The Impact of Nurse Staffing Levels on Patient Safety Outcomes: A Systematic Review of the Literature

This is a sample Master’s (MSc) dissertation written by an AHC subject expert in Nursing to illustrate distinction / first-class standard. It uses a secondary-research (systematic literature review) design: it synthesises published studies and reports no new primary data. Use it as a model for structure, critical synthesis, methodology reporting and Harvard referencing, not as material to submit as your own.

Abstract

Nurse staffing has become one of the most intensely researched and politically contested topics in health services research, driven by a persistent tension between the cost of employing registered nurses and the evidence that understaffing harms patients. This dissertation systematically reviews the international quantitative and mixed-methods literature examining the relationship between nurse staffing levels and patient safety outcomes in adult acute hospital settings. Guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) framework, a structured search of MEDLINE, CINAHL, the Cochrane Library and Scopus was conducted, supplemented by reference-list and grey-literature searching. Studies were screened against explicit inclusion and exclusion criteria, appraised for methodological quality using the Critical Appraisal Skills Programme and Mixed Methods Appraisal Tool instruments, and synthesised thematically because clinical and statistical heterogeneity precluded meta-analysis. A narrative account of the search identified a body of large observational cohort studies, cross-sectional multi-site surveys, systematic reviews and two influential policy-driven natural experiments. Four themes were constructed from the synthesis: the association between registered-nurse staffing and patient mortality; the mediating role of missed or unfinished nursing care; the relationship between staffing and specific adverse events such as falls, pressure ulcers and healthcare-associated infection; and the influence of contextual factors including skill mix, the practice environment and shift patterns. The review argues that the evidence for a relationship between lower registered-nurse staffing and worse safety outcomes is now substantial, consistent in direction and biologically plausible, with missed care emerging as the most credible explanatory mechanism, but that the observational nature of most studies, inconsistent measurement and a scarcity of economic evaluation continue to limit causal certainty. Recommendations are offered for policy, nursing management and future research, including the prioritisation of prospective and quasi-experimental designs and the standardised reporting of staffing exposures. The dissertation contributes an integrated appraisal of a fragmented literature and clarifies where the evidence is strong enough to guide practice and where it remains provisional.

Keywords: nurse staffing, nurse-to-patient ratio, patient safety, missed nursing care, mortality, systematic review, skill mix.

Table of Contents

1. Introduction 1.1 Background to the Study 1.2 Problem Statement 1.3 Research Aim and Objectives 1.4 Research Questions 1.5 Significance of the Study 1.6 Structure of the Dissertation 2. Background and Context 2.1 Defining Nurse Staffing and Patient Safety 2.2 The Policy Context 2.3 Theoretical Frameworks 2.4 Rationale for a Systematic Review 3. Methodology 3.1 Research Philosophy and Design 3.2 Databases and Information Sources 3.3 Search Strategy 3.4 Inclusion and Exclusion Criteria 3.5 Study Selection (PRISMA) 3.6 Data Extraction 3.7 Quality Appraisal 3.8 Synthesis Approach 3.9 Ethical Considerations 4. Findings 4.1 Overview of Included Studies 4.2 Theme One: Staffing and Mortality 4.3 Theme Two: Missed and Unfinished Care as Mechanism 4.4 Theme Three: Staffing and Specific Adverse Events 4.5 Theme Four: Skill Mix, Practice Environment and Shift Patterns 5. Discussion 6. Conclusion and Recommendations 7. References

Chapter 1: Introduction

1.1 Background to the Study

The idea that having enough nurses at the bedside keeps patients safe is intuitively obvious, yet it has taken several decades of research to move it from professional conviction to an evidence base capable of informing policy. Nurses are the largest professional group in almost every health system and the staff most continuously present with patients. They administer medication, monitor for clinical deterioration, coordinate care between specialities and perform the surveillance function on which the early detection of complications depends. When there are too few nurses for the number and acuity of patients, the argument runs, this surveillance is degraded, essential care is delayed or omitted, and preventable harm follows. Translating that argument into defensible evidence, however, has proved methodologically demanding, because staffing is not randomly assigned to patients and because the outcomes of interest — death, serious complications, patient experience — are shaped by innumerable factors beyond the number of nurses on a shift.

The modern evidence base is usually traced to two landmark studies published within months of one another at the turn of the millennium. Needleman et al. (2002), analysing administrative data from hundreds of United States hospitals, reported that higher levels of care by registered nurses were associated with lower rates of adverse outcomes including urinary tract infection, pneumonia, shock and failure to rescue. In the same year Aiken et al. (2002) reported, from a survey of more than ten thousand nurses linked to surgical patient records in Pennsylvania, that each additional patient per nurse was associated with a seven per cent increase in the likelihood of a patient dying within thirty days of admission and a comparable increase in failure to rescue. These findings gave empirical weight to what had been a largely professional claim, and they catalysed a research programme that now spans dozens of countries and hundreds of studies.

In the two decades since, the field has grown both in volume and in sophistication. Large international consortia such as RN4CAST have collected standardised data across European and other health systems (Sermeus et al., 2011; Aiken et al., 2014); systematic reviews and meta-analyses have attempted to pool the accumulating evidence (Kane et al., 2007; Shekelle, 2013); and attention has shifted from simply demonstrating an association towards understanding the mechanism through which staffing affects outcomes, with the concept of missed or unfinished nursing care emerging as a leading candidate (Ball et al., 2014; Griffiths et al., 2018). At the same time, high-profile care failures such as those examined by the Mid Staffordshire NHS Foundation Trust Public Inquiry (Francis, 2013) placed staffing at the centre of policy debate in the United Kingdom, prompting national guidance and, in some jurisdictions, legislated minimum ratios.

1.2 Problem Statement

Despite this substantial and growing literature, the evidence remains difficult to interpret and to apply. The great majority of studies are observational, which leaves open the possibility that hospitals which invest in nursing also differ systematically in other respects that influence outcomes. Staffing itself is measured in many different ways — patients per nurse, nursing hours per patient day, registered-nurse skill mix, whole-time equivalents — which complicates comparison across studies. Outcomes range from all-cause mortality to specific nurse-sensitive indicators, and are drawn from data sources of varying reliability. The result is a body of evidence that is broadly consistent in its direction but heterogeneous in its detail, and about which reasonable disagreement persists regarding the strength of the causal claim and the appropriate policy response. There is a continuing need for integrative appraisal that draws the fragmented literature together, weighs its methodological quality and clarifies both what can be concluded with confidence and what remains uncertain. This dissertation responds to that need through a systematic review of the international evidence.

1.3 Research Aim and Objectives

The aim of this dissertation is to systematically review and critically synthesise the international research evidence on the relationship between nurse staffing levels and patient safety outcomes in adult acute hospital settings.

