Author: Robert Hirst/ Editor: Govind Oliver / Codes: SLO10 / Published: 18/01/2020

The CERA study was devised in the aftermath of hearing about the first deferral of the SHED study (see our website for more information about the current state of it). We were interested in seeing how the pandemic was affecting clinicians working on the frontline. We based our study around the pandemic interval framework (seen below):

From the Preparedness & Response Framework for Influenza Pandemics

We sadly now find ourselves in the midst of another wave of this pandemic. The findings of the CERA study are as apposite now as they were then. We want to share some of the findings here. The full pre-print can be found here.  

Background

Clinicians working across emergency medicine (EM), anaesthetics & intensive care medicine (ICM) have been vital to the treatment of patients with COVID-19 (which is not to diminish the valiant efforts of everyone in GPs, hospitals, care homes, paramedics, and everywhere else). 

The COVID-19 Pandemic has imposed unprecedented (there’s that word) demands to workload intensity & personal health risks to clinicians1. We know that exposure to infectious disease outbreaks & elevated psychological distress is associated with increase rates of sickness, absenteeism, impaired work performance and development of physical and mental health problems. Most of the data previously collected has been snapshot data – there is little longitudinal data examining the evolving and cumulative effects on the psychological wellbeing of frontline doctors.

Study aims

Assess prevalence & degree of psychological distress and trauma of doctors during the acceleration, peak, and deceleration phases of the first wave of the COVID-19 pandemic. Establish which personal & professional factors were significantly associated with psychological distress & trauma.  

Methods

The CERA study was a prospective, online, longitudinal survey administered to all doctors working in EM, Anaesthetics or ICM in the UK & Ireland during the first wave of the pandemic. There were no exclusion criteria.

The survey was distributed through training networks, training faculties or Royal College Networks via email and instant messaging groups during the acceleration, peak, and deceleration phase (30 days post-peak). Determination of the peak for timing of the peak survey was determined by national case & national death rates. Dates for these can be found in the preprint, but differed between the UK & Ireland. The survey link was kept off wider social media to prevent international contamination of data.

Our primary outcomes were the General Health Questionnaire-12 (GHQ-12), which was used to assess psychological distress, and the Impact of Events Scale – Revised (IES-R), which was used to measure trauma. The GHQ-12 was administered at all three surveys and the IES-R was administered at the peak & deceleration surveys. A number of other personal and professional factors thought to impact upon distress & trauma from the literature and from discussion withing the steering group were collected across the surveys. 

Statistics

For the GHQ-12 we reported two measures – bimodal scoring (with responses scoring 0, 0, 1, 1) and Likert scoring (0, 1, 2, 3). We used bimodal for assessing the prevalence of psychological distress (which is conventional in clinical practice2) and Likert for comparison across time. We set our score for distress as >3. An IES-R score >24 indicated a clinically significant traumatic stress response, and >33 indicated a diagnosis of probable post-traumatic stress disorder (PTSD).

Our final analysis cohort was composed of those who completed all three surveys. The change over time in scores was examined using a repeated measures linear mixed-effect model, with survey phase as a fixed effect & a participant level random effect. 

To identify the potential effects of our personal and professional factors on GHQ-12 or IES-R scores over time, further models with each of our factors were added individually as a single additional covariate, including an interaction term with survey phase. We used Nagakawa’s marginal R2 to measure the proportion of outcome variance accounted for by the model (excluding random effects). Values ranged from 0 to 1, with occurring when the model returns the population average and 1 occurring when the model perfectly predicts the outcome. Easy, right? 

Results

Of an estimated 34,188 doctors working across EM, Anaesthetics and ICM in the UK, the acceleration survey received 5440 responses (a 15.9% response rate). Follow-up responses during the peak period were 3896 (71.6%) and 3079 (56.6%) during the deceleration period. 

Of our final analysis cohort, 54.8% were working in EM, 36.2% were working in anaesthetics and 17.2% were working in ICM, with some working across multiple specialities. 

The prevalence of distress was 44.7% during acceleration, 36.9% at the peak, and 31.5% at deceleration. Median scores were higher in Anaesthetics & ICM during the acceleration phase (14.4% and 14.0% compared to EM 13.3%), but all decreased throughout the first wave.  

The prevalence of trauma was 23.7% at the peak and 17.7% at deceleration. The prevalence of probable PTSD was 12.6% at the peak and 10.1% at deceleration. The prevalence of trauma & probable PTSD were highest in EM & ICM compared to Anaesthetics (13.9% & 13.6% compared to 10.8%) at the peak. 

The factors most strongly associated with distress was worry concerning infecting family members and worry of becoming infected. The factors that were most strongly associated with trauma were worry of infecting family members, worry about exacerbating an established mental health condition, personal infection, and ethnicity. 6.9% of respondents had been diagnosed with COVID-19 by the deceleration, but a positive diagnosis was not predictive of trauma.

Discussion

The rates of psychological distress & trauma peaked at 44.7% and 23.7% respectively, far higher than for the general population3. There was some natural recovery in terms of both distress & trauma, but a significant proportion of patients still experiencing residual distress or PTSD symptoms 30-days post-pandemic. 

The factors most associated with elevated distress & trauma related to familial and personal safety, highlighting the importance of adequate PPE for this and future pandemics. The impact of the pandemic on mental health and those with mental health conditions was also illustrated, with this the factor most likely to be associated with trauma. The role of ethnicity was also identified as a novel predictor of trauma, perhaps unsurprising giving the higher rates of reported mortality in ethnic minority groups4. 

It is clear from the data that vulnerability to poorer psychological outcomes may be predicted by certain characteristics, and potentially mitigated through targeted intervention. Previous work around the SARS pandemic reflected that psychological distress is likely to persist5, so it is important that we identify & intervene now. If you’re interested in reading more about this study, you can find the full pre-print here.  

References

  1. McCabe R, Schmit N, Christen P, et al. (2020). Adapting hospital capacity to meet changing demands during the COVID-19 pandemic. BMC Medicine 18: 329.
  2. Goldberg, D. et al. (1997). The validity of two versions of the GHQ in the WHO study of mental illness in general health care. Psychological Medicine 27(1): 191-7. 
  3. Rettie, H., Daniels, J. Coping and Tolerance of Uncertainty: Predictors and Mediators of Mental Health During the COVID-19 Pandemic. Am Psychol 2020. 
  4. Office of National Statistics. (2020). Coronavirus (COVID-19) related deaths by ethnic group, England and Wales. (Accessed 14th January 2021). 
  5. Allan et al., (2020). The prevalence of common and stress-related mental health disorders in healthcare workers based in pandemic-affected hospitals: a rapid systematic review and meta-analysis. European Journal of Psychotraumatology 11(1): 1810903.