The Uncertain Science of Predicting Death | Oncology | JAMA Network Open | JAMA Network
[Skip to Navigation]
Views 2,548
Citations 0
Invited Commentary
Oncology
April 1, 2020

The Uncertain Science of Predicting Death

Author Affiliations
  • 1Division of Hematology and Oncology, Vanderbilt University School of Medicine, Nashville, Tennessee
  • 2Division of Pediatric Hematology/Oncology, Vanderbilt University School of Medicine, Nashville, Tennessee
JAMA Netw Open. 2020;3(4):e201736. doi:10.1001/jamanetworkopen.2020.1736

With their study, Seow et al1 call attention to the complementary role that patients play in shaping end-of-life (EOL) care and the potential utility of a patient-oriented prognostic tool in influencing their experience. The investigators sought to develop a prognostic tool that is geared toward patients and families, making excellent use of a large and representative data set from Ontario, Canada, that is unique for its longitudinal measurement of patient-reported outcomes and symptom scores that are not available in most other cancer registries. Using the Edmonton Symptom Assessment Scale and Palliative Performance Scale to predict mortality, along with various other clinical factors, the investigators created a novel model for predicting mortality among patients with cancer irrespective of cancer type and laboratory values.

Despite increased attention to quality of EOL care, patients with cancer continue to receive aggressive care near death and experience poor EOL outcomes.2 Although 80% of patients prefer to die at home, most die in hospitals and without a prior goals-of-care conversation,3,4 and nearly 25% of patients with cancer receive chemotherapy within 2 weeks of death.2 Outcomes are shaped by multiple factors, including poor physician prognostic awareness and optimism bias. With advances in therapeutics, patients and practitioners may anchor on the explanation of advanced cancer as treatable but not curable but neglect the harder discussions of how declining functional status and the biological limits of late-line therapies will inevitably limit the value of additional treatments. Therefore, physicians have been the focus of recent attempts to improve prognostication and palliative care implementation through electronic health record automation and machine learning methods.3,4

Efforts to develop and standardize patient-facing prognostic tools to help patients understand their own mortality and disease trajectory, as well as tools to empower patients and families to advocate for care consistent with their goals and values, are also necessary. Patients substantially overestimate their own likely survival, and discordance between physician and patient understanding is substantial, including on fundamental issues, such as whether cure is likely.5 Furthermore, heterogeneity of physician practice, as opposed to relevant clinical factors, often determines the timing and likelihood of EOL discussions.6

A prognosis framework based on functional decline and symptoms is intuitive and conceptually useful in orienting patients and practitioners toward linking the day-to-day changes that patients and families experience to their larger prognostic significance. Of note, the study’s finding that any change in appetite is predictive of mortality is an important one, highlighting the critical need for further research into cancer-related cachexia. Although some of the other metrics used in the model are circular, such as receiving EOL home care, this not an inherent weakness in a patient-oriented model. As discussed, many patients are unaware of even basic prognostic significance. The significance of EOL home care noted in the article may also reflect a difference in the Canadian health system, in which there is no distinct hospice service with prognosis and therapy requirements for enrollment.

Although the analysis is comprehensive, the model’s widespread implementation and dissemination as a patient-facing tool will require further study, particularly in implementation for patients with varying levels of health literacy. Although the numeric probability of survival in days may be of interest to some patients, a discussion with a practitioner of what this means—in hours to days, days to weeks, weeks to months, or months to years—will be needed to assist patients in appropriate legacy planning, EOL preparedness, and medical decision-making. Fundamentally, the model provides information, with the eventual goal to empower patients to speak with practitioners about palliative care. The translation of knowledge to behavior and of information to empowerment and activation requires further intervention for successful implementation.