The objectives are:

1. To identify, through a structured and reproducible search, the primary studies and systematic reviews examining the relationship between nurse staffing levels and patient safety outcomes. 2. To appraise the methodological quality of the included studies using recognised critical-appraisal instruments. 3. To synthesise the evidence thematically in order to characterise the nature, direction and consistency of the relationship between staffing and safety. 4. To evaluate the mechanisms proposed to explain that relationship, in particular the role of missed nursing care. 5. To draw evidence-based conclusions and formulate recommendations for policy, nursing management and future research.

1.4 Research Questions

The review is guided by one primary and three subsidiary questions:

• Primary: What is the relationship between nurse staffing levels and patient safety outcomes in adult acute hospital settings?
• Subsidiary 1: Through what mechanisms is nurse staffing understood to influence patient safety?
• Subsidiary 2: How do contextual factors such as skill mix, the practice environment and shift patterns modify the staffing–safety relationship?
• Subsidiary 3: What is the methodological strength of the available evidence, and what are its principal limitations?

1.5 Significance of the Study

The significance of the topic is at once clinical, economic and political. Clinically, if staffing shapes safety, then decisions about establishment size are decisions about patient survival and morbidity, and belong squarely within the domain of clinical governance rather than being treated as purely financial. Economically, registered-nurse staffing is one of the largest controllable costs in any hospital budget, so the question of how much nursing is enough carries substantial resource implications and cannot be answered by reference to safety alone. Politically, staffing has become a focus of workforce campaigning, professional advocacy and, increasingly, legislation, with mandated minimum ratios adopted in California, Victoria and Queensland and safe-staffing legislation enacted in parts of the United Kingdom. A rigorous synthesis of the evidence is therefore of direct relevance to nurse managers setting establishments, to executives balancing safety against cost, and to policymakers weighing the case for regulation. For the discipline of nursing, the topic also carries symbolic weight, because it concerns the demonstration that nursing work has measurable, and in some cases life-or-death, consequences for patients.

1.6 Structure of the Dissertation

The dissertation proceeds in six chapters. Chapter Two sets out the background and conceptual context, defining the key constructs, sketching the policy landscape and introducing the theoretical frameworks that inform the interpretation of the evidence. Chapter Three describes the systematic review methodology in detail, including the search strategy, selection process, quality appraisal and approach to synthesis. Chapter Four presents the findings as a thematic synthesis organised around four themes. Chapter Five discusses the findings in relation to the research questions, the wider literature and the strength of the causal claim. Chapter Six concludes and offers recommendations for policy, practice and future research.

Chapter 2: Background and Context

2.1 Defining Nurse Staffing and Patient Safety

Any review of this literature must begin by clarifying two constructs that are more complicated than they first appear. Nurse staffing is not a single variable but a family of related measures, and the choice between them has consequences for what a study can show. The most intuitive measure is the nurse-to-patient ratio, usually expressed as the number of patients per registered nurse on a shift; this is the measure used in mandated-ratio policies and in much of the North American literature. A second common measure is nursing hours per patient day, which aggregates staffing over a longer period and is less sensitive to shift-level fluctuation. A third is skill mix, the proportion of the nursing workforce that is registered as opposed to comprising healthcare assistants or unlicensed support workers. Each captures a different facet of the staffing construct: ratios emphasise the immediate workload of an individual nurse, hours per patient day emphasise the total nursing resource available to a patient over a stay, and skill mix emphasises the level of professional judgement embedded in that resource. Studies that measure staffing differently are, in an important sense, asking slightly different questions, and this heterogeneity is a recurring theme in the appraisal that follows.

Patient safety is likewise a broad construct. The Institute of Medicine’s foundational definition frames safety as freedom from accidental injury during the process of care, and in the nursing literature this is operationalised through so-called nurse-sensitive outcomes — those most plausibly influenced by the quantity and quality of nursing care. These include failure to rescue (death following a complication that might have been detected and treated), in-hospital and thirty-day mortality, healthcare-associated infections, pressure ulcers, patient falls, medication administration errors and nurse-rated quality and safety of care. Not all of these are equally proximate to nursing: mortality is a distal outcome shaped by many non-nursing factors, whereas missed observations or omitted care are more directly attributable to staffing. The review therefore treats the range of outcomes as a spectrum of proximity to nursing action rather than as an undifferentiated category.

2.2 The Policy Context

The research literature has developed in close dialogue with policy, and the two cannot be fully separated. The most consequential policy intervention has been the introduction of mandated minimum nurse-to-patient ratios. California implemented statewide minimum ratios in medical-surgical and other units from 2004, creating a natural experiment that has been extensively studied. More recently, the Australian state of Queensland introduced minimum ratios in 2016, and the subsequent evaluation by McHugh et al. (2021) provided some of the strongest quasi-experimental evidence in the field. In the United Kingdom, the failures at Mid Staffordshire, where inadequate staffing was implicated in poor and sometimes dangerous care, prompted the Francis Report (Francis, 2013) and a subsequent focus on safe staffing, including guidance from the National Institute for Health and Care Excellence (NICE, 2014) and safe-staffing legislation in Wales and Scotland. Professional bodies, notably the Royal College of Nursing (RCN, 2021), have campaigned for staffing to be given legal underpinning and have published position statements marshalling the evidence. This policy activity matters for the review because it has shaped both the questions researchers ask and the data available to answer them; the mandated-ratio jurisdictions in particular have generated the closest approximations to experimental evidence.

2.3 Theoretical Frameworks

Several theoretical frameworks help to explain why staffing might affect safety and to organise the interpretation of the evidence. The most influential is the concept of failure to rescue, articulated in the work of Aiken and colleagues, which holds that complications occur in hospitalised patients regardless of nursing, but that whether a complication proves fatal depends on its timely detection and management — a surveillance function performed largely by nurses. On this account, adequate staffing does not prevent complications so much as prevent them from becoming fatal, which is why failure to rescue rather than complication incidence is often the more staffing-sensitive outcome.

A second and increasingly central framework is that of missed, unfinished or rationed nursing care. Where nursing time is insufficient for the required workload, nurses must prioritise, and certain activities — repositioning, mouth care, timely observations, patient education, comfort and communication — are deferred or omitted. Missed care thus functions as the proximal mechanism linking the distal cause (low staffing) to the distal effect (adverse outcomes). This concept, developed by Kalisch et al. (2009) and elaborated internationally through the RN4CAST and Missed Care programmes (Ball et al., 2014; Griffiths et al., 2018), provides a plausible causal pathway that the older mortality studies could only infer.

A third framework, drawn from the wider patient-safety literature, is the systems perspective associated with Reason’s model of organisational accidents, in which harm results from the alignment of latent conditions and active failures rather than from individual error alone. Understaffing is best conceived within this framework as a latent condition that weakens the defences of the system and increases the probability that active failures will reach the patient. Together these frameworks move the discussion beyond a simple correlation towards a mechanistic account, and they inform the thematic structure of the findings chapter.

