The clinical judgement of healthcare professions has greater predictive accuracy than risk models for intensive care unit admission
The clinical judgement of the healthcare professionals treating a patient has a better predictive accuracy for whether an individual should be admitted to an intensive care unit (ICU) compared with the use of a risk stratification model, according to a systematic review by a Dutch team based in Amsterdam.
The outcomes for critically ill patients are often time-sensitive and research suggests how emergency department patients whose transfer to intensive care exceeds 6 hours, not only increases the length of stay in hospital but also mortality. Consequently, several early warning scoring (EWS) systems have been developed to help risk stratify patients within an emergency department (ED) to determine outcomes such as mortality or to predict the need for intensive care admission. But does the development of these models mean that clinical judgement is no longer necessary? The available evidence to date suggests that such risk stratification models may be no better than clinical judgement. For example, one study found that a simple clinical assessment by healthcare staff was superior to a formalised triage system to predict short-term mortality among emergency department patients. Similarly, in a study of the Canadian Syncope Risk Score (CSRS) used in an ED for syncope risk stratification, the authors found that the tool had similar predictive accuracy to clinical judgement. Determining whether risk stratification models perform better than simple clinical judgement was the aim the current study by the Dutch team. They focused on several areas: the need for ICU admission, severe adverse events and finally, clinical deterioration and mortality. The team included studies in which the authors had compared either risk-stratification or an EWS with clinical judgement among adult patients in an acute setting, i.e., either at the ED or pre-hospital assessment. The outcomes of interest were the need for ICU admission, severe adverse events, clinical deterioration and death.
Clinical judgement vs adverse patient outcomes
The literature search identified only 6 relevant studies with 6419 participants of which 4 studies were deemed to be at a high risk of bias. Due to the small number of studies, no meta-analysis was undertaken and descriptive analyses were used instead.
For ICU admission, in one sepsis study, the sensitivity of clinical judgement was 91% (95% CI 0.83 – 0.99) and specificity 71% (95% CI 0.60 – 0.82) which was superior to PIRO and MEDS (all p < 0.001). For the second sepsis study which considered mortality, there was no difference between clinicians and the risk-stratification models.
For the prediction of clinical deterioration, only one study was available and there was no difference between the two different forms of assessment. Finally, for the prediction of severe adverse events (ICU admission, cardiac arrest and death), one study reported on the use of MEWS or the clinical judgement of nurses, and while sensitivities were similar (56.6% vs 61.8%, statistical significance was not reported), nurse’s judgement had a higher specificity (94.1% vs 88.5%, nurses vs MEWS).
The authors concluded that while there were only limited data available, it appeared that clinical judgement was superior to risk stratification models for predicting the need for ICU admission and for prediction of severe adverse events. However, both approaches were similar for prediction of clinical deterioration and mortality.
Veldhuis LI et al. Performance of early warning and risk stratification scores versus clinical judgement in the acute setting: a systematic review Emerg J Med 2022