In Reply We agree that observational comparison of outcomes across populations is dependent on the distribution of all attributes with the potential to affect the outcome of interest. For attributes that are captured by the database (eg, age, comorbidity, stage), a number of strategies were used to create balance (propensity matching) or adjust for imbalances (Cox proportional hazards). For attributes that are not captured by the database (eg, performance status [PS], smoking status), one must first hypothesize as to whether an imbalance is likely. First, we identified populations by a common event that tends to create balance in unmeasured attributes. All patients had undergone and survived lung cancer surgery. This would suggest that all patients had a reasonable PS (in the study by Kawaguchi et al,1 75% of patients who underwent surgery had a PS of 0). Whereas PS and smoking have not been examined in relationship to the timing of adjuvant chemotherapy in lung cancer per se, neither has been associated with delays for other tumor types.2,3 Therefore, we believe that imbalances, should they exist for smoking and PS, would likely be subtle.
Boffa DJ. Delayed Adjuvant Chemotherapy and Survival After Lung Cancer Surgery—Reply. JAMA Oncol. 2018;4(2):262–263. doi:10.1001/jamaoncol.2017.2760
Customize your JAMA Network experience by selecting one or more topics from the list below.
Create a personal account or sign in to: