Article Text

Download PDFPDF

Psychological impact of working in paediatric intensive care. A UK-wide prevalence study
  1. Gareth A L Jones1,
  2. Gillian A Colville2,
  3. Padmanabhan Ramnarayan3,
  4. Kerry Woolfall4,
  5. Yvonne Heward5,
  6. Rachael Morrison5,
  7. Amy Savage6,
  8. James Fraser7,
  9. Michael J Griksaitis8,
  10. David P Inwald9
  1. 1 Faculty of Medicine, Imperial College London, London, UK
  2. 2 Paediatric Psychology Service, St George's University Hospitals NHS Foundation Trust, London, UK
  3. 3 Children's Acute Transport Service, Great Ormond Street Hospital for Children NHS Trust, London, UK
  4. 4 Institute of Psychology, University of Liverpool, Liverpool, UK
  5. 5 Paediatric Intensive Care Unit, Birmingham Women's and Children's NHS Foundation Trust, Birmingham, UK
  6. 6 Child Clinical Psychology, University Hospital Southampton NHS Foundation Trust, Southampton, UK
  7. 7 Paediatric Intensive Care Unit, Bristol Children's Hospital, Bristol, UK
  8. 8 Paediatric Intensive Care Unit, University Hospital Southampton NHS Foundation Trust, Southampton, UK
  9. 9 Paediatric Intensive Care Unit, St Mary's Hospital, Imperial College Healthcare NHS Trust, London, UK
  1. Correspondence to Dr David P Inwald, PICU, Imperial College Healthcare NHS Trust, London W2 1NY, UK; d.inwald{at}


Objective To determine the prevalence of work-related psychological distress in staff working in UK paediatric intensive care units (PICU).

Design Online (Qualtrics) staff questionnaire, conducted April to May 2018.

Setting Staff working in 29 PICUs and 10 PICU transport services were invited to participate.

Participants 1656 staff completed the survey: 1194 nurses, 270 physicians and 192 others. 234 (14%) respondents were male. Median age was 35 (IQR 28–44).

Main outcome measures The Moral Distress Scale-Revised (MDS-R) was used to look at moral distress, the abbreviated Maslach Burnout Inventory to examine the depersonalisation and emotional exhaustion domains of burnout, and the Trauma Screening Questionnaire (TSQ) to assess risk of post-traumatic stress disorder (PTSD).

Results 435/1194 (36%) nurses, 48/270 (18%) physicians and 19/192 (10%) other staff scored above the study threshold for moral distress (≥90 on MDS-R) (χ2 test, p<0.00001). 594/1194 (50%) nurses, 99/270 (37%) physicians and 86/192 (45%) other staff had high burnout scores (χ2 test, p=0.0004). 366/1194 (31%) nurses, 42/270 (16%) physicians and 21/192 (11%) other staff scored at risk for PTSD (χ2 test, p<0.00001). Junior nurses were at highest risk of moral distress and PTSD, and junior doctors of burnout. Larger unit size was associated with higher MDS-R, burnout and TSQ scores.

Conclusions These results suggest that UK PICU staff are experiencing work-related distress. Further studies are needed to understand causation and to develop strategies for prevention and treatment.

  • intensive care
  • paediatric staffing
  • burnout
  • post traumatic stress
  • moral distress

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

What is already known on this topic?

  • Staff working in intensive care are at risk of psychological morbidity, impacting on staff well-being and mental health, staff retention and patient care.

  • The prevalence of burnout, moral distress and post-traumatic stress has not been assessed before in a national UK paediatric intensive care unit (PICU) staff survey.

What this study adds?

  • UK PICU staff are experiencing work-related distress; junior nurses are at the highest risk of moral distress and PTSD symptoms, and junior doctors of burnout.

  • Mean unit occupancy >15 patients is a risk factor for work-related distress.

  • Free-text comments provided by participants suggest potential strategies for treatment and prevention.


