Protective association of work environment on nurses' posttraumatic stress disorder symptoms during the pandemic

FigureThe COVID-19 pandemic has had a devastating effect on the nursing workforce. Although research conducted during the last 2 to 3 decades had clearly quantified that factors such as an unsupportive practice environment and high patient-to-nurse ratios were associated with nurses' intentions to leave, the exodus of nurses during the COVID-19 pandemic was unprecedented. According to a study published in Health Affairs, the national supply of RNs decreased by 100,000 nurses in 1 year during the height of the pandemic.1 New data show that, in reality, these same nurses, rather than leaving the profession, had shifted to different employers.2 Another report said that US hospitals experienced a 23% total turnover rate in 2022 among their RN employees.3 The authors concluded that such turnover rates have taken a financial toll on hospital budgets, with recruitment costs for one RN averaging $52,000, resulting in financial costs per hospital of approximately $6.6 million to $10.5 million per year. A unique phenomenon potentially influencing intent to leave in the context of the pandemic is posttraumatic stress disorder (PTSD). A systematic review and meta-analysis of pandemic-related PTSD in frontline nurses revealed widespread PTSD, with the authors urging both organizational-level change and psychological interventions to enable nurses' recovery.4 A protective factor that may mitigate PTSD is a supportive work environment. In a study of more than 21,000 physicians and nurses working at 60 Magnet® hospitals throughout the US during the COVID-19 pandemic, 63% of nurses reported chaotic work environments.5 Hospital organizational features, such as the work environment, are postulated to affect nurse job outcomes, both positively and negatively.6 A supportive nursing practice environment includes greater nurse autonomy, increased nurse control, more availability of resources, and improved nurse-physician relationships for better patient and nurse outcomes alike. Conversely, an unsupportive nursing work environment makes the nurse vulnerable to negative outcomes. Rather than relying solely on wellness and resilience programming to improve the mental health of nurses, addressing the nursing work environment may achieve a systemwide improvement and takes the onus off nurses to manage their own workplace mental health.5,7,8 The purpose of this article is to determine whether a supportive work environment in the presence of PTSD symptoms decreases the odds of intent to leave in the pandemic era. METHODS A cross-sectional survey was conducted using a list of all RNs with active licenses issued by the New Jersey Division of Consumer Affairs, who had a home address in New Jersey and an email address so that the authors could contact them electronically. The Rutgers University Institutional Review Board reviewed the study and granted expedited approval. About 79% of licensed RNs with New Jersey addresses supplied an email. The invitation was sent from the first author's university email with embedded university letterhead to 100,463 RNs who met these criteria. The authors used a modified Tailored Design Method survey methodology as outlined by Dillman, which entails using specific components in the invitation letter and sending three follow-up invitations.9 The email included an electronic link to a Qualtrics survey that was prefaced by informed consent. The letter presented additional inclusion criteria, such as working in acute care during the pandemic and being currently employed in acute care. There was no way to calculate the response rate among licensed nurses who met these criteria because no denominator was available for these additional inclusion criteria. Overall, 1,817 nurses opened the invitation. Of those, 1,049 recipients passed the consent and screening inclusion criteria questions. In total, 824 nurses answered three key items: intent to leave the job within the next 12 months, the impact of PTSD symptoms, and work environment conditions. This subset was considered the analytic sample, and their responses were subjected to analyses. Intent to leave A nurse's declaration of intention to leave their current nursing job was operationalized as a single item, such as, “Do you intend to leave your job in the next 12 months? (Yes/No).”10,11 Impact of event Impact of event is theoretically defined as an appraisal of an individual's distress regarding a significant life experience.12 In a recent study, the Impact of Event Scale (IES-6) was validated independently as a screening tool for PTSD.13,14 A Cronbach alpha of .80 demonstrated reliability, or internal consistency, for the IES-6.15 Criterion validity was evident in the correlation of the IES-6 to the original IES-R, such as subscale correlations ranging from .78 to .94.15 Respondents were asked to rate each item during the past 7 days, with respect to COVID-19 during the last 3 years. An example item is, “I was aware that I still had a lot of feelings about it, but