2.4 Rationale for a Systematic Review

Given the volume, heterogeneity and policy salience of this literature, a systematic review is the appropriate design. A narrative or unsystematic review risks selective citation and cannot demonstrate that the evidence has been identified and appraised in a reproducible way. A systematic review, by contrast, applies an explicit and documented method to the identification, selection, appraisal and synthesis of studies, minimising bias and enabling the reader to judge the trustworthiness of the conclusions. Because the primary studies vary so widely in their measures and populations, a formal meta-analysis is not attempted; instead a thematic synthesis is used, an approach well suited to integrating heterogeneous quantitative and mixed-methods evidence around a set of interpretive themes.

Chapter 3: Methodology

3.1 Research Philosophy and Design

This dissertation adopts a secondary-research design in the form of a systematic literature review. No primary data were collected; the study synthesises existing published research. Philosophically, the review sits broadly within a post-positivist tradition, in that it treats the quantitative outcome studies as generating fallible but cumulative knowledge about a real relationship between staffing and safety, while acknowledging, in a manner sympathetic to critical realism, that this relationship is context-dependent and mediated by mechanisms such as missed care that are not directly observable in outcome data. This positioning has practical consequences for the synthesis: it justifies attending not only to whether studies find an association but to the mechanisms and contexts that explain when and why the association holds, which is the rationale for the thematic rather than purely aggregative approach.

The review was designed and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 statement (Page et al., 2021), which provides a checklist and flow diagram for the transparent conduct and reporting of systematic reviews. Because the review integrates quantitative and mixed-methods studies around interpretive themes rather than pooling effect sizes, elements of the integrative review approach of Whittemore and Knafl (2005) and the thematic synthesis method of Thomas and Harden (2008) were also incorporated.

3.2 Databases and Information Sources

Four electronic bibliographic databases were searched to balance sensitivity and relevance. MEDLINE (via Ovid) was searched as the principal biomedical database; the Cumulative Index to Nursing and Allied Health Literature (CINAHL) was searched as the leading nursing-specific database; the Cochrane Library was searched for systematic reviews and any controlled trials; and Scopus was searched as a large multidisciplinary database to capture health-services and policy literature that the clinical databases might miss. To reduce the risk of publication and database bias, these were supplemented by three additional sources of studies: backward citation searching of the reference lists of all included studies and relevant reviews; forward citation searching of the seminal studies; and targeted grey-literature searching of the websites of the Royal College of Nursing, the National Institute for Health and Care Excellence and comparable bodies for reports and position statements.

3.3 Search Strategy

The search strategy was constructed around the population and exposure and outcome of interest, using a combination of controlled vocabulary (for example MeSH terms such as “Nursing Staff, Hospital” and “Personnel Staffing and Scheduling”) and free-text keywords, combined with the Boolean operators AND and OR. Three concept blocks were built and then combined. The first, capturing the exposure, combined terms such as “nurse staffing”, “nurse-to-patient ratio”, “nursing hours”, “skill mix”, “staffing level” and “workforce”. The second, capturing the outcome, combined “patient safety”, “mortality”, “failure to rescue”, “adverse event”, “medication error”, “pressure ulcer”, “falls”, “healthcare-associated infection”, “missed care” and “quality of care”. A third block restricted the setting to acute or hospital care. Truncation and wildcard symbols were used to capture variant spellings and word endings, and search terms were adapted to the syntax of each database. The full search string for each database was recorded to ensure reproducibility, in line with PRISMA reporting expectations. Searches were limited to English-language publications and, to capture the modern evidence base while allowing the seminal turn-of-the-millennium studies to be included, to the period from 2000 onwards.

3.4 Inclusion and Exclusion Criteria

Explicit criteria were defined a priori to guide selection and to make the process reproducible. Studies were included if they: examined a relationship between a measure of nurse staffing (ratio, hours per patient day, skill mix or equivalent) and one or more patient safety outcomes; were conducted in adult acute or hospital inpatient settings; used a quantitative or mixed-methods design (including cohort, cross-sectional, case-control, quasi-experimental and systematic-review designs); and were published in English in a peer-reviewed journal or as a substantive report from a recognised body. Studies were excluded if they: were set exclusively in paediatric, neonatal, long-term care, community or primary-care settings, where the staffing–outcome relationship differs materially; examined only nurse outcomes such as burnout or turnover without a patient safety outcome; were editorials, commentaries, letters or conference abstracts without extractable data; or were not available in English. The restriction to adult acute settings was made to preserve the conceptual coherence of the synthesis, since the mechanisms and outcomes in, for example, neonatal intensive care differ substantially from those on adult wards.

3.5 Study Selection (PRISMA)

The selection process followed the PRISMA flow. Records retrieved from the four databases were exported to reference-management software and de-duplicated. The remaining records were screened in two stages. In the first stage, titles and abstracts were screened against the inclusion and exclusion criteria, and records that were clearly irrelevant were removed. In the second stage, the full texts of the potentially eligible records were retrieved and assessed in detail against the criteria, with the reason for each exclusion at this stage recorded. Studies identified through citation and grey-literature searching were subjected to the same full-text assessment. The outcome of this process is conventionally summarised in a PRISMA flow diagram reporting the number of records identified, screened, assessed for eligibility, excluded with reasons, and finally included; in this review the process moved from a large initial yield through progressive screening to a final set of studies spanning large cohort analyses, multi-site cross-sectional surveys, systematic reviews and two policy-based natural experiments.

3.6 Data Extraction

Data were extracted from each included study using a standardised extraction form designed for the review. For each study the form captured bibliographic details; country and setting; design; sample size and unit of analysis (patient, nurse, ward or hospital); the staffing measure used; the outcomes measured and their data source; the analytic method and any adjustment for confounding; the principal findings; and the authors’ stated limitations. Extracting these fields in a structured way served two purposes: it enabled the comparison of studies that used different measures and reported results in different metrics, and it disciplined the reviewer to attend to methodological features, such as case-mix adjustment, that bear directly on the credibility of a finding. Extracted data were tabulated to provide an at-a-glance summary of the evidence and to support the construction of themes.

3.7 Quality Appraisal

Because a systematic review is only as trustworthy as the studies it synthesises, each included study was appraised for methodological quality. Given the mix of designs, two complementary instruments were used. The relevant Critical Appraisal Skills Programme (CASP) checklists were applied to cohort and systematic-review studies, prompting appraisal of the appropriateness of the design, the handling of confounding, the completeness of follow-up and the precision of the results. The Mixed Methods Appraisal Tool (MMAT) of Hong et al. (2018) was used for studies combining quantitative and qualitative components. Appraisal was used not as a mechanical scoring exercise leading to the exclusion of weaker studies, but to weight the contribution of each study to the synthesis, so that findings resting on large, well-adjusted cohort or quasi-experimental designs were given more interpretive weight than those from small, unadjusted cross-sectional surveys. The principal quality concerns encountered — residual confounding, the cross-sectional conflation of exposure and outcome, and variable measurement of staffing — are discussed in the findings and discussion chapters where they bear on particular conclusions.

