Skip to main content

Table 2 Logistic regression analysis for factors associated with weapon use among forensic psychiatric patients in Ontario

From: Weapon use during the index offense: a study among forensic psychiatry patients in Ontario, Canada

Variables

Null hypothesis

Bivariate regression analysis

Multivariate regression analysis

Crude odds ratio (95% confidence interval)

p-value

Adjusted odds ratio (95% confidence interval)

p-value

Patients’ characteristics

Age (years)

 

Age of the patients at the index offence has no effect on the log odds of having used a weapon during the index offence

0.99 (0.98–1,00

0.134

1.00 (0.99–1.01)

0.828

Gender (as reported by patient)

Male

There is no linear relationship between the gender of patients and log odds of having used a weapon during the index offence

1

 

1

 

Female

0.90 (0.61–1.32)

0.582

1.07 (0.69–1.67)

0.752

Previous hospitalization for a psychiatric condition

No

There is no linear relationship between having history of being hospitalized prior to index offence and log odds of having used a weapon during the index offence

1

 

1

 

Yes

0.42 (0.28–0.63)

< 0.001

0.43 (0.27–0.67)

< 0.001

Primary psychiatry diagnosis

Psychotic disorder

There is no linear relationship between the primary diagnosis and log odds of having used a weapon during the index offence

1

 

1

 

Mood disorder

0.59 (0.33–1.06)

0.079

0.65 (0.34–1.24)

0.190

Neurodevelopmental disorder

1.31 (0.59–2.93)

0.509

1.71 (0.67–4.36)

0.258

Personality disorder

0.33 (0.14–0.78)

0.011

0.43 (0.16–1.16)

0.097

Others

1.40 (0.69–2.83)

0.353

1.89 (0.85–4.22)

0.118

Intoxication at the time of the index

No

There is no linear relationship between being intoxicated during the index offence and log odds of having used a weapon during the index offence

1

 

1

 

Drugs

1.73 (0.91–3.29)

0.096

1.92 (0.93–3.94)

0.076

Alcohol

0.83 (0.44–1.56)

0.562

0.58 (0.29–1.18)

0.136

Yes (not specified)

4.43 (0.49–39.80)

0.184

2.22 (0.22–21.98)

0.495

Both (alcohol and drugs)

2.77 (0.53–14.36)

0.226

1.98 (0.34–11.34)

0.443

Unknown

0.61 (0.20–1.85)

0.388

0.55 (0.17–1.78)

0.319

Victim characteristics

Victim gender

Male

There is no linear relationship between the victim’s gender and log odds of having used a weapon during the index offence

1

 

1

 

Female

0.68 (0.51–0.91)

0.009

0.54 (0.38–0.76)

< 0.001

Unknown

0.31 (0.17–0.54)

< 0.001

0.58 (0.30–1.12)

0.107

Victim age

Age in years at the time of the offence

Victim’s age of the patients at the index offence has no effect on the log odds of having used a weapon during the index offence

1.02 (1.00–1.03)

0.027

  

Victim relationship

Stranger – adult

Victim’s relation to the patient has no effect on the log odds of having used a weapon during the index offence

1

 

1

 

Stranger – child

0.49 (0.17–1.41)

0.186

0.44 (0.14–1.34)

0.149

Acquaintance – adult

1.97 (1.25–3.11)

0.003

1.70 (1.06–2.74)

0.029

Acquaintance – child

0.63 (0.56–7.09)

0.712

0.38 (0.03–4.98)

0.461

Friend

2.11 (0.74–6.01)

0.160

1.18 (0.38–3.66)

0.768

Parent

2.41 (1.51–3.83)

< 0.001

2.64 (1.60–4.31)

< 0.001

Son/daughter

0.85 (0.33–2.15)

0.725

0.69 (0.25–1.92)

0.479

Sibling

2.66 (1.20–5.91)

0.016

3.14 (1.37–7.16)

0.007

Lover/partner/spouse

2.37 (1.20–5.91)

0.013

2.74 (1.32–5.60)

0.007

Other family members

6.35 (2.10–19.14)

0.001

6.61 (2.13–20.54)

0.001

Law enforcement professionals

0.46 (0.26–0.79)

0.005

0.39 (0.21–0.71)

0.002

Healthcare/support staff

0.25 (0.13–0.51)

< 0.001

0.27 (0.13–0.55)

< 0.001

Co-habitant/co-patient

0.98 (0.41–2.31)

0.957

0.82 (0.34–2.02)

0.672

Others

1.90 (0.52–6.92)

0.328

1.54 (0.41–5.80)

0.526

Unknown

0.36 (0.12–1.13)

0.081

0.50 (0.14–1.70)

0.265

  1. Model statistics All the variables included in the final model had a VIF of less than 2 and a mean VIF of 1.05. In addition, the final model had a goodness of fit chi-square value of 657.29 and a p-value of 0.227, meaning that we did not have enough evidence to reject the null hypothesis that the model had a good fit and that the probability of having a chi-square value this high or higher was above 0.05