Paediatric intensive care is a difficult environment in which to work. Staff need skills to manage both acute critical illness and palliative care at the end of life. Paediatric intensive care unit (PICU) admissions in the UK are rising above levels expected from the growing population.1 More and more patients have life-limiting conditions and multiple PICU admissions prior to death.2 3 Recently, some cases have attracted high-profile scrutiny in the Courts and on social media. As a result of these factors, the emotional costs of working in PICU are recognised as making it one of the most challenging in healthcare.4

In the USA, there has been increasing awareness of staff mental health in intensive care settings, and its impact on patient care. This has led to the Joint Critical Care Societies’ statement on professional burnout, which recognises a need for more staff support and research.5 In the UK, paediatricians represent around 5% of referrals to the NHS Practitioner Health Programme, a confidential mental health and addiction service for doctors, an over-representation compared with other specialties. Neonatologists and intensive care specialists form a significant subset of the paediatricians.6 While strategies to support the workforce have been identified, the evidence base for interventions is scant, and provision is patchy and varies from unit to unit.

Overall prevalence of work-related psychological distress in PICU staff is unknown though a survey in one UK PICU has shown high levels of burnout and post-traumatic stress disorder (PTSD).7 Significant moral distress has been reported in adult intensive care unit (ICU) staff in the UK8; in a single-site North American mixed paediatric/neonatal intensive care unit staff cohort9; and in a UK PICU following a recent high-profile case.10

The aim of this study was to assess the prevalence of moral distress, burnout and post-traumatic stress in staff working in UK PICUs and in PICU transport teams.


Moral distress is felt when there is a discrepancy between a health professional’s moral judgement about best treatment for a patient and the treatment they receive.11 It was assessed in this study by the Moral Distress Scale-Revised (MDS-R).12 This comprises 21 items: 6 about end-of-life care; 5 about staffing and resources; 4 about communication; 4 items on decision-making; and 2 on witnessing unethical behaviour.

Respondents rate each item in terms of frequency and intensity of subjective disturbance. Overall scores range from 0 to 336. The MDS-R also contains a final section on the respondent’s attitude to the idea of leaving their position as a result of moral distress. The scale has been shown to have good content validity and reliability (Cronbach’s alpha 0.84–0.90) across a number of different samples.9 13 There are no specific ‘cut-off’ values but as previous work has shown a score of 90 or above identifies individuals with a significant level of distress,14 we used 90 as the threshold to indicate significant distress on this dimension in this study.

Burnout is a widely recognised but poorly defined manifestation of occupational stress,5 associated with medical error, depression and suicide.15 It is most often assessed using the Maslach Burnout Inventory (MBI).16 In our study we used shortened 9-item version of the original MBI, known as the abbreviated MBI (aMBI),17 which has been found to retain the factor structure of the original instrument.18 Scores on aMBI were prorated using full-scale cut-offs of ≥27 for high emotional exhaustion (EE) and ≥10 for high depersonalisation (DP).16 Following the methodology of a recent landmark epidemiological study on physicians in the USA,19 burnout was defined as a threshold of a top tercile score in either the aMBI-EE subscale (≥9/18) or the aMBI-DP subscale (≥6/18).

PTSD occurs in response to a traumatic event and manifests as hyper-arousal, re-experiencing events in memories or nightmares, avoidance of similar situations and generalised negative feelings or beliefs.20 Risk of PTSD was assessed using the Trauma Screening Questionnaire (TSQ).21 This is used to identify the number of post-traumatic stress symptoms an individual has experienced in the previous 2 weeks, in relation to a particular event (here defined as ‘stressful incident at work’). Scores range from 0 to 10 with a score of ≥6 indicating increased risk of PTSD. The TSQ has been demonstrated to have an overall test efficiency of 90% (sensitivity of 0.85 and specificity of 0.89) in identifying PTSD following assault22; this was the threshold we used.

Eligible staff were those employed by any participating site and working in any capacity in PICUs or transport teams. Twenty-nine UK PICUs, 17 general units, 8 ‘mixed’ (cardiac and general) units, 4 cardiac ICUs and 10 centralised PICU transport teams took part. Staff were invited via email and local study posters to complete anonymous online questionnaires using Qualtrics over a 6-week period, May to June 2018. Information collected included professional role, length of service and unit. There was an opportunity for participants to leave free-text comments. Signposting to psychological support services was provided for staff who sought help after they had completed the survey.