3.8 Synthesis Approach

A formal meta-analysis was considered but rejected on the grounds of heterogeneity. The included studies differed too greatly in their staffing measures, outcomes, populations and analytic methods for the pooling of effect estimates to be meaningful; combining, for example, an odds ratio for thirty-day mortality per additional patient per nurse with a correlation between nursing hours and pressure-ulcer rates would produce a spuriously precise but substantively uninterpretable summary. Instead, a thematic synthesis following the approach of Thomas and Harden (2008) was undertaken. The findings of each study were read closely and coded; codes were grouped into descriptive themes; and these were developed into the four analytical themes that structure Chapter Four. This approach preserves the richness and context of the individual studies while allowing patterns across them to be identified, and it accommodates the mixed quantitative and qualitative nature of the evidence, including the missed-care literature that supplies the mechanistic account.

3.9 Ethical Considerations

As a secondary study using only published data, the review did not require ethical approval and posed no risk to participants. Nonetheless, ethical conduct was observed in the integrity of the process: the search and selection were reported transparently to avoid selective citation; the appraisal was conducted honestly, including acknowledgement of evidence that complicates the review’s conclusions; and all sources were accurately cited to avoid plagiarism and to give due credit. The commitment to representing the evidence fairly, including its limitations, is itself an ethical obligation in a field where the conclusions carry implications for patient safety and for the allocation of scarce resources.

Chapter 4: Findings

4.1 Overview of Included Studies

The synthesised literature is dominated by large observational studies conducted in high-income health systems, principally the United States, the United Kingdom, other European countries and Australia. Three design types recur. The first is the large retrospective cohort or administrative-data study, which links routinely collected patient outcomes to measures of staffing, exemplified by the work of Needleman et al. (2002, 2011). The second is the multi-site cross-sectional survey, which combines nurse-reported data on staffing and the practice environment with patient outcomes drawn from administrative sources, exemplified by Aiken et al. (2002, 2014) and the RN4CAST consortium (Sermeus et al., 2011). The third is the systematic review and meta-analysis, which pools or narratively integrates the primary studies, exemplified by Kane et al. (2007) and by the reviews of missed care (Recio-Saucedo et al., 2018). To these are added two policy-driven natural experiments — the Californian mandated-ratio studies and the Queensland evaluation of McHugh et al. (2021) — which, by exploiting an externally imposed change in staffing, come closer than the observational studies to supporting causal inference. The four themes constructed from this body of evidence are presented in turn.

4.2 Theme One: Staffing and Mortality

The relationship between registered-nurse staffing and patient mortality is the most extensively studied and, arguably, the most consequential aspect of this literature, and the evidence here is both substantial and broadly consistent. The foundational finding is that of Aiken et al. (2002), whose analysis of surgical patients in Pennsylvania hospitals reported that each additional patient in a nurse’s average workload was associated with a seven per cent increase in the odds of death within thirty days of admission and a seven per cent increase in failure to rescue. The importance of this study lies not only in its scale but in its articulation of the failure-to-rescue mechanism: nurses do not prevent surgical complications, but adequate staffing allows those complications to be detected and acted upon before they become fatal.

This finding has been replicated across diverse health systems, lending it a robustness that no single study could provide. The RN4CAST study of Aiken et al. (2014), analysing data from more than four hundred thousand surgical patients across three hundred hospitals in nine European countries, found that an increase in a nurse’s workload by one patient was associated with a seven per cent increase in the likelihood of an inpatient dying within thirty days of admission — a figure strikingly close to the earlier Pennsylvania estimate despite the different setting and time period. The same study reported that every ten per cent increase in the proportion of nurses holding a bachelor’s degree was associated with a seven per cent decrease in the likelihood of death, introducing education as well as headcount into the staffing–mortality relationship and complicating any simple reading based on numbers alone.

The English evidence points in the same direction and thereby extends the finding beyond the United States. Rafferty et al. (2007), analysing cross-sectional survey data from more than three thousand nurses linked to the discharge records of surgical patients across thirty English hospitals, reported that patients and nurses in hospitals with the most favourable staffing levels experienced substantially better outcomes than those in the least favourably staffed hospitals, with patients in the highest-workload quartile facing a markedly higher likelihood of death and failure to rescue than those in the lowest. That an English health system organised and funded quite differently from the American one should yield a comparable pattern is itself evidentially significant, because it reduces the plausibility that the association is peculiar to the market-based financing or organisational features of any single country.

The mortality evidence is strengthened by studies that address the observational literature’s central weakness, namely the possibility that hospitals with better staffing differ systematically in other ways. Needleman et al. (2011) took an important step in this direction by examining staffing at the level of the individual shift and patient rather than the hospital, analysing whether exposure to shifts with below-target staffing was associated with mortality within a single institution. They found that exposure to units with staffing below target was associated with increased mortality, as was high patient turnover, a design that reduces the confounding introduced by comparing different hospitals because patients within the same institution are compared across better- and worse-staffed shifts. This shift-level approach represents a methodological advance and provides some of the more credible evidence that the association is not merely an artefact of hospital-level differences.

The strongest evidence in this theme, in terms of causal inference, comes from the policy natural experiments. McHugh et al. (2021), evaluating the introduction of mandated minimum nurse-to-patient ratios in Queensland, Australia, compared hospitals subject to the new ratios with comparison hospitals that were not, using a difference-in-differences design. They reported that the intervention hospitals achieved improvements in staffing and that these were accompanied by reductions in mortality and readmissions, alongside an estimated cost saving from avoided readmissions and length of stay that offset the cost of the additional nurses. Because the change in staffing was externally imposed rather than chosen by the hospitals, this study comes closer than the observational literature to demonstrating that changing staffing changes outcomes, and its finding that the intervention was cost-neutral or cost-saving is particularly significant for the policy debate.

The systematic-review evidence corroborates the primary studies while adding nuance. Kane et al. (2007), in a meta-analysis for the United States Agency for Healthcare Research and Quality, found that higher registered-nurse staffing was associated with lower hospital-related mortality, with the association strongest in intensive care and surgical patients, but cautioned that the observational designs of the included studies precluded firm causal conclusions and that the magnitude of the association varied. Shekelle (2013), reviewing the evidence as a candidate patient-safety strategy, similarly concluded that the association between higher nurse staffing and lower mortality was consistent across a large number of observational studies but that residual confounding could not be excluded. The convergence of these appraisals is instructive: they accept the consistency and plausibility of the association while withholding the strongest causal language, a judgement this review shares.

