Table 3.
Factor | Meta-regression (1) Crude mortality rate (CMR) of suicide by study factors | Meta-regression (2) Standardised mortality ratio (SMR) of suicide by study factors | ||||||
---|---|---|---|---|---|---|---|---|
Number of studies (n = 19) | CMR (95% CI) per 100,000 person years | p value | I2 | Number of studies (n = 11) | SMR (95% CI) | p value | I2 | |
Type of incarceration facility | ||||||||
Jail | 1 | 9.2 (0.0, 163.2) | 0.033 | 99.0% | 1 | 1.0 (0.0, 12.7) | 0.416 | 97.5% |
Prison | 16 | 113.5 (74.5, 152.4) | 9 | 8.4 (4.3, 12.5) | ||||
Youth detention | 2 | 256.5 (139.0, 374.0) | 1 | 9.2 (0.0, 22.1) | ||||
Study design | ||||||||
Prospective | 1 | 204.0 (19.2, 388.8) | 0.350 | 99.2% | 1 | 6.7 (3.6, 9.9) | 0.047 | 98.0% |
Retrospective | 18 | 117.5 (73.4, 160.7) | 10 | 18.4 (7.4, 29.4) | ||||
Interval censoring | ||||||||
No | 7 | 140.7 (70.2, 211.1) | 0.494 | 99.2% | 4 | 11.0 (5.4, 16.7) | 0.131 | 97.3% |
Yes | 12 | 111.4 (58.0, 164.8) | 7 | 5.9 (1.8, 10.0) | ||||
Total length of follow-up (years)a | ||||||||
≤ 10 years | 9 | 123.4 (60.6, 186.1) | 0.955 | 99.2% | 6 | 10.1 (4.9, 15.3) | 0.195 | 98.1% |
> 10 years | 10 | 121.0 (61.6, 180.5) | 5 | 5.8 (1.14, 11.9) | ||||
Countryb | ||||||||
Australia | 6 | 166.5 (137.8, 195.1) | < 0.001 | 96.7% | 4 | 6.6 (0.3, 12.8) | 0.337 | 95.1% |
Canada | 1 | 58.6 (0.0, 118.6) | 1 | 4.3 (0.0, 15.8) | ||||
Finland | 1 | 340.9 (255.4, 426.5) | – | – | ||||
Netherlands | – | – | – | – | 1 | 6.7 (0.0, 19.3) | ||
Norway | 1 | 82.8 (22.9, 142.8) | – | – | ||||
Sweden | 2 | 200.6 (153.4, 247.8) | 1 | 18.4 (5.1, 31.7) | ||||
England and Wales | 2 | 117.0 (72.1, 162.1) | 1 | 13.5 (1.9, 25.1) | ||||
US | 6 | 34.9 (10.4, 59.3) | 3 | 5.5 (0.0, 12.5) | ||||
Male only samplesb | ||||||||
No | 18 | 126.6 (83.5, 169.8) | 0.344 | 99.24% | – | – | – | – |
Yes | 1 | 40.9 (0.0, 221.6) | – | – |
CMR crude mortality rate, SMR standardised mortality ratio, 95% CI 95% confidence interval
aIncluded studies for meta-analysis 1 had a median follow up length of 10 years, range 0–25 years. Included studies for meta-analysis 2 had a median follow-up length of 10.8 years, range 0–25 years
bAs no studies included in the primary analysis reported SMRs for men only, the “Male only samples” variable was not included in meta-regression 2