Table 2

Logistic regression analyses with a binary response (becoming a father < 22 years or not)
Total Young fathers Single predictor Multipredictor model 1 Multipredictor model 2
n % OR (95% CI) p OR (95% CI) p OR (9 CI) p
Age of the boy’s mother
Old 2225 4.6
Young 151 8.1 1.8 (1.2–2.8) 0.005 1.0 (0.6–1.9) 0.940
Age of the boy’s father
Old 2269 4.8
Young 149 11.4 2.6 (1.5–4.4) <0.001 2.0 (0.98–4.2) 0.059 1.9 (1.1–3.4) 0.029
Educational level of the boy’s mother
High 760 2.4
Low 1789 6.2 2.7 (1.6–4.5) <0.001 1.9 (1.1–3.4) 0.024 2.2 (1.3–3.8) 0.003
Educational level of the boy’s father
High 509 2.6
Low 1833 5.9 2.4 (1.3–4.3) 0.003 1.6 (0.8–3.0) 0.187
Family structure
Intact 2182 5.0
Non-intact 436 5.3 1.1 (0.7–1.7) 0.776
School performance
Average or above 2241 4.9
Below average 425 7.1 1.5 (0.98–2.3) 0.064 0.9 (0.5–1.5) 0.734
Parent and teacher report of psychopathology (SD)
Conduct problems (3.0) 2604 - 1.5 (1.3–1.7) <0.001 1.4 (1.1–1.8) 0.004 1.4 (1.2–1.6) <0.001
Hyperactivity (2.4) 2587 - 1.3 (1.1–1.5) <0.001 1.0 (0.8–1.2) 0.810
Emotional problems (1.9) 2597 - 0.9 (0.8–1.1) 0.366
Child report of psychopathology
CDI total score (5.9) 2651 - 1.3 (1.1–1.5) <0.001 1.2 (1.004–1.4) 0.045 1.2 (0.99–1.4) 0.067
Police register information*
No criminal offenses 2101 4.3 -
1-2 offenses 401 7.5 1.8 (1.2–2.7) 0.008 - 1.3 (0.8–2.1) 0.296
3-5 offenses 108 5.6 1.3 (0.6–3.0) 0.546 - 0.7 (0.2–1.9) 0.425
>5 offenses 111 14.4 3.7 (2.1–6.6) <0.001 - 2.6 (1.3–5.2) 0.008

OR = odds ratio, CI = confidence interval. The variables associated with becoming a young father at significance level p < 0.1 were chosen to the Multipredictor model 1 which included the childhood variables. The variables which remained significant at level p < 0.1 in model 1 were chosen to the model 2 which included the variable on adolescent criminal offenses. OR's > 1 indicate increased likelihood of becoming a young father. The odds ratios for continuous variables (related to psychopathology) were calculated for the change in one SD unit. *Overall p-value was <0.001 in the single predictor model and 0.032 in the multipredictor model.

Lehti et al.

Lehti et al. BMC Public Health 2012 12:560   doi:10.1186/1471-2458-12-560

Open Data