Correlates of turnover intention among nursing staff in the COVID-19 pandemic: a systematic review

To achieve the study aim, a systematic literature search was conducted. Based on the three PEO components [25] population (i.e., nursing staff), exposure (i.e., COVID-19 pandemic), and outcome (i.e., turnover intention), a search string was built (see Additional file 1). As mentioned above, there is no clear definition of “turnover intention”. This study refers to all types: “Turnover intention” is used as an umbrella term for “intention to leave”, “intention to quit”, “consideration of attrition”, and “consideration of resignation” [21, 22]. Positively formulated terms such as “intention to stay” are excluded from the scope of this review because they describe different constructs with different work-related factors as correlates [26, 27].The search string included synonyms in English and German with truncation as well as Medical Subject Heading terms with automatic explosion and was applied to all fields. A filter for publication in the years 2020 and 2021 was set. The search was conducted on 14 October 2021. In order not to miss any new publications during the analysis and writing process, an alert for the search was installed. The last alert results were accounted for on 31 December 2021.

The inclusion criteria were: a) nursing staff with patient contact as the target population, b) data assessment during the COVID-19 pandemic, and c) original peer-reviewed articles d) in German or English e) using a quantitative design with f) turnover intention as an outcome and g) at least one potential correlate, i.e., working conditions or demographic characteristics. Exclusion criteria were: a) target populations consisting of others than fully educated nursing staff (with patient contact) without profession-specific analyses and b) solely positive constructs, e.g., intention to stay, as outcomes. During data extraction, a third exclusion criterion was formulated: c) turnover intention assessment item implies possible correlates (e.g., “considered leaving nursing because of the workload, stress, and fear” [28]) other than COVID-19 (e.g., “Due to COVID-19, next year I will probably look for a new job outside this organization” [29]).

To cover a wide range of medical, psychological, and nursing science sources, the databases MEDLINE, CINAHL, PsycINFO, PSYNDEX, PsycArticles, and SocINDEX were searched via the platform EBSCOhost. The resulting articles were exported to the bibliographic management software Endnote 20.1. Duplicates were automatically eliminated by EBSCOhost comparing the citation metadata title, author, date published, and ISSN or journal name or ISBN. Two researchers (K.T. and K.H.) screened titles and abstracts independently using Rayyan, a web application to perform collaborative systematic reviews. Duplicates not detected by EBSCOhost were removed. Conflicting judgments from both reviewers were discussed until a consensus was reached. After this, full-text articles were screened based on the inclusion and exclusion criteria, again independently by the same researchers (K.T. and K.H.). Eventually, data were extracted and quality appraisal performed, evenly divided between all authors. K.H. checked all extracted data by comparing them with the original articles. According to Stone et al. [30], we did not exclude articles of low quality in order to avoid selection bias. For the included articles, the references were tracked and citing literature was cross-checked on 31 December 2021, again evenly divided between all authors. The final set of articles was synthesized qualitatively because of the methodological heterogeneity of the found studies.

Table 1 presents the data extracted from the 19 included publications. The studies were conducted in the Middle East (incl. Pakistan; n = 7), Southeast Asia (n = 5), North America (n = 3), Europe (n = 2), Australia, and Taiwan (n = 1, each) in the years 2020 and 2021 and reported samples between n = 64 and n = 1,705 (mean = 398). The instruments to assess turnover intention mostly comprised one or two items (n = 11), up to 15 items, but were not always described in detail. All studies except for one which followed a mixed-methods approach used a cross-sectional design analysing self-report questionnaire data. Several inference-statistical methods were applied to check the found associations for significance, mostly regression analyses (n = 12). The quality appraisal of the publications resulted in sum scores between 4 and 11 out of 12 possible points (mean = 8.6) [31].

Figure 2 displays which factors showed significant associations with turnover intention. These factors can be clustered into the domains “individual factors” and “organizational factors”. Individual factors refer to psychological and demographic characteristics. Organizational factors comprise work demands, ethical issues, and aspects of employer support.

