Understanding turnover in healthcare and welfare sectors of high-income countries: an umbrella review

Abstract

Background

This umbrella review aims to systematically synthesize the current evidence on the factors associated with personnel turnover in the healthcare and welfare sectors, the methodologies used to measure turnover, and the effectiveness of interventions designed to reduce turnover.

Methods

Following PRISMA-P guidelines, we analyzed data from January 2013 to July 2023 across multiple databases, including PubMed, Web of Science, CINAHL, Cochrane Database of Systematic Reviews, and Embase. Two reviewers independently screened titles and abstracts for predefined eligibility criteria. They also collaboratively examined 10% of the full texts to ensure compliance with these criteria. Study quality was evaluated using AMSTAR 2, and a Corrected Covered Area analysis was conducted.

Results

The analysis included 37 studies with a total of 511 primary studies. Turnover factors were grouped into three categories: socio-demographics and health status, work environment characteristics, and personal attitude and functioning, highlighting a wide range of factors that influence both turnover intention and actual turnover. Various measurement methodologies were identified, with a notable lack of standardization. Interventions targeting work-related factors showed mixed effectiveness, underscoring the importance of context-specific strategies.

Conclusions

The findings show the complexity and context-specific nature of turnover in healthcare and welfare sectors, requiring targeted, context-specific interventions to address the diverse factors involved. Standardization of measurement methodologies is necessary for comparing turnover rates and the effectiveness of interventions. Future research should focus on filling existing gaps, particularly in non-hospital settings, and on developing and evaluating multifaceted interventions.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-025-12966-5.

Background

The healthcare and welfare sectors face major challenges in managing their workforces. This is demonstrated by a predicted shortage of almost 18 million healthcare professionals (HCPs) by 2030 [95]. Similarly, in social services, a 2022 survey by Social Employers and the European Federation of Public Service Unions (EPSU) found that 19 out of 22 European countries reported worsening staff shortages since 2021, with nearly one-third facing severe gaps of over 10% unfilled positions [36]. In the United States, workforce instability in social services also remains a major concern, with 22% of professionals frequently considering leaving their jobs due to stress, poor work-life balance, and low job satisfaction [35]. In the healthcare sector, the situation is exacerbated by growing patient dependency due to aging populations and increased comorbidities [12, 52]. In some countries, an aging workforce and an increase in retirement increase labor shortages [12]. Additionally, it is difficult to attract new HCPs such as nurses because younger generations often view the profession as unattractive because of low salaries and job status [96]. All of these challenges have recently grown larger due to the COVID-19 pandemic, which has increased the demand for HCPs. For example, the pandemic has led to changes in the work environment, impacting their mental and physical health, which makes them more likely to consider leaving their jobs [32, 56].

In both the healthcare and welfare sectors, the increasing trend of HCPs considering to leave their positions has several consequences across patient (or client), team, and organizational levels. In the healthcare sector and at the patient level, studies have shown a link between staff turnover and negative patient outcomes, such as an increase of medical errors [69], longer hospital stays [48], and lower patient satisfaction [91]. At the team level, frequent turnover leads to a less stable work environment and lower job satisfaction among HCPs [23]. This dissatisfaction could make even more HCPs want to leave their careers, causing a loss of important knowledge and experience in the field [44, 46]. At the organizational level, turnover results in a loss of expertise, heavier workloads for the remaining staff, and higher costs for hiring and training new staff [34, 81]. For example, the cost of each individual turnover is estimated to be three times the salary of a nurse [78]. Similarly, in welfare settings, staff turnover can interrupt service continuity and undermine service quality. For example, the loss of HCPs can lead to a reduction in the efficacy of welfare programs, which could impact client well-being and satisfaction [1, 65]. In addition, staff turnover in welfare settings can lead to an unstable work environment, affecting employee morale and job satisfaction, which in turn may further exacerbate turnover rates [31, 50].

In summary, this all shows that the adverse effects of turnover are far-reaching and cannot be ignored. In order to get a better understanding of these dynamics, this umbrella review focuses on factors associated with turnover intention and actual turnover of HCPs. On the one hand, although there is a lot of research on turnover within the healthcare sector, it tends to be fragmented, often concentrating on specific settings such as acute care hospitals, or is limited to particular professions like nursing (e.g. [41]). The welfare sector, on the other hand, has not received the same level of scholarly attention. The reasons for this imbalance are multifaceted, including the diversity of welfare services which range from child welfare and family support to services for the elderly and disabled, each with its own unique challenges and operational dynamics. This diversity complicates the standardization of outcome measurements, making research in this field more challenging [1]. Moreover, the indirect impact of turnover on client outcomes in the welfare sector, such as the quality of child and family services or the efficacy of social support programs, may not be as immediately quantifiable as in health care, where outcomes like infection rates provide more clear indicators of service quality [67]. This could contribute to the comparatively lower research emphasis on turnover within welfare settings.

These challenges highlight the need to examine the underlying factors of employee turnover. Before proceeding, we want to take a closer look at employee turnover, differentiating between ‘turnover intention’ and ‘actual turnover’.

Employee turnover

Turnover intention, also known as the intention to leave, refers to the desire or plan of an employee to quit their current job within a certain time period [68]. This concept is grounded in the theory of planned behavior [2], suggesting that turnover intention is a precursor to actual turnover behavior, starting with job dissatisfaction and culminating in the physical act of leaving. Actual turnover is the act of leaving a job. The 'turnover rate' is the most common indicator used in studies to assess actual turnover, usually calculated by dividing the number of terminations by the total number of employees during a specific time frame, commonly a fiscal or calendar year [60], reflecting an organizational perspective on turnover.

Employee turnover can be described as voluntary (resignation or natural turnover like retiring) or involuntary (dismissal or lay-off by restructurings), and whether it could have been avoidable or unavoidable. It can also mean leaving the department, organization, or profession completely, or a mix of different situations [41, 54]. Adding to this complexity, next to this more individual perspective (in the case of or the nature of an individual exit), employee turnover can also focus on the organizational perspective, indicated by a turnover rate (‘how high are the turnover rates of this organization’). While some degree of turnover is both expected and beneficial to an organization, striking a balance is key. For instance, in the nursing profession, an optimal turnover rate is often considered to be between 5 and 10% [39, 55]. However, because there is not a clear and consistent way to define – and consequently measure – turnover, the estimated rate of (nursing) turnover worldwide ranges from 4 to 54% [38].

The relationship between turnover intention and actual turnover is complex, with some studies showing weak correlations while others present turnover intention as a strong predictor of actual turnover [40, 85]. Therefore, both outcomes are taken into account in this study.

Factors and interventions of turnover (intention)

In the quest to reduce staff turnover, both intended and actual turnover are influenced by a complex interplay of factors. Individual aspects such as workload, lack of autonomy, and personal demographics intertwine with group aspects such as staffing levels and managerial styles [41]. Over the past decades, several conceptual models and classification systems have been developed to understand ‘turnover’, often focusing on individual, job-related, and/or organizational factors [43]. These models identify a wide range of reasons why professionals leave or express an intention to leave. Recent research has shifted towards understanding the factors as a means to improve retention [29*, 30*, 41]. With a better understanding of these contributing factors, they can then turn their attention to interventions that have been developed to address these challenges and improve retention in the healthcare workforce. Over the years, a variety of interventions targeting different levels of the healthcare and welfare system have been developed and implemented, including strategies like improving work conditions, offering professional development opportunities, and fostering a positive organizational culture. However, the effectiveness of these interventions may vary based on the context and specific challenges faced by HCPs in different settings [4, 33, 37, 87].

Scope and terminology

Given the diversity of services, professions, and research approaches in this field, we want to clarify the scope and terminology used in this review. Although the healthcare and welfare sectors are often studied separately, they share similar workforce challenges, such as high turnover rates, work-related stress, and difficulties in recruitment and retention. In addition, in practice, the boundaries between healthcare and welfare are often blurred, for instance in long-term care and community-based services. Studying them together allows for a more comprehensive understanding of common factors influencing turnover, while also accounting for sector-specific dynamics.

In this review, the healthcare sector is defined as encompassing services provided by hospitals, primary care facilities, nursing homes, and outpatient clinics. The welfare sector refers to social service settings such as child welfare agencies, disability and elder care services, and community-based behavioral and mental health organizations. Accordingly, healthcare and welfare professionals (HCPs) include, among others, nurses, physicians, psychologists, social workers, child welfare workers, case managers, and various support staff.