Like most behavioral change, improving EOL outcomes has proven to be a complex problem that requires multilevel interventions with careful study of barriers and facilitators for stakeholders. Patient prognostic awareness is one of many necessary components toward achieving EOL outcomes that align with patient goals and values. As physicians, we must prioritize discussing prognosis and medical decision-making within a framework of overarching, values-driven patient goals and realistic treatment expectations. Shared decision-making is critical in the evolving determination of the value of medical therapy in the context of expected toxic effects, quality of life, and prognosis. Of importance, patients’ perspectives change over time: some bargain down to accommodate acceptance of previously intolerable debility, whereas others no longer bear increasing burdens of disease as they experience an increasing realization of their own mortality.

Although cancer is associated with high mortality, death remains remarkably unpredictable. As we strive to improve tools to predict mortality, we must also strive to improve our management of the uncertainty of disease. When access to palliative and quality EOL care are dependent on a predictable course of illness and certain death, we miss the opportunity to help patients with an unpredictable disease trajectory. As treatments continue to improve, the EOL trajectories of many cancers are changing to mimic the longer, less predictable EOL courses traditionally seen in nonmalignant advanced chronic conditions.7 In the changing and complex landscape of oncology care, including novel therapeutics, targeted and personalized medicine, evolving care delivery and payment models, and quality metrics, early and accurate predictors of mortality and morbidity will become even more important.

Back to top
Article Information

Published: April 1, 2020. doi:10.1001/jamanetworkopen.2020.1736

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 Baum LVM et al. JAMA Network Open.

Corresponding Author: Laura Van Metre Baum, MD, Division of Hematology and Oncology, Vanderbilt University Medical Center, 2220 Pierce Ave, Nashville, TN 37232-6307 (laura.baum@vumc.org).

Conflict of Interest Disclosures: None reported.

Additional Contributions: Laura Cantino, MD, Department of Palliative Medicine, Kaiser Permanente Northern California, Oakland, provided helpful review of this article and was not compensated for her help.

References
1.
Seow  H, Tanuseputro  P, Barbera  L,  et al.  Development and validation of a prognostic survival model with patient-reported outcomes for patients with cancer.  JAMA Netw Open. 2020;3(4):e201768. doi:10.1001/jamanetworkopen.2020.1768Google Scholar
2.
Prigerson  HG, Bao  Y, Shah  MA,  et al.  Chemotherapy use, performance status, and quality of life at the end of life.  JAMA Oncol. 2015;1(6):778-784. doi:10.1001/jamaoncol.2015.2378PubMedGoogle ScholarCrossref
3.
Parikh  RB, Manz  C, Chivers  C,  et al.  Machine learning approaches to predict 6-month mortality among patients with cancer.  JAMA Netw Open. 2019;2(10):e1915997-e1915997. doi:10.1001/jamanetworkopen.2019.15997PubMedGoogle ScholarCrossref
4.
Avati  A, Jung  K, Harman  S, Downing  L, Ng  A, Shah  NH.  Improving palliative care with deep learning.  BMC Med Inform Decis Mak. 2018;18(4)(suppl 4):122. doi:10.1186/s12911-018-0677-8PubMedGoogle ScholarCrossref
5.
Weeks  JC, Catalano  PJ, Cronin  A,  et al.  Patients’ expectations about effects of chemotherapy for advanced cancer.  N Engl J Med. 2012;367(17):1616-1625. doi:10.1056/NEJMoa1204410PubMedGoogle ScholarCrossref
6.
Liu  P-H, Landrum  MB, Weeks  JC,  et al.  Physicians’ propensity to discuss prognosis is associated with patients’ awareness of prognosis for metastatic cancers.  J Palliat Med. 2014;17(6):673-682. doi:10.1089/jpm.2013.0460PubMedGoogle ScholarCrossref
7.
Amblàs-Novellas  J, Murray  SA, Espaulella  J,  et al.  Identifying patients with advanced chronic conditions for a progressive palliative care approach: a cross-sectional study of prognostic indicators related to end-of-life trajectories.  BMJ Open. 2016;6(9):e012340. doi:10.1136/bmjopen-2016-012340PubMedGoogle Scholar
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    ×