Koepke et al. raise three issues, first regarding the evidence supporting the hypotension prediction index alert threshold for future trials. Selection of a hypotension prediction index alert threshold must balance sensitivity and specificity, with the goal of predicting meaningful hypotension while avoiding false alerts. Preferably, sensitivity and specificity should be high for an optimal alert threshold. However, evaluating hypotension prediction index sensitivity is challenging because clinician responses to alerts can prevent impending hypotension that might otherwise have occurred. The prediction window also changes sensitivity and specificity, with predictions being less accurate over longer event horizons. For example, two minutes before hypotensive events, a hypotension prediction index threshold of 47 can be considered optimal, whereas it is 32 five minutes before events (table 1). Therefore, choosing the alert threshold based only on sensitivity and specificity...

You do not currently have access to this content.