Univariate and multivariate analyses were performed to identify risk factors using Stata V.14 (StataCorp, USA). Risk factors studied were respondents’ years in service, age, gender, grade (consultant, junior doctor, senior nurse, junior nurse, other), primary working environment (cardiac, general or mixed unit (cardiac and general), or transport) and unit size divided into quartiles for average bed occupancy (first quartile: 2–9; second quartile: 10–12; third quartile: 13–15; and fourth quartile: 16–28). In multivariate analysis the senior nurse group included advanced nurse practitioners (ANP) and senior nurses.

The electronic survey was designed so that it was not possible for respondents to move from one section to the next without answering all questions. Surveys which were marked complete but had incomplete data when the study closed (n=2) were excluded from data analysis. Qualitative free-text responses were analysed thematically23 whereby key themes were identified in responses to each free-text question. Informed by the constant comparative approach,24 the aim was to provide insight into respondents’ views and recommendations (eg, topics for future work). NVivo V.10 software was used to assist the organisation and coding of data.



A total of 1656/3775 (44%) staff from 27/29 PICUs and 10/10 PICU transport services completed the survey. Mean unit bed occupancy ranged from 2 to 28 patients.25 Respondents were 1194 nurses, 270 physicians and 192 other members of staff. Among nursing staff, 940 were classified as ‘junior’ (band 6 and below), 254 as ‘senior’ (band 7 and above) of whom 57 were ANPs. Among medical staff, 166 were consultant grade and 104 were in training grades—42 senior trainees (ST7 or above), 60 junior trainees (ST6 or below) and 2 unclassified (no response). The ‘other’ group included 65 physiotherapists, 21 pharmacists, 19 dieticians, 9 HCAs and 75 others (psychologists, speech and language therapists, occupational therapists, play therapists, ambulance technicians, physician associates, medical device technicians, clerical staff and housekeeping staff).

Descriptive analysis

Two hundred and thirty-four (14%) respondents were male. Median age was 35 (IQR 28–44) (see figure 1) and median number of years in service on PICU was 6 years (IQR 2–14). Overall, 502/1656 (30%) reported a significant level of moral distress (≥90 for MDS-R), 779 (47%) scored in high range for burnout (≥6/18 on aMBI-DP subscale or ≥9/18 on aMBI-EE subscale) and 429 (26%) scored at risk of developing PTSD (≥6 on TSQ scale). There was a significant overlap across the various domains, with many staff scoring positively for more than one measure (figure 2).

Figure 1

Histogram showing age distribution of respondents to the survey. Y axis—n (respondents), x axis—age banding.

Figure 2

Overlap between the various measures of psychological morbidity: figure illustrates extent and overlap of high scoring staff for each of the three types of work-related distress surveyed (n=1656). Moral distress defined as Moral Distress Scale-Revised ≥90. Burnout defined as a high score on the abbreviated Maslach Burnout Inventory for emotional exhaustion (subscale ≥9, equivalent to ≥27 on the full MBI) and/or depersonalisation (subscale ≥6, equivalent to ≥10 on the full MBI). Post-traumatic Stress defined as Trauma Screening Questionnaire ≥6. MBI, Maslach Burnout Inventory.

Overall, 645 respondents (39%) had considered leaving PICU at some point, and 370 (22%) were considering leaving when the survey was taken. Three hundred and ninety-four (24%) reported that they had previously sought support for psychological issues, often using more than one source of support; 209 from a friend or family member, 191 from their general practitioner, 134 from their employer occupational health service, 163 from a senior work colleague, 93 from a work colleague of similar seniority and 70 from another professional (eg, counsellor, staff support worker, clinical psychologist).

Univariate comparisons by professional group

Nurses had a higher mean MDS-R, aMBI-EE and TSQ score than physicians and others but physicians reported the highest aMBI-DP mean scores (table 1). MDS-R score ≥90 was reported by 435/1194 (36%) nurses, 48/270 (18%) physicians and 19/192 others (10%) (χ2 test, p<0.00001). A high burnout score was reported by 594/1194 (50%) nurses, 99/270 (37%) physicians and 86/192 (45%) others (χ2 test, p=0.0004). Those scoring at risk for PTSD were 366/1194 (31%) nurses, 42/270 (16%) physicians and 21/192 (11%) others (χ2 test, p<0.00001).

