Table 3.
Cox proportional hazard model to estimate risk of completed suicide after adjustment for measured risk factors and confounders
| Analysis covariates | Hazard ratio* (95% CI) |
|---|---|
| Venlafaxine v citalopram | 1.70 (0.76 to 3.80); 2.44 (1.12 to 5.31)† |
| Venlafaxine v fluoxetine | 1.63 (0.74 to 3.59); 2.85 (1.37 to 5.94)† |
| Venlafaxine v dothiepin | 1.31 (0.53 to 3.25); 2.54 (1.07 to 6.02)† |
| Men v women | 2.15 (0.53 to 8.67) |
| Age at start of treatment episode: | |
| 30-59 v 18-29 | 0.56 (0.17 to 1.77) |
| ≥60 v 18-29 | 0.33 (0.07 to 1.51) |
| Sex-age interaction: | |
| Men aged 30-59 | 1.83 (0.38 to 8.80) |
| Men aged ≥60 | 2.31 (0.32 to 16.47) |
| Overlap with other antidepressants (yes v no) | 0.75 (0.15 to 3.65) |
| First 30 days of treatment episode (yes v no) | 2.85 (1.51 to 5.39) |
| Duration of first prescription (≤14 v ≥15 days) | 1.35 (0.55 to 3.31) |
| Drug history before treatment episode: | |
| Same antidepressant (yes v no) | 0.51 (0.26 to 0.99) |
| Other antidepressants ever before (yes v no) | 0.56 (0.21 to 1.50) |
| Other antidepressants in year before (yes v no) | 2.21 (0.91 to 5.39) |
| No of other previous antidepressants (3 v 1 or 2) | 1.00 (0.48 to 2.09) |
| History of antidepressant multitherapy (yes v no) | 1.40 (0.59 to 3.30) |
| Suicidal behaviour in year before treatment episode (yes v no) | 7.51 (2.82 to 19.96) |
| Admission to hospital for psychiatric disorder in month before treatment episode (yes v no) | 2.53 (0.74 to 8.68) |
| Psychiatric referral in year before treatment episode (yes v no) | 0.99 (0.23 to 4.32) |
| Psychotropic therapy in 60 days before treatment episode: | |
| Any psychotropics (yes v no) | 1.96 (0.86 to 4.42) |
| Antipsychotics (yes v no) | 2.46 (1.05 to 5.80) |
| Mood stabilisers (yes v no) | 0.71 (0.22 to 2.34) |
| Any psychiatric morbidity in year before treatment episode (yes v no) | 0.84 (0.39 to 1.82) |
| Medical history: | |
| Bipolar disorder (yes v no) | 4.94 (1.30 to 18.84) |
| Depression (yes v no) | 1.15 (0.34 to 3.86) |
| Mixed anxiety and depression (yes v no) | 0.71 (0.33 to 1.50) |
| Substance misuse (yes v no) | 1.30 (0.54 to 3.13) |
*Measures relative risk of completed suicide adjusted for each analysis covariate after adjustment for all other variables in model, including treatment.
†Unadjusted values.