Table 4. Adjusted hazard ratios of suicides by depressive symptoms according to follow-up period.a .
From 2001 to 2004 (n = 10,238) | From 2005 to 2008 (n = 9,877) | ||||||
---|---|---|---|---|---|---|---|
Categories of depressive symptoms (score) | No. of suicidesb | p-value | HR (95% CI) | No. of suicidesb | p-value | HR (95% CI) | |
Total BDI score (0–63) | Five-score increase | 16 | 0.02 | 1.3 (1.1–1.8) | 25 | 0.16 | 1.1 (1.0–1.4) |
2-group analysisc | No-to-moderate (0–30) | 4 | 1.0 (Reference) | 10 | 1.0 (Reference) | ||
Severe (31–63) | 12 | 0.02 | 5.2 (1.3–19.8) | 15 | 0.06 | 2.7 (0.9–7.6) | |
5-group analysisc | No (0–13) | 1 | 1.0 (Reference) | 2 | 1.0 (Reference) | ||
Mild (14–21) | 3 | 0.70 | 1.6 (0.1–17.6) | 3 | 0.87 | 1.2 (0.2–8.4) | |
Moderate (22–30) | 1 | 0.99 | 0.0 | 4 | 0.80 | 1.3 (0.2–9.2) | |
Severe (31–39) | 4 | 0.33 | 3.3 (0.3–37.8) | 8 | 0.21 | 3.3 (0.5–20.9) | |
Extreme (40–63) | 8 | 0.13 | 7.1 (0.6–88.3) | 7 | 0.23 | 3.3 (0.5–23.8) | |
Trend testd | 16 | 0.03 | 1.9 (1.1–3.3) | 25 | 0.09 | 1.4 (0.9–2.2) |
Notes.
- BDI
- Beck Depression Inventory
- CI
- confidence interval
- HR
- hazard ratio
Hazard ratios were calculated using a Cox proportional hazards model over multiple imputed data, after adjustment for age at enrollment, smoking status, drinking status, body mass index, selfrated health, marital status, educational status, household monthly income and participant group.
Number of suicides may not match those of the other groups of depressive symptoms due to rounding of averages over multiple imputed data.
Cut-off score based on quartiles and the last decile.
Five categories of depressive symptoms were analyzed as ordinal variables.