Table 3. Difference-in-Differences Analyses for Proportion of Individuals With a New Suicide Prevention Safety Plan Documentation and Mortality Outcomes in Subsequent 6 Months.
Outcome | REACH VET era | Pre–REACH VET era | Difference-in-difference, mean (95% CI)a | |||||
---|---|---|---|---|---|---|---|---|
Risk tier, mean (SD), proportion | Difference, mean (95% CI)b | Risk tier, mean (SD), proportion | Difference, mean (95% CI)b | Unadjusted | Adjusting for age and sex | |||
Top 0.1% | Top 0.3%-0.1% | Top 0.1% | Top 0.3%-0.1% | |||||
With a new safety planc | 0.1680 (0.3739) | 0.0558 (0.2295) | 0.1123 (0.1046 to 0.1199) | 0.1284 (0.3345) | 0.0523 (0.2226) | 0.0761 (0.0702 to 0.0820) | 0.0362 (0.0266 to 0.0458) | 0.0768 (0.0561 to 0.0974) |
All-cause mortalityd | 0.0217 (0.1456) | 0.0317 (0.1752) | −0.0100 (−0.0121 to −0.0079) | 0.0195 (0.1381) | 0.0286 (0.1668) | −0.0092 (−0.0113 to −0.0071) | −0.0009 (−0.0038 to 0.0021) | 0.0000 (−0.0001 to 0.0001) |
Cause-specific mortalitye | ||||||||
Suicide | 0.0028 (0.0526) | 0.0016 (0.0395) | 0.0012 (0.0005 to 0.0019) | 0.0025 (0.0501) | 0.0018 (0.0427) | 0.0007 (0.0000 to 0.0013) | 0.0005 (−0.0004 to 0.0015) | 0.0007 (−0.0006 to 0.0019) |
Nonsuicide external cause | 0.0052 (0.0720) | 0.0043 (0.0651) | 0.0010 (−0.0001 to 0.0020) | 0.0038 (0.0617) | 0.0030 (0.0546) | 0.0008 (0.0000 to 0.0016) | 0.0001 (−0.0012 to 0.0014) | 0.0001 (−0.0015 to 0.0018) |
All-causef | 0.0223 (0.1476) | 0.0320 (0.1760) | −0.0097 (−0.0121 to −0.0073) | 0.0192 (0.1371) | 0.0283 (0.1659) | −0.0092 (−0.0112 to −0.0071) | −0.0005 (−0.0037 to 0.0026) | 0.0001 (−0.0001 to 0.0002) |
REACH VET era − pre–REACH VET Era.
Mean difference (top 0.1% − top 0.3% to 0.1%).
Cohorts were limited to those with no safety plan in the prior 2 years, for a total of 77 625 individuals.
Mortality ascertained from the Vital Status File. Cohorts were limited to those for whom vital status could be ascertained, for a total of 173 305 individuals.
Mortality ascertained from the Mortality Data Repository. Cohorts were limited to those for whom cause of death could be ascertained, for a total of 152 515 individuals.
Adjusted analyses used a modified Poisson model.