Table 1

Univariate comparison by staff group (n=1656)

Regression analyses

All staff groups were at higher risk of burnout when compared with consultants, with junior doctors at highest risk. Nurses, whether junior or senior, were at higher risk of PTSD. Junior nurses were at higher risk of moral distress. Male gender was associated with lower risk of moral distress. Larger unit size was associated with high MDS-R, burnout and TSQ scores, although unit type was not. Findings in relation to age and years in service were seemingly at odds, with increasing age being associated with a lower risk of moral distress and burnout but numbers of years in service in PICU associated with both (table 2).

Table 2

Logistic regression showing factors related to staff distress (n=1656)

Logistic regression analyses showed that ‘considering leaving now’ was associated with being a junior nurse (compared with consultants, OR 2.5, 95% CI 1.4 to 4.6, p=0.002) and working in a larger unit (fourth quartile for unit size compared with first quartile, OR 1.7, 95% CI 1.1 to 2.7, p=0.014). The only significant risk factor for ‘sought help’ was years in service (for every year in service, OR 1.03, 95% CI 1.00 to 1.06, p=0.03); data not shown.

Qualitative analysis of free-text comments

At the end of the questionnaire participants’ comments were sought on content of the questionnaire and topics to explore in future work. Just over a third 588/1656 (36%) of participants provided a response; of these 98/588 (17%) suggested topics for future research. Some comments suggested that staff found completing the questionnaire challenging as it raised issues often ‘considered “taboo” subjects in my work environment’ (P305).

Thirty-eight topics were identified (table 3). Thirty participants thought support for PICU staff is insufficient and suggested tailored psychological support. Many highlighted insufficient staffing levels and capacity. Work patterns, poor relationships with colleagues and low pay were also described as having a negative impact. Staff suggested that future work should explore the impact of challenging relationships with families, withdrawal of treatment and child death on staff psychological well-being, and the exacerbation of staff distress by negative media coverage. Five participants commented on moral distress caused by providing ‘futile care’ (P174), which may not be ‘what is best for the patient’ (P614), and difficulty caused by ‘unrealistic expectations of society about what we can do for some children’ (P367).

Table 3

Qualitative analysis of free-text comments. Topics identified as important to explore in future research on the psychological impact of working in paediatric critical care (n=98)

In terms of possible solutions, seven staff suggested education and training, while three stated that leaving PICU, or reducing clinical work, had been effective in reducing work-related stress. In contrast, a few participants provided positive comments about the value of working in teams in which they can ‘communicate openly’ (P18), ‘express distress and …readily access help in debriefing after horrible events’ (P17).


Those working in critical care recognise that stress impacts on both personal well-being and patient care; a recent research prioritisation exercise by the UK Paediatric Intensive Care Society identified staff mental health and stress as the most important topic in need of research (LN Tume, personal communication. Oral Presentation at PICS, Bristol UK, 2018). Our results, from the largest survey of PICU staff to date, confirm this, showing significant rates of moral distress, burnout and risk of post-traumatic stress. The findings were broadly consistent with the literature in that female gender, nursing background and inexperience were associated with increased risk.8 9 13 14 26

There was no clear association with type of unit, although staff working on the largest units were at increased risk of high levels of moral distress, EE and PTSD, with junior nurses in particular considering leaving their posts. The findings suggest that there may be an optimal size of unit in terms of staff well-being. Of note, ‘transport service’ as a unit type was omitted from the model due to collinearity; likely due to the higher proportion of older and male staff members in transport teams, both associated with a lower risk of psychological morbidity (table 2).