During the COVID-19 pandemic, several psychological characteristics of the participating nurses were found to be associated with turnover intention (see Fig. 2). Several studies identified anxiety or fear of COVID-19 or perceived threat by COVID-19 as a correlate of turnover intention [29, 34‐36, 38]. Besides general health [45], further psychological symptoms showed relationships with turnover intention, e.g., (compassion) fatigue, burnout, or PTSD symptoms [40, 42, 44, 45]. Discrimination [41], stigma [43], or low pride in being a nurse [33] were found to be associated with turnover intention, whereas resilience was negatively correlated with the outcome [32, 40, 41]. Male nurses reported lower levels of turnover intention [45], as did well-educated and specially trained nurses [43]. Regarding legal status, the results were inconclusive [45, 47].

Nurses who were (recently) deployed into critical care [43, 47, 49], received or cared for COVID-19 patients [33, 42, 47] or experienced an infection themselves or in their team reported higher levels of turnover intention [42]. Further job-related factors associated with turnover intention were low job control [42, 45], increased working hours or workload [43, 49], and job stress/strain in general as well as job insecurity [45, 47]. Nurses satisfied with their work or pay reported lower levels of turnover intention [42]. The findings concerning work experience were inconclusive [42, 47].

Several studies showed that ethical issues were associated with turnover intention, among them moral distress (positively) [50] as well as ideological contract [29] and quality of care (both negatively) [40, 53]. Employer support forms a third domain of organizational issues that are connected to turnover intention. Nurses who experienced supporting leadership, employer or social support, or a good organizational culture reported lower levels of turnover intention [42, 43, 45, 52]. When nurses felt poorly prepared [42] or not connected to their work environment [33], they reported higher levels of turnover intention. Lacking communication about coronavirus planning or lacking safety [53] as well as violence and mobbing [49] were additionally identified as correlates of turnover intention.

Psychological characteristics like anxiety, fear, and perceived threat of COVID-19 are factors associated with nurses’ turnover intention [29, 34‐36, 38] which emphasizes the existential threat many nurses are experiencing in this pandemic. Also in the severe acute respiratory syndrome (SARS) pandemic, the perceived risk of death from SARS was an important predictor of nurses' turnover intention [54]. Leadership support [36] and ideological contract [29] may reduce the influence of fear on turnover intention. Even outside of crisis situations, the protective effect of ideological motives on turnover intention is known [55].

Turnover intention is also associated with (psychological) health factors and symptoms [40, 42, 44, 45]. The association between PTSD and turnover intention was also evident in the Middle East respiratory syndrome (MERS) epidemic, although, as with anxiety, leadership support was able to mitigate the relationship between PTSD symptoms and turnover intention [56]. Relieving health-impaired nursing staff through suited offers of health promotion interventions should be carefully considered, but more research is needed on how adaptive strategies can reduce the long-term impact of mental health threats like burnout [44]. Furthermore, personnel with leadership tasks should be trained to offer their staff the support they need especially in times of crises. The positive relationship between resilience and retention [32, 40, 41] has been consistently reported in pre-pandemic literature [57]. This review shows that resilience also mitigates the associations of compassion fatigue [40] and COVID-19-associated discrimination [41] with turnover intention. COVID-19 associated discrimination and stigma against nurses [41, 43] may be due to fear of infection [58] and were repeatedly described as a serious problem [59‐61]. Strategies to improve retention could start with resilience promotion in the form of mindfulness-based stress therapy [62]. Firstly, since it is conceivable that the experience of stigmatization was particularly severe in the first phase of the pandemic due to the novelty and unfamiliarity of the virus [63], nurses’ resilience should be given special protection and fostering especially in the onset of a pandemic. Secondly, since a higher rate of compassion fatigue was mostly observed in nurses assigned to critical care units, emergency departments, and units designated for treating and managing patients with COVID-19 [64], nurses in these areas of work should be favoured for interventions. Additionally, public campaigns are conceivable to reduce nurses’ stigmatization and discrimination by the public and foster nurses’ pride in their work.

Concerning demographic characteristics, there are diverging results with regard to marital status and gender [45, 47]. This inconsistency is also reflected in the results of other studies. Depending on the population and setting examined, these play no [65] up to a significant role [66] in the context of nurses’ turnover intention. Analogous to the result of Mirzaei et al. [45], male nurses are more likely to be prone to turnover intention in some countries outside of pandemic situations [24] with strong differences in relevance of gender as a contributing factor to turnover intention between countries [67]. The diverging results regarding demographic factors can be due to different cultural conditions, family structures, and gender roles of nurses in different countries and settings. Age does not seem to be an associated factor of nurses’ turnover intention in the COVID-19 pandemic which is consistent with findings from the SARS pandemic in the years 2002 to 2004 [54].