Taken as a whole, the mortality theme presents a picture of remarkable consistency. Across countries, decades, data sources and analytic strategies, lower registered-nurse staffing is associated with higher patient mortality, and the magnitude of the association clusters around a similar figure of roughly a seven per cent increase in the odds of death per additional patient per nurse. The failure-to-rescue mechanism gives the association biological plausibility, the shift-level and quasi-experimental designs address some of the confounding concerns, and the policy evaluations suggest that the relationship is amenable to intervention. The residual uncertainty concerns causation rather than association, and turns on the observational nature of most of the evidence — a matter taken up in the discussion.

4.3 Theme Two: Missed and Unfinished Care as Mechanism

If the mortality theme establishes that staffing is associated with the most serious outcome, the missed-care theme addresses the question of how, and it is here that the literature has advanced most notably in the past decade. The concept of missed, unfinished or rationed nursing care refers to the necessary nursing activities that are delayed, partially completed or omitted when nursing time is insufficient for the workload. Kalisch et al. (2009) provided an early conceptualisation, identifying commonly missed activities including ambulation, turning, timely medication administration, feeding, patient teaching, emotional support and surveillance, and arguing that missed care represents an error of omission that mediates between structural conditions such as staffing and patient outcomes.

The empirical development of this concept owes much to the RN4CAST and subsequent missed-care programmes. Ball et al. (2014), analysing data from English hospitals within the RN4CAST study, found that missed care was widespread, with nurses reporting that one or more necessary care activities had been left undone on their last shift in the large majority of cases, and, crucially, that the amount of missed care increased as the number of patients per nurse increased. This established the first half of the causal chain: lower staffing predicts more missed care. The activities most frequently reported as missed — comfort and talking with patients, educating patients, and developing care plans — are those that are less immediately visible and more easily deferred under time pressure, while activities perceived as more urgent are protected.

The second half of the chain — that missed care in turn predicts worse outcomes — has been more difficult to demonstrate but is increasingly supported. Recio-Saucedo et al. (2018), in a systematic review of the impact of nursing care left undone on patient outcomes, found that missed care was associated with a range of adverse outcomes including medication errors, infections, falls, pressure ulcers, readmissions and patient dissatisfaction, although they noted that much of the evidence relied on nurse-reported rather than independently measured outcomes and that the field was still maturing. Ball et al. (2018), returning to the RN4CAST surgical dataset, went further by examining missed care as a mediator, and found that the association between lower nurse staffing and higher post-operative mortality was partly explained by the amount of care left undone, providing direct empirical support for the mediating role that the concept had been theorised to play.

The systematic review by Griffiths et al. (2018) of the association between nurse staffing and the omission of nursing care consolidated this literature, concluding that there was consistent evidence that lower registered-nurse staffing was associated with more missed care, that the relationship was more consistent for registered nurses than for healthcare assistants, and that missed care offered a plausible and increasingly evidenced pathway from staffing to harm. The subsequent Missed Care Study reported by Griffiths et al. (2019), which linked staffing and mortality within a single English hospital trust, found that days on which registered-nurse staffing was low were associated with increased risk of death, and that this risk was not offset by higher levels of healthcare-assistant staffing — a finding that both supports the mediating role of registered-nurse care specifically and cautions against substituting unregistered for registered staff.

The strength of this theme lies in its provision of a coherent and plausible mechanism, which is precisely what the earlier mortality studies lacked. By demonstrating both that low staffing increases missed care and that missed care is associated with adverse outcomes, and by beginning to test missed care formally as a mediator, this literature converts a bare statistical association into a causal story with an identifiable intervening process. Its principal weakness is measurement: missed care is usually captured by nurse self-report, which is subject to recall and social-desirability bias, and the outcomes to which it is linked are sometimes also nurse-reported, raising the possibility of common-method bias. Jones et al. (2015), in a review of the unfinished-care literature, drew attention to the proliferation of overlapping terms and instruments — missed care, care left undone, rationed care, implicit rationing — and argued that conceptual and measurement inconsistency was hampering the accumulation of evidence, a caution that tempers but does not overturn the theme’s contribution.

4.4 Theme Three: Staffing and Specific Adverse Events

Beyond mortality and its mechanism, a substantial body of evidence links staffing to specific nurse-sensitive adverse events, and this theme both broadens the safety picture and reveals its unevenness. The foundational study again is Needleman et al. (2002), which reported that higher registered-nurse staffing was associated with lower rates of urinary tract infection, upper gastrointestinal bleeding, hospital-acquired pneumonia, shock or cardiac arrest and failure to rescue among medical patients, though the associations were weaker and less consistent for surgical patients and for some outcomes. The pattern suggests that staffing is more strongly related to those complications whose prevention or early detection depends on nursing surveillance and basic care than to those driven predominantly by medical or surgical factors.

Pressure ulcers and falls, being highly nurse-sensitive, feature prominently. The meta-analysis by Kane et al. (2007) found that higher registered-nurse staffing was associated with lower rates of these events, and numerous single-site and multi-site studies have reported similar associations, consistent with the theoretical expectation that the repositioning, mobilisation, toileting and surveillance that prevent pressure damage and falls are among the first activities to be rationed when staffing is short. The Australian study by Twigg et al. (2011), which examined the introduction of a nursing-hours-per-patient-day staffing method in Western Australia, reported reductions in several nurse-sensitive outcomes including mortality, failure to rescue and some hospital-acquired complications following the increase in staffing, providing quasi-experimental support for a causal reading in relation to these events.

Healthcare-associated infection presents a more complex picture. Several studies report associations between lower staffing and higher infection rates, plausibly mediated by lapses in hygiene, catheter care and surveillance under time pressure, and the biological pathway is credible. However, infection outcomes are influenced by many factors beyond nursing, including antimicrobial stewardship, device management protocols and case mix, so the staffing signal is harder to isolate, and the evidence, while suggestive, is less consistent than for mortality or pressure ulcers.

Medication administration errors, though intuitively linked to staffing through the pressure and interruption that accompany busy shifts, yield the least consistent evidence in this theme. Some studies report more errors under lower staffing, but the outcome is notoriously difficult to measure, relying either on incident reporting, which is affected by reporting culture and may paradoxically rise when staffing improves and reporting increases, or on direct observation, which is resource-intensive and rarely conducted at scale. This measurement difficulty, rather than a genuine absence of relationship, is the more likely explanation for the inconsistency, and it illustrates a general point that runs through this theme: the strength and consistency of the staffing–outcome association depends heavily on how proximate the outcome is to nursing action and how reliably it can be measured.

The overall contribution of this theme is to show that the staffing–safety relationship is not confined to mortality but extends across a range of complications, while also revealing that it is uneven. It is strongest and most consistent for outcomes that are both nurse-sensitive and reliably measured, such as failure to rescue and pressure ulcers, and weaker or more equivocal for outcomes that are either driven substantially by non-nursing factors, such as some infections, or difficult to measure, such as medication errors. This unevenness is not evidence against the relationship but a predictable consequence of the varying proximity of different outcomes to nursing care.

