Organisational factors associated with healthcare workforce development, recruitment, and retention in the United Kingdom: a systematic review

Inclusion criteria were studies conducted in the UK, published in the English language in the last five years (2018 to 2023), peer-reviewed research articles, employed quantitative research designs (e.g., cross-sectional and longitudinal), and recruited nurses and/or various allied health professionals working in all healthcare settings as the study sample. To include nurses and allied health professionals working on a diverse range of roles, there was no restriction based on their roles. Papers published in English in the last five years (2018 to 2023), adopted a quantitative design (including mixed methods paper with the intention to extract only the quantitative results), and were conducted in the UK were selected. The primary predictor or independent variable of interest was indicators related to the organisational practice environment, and the primary outcomes were recruitment, retention, intention to leave/stay, and turnover. The eligibility criteria are further presented in Table 1.

Exclusion criteria were studies conducted outside the UK, non-English-language publications, published before 2018, non-peer-reviewed sources (e.g., theses, literature reviews, editorials), non-quantitative research designs (e.g., qualitative and case studies), and other groups not falling under the specified allied health professions (such as pharmacists, clinicians, students and patients). Additionally, outcomes related to the quality of care and mental well-being were outside the scope of this review.

We used the Population Intervention, Comparison and Outcome (PICO) framework (without the optional ‘comparison’ element as this was not relevant to our research question) to guide the search [27, 28].

The National Heart, Lung and Blood Institute (NHLBI) Study Quality Assessment Tool [39] for observational cohort and cross-sectional studies was used to assess the quality of the included studies for risk of bias (Table 3). The tool comprises 15 questions that thoroughly examine various aspects of the studies, such as their aims, sample size, design, outcome and independent measures, confounding variables, and inclusion/exclusion criteria. Two reviewers (EA and PA) independently assessed each question by assigning a rating of ‘yes’ for a low risk of bias, ‘no’ for a high risk of bias, ‘Not Reported’ when no supporting information was available, and ‘NA’ if the criteria were not applicable. Any discrepancies in the quality ratings between reviewers were resolved through discussions and consensus with the third reviewer. To ensure the validity and reliability of the review’s conclusions, studies rated as ‘poor’ will be removed as their results may be unreliable.

Given the methods of analyses and outcomes reported in the included studies, a narrative synthesis, compared to a meta-analysis, was a better fit for synthesising the results. The narrative synthesis followed the methodologies proposed by the Cochrane Consumers and Communication Review Group’s Data Synthesis and Analysis document [40] and the Guidance on the Conduct of Narrative Synthesis in Systematic Reviews by the UK Economic and Social Research Council Methods Programme [41]. The results were integrated based on the reported organisational practice environment factors.

All potential records (n=4216) generated from the search of databases were collated and uploaded into Endnote© Version X8. Duplicates (n=1043) were identified and removed using the Endnote functions, leaving 3173 records. Screening of titles and abstracts resulted in the exclusion of 3001 papers. The full texts of the remaining 111 papers were screened, resulting in the exclusion of 101 papers due to inclusion criteria, study design and location. The remaining 10 papers (32–41) were included in the review (see Fig. 1 for the PRISMA flowchart showing the selection process). No additional papers were identified through the reference lists of the included studies.

A summary table of the characteristics of the included studies is provided in Table 2. Most of the studies recruited participants working in hospital settings, with two studies conducted in a care home/community nursing setting [33, 37]. While our primary focus was on nursing and allied health professionals with no restriction based on role, it is important to highlight that the studies included in our review featured a range of distinct roles within these staff groups. These encompassed clinical and non-clinical roles, frontline clinical staff, and managerial positions. To further elucidate the composition of the study samples, we have detailed the specific staff groups investigated in each study within Table 2. All the studies incorporated a quantitative component, with two adopting a mixed methods approach [31, 35]. Nearly all the studies were cross-sectional; only one had a longitudinal design [33]. The sample sizes varied across the studies, ranging from 116 to 36,850 participants, with most studies reporting nurses as their participants. The analyses employed in the included studies were varied, with correlation and regression analyses commonly reported. Multiple tools were used to assess outcomes (see Appendix 2, Additional File 1 for a description of the indicators reported and how they were measured in each study). Authors employed various measurements in their studies to capture specific dimensions of interest. For example, when investigating intention to leave, some studies used single-item questions such as “Are you considering leaving your job?” [29, 37] or multi-item Likert scale-based questions to gauge the degree of intention to leave [30, 34]. Actual turnover was assessed through self-report measures, where care home managers, for example, reported the number of staff left over a given period [33]. Retention was measured using questionnaires that examined factors influencing retirement age and timing [31] or was calculated using organisational workforce data [36].

