Surgery at the End of Life—Aggressive But Necessary?

Elsewhere in JAMA Network Open, Greenberg et al¹ leverage California Patient Discharge Data to characterize patients who enrolled in hospice after gastrointestinal tract surgery. The authors identified 2688 patients discharged to hospice after surgery for a digestive disorder in California between January 1, 2015, and December 31, 2019. Of these, 2389 patients (88.9%) presented urgently or emergently for surgery or interventional procedures, most frequently owing to cancer or a bowel obstruction. In the 3 years preceding this surgical hospitalization, this cohort had a mean of 2.21 hospital admissions per patient, with 3594 of 5953 admissions (60.4%) within 1 year of hospice enrollment. Younger patients, those who belonged to racial and ethnic minority groups, those with limited English proficiency, and/or those insured through Medicaid were more likely to be readmitted after hospice enrollment. Using the Distressed Community Index as a measure of economic status, those in the midtier, at-risk, or distressed group had significantly higher rates of readmission.

This work highlights the challenge in using large data sets to capture the nuances of care at the end of life. As the authors emphasize, surgical care specifically is associated with worse quality of life, common misconceptions about curative intent, and poor communication resulting in goal-discordant care. Although certainly true, this study may not be fully equipped to support these assertions. Furthermore, when considering the trajectory of cancer, their findings speak in favor of surgery’s role at the end of life. In end-of-life care literature, overtreatment is defined as a medical intervention that is extremely unlikely to help a patient, is misaligned with a patient’s wishes, or both.² With respect to the latter definition, large data sets do not capture patient wishes, and conclusions about the merit of surgery are not possible. Large data sets are potentially capable of assessing the ability of an intervention to help, but this requires first asking, “What are we helping?” Overtreatment would be evident if the goal of surgery was to prolong life in a patient with terminal cancer. However, interventions performed for symptomatic relief in a patient facing life-threatening cancer are in accordance with the priorities of palliative care.³ The authors found that bowel resection, gastric bypass, and fecal diversion were the most frequently performed operations, implying that surgery was provided with palliative intent. Without these interventions, patients would likely require hospitalization and experience a painful death.

The authors report that 88.9% of their cohort underwent urgent or emergent operations, supporting the critical need for advance care discussions with attention to the likely natural course of patients’ disease (ie, period of gradual progression followed by becoming increasingly symptomatic and finally, a short period of evident decline) before they present for surgery.⁴ However, the data source does not capture whether these discussions might already be happening, which is arguably feasible given the numerous hospitalizations before the surgical intervention. Furthermore, although there may be a typical cancer trajectory, cancer can also be insidious and unpredictable. As such, it may not be truly accurate to conclude that surgery near death is a failure in the care pathway leading to untimely or aggressive care.

For decades, experts have voiced concerns about the methodological difficulties in using big data for end-of-life care research. The 2003 report by Lunney et al,⁵ Describing Death in America: What We Need to Know, documents these persistent issues. The foremost issue mentioned is the need to obtain information from varying perspectives. These include the patients, their loved ones, and their clinicians. Administrative and clinical registry data have no insights into the thoughts of these actors or the intimate discussions that occur. Without these perspectives, and consequently the inability to define patient and family values, such data are inherently limited in answering value-based questions such as, “Is this goal-concordant care?” or “Will this be beneficial to the patient?” This report also cautions about the reliability and validity of data gathered for purposes other than end-of-life care research. Although it was shown that this cohort had a mean of 2.21 hospitalizations per patient before surgery leading to hospice, there is limited understanding of these hospitalizations, such as whether palliative care was involved and/or goals-of-care discussions had been conducted. Such knowledge is critical to understanding whether high quality end-of-life care was provided.

The analysis by Greenberg et al¹ reveals multiple opportunities to elicit patient values and discuss future care among patients with gastrointestinal tract cancer undergoing surgical intervention and suggests that such conversations should begin at the time of diagnosis and continue even after hospice discharge. Unfortunately, for this cohort, it is unknown whether these conversations occurred and whether patient values were respected. Often, certain data sources lack granular details to draw value-laden conclusions about the appropriateness of care. Going forward, merging data to include information about when and how palliative care was involved might facilitate circumstantial conclusions about appropriateness of care. Inevitably, this question can only be fully answered by assessing the conversations among patients, their loved ones, and their clinicians and understanding the complex circumstances surrounding death due gastrointestinal tract cancer.

Dying of cancer is a dynamic process characterized by uncertainty and the need for ongoing conversations to elicit patient values and needs. Such conversations are critical to guide goal-directed treatments, with options that may change with time. Herein, the authors characterize patients who transitioned to hospice after gastrointestinal tract surgery and find that end-of-life care is burdensome and suboptimal. Further, the authors highlight those at highest risk of readmission and offer potential interventions to mitigate disparities at the end of life at multiple levels. From the clinician standpoint, prognostication is difficult, and predicting exactly when a patient will have an obstruction is unachievable. Therefore, proper use of surgery among hospice patients with cancer may remain value laden. However, we support the authors’ opinions that advance care planning and discussions about care goals are important opportunities to improve end-of-life care for patients undergoing surgery. To operate preemptively and/or without aligning preferences potentially subjects a vulnerable population to unnecessary surgery—an aggressive yet avoidable approach.

Published: July 6, 2022. doi:10.1001/jamanetworkopen.2022.20382

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2022 Millis MA et al. JAMA Network Open.

Corresponding Author: Pasithorn A. Suwanabol, MD, MS, 2922 Taubman Center, 1500 E. Medical Center Dr, Ann Arbor, MI 48109 (pasuwan@med.umich.edu).

Conflict of Interest Disclosures: None reported.