4.5 Theme Four: Skill Mix, Practice Environment and Shift Patterns

The final theme concerns the contextual and compositional factors that modify the staffing–safety relationship, and it is important because it cautions against reducing the issue to headcount alone. Three factors recur in the literature: skill mix, the practice environment, and shift patterns.

Skill mix — the proportion of the nursing workforce that is registered rather than comprising healthcare assistants — emerges as a critical qualifier. The recurring finding, articulated most clearly by Aiken et al. (2014) and reinforced by Griffiths et al. (2016, 2019), is that it is registered-nurse staffing specifically, rather than total nursing staff, that drives the safety benefit, and that increasing the number of lower-cost healthcare assistants does not compensate for a shortfall in registered nurses. The Missed Care Study of Griffiths et al. (2019) found that low registered-nurse staffing was associated with increased mortality and that higher healthcare-assistant staffing did not offset this risk; indeed some analyses suggest that a richer assistant-to-registered ratio may be associated with worse outcomes, possibly because it reflects the substitution of unregistered for registered labour. This finding carries direct and uncomfortable policy implications, since substituting assistants for registered nurses is a common cost-containment strategy, and the evidence suggests it may compromise safety.

The practice environment — encompassing managerial support, nurse involvement in decision-making, the quality of nurse–physician relationships and organisational investment in care quality — is the second modifying factor. Studies within the RN4CAST tradition consistently find that the association between staffing and outcomes is stronger or weaker depending on the quality of the practice environment, and that good staffing and a good practice environment together produce better outcomes than either alone. Aiken and colleagues’ work suggests that improving the practice environment can amplify the benefit of a given level of staffing, implying that staffing operates within an organisational context rather than in isolation. This is theoretically important because it locates staffing within the systems perspective introduced in Chapter Two: numbers matter, but so do the conditions under which those numbers work.

Shift patterns, particularly the widespread adoption of twelve-hour shifts, constitute the third factor and a more recent focus of research. Dall’Ora et al. (2016) and related work from the same programme found that long shifts of twelve hours or more were associated with higher levels of burnout, job dissatisfaction and, in some analyses, missed care and lower-rated quality, raising the possibility that how nursing time is distributed across the day affects safety independently of the total hours provided. The evidence here is less mature than for staffing levels, and the outcomes studied are often intermediate (burnout, missed care) rather than patient harm, but it adds an important temporal dimension to the staffing question and suggests that the organisation of the working day is a legitimate safety concern.

The collective contribution of this theme is to refine the crude proposition that more nurses mean safer patients into a more precise and defensible claim: that more registered nurses, working in a supportive practice environment and under sustainable shift arrangements, are associated with safer patients, and that these compositional and contextual factors are not incidental but integral to the relationship. This refinement matters both scientifically, because it specifies the conditions under which the staffing effect holds, and practically, because it warns against the false economy of meeting a headline staffing number through skill dilution.

Chapter 5: Discussion

The purpose of this chapter is to interpret the four themes in relation to the research questions, to weigh the strength of the causal claim, to situate the findings within the wider literature and to acknowledge the limitations of the review.

The primary research question asked what the relationship is between nurse staffing levels and patient safety outcomes in adult acute settings. The synthesis supports a clear answer: there is a consistent, international and biologically plausible association between lower registered-nurse staffing and worse patient safety outcomes, most robustly demonstrated for mortality and failure to rescue and extending, with varying strength, to a range of nurse-sensitive adverse events. The consistency of this association is its most striking feature. That studies conducted in different countries, health systems, decades and using different data sources and analytic methods converge on a similar direction and, for mortality, a similar magnitude, is a powerful argument against the association being an artefact of any single setting or method. In the terms of the classic Bradford Hill considerations for causation, the evidence displays strength that is modest but reliable, consistency that is high, biological plausibility supplied by the surveillance and missed-care mechanisms, a coherent dose–response relationship in the observation that each additional patient per nurse incrementally raises risk, and, increasingly, experimental support from the mandated-ratio evaluations.

The first subsidiary question asked about mechanism, and here the review’s principal contribution is to show that the field has moved beyond bare association towards a credible causal pathway. The missed-care literature demonstrates that low staffing increases the omission of necessary care and that such omission is associated with adverse outcomes, and the mediation analysis of Ball et al. (2018) provides direct evidence that missed care partly explains the staffing–mortality relationship. This mechanistic account is what elevates the staffing evidence above a mere correlation, because it specifies a plausible and partly demonstrated process through which the cause produces the effect. It also has practical value: if missed care is the mechanism, then measuring and reducing missed care becomes a proximal target for intervention and monitoring, potentially more actionable than mortality, which is distal and multiply determined.

The second subsidiary question concerned contextual modifiers, and the synthesis shows that the staffing effect is conditioned by skill mix, practice environment and shift patterns. The most consequential of these findings is that the safety benefit attaches specifically to registered-nurse staffing and is not reproduced by substituting healthcare assistants. This has a direct bearing on a common managerial temptation. Because registered nurses are expensive, skill dilution is an attractive route to meeting a staffing number at lower cost, but the evidence indicates that this route may compromise the very safety the staffing is intended to protect. The practice-environment findings similarly caution against treating staffing as a purely numerical problem, since the same number of nurses produces better outcomes in a supportive organisational context. These modifiers do not weaken the staffing argument; they sharpen it, by specifying that it is registered-nurse staffing, embedded in a good environment and sustainable shift patterns, that delivers the benefit.

The third subsidiary question asked about the methodological strength of the evidence, and honesty requires acknowledging its central limitation: the great majority of studies are observational, and observational evidence, however consistent, cannot fully exclude confounding. Hospitals that staff their wards well are likely to differ from those that do not in ways — funding, leadership, medical staffing, case mix — that also affect outcomes, and although the better studies adjust for measured confounders and although the shift-level designs of Needleman et al. (2011) and Griffiths et al. (2019) address within-hospital confounding, unmeasured confounding remains possible. This is precisely why the mandated-ratio natural experiments, and above all the Queensland evaluation of McHugh et al. (2021), carry such weight: by exploiting an externally imposed change in staffing, they approximate the counterfactual that observational studies can only estimate, and their finding that increased staffing reduced mortality and readmissions at little or no net cost provides the strongest available support for a causal interpretation. The convergence of consistent observational evidence, a demonstrated mechanism and quasi-experimental confirmation is, taken together, considerably more persuasive than any single strand alone, even if the field still lacks the randomised evidence that would settle the question definitively — evidence that is, for obvious ethical reasons, unlikely ever to be generated by deliberately understaffing wards.

