Clinical Prediction Model Calculates Risk Death Advanced Cancer Aptitude Health

Clinical Prediction Model Calculates 1-Year Risk of Death in Advanced Cancer

For patients with advanced cancer, timely and accurate prognosis is needed to guide decisions about palliative care. Historically, clinicians have used clinical variables, laboratory values, or the “surprise” question (“Would I be surprised if this patient died in the next year?”) to make predictions about patients’ remaining life expectancy and identify patients with palliative care needs. In a recent JAMA Network publication, Catherine Owusuaa, MD, and colleagues conducted a multicenter, prospective, prognostic study to develop and validate a clinical prediction model using these 3 tools to calculate the 1-year risk of death among patients with advanced cancer. The study included 867 patients with locally advanced or metastatic cancer from 6 hospitals (inpatient and outpatient clinics) in the Netherlands. The most common cancer types were breast, lung, and gastrointestinal.

The model developed by Dr Owusuaa and colleagues combines the “surprise” question, clinical characteristics, and laboratory values to calculate mortality risk. The “surprise” question was answered primarily by attending medical specialists, but also by nurse practitioners and residents, with no significant differences in accuracy between these groups. Clinical characteristics included age, sex, comorbidity, cancer type, metastases, Eastern Cooperative Oncology Group performance status, food intake, weight loss, pain, dyspnea, and fatigue. Laboratory values included hemoglobin, C-reactive protein, and serum albumin. The analysis confirmed that the extended model, which included the “surprise” question, clinical characteristics, and laboratory values, had better discrimination ability than a simple model (“surprise” question only) or clinical model (“surprise” question and clinical characteristics).

High level
Incorporation of the extended clinical prediction model into electronic medical records can serve as a reminder for clinicians across the care continuum to be aware of patients who are at greater risk of dying within 1 year. It could also be implemented as part of a digital advance care planning program, as well as to support communication with patients and their caregivers.

Ground level
The choice of model to predict survival should be based on available patient information. The results of this study suggest that the clinical prediction model developed by Dr Owusuaa and colleagues can be used by clinicians to identify patients at risk of dying within 1 year who may benefit from palliative care and advance care planning. However, if its use is limited by an ability to obtain laboratory values, the clinical model (“surprise” question and clinical characteristics) may be a good alternative. The model requires external validation but has the potential to help support clinicians in initiating conversations with patients and their caregivers regarding advance care planning, and may aid in tailoring treatment decisions to improve patients’ quality of life in the last period of their lives.