Machine learning determination of motivators of terminal extubation during the transition to end-of-life care in intensive care unit

Abstract

Procedural aspects of compassionate care such as the terminal extubation are understudied. We used machine learning methods to determine factors associated with the decision to extubate the critically ill patient at the end of life, and whether the terminal extubation shortens the dying process. We performed a secondary data analysis of a large, prospective, multicentre, cohort study, death prediction and physiology after removal of therapy (DePPaRT), which collected baseline data as well as ECG, pulse oximeter and arterial waveforms from WLST until 30 min after death. We analysed a priori defined factors associated with the decision to perform terminal extubation in WLST using the random forest method and logistic regression. Cox regression was used to analyse the effect of terminal extubation on time from WLST to death. A total of 616 patients were included into the analysis, out of which 396 (64.3%) were terminally extubated. The study centre, low or no vasopressor support, and good respiratory function were factors significantly associated with the decision to extubate. Unadjusted time to death did not differ between patients with and without extubation (median survival time extubated vs. not extubated: 60 [95% CI: 46; 76] vs. 58 [95% CI: 45; 75] min). In contrast, after adjustment for confounders, time to death of extubated patients was significantly shorter (49 [95% CI: 40; 62] vs. 85 [95% CI: 61; 115] min). The decision to terminally extubate is associated with specific centres and less respiratory and/or vasopressor support. In this context, terminal extubation was associated with a shorter time to death.

Figure 1

Flowchart of the study enrolment. Note: TTD = time to death; WLST = withdrawal of life-sustaining treatments.

Figure 2

Frequency of terminal extubation across centres. Y-axis: study centre ID and country. Note: Group 1 includes all centres in the Czech Republic and one Canadian centre. Group 2 includes all remaining centres in Canada and the centre in the Netherlands.

Figure 3

Visual interpretation of the random forest model for 15 most important features: (a) Accumulated local effects plots describing how features influence the prediction of a RF model (terminal extubation) on average (red horizontal line). Y-axis: % change in probability of extubation. The validity of the curves is limited in areas with few data—see the rug plot on the x-axis. The corresponding partial dependence plots (PDP) curves are shown in Fig. S5. (b) Permutation feature importance measured as the factor by which the model’s classification error (CE) increases when the feature is shuffled in the test data (permuted CE/original CE [4.7%]).

Figure 4

Plot of survival after WLST derived from the Cox regression model. (a) Univariate Cox regression comparing patients with and without terminal extubation. (b) Adjusted Cox regression comparing patients with and without terminal extubation. (c) Adjusted Cox regression comparing patients with different total circulatory drugs doses. (d) Adjusted Cox regression comparing patients with peak inspiratory pressure above and below 30 cmH₂O. Note: Circ_total_ranked = total ranked circulatory drugs dose, PIP = peak inspiratory pressure.