Table 1.
Comparison of effect estimate by various subgroups
| Grouping variable | Subgroup | Numb. of studies | Sample | RR (95% CI) | Prediction intervals | Tau2 | I2 (95%CI) | H (95%CI) | P-value for heterogeneity | P-value Egger’s test | P-value difference in subgroups |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Overall (any injury) | – | 11 | 38,662,538 | 0.47 (0.28–0.80) | 0.07–3.27 | 0.669 | 99.9 (99.9–99.9) | 34.7 (33.2–36.4) | < 0.0001 | 0.361 | NA |
| By injury site | Head | 6 | 32,964,587 | 0.49 (0.22–1.08) | 0.03–8.65 | 0.910 | 99.3 (99.1–99.4) | 11.6 (10.1–13.4) | < 0.0001 | 0.046 | 0.090 |
| Spinal | 7 | 5,281,081 | 0.56 (0.37–0.84) | 0.15–2.01 | 0.206 | 93.1 (88–96) | 3.8 (2.9–5.0) | < 0.0001 | 0.328 | ||
| Facial | 2 | 409,016 | 0.75 (0.40–1.43) | – | 0.207 | 96.9 | – | < 0.0001 | – | ||
| Neck | 2 | 855 | 0.69 (0.07–6.44) | – | 1.761 | 56.6 | – | 0.129 | – | ||
| Thoracic | 2 | 855 | 0.96 (0.74–1.25) | – | 0.0 | 0.0 | – | 0.754 | – | ||
| Abdomen | 2 | 4621 | 0.93 (0.84–1.04) | – | 0.0 | 0.0 | – | 0.912 | – | ||
| Upper limb | 1 | 766 | 1.05 (0.83–1.34) | – | – | – | – | – | – | ||
| Lower limb | 1 | 766 | 0.77 (0.58–1.04) | – | – | – | – | – | – | ||
| By crash type | Motor vehicle collision | 5 | 5,692,713 | 0.64 (0.46–0.89) | 0.19–2.17 | 0.119 | 98.8 (98–99) | 9.0 (7.4–10.8) | < 0.0001 | 0.670 | 0.001 |
| Rollover | 3 | 3,276,677 | 0.43 (0.20–0.93) | 0.0–963.4 | 0.465 | 99.2 (99–99.6) | 11.4 (9.1–14.3) | < 0.0001 | 0.268 | ||
| Multiple | 3 | 3447 | 0.37 (0.30–0.44) | 0.11–1.27 | 0.0 | 0.0 (0.0–90) | 1.0 (1.0–3.1) | 0.506 | 0.025 | ||
| Front | 1 | 18,296,847 | 0.18 (0.18–0.19) | – | – | – | – | – | – | ||
| Side | 1 | 8,571,748 | 0.15 (0.14–0.15) | – | – | – | – | – | – | ||
| Rear | 1 | 2,821,106 | 0.09 (0.08–0.09) | – | – | – | – | – | – | ||
| By study sample size | < median (6128) | 5 | 4662 | 0.51 (0.20–1.30) | 0.02–12.46 | 0.783 | 94.5 (90–97) | 4.3 (3.2–5.8) | < 0.0001 | 0.121 | 0.594 |
| ≥ median (6128) | 6 | 38,657,876 | 0.45 (0.23–0.86) | 0.04–5.10 | 0.659 | 99.9 (99.9–99.9) | 48.7 (46.3–51.2) | < 0.0001 | 0.340 | ||
| By study type | Hospital | 3 | 6189 | 0.39 (0.01–1.72) | 0.0–5025 | 1.079 | 61.1 (0–89) | 1.6 (1.0–3.0) | 0.077 | 0.038 | 0.967 |
| Patient databases | 8 | 38,656,349 | 0.48 (0.27–0.84) | 0.06–3.88 | 0.648 | 99.9 (99.9–99.9) | 40.6 (38.6–42.6) | < 0.0001 | 0.452 | ||
| Publication year | Before 2010 | 3 | 5,270,440 | 0.36 (0.33–0.38) | 0.22–0.57 | 0.0 | 0.0 (0.0–90) | 1.0 (1.0–3.1) | 0.530 | 0.036 | 0.269 |
| On or after 2010 | 8 | 33,392,098 | 0.54 (0.30–0.98) | 0.06–4.75 | 0.697 | 99.9 (99.9–99.9) | 41.4 (39.4–43.4) | < 0.0001 | 0.388 |
NB: Absent predictive intervals could not be estimated due to inadequate number of studies (less than 3); absent I2 confidence intervals could not be estimated due to degrees of freedom less than 2 and Egger’s test was inestimable for groups with less than 3 studies