Effect of continuous deep sedation on survival in patients with advanced cancer (J-Proval): a propensity score-weighted analysis of a prospective cohort study

Summary

Background

Continuous deep sedation (CDS) before death is a form of palliative sedation therapy that has become a focus of strong debate, especially with respect to whether it shortens survival. We aimed to examine whether CDS shortens patient survival using the propensity score-weighting method, and to explore the effect of artificial hydration during CDS on survival.

Methods

This study was a secondary analysis of a large multicentre prospective cohort study that recruited and followed up patients between Sept 3, 2012, and April 30, 2014, from 58 palliative care institutions across Japan, including hospital palliative care settings, inpatient palliative care units, and home-based palliative care services. Adult patients (aged ≥20 years) with advanced cancer who received care through the participating palliative care services were eligible for this secondary analysis. Patients with missing data for outcome variables or who lived for more than 180 days were excluded. We compared survival after enrolment between patients who did and did not receive CDS. We used a propensity score-weighting method to control for patient characteristics, disease status, and symptom burden at enrolment.

Findings

Of 2426 enrolled patients with advanced cancer, we excluded 289 (12%) for living longer than 180 days and 310 (13%) with missing data, leaving an analysis population of 1827 patients. 269 (15%) of 1827 patients received CDS. Unweighted median survival was 27 days (95% CI 22–30) in the CDS group and 26 days (24–27) in the no CDS group (median difference −1 day [95% CI −5 to 4]; HR 0·92 [95% CI 0·81–1·05]; log-rank p=0·20). After propensity-score weighting, these values were 22 days (95% CI 21–24) and 26 days (24–27), respectively (median difference −1 day [95% CI −6 to 4]; HR 1·01 [95% CI 0·87–1·17]; log-rank p=0·91). Age (p_interaction=0·67), sex (p_interaction=0·26), performance status (p_interaction=0·90), and volume of artificial hydration (p_interaction=0·14) did not have an effect modification on the association between sedation and survival, although care setting did have a significant effect modification (p_interaction=0·021).

Interpretation

CDS does not seem to be associated with a measurable shortening of life in patients with advanced cancer cared for by specialised palliative care services, and could be considered a viable option for palliative care in this setting.

Funding

Japanese National Cancer Center Research and Development Fund.

Introduction

Patients nearing death often have distressing symptoms.¹ Despite advances in palliative medicine, some symptoms are refractory to intensive palliative care and palliative sedation therapy is used.2, 3 The concept of palliative sedation therapy was initially introduced in the early 1990s, and academic associations worldwide have since clarified definitions and developed clinical guidelines.4, 5, 6

Continuous deep sedation until death (CDS) is a type of palliative sedation therapy. It is more controversial than other types of palliative sedation therapy with respect to many aspects, including whether it shortens patient survival, especially when provided without artificial hydration therapy, and whether the physician's intent is purely symptom palliation.7, 8 CDS might be potentially life shortening as a result of concomitant withholding of artificial hydration, pharmacological effects of high-dose sedatives on the respiratory and circulation system, or both. For evidence-based discussion, the clarification of whether CDS, not palliative sedation therapy as a whole, shortens patient survival is of great value.

Until now, many empirical studies have addressed the potential effect of palliative sedation therapy on patient survival.9, 10, 11, 12, 13, 14, 15, 16, 17 A systematic review⁹ identified no significant differences in survival between patients who were sedated and those who were not. However, these previous studies had many limitations. First, the definition of sedation varied, and few studies specifically investigated CDS. Second, patients were recruited from one or a few selected institutions, and therefore the generalisability is limited. Third, except for one matched-cohort study,¹¹ known prognostic factors were not adjusted for. Finally, no published research has addressed the effects of artificial hydration on patient survival, despite this topic being a strong focus of debate.18, 19

Research in context

Evidence before this study

We searched PubMed and palliative care journals (Journal of Pain and Symptom Management, Palliative Medicine, Journal of Palliative Medicine, and Supportive Care in Cancer) with the terms (“palliative care”[All Fields] OR “end-of-life”[All Fields] AND (“hypnotics and sedatives”[MeSH Terms] OR “sedation”[All Fields] OR “midazolam”[All Fields])) AND “survival”[All Fields]) for articles published in English between Jan 1, 1990, and Dec 31, 2014. After excluding articles that did not refer to survival effects of palliative sedation therapy, we found one systematic review and 11 relevant articles. These studies had considerable limitations—eg, small sample size, single-centre study, retrospective design, mixed procedure of palliative sedation therapy, and under-adjustment or non-adjustment for potential confounders. Only one study from Italy was multicentre and attempted to control for patient characteristics; however, some important known confounders including disease status, physical function, and symptom burden were not adjusted for. Therefore, our understanding of whether palliative sedation therapy shortens life, especially continuous deep sedation (CDS), might be biased by the shortage of information about the medical condition of the patients involved.

Added value of this study

This study was a large multicentre prospective cohort study done in three palliative care settings (hospital palliative care settings, inpatient palliative care units, and home-based palliative care services). We used the propensity score-weighting method to adjust for baseline patient characteristics, including disease status, recent chemotherapy, physical and mental function, and symptom burden at enrolment. We observed no measurable life-shortening effects of CDS in the total population or any subgroups. Because a randomised controlled trial to examine survival effects of palliative sedation therapy is unfeasible and unethical, our results provide one of the best sources of evidence on this issue.

Implications of all the available evidence

Our findings suggest that CDS in patients with advanced cancer, cared for by palliative care specialists, did not hasten patient death. This finding might alleviate concerns and feelings of guilt that families and clinicians might have when deciding whether to give a patient CDS. CDS can be provided as a last resort for dying patients with intolerable suffering if it is given to patients with advanced disease refractory to medical treatments and a short life expectancy, and the treatment decision is made through multidisciplinary team discussion including palliative care specialists.

The primary aim of this study was thus to examine whether CDS shortens patient survival using data from a large-scale multicentre prospective cohort study after controlling for potential confounders using the propensity score-weighting method, and to explore whether artificial hydration benefits or harms patient survival.