Longitudinal epidemiology of pain severity and interference among women with metastatic breast cancer Public Deposited

Downloadable Content

Download PDF
Last Modified
  • March 21, 2019
  • Castel, Liana D.
    • Affiliation: Gillings School of Global Public Health, Department of Epidemiology
  • Knowledge is limited about risk factors for cancer pain experienced over the course of disease in specific tumor types. In this study, we assessed pain hazards using data originally collected over 51 weeks in a clinical trial among 1,124 women with metastatic breast cancer; pain was measured by the Brief Pain Inventory (BPI) severity and interference with daily living 0-10 subscales. Under a continuous time assumption, we conducted univariate (per-cutpoint) and multivariate (cutpoints 3, 4, 5, 6, and 7 on the BPI) proportional hazards analyses to estimate effects of baseline characteristics on pain hazards. For the severity scale, compared with Caucasian race, non-Caucasian race was associated with 2.29 times the hazard of reaching severity cutpoint 7 versus 1.38 for cutpoint 3, all other covariates held constant. For the interference scale, compared with active baseline Eastern Cooperative Oncology Group (ECOG) status, restricted baseline ECOG status was associated with 2.97 times the hazard of reaching interference cutpoint 7 versus 2.00 for cutpoint 3. Under a categorical (interval-censored) time assumption, we used piecewise exponential models to estimate associations of baseline and time-dependent characteristics with "survival" rates for not yet reaching a score of 7 or above on each subscale, per 80-day interval. Estimated survival rates at the first interval were 0.92 for Caucasian women versus 0.80 for non-Caucasian women; for the interference scale, these rates were 0.80 versus 0.70, respectively. In subsequent intervals, rates declined similarly for Caucasian and non-Caucasian women, but for both pain outcomes, the cumulative survival rate for Caucasians in the last interval was still higher than that of non-Caucasians in the first interval. In confirming associations of ECOG performance status (both as a baseline and timedependent covariate) and race with pain hazards over time in metastatic breast cancer, our findings inform individualized prognoses for pain outcomes according to baseline patient attributes. Early intervention and more aggressive pain management strategies can be tailored to these personalized prognoses over the course of treatment, to delay first occurrence of higher pain scores among those at greatest risk. Future research should specifically target potential sources of racial disparities in cancer pain.
Date of publication
Resource type
Rights statement
  • In Copyright
  • Hartmann, Katherine Eubanks
  • Open access

This work has no parents.