I didn't deal with them.” The five Likert-type responses ranged from 0 (not at all) to 4 (extremely). Based on the recommended scoring, a mean item score was computed for the IES-6, hypothetical range 0 to 4. Mean item scores on the impact-of-event measure are routinely dichotomized using tested and published cut points.13,14,16 Dichotomization was conducted for the impact of the COVID event score, which has a tested and established mean item cut point of 1.75 when screening for PTSD symptoms reflects an average answer of “moderately.”13,14 Work environment/working conditions Work environment, or working conditions, is theoretically defined as the set of organizational attributes that support professional, clinical nursing practice, including decentralized authority, supportive managers, interdisciplinary collaboration, open and effective communication methods, sufficient resources, and care continuity.17 This study took the measurement approach used by Aiken and colleagues and simplified the measure of work environment to a single item, operationally defined as a single question: “How would you rate the work environment in your current job at this hospital? (Excellent, Good, Fair, Poor).”18,19 Demographic and control variables Nurse demographic characteristics included gender, marital status, race, ethnicity, educational status, job title, unit, employment status, shift, years in the current RN role, and age. Nurse staffing was included as a control variable, given its theoretical and clinical relevance. To evaluate whether nursing unit type was associated with the effect of staffing, we ran a sensitivity model including nursing unit type and found that this was insignificant. Nurse staffing was measured as the number of patients cared for on the last shift worked.20,21 Data analysis The nurse sample and key variables were described using means, standard deviations, and percentage distributions. Bivariate tests were used to explore associations between intent to leave the job and demographic characteristics, including chi-square for nominal variables. Only the demographics that were significant to intent to leave were included in subsequent regression models. Cross-tabulations of key variables were conducted to examine bivariate associations. The extent of missing data for all variables was found to be minimal (for example, less than 5%) except for patient-to-nurse ratio, which was 6.7%. There were 55 missing values, and another 20 were set to missing due to exceeding 20 patients per nurse. Logistic regression was used to analyze the independent variable, impact of event (PTSD symptoms), and the dependent variable, intent to leave. Four models were estimated using multivariate logistic regression. Model 1 was a simple bivariate (unadjusted) model. Model 2 controlled for marriage- and job-related variables. Model 3 added work environment. Model 4 added the control variable, patient-to-nurse ratio. Nurses with missing values for multiple demographic variables that were included in the regression model were excluded from the sample to assure model convergence. Statistics were computed using Stata 17 software.22 RESULTS As presented in Table 1, the majority of the 824 New Jersey RNs were female (88%), married (61%), diverse (72% White and 28% non-White), non-Hispanic (92%), held a bachelor's degree (52%), and had a mean age of 47.3 years (SD 12.2). A total of 78% had a job title in direct care nursing, 27% worked on a medical-surgical unit, 78% were employed full-time, 72% worked a 12-hour shift, and 43% had less than or equal to 5 years of RN experience in their current role. The average number of patients per nurse was 4.9 (SD 3.1). TABLE 1: - Demographic characteristics of study participants (N = 824) Characteristics % Gender (n = 819) Male 10.1 Female 88.4 Other 1.5 Marital status (n = 815) Nonmarried 39.4 Married 60.6 Race (n = 813) Non-White 27.9 White 72.1 Ethnicity Non-Hispanic 92.1 Hispanic 7.9 Educational status (n = 817) Associate 12.6 Bachelor's 52.2 Graduate 36.2 Job title (n = 820) Clinical nursing 77.6 Nursing management 10.7 Advanced practice nursing 4.8 Other 7.0 Unit (n = 818) Outpatient 18.3 Critical care 20.3 Medical-surgical 27.0 Labor and delivery/postpartum and women's health 8.6 Other 25.8 Employment status (n = 819) Less than full-time 21.5 Full-time 77.7 Other 0.9 Shift (n = 820) 8 hours 11.1 12 hours 71.6 Other 17.3 Years in current RN role (n = 817) Less than or equal to 5 years 42.7 Between 6 and 20 years 35.4 Greater than 20 years 21.9 Age Mean SD In years 47.3 12.2 Note: Total percentages for each variable may not equal 100% due to missing data and rounding. Chi-square tests of significance determined that intent to leave the job was statistically significantly associated with characteristics such as marital status (χ2(1) = 11.3, P = .001), unit type (χ2(4) = 11.1, P = .025), shift length (χ2(2) = 6.3, P = .044), and years in the current RN role (χ2(2) = 19.5, P < .001) (see Table 2). Thus, only marital status, shift, unit, and years in the current RN role were controlled when estimating