they arise. Without continuous monitoring, safety concerns may not be detected until they affect a large number of patients, and in the past several drugs have been withdrawn from the market after post-marketing surveillance revealed previously unrecognised risks. Real-world evidence serves as an ongoing safeguard: the safety data collected through it let regulatory authorities act swiftly when new safety concerns arise. Should a concerning trend, such as a rise in adverse event reports, be identified, regulatory agencies can take timely action, including issuing warnings, modifying labelling or even withdrawing a product from the market. Regulatory agencies such as the EMA and the FDA acknowledge the growing importance of real-world evidence in post-market surveillance, and in 2025 both were increasingly relying on it to monitor the safety of drugs and medical devices once they were available to the public.
Long-term effects
Trials generally follow patients for a limited period, often a few months to a few years, while the full range of a medicine's long-term effects may not appear until later, sometimes much later. Some safety issues, such as organ toxicity or cardiovascular complications, may not emerge until years after treatment starts. Real-world data provide the longitudinal data needed to monitor these long-term effects, giving a more complete picture of a drug's safety profile. By following patients over extended periods, real-world data can identify chronic side effects or benefits that only show with prolonged use, and such longitudinal data are particularly critical for drugs intended for long-term use, such as those for chronic conditions like diabetes or cardiovascular disease. Post-authorisation safety studies can serve as valuable sources for publications on the long-term safety of new medicines. One such study is ACROSTUDY, a global non-interventional safety surveillance study of the long-term treatment of acromegaly with pegvisomant, a growth hormone receptor antagonist. Trial data had raised concerns about potential liver function abnormalities and pituitary tumour growth, so marketing authorisation was granted on the condition that a post-authorisation safety study be conducted, and that study then generated reassuring safety data.
Populations trials exclude
Trials often impose strict inclusion and exclusion criteria, which makes their populations homogeneous, so their findings may not be fully generalisable to the broader, more diverse population. Their findings may not be fully representative of how a drug performs across different demographic groups, including patients of various ages, ethnicities or with multiple underlying health conditions. A medication that shows efficacy in young, healthy trial participants may, for example, show a different safety profile in older people or in patients with multiple comorbidities. Real-world evidence gives insight into drug interactions and treatment effectiveness across varied demographic groups, including those under-represented in trials. Real-world evidence can reveal how demographic and health factors influence the safety and effectiveness of treatments in real-world populations. Real-world studies can include diverse populations, and so give a more accurate picture of how a drug affects different segments of the population. This is especially important when assessing safety for vulnerable groups, including the elderly, pregnant women and people with comorbidities.
A study of the treatment outcomes and safety of growth hormone replacement during pregnancy in women with growth hormone deficiency used the Pfizer International Metabolic Database (KIMS), which collected real-world data on adults with hypopituitarism, and from the 201 pregnancies identified its authors concluded that there was no relationship between pregnancy and growth hormone replacement. Whether growth hormone replacement improves life expectancy in patients with growth hormone deficiency had long remained unanswered, and a mortality study using real-world data from KIMS found that patients whose hypopituitarism was due to craniopharyngioma or aggressive tumours continued to show increased mortality, while those with other underlying causes had life expectancy comparable to the general population. A study of mortality in patients with drug-resistant epilepsy used the Clinical Practice Research Datalink, a real-world research database of deidentified primary care data from a network of general practices across the UK. Because drug-resistant epilepsy is not coded as a distinct diagnosis in that database, its authors defined specific criteria to identify the cohort, and then assessed the prevalence of comorbidities and all-cause and epilepsy-related mortality between 1 January 2011 and 31 March 2021, finding mortality approximately four times higher than in the general UK population.
Rare adverse events
Some adverse events are infrequent and may not be detectable in the relatively small sample sizes of trials. Trials are effective at detecting common side effects, but they often lack the statistical power to identify rare or infrequent adverse events. Real-world data, which involve larger and more heterogeneous patient populations, are better equipped to detect these rare occurrences. Studies using large real-world datasets can identify rare but potentially serious side effects that may not have been apparent in the trial phase. Severe allergic reactions, for example, may affect only a small subset of patients in trials but become more apparent as the drug is used by a larger and more diverse population. Some adverse reactions, such as specific cancers or severe allergic responses, may occur in only a small subset of patients but could have significant health implications if not identified and addressed in a timely manner. Meta-analyses found, for example, that insulin therapy seemingly has an association with increased overall cancer risk, and has significant associations with colorectal and pancreatic cancers.
Limits and place beside trials
Despite challenges of data quality, confounding variables and causal inference, real-world evidence plays a crucial role in keeping safety under continuous monitoring and informing regulatory decisions after approval. In the view of one 2025 review, real-world data and the evidence drawn from them are a critical supplement to traditional trials, giving insight into the safety and effectiveness of treatments in diverse real-world populations, and by continuously monitoring safety data and identifying potential risks, real-world evidence ensures that treatments continue to benefit patients well beyond initial approval. The same review expects the role of safety monitoring and post-market surveillance to continue to expand as healthcare systems globally integrate real-world data.