Table 1.

Likelihood of a positive opioid urine analysis for 7- and 28-day buprenorphine/naloxone taper groups across the 4-week treatment period using generalized estimation equations: impact of missing data treatment choice on outcomes

	Listwise deletion (n = 394)		Positive UA imputation (n = 516)		Multiple imputation (n = 516)
Covariates	Odds ratio	(95% CI)	Odds ratio	(95% CI)	Odds ratio	(95% CI)
Trial arm (reference = 7 days)	0.546*	(0.371–0.802)	0.432*	(0.341–0.549)	0.572*	(0.424–0.775)
Age (per year)	0.981*	(0.965–0.997)	0.991 ns	(0.981–1.00)	0.984*	(0.971–0.997)
Sex (reference = male)	1.518*	(1.039–2.219)	1.187 ns	(0.932–1.511)	1.524*	(1.089–2.133)
Race (reference = Caucasian)	0.931 ns	(0.806–1.075)	0.920 ns	(0.841–1.005)	0.964 ns	(0.850–1.093)
Time (per week)	1.202*	(1.035–1.398)	1.268*	(1.160–1.387)	1.183*	(1.184–1.383)
Baseline opioid UA (reference = UA–)	8.886*	(5.912–13.356)	5.562*	(4.020–7.694)	9.873*	(7.194–13.617)
Baseline cocaine UA (reference = UA–)	1.460 ns	(0.975–2.185)	1.358*	(1.053–1.751)	1.191 ns	(0.867–1.639)

The dependent measure was a negative (UA–) or positive (UA+) opioid UA at the end of the taper (0 = UA–; 1 = UA+). Trial arm represents the 7- and 28-day taper groups (0 = 7 days; 1 = 28 days).

UA, urine analysis; ns, non-significant.

^*p < 0.05.