CONTEXT: Understanding the longitudinal transitions of performance status among persons with cancer can assist providers in determining the appropriate time to initiate palliative care support. OBJECTIVES: To model longitudinal transitions of performance status in cancer outpatients, to determine the probabilities of improvement and deterioration in performance status over time, and to evaluate the factors associated with rates of transitions. METHODS: This population-based, retrospective, cohort study comprised adult outpatients diagnosed with any type of cancer and assessed for performance status throughout their observation period using the Palliative Performance Scale (PPS; scale 0-100; 0 indicates death). At every PPS assessment, patients were assigned to one of four states: stable state (PPS score 70-100), transitional state (PPS score 40-60), end-of-life state (PPS score 10-30), or dead. A Markov multistate model under the presence of interval censoring was used to examine the rate of state-to-state transitions. RESULTS: There were 11,374 patients representing nearly 71,000 assessments. Patients with lung cancer in the transitional state had a 27.7% chance of being dead at the end of one month vs. 17.5% in patients with breast cancer. The average time spent in the transitional state was 6.6 weeks for patients diagnosed with gastrointestinal cancer vs. 8.8 weeks for patients with breast cancer. The rate at which one moves from the transitional state to death was higher for patients with lung cancer than those with breast cancer. CONCLUSION: We estimated the probability and direction of change in performance status in cancer outpatients. Entry into the transitional state may serve as an indicator for referral for palliative care support. Mean end-of-life sojourn times are too short to allow meaningful integration of palliative care.
CONTEXT: Understanding the longitudinal transitions of performance status among persons with cancer can assist providers in determining the appropriate time to initiate palliative care support. OBJECTIVES: To model longitudinal transitions of performance status in cancer outpatients, to determine the probabilities of improvement and deterioration in performance status over time, and to evaluate the factors associated with rates of transitions. METHODS: This population-based, retrospective, cohort study comprised adult outpatients diagnosed with any type of cancer and assessed for performance status throughout their observation period using the Palliative Performance Scale (PPS; scale 0-100; 0 indicates death). At every PPS assessment, patients were assigned to one of four states: stable state (PPS score 70-100), transitional state (PPS score 40-60), end-of-life state (PPS score 10-30), or dead. A Markov multistate model under the presence of interval censoring was used to examine the rate of state-to-state transitions. RESULTS: There were 11,374 patients representing nearly 71,000 assessments. Patients with lung cancer in the transitional state had a 27.7% chance of being dead at the end of one month vs. 17.5% in patients with breast cancer. The average time spent in the transitional state was 6.6 weeks for patients diagnosed with gastrointestinal cancer vs. 8.8 weeks for patients with breast cancer. The rate at which one moves from the transitional state to death was higher for patients with lung cancer than those with breast cancer. CONCLUSION: We estimated the probability and direction of change in performance status in cancer outpatients. Entry into the transitional state may serve as an indicator for referral for palliative care support. Mean end-of-life sojourn times are too short to allow meaningful integration of palliative care.