Table 4:

Five-year outcomes by self-harm and history of mental health conditions*

OutcomeNo. of adolescents with no history of mental health conditions
n = 11 737
No. of adolescents with history of mental health conditions
n = 4655
Time-to-event analyses
No. (%) of adolescents with no self-harm visits
n = 7693
No. (%) of adolescents with ≥ 1 self-harm visits
n = 4044
No. (%) of adolescents with no self-harm visits
n = 3038
No. (%) of adolescents with ≥ 1 self-harm visits
n = 1617
Self-harm HR (95% CI)History of mental health conditions HR (95% CI)Self-harm × mental health history HR (95% CI)
Readmissions to ED or hospital for self-harm365 (4.7)1144 (28.3)272 (9.0)482 (29.8)6.85 (6.11–7.69)1.92 (1.64–2.24)7.17 (6.25–8.34)
Overall mortality22 (0.3)49 (1.2)12 (0.4)9 (0.6)4.26 (2.57–7.05)1.38 (0.69–2.79)1.94 (0.88–4.25)
Suicides6 (0.1)36 (0.9)≤ 56 (0.4)11.48 (4.79–27.5)1.69 (0.48–5.98)4.76 (1.51–15.00)
Mean ± SDMean ± SDMean ± SDMean ± SDp valuep valuep value
Lower-bound estimates of 5-year health care costs, $16 079 ± 44 01130 600 ± 58 89126 591 ± 72 31029 859 ± 67 002< 0.001< 0.001< 0.001
  • Note: CI = confidence interval, ED = emergency department, HR = hazard ratio,

  • * Time-to-event analyses were Cox regressions of the time to the outcome regressed on whether the adolescent had a self-harm emergency department visit (self-harm column), whether the adolescent had a mental health code associated with the index emergency department visit or any other medical service in the last year (history of mental health conditions column), and the interaction between those markers (self-harm × mental health history column). In the cell with “≤ 5,” the actual count and percentage were suppressed to prevent reidentification of patients. The bottom row of the table reports a 2 × 2 analysis of variance for the logs of the lower-bound 5-year cumulative costs.