In addition, throughout this review, we use the term “factors” as a neutral, overarching label for all variables associated with turnover, regardless of their temporal or causal status. The broader literature often refers to “antecedents” (variables that precede turnover in time), “predictors” (variables that statistically predict turnover), or “determinants” (variables implying a causal relationship). However, we avoid the term “determinants”, as the observational nature of most included studies does not support strong causal inferences.

Research questions

With these challenges in mind, our study aims to systematically address the multifaceted nature of turnover by examining its associated factors across different domains (individual, group or team level, and organizational level) and contexts (healthcare and welfare setting), and considering the recent changes brought about by the COVID-19 pandemic. For this purpose, we have formulated three research questions:

• The primary research question guiding this study is:

  1)"Which factors are associated with personnel turnover among HCPs in the healthcare and welfare sectors as found in systematic reviews of the last 10 years?"

• The two secondary research questions include:

2a)"What are the commonly used methods for measuring personnel turnover in the healthcare and welfare sectors as described in systematic reviews of the last 10 years?"

2b)"What types of interventions, described in systematic reviews of the last 10 years, have proven to be effective in reducing personnel turnover in the healthcare and welfare sectors?"

By answering these questions, we hope to contribute to the development of strategies to improve retention and, consequently, the quality of healthcare and welfare services.

Methods

Search strategy and study selection

A systematic search of the literature was conducted in July 2023, based on the guidelines of the Preferred Reporting Items for Systematic review and Meta-analysis Protocols (PRISMA-P) 2015 statement [64]. The electronic databases used for the search included PubMed, Web of Science, CINAHL, Cochrane Database of Systematic Reviews, and Embase. Initially, potential Medical Subject Headings (MeSH) terms were identified via PubMed. In this study, we focus on 'turnover intention' and 'actual turnover' and their synonyms. However, positively phrased terms such as 'intention to stay' are excluded from this umbrella review as they represent distinct constructs with different work-related correlates [66]. In addition, to ensure comparability of findings, we limited our scope to studies from high-income, non-rural contexts, excluding LMICs, UMICs, and rural/remote areas due to differences in resources and workforce structures that warrant separate analyses. Furthermore, we limited the search to studies published from January 2013 onward to ensure that the review reflects recent workforce turnover evidence in contemporary healthcare and welfare contexts. Older studies were excluded to avoid outdated data and maintain the relevance of our findings. Finally, systematic reviews were included if they analyzed peer-reviewed primary studies. Reviews that incorporated studies with grey literature (e.g., reports, theses, policy documents) as primary data sources were excluded. This was done to ensure a consistent quality standard across the evidence base.

After identifying the main MeSH terms, non-MeSH entry terms and synonyms that met the inclusion criteria were added to complete the search string. This search strategy was then converted or modified to run on the other databases. Appendix 1 provides the full details of the search strategy. The selection process followed the inclusion and exclusion criteria listed below in Table 1.

Table 1.

Inclusion and exclusion criteria for literature selection

	Inclusion	Exclusion
Population	Professionals in the healthcare and welfare sectors, with a focus on nursing, caregiving, child support work, social work, and logistical support	Studies focused solely on specific (hospital) departments or specialized professional categories not in our focus. Studies conducted exclusively in low- and middle-income countries, UMICs, rural and remote areas
Issue of Interest	Studies addressing factors, measurement methods, or interventions related to personnel turnover in the healthcare and welfare sectors	Studies not addressing any of the three research questions
Design	Systematic reviews and/or meta-analyses that have been peer-reviewed	Non-systematic reviews (e.g., scoping, narrative, meta-review). Studies incorporating grey literature in their research design
Language	Articles published in English	Articles published in other languages
Date of Publication	Articles published between January 2013 and July 2023	Articles published before January 2013

Screening

Duplicates were removed from the initial search results using Endnote and Rayyan. Two reviewers (masked for review process) independently screened the titles and abstracts of the remaining articles, achieving a Kappa of 0.71, indicating substantial agreement. Any discrepancies between the reviewers were resolved by consulting a third reviewer (masked for review process). Next, two reviewers independently analyzed 10% of the selected full-text articles to confirm their adherence to the inclusion and exclusion criteria. Additionally, the reference lists of all included publications were examined, and a forward citation tracking was applied to identify any relevant articles missed in the initial search or published subsequently.

Data extraction

Data extraction was conducted by one reviewer (masked for review process) using predefined extraction forms and spreadsheets, and 10% of the data was double-checked by a second reviewer (masked for review process). Given the heterogeneity of the data, which precluded a meta-analysis, results are presented descriptively, incorporating both qualitative and quantitative aspects in line with the SWiM (Synthesis Without Meta-analysis) guidelines [16]. The results are organized according to the primary and secondary research questions concerning the factors, measurement methods, and interventions associated with personnel turnover in the healthcare and welfare sectors.

Synthesis

The characteristics and findings of the included studies are presented through tables and narrative summaries. To address the primary research question and manage the diversity and volume of reported factors of turnover, several steps were taken to organize the data. Initially, every potential factor mentioned in the systematic reviews was extracted, this included the frequency with which a factor was cited by a primary study. Subsequently, an overview was constructed, detailing all factors cited at least once in the systematic reviews, as well as the number of systematic reviews in which a factor was mentioned multiple times. Finally, a table was created that only included the factors that were mentioned at least twice in systematic reviews. The latter was to ensure that the factors included were meaningful and not isolated antecedents. Here, a distinction was made between turnover intention and actual turnover. Furthermore, the terminology used in the original reviews was retained (e.g., with some reviews referencing 'job satisfaction' and others 'job dissatisfaction'). In addition, we grouped the factors into three main categories: socio-demographics and health status, work environment characteristics and personal attitude and functioning. The work-related factors were then further subdivided into (a) job content, (b) working conditions, (c) terms of employment, and (d) social relations at work [53, 79]. This structuring implies a combined deductive-inductive approach: starting with a broad extraction of factors (inductive), we then categorized them into predefined theoretical domains (deductive). This process was executed by two authors (masked for review process). Finally, a sunburst visualization was made that attempted to streamline the table for a more intuitive understanding by 1) combining turnover intention and actual turnover for a unified perspective, 2) including only items that demonstrated a positive, negative, or mixed association (i.e., focusing on associations with clear impacts), 3) standardizing all items on a negative scale for easier comparison (thus, for example, retaining only pay dissatisfaction), and 4) concentrating exclusively on work environment characteristics. This focus allows us to zoom in on the aspects of the workplace that directly impact staff retention, providing clear directions for making meaningful changes to reduce turnover.

For the secondary research questions, (masked for review process) conducted the data analysis and synthesis. Finally, all the findings were discussed among all authors in a series of meetings until a consensus was reached on the interpretation and implications of the findings.

Methodological quality and corrected covered area

The quality appraisal of the included reviews was conducted using the AMSTAR 2 (A MeaSurement Tool to Assess systematic Reviews) tool, which consists of 16 items that assess various domains of systematic review methodology, including the selection of studies, data extraction, assessment of risk of bias, and the synthesis of results [80]. The AMSTAR 2 tool categorizes the overall confidence in the results of the review as high, moderate, low, or critically low. A review is rated as 'high" if it has no or one non-critical weakness. It is rated as 'moderate' if it has more than one non-critical weakness. A 'low' rating indicates the presence of one critical flaw with or without non-critical weaknesses. A 'critically low' rating indicates the presence of more than one critical flaw with or without non-critical weaknesses. The critical domains in the AMSTAR 2 tool include the adequacy of the literature search, justification for excluding individual studies, risk of bias from individual studies being included in the review, appropriateness of meta-analytical methods, consideration of risk of bias when interpreting the results of the review, and assessment of the presence and likely impact of publication bias. This tool has been widely used in previous similar umbrella reviews and it is considered to be a valid and reliable instrument [59, 62]. Part of this quality appraisal (10%) was conducted by two reviewers (masked for review process). Disagreement between the two reviewers was solved via a consensus discussion. When no consensus could be reached, a third reviewer (masked for review process) was consulted.

Additionally, the Corrected Covered Area (CCA) was calculated for each research question, following the approach of Hennessy & Johnson [45]. The CCA is used to identify the degree of overlap between primary studies included in different reviews. The first step in this process involved constructing a citation matrix of the primary studies included in the reviews. Then, the CCA was calculated across the entire matrix. Furthermore, when a low overall CCA was determined, the citation matrix was further examined to identify any reviews with complete or nearly complete overlap in subsamples of primary studies.

