We spend a large proportion of our lives at work, so we must pay attention to our own well-being. A casual stroll through Twitter may reveal any number of well-being initiatives—from early morning yoga classes to communal choirs to meditation mornings. While many of these appear to work subjectively, there is clearly a need for a better measure of well-being. Enter the NFR.
What is the NFR?
The Need For Recovery scale measures the subjective feeling of the need to take a break and recuperate from the emotional and physical demands of a day at work.
Originally derived from a sample of 68775 shift workers in the Netherlands, it comprises 11 dichotomous Yes/No questions.
In the emergency department, several extrinsic factors may affect our Need For Recovery. These include departmental crowding, lack of resources, and ambulances queuing out the door. The assumption is that the daily accumulation of such tiring shifts, without a restorative break, can impact our long-term health and increase occupational stress.
To assess our current performance, it might be worth taking a snapshot of the adult and paediatric emergency department workforce. PERUKI and TERN combined their might to do just this.
Population
This study was coordinated through the Trainee Emergency Research Networks (TERN) of the UK and Ireland and Paediatric Emergency Research in the UK and Ireland (PERUKI).
A total of 4247 emergency physicians* from 112 emergency departments completed the survey.
For this study, an emergency physician was defined as any doctor working in the ED, whether a seasoned veteran or a postgraduate year 1 doctor completing their six-month term in emergency medicine.
Exposure
Participants were invited to participate in an online version of the Need for Recovery instrument. Although developed for industry, it has been piloted in the Emergency Department staff. Graham et al. (2020) trialled the survey in a single ED and achieved an 80.3% response rate (168/209).
These subjects had to complete the 11-point NFR survey and 32 additional questions. This more extensive survey raised the ante, adding 44 items to the NFR. The quality of the web-based survey was maximised using the CHERRIES checklist.
Snapshot data was collected over six weeks from 3rd June 2019, long before anybody had even heard of SARS-CoV-2.
Outcomes
Before we look at the Need For Recovery, it is worth examining the demographic data.
3445 (83.5%) survey respondents worked full-time. 609 (14.7%) worked at 80% or fewer full-time equivalents.
2886 (70.3%) worked more than one weekend in four, with a shocking 1479 (36%) working every other weekend!
36.2% worked for at most 4 consecutive days, with the majority working between 5 and 7 days in a row. 13.5% (554) had been rostered for eight days straight.
Emergency physicians score higher than paediatricians, miners, paramedics, everyday folk, nurses, merchant sailors and truck drivers.
The median NFR score (remember, you want it as low as possible) was 70.0 (with 95% confidence intervals ranging from 62.0 to 78.0). It appeared to be higher in the more junior cohort – those who had been in the department for less than a year. This group scored an average of 72.7 with an upper limit of 90.9. Those who had spent more time on the job seemed to recover better, with a median NFR of 63.6. Those lucky individuals who had made a career in emergency medicine and had spent over ten years in the post had the lowest score, 54.5.
As one might expect, higher scores were associated with a full-time work commitment, a burdensome weekend roster and a higher number of consecutive days worked.
Lower scores were found in consultants and those who worked less than full-time. Access to study leave and annual leave was also associated with lower scores. Luckily for those of us who have pursued a career in paediatric emergency medicine, this was also associated with a lower NFR score.
The authors conclude that three modifiable risk factors are associated with a higher NFR: access to annual leave, study leave, and the proportion of out-of-hours work. Non-modifiable factors include male gender, seniority, generally good physical health, and working in a Paeds ED.
Risk of bias
So far, we have reported the facts, as presented, but let’s get a little nerdy, as Ken Milne would say, and look at the paper using Burns and Kho’s (2015) assessment guide for survey reports.
Was a clear research question posed?
Yes, the researchers wanted to examine Need For Recovery scores in a cohort of doctors working in emergency departments throughout the United Kingdom and Ireland.
Was the target population defined?
Yes but… – the target population was defined as any registered doctor who had a fixed contract position (i.e. not a locum) in an emergency department in the UK and Ireland.
There are 183 Type 1 Emergency Departments in England alone. These are consultant-led, 24 hours a day, 7 days a week, 365 days a year services. The authors asserted that they wanted to have over 50% of their respondents from Type 1 centres but this data is not clear in the data presented. Are the majority of respondents from centres that do not have good consultant support?
