Table 4.

 Univariable and multivariable associations between covariates and drug dependent cause specific mortality

Cohort characteristic	Unadjusted hazard ratio (95% CI)	Adjusted hazard ratio (95% CI)
Male sex	1.24 (0.70 to 2.21)	NA
Age (per year)	0.98 (0.95 to 1.01)	1.00 (0.98 to 1.01)
Social class (Carstairs):
1-4 (reference category)	1.0	NA
5-7	1.12 (0.63 to 2.02)	NA
Mean methadone dose ≥60 mg	0.68 (0.29 to 1.58)	NA
Overusing methadone	4.52 (2.47 to 8.26)*	1.85 (0.91 to 3.80)
Methadone breaks	0.51 (0.39 to 0.68)*	1.01 (0.74 to 1.40)
Duration of methadone treatment (years)	0.76 (0.69 to 0.83)*	0.93 (0.92 to 0.95)*†
Time since last methadone prescription filled (months):
≤1 (reference category)	1.0	1.0
2-3	0.52 (0.22 to 1.24)	0.64 (0.26 to 1.04)
4-6	0.40 (0.10 to 1.67)	0.24 (0.06 to 1.01)
>6	0.34 (0.19 to 0.62)*	0.02(0.00 to 0.05)*
Psychiatric admission	2.23 (1.22 to 4.07)*	2.41 (1.25 to 4.64)*
Having urine tested	0.27 (0.16 to 0.45)*	0.52 (0.26 to 1.04)
Co-prescribing:
Benzodiazepines	2.73 (0.85 to 8.76)	4.35 (1.32 to 14.30)*
Antipsychotics	0.18 (0.056 to 0.58)*	0.27 (0.08 to 0.89)*
Antidepressants	0.29 (0.16 to 0.53)*	0.51 (0.30 to 0.98)*
Opioid analgesics	0.65 (0.39 to 1.10)	0.72 (0.41 to 1.26)

Final regression model mutually adjusted for all significant covariates; male sex, comorbidity (Charlson index), social class, mean methadone dose (≥60 mg daily), and volume of methadone prescribing by practice were assessed unadjusted but had no significant influence on cause specific mortality (P>0.2) and so were excluded from final model.

NA=not applicable.

*P<0.05.

†Variable treated as time varying coefficient in Cox regression model and hazard ratio reported at median follow-up of 4.38 years.