Performance of the Hypotension Prediction Index May Be Overestimated Due to Selection Bias

Abstract

The Hypotension Prediction Index is a proprietary prediction model incorporated into a commercially available intraoperative hemodynamic monitoring system. The Hypotension Prediction Index uses multiple features of the arterial blood pressure waveform to predict hypotension. The index publication introducing the Hypotension Prediction Index describes the selection of training and validation data. Although precise details of the Hypotension Prediction Index algorithm are proprietary, the authors describe a selection process whereby a mean arterial pressure (MAP) less than 75 mmHg will always predict hypotension. We hypothesize that the data selection process introduced a systematic bias that resulted in an overestimation of the current MAP value’s ability to predict future hypotension. Since current MAP is a predictive variable contributing to Hypotension Prediction Index, this exaggerated predictive performance likely also applies to the corresponding Hypotension Prediction Index value. Other existing validation studies appear similarly problematic, suggesting that additional validation work and, potentially, updates to the Hypotension Prediction Index model may be necessary.

Hypotension prediction index guided versus conventional goal directed therapy to reduce intraoperative hypotension during thoracic surgery: a randomized trial

“…A recent study by Enevoldsen et al [ 20 ] discussed a problematic selection bias in the development of the HPI algorithm which might overestimate the probability of occurrence of a hypotensive event after HPI levels have reached a certain threshold (provided no therapeutic intervention was performed after the alarm). While these concerns are valid from a statistical point of view, the goal of HPI guided hemodynamic monitoring is to reduce the incidence and duration of hypotensive events, which was achieved both in our study and study by Wijnberge et al…”

Prediction and Prevention of Intraoperative Hypotension with the Hypotension Prediction Index: A Narrative Review

“…The HPI is a real-time and continuous predictor of IOH that has the potential to decrease the incidence and cumulative duration of IOH. Concerns raised elsewhere [ 41 ] regarding a selection bias in the development of the model need further investigation in order to weigh whether this issue can severely flaw the algorithm performance in a clinically relevant manner. Despite this, the majority of the current literature supports its use, and the initial results are promising.…”

“…In their cohort, the investigators excluded hypotensive events caused by the clinical intervention (e.g., vascular clamping, patient positioning) and arbitrarily used a binary definition of hypotension (hypotensive events defined as MAP < 65 mmHg and nonhypotensive events defined as MAP > 75 mmHg) leaving a gray zone in between. A recent analysis of the HPI algorithm underlined the problems generated by this selection bias [ 41 ]. In fact, Enevoldsen and Vistisen [ 41 ] analyzed data from the original [ 13 ] as well as subsequent validation studies and found that the AUC of all studies was skewed towards high specificity.…”

“…A recent analysis of the HPI algorithm underlined the problems generated by this selection bias [ 41 ]. In fact, Enevoldsen and Vistisen [ 41 ] analyzed data from the original [ 13 ] as well as subsequent validation studies and found that the AUC of all studies was skewed towards high specificity. This, as explained by the authors and an accompanying editorial [ 34 ], could potentially lead to an overestimation of the risk of hypotension with resulting overtreatment.…”

Selection Bias in the Hypotension Prediction Index: Reply

“…We have read the interesting comment by Giustiniano and Nisi 1 about our article. 2 The authors highlight that the diastolic arterial blood pressure can decrease significantly when the Hypotension Prediction Index (hereafter the Index ) reaches approximately 50 and argue for the possible need for treatment when diastolic arterial blood pressure is very low. We appreciate the comment and would like to highlight additional aspects that probably need to be considered in this context.…”