Moral distress

The mean score of 69 for the sample on MDS-R was comparable to that reported in adult intensive care staff in the UK (n=171),8 a mixed group of health professionals in the USA (n=592)13 and paediatric/neonatal intensive care staff in Canada (n=2822),14 but lower than that reported in two single-site PICU studies in Canada (mean=102, n=206)9 and the UK (mean=96, n=50).10 These single-site studies were at large tertiary centres consistent with our finding that staff in larger units reported higher levels of moral distress. The main reason cited for moral distress related to perceived futile treatment at the end of life.8 13 The finding that 22% of staff were considering leaving their job because of moral distress is concerning and is higher than rates previously reported (7%–20%).8 12 13


Overall prevalence of burnout among respondents was 47%. This is higher than that reported in a recent survey of intensive care staff in the UK, which used the same tool (32% for adult ICU staff; 42% for paediatric ICU staff).27 It is also higher than the prevalence found in US physicians from across all specialties (32%–42%) and the US general population (28%).19 A more recent US study in paediatric ICU physicians identified a prevalence of burnout of around 20%,28 though a different tool was used to assess burnout. Though we have used a validated measure, this highlights one of many problems with burnout research—different instruments, different cut-offs29 and even the notion that the very act of measuring burnout might cause symptoms.30

Post-traumatic stress

Twenty-six per cent of all respondents and 31% of nurses scored above a recognised clinical cut-off on the screening tool which assesses risk of PTSD. These rates are at the upper end of those found in other studies of intensive care staff (17%–38.5%)31 and much higher than the general population (7%).32

Implications for intervention development

Recent work has emphasised the importance of interventions for burnout at both individual and organisational level.15 The qualitative comments in this study support this, suggesting individual psychological support, interventions to facilitate communication within teams and to address systemic problems such as low staffing levels, shift patterns and low pay for nurses. For post-traumatic stress there are established treatments available.33 Recent work has also focused on the possibility of ‘post-traumatic growth’34 in PICU staff; the notion that individuals can have positive effects resulting from stressful events.35 Thus, it may be important when designing interventions to also look at factors associated with resilience and well-being.


The strengths of the study include the size of the sample; participation of every PICU in the country; inclusion of allied health professionals and other ancillary staff; examination of three different forms of staff stress; individual qualitative comments; and anonymisation. The study design was also a strength, in that the survey was set up in such a way that it was only possible to move on to a new page if all previous items were complete; this meant that there were almost no missing data, which is unusual for a study of this size.


The results may not be representative of all staff working in PICU; however, the response rate was higher than in other online surveys of this type.36 Respondents were informed that one of the elements the survey was looking at was burnout; whereas ideally respondents to the MBI should be unaware that it is a burnout measure to prevent sensitisation to the ‘general issue of burnout’.16 Short-form self-report questionnaires were used to reduce the burden on participants but cannot be regarded as diagnostic. Despite this, there are a number of studies which demonstrate the validity of abbreviated measures in this field18 19 and the PTSD screening instrument used has been recommended as a robust measure in terms of its association with a gold standard clinical interview in a recent review.37


All staff groups working on PICU, especially nurses, are at risk. Junior nurses are at highest risk of moral distress, and junior doctors of burnout. Mean unit occupancy >15 patients is a risk factor for all types of psychological morbidity measured in this study. Participants in our study identified contributory factors, as well as possible solutions, to psychological distress, in qualitative free-text responses. Further work, focusing on staff who feel fulfilled and happy in their work, may help determine individual and institutional factors associated with resilience and well-being, though understanding these may not necessarily help staff members who have serious psychological problems. Prospective studies are required to determine which interventions, if any, may be helpful to prevent and to treat psychological morbidity in PICU staff. This is vital for staff recruitment and retention and most importantly, for good patient care.



  • Twitter @MJGriksaitis, @dlawni

  • Contributors GALJ contributed to the literature search, study design, data collection, data analysis, data interpretation and writing. GAC contributed to the literature search, data analysis, data interpretation and writing. PR and KW contributed to data analysis, data interpretation and writing. YH, RM and AS contributed to data interpretation and writing. JF and MJG contributed to study design, data interpretation and writing. All authors have seen and approved the final manuscript.

  • Funding The investigation was funded by a grant from the Paediatric Intensive Care Society, UK. GALJ was funded by the NIHR Academic Clinical Fellowship scheme.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Ethics approval The study was registered with the HRA and received research ethics committee approval (IRAS: 218720, HRA REC: 17/HRA/0192).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement Data are available upon reasonable request.