models in which intent to leave the job was the dependent variable. TABLE 2: - Bivariate analysis for all categorical demographic variables and intent to leave the job (N = 824) Intent to leave the job Independent variables No Yes χ2 P % % Gender (n = 819) 4.4 .112 Male 68.7 31.3 Female 57.2 42.8 Other Marital status (n = 815) 11.3 .001 Nonmarried 51.1 48.9 Married 63.0 37.0 Race (n = 813) 3.0 .084 Non-White 53.7 46.3 White 58.6 41.4 Ethnicity (n = 811) 2.1 .147 Non-Hispanic 59.3 40.7 Hispanic 50.0 50.0 Educational status (n = 817) 1.6 .441 Associate 58.3 41.8 Bachelor's 60.5 39.5 Graduate 55.7 44.3 Job title (n = 820) 6.8 .079 Clinical nursing 56.5 43.6 Nursing management 61.4 38.6 Advanced practice nursing 60.0 41.0 Other 73.7 26.3 Unit (n = 818) 11.1 .025 Outpatient 57.3 42.7 Critical care 54.8 45.2 Medical-surgical 52.0 48.0 Labor and delivery/postpartum and women's health 64.3 35.7 Other 66.3 33.7 Employment status (n = 819) 1.1 .563 Less than full-time 60.8 39.2 Full-time 58.0 42.0 Other 42.8 57.1 Shift (n = 820) 6.3 .044 8 hours 65.9 34.1 12 hours 55.7 44.3 Other 64.8 35.2 Years in current RN role 19.5 < .001 Less than or equal to 5 years 49.6 50.4 Between 6 and 20 years 64.7 35.3 Greater than 20 years 65.4 34.6 Half (50%) of this study's sample had IES-6 mean scores that exceeded the 1.75 cut-point, indicating positive screening for PTSD symptoms. The mean and SD for the IES-6 mean scores was 0.50 (SD 0.50). Overall, 42% of nurses intended to leave their job. Most nurses viewed their work environment as fair or poor (58%). There was a statistically significant association between intent to leave and PTSD symptom scores (χ2(1) = 43.1, P < .001) (see Table 3). Of the 286 who reported nonthreshold PTSD symptoms, 60% didn't intend to leave their jobs. Among the 219 nurses who did report threshold-level PTSD symptoms, 63% intended to leave their jobs. Moreover, a chi-square test of significance determined a statistically significant association between work environment and intent to leave (χ2(1) = 75.3, P < .001). Among the 260 nurses who rated their jobs as good/excellent, 75% didn't intend to leave their jobs. In contrast, of the 263 nurses who rated their jobs as poor/fair, 55% intended to leave their jobs. TABLE 3: - Cross-tabulation of intent to leave and PTSD symptoms or work environment (N = 824) Intent to leave PTSD symptoms No Yes No PTSD symptoms 69.4 30.6 Yes, PTSD symptoms 46.8 53.2 Intent to leave Work environment No Yes Poor/fair 45.4 54.6 Good/excellent 75.6 24.4 Multivariate logistic regression analyses were conducted to determine the relationships between impact of the COVID event (PTSD symptoms), work environment, and marriage and job characteristics upon intent to leave the job (see Table 4). The negative psychological impact of the COVID event (PTSD symptoms) was statistically significantly associated with intent to leave the job, χ2(1) = 43.6, P < .001. Nurses with PTSD symptoms have 2.6 times the odds of intending to leave the job compared with nurses without PTSD symptoms (OR = 2.58, 95% CI 1.94, 3.43, P < .001) (Model 1). Models 2, 3, and 4 controlled for covariates, including marriage- and job-related characteristics. In Model 2, PTSD symptoms were statistically significantly associated with intent to leave, χ2(12) = 75.3, P < .001. When controlling for covariates, nurses with PTSD symptoms had 2.6 times the odds of intending to leave the job compared with nurses without PTSD symptoms (OR = 2.56, 95% CI 1.90, 3.44, P < .001) (Model 2). TABLE 4: - Effects of PTSD, work environment, nurse staffing, and patient-to-nurse ratio on intent to leave when controlling for marital status and job-related characteristics Model 1 (N = 824) Model 2 (n = 818)∗ Model 3 (n = 818)∗ Model 4 (n = 818)∗ Unadjusted odds ratio 95% CI P Adjusted odds ratio 95% CI P Adjusted odds ratio 95% CI P Adjusted odds ratio 95% CI P LL UL LL UL LL UL LL UL Predictor variables Impact of event (Ref. No PTSD symptoms) PTSD symptoms 2.58 1.94 3.43 < .001 2.56 1.90 3.44 < .001 2.19 1.61 2.98 < .001 2.11 1.55 2.88 < .001 Work environment (Ref. Good/excellent) Poor/fair 3.34 2.41 4.61 < .001 3.23 2.33 4.47 < .001 Patient-to-nurse ratio 1.07 1.01 1.13 .025 Note: Nurses with missing values for multiple demographic variables that were included in the regression model were excluded from the sample to assure model convergence. The following control variables were included in Models 2, 3, and 4: marital status and job characteristics, such as unit type, shift length, and time in the current RN role. Subsequently, when controlling for both work environment and covariates, PTSD symptoms were statistically significantly associated with intent to leave the job, χ2(13) = 131.8, P < .001. In Model 3, when controlling for marriage- and job-related covariates, nurses with PTSD symptoms have 2.2 times the odds of intending to leave the job compared with nurses without PTSD symptoms (OR = 2.19, 95% CI 1.61, 2.98, P < .001). Nurses with poor to fair work environments have 3.3 times the odds of intending to leave their jobs compared with those nurses who