Predictors and outcomes of withholding and withdrawal of life-sustaining treatments in intensive care units in Singapore: a multicentre observational study

Clare Fong, Wern Lunn Kueh, Sennen Jin Wen Lew, Benjamin Ho, Yu-Lin Wong, Yie Hui Lau, Yew Woon Chia, Hui Ling Tan, Ying Hao Christopher Seet, Wen Ting Siow, Graeme MacLaren, Rohit Agrawal, Tian Jin Lim, Shir Lynn Lim, Toon Wei Lim, Vui Kian Ho, Chai Rick Soh, Duu Wen Sewa, Chian Min Loo, Faheem Ahmed Khan, Chee Keat Tan, Roshni Sadashiv Gokhale, Chuin Siau, Noelle Louise Lim, Chik-Foo Yim, Jonathen Venkatachalam, Kumaresh Venkatesan, N Chia, Mei Fong Liew, Guihong Li, Su Mon Myat, Zena Zena, Shuling Zhuo, Ling Ling Yueh, Caroline Shu Fang Tan, Jing Ma, Siew Lian Yeo, Yiong Huak Chan, Jason Phua, for Sicm-nicer groupJournal of Intensive Care 2024 · DOI

Abstract

Differences in the independent predictors of withholding and withdrawal of life-sustaining treatments (LST) exist. Even after accounting for baseline characteristics, both withholding and withdrawal of LST independently predict hospital mortality. Later mortality in patients who had LST withdrawn compared to withholding suggests that the decision to withdraw may be at the point when medical futility is recognised.

Introduction

In critical care, limiting life-sustaining treatment (LST) generally takes two forms: withholding (deciding not to start a treatment, such as mechanical ventilation) or withdrawal (stopping a treatment already in progress). While international guidelines often state there is no ethical difference between the two, this "Western" consensus is not universally accepted. In many Asian cultures, withdrawing treatment is often viewed as a more difficult "active" step compared to the "passive" omission of withholding.

This study, published in the Journal of Intensive Care, explores these complex end-of-life decisions in Singapore’s intensive care units (ICUs). Singapore, with its multi-ethnic and multi-religious population, provides a unique environment to study how clinical, cultural, and socioeconomic factors influence these decisions in a high-income Asian society.

Methods

The researchers conducted a multicentre observational study using data from the National Intensive Care Unit Repository (NICUR), covering 21 adult ICUs across nine public hospitals in Singapore. The study analyzed 8,907 unique patients (9,723 admissions) between 2014 and 2019.

Researchers utilized Multivariable Generalised Mixed Models (GLMM) to account for institutional variations. Clinical severity was measured using the APACHE II score, and socioeconomic status was proxied via hospital paying classes (government subsidy levels). Baseline characteristics of the study population are summarized below.

Table 1: Baseline characteristics of patients categorized by LST limitation status.

Key Insights and Predictors

The "Action vs. Omission" Bias

Withholding was found to be twice as common as withdrawal (13.0% vs 6.2%). This supports the hypothesis that clinicians and families find it easier to forgo starting a treatment than to stop one already in place, reflecting a significant psychological and cultural barrier to withdrawal.

Cultural and Religious Divergence

Demographic analysis revealed that being of Chinese race or identifying as Hindu or Muslim independently predicted the withholding of treatment, but not its withdrawal. This suggests that cultural and religious frameworks may provide clearer guidance on what treatments should not be initiated, rather than providing a mandate for when to stop ongoing care.

Socioeconomic Signals

Interestingly, a lower hospital paying class (indicating higher government subsidy) was a unique predictor for withdrawal. The authors suggest that as costs accumulate during a long, potentially futile ICU stay, financial considerations might eventually influence the decision to move toward withdrawal in a way they do not for initial withholding.

Survival and Outcomes

Both withholding and withdrawal were strong independent predictors of mortality. Even after adjusting for illness severity, withholding LST increased the odds of death by 13.8 times, while withdrawal increased it by 38.3 times.

Figure 5: Kaplan Meier Survival Curve showing the distinct survival trajectories for patients with withheld vs. withdrawn LST.

Patients who had treatment withdrawn had longer ICU stays and higher overall mortality (91.8%) than those where treatment was withheld (82.1%). The survival curves indicate that withdrawal is often the "final" decision made once medical futility becomes undeniable after a trial of therapy, whereas withholding often occurs earlier in the clinical course.

Conclusion and Future Directions

The study highlights that withholding and withdrawal are not practiced interchangeably in Singapore. While clinical severity (APACHE II) and functional status are common predictors for both, cultural and socioeconomic factors uniquely influence the decision to withhold versus withdraw.

Future research is needed to move beyond mortality metrics and explore "patient-centric outcomes," such as the quality of death and the long-term psychological impact on surviving family members in the Asian context.

Summary updated

Abstract

Differences in the independent predictors of withholding and withdrawal of LST exist. Even after accounting for baseline characteristics, both withholding and withdrawal of LST independently predict hospital mortality. Later mortality in patients who had LST withdrawn compared to withholding suggests that the decision to withdraw may be at the point when medical futility is recognised.
BibTeX
@article{fong2024predictors,
  author = {Clare Fong and Wern Lunn Kueh and Sennen Jin Wen Lew and Benjamin Ho and Yu-Lin Wong and Yie Hui Lau and Yew Woon Chia and Hui Ling Tan and Ying Hao Christopher Seet and Wen Ting Siow and Graeme MacLaren and Rohit Agrawal and Tian Jin Lim and Shir Lynn Lim and Toon Wei Lim and Vui Kian Ho and Chai Rick Soh and Duu Wen Sewa and Chian Min Loo and Faheem Ahmed Khan and Chee Keat Tan and Roshni Sadashiv Gokhale and Chuin Siau and Noelle Louise Lim and Chik-Foo Yim and Jonathen Venkatachalam and Kumaresh Venkatesan and N Chia and Mei Fong Liew and Guihong Li and Li Li and Su Mon Myat and Zena Zena and Shuling Zhuo and Ling Ling Yueh and Caroline Shu Fang Tan and Jing Ma and Siew Lian Yeo and Yiong Huak Chan and Jason Phua and for SICM-NICER group},
  title = {{Predictors and outcomes of withholding and withdrawal of life-sustaining treatments in intensive care units in Singapore: a multicentre observational study}},
  journal = {Journal of Intensive Care},
  year = {2024},
  doi = {10.1186/s40560-024-00725-3},
}
Figure