TABLE 2.

Risk for adverse events among adults after the start of an opioid analgesic episode, by antidepressant receipt and controlling for patient characteristics^a

	All overdose and self-harm events (N=836,091)		Opioid overdose (N=836,067)		Nonopioid controlled substances overdose (N=836,055)		Other overdose (N=836,060)		Self-harm (N=836,068)
Characteristic	AOR	95% CI	AOR	95% CI	AOR	95% CI	AOR	95% CI	AOR	95% CI
Antidepressant effect	1.19*	1.03–1.37	1.18*	1.02–1.37	1.18*	1.02–1.37	1.18*	1.02–1.36	1.16	1.00–1.34
Antidepressant effect after 6 weeks	.79*	.65–.97	.78*	.64–.96	.76*	.62–.94	.79*	.65–.97	.82*	.67–1.00

^aThe data came from the deidentified Optum Clinformatics Data Mart database, 2007–2017. Adjusted odds ratios (AORs) were estimated with separate multivariable logistic models. We controlled for year fixed effects (calendar year of episodes), a set of indicators for the number of weeks since the opioid treatment episode began, sex, age, and length of continuous insurance enrollment as of the beginning of the opioid episode. Standard errors were corrected and clustered on patient groups. The number of observations in the competing risk models varied slightly by the type of adverse event of interest because we censored all the observations during and after the week of an event that was not the event of interest.

^*p<0.05.