report good to excellent work environments (OR = 3.34, 95% CI 2.41, 4.61, P < .001). Combining the effects of having PTSD symptoms with a poor/fair work environment is equal to 7 times the odds of intending to leave the job compared with nurses without PTSD symptoms (OR = 7.30, 95% CI 4.76, 11.20, P < .001) when controlling for covariates. When controlling for work environment, covariates, and the number of patients cared for on the last shift, PTSD symptoms were significant, χ2(15) = 136.9, P < .001. In Model 4, when controlling for covariates, nurses with PTSD symptoms have 2.1 times the odds of intending to leave the job compared with nurses without PTSD symptoms (OR = 2.11, 95% CI 1.55, 2.88, P < .001). Nurses with poor to fair work environments are almost 3 times more likely to intend to leave their jobs than those nurses who report good to excellent work environments (OR = 3.23, CI 2.33, 4.47, P < .001). For every additional patient in the control variable, the odds of intending to leave increase by 7% (OR = 1.07, 95%, CI 1.01, 1.13, P = .03). Within Model 4, the control variable, patient ratio, is significant (P = .025). The patient ratio is also significant in multivariate logistic regression with PTSD symptom scores and covariates, without work environment (P = .004). In addition, in Models 2, 3, and 4, being married was a significant covariate (P = .016, .041 and .044, respectively), as was having 6 to 20 years of experience in the current RN job role (P = .003, P < .001, and P < .001, respectively). Being married reduces the odds of intent to leave by 31% (OR = 0.689, 95% CI 0.509, 0.934) and, among nurses with more experience, the odds of intent to leave are reduced by 40% (OR = 0.596, 95% CI 0.426, 0.834). DISCUSSION We were motivated to examine whether the work environment had a protective effect on nurses' intent to leave during the pandemic. The combined effects of having PTSD symptoms plus a poor or fair work environment was equal to 7 times the odds of intending to leave the job for these nurses compared with nurses without PTSD symptoms in good or excellent work environment when controlling for covariates. Our intrapandemic data revealed a small protective effect of the work environment as hypothesized. Thus, improving the work environment may be a complementary management strategy in addition to offering mental healthcare services to those with PTSD symptoms. In this study, 42% of nurses intended to leave their jobs within the next 12 months. Our findings are in line with comparable COVID-19 era studies that reported intent-to-leave rates among nurses at 36%, 44%, and 22%.2,23,24 Nurses in our study reported PTSD symptoms at 50%. Prior to the onset of the pandemic, PTSD symptoms ranged between 8% and 21% among nurses.25 In the US, studies reported PTSD symptoms at 33%, with ICU nurses at 22.2%.24,26 A systematic review pooled 25 COVID-era studies and estimated the range of PTSD symptoms to be between 0.4% and 37.4%.4 However, the state of New Jersey had the highest COVID rate of state in the which for the PTSD symptom rate among nurses in this of nurses with PTSD symptoms in this study also intended to leave their jobs. of nurses in this study reported a poor to fair work environment. This is in with another US study that a work environment in the and In the current study, 55% of those nurses who reported poor/fair working environments also intended to leave. In contrast, a good to excellent work environment entails nurse autonomy, sufficient and resources, and relationships between nurses and Prior and a recent study that supportive work environments are workplace negative psychological effects and control the work environment is a factor in potentially intent to leave among those PTSD symptoms. In this study, the control effects on intent to leave were small but The average patient per nurse was 4.9 Prior to the pandemic, a patient-to-nurse ratio, with was to in improving nurse and patient the of is a The responses from those who didn't may have Although this study had 824 are approximately active New Jersey RNs in those from PTSD symptoms were more likely to the survey and, the may not be of the state nursing workforce. Of those nurses who did we the worked at the same or different is about those nurses who the to the This survey only those in the state of New a small of the this study on a national would about the nursing workforce. Although not a for mental healthcare for the nurse may be to intent to leave among nurses who report PTSD symptoms by improving the work environment. The Scale of the Nurse Work five which are used to measure work environment but also a for survey their nurses, their scores to values, and on the that has the There are and tested interventions that be such as nurse role and nurse-physician findings the but significant, impact of good and excellent work environments and PTSD symptoms on nurse outcomes. This study that supportive work in the presence of PTSD symptom may the intention to leave. to work environment and reported PTSD symptoms and intent to leave likely in a of RN