Results

Results of the search strategy

In Fig. 1, we have outlined the process of selecting studies for inclusion, as well as the grounds for exclusion. A list of all articles that were excluded can be found in Appendix 2. Out of the total, 37 reviews satisfied the inclusion criteria and were included into the review.

Fig. 1.

Flowchart of inclusion process

Study characteristics

The majority (34 out of 37) are systematic reviews, with three being a combination of systematic reviews and meta-analyses. Most reviews (21 of the 37) were published between 2021 and 2023. We included a total of 917 primary studies, ranging from as few as five to as many as 129 per review.1 However, only a subset of these primary studies directly answered the research questions posed. Therefore, the total of included primary studies per review in this umbrella review is slightly lower: there are 511 studies in total, with the number ranging from a minimum of one study to a maximum of 55 studies. Of these, 92% are quantitative, 5% are qualitative, and 3% use mixed methods. Geographically, the majority of these studies were conducted in the USA (43%), followed by contributions from Canada (7%), China (5%), South Korea (4%), and the UK (3%). Fig. 2 provides an overview.2 Furthermore, the majority of the reviews (27 out of 37) were conducted in the healthcare sector, predominantly in hospital settings. Settings in the welfare sector include community behavioral health agencies, outpatient mental health providers, and child welfare services. Lastly, the type of professionals involved in the studies across the reviews was predominantly nurses. Other professions included, among others, physiotherapists, occupational therapists, healthcare administrative or management staff, child welfare workers, and social workers. Table 2 summarizes the characteristics of the included studies.

Fig. 2.

Overview of geographic distribution of primary studies included in the umbrella review. Note: Studies conducted exclusively in LMICs, UMICs, or rural/remote areas were excluded based on our eligibility criteria. However, some included studies may cover these regions as part of broader international analyses. Therefore, Fig. 2 reflects the distribution of studies as included in our review, but is not intended to represent a global overview

Table 2.

General characteristics of the included systematic reviews & relevant primary studies addressing the research questions