According to 2018 data, 26% of advertised UK EM consultant places are unfilled. I would be interested to know if individual departmental data could be pulled out and benchmarked against the national average.
Was a systematic approach used to develop the questionnaire?
Yes – the technique was clearly outlined in the methods.
Was the questionnaire tested?
Yes – the questionnaire developed was very similar to that used by Graham et al. The only key difference being around some of the ancillary questions, rather than those analysed in this paper
Were they administered in such a way that response and non-response bias was limited?
Yes – it was distributed in an appropriate way.
Was the response rate reported?
No – though all principal investigators should be applauded for the large number of surveys completed, it is not clear what the denominator is. How many emergency physicians, as defined by the study group) were working during the six week period? Were those doctors who did not even attempt to complete the survey just too exhausted to do so?
Each site PI did provide an best-guess estimate of the number of potential respondents (accounting for sick leave, sabbaticals, annual leave etc.). This number was then used as the departmental denominator, with each site aiming for a 70% response rate. The actual response rate is not mentioned in the paper.
Were the results clearly and transparently reported?
Sort of – there were actually 5107 unique visits to the survey site but only 4247 eligible for analysis. The NFR scores were then calculated as long as a minimum of 8 of the 11 questions were answered. The authors do not mention how they handled the missing data. If only 8/11 answers were provided, how would the addition of 3 further data points affect the results?
The demographic data was clearly reported, barring a few items, as described above, and the rest of the results were presented neatly.
Although I agree with the majority of the authors conclusions I am not so sure I would agree with the assertion that NFR score is unrelated to hours worked. It would seem from the data provided in table 2. those working less than full-time had a lower score.
Where to from here?
The ability to bounce back after a hard day at work is a marker of our general well-being. Whilst not everyone can switch to working in PEM, the data from Cottey et al. suggest that modifiable factors improve one’s ability to recover. Interestingly, none are within the junior doctors’ locus of control – access to study, annual leave, and better rostering. They are fully in the hands of others. This lack of autonomy can also lead to a lack of motivation.
The term, Need for Recovery, suggests that the onus is on the individual as if they were an elite athlete resting between races. It is not. It is the system that needs to change.
In their limitations, the authors note that this is a snapshot of data taken at a single moment. There is little in the medical sphere against which to benchmark this data. A similar survey could be carried out in a different setting, geography, or time. Are these ratings typical of all hospital areas or just the NHS? Do different craft groups score as highly? Are Need for Recovery scores as high in Australia and New Zealand (Editor’s note: I’ll work on that one), or are they a product of just working in the ED, regardless of the temperature outside? Are these scores a product of the time? The survey was carried out in the Northern Hemisphere summer of 2019. What would those scores be like now, in a COVID-ravaged world?
One more thing
There is one non-modifiable risk factor that we have not discussed – gender.
Male and female respondents were equal, but men had a much lower NFR than women – 65.6% (CI 60.8 – 66.5) vs 72.7% (70.5 – 75). But women are much more likely to be the primary caregiver, you might argue. And you would be right, but even if this is taken into account, women seem to have a higher need for recovery. This seems to hold in other studies examining gender differences and fatigue.
References:
Burns KE, Kho ME. How to assess a survey report: a guide for readers and peer reviewers. Cmaj. 2015 Apr 7;187(6):E198-205.
Eysenbach, G., 2004. Improving the quality of Web surveys: the Checklist for Reporting Results of Internet E-Surveys (CHERRIES). Journal of medical Internet research, 6(3), p.e34.
Graham B, Cottey L, Smith JE, et al Measuring ‘Need for Recovery’ as an indicator of staff well-being in the emergency department: a survey study. Emerg Med J Published Online First: 2020. doi: 10.1136/emermed-2019-208797
Van Veldhoven, M.J.P.M. and Broersen, S., 2003. Measurement quality and validity of the “need for recovery scale”. Occupational and environmental medicine, 60(suppl 1), pp.i3-i9.
Winwood, P.C., Winefield, A.H. and Lushington, K., 2006. Work‐related fatigue and recovery: the contribution of age, domestic responsibilities and shiftwork. Journal of Advanced Nursing, 56(4), pp.438-449.
Wood, M., 2005. Bootstrapped confidence intervals as an approach to statistical inference. Organizational Research Methods, 8(4), pp.454-470.