Several further limitations of the evidence base deserve emphasis. Measurement heterogeneity pervades the field: staffing is operationalised in incompatible ways across studies, outcomes range from reliably measured mortality to poorly measured medication errors, and missed care is captured largely by self-report. This heterogeneity both precludes meta-analysis and complicates the aggregation of findings, and it points to a need for standardised measurement. Economic evidence, though beginning to accumulate through studies such as McHugh et al. (2021), remains thin, which is a serious gap given that the policy debate ultimately turns on whether the safety benefit justifies the cost. And the evidence is overwhelmingly drawn from high-income countries, limiting its transferability to the low- and middle-income settings where staffing shortages are most acute and where the mechanisms may operate differently.

The findings of this review align closely with the conclusions of the major prior syntheses, which lends confidence to its account. Kane et al. (2007) and Shekelle (2013) reached similar judgements about the consistency of the association and the residual uncertainty about causation, and the more recent missed-care reviews (Griffiths et al., 2018; Recio-Saucedo et al., 2018) support the mechanistic reading advanced here. Where this review adds value is in integrating the mortality, mechanism, adverse-event and contextual literatures into a single appraisal and in showing how they reinforce one another: the mortality studies establish the outcome, the missed-care studies supply the mechanism, the adverse-event studies map the breadth of the effect, and the contextual studies specify the conditions, so that the whole is more compelling than the sum of its parts.

The limitations of the review itself must also be acknowledged. As the work of a single reviewer, the screening, extraction and appraisal were not independently duplicated, which introduces the possibility of selection and appraisal bias that a two-reviewer process would mitigate. The restriction to English-language publications risks language bias and may under-represent evidence from non-English-speaking systems. The decision not to conduct a meta-analysis, though justified by heterogeneity, means the review offers a narrative rather than a quantitative summary and cannot provide a pooled effect estimate. And, as with all reviews, the conclusions are constrained by the quality of the underlying studies, which for the reasons discussed remain predominantly observational. These limitations temper the strength of the claims but do not, in the reviewer’s judgement, overturn the central finding, which rests on a convergence of evidence too broad and consistent to be readily dismissed.

Chapter 6: Conclusion and Recommendations

6.1 Conclusion

This dissertation set out to systematically review the international evidence on the relationship between nurse staffing levels and patient safety outcomes in adult acute hospital settings. Drawing on a structured search, explicit selection criteria, formal quality appraisal and a thematic synthesis, it has argued that the evidence supports a consistent, plausible and increasingly well-explained association between lower registered-nurse staffing and worse patient safety, most robustly for mortality and failure to rescue and extending across a range of nurse-sensitive outcomes. The review has shown that this relationship is not a bare statistical correlation but is underpinned by an identifiable mechanism, the omission of necessary nursing care under time pressure, and that it is conditioned by skill mix, the practice environment and shift patterns, such that the safety benefit attaches specifically to registered nurses working in supportive and sustainable conditions rather than to nursing headcount alone. It has also argued, with appropriate caution, that while the observational nature of most of the evidence precludes definitive causal proof, the convergence of consistent observational findings, a demonstrated mechanism and quasi-experimental confirmation from mandated-ratio evaluations makes a causal interpretation the most reasonable reading of the evidence, and one that is sufficient to guide policy and practice even in the absence of the randomised trials that will, for ethical reasons, never be conducted.

The overarching conclusion is therefore that nurse staffing is a patient safety issue and should be governed as such. Decisions about registered-nurse establishments are, on the evidence reviewed here, decisions about patient survival and morbidity, and belong within the domain of clinical governance rather than being treated as purely financial matters to be minimised. At the same time, the review has been candid about the limits of the evidence: its measurement heterogeneity, its thin economic base and its concentration in high-income settings all represent genuine gaps that qualify the confidence with which specific staffing thresholds can be prescribed.

6.2 Recommendations for Policy

For policymakers, the evidence supports treating registered-nurse staffing as a determinant of patient safety and provides a defensible basis for regulatory attention. The mandated-ratio evaluations, particularly the finding of McHugh et al. (2021) that increased staffing reduced mortality and readmissions at little or no net cost, weaken the assumption that safe staffing is unaffordable and strengthen the case for minimum-staffing frameworks accompanied by robust evaluation. Where minimum standards are set, the evidence on skill mix argues strongly that they should specify registered-nurse staffing rather than total nursing staff, so as to prevent the safety standard being met through skill dilution. Policy should also invest in the standardised, routine collection of staffing and nurse-sensitive-outcome data, without which the ongoing monitoring and evaluation that the field requires cannot be achieved.

6.3 Recommendations for Nursing Management and Practice

For nurse managers and directors of nursing, the review suggests that establishment-setting should be evidence-informed, acuity-sensitive and protective of registered-nurse numbers, and that the temptation to substitute unregistered for registered staff as a cost-containment measure should be resisted in light of the evidence that such substitution does not deliver equivalent safety. Managers are also encouraged to monitor missed care as a proximal, actionable indicator of staffing adequacy, since it is more sensitive and more timely than mortality and provides an early warning of unsafe conditions. Attention to the practice environment — supportive leadership, nurse involvement in decision-making and sustainable shift patterns — emerges from the evidence as a complement to, not a substitute for, adequate numbers, and should form part of any staffing strategy.

6.4 Recommendations for Future Research

For researchers, the review identifies several priorities. The first is the pursuit of prospective and quasi-experimental designs, exploiting policy changes and natural experiments where randomised trials are infeasible, in order to strengthen causal inference beyond what observational studies can provide. The second is the standardisation of measurement, both of staffing exposures and of missed care, to enable the accumulation and pooling of evidence that current heterogeneity prevents. The third is the extension of economic evaluation, so that the cost-effectiveness as well as the effectiveness of staffing interventions can be established, since this is the terrain on which policy is ultimately decided. The fourth is the extension of research to low- and middle-income settings, where staffing shortages are most severe and where the transferability of the high-income evidence cannot be assumed. Pursuing these priorities would convert an already substantial and consistent evidence base into one capable of supporting confident and specific guidance on how many nurses, of what registration, are needed to keep patients safe.

References

Aiken, L.H., Clarke, S.P., Sloane, D.M., Sochalski, J. and Silber, J.H. (2002) ‘Hospital nurse staffing and patient mortality, nurse burnout, and job dissatisfaction’, JAMA, 288(16), pp. 1987–1993.

Aiken, L.H., Sloane, D.M., Bruyneel, L., Van den Heede, K., Griffiths, P., Busse, R., Diomidous, M., Kinnunen, J., Kózka, M., Lesaffre, E., McHugh, M.D., Moreno-Casbas, M.T., Rafferty, A.M., Schwendimann, R., Scott, P.A., Tishelman, C., van Achterberg, T. and Sermeus, W. (2014) ‘Nurse staffing and education and hospital mortality in nine European countries: a retrospective observational study’, The Lancet, 383(9931), pp. 1824–1830.

Ball, J.E., Bruyneel, L., Aiken, L.H., Sermeus, W., Sloane, D.M., Rafferty, A.M., Lindqvist, R., Tishelman, C. and Griffiths, P. (2018) ‘Post-operative mortality, missed care and nurse staffing in nine countries: a cross-sectional study’, International Journal of Nursing Studies, 78, pp. 10–15.