Overall, the quality ratings varied across studies, with most studies rated as fair or good. The specific quality ratings for each study are provided in Table 3. Six of the ten papers were rated as good, while four were rated as fair. All studies clearly stated their research aim/question. The study population was also clearly specified and defined in all studies. The participation rate of eligible persons was below 50% in most studies, except for two studies where this information was not reported. All study subjects were selected or recruited from the same or similar populations, and the timeframe was considered sufficient to observe associations between exposure and outcome. Most studies did not provide sample size justification or power calculation; only four studies reported this information [29, 31, 35, 37]. Potential confounding variables were measured and adjusted for in most studies except for four [29, 36‐38].

The associations between organisational environmental factors and healthcare staff intention to leave/actual turnover/retention is summarised in Table 4.

Four of the included studies [29, 33, 34, 37] investigated associations between workplace challenges (job stress, burnout and working conditions) and staff turnover or intention to leave. The study reporting burnout (defined by the WHO as “a syndrome conceptualised as resulting from chronic workplace stress that has not been successfully managed” [42]) found no association with staff turnover [33]. One study reported a positive association between job stress (that is, the mental situation individuals may experience when they are confronted with job demands and pressures that do not align with their skills and capabilities [43]), and nurses’ intention to leave [34]. Two studies reported the effect of working conditions on intention to leave [29, 37], with one focusing on staff experience working in COVID-19 high-risk areas [29]. The COVID-19-focused study found no difference in turnover intention between those working in higher- or lower-risk areas. Another study found that staff intention to leave significantly decreased with improvements in working conditions (0.49 (0.34–0.70), p < 0.001) [37].

Workplace aggression, in this context, refers to encountering actions that pose potential harm, are actively avoided by the target, and take place during the target’s work duties [44]. Only one study investigated the impact of workplace aggression on staff intention to leave. The study by Cheng et al. found that experiencing aggression from either patients or colleagues had a damaging effect on staff intention to leave, with the impact of aggression from colleagues twice that from patients [30].

The results of the two studies that reported on moral distress were consistent. Colville et al. found that moral distress (resulting from situations when someone is aware of the correct course of action but institutional constraints create significant obstacles to following through with the right decision) predicts staff intention to leave and turnover [32]. Similarly, Witton et al. reported that moral distress was negatively correlated with intent to stay; nurses who stated they had high rates of moral distress were more likely to consider leaving their current employer [38].

One of the included studies considered job embeddedness (the degree of connection employees have to their jobs through a network of social relationships and factors [45]), with a negative association between this organisational practice environment factor and nurses’ intention to leave reported [34].

Two studies reported the effect of staff involvement in leadership and management. Quek et al. found that higher levels of distributed leadership significantly predicted lower turnover intention [35]. Robinson et al. investigated the association between involvement in improvement (measured using three indicators: the ability to make suggestions to improve their work, involvement in decisions on changes that affect their work area, and ability to make improvements in their area of work) and the retention of nurses working across different types of NHS Trusts (Community, Specialist, Mental Health and Acute). Only two indicators of improvement (ability to make suggestions to improve their work and ability to make improvements in their area of work) were significantly correlated with RN retention, and this was seen in those working in mental health and acute NHS trusts [36]. For RNs in Mental Health NHS Trusts, retention was positively correlated with their ability to make suggestions to improve their work and their ability to make improvements in their area of work. However, in Acute NHS Trusts, a negative correlation was seen between RNs’ ability to make improvements in their area of work and their retention.

Two studies reported on aspects of organisational support. The COVID-19-focused study by Blake et al. found no significant differences in turnover intentions between staff who accessed a supported wellness centre set up in UK hospitals to mitigate the psychological impact of the pandemic and those who did not [29]. The other study reported that support from managers reduced the odds of staff indicating an intention to leave [37].

With a focus on older nurses (over 55 years of age), Cleaver et al. examined the relationship between shift patterns and the odds of working beyond retirement. The odds of working beyond retirement are significantly higher if staff can reduce their working hours and choose when to work or have a fixed working pattern. However, other factors, such as cessing to work shifts, nights and weekends, were not significantly associated with the intention to work beyond retirement [31].