Anxiety, Depression and Post Traumatic Stress Disorder after critical illness: a UK-wide prospective cohort study

Hatch. Critical Care 2018; 22:310. doi.org/10.1186/s13054-018-2223-6

Clinical Question

  • In critically ill adults that survive their acute illness, what is the incidence of depression, anxiety and post-traumatic stress disorder?


  • Intensive care interventions aim to increase survival in those with critical illness, but the psychopathological component has not been studied in great depth
  • Survivors of critical illness are thought to suffer anxiety, depression and post-traumatic stress disorder (PTSD) in up to 46%, 29% and 34% of cases, respectively
  • Studies to date have been small or in specific subsets of critically ill populations. This study aimed to be a large cohort study to describe the pattern of psychopathology after critical illness and assess any association with mortality


  • Prospective, cohort, observational study
  • Multi-centre, single country
  • Three phases of recruitment:
    • Phase 1 included postal surveys at 3, 12 and 24 months
    • Phase 2 included postal surveys at 3, 12 and 24 months but was also a study-within-a-study RCT investigating survey burden and response rates
    • Phase 3 included postal surveys at 3 and 12 months, and included an initial visit prior to discharge by a research nurse and a telephone follow-up if they did not respond to the postal survey
  • Since a self-reporting survey was not considered diagnostic of a disease, the ‘caseness’ of each disease was assessed – this is the degree to which accepted diagnostic criteria are applicable to a given patient according to their survey answers
  • Patients were matched to records within the UK’s ICNARC Case Mix Programme database, to obtain diagnoses, severity of illness scores and admission details
  • The NHS Summary Care Record database and individual General Practitioners were consulted to identify patients who had died or would not be able to answer questionnaires
  • Survival status was obtained by matching patients to records held by the Office of National Statistics


  • 39 UK institutions recruited patients with a mixture of university hospitals, university-affiliated hospitals and district general hospitals
    • 26 recruited during Phase 1 and 2
    • 31 recruited during Phase 3
      • Of which 18 had also recruited for Phase 1 and 2
  • Recruitment occurred from Nov 2006 to May 2013
    • Phase 1 – Nov 2006 to May 2008
    • Phase 2 – May 2008 to Oct 2010
    • Phase 3 – May 2012 to May 2013


  • Inclusion: All patients who received at least 24 hours of “Level 3 Care” in an Intensive Care Unit
  • Exclusion: Age under 16 years; not registered with a General Practitioner; taking part in another follow-up study; unable to match record in ICNARC database; withdrawal of consent
  • 21,633 patients were screened; 19,822 were eligible; 13,155 were sent the questionnaire after exclusion of those that had died or were withdrawn after consulting their GP
  • Demographics of those that replied compared to all those sent a questionnaire were very similar (responders vs whole cohort)
    • Median age: 64 vs 63 years
    • APACHE 2 score: 15 vs 15
    • Median ICU Length of Stay: 3 vs 3 days



  • 4,942 patients (38%) completed at least one survey
    • 2,943 patients (22%) actively declined by returning a blank survey form
  • Primary outcome:  Caseness at 3 months was similar to results at 12 months
    • Anxiety: 45.7%
    • Depression: 41.0%
    • PTSD: 22.0%
  • Secondary outcome:
    • Cross-over of caseness: shown in the Venn Diagram below:
      • 55.2% of patients met thresholds for caseness of at least one of the three conditions
      • 35.8% of patients met thresholds for caseness of more than one condition

Venn Diagram describing cross-over of caseness amongst responders (area represents proportion; green = PTSD, orange = Depression, blue = Anxiety)

    • Survival:
      • Caseness for depression was associated with a higher likelihood of death within 2 years of discharge
        • Hazard Ratio: 1.47 (after adjustments for known confounders; 95% CI 1.19–1.80; P < 0.001)
      • This link was not found for anxiety or PTSD

Authors’ Conclusions

  • In those who respond to a postal survey after a critical illness, the incidence of psychopathology is more than 50%, and patients with depression are nearly 50% more likely to die during the first 2 years after discharge from an Intensive Care Unit compared to those without depression


  • Extremely important, patient-centred study question
  • Large, multi-centre study
  • Utilised large national datasets with excellent data quality to extrapolate covariates and outcomes
  • Highly validated tools used to assess caseness of diseases
  • Appropriate methodology, ethical approval, statistical methods and reporting framework


  • Cohort study has inherent biases and is less able to demonstrate causation than a randomised, controlled trial
    • An association was demonstrated between depression and 2-year survival, but this cannot be said to be causal
    • Known and unknown confounding factors cannot always be accounted for – for example, pre-morbid psychopathologies were not known but are likely to have effected the incidence after critical illness
  • The response rate was low, so there may be selection bias
    • Are responders more or less likely to suffer psychopathologies compared to non-responders?
    • What was the reason that 22% chose to actively not complete the survey, and how did they differ from those that did complete the survey?
  • The study-within-a-study RCT demonstrated that asking patients to complete questionnaires about depression, anxiety and PTSD leads them to report their quality of life as lower than those who were not asked to complete these questionnaires
    • A bias that occurs as a result of taking the measurement itself is known as the Hawthorne Effect
  • Three different phases may have led to changing biases over time
    • However, the recruitment criteria and questionnaires did not change over time so this likely to be of negligible effect
  • Incidence of a disease has to be inferred from the caseness acquired from a postal survey tool
    • An assessment by a qualified, skilled clinician may come to a different conclusion than a questionnaire implies
  • This result may not be extrapolatable to populations beyond the United Kingdom

The Bottom Line

  • Although these data have limitations, they form the best evidence available in this important area of patient care after critical illness
  • Clinicians should be aware of the high incidence of anxiety, depression and PTSD when considering patients for admission, life-prolonging invasive therapies, withdrawal of care or follow-up after critical illness
  • Families and patients should be made aware of the psychological burden after critical illness and encouraged to seek help, although further research is required to identify therapies that are effective

External Links


Summary author: Duncan Chambler
Summary date: 10 April 2019
Peer-review editor: David Slessor

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.