Socioeconomic correlates of incident and fatal opioid overdose among Swedish people with opioid use disorder

Table 2 Associations of incident opioid overdose (OD). N = 22,079. Cox Regression Models (Hazard ratios and 95% CIs)

	UNIVARIATE MODELS	MULTIVARIATE MODEL
Male sex	1.43 (1.34; 1.53)*	1.29 (1.20; 1.38)*
Age at OUD registration	0.97 (0.97; 0.97)*	1.04 (1.03; 1.05)*
Year of birth	1.03 (1.03; 1.04)*	1.07 (1.06; 1.08)*
Country of birth
Nordic countries (vs. Sweden)	1.07 (0.89; 1.30)^a	1.08 (0.88; 1.32)
Europe (vs. Sweden)	0.75 (0.64; 0.87)^a*	0.84 (0.72; 0.98)*
Asia (vs. Sweden)	0.58 (0.51; 0.67)^a*	0.63 (0.55; 0.74)*
Outside Europe/Asia (vs. Sweden)	0.96 (0.80; 1.15)^a	0.99 (0.82; 1.19)
Criminal conviction	1.85 (1.73; 1.99)^a*	1.53 (1.42; 1.65)*
Education (years)	0.92 (0.90; 0.93)^a*	0.96 (0.94; 0.97)*
School Grades	0.91 (0.88; 0.94)^a*	N/A^b
IQ	0.87 (0.81; 0.92)^a*	N/A^b
Resilience	0.91 (0.86; 0.97)^a*	N/A^b
Parental education (years)	0.98 (0.96; 0.99)^a*	N/A^b
Social welfare	1.60 (1.51; 1.71)^a*	1.31 (1.22; 1.39)*
Income	0.75 (0.69; 0.82)^a*	NS
Neighborhood deprivation	1.02 (1.00; 1.03)^a*	NS
Married	0.78 (0.72; 0.85)^a*	NS
1 ≤ children	1.01 (0.94; 1.08)^a	NS
Distance to mother		N/A^b
0–10 km (vs. same place)	0.95 (0.88; 1.03)^a	N/A^b
10–50 km (vs. same place)	1.18 (1.05; 1.32)^a*	N/A^b
50+ km (vs. same place)	1.36 (1.11; 1.66)^a*	N/A^b
Distance to father		N/A^b
0–10 km (vs. same place)	1.06 (0.96; 1.16)^a	N/A^b
10–50 km (vs. same place)	1.15 (1.01; 1.30)^a*	N/A^b
50+ km (vs. same place)	1.23 (1.00; 1.52)^a*	N/A^b
Prior OUD/opioid OD	1.96 (1.73; 2.22)*	1.70 (1.49; 1.94)*
Inpatient registration of OUD	1.31 (1.23; 1.40)^a*	1.28 (1.20; 1.36)*
Psychiatric disorder	1.96 (1.84; 2.09)^a*	1.76 (1.64; 1.89)*

HR Hazard ratio, CI confidence interval, OD overdose, OUD opioid use disorder, N/A not available, NS not significant
^aControlled for sex, age at registration, year of birth, prior OUD/opioid OD
^bVariable not included in the multivariable analysis due to a relatively large proportion of missing information. These variables were not missing due to poor register quality; rather, they were missing because not all individuals were represented in all registers (e.g. The Conscript Register includes almost exclusively men)
* p < 0.05

ISSN: 1747-597X