General information				Specific information (only studies that answered our research question(s))
Authors, Year	Objectives	Search Strategy	Total number of studies	Number of relevant studies	Study design	Geographical Location*		Industry	Type of professionals
Aparício, C. et al. [3*], Systematic review	Explore the evidence of preceptorship programs and clinical supervision programs on nurse retention in hospital settings	Databases: PubMed, CINHAL, Medline, Web of Sciences, ScienceDirect Time span: 2009–2019	6	3	Quantitative = 3	Australia = 1, USA = 1, Ethiopia = 1		Healthcare sector—hospital settings	Newly qualified nurses, nurse preceptors, nurse supervisors
Bae, S.H. [6*], Systematic review	Investigate and synthesize the impacts of nurse staffing and work schedules on nurse turnover in acute hospitals	Databases: CINAHL, Cochrane Library, DBpia, EBSCO, PubMed, PsycINFO, RISS, Web of Science Time Span: 2000–2021	14	14	Quantitative = 14	South Korea = 4, USA = 9, Canada = 1		Healthcare sector—hospital settings	Staff nurses
Bae, S.H. [5*], Systematic review	Investigate factors contributing to newly licensed registered nurses turnover in acute care hospitals	Databases: CINAHL, Cochrane Library, DBpia, EBSCO, PubMed, PsycINFO, RISS, Web of Science Time span: 2000–2021	10	10	Quantitative = 10	USA = 2, South Korea = 5, Japan = 3		Healthcare sector—hospital settings	Newly licensed registered nurses
Bahlman-Van Ooijen, W. et al. [8*], Systematic review	Provide a systematic overview of the qualitative evidence available on the motivations for nurses to leave the nursing profession	Databases: CINAHL, PsycINFO, PubMed Time span: 2010–2023	9	9	Qualitative = 9	Iran = 3, Canada = 1, Finland = 2, Netherlands = 1, Germany = 1, China = 1		Healthcare sector—various healthcare settings	Nurses
Bos, A. et al. [10*], Systematic review	Understand how a nursing home's profit status impacts financial performance, employee well-being, and client well-being, focusing specifically on private for-profit and private not-for-profit nursing homes	Databases: PiCarta, Scopus, PubMed, Google Scholar, Web of Science Time span: 2004–2014	50	4	Quantitative = 4	USA = 4		Healthcare sector -Nursing home	Registered nurses, licensed vocational nurses, certified nurse assistants
Brabson, L.A. et al. [9], Systematic review	Identify factors associated with turnover in community behavioral health settings and synthesize recommendations to address problematic levels of turnover	Databases: PsycINFO, Medline/PubMed, SSCI Time span: No restriction	16	16	Quantitative = 15, Mixed-method = 1	USA = 16		Welfare sector—community behavioral health agencies	Community behavioral health roles
Brook, J. et al. [11*], Systematic review	Identify the characteristics of successful interventions to reduce turnover and increase retention of early career nurses	Databases: Google Scholar, Cochrane Library, Academic Search Complete, Medline, Health Policy Reference Centre, EMBASE, Psychinfo, CINAHL Time span: 2001–2017	53	53	Quantitative = 53	USA = 48, Australia = 2, Canada = 1, UK = 1, Taiwan = 1		Healthcare sector—hospital settings	Early career nurses, newly hired/licensed/qualified nurses, nurses with no experience in a particular specialty
Butler, M. et al. [15], Systematic review and meta-analysis	Explore the effect of hospital nurse-staffing models on patient and staff-related outcomes, identify which staffing model(s) are associated with better outcomes, and assess the impact of staffing model(s) on cost outcomes	Databases: Cochrane CENTRAL, MEDLINE Ovid, Embase Ovid, NHSEED, CINAHL EBSCO, ICTRP, ClinicalTrials.gov, ISI Web of Science Time Span: 2009–2018	19	4	Quantitative = 4	USA = 3, Netherlands = 1		Healthcare sector—hospital settings	Hospital nurses (registered, practical, unlicensed personnel or global equivalents)
Carpenter, J. et al. [17], Systematic review	Review the effectiveness and cost-effectiveness of supervision in child welfare in relation to outcomes for consumers, staff, and organizations	Databases: ASSIA, British Education Index, Campbell Collaboration, CINAHL, Cochrane Library, Medline, PsycInfo, Social Care Online, Social Services Abstracts, Social Work Abstracts, EBSCO Host EJS, IngentaConnect Time span: After 2000	22	9	Quantitative = 8, Qualitative = 1	USA = 9		Welfare sector—child welfare services sector	Child welfare workers, social workers, residential/school social workers, caseworkers, child protection/youth workers, newly qualified/educational social workers
Chan, Z.C.Y. et al. [18], Systematic review	Present the findings of a literature review regarding nurses' intention to leave their employment or the profession	Databases: British Nursing Index, MEDLINE, CINAHL, PsycINFO, Sociological Abstracts, Cochrane Time span: 2001–2010	31	31	Quantitative = 29, Qualitative = 2	Canada = 4, China = 2, European countries = 2, Finland = 1, Japan = 3, Jordan = 4, Korea = 1, Lebanon = 1, Macao = 1, South Africa = 1, Sweden = 2, Taiwan = 5, UK = 2, USA = 2		Healthcare sector—hospital settings	Registered nurses (RN) in non-specialty wards
Chen, C.M. et al. [19], Systematic review	Conduct a systematic literature review and examine the effectiveness and application of mentorship programs for recently registered nurses	Databases: Medline (Ovid), Cochrane Library, CINAHL, PubMed, Index to Taiwan Periodical Literature System, Chinese Electronic Periodical Services (CEPS), National Digital Library of Theses and Dissertations in Taiwan Time span: 1999–2011	5	2	Quantitative = 2	USA = 1, Taiwan = 1		Healthcare sector—hospital settings	Recently registered nurses (mentees), experienced nurses (mentors)
Cummings, G.G. et al. [25], Systematic review	Examine the relationships between leadership styles and outcomes for the nursing workforce and their work environments, and determine how nursing leadership styles influence outcomes for nurses, nursing environments, and the nursing workforce	Databases: CINAHL, Medline, PsychInfo, ABI, ERIC, Sociological Abstracts, Embase, Cochrane, Health Star, Academic Search Premier Time span: 1985–2017	129 studies (53 original and 76 updated)	9	Quantitative = 9	USA = 5, Canada = 4		Healthcare sector	Nurses
Daouk-O¨Yry, L. et al. [28*], Systematic review	Develop an integrative framework of nurse absenteeism and turnover in hospitals that accounts for both voluntary and involuntary attendance behavior	Databases: PubMed, CINAHL Plus Time span: 2007–2013	41	28	Quantitative = 17, Qualitative = 3, Mixed-method = 8	Kuwait = 1, UK = 2, USA = 19, Jordan = 1, New Zealand = 1, Sweden = 2, Japan = 3, Canada = 1, Belgium = 1, Australia = 1		Healthcare sector—hospitals and medical centers	Nurses
De Vries, N. de et al. [29*], Systematic review	Explore the prevalence of nurses and physicians intending to leave their position in EU hospitals and understand the main determinants influencing job retention among these professionals	Databases: PubMed, Embase, CINAHL Time Span: Last 10 years as of 2021	Total: 345. EU studies: 37. Non-EU studies: 308	37	Quantitative = 35, Qualitative = 2	Australia = 1, Belgium = 2, Czech Republic = 1, Denmark = 1, EU = 1, Finland = 1, France = 1, Germany = 3, Iceland = 1, Ireland = 1, Italy = 3, Lebanon = 1, Lithuania = 1, Netherlands = 2, Norway = 3, Poland = 2, South Korea = 1, Sweden = 3, Switzerland = 2, Turkey = 6, UK = 2, USA = 2		Healthcare sector—hospital settings	Nurses and physicians
De Vries, N. et al. [30], Systematic review	Identify and analyze interventions that minimize nurse and physician job retention in hospitals	Databases: CINAHL, Embase, PubMed Time Span: 2012–2022	55	55	Quantitative = 47, Qualitative = 5, Mixed-method = 5	US = 23, Canada = 2, Oman = 1, Thailand = 1, UK = 3, Turkey = 1, Taiwan = 4, China = 6, Australia = 3, France = 1, Sweden = 1, Mexico = 1, Korea = 2, Finland = 1, South-Korea = 2, Malaysia = 2, India = 1, Iran = 1		Healthcare sector—hospital settings	Healthcare professionals (primarily nurses, physicians, combination of both)
Jarden, R.J. et al. [49*], Systematic review	Determine the published prevalence, predictors, barriers, and enablers of new graduate registered nurse wellbeing, work wellbeing, and mental health	Databases: CINAHL, MEDLINE, EMBASE, PsycINFO Time span: 2009–2019	34	9	Quantitative = 9	Canada = 2, South Korea = 3, USA = 1, Japan = 2, Taiwan = 1		Health care organizations—various settings	New graduate registered nurses in first year of practice
Ke, Y.T. et al. [51*], Systematic review	Determine the effects of nursing preceptorship on the competence, job satisfaction, professional socialization, and retention of new nurses	Databases: Index of Taiwan Periodical Literature System, Airiti Library, CINAHL, Cochrane Library, PubMed/MEDLINE Time span: Until June 2015	6	2	Quantitative = 2	USA = 1, Japan = 1		Healthcare sector	New nurses, preceptors
Lartey, S. et al. [57*], Systematic review	Address the effectiveness of strategies for retaining experienced registered nurses	Databases: CINAHL, PsychInfo, EMBASE, Medline, Cochrane library, SCOPUS Time span: No restriction	12	12	Quantitative = 12	USA = 9, Canada = 1, Sweden = 1, Italy = 1		Healthcare sector—hospital settings	Mostly registered nurses, some nurse managers, some nurses
Lee, J. (2021), Systematic review	Examine and describe the literature on nursing home nurses' turnover intentions in their workplace	Databases: CINAHL, PubMed, Cochrane Library, PsycINFO, RISS, Dbpia Time Span: 2009–2019	6	6	Quantitative = 6	Korea = 1, USA = 3, Taiwan = 2		Healthcare sector—nursing home	Registered nurses, licensed practical nurses
Lu, H. et al. [61*], Systematic review	Identify a comprehensive knowledge of the job satisfaction of qualified general nurses working in acute care hospitals and its associated factors	Databases: PubMed, Web of Science, CINAHL, Embase, PsycINFO, Applied Social Sciences Index, CNKI, WanFang, SinoMed, VIP Time span: 2012–2017	59	7	Quantitative = 7	China = 2, Papua New Guinea = 1, Turkey = 2, Hong Kong = 1, Taiwan = 1		Healthcare sector—hospital settings	Qualified general nurses in acute care hospitals
Padín, P.F. et al. [70], Systematic review	Synthesize and analyze the available evidence on burnout among health social workers from an international context within a broad temporal range	Databases: PubMed, Scopus, Web of Science (WOS), Higher Council for Scientific Research (CSIC) Time span: 2000–2020	14	4	Quantitative = 4	USA = 4		Mental health, social health, social workers	Health social workers in different specializations
Park, H. et al. [71*], Systematic review	Review the effectiveness of policies addressing nursing workforce shortages	Databases: MEDLINE, Embase, CINAHL, PubMed Time span: 2009–2019	12	7	Quantitative = 5, Qualitative = 1, Mixed-method = 1	New Zealand = 1, USA = 1, Canada = 3, Kenya = 1, Zimbabwe = 1		Healthcare sector	Nurses, graduate and young nurses (below 30), midwives, physicians, environmental health technicians, nurse aids, community health volunteers
Park, J.H. et al. [72*], Systematic review	Evaluate whether e-healthcare interventions improve burnout and other mental health aspects of nurses	Databases: MEDLINE (via PubMed), EMBASE (via Elsevier), Cochrane Library Central Register of Controlled Trials, Cumulative Index of Nursing and Allied Health Literature, Allied and Complementary Medicine Database, PsycARTICLES, Google Scholar Time span: Until January 2021	7	1	Quantitative = 1	South Korea = 1		Healthcare sector	Nurses
Pedrosa, J. et al. [73*], Systematic review	Identify scientific evidence on factors of organizational culture associated with nurses' turnover	Databases: CINAHL® Complete; Nursing & Allied Health Collection; Cochrane Central Register of Controlled Trials; Cochrane Database of Systematic Reviews (CDSR); Cochrane Methodology Register (CMR); Library, Information Science & Technology Abstracts (LISTA); MedicLatina; MEDLINE® Complete Time span: 2014–2018	9	7	Quantitative = 7	China = 2, Italy = 1, UK = 1, Thailand = 1, USA = 1, South Korea = 1		Healthcare sector	nurses
Poon, Y.S.R. et al. [74*], Systematic review	Examine the factors affecting turnover intention among healthcare workers during the COVID-19 pandemic	Databases: PubMed, Embase, Scopus, CINAHL, Web of Science, PsycINFO Time Span: 2020–2022	43	43	Quantitative = 39, Qualitative = 2, Mixed-method = 2	Australia = 2, Canada = 1, China = 2, Egypt = 3, Germany = 2, Indonesia = 1, Iran = 3, Italy = 2, Lebanon = 3, Oman = 4, Pakistan = 4, Peru = 1, Philippines = 4, Qatar = 1, Saudi Arabia = 2, Turkey = 2, UAE = 1, UK = 2, USA = 9		Healthcare sector—various healthcare settings	healthcare professionals (doctors, nurses, pharmacists, physiotherapists, occupational therapists, administrative or management staff)
Rodríguez-García, M.C. et al. [76*], Systematic review	Investigate how Magnet hospital status affects outcomes for nursing professionals, patients, and healthcare organizations	Databases: CINAHL, ProQuest, PubMed, La Biblioteca Cochrane Plus Time span: 2010–2018	21	4	Quantitative = 4	USA = 4		Healthcare sector—hospital settings	nurses
Romppanen, J. et al. (2016), Systematic review	Gather, assess, and synthesize current research knowledge on interventions aiming to improve nurses’ well-being at work, including different facets of well-being	Databases: CINAHL, Cochrane, EBSCO, PubMed, PsykInfo, Scopus Time span: 2009–2015	10	1	Quantitative = 1	Canada = 1		Healthcare sector—hospitals, mental health facilities, academic health center, nursing home	primarily nursing staff, other healthcare personnel in five studies
Shin, S. et al. [82*], Systematic review and meta-analysis	Systematically assess empirical studies on the relationship between nurse staffing and nurse outcomes through meta-analysis	Databases: CINAHL, PubMed, PsycINFO, Cochrane Library, EBSCO, RISS, Dbpia Time span: 2000–2016	13	3	Quantitative = 3	USA = 3		Healthcare sector—hospital settings	nurses
Stemmer, R. et al. [83*], Systematic review	Investigate the association of unfinished nursing care on nurse outcomes	Databases: CINAHL, Cochrane Library, Embase, Medline via PubMed, ProQuest, PsycINFO, Scopus Time span: No restriction	9	3	Quantitative = 3	Korea = 1, USA = 2		Healthcare sector—hospital settings	registered nurses, nurse aids, nurse assistants, various education levels
Stevanin, S. et al. [84*], Systematic review	Describe and summarize workplace characteristics of three nursing generations: Baby Boomers, Generations X, and Y	Databases: PubMed, CINAHL, PsycINFO, Scopus Time span: 1991- 2017	33	8	Quantitative = 6, Qualitative = 1, Not declared = 1	USA = 2, New Zealand = 1, Canada = 3, Australia = 1, Japan = 1		Healthcare sector—hospital settings	registered nurses, nurse managers (in one study)
Tolksdorf, K.H. et al. [86], Systematic review	Identify factors associated with nurses' turnover intention during the COVID-19 pandemic	Databases: MEDLINE, CINAHL, PsycINFO, PSYNDEX, PsycArticles, SocINDEX Time Span: 2020–2021	19	19	Quantitative = 19	Australia = 1, Canada = 2, Egypt = 1, Indonesia = 1, Iran = 1, Lebanon = 1, Pakistan = 2, Philippines = 4, Qatar = 1, Romania = 1, Taiwan = 1, Turkey = 1, UK = 1, USA = 1		Healthcare sector—hospital settings	majority nurses, one study included physicians
Vázquez-Calatayud, M. et al. [89], Systematic review	Identify effective interventions that promote the retention of newly graduated registered nurses in the hospital setting	Databases: PubMed, CINAHL, Scopus, PsycINFO, Cochrane Library Time span: 2012–2022	9	5	Quantitative = 5	USA = 3, Taiwan = 1, China = 1		Healthcare sector—hospital settings	newly graduated registered nurses
Wei, C.W. et al. [92*], Systematic review	Synthesize recent literature on spiritual issues in nursing and provide evidence-based recommendations to improve nursing management and quality of care	Databases: CEPS, Medline/PubMed, Embase, Cochrane library, Google Scholar Time span: No restriction	7	1	Quantitative = 1	Taiwan = 1		Healthcare sector—hospital settings	nurses
Woodward, K.F. et al. [94*], Systematic review	Understand factors associated with registered nurse work outcomes in the United States and examine the inclusion of equity and wellness concepts	Databases: PubMed, CINAHL Time span: 2010–2020	34	27	Quantitative = 27	USA = 27		Healthcare sector—hospital settings	clinical registered nurses
Zhu, L.L. et al. [99*], Systematic review and meta-analysis	Elaborate on the relationship between work engagement, perceived organizational support, and the turnover intention of nurses by analyzing potential moderators	Databases: PubMed, Embase, Cochrane Library, Web of Science, Medline, Scopus Time Span: Until 2022	40	40	Quantitative = 40	Australia = 4, Belgium = 1, Canada = 2, China = 10, Cyprus = 1, Egypt = 2, Ireland = 1, Iran = 3, Israel = 1, Italy = 5, Japan = 1, Jordan = 1, Poland = 1, Portugal = 1, Sweden = 1, Taiwan = 2, UK = 2, USA = 1		Healthcare sector—hospitals, medical institutions, nursing homes	nurses
Zhang, Y. et al. (2016), Systematic review	Evaluate the effectiveness of a mentoring program on the mentor, mentee, and organization	Databases: Cochrane Library, Medline, Ovid, Elsevier, Embase, CINAHL, CBM, CNKI, WanFang Data Time span: No restriction	9	4	Quantitative = 4	USA = 4		Not mentioned—nursing	newly graduated nurses
Zhang, Y. et al. [98*], Systematic review and meta-analysis	Quantitatively assess the correlation between horizontal violence and turnover intention in nurses	Databases: Cochrane Library, PubMed, EMBASE, CINAHL, SinoMed, CNKI, Wanfang Time span: Until March 2022	14	14	Quantitative = 14	Canada = 5, USA = 4, Korea = 2, China = 1, Pakistan = 1, Turkey = 1		Not mentioned—nursing	nurses

