Do the Trajectories of Dyspnea Differ in Prevalence and Intensity by Diagnosis at the End of Life? A Consecutive Cohort Study

Abstract

Context: Breathlessness reportedly worsens as death approaches for many people, but the differences in intensity and time course between underlying causes are not well described. Objectives: To determine differences in the intensity of breathlessness by diagnosis over time as death approaches in a consecutive cohort seen by a specialist palliative care service. Methods: Patients referred to Silver Chain Hospice Care Service over a period of four years (January 2004 to December 2007) had dyspnea evaluated at every clinical encounter until death. A numeric rating scale (NRS) was used to measure the intensity. Patients were categorised into five clusters (lung cancer, secondary cancer to lung, heart failure, end-stage pulmonary disease, and no identifiable cardiorespiratory cause) at three time points (60-53 [T3], 30-23 [T2], and 7-0 [T1] days before death [To]. Group differences were assessed using analysis of variance. Joinpoint regression models defined significant changes in mean breathlessness intensity.Results: For 5,862 patients, data were collected an average of 20 times (median: 13; 116,982 data points) for an average of 86 days (median: 48). Breathlessness was significantly higher at all three time points in people with noncancer diagnoses. Breathlessness increased significantly at days 10 and 3 before death for people with cancer (P<0.001 for both), but remained unchanged, albeit significantly higher for patients with noncancer diagnoses. In the three months leading to death, the prevalence of "no breathlessness" decreased from 50% to 35%, and the proportion of patients with severe breathlessness (>7 out of 10) increased from 10% to 26%. Conclusion: Prevalence of breathlessness increases rapidly at life's end, especially for people with primary lung cancer; the levels of breathlessness became close to those experienced by people with noncancer diagnoses despit symptom control measures.