Protecting the Well-Being of Healthcare Workers During the COVID-19 Pandemic

Background

Frontline health care workers (HCWs) are facing unprecedented stressors during the COVID-19 pandemic, resulting in increased risk of psychological distress and mental health disorders. Research from past infectious disease outbreaks suggests that the effects of COVID‑19 on HCWs will be profound and long-lasting. The well-being of HCWs has a significant effect on patient experience of care. Before the pandemic, studies showed that HCWs with burnout have lower productivity, higher system costs due to worker turnover, more self-reported medical errors, lower quality of care, and worse relationships with their patients. However, interventions with rigorous evaluations to protect HCWs’ mental well-being during disease outbreaks or other disasters are scarce. Stress First Aid (SFA) is an evidence-informed and promising intervention to mitigate the psychosocial impact of COVID-19 on HCWs. It is a good fit for HCWs because it is designed to address ongoing workplace stressors, is a self-care peer-to-peer support model that is associated with less stigma than formal mental health care, and can be delivered by non–mental health professionals.

Objectives

Our objectives were to (1) test the effect of SFA vs usual care (UC) already provided by hospitals and health centers on HCW well-being, (2) understand and document any UC activities in place to support HCW well-being before implementing SFA across sites, and (3) assess the experiences of HCWs with SFA and their impact on HCW well-being.

Methods

We recruited a diverse set of hospitals and health centers (8 matched pairs of hospitals and 6 pairs of health centers) to participate in a cluster randomized controlled trial (cRCT) (ClinicalTrials.gov identifier NCT04723576). Participating sites in each pair were randomly assigned to SFA or UC (ie, the psychosocial support currently being received by HCWs at that site). Each site identified a leader to provide organizational support of the study; SFA sites also identified at least 1 site champion to be trained in the intervention. Using a train-the-trainer model, site champions trained their peers in selected HCW teams or units to implement SFA over an approximately 8-week period. We surveyed HCWs before and after the implementation period. Primary outcomes were (1) general psychological distress (measured by the Kessler Psychological Distress Scale [K6]) and (2) symptoms of and probable diagnosis of posttraumatic stress disorder (PTSD) (measured by the PTSD Checklist for Diagnostic and Statistical Manual of Mental Disorders [Fifth Edition]). Secondary outcomes included sleep impairment, workplace stress, burnout, resilience, and moral distress. In addition, through in-depth qualitative interviews with leaders, site champions, and HCWs, we assessed the implementation of SFA, including acceptability, feasibility, and uptake. We used intention-to-treat (ITT) analysis to evaluate the impact of SFA on these outcomes. We used Dedoose to code excerpts from the qualitative data.

Results

Both the preintervention and postintervention surveys were completed by 2077 HCWs (response rates, 41% for HCWs from health centers and 26% for HCWs from hospitals, for a total response rate of 28%). The HCWs in the health center sample were younger and more likely to have been in the profession for at least 5 years. Among HCWs in the hospital sample, there were more people identifying as non-White Hispanic. Adherence to the SFA intervention was 70% for health centers and 32% for hospitals. The ITT analyses revealed no overall treatment effect. The change in general psychological distress was 0.238 on the K6 (P = .395), and the change in PTSD symptom score was 0.189 (P = .768). Because we observed consistently strong age effects across outcomes, we ran post hoc analyses to examine the heterogeneity of the treatment effect by age group. The adjusted difference between treatment and control groups for HCWs 30 years of age and younger in the health center sample is a reduction of more than 4 points on the 0 to 24 scale for psychological distress (P = .012; 95% CI, −8.1 to −1.0) and nearly 7 points on the 0 to 80 scale on the PTSD symptom score measure (P = .040; 95% CI, −13.2 to −0.32), both of which are clinically meaningful effect sizes. This strong and consistent effect was not observed in other age groups. Although the cRCT did not find an overall treatment effect, interviewees reported that SFA brought problem-solving and awareness to the forefront, gave HCWs techniques to avoid putting the stress on patients, and facilitated the verbalization of concerns with supervisors. Key themes discussed related to SFA included empathy for patients, empathy for staff, and support for peers. Site leaders for SFA and UC sites reported similar resources already in place to support HCWs, except for informal peer discussions, which were more common among UC hospitals than SFA hospitals. Stress First Aid implementation lessons included communication across organizations and improvement in patient care.

Conclusions

Our study did not find an overall significant effect of SFA on HCW well-being. However, this rigorous mixed-methods evaluation yielded a number of richly detailed insights into HCW experiences during the COVID-19 pandemic and lessons about how best to engage with frontline workers in health care settings.

Limitations

Our study had some limitations, many of which related to its occurrence during the COVID-19 pandemic. Although we designed the study to be run in 3 cohorts or “waves,” the cohorts were less distinct than we had intended, and many of the sites within different cohorts had overlapping timelines. Fortunately, this timeline issue affected both sites in a pair almost equally because the site assigned to UC simply fielded their survey later (at the same time as its paired SFA site). It is unclear whether the “dose” of the intervention was sufficient; we may have not seen an effect because we did not choose an intensive enough implementation of this intervention. Another limitation is that our follow-up period may have been too short to sufficiently appreciate the impact of our intervention on the chosen outcomes. Finally, it was difficult to identify adequately matched hospital pairs; there may have been unmeasured differences between hospitals, but this should have affected each group to the same extent.