This table presents an overview of the included systematic reviews and a selection of relevant primary studies that specifically contribute to our research questions. The 'Total number of studies' column refers to the total number of studies reviewed in each systematic review. The 'Number of relevant studies' column indicates the number of studies that directly respond to our research questions and meet the inclusion criteria (cf. Table 1)

^*Some studies included more than 1 country

Quality appraisal

Appendix 3 presents the quality appraisal scores for the individual reviews. The results of the quality appraisal indicate a lack of rigor in the majority of the 37 included reviews assessed using the AMSTAR 2 tool. Twenty-three reviews (62%) received a low overall score, with an additional nine reviews (24%) rated as critically low, indicating that 86% of the studies fall into lower quality categories. In contrast, a small fraction of the reviews achieved a moderate (11% or 4 reviews) or high (3% or 1 review) overall score. All reviews incorporated the PICO components into their research questions and criteria, and employed a comprehensive literature search strategy. Additionally, while the majority of the reviews adequately explained their choice of study designs for inclusion (21 out of 37), conducted study selection in duplicate (26 out of 37), and performed data extraction in duplicate (22 out of 37), only a limited (nine out of 37) number included a research protocol. Furthermore, a very small number (six out of 37) fully provided and justified a list of excluded reviews, and many did not account for the risk of bias in individual studies when interpreting or discussing the review results (10 out of 37) or did not satisfactorily explain and discuss any observed heterogeneity in the review results (eight out of 37).

Corrected covered area

The CCA values obtained were 1.28% for research question 1a (measurements), 0.38% for research question 1b (factors), and 1.02% for research question 2 (interventions). These low CCA values indicate minimal overlap of primary studies across the included reviews, suggesting that the different reviews analyzed a broad spectrum of unique primary studies [45]. Next, the citation matrix was examined to identify any reviews with (nearly) complete overlap in subsamples of the primary studies. For research questions 1 and 2a, two reviews with a high overlap or CCA of 29% were identified [74*, 86]. This substantial overlap suggested that these reviews largely analyzed the same set of primary studies, which could result in redundant findings. To circumvent this redundancy, our synthesis and thematic analysis focused on the most comprehensive and highest-quality review among the two [73*].

Q1: Factors influencing staff turnover in healthcare and welfare sectors

Out of the included 37 reviews, 25 addressed this research question. As mentioned in the methodology section, we identified all potential factors mentioned in these reviews, noting the frequency of each citation. Table 3 summarizes the factors cited in more than one review, using the same terminology as in the original reviews. Out of the 25 reviews, 13 reviews (52%) focused on the factors of turnover intention, five reviews (20%) on actual turnover, and seven reviews (28%) on both.

Table 3.

Summary of factors

	A. Socio-demographics and health status		B. Work environment characteristics									C. Personal attitude and functioning
			Job content		Working conditions			Terms of employment		Social relations at work
Turnover Intention	Poor mental health & psychosocial well-being	+	Low decision latitude	+	Poor work environment		+	Lack of car. dev. opportunities	+	Bullying behavior	+	Burnout/fatigue	+
	Age	±	High level of job strain	+	Poor patient safety		+	Working in shifts	+	Low support/interaction colleagues	+	Work-related stress	+
	Gender	±	High workload	+	Inadequate staffing		+	Work-life imbalance	+	Perceived organizational support deficit	+	Lack of job satisfaction	+
	More/less work experience	±	Empowerment	-	Magnet designation		-	Dissatisfaction with pay/benefits	+	Poor leadership (style)	+	Low career satisfaction/ professional fulfillment	+
	Education	±/0	Job autonomy	-	Positive work environment		-	Effort-reward imbalance	+	Workplace violence	+	Not feeling connected to organization	+
	Married vs. Single nurses	±/0			Working in desired place of work		-	Growth opportunities	-	Support/interaction with colleagues	-	Low work-related stress	-
					Adequate staffing		-	Good work-life balance	-	Support supervisor/organization	-	High work engagement	-
					Ethical decision making		-	Pay satisfaction	-			Professional commitment	-
												Organizational commitment	-
												High perceived quality of care	-
												Job satisfaction	-
Actual turnover	Poor physical health conditions	+	High workload	+		Non Magnet hospital units	+	Dissatisfaction with pay/benefits	+	Lack of support supervision	+	Burnout/fatigue	+
	Age or generation (mixed)	±	Patient load	0		Unrealistic expectations of employer	+	Working full time	+	Poor leadership (style)	+	Job dissatisfaction	+
	Having more than 1 job	-				Unsupportive org. culture	+	Voluntary overtime	-	Social support (peers)	-	Job embeddedness	-
						Number of nurses per 100 beds	-	Mandatory overtime	0			Job satisfaction	-
								Type of shift	0			Organizational commitment	-
								Typical work schedule	0

