Which factors shape decisions to withdraw life support in stroke patients?

The decision to withdraw life-sustaining therapy (WLST) among acute hospitalized stroke patients is determined by several factors, including age, race, level of consciousness, state region, insurance status, ambulation status at baseline, and stroke centre type, suggests a study presented at ISC 2023.

“The decision to withhold or WLST is common after acute stroke,” according to the researchers, who then sought to investigate factors associated with WLST in hospitalized acute stroke patients.

Patients with acute ischaemic stroke (AIS), intracerebral haemorrhage (ICH), and subarachnoid haemorrhage (SAH) across 152 Florida hospitals that participated in the prospective Florida Stroke Registry from 2008 to 2021 were recruited in this study.

The research team carried out importance plots to generate the predictive factors associated with WLST. They also produced area under the receiver operating characteristic curves (AUC) to generate logistic regression (LR) and random forest (RF) models. A 75/15/15 ratio was used for training/testing/validation.

Of the eligible patients, 309,393 had AIS, 47,485 had ICH, and 16,694 had SAH. During hospitalization, 9 percent, 28 percent, and 19 percent subsequently had WLST, respectively. [ISC 2023, abstract 14]

Patients who had their life support withdrawn were more likely to be older (77 vs 69 years), women (57 percent vs 49 percent), White (76 percent vs 67 percent), have greater stroke severity at presentation (National Institutes of Health Stroke Scale ≥5; 29 percent vs 19 percent), be treated in comprehensive stroke centres (52 percent vs 44 percent), have Medicare insurance (53 percent vs 44 percent), undergo surgical treatments (1.2 percent vs 0.3 percent), have impaired levels of consciousness (38 percent vs 12 percent), and less likely to be uninsured (8 percent vs 13 percent).

Factors significantly predictive of the decision to WLST among AIS patients were as follows: age, stroke severity, state region, insurance status, stroke centre type, race, and level of consciousness (RF AUC, 0.93; LR AUC, 0.85).

In ICH the predictors were age, impaired level of consciousness, state region, race, insurance status, stroke centre type, and ambulation status at baseline (RF AUC, 0.76; LR AUC, 0.71), while those in SAH included age, impaired level of consciousness, state region, insurance status, race, and stroke centre type (RF AUC, 0.82; LR AUC, 0.72).

In an earlier study involving Japanese physicians, findings showed that these healthcare providers were more likely to hold greater negative attitudes toward the withdrawal of life-support care than its withholding. [BMC Med Ethics 2007;doi:10.1186/1472-6939-8-7]

Japanese physicians also favoured withholding invasive life-sustaining treatments such as the attachment of a respirator over less invasive and long-term treatments such as tube feeding. [BMC Med Ethics 2007;doi:10.1186/1472-6939-8-7]

“Discrepancies were demonstrated between attitudes and actual behaviours,” the authors said. “Physicians may need systematic support for appropriate decision-making for end-of-life care.”

Another study suggested improvements in prognosis estimation and understanding of patient preferences to promote patient-centred care when dealing with life-sustaining therapies. [Rev Neurol (Paris) 2011;167:468-73]