We thank Dr. Pivalizza for his insightful observations on our article.1 We agree that our work does not demonstrate a causal relationship between a history of opioid use and postsurgical readmission, and that our title mistakenly suggests that possibility. Because this study was observational, we expected many covariates to be differentially distributed between opioid abuse or dependence and non-opioid abuse or dependence groups. This relationship was indeed evident from table 2,1 where 29 out of 32 covariates had significant P values. However, we did not choose adjustment covariates based upon P values. We chose covariates a priori because of the possibility that they might confound the association between opioid abuse or dependence or that they were known surrogate markers for the severity of a patient’s postoperative condition (e.g., transfusion).2
As Dr. Pivalizza notes, many comorbidities (e.g., peptic ulcer disease) had no obvious relationship with opioid abuse/dependence and were equally or more strongly associated with readmission. We agree that those comorbidities may themselves predispose to readmission. Even after adjustment for all of these comorbidities, however, the opioid association remained an independent risk factor for readmission. Our data suggest that the relationship between opioid abuse/dependence and readmission is roughly equal in effect to comorbidities such as peptic ulcer or liver disease.
Although the association we found remains significant after our extensive adjustment, we recognize that a retrospective observational trial cannot prove causality. A better title for our paper might be “Opioid Abuse or Dependence Is Associated with Increased Readmission Rates after Major Operating Room Procedures.”
The authors declare no competing interests.