Aging a Decade in a Day: Age Cutoff Bias in Adjuvant Therapy Allocation for Early-Stage Breast Cancer

Abstract

Purpose: Physicians may expedite interpretation of data presented as a continuous variable by binning the data into "high" and "low" subgroups (cutoff heuristic). Use of this cognitive shortcut with age may lead to fewer nuanced or inappropriate decisions. We hypothesized an age cutoff heuristic may lead to non-evidence-based adjuvant treatment allocation among patients with early-stage breast cancer.

Methods and materials: Two cohorts with strong indications for adjuvant treatment regardless of age that underwent lumpectomy for early-stage breast cancer between 2004 and 2017 were identified in the National Cancer Database. Cohort 1 had higher-risk features (estrogen receptor negative, endocrine therapy not planned, final margins positive, or size >3 cm; n = 160,990) and was appropriate for radiation. Cohort 2 had hormone receptor positivity with tumors >5 mm (n = 394,946) and was appropriate for endocrine therapy. Multivariable logistic regressions with odds ratios (ORs) and 99.8% confidence intervals (CIs) were performed to determine whether any single year-over-year age difference was independently associated with a difference in likelihood of adjuvant therapy recommendation.

Results: In cohort 1, radiation recommendation decreased sharply at age 70, ranging from 90% to 92% between the ages of 50 and 69 years to 81% for those aged 70 years. Multivariable logistic regressions showed year-over-year age difference was an independent predictor for adjuvant radiation recommendation at only age 70 versus 69 (OR, 0.47; CI, 0.39-0.57; P < .001). For cohort 2, endocrine therapy recommendation showed a small decline at age 70, and year-over-year age difference was a predictor of endocrine therapy recommendation at only age 70 versus 69 (OR, 0.86; CI, 0.74-0.99; P = .001).

Conclusions: We observed a unique decline in appropriate adjuvant therapy recommendation between ages 69 and 70. This suggests use of an age cutoff heuristic to process patient age in this population as a categorical, binary variable. This is a previously undescribed phenomenon in early-stage breast cancer.