High levels of social support from supervisors and colleagues seem to be associated with lower turnover intention [36, 43, 45]. This finding could be explained through the process of stress reduction [68] and has been evident under normal conditions [69] and in the MERS epidemic [56]. High leadership support may also be able to decrease the relationship between fear of COVID-19 and turnover intention [36]. The central role that supervisor support plays in influencing turnover intention has been widely acknowledged [17], and the positive effect of both leadership support [36, 43, 45] and leadership style [42] emphasizes the importance of social relationships, appreciation, and protection apart from purely monetary remuneration and also opens up opportunities for managers to improve satisfaction and retention among their employees. Nevertheless, pay satisfaction could also be a relevant factor [52] that should be considered in nurse retention efforts, but due to the poor quality of Widodo’s study [52] and the stronger evidence regarding the importance of leadership support in this review, the latter should be treated a priority.

Since factors related to organizational culture [52] and feelings towards team climate [33] as well as exposure to violence and mobbing [49] also seem to be relevant factors to nurses’ turnover intention, employers should not neglect interpersonal conditions in their organization and promote a positive workplace culture [70] as well as violence prevention measures and the implementation of support systems. Since it was also found that there was a correlation between turnover intention and nurses’ feeling that they are poorly prepared [42], their safety was not prioritized [53], and management communication on pandemic planning [53] was insufficient, those measures could create a well-founded sense that the safety of nurses is a high priority.

Working in [33, 42, 47] and redeployment to [43] COVID-19 patient care as well as a general department change [49] emerged as correlates of turnover intention. This work context seems to be particularly critical, which may be due to the numerous stress factors present in this setting. In the MERS epidemic, nurses involved in the direct care of suspected patients were also prone to increased turnover intention [49]. However, the SARS pandemic also showed that nurses caring for SARS patients were less likely to consider leaving. This finding was probably related to the nurses having received relevant training, which enabled them to better assess the risks so that they were less affected by fear [54]. The results of Li et al. could confirm this assumption, as they identified education and training concerning the COVID-19 pandemic as a protective factor for turnover intention [43]. This finding is in line with results from the SARS pandemic [71]. COVID-19 training should include information about the proper utilization of available resources, the nature of the virus, precautionary measures to avoid transmission, number of new and recovered cases reported per day as well as hospital protocols [36] and could be executed remotely to maintain social distancing. The results by Nashwan et al. [47] also point out that nurses dealing with COVID-19 patients in intensive care units for the past three to six months are particularly at risk of turnover intention. Even before the COVID-19 pandemic, intensive care nurses were known to be more likely to quit due to prolonged exposure to traumatic experiences and stress [72]. However, this component opens up a time window for targeted preventive interventions in the first three months of employment in COVID-19 intensive care or indicates that employment in this area could be limited in time. In addition, the NEXT study showed that there is also a six-month window between the formation of turnover intention and the actual dismissal of nurses during which preventive measures could be taken [73].

Increased working hours were found to be correlates of turnover intention in two studies [43, 49] as well as an increased number of patients [49] and higher job strain [45]. Additionally, nurses’ feelings of not being in control of the situation at work [42] and a low decision latitude and job insecurity [45] contribute to this intention. A high workload with little room for decision-making results in stress [74], which itself has the potential to contribute to turnover intention as well [44, 47]. However, the influence of workload on stress can again be reduced by supervisor support [75]. While the workload is unlikely to be reduced in the short term, especially in the early phase of a pandemic, in terms of decision latitude, it is possible to provide nurses with learning opportunities and participation in decision-making processes, e.g., concerning the implementation of infection protection measures. In addition, since the feeling that management communication on pandemic planning was insufficient [53] is associated with nurses’ turnover intention, involving nurses in the pandemic planning could at the same time eliminate this lack of communication. Furthermore, this could lead to a better understanding of the pandemic situation, reduce fear, and promote self-efficacy as well as job control. Stress reduction measures and learning adaptive coping strategies [44] could also reduce nurses’ turnover intention in this pandemic, if stress reduction itself is not feasible due to the crisis situation. Conversely, nurses with high levels of stress and maladaptive coping strategies can be identified as a particularly vulnerable group for turnover intention. Other studies have shown that stress and management problems outside of pandemics have both direct and indirect effects on job satisfaction and the intention to leave the company [76]. There is also a vicious circle between job stress and job satisfaction: intense stress leads to job dissatisfaction, which in turn increases the stress [77]. In the SARS pandemic, stress even proved to be the most important predictor of nurses' intention to leave [54]. Lavoie-Tremblay et al. [42] found that work satisfaction also could be a relevant protective factor of nurses’ turnover intention in the COVID-19 pandemic.