Higher +; Lower -; Mixed results ±; No difference: 0

Regarding factors for turnover intention, a total of 46 factors were identified, categorized into socio-demographics and health status (6) work environment characteristics (28), and personal attitude and functioning (11). Among socio-demographics and health status, poor mental health and (poor) psychosocial well-being were found to be positively correlated with turnover intention, indicating that individuals experiencing lower mental and psychosocial well-being are more likely to express a desire to leave their current positions. However, the majority of socio-demographic factors, such as age, gender, work experience, and education, showed mixed results. In addition, 'education' and 'marital status' (married vs. single), appears in multiple categories, being both positively and negatively correlated, as well as showing no clear association. Regarding the work environment characteristics, six work environment characteristics were positively correlated with turnover intention, highlighting detrimental aspects such as low decision latitude, high job strain, high workload, poor work environment, inadequate staffing, and several others including negative social relations at work like bullying behavior and poor leadership. Conversely, 12 characteristics were found to have a negative correlation with turnover intention, suggesting protective factors like empowerment, job autonomy, positive work environments, and supportive social relations at work. Within the personal attitude and functioning category, five factors were identified that positively correlate with turnover intention: burnout/fatigue, work-related stress, job dissatisfaction, low career satisfaction, and a lack of organizational connection. In contrast, six factors—reduced work-related stress, heightened work engagement, strong professional and organizational commitment, increased perceived quality of care, and job satisfaction—were linked to lower turnover intention, acting as potential protective factors.

For actual turnover, a total of 23 factors were identified and again categorized into socio-demographics and health status (3), work environment characteristics (14), and personal attitude and functioning (5). Among socio-demographics and health status, again, poor physical health conditions was positively associated with actual turnover. Age also showed mixed associations, mirroring the same mixed patterns observed with turnover intentions. Finally, holding more than one job was associated to lower actual turnover. Regarding the work environment characteristics, on the one hand, factors like high workload, non-Magnet hospitals (i.e., institutions lacking Magnet® recognition for nursing excellence and quality of care), and poor leadership style were positively correlated with actual turnover. On the other hand, factors like number of nurses per 100 beds and voluntary overtime were negatively correlated with actual turnover. Mandatory overtime, type of shift and typical work schedule showed no significant association, suggesting that these aspects may not be critical in the decision to leave.

Mixed findings in socio-demographics

Revisiting the mixed findings in socio-demographic factors, we delved into the systematic reviews again to get more details behind these varied outcomes. Starting with age, we saw – in general – that younger nurses are more prone to leaving, driven by workplace challenges, reward imbalances, and personal life impacts. On the other side, older nurses are less inclined to leave, hinting at age-related differences in their approach to turnover [28*, 29*, 58, 84*]. Yet, some research also shows mixed results in how age affects turnover intentions or actual turnover [94*].

Looking at gender, the evidence is more diverse: certain reviews suggest that male nurses are more likely to leave, while others find them less inclined to turnover intentions [6*, 29*, 86]. However, males are frequently underrepresented in sample sizes, especially within nursing studies, which could influence the robustness of these findings.

Next, working experience also showed mixed results, that is: while employees with less working experience often show lower intentions to leave [58], this trend is not always consistent. For instance, one review found that nurses with less than five years of working experience were actually less inclined to remain in their positions [28*]. In addition, nurses with extensive tenure at their current workplace, especially those over 35 years old, show a higher propensity to leave [28*].

The findings on marital status show that single nurses can have both higher and lower intentions to leave [6*, 18, 58], while some research also yields inconclusive results on this relationship [86].

Lastly, regarding education level, some research suggests that nurses with higher education levels, such as a Master's degree, have been associated with higher likelihoods of leaving current employment [30]. However, other reviews show lower turnover intentions among well-educated nurses [6*, 28*, 86] or no significant relationship between educational level and turnover intentions [6*, 58].

Interplay of factors of turnover intention and actual turnover

While looking at the work environment characteristics and personal attitude and functioning factors for both turnover intention and actual turnover, we see consistencies and oppositions among factors. For example, job autonomy, associated with decreased turnover intentions, could stand in contrast to low decision latitude, which correlates with increased desires to leave. Likewise, a positive work environment is inversely related to a poor work environment and job satisfaction directly opposes job dissatisfaction.

This pattern of both overlapping and opposing factors highlights the complex nature of factors influencing turnover behaviors. To provide a clearer understanding, Fig. 3, presents a streamlined overview of the factors after removing redundancies, as described in the method section. It features a sunburst visualization, which highlights the key factors across various work environment categories. Within the domain of job content, factors include high patient load, elevated job strain, high workload, along with low empowerment, job autonomy, and decision latitude. Working conditions encompass poor work environment and patient safety, inadequate staffing, non-Magnet hospital units, not working in the desired location, absence of ethical decision making, unrealistic employer expectations, unsupportive organizational culture, and a low number of nurses per 100 beds. The terms of employment include the following factors: a lack of career development opportunities, working in shifts, work-life imbalance, dissatisfaction with pay and benefits, effort-reward imbalance, no growth opportunities, and working full-time. Finally, ‘social relations at work’ includes low support or interaction with colleagues, poor leadership styles, workplace violence, bullying behavior, and perceived organizational support deficits.

Fig. 3.

Sunburst diagram of key factors across work environment characteristics

Q2a: Measurements of staff turnover in healthcare and welfare sectors

Out of 37 included reviews, we found that only 11 clearly detailed the methods used to measure turnover. Within these, four reviews (36.4%) investigated actual turnover, three reviews (27.2%) concentrated on turnover intention, and the remaining four (36.4%) examined both types, thereby confirming that a distinction is commonly made between these two concepts in the literature.

For the turnover intention, the definitions varied among the reviews, including terms such as intention to quit and intention to change jobs [70, 86, 88]. It was predominantly self-reported by the employees, representing perception-based measures [29*, 74*, 86, 94*]. The methods employed for measuring turnover intention varied considerably; quantitative studies ranged from using single-item questions to employing larger, validated scales such as the Turnover Intention Scale (TIS-6) [29*, 74*, 86, 94*]. Furthermore, qualitative studies typically conducted semi-structured interviews with individual participants [74*].

For actual turnover, this was expressed in diverse ways, often distinguishing between voluntary and involuntary turnover [57, 87, 94*]. Turnover was frequently expressed as a percentage, referred to as the 'turnover rate,' and reported by senior management personnel [6*, 9*, 25, 29*, 57]. Additionally, other measurements, such as the accession rate, vacancy rate, and resignation data, were also used in conjunction with the turnover rate [25, 57]. In one review, the terms 'retention' and 'turnover' were often used interchangeably [57].

Furthermore, reviews describing actual turnover and/or turnover intention measurements seldom make clear distinctions between leaving the unit, leaving the organization, and leaving the profession (e.g. [49*, 75]). Some reviews just mentioned general turnover or turnover intention measurements or did not specify any measurement used. Additionally, the timeframe for measuring turnover showed significant variation, ranging from six weeks to four years, albeit most studies predominantly measured turnover for a duration of one year [6*, 9*]. As time progressed, an increase in the average turnover rate was observed [6*]. Finally, the reviews themselves suggest that there exists a wide variety of methods used for measuring turnover, and notably, not all primary studies provided detailed information regarding the measurements used. For this reason, the systematic reviews often recommended cautious and careful interpretation of the results across all studies reviewed.

Q2b: Interventions to reduce staff turnover in healthcare and welfare sectors

A total of 12 (out of 37) systematic reviews answered this question. All interventions targeted work-related factors. We organized our findings again using the subdivision (a) job content, (b) working conditions, (c) terms of employment, and (d) social relations at work.

Firstly, related to job content, primary and team nursing methods showed mixed results in staff turnover in one systematic review [14], whereas the effectiveness of stress coping programs on retention varied in another study [29].

