Labelling 'unexpected' deaths on a neurology service.

Area of Science:

• Neurology
• Clinical Risk Prediction
• Medical Outcomes

Background:

• Accurate prediction of patient mortality is crucial for clinical decision-making and resource allocation.
• Dichotomizing patient risk based on a priori perceptions can aid in stratifying patient populations.
• Neurology services manage a diverse patient population with varying prognoses.

Purpose of the Study:

• To develop and validate a list of descriptive phrases for classifying neurology admissions by perceived 30-day mortality risk.
• To assess the predictive accuracy of this risk classification for 30-day mortality.
• To determine if the classification supports distinguishing between expected and unexpected deaths.

Main Methods:

• Development of a list of descriptive phrases to categorize admissions as 'higher than average' or 'lower than average' risk of 30-day death.
• Application of the list to 500 consecutive neurology admissions.
• Calculation of relative risk of death for the 'higher-risk' group.

Main Results:

• The 'higher-risk' classification identified 20% of admissions.
• These admissions carried a 17-fold relative risk of 30-day mortality compared to the 'lower-risk' group.
• The list demonstrated validity as a predictor of 30-day outcomes.

Conclusions:

• The developed risk stratification tool is a valid and informative predictor of 30-day mortality in neurology admissions.
• This classification system can assist in differentiating between expected and unexpected patient deaths.
• The findings support the utility of subjective risk perception in objective outcome prediction.