The scale of the problem that motivated the review is considerable. In the United Kingdom’s National Health Service, inpatient falls are a major cause of morbidity and mortality, particularly among older and frailer patients. Although falls prevention guidelines exist, the evidence supporting the efficacy of interventions in hospital settings remains thin. Yet hospitals are still benchmarked against one another using a crude national metric based on occupied bed days, a denominator that says nothing about how many staff were on shift, how easily nurses could see at-risk patients, or how the physical layout of a ward either helps or hinders supervision. The Nottingham team set out to test whether this measurement, when considered in the context of the settings in which it is reported, actually holds up.

To answer that question, the researchers conducted a systematic review to PRISMA guidelines, the international standard for transparent evidence synthesis. They searched nine databases and screened 3,541 articles, ultimately including 103 interventional studies from 19 countries that used falls per 1,000 occupied bed days as a metric. Rather than simply tallying numbers, the team extracted narrative data, author-reported limitations and contextual variables, and analysed them thematically using NVivo 15 software. Crucially, they framed their interpretation with the Systems Engineering Initiative for Patient Safety model, a Human Factors framework that treats safety as an emergent property of interactions between people, technology, tasks, organisations and the physical environment rather than the product of individual carelessness.

The first key finding is startling in its simplicity: the definition of a fall varies between study sites, and in 42 of the 103 full-text studies, exactly half, no definition was stated at all. This is not a trivial omission. If one hospital counts only witnessed falls to the floor while another includes assisted lowering, slides from beds and near-misses, their headline rates are measuring different phenomena. The review found multiple different interpretations of what constitutes a fall circulating across the literature, which means that any cross-hospital or cross-country comparison built on these figures rests on shifting semantic ground. A patient safety metric that cannot agree on what it is counting cannot plausibly serve as a universal yardstick.

The second finding concerns how the measure is expressed. Even when falls are defined, there is a lack of consistency in how the resulting rate is calculated and presented, producing comparisons between hospitals that are simply not equivalent. Occupied bed days are an imperfect denominator in the best of circumstances: they blend together patients of vastly different mobility, acuity and risk. A rehabilitation ward full of ambulant elderly patients and an intensive care unit with sedated, bed-bound patients may report the same denominator while facing entirely different fall hazards. When expression conventions differ on top of that, the review concludes, benchmarking and performance reporting may not be an accurate reflection of the challenges wards actually face, suggesting the measure may not be a useful comparator at all.

Third, the type of fall and how it is classified may mean that not all falls are reported, due to subjectivity at ward level. Classification decisions are made by busy clinicians in real time, and judgement calls about whether an incident qualifies as a reportable fall introduce human variability into the data stream itself. A slide from a chair, a controlled descent to the floor with staff assistance, or a stumble caught before impact might be logged in one ward and dismissed in another. This subjectivity does not merely add statistical noise; it systematically distorts the picture, potentially rewarding units that classify generously and penalising those that document scrupulously.

The fourth finding may be the most consequential for anyone designing safety policy: there are multiple contextual challenges that cause substantial variability in fall rates, and almost none of them are measured. Ward design and patient visibility, for example, profoundly affect the ability of staff to monitor at-risk patients, yet they appear nowhere in the standard metric. The number of staff available on each shift has also not been measured as a variable, despite being an obvious determinant of how closely vulnerable patients can be watched. Two hospitals could have identical reported fall rates while one operates single-storey bays with clear sightlines and generous staffing and the other runs cramped, partitioned wards with chronic shortfalls. Treating their numbers as comparable, the review argues, is an exercise in false equivalence.

By mapping these findings onto the SEIPS model, the authors make a broader theoretical point with practical teeth. Falls are not simply failures of individual vigilance or patient weakness; they are shaped by interactions between technology, people and the environment. A call bell that is out of reach, a bed height that is hard to adjust, a corridor with poor lighting, a shift roster that leaves one nurse covering too many high-risk patients, each of these system-level factors contributes to risk in ways that a per-bed-day rate cannot capture. The review reinforces the need for a systems perspective and Human Factors methodologies to tackle this pervasive patient safety challenge, an approach that has transformed safety in aviation and other high-stakes industries and is increasingly argued to be overdue in healthcare.

The implications for policy and practice are significant. National benchmarking exercises, league tables and performance dashboards built on falls per 1,000 occupied bed days risk drawing the wrong conclusions about which hospitals are safe and which need support. If the metric cannot distinguish between genuine improvement and artefacts of definition, classification or context, then improvement efforts may be misdirected, and wards facing the hardest conditions may be unfairly branded as underperforming. The review’s authors suggest that meaningful comparison would require acknowledging varying ward layouts, differences in patient visibility and staffing levels, variables that currently go unmeasured. Richer, context-aware measurement could also sharpen the evaluation of prevention interventions, an area where the evidence base remains weak despite decades of guideline production.

For patients and families, the message is not that falls data is worthless, but that it must be read with care. Falls in hospital remain a serious threat, and measuring them is the first step towards preventing them. What this review demonstrates is that the first step itself needs scrutiny: half of the studies in the field do not even define their central event, and the standard denominator ignores the very conditions that make falls likely. The Nottingham team, funded by the Engineering and Physical Sciences Research Council and supported by NIHR research centres, has produced a rare thing in patient safety research, a study that questions the ruler rather than the measured. If hospitals and health systems take its findings seriously, the future of fall prevention may depend less on counting falls and more on understanding the systems in which they occur.

Subject of Research: Measurement and benchmarking of inpatient falls in hospitals using a systems perspective

Article Title: A systematic review of the measurement of falls in hospitals

Article References: A systematic review of the measurement of falls in hospitals. (n.d.). https://doi.org/10.1186/s12913-026-15780-9

Image Credits: AI Generated

DOI: 10.1186/s12913-026-15780-9

Keywords: hospital falls, patient safety, systematic review, occupied bed days, benchmarking, SEIPS model, human factors, NHS, ward design, staffing levels, falls measurement, health services research