Secondly, regarding the working conditions, transition programs to specialty units, job environment interventions, and technological innovations generally improved retention, although robots increased time pressure and missed care [30*]. As for the Magnet hospital models, their impact on retention was found to be inconsistent [57]. Providing extra staffing did not reduce nurse turnover or their intent to leave [30*]. The effectiveness of self-scheduling in reducing turnover is uncertain due to the low level of evidence and high risk of bias [14]

Thirdly, regarding the terms of employment, wage increases, orientation and transition to practice programs, preceptorship (guided practical training with an experienced clinician), mentorship (supportive guidance from an experienced professional), and residency programs have consistently shown positive impacts on reducing turnover and increasing retention rates [11*, 19, 30*, 51*, 71*, 89*, 97*]. However, strategies such as full-time commitment, shift structure changes, and externships have produced mixed results, with some studies recording positive impacts and others showing no significant change or negative outcomes [11*, 29*].

Fourthly, regarding social relations at work, interventions aimed at providing social support, improving interpersonal relations, and enhancing teamwork have generally been successful in reducing turnover intentions [29*, 59, 79].

Most interventions studied were short-term, typically lasting less than a year [12, 90]. Only a smaller portion of the primary studies examined long-term interventions lasting over a year [89*].

Discussion

The umbrella review aggregated results from 37 systematic reviews and a total of 511 primary studies, focusing on personnel turnover in the healthcare and welfare sectors, with a majority of the studies involving healthcare personnel, especially nurses. Firstly, the analysis identified a broad spectrum of factors, divided into socio-demographic factors, work environment characteristics, and personal attitude and functioning. Socio-demographic factors include several factors (such as age) which showed mixed effects on turnover intentions and actual turnover. Work environment characteristics were further subdivided into job content, working conditions, terms of employment, and social relations at work. Personal attitude and functioning includes individual attributes such as burnout and job dissatisfaction. Secondly, different methods are used, with turnover intention mainly self-reported by employees and actual turnover rates most often reported by senior management. Thirdly, while several interventions have been effective in reducing turnover, their impact remains uncertain due to low evidence levels, high bias risk, and contextual differences across settings, indicating that outcomes are context-specific.

Factors influencing staff turnover in healthcare and welfare sectors

The diverse range of factors associated with personnel turnover in the healthcare and welfare sectors, illustrates a potentially complex interplay among them. This diversity indicates that turnover is influenced by multiple factors that vary across different contexts, including geographical locations, healthcare systems, and cultural contexts. In addition, despite using our own sub-categorization in this umbrella review, we observed various categorization schemes in the individual systematic reviews, such as individual, work-related, and organizational factors (e.g. [9*, 18, 86]. This variety in categorization, together with the geographical and setting-specific gaps, underscores the need for a more holistic approach, such as a multifactorial, tailored, and systems-thinking framework that can adapt to the unique challenges of different healthcare and welfare settings [28*].

Looking at the socio-demographic factors, the mixed results call for alternative explanations. For example, age and working experience level show mixed evidence as antecedents of turnover intention and/or actual turnover. A potential explanation might be the ‘reality shock’ during a first encounter or ‘wake-up-call’ when entering the profession. Although junior professionals are familiar with the healthcare setting through their education, it might still feel more overwhelming than expected or ‘eye-opening’ causing them to leave the job and profession at an early stage. A similar mechanism in the teaching profession is called the ‘praxis shock’ [27]. Conversely, more experienced professionals, who have invested significant time in their careers, may demonstrate greater commitment, reducing their likelihood of leaving [47]. At the same time, senior professionals might also be more likely to leave the profession, for example, because they might give up hope that things would change [20].

Regarding the work environment characteristics, we noticed that these are mainly individual level factors, and that there is little research on team and organizational level factors like team psychosocial safety and organizational culture. The psychosocial aspects at the team level, such as social support, team cohesion, and perceived fairness, can influence turnover intentions [13, 90]. Effective leadership, in particular, can mitigate stress and provide a supportive work environment, thereby reducing turnover intentions [13, 24*]. Moreover, the sense of belonging and support within a team can buffer the stress of the healthcare and welfare environments [93]. Looking at the organizational factors, a negative organizational culture, characterized by poor communication, lack of trust, and limited employee involvement, could also contribute to higher turnover rates [21].

Finally, looking at the personal attitude and functioning factors, we can see both negative factors such as burnout and job dissatisfaction, and positive influences like job satisfaction, underscoring the impact of individual well-being and job fulfillment on turnover.

Measurements of staff turnover in healthcare and welfare sectors

Our umbrella review shows the complexity of the concept of staff turnover in the healthcare and welfare sectors. The definitions and terms used to describe turnover intention and actual turnover vary across reviews, reflecting inconsistency in describing these constructs [9*]. Moreover, a multitude of methods and indicators was used to measure turnover intention and actual turnover, and detailed information regarding the measurements used was often missing.

These inconsistencies can lead to misinterpretations and hinder comparative studies, showing a need for standardization of definitions and measurements. This is especially important considering that intentions often overestimate actual actions, such as actual turnover. This means that studies measuring turnover intention may not accurately reflect actual turnover, as various factors can influence an individual's decision between the time of experiencing an intention to leave and actually leaving [22, 88]. For example, an individual might have an intention to leave due to current dissatisfaction, but later decide to stay due to changes in their personal circumstances, workplace environment, or the unavailability of better alternatives. Therefore, relying solely on turnover intention as a proxy for actual turnover may result in misleading conclusions and ineffective interventions. However, including both turnover intention and actual turnover would uncover potential routes to avoid actual turnover. This underlines the importance of using a combination of methods to measure both turnover intention and actual turnover, as well as the need for standardization in the methods used across studies to ensure comparability and validity of the findings. This means that future research should focus on using uniform definitions and scales for measuring turnover to facilitate comparison across studies. In addition, longitudinal designs should be prioritized to capture the progression from turnover intention to actual departure, and emphasize context-specific studies that account for varying healthcare settings and cultural nuances.

In addition, there is a lack of distinction in the measurement of turnover between leaving the organization and leaving the profession. Moreover, internal transfers within the organization (i.e., leaving the team) are also possible. These types of turnover however, have distinct causes and consequences [42, 47]. For instance, leaving an organization may be driven by dissatisfaction with the immediate work environment, while leaving the profession might result from broader disillusionment or stress [75]. Additionally, turnover at the team level, which is seldom addressed, can significantly disrupt team dynamics and cohesion, team performance and – as a consequence – impacting patient care [7]. Future research should aim to differentiate these forms of turnover more clearly, allowing for more targeted interventions.

Interventions to reduce staff turnover in healthcare and welfare sectors

This review has identified a range of interventions that have proven effective in reducing turnover and increasing retention rates. However, it should be noted that the results of various interventions are mixed. For example, job content interventions such as primary and team nursing methods have yielded mixed results; some settings demonstrate benefits, while others do not see as clear an impact, indicating the importance of fitting the intervention to the specific workplace environment [11*, 14]. Similarly, stress coping programs varied in effectiveness but generally showed positive retention outcomes in high-stress units, suggesting their value when tailored to address unit-specific demands [29*]. Working condition improvements were not universally effective, with success depending on how they integrated into the staff's workflow and workload [30*]. Conversely, adjustments to terms of employment through financial incentives and structured growth opportunities have been consistently effective, particularly in supporting the retention of early-career professionals [11*, 19]. These findings show the need for interventions that are not only supportive but also provide clear pathways for professional advancement. Finally, interventions that target enhancements in social relations at work, have also been generally successful in reducing turnover intentions across various healthcare settings [30*, 57*, 77*]. The positive impact of these interventions indicates an important role of a supportive and collaborative work environment for staff retention. To summarize, multifaceted interventions that target both individual and work environment factors could help to reduce staff turnover [18, 57*]. There is no one-size-fits-all solution and effective retention strategies should be as dynamic as the work environments they aim to improve, with an understanding of the contextual factors that influence turnover behavior.

Furthermore, when taking a closer look at the interventions discussed in the reviews, several limitations were identified, including potential bias in participant recruitment, small sample sizes, inconsistent and incomplete descriptions of interventions, variations in evaluation methods, and difficulties in making comparisons across studies due to varying outcome measures [3, 11*, 19, 49*]. These limitations mean that the findings should be interpreted with caution, and the generalizability of the results might be limited. The lack of randomized controlled trials, which provide more rigorous evidence, also underscores a gap in the literature and a need for higher-quality research.

