Table 5.
Summary of meta-analytic results, subgroup analyses, and publication biases
| Outcome | Studies | Meta-analytic results | Subgroup analyses | Publication biases | |||
|---|---|---|---|---|---|---|---|
| SMD [95% CI] | Timepoint | Program type | Occupation | Study design | Linear regression test of funnel plot asymmetry | ||
| Absenteeism | 4 | 0.01 [− 0.19; 0.21] | – | – | – | – | – |
| Alcohol | 10 | − 0.08 [− 0.21; 0.06] | – | – | – | – | – |
| Antithrombin | 1 | 0.49 [− 0.45; 1.44] | – | – | – | – | – |
| Anxiety | 21 | − 0.20 [− 0.31; − 0.10] | 0.61 | 0.31 | 0.76 | – | − 0.10 [− 0.23; 0.02], p < 0.01 * |
| Blood pressure | 8 | − 0.17 [− 0.38; 0.04] | – | – | – | – | – |
| Burnout | 15 | − 0.45 [− 0.64; − 0.26] * | 0.15 | 0.04* | 0.46 | – | − 0.28 [− 0.47; − 0.09], p = 0.07 |
| Coping | 5 | 0.41 [0.02; 0.80] * | < 0.01* | < 0.01* | < 0.01* | – | – |
| Cortisol | 6 | − 0.20 [− 0.44; 0.04] | – | – | – | – | – |
| Depression | 18 | − 0.46 [− 0.71; − 0.21] * | 0.05* | < 0.01* | 0.80 | 0.88 | − 0.17 [− 0.47; 0.12], p = 0.03 * |
| DHEA | 2 | − 0.29 [− 0.80; 0.21] | – | – | – | – | – |
| Drug use | 2 | − 0.05 [− 0.35; 0.24] | – | – | – | – | – |
| General symptoms | 7 | − 0.70 [− 1.14; − 0.26] * | < 0.01* | < 0.01* | 0.46 | – | – |
| Heart rate | 21 | − 0.27 [− 0.40; − 0.14] | 0.04* | 0.01* | 0.04* | – | − 0.23 [− 0.37; − 0.09], p = 0.10 |
| Prolactin | 1 | − 0.07 [− 0.56; 0.41] | – | – | – | – | – |
| PTSD | 9 | − 0.33 [− 0.55; − 0.11] * | < 0.01* | 0.10 | 0.03* | – | – |
| Resilience | 22 | 0.27 [0.13; 0.42] * | 0.02* | < 0.01* | 0.29 | 0.78 | 0.27 [0.13; 0.42], p = 0.62 |
| Stress | 25 | − 0.35 [− 0.51; − 0.20] * | 0.49 | 0.14 | 0.02* | 0.80 | − 0.21 [− 0.37; − 0.05], p = 0.55 |
| Suicidality | 2 | 0.33 [− 0.07; 0.73] | – | – | – | – | – |
| Well-being | 20 | 0.46 [0.26; 0.66] * | 0.63 | 0.24 | < 0.01* | 0.45 | 0.46 [0.26; 0.66], p = 0.16 |
The standardized mean difference (SMD) is a method of pooling continuous outcomes (i.e., scores on rating scales) in meta-analysis. It is preferred over mean differences when there are differences in how the outcome is measured across studies. The asterisk (*) indicates that the pooled estimate is statistically significant at p < 0.05 (i.e., the confidence interval (CI) does not overlap with the null). Subgroup analyses were only performed where at least ten studies were included. Statistically significant subgroup analyses (p < 0.05) are marked with an asterisk (*). The linear regression test of funnel plot asymmetry is used to appraise publication bias in the pooled estimate for any individual outcome in the meta-analysis. p values smaller than 0.05 (marked with an asterisk) are considered statistically significant and indicate evidence of publication bias in that outcome’s pooled estimate. DHEA, dehydroepiandrosterone; PTSD, posttraumatic stress disorder