Skip to main content

Table 2 Multinomial regression analysis where the dependent variable was change in crime status at follow-up

From: Desistance from crime following substance use treatment: the role of treatment retention, social network and self-control

 

Desisted vs. Continued crime

No crime vs. Continued crime b

 

Unadjusted OR (95% CI), n = 293

p-value

Adjusted model OR (95% CI), n = 277a

p-value

Unadjusted OR (95% CI), n = 102

p-value

Adjusted model OR (95% CI), n = 92a

p-value

Demographics

 Age at baseline

1.01 (0.98–1.04)

p = 0.64

1.05 (1.00–1.10)

p = 0.041

1.14 (1.09–1.19)

p < 0.001

1.14 (1.07–1.21)

p < 0.001

 Being female

1.66 (0.83–3.31)

p = 0.15

1.05 (0.45–2.45)

p = 0.92

3.20 (1.32–7.72)

p = 0.010

2.44 (0.77–7.69)

p = 0.13

Treatment

 Index treatment: Long-term inpatient (ref: OMT)

1.55 (0.88–2.73)

p = 0.13

3.71 (1.12–12.29)

p = 0.032

0.14 (0.04–0.43)

p = 0.001

0.86 (0.07–10.66)

p = 0.91

Treatment status (ref: interrupted treatment)

 - Ongoing treatment at T1

1.76 (0.91–3.43)

p = 0.10

2.90 (1.01–8.36)

p = 0.049

13.33 (2.87–62.00)

p < 0.001

6.30 (0.93–42.88)

p = 0.06

 - Completed according to plan

1.97 (0.88–4.42)

p = 0.10

0.53 (0.17–1.64)

p = 0.27

2.50 (0.36–17.17)

p = 0.35

1.71 (0.11–25.66)

p = 0.70

Substance use

 Reduction in number of substances used

1.23 (1.11–1.36)

p < 0.001

1.09 (0.97–1.22)

p = 0.15

1.02 (0.90–1.17)

p = 0.73

0.90 (0.73–1.11)

p = 0.32

Stimulant use (ref: continued use)

 - Ended use

3.86 (2.01–7.42)

p < 0.001

1.78 (0.71–4.49)

p = 0.22

2.80 (0.89–8.77)

p = 0.08

5.38 (1.11–26.11)

p = 0.036

 - Not used in study period

5.12 (2.12–12.36)

p < 0.001

4.86 (1.72–13.70)

p = 0.003

18.9 (5.89–60.46)

p < 0.001

11.98 (2.80–51.21)

p < 0.001

Other factors

 Change in primary social network (ref: Continued substance using network)

  - Left substance using network

3.86 (1.89–7.88)

p < 0.001

2.87 (1.15–7.18)

p = 0.024

5.10 (1.22–21.25)

p = 0.025

12.60 (2.34–67.88)

p = 0.003

  - No substance using network in study period

3.25 (1.60–6.59)

p = 0.001

2.18 (0.93–5.11)

p = 0.07

13.48 (3.55–51.16)

p < 0.001

9.71 (2.10–44.99)

p = 0.004

  Improvement in self-control score (BSCS)

1.09 (1.05–1.12)

p < 0.001

1.08 (1.04–1.13)

p < 0.001

1.03 (0.99–1.07)

p = 0.20

1.04 (0.98–1.11)

p = 0.22

  1. N = 334. OR Odds ratio. aOR Adjusted odds ratio. CI Confidence interval. Significant ORs (p < 0.05) are in bold. Missing data: Treatment status, n = 6; social network, n = 1; self-control score, n = 6; number of substances, n = 9. a 92% (n = 315) of participants had valid responses for all variables in the adjusted model. b Findings from the additional comparison of No crime vs Continued crime should be considered preliminary as the smaller sample size (n = 92) results in wide confidence intervals (CIs) and uncertain estimates
  2. Model fit: According to the likelihood ratio chi-square test, the multinomial regression model is a significant improvement in fit compared to the null model (X2(22)=137.0, p < 0.001). The Pearson’s chi-square goodness-of-fit test also indicated a good model fit (X2(606) = 584.6, p = 0.73)
Back to article page

BMC Psychiatry

ISSN: 1471-244X

Contact us