Differing results were present regarding work experience. Lavoie-Tremblay et al. [42] found that less experienced nurses are more prone to turnover intention while Nashwan et al. [47] found that nurses with five to ten years of work experience are more at risk. So far, both under normal circumstances [78, 79] and during the outbreaks of SARS [54] and MERS [56], less experienced nurses were more at risk of turnover intention. This finding could be explained by the fact that more experienced and therefore mostly older nurses find it more difficult to change jobs due to family obligations or a stronger sense of duty towards their organization due to longer employment [54]. The fact that Nashwan et al. [47] found experienced nurses more likely to report turnover intention may be due to a perceived threat to their own health or to the health of their families and may indicate the risk of losing experienced nurses in the current pandemic. However, because of the weaker quality of Nashwan’s study [47] and the fact that their results do not coincide with previous knowledge, this evidence should be considered with caution. Nevertheless, a self-infection with COVID-19 or that of a team member is likely to increase turnover intention [42] which emphasizes the need for sufficient and adequate personal protection equipment.

The moral distress nurses face due to system-related factors [50] and their work environment [51] could increase turnover intention. Moral distress is connected to perceived quality of care [80], which itself was shown to be associated with turnover intention [40, 53] in this review. Petrisor et al. [50] pointed out that in the pandemic situation, intensive care nurses could have benefitted from interventions targeting the organizational aspects of workflow since root causes of moral distress should be targeted. Additionally, the opportunity to get consultation by ethics committee in case of moral distress could be considered [51].

This study shows some strengths and limitations. Through data assessment, analysis, and interpretation, a team of researchers with different professional backgrounds (i.e., nursing, public health, psychology) was involved. We used researcher triangulation to ensure data quality. After our initial search covering this up-to-date topic, an alert was installed in order not to miss any new publications during the analysis and writing process. This way, publications until 31 December 2021 were accounted for to grant the most recent coverage. Furthermore, articles from all countries were included to offer a wide range of perspectives and experiences, e.g., because countries were affected differently by the pandemic. One might argue that this is also a weakness because health care systems might not be comparable across nations, but some relationships were found in several countries (e.g., association of turnover intention with moral distress) and get more emphasis this way.

However, there are also some limitations to report. Our systematic search only included publications in English or German. Therefore, we might have missed peer-reviewed articles from national journals. However, we did not include any articles in German, which might lead to the notion that we covered the majority of relevant publications. Furthermore, we did not accept preprints, although their quality is comparable to peer-reviewed articles [81]. With our strategy to create an alert and to thereby include all relevant articles until 31 December 2021, we tried to account for the most recent peer-reviewed publications.

Due to the heterogeneity of the included articles and the incoherent operationalization of turnover intention, we could not conduct a quantitative meta-analysis. The description of how turnover intention was assessed was incomplete in so many articles that we cannot give any synthesized information on factors specifically associated with organizational or professional turnover.

This study was developed from a thesis by the main researcher (K.T.). To account for the latest publications and to ensure data quality through researcher triangulation, we repeated the whole selection process with two reviewers (K.T. and K.H.). The knowledge of the thesis could have influenced the researchers’ judgement during the selection process. However, we discussed each result thoroughly in case of doubts no matter if it was included in the thesis or not. By the time this study was realised, the main researcher (K.T.) was enrolled as a student but closely supervised by U.T. and K.H., who has experience in conducting systematic reviews [82].