Self-extubation risk assessment tool: predictive validity in a real-life setting

Nurs Crit Care. 2008 Nov-Dec;13(6):310-4. doi: 10.1111/j.1478-5153.2008.00305.x.


Aim: To evaluate the predictive validity of the self-extubation risk assessment tool (SERAT) in intensive care patients.

Background: Unplanned extubation is an important complication in intensive care units (ICUs). Physicians and nurses working in the ICU would benefit by having access to a tool that could reliably identify patients at risk for unplanned extubation. The SERAT is a risk stratification scheme developed to identify patients at risk for deliberate self-extubation.

Design: A prospective, diagnostic study.

Methods: Over a 3-month period, 256 patients who were admitted in one of five ICUs in four hospitals in Flanders (Belgium) were studied. The Glasgow Coma Scale and the Bloomsbury Sedation Score were completed by nurses at the start of each shift, i.e. three times per day. Independent nurse researchers collected data on planned or unplanned extubation and placed the data in the SERAT classification scheme.

Results: Five self-extubations and three accidental extubations occurred during the 3-month study period, yielding an incidence of 4.47% that corresponded to 0.56 unplanned extubations per 100 ventilation days. Using the highest accuracy model, we determined that the SERAT had a sensitivity of 100%, specificity of 90%, negative predictive value of 100%, positive predictive value of 1.2% and accuracy of 90%.

Conclusions: Although the SERAT can correctly identify patients at risk for deliberate self-extubation, its use also produces a high number of false-positive identifications. Further research is necessary to evaluate how the false-positive rate can be reduced, and subsequently, the predictive validity of the SERAT can be improved.

Relevance to clinical practice: Because of the high number of false positives, the use of the SERAT in clinical practice to date is not advocated. The positive predictive value has to be improved to avoid the implementation of intensive interventions in patients who are not at risk.

Publication types

  • Validation Study

MeSH terms

  • Accidents / statistics & numerical data
  • Belgium
  • Conscious Sedation / adverse effects
  • Critical Care* / methods
  • Device Removal
  • Equipment Failure
  • Female
  • Glasgow Coma Scale
  • Humans
  • Intubation, Intratracheal / adverse effects
  • Intubation, Intratracheal / nursing*
  • Intubation, Intratracheal / psychology
  • Likelihood Functions
  • Male
  • Nursing Assessment / methods*
  • Nursing Assessment / standards
  • Nursing Evaluation Research
  • Predictive Value of Tests
  • Prospective Studies
  • Risk Assessment / methods*
  • Risk Assessment / standards
  • Risk Factors
  • Sensitivity and Specificity
  • Treatment Refusal