In addition to these limitations, several other considerations should be noted. Firstly, no interventions were found targeting the organizational level, e.g., organizational culture, despite the potential impact of organizational culture on turnover. There is thus a need for more research to fill this gap. Secondly, contextual factors should be considered when implementing interventions. An intervention that is successful in one setting (e.g., a large hospital) may not be as effective in another (e.g., a small nursing home) due to differences in workforce characteristics and other contextual factors. Thirdly, there is a need for cost-effectiveness evaluations of the interventions. While some interventions may be effective in reducing turnover and increasing retention, they may also be costly to implement [11*]. Lastly, there is a need for studies that assess the long-term effects of interventions since many studies only assess the impact of interventions in the short term [11*].

Implications for research

This umbrella review highlights several areas for future research. Firstly, the mixed findings on socio-demographic factors such as age show the need for further research to understand their impact on turnover decisions and develop effective retention strategies. Secondly, there is a surge in research published between 2021 and 2023, potentially linked to increased turnover or turnover intention during the COVID-19 pandemic, indicating a need to investigate the persistence of identified factors post-pandemic [74*]. Thirdly, most primary studies were conducted in the USA, primarily in hospitals, and involved mainly nurses. This indicates a potential research gap in other geographic regions and healthcare and welfare settings, necessitating more research to understand the generalizability of the findings. Fourthly, the minimal overlap in primary studies across the included reviews, as indicated by low CCA values, suggests that the different reviews analyzed a broad spectrum of unique primary studies. Although some redundancy was noted and addressed in our synthesis, the variety in reviews also points to areas within the healthcare and welfare sectors that might not have been fully captured, suggesting there is much more to learn about the factors of personnel turnover in the healthcare and welfare sectors. Fifthly, the predominant reliance on quantitative methods suggests a potential gap in understanding the nuances and complexities of the subject, which cannot be fully captured by quantitative methods alone. Therefore, more qualitative and mixed-methods studies are recommended. Sixthly, concerning the concept and measurement of staff turnover, there is a need for standardized terminology and outcomes to allow for better comparability across studies [14, 86]. For example, to address this need, one could consider assembling an interdisciplinary task force to propose a broadly accepted definition and measurement tool for staff turnover, potentially through a Delphi study to ensure consensus among experts. Finally, regarding interventions, future research might benefit more from adopting hybrid study designs that not only assess effectiveness but also emphasize implementation factors, or from engaging in realist evaluation studies that offer deeper insights into 'what works, for whom, in what circumstances, and why [26, 63]. Additionally, for a better comparison across studies and a clearer understanding of the relationships, future research should standardize the reporting of interventions, assess the feasibility and sustainability of interventions in different contexts, and explore aspects such as theoretical frameworks and long-term measurements [89*, 99*].

Implications for practice & policy

The surge in recent research highlights a growing emphasis on understanding staff turnover [30*]. However, due to variations in measurement methodologies, obtaining an understanding of the actual turnover rate proves challenging. In addition, often only fragmented information is available (e.g., some– mostly larger– organizations that gather data on their staff turnover) and no overall picture of turnover statistics exists at a national level, in most countries (there are exceptions, e.g. the Netherlands). Without access to reliable data, it becomes difficult for policymakers to make informed decisions. We suggest establishing a national turnover database to standardize data collection and ensure regular reporting by healthcare and welfare organizations. This would facilitate better analysis and understanding of turnover trends, enabling policymakers to make more informed retention strategies. Collaboration between researchers and policymakers should be encouraged to analyze this data and develop targeted interventions.

Furthermore, addressing personnel turnover in healthcare and welfare sectors requires a multifaceted, context-specific approach, informed by the specific factors of each organization. Managers and HR professionals must select interventions aligned with their healthcare workers' needs and the organization’s mission, vision, and values [30*]. Practice implications can include, for example, work environment evaluations, employee support, leadership training, career development, and promoting work-life balance. Policy implications can involve standardizing turnover measurements, promoting research, considering contextual factors, long-term effects evaluations of interventions, and addressing COVID-19 specific challenges.

Limitations

This review has several limitations. First, in this overview of systematic reviews, our understanding and interpretation of the primary studies are based on the reporting and interpretation of the review authors. This introduces a layer of interpretation between the original data and our analysis, as biases, errors, or interpretations made by the original authors of the primary studies and the systematic review authors, combined with our reliance on their reporting and interpretation, could potentially affect the accuracy and reliability of our findings. Second, the methodological quality of the included systematic reviews was highly variable, with many reviews showing room for improvement. Third, we applied narrow inclusion criteria to maintain a comparable scope across studies. For instance, we excluded grey literature, which may have introduced publication bias, as relevant evidence from non-peer-reviewed sources was not considered. In addition, we focused on studies conducted in high-income, primarily urban settings. Studies set exclusively in low- and middle-income countries or rural and remote areas were excluded to reduce heterogeneity in system resources and workforce structures. However, turnover dynamics may differ substantially in these underrepresented settings, which limits the external validity of our findings. Fourth, as with all systematic (umbrella) reviews, while efforts were made to include all relevant studies, there is a possibility that some relevant studies were inadvertently excluded during the review process. In addition, with the inevitable time lag from the publication of primary studies to their inclusion in a pertinent review, we are likely to have missed some of the more recent literature, particularly studies emerging in the wake of the COVID-19 pandemic that begin to address its long-term effects. Fifth, not all records were double-screened by two independent reviewers. Given the large number of records, a random sample was checked for consistency. This pragmatic approach may have introduced a risk of selection or extraction bias. Sixth, there was heterogeneity in the included studies in terms of design, methods, and settings. This makes it difficult to make direct comparisons across studies and may limit the generalizability of the findings. Furthermore, the majority of the studies included in this umbrella review were conducted in the USA, and the predominance of studies conducted in healthcare settings, and particularly involving nurses, suggests that the findings may not be generalizable to the whole healthcare and welfare sector. Finally, while we consistently used the term factors to refer to variables associated with turnover, the terminology used in the included systematic reviews and their primary studies was not uniform. Terms such as antecedents, predictors, or determinants were sometimes used without clear conceptual distinction. This variability may have influenced how relationships were reported and interpreted, and should be kept in mind when interpreting the findings of this review.

Conclusion

This umbrella review reveals that turnover in the healthcare and welfare sectors is influenced by a complex mix of factors, including socio-demographic and health status, work environment characteristics, and personal attitudes and functioning. In addition, there is inconsistency in measuring turnover intention and actual turnover, underscoring a need for standardization. Furthermore, while several interventions have shown potential in reducing turnover, their widespread applicability is limited due to methodological disparities and potential biases. These results call for rigorous study designs and standardized reporting in future research. In addition, the surge in research on staff turnover, potentially driven by the COVID-19 pandemic, underscores the need to broaden our research horizons both in terms of location and context.

Abbreviations

AMSTAR 2

A Measurement Tool to Assess Systematic Reviews, version 2

CCA

Corrected Covered Area

CINAHL

Cumulative Index to Nursing and Allied Health Literature

COVID-19

Coronavirus Disease 2019

EPSU

European Federation of Public Service Unions

HCP

Healthcare Professional

Kappa

A statistic that measures inter-rater agreement

LMICs

Low- and middle-income countries

MeSH

Medical Subject Headings

PICO

Population, Intervention, Comparison, Outcome

PRISMA-P

Preferred Reporting Items for Systematic review and Meta-Analysis Protocols

RN

Registered Nurse

SWiM

Synthesis Without Meta-analysis

TIS-6

Turnover Intention Scale (6 items)

UMICs

Upper-middle-income countries

Authors’ contributions

Herlinde Wynendaele: Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Validation, Writing – original draft; Els Clays: Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Supervision, Validation, Writing – review & editing; Ellen Peeters: Conceptualization, Methodology, Validation, Writing – review & editing; Yannai DeJonghe: Formal analysis, Validation, Writing – review & editing; Ann Van Hecke: Conceptualization, Funding acquisition, Methodology, Supervision, Writing – review & editing; Jeroen Trybou: Conceptualization, Funding acquisition, Methodology, Project administration, Supervision, Writing – review & editing.

Funding

Financial support was obtained from the Flemish Policy Research Center for Welfare, Public Health and Family (in Dutch: Steunpunt Welzijn, Volksgezondheid en Gezin).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.