Ball, J.E., Murrells, T., Rafferty, A.M., Morrow, E. and Griffiths, P. (2014) ‘”Care left undone” during nursing shifts: associations with workload and perceived quality of care’, BMJ Quality & Safety, 23(2), pp. 116–125.

Dall’Ora, C., Griffiths, P., Ball, J., Simon, M. and Aiken, L.H. (2016) ‘Association of 12 h shifts and nurses’ job satisfaction, burnout and intention to leave: findings from a cross-sectional study of 12 European countries’, BMJ Open, 6(2), e008331.

Francis, R. (2013) Report of the Mid Staffordshire NHS Foundation Trust Public Inquiry. London: The Stationery Office.

Griffiths, P., Ball, J., Drennan, J., Dall’Ora, C., Jones, J., Maruotti, A., Pope, C., Recio-Saucedo, A. and Simon, M. (2016) ‘Nurse staffing and patient outcomes: strengths and limitations of the evidence to inform policy and practice. A review and discussion paper based on evidence reviewed for the National Institute for Health and Care Excellence Safe Staffing guideline development’, International Journal of Nursing Studies, 63, pp. 213–225.

Griffiths, P., Maruotti, A., Recio-Saucedo, A., Redfern, O.C., Ball, J.E., Briggs, J., Dall’Ora, C., Schmidt, P.E. and Smith, G.B. (2019) ‘Nurse staffing, nursing assistants and hospital mortality: retrospective longitudinal cohort study’, BMJ Quality & Safety, 28(8), pp. 609–617.

Griffiths, P., Recio-Saucedo, A., Dall’Ora, C., Briggs, J., Maruotti, A., Meredith, P., Smith, G.B. and Ball, J. (2018) ‘The association between nurse staffing and omissions in nursing care: a systematic review’, Journal of Advanced Nursing, 74(7), pp. 1474–1487.

Hong, Q.N., Fàbregues, S., Bartlett, G., Boardman, F., Cargo, M., Dagenais, P., Gagnon, M.-P., Griffiths, F., Nicolau, B., O’Cathain, A., Rousseau, M.-C., Vedel, I. and Pluye, P. (2018) ‘The Mixed Methods Appraisal Tool (MMAT) version 2018 for information professionals and researchers’, Education for Information, 34(4), pp. 285–291.

Jones, T.L., Hamilton, P. and Murry, N. (2015) ‘Unfinished nursing care, missed care, and implicitly rationed care: state of the science review’, International Journal of Nursing Studies, 52(6), pp. 1121–1137.

Kalisch, B.J., Landstrom, G.L. and Hinshaw, A.S. (2009) ‘Missed nursing care: a concept analysis’, Journal of Advanced Nursing, 65(7), pp. 1509–1517.

Kane, R.L., Shamliyan, T.A., Mueller, C., Duval, S. and Wilt, T.J. (2007) ‘The association of registered nurse staffing levels and patient outcomes: systematic review and meta-analysis’, Medical Care, 45(12), pp. 1195–1204.

McHugh, M.D., Aiken, L.H., Sloane, D.M., Windsor, C., Douglas, C. and Yates, P. (2021) ‘Effects of nurse-to-patient ratio legislation on nurse staffing and patient mortality, readmissions, and length of stay: a prospective study in a panel of hospitals’, The Lancet, 397(10288), pp. 1905–1913.

National Institute for Health and Care Excellence (2014) Safe staffing for nursing in adult inpatient wards in acute hospitals (SG1). London: NICE.

Needleman, J., Buerhaus, P., Mattke, S., Stewart, M. and Zelevinsky, K. (2002) ‘Nurse-staffing levels and the quality of care in hospitals’, New England Journal of Medicine, 346(22), pp. 1715–1722.

Needleman, J., Buerhaus, P., Pankratz, V.S., Leibson, C.L., Stevens, S.R. and Harris, M. (2011) ‘Nurse staffing and inpatient hospital mortality’, New England Journal of Medicine, 364(11), pp. 1037–1045.

Page, M.J., McKenzie, J.E., Bossuyt, P.M., Boutron, I., Hoffmann, T.C., Mulrow, C.D., Shamseer, L., Tetzlaff, J.M., Akl, E.A., Brennan, S.E., Chou, R., Glanville, J., Grimshaw, J.M., Hróbjartsson, A., Lalu, M.M., Li, T., Loder, E.W., Mayo-Wilson, E., McDonald, S., McGuinness, L.A., Stewart, L.A., Thomas, J., Tricco, A.C., Welch, V.A., Whiting, P. and Moher, D. (2021) ‘The PRISMA 2020 statement: an updated guideline for reporting systematic reviews’, BMJ, 372, n71.

Rafferty, A.M., Clarke, S.P., Coles, J., Ball, J., James, P., McKee, M. and Aiken, L.H. (2007) ‘Outcomes of variation in hospital nurse staffing in English hospitals: cross-sectional analysis of survey data and discharge records’, International Journal of Nursing Studies, 44(2), pp. 175–182.

Recio-Saucedo, A., Dall’Ora, C., Maruotti, A., Ball, J., Briggs, J., Meredith, P., Redfern, O.C., Kovacs, C., Prytherch, D., Smith, G.B. and Griffiths, P. (2018) ‘What impact does nursing care left undone have on patient outcomes? Review of the literature’, Journal of Clinical Nursing, 27(11–12), pp. 2248–2259.

Royal College of Nursing (2021) Nursing workforce standards: supporting a safe and effective nursing workforce. London: Royal College of Nursing.

Sermeus, W., Aiken, L.H., Van den Heede, K., Rafferty, A.M., Griffiths, P., Moreno-Casbas, M.T., Busse, R., Lindqvist, R., Scott, A.P., Bruyneel, L., Brzostek, T., Kinnunen, J., Schubert, M., Schoonhoven, L. and Zikos, D. (2011) ‘Nurse forecasting in Europe (RN4CAST): rationale, design and methodology’, BMC Nursing, 10, 6.

Shekelle, P.G. (2013) ‘Nurse–patient ratios as a patient safety strategy: a systematic review’, Annals of Internal Medicine, 158(5 Pt 2), pp. 404–409.

Thomas, J. and Harden, A. (2008) ‘Methods for the thematic synthesis of qualitative research in systematic reviews’, BMC Medical Research Methodology, 8, 45.

Twigg, D., Duffield, C., Bremner, A., Rapley, P. and Finn, J. (2011) ‘The impact of the nursing hours per patient day (NHPPD) staffing method on patient outcomes: a retrospective analysis of patient and staffing data’, International Journal of Nursing Studies, 48(5), pp. 540–548.

Whittemore, R. and Knafl, K. (2005) ‘The integrative review: updated methodology’, Journal of Advanced Nursing, 52(5), pp. 546–553.