Factors influencing risk for HIT: type of heparin, patient population, and gender (fixed-effects statistical approach)
. | . | 95% CI . | . | . | |
|---|---|---|---|---|---|
. | Common OR for HIT . | Lower . | Upper . | P . | |
| Group (no. of studies) | |||||
| Overall effect of heparin type: UFH vs LMWH (7) | 5.29 | 2.84 | 9.86 | < .0001 | |
| Overall effect of patient type: surgical vs medical (7)* | 3.25 | 1.98 | 5.35 | < .0001 | |
| Overall effect of gender: female vs male | 2.37 | 1.37 | 4.09 | .0015 | |
| Interactions (no. of studies)† | |||||
| Female (7) | 9.22 | 3.87 | 21.97 | < .0001 | |
| Male (7) | 1.83 | 0.64 | 5.23 | .291 | |
| Females vs males | — | — | — | .020 | |
| Surgical (4) | 13.93 | 4.33 | 44.76 | < .0001 | |
| Medical (3) | 1.75 | 0.73 | 4.22 | .233 | |
| Surgical vs medical | — | — | — | .005 | |
| Female surgical (4) | 17.39 | 4.22 | 71.70 | < .0001 | |
| Female medical (3) | 3.75 | 1.16 | 12.17 | .025 | |
| Female surgical vs female medical | — | — | — | .103 | |
. | . | 95% CI . | . | . | |
|---|---|---|---|---|---|
. | Common OR for HIT . | Lower . | Upper . | P . | |
| Group (no. of studies) | |||||
| Overall effect of heparin type: UFH vs LMWH (7) | 5.29 | 2.84 | 9.86 | < .0001 | |
| Overall effect of patient type: surgical vs medical (7)* | 3.25 | 1.98 | 5.35 | < .0001 | |
| Overall effect of gender: female vs male | 2.37 | 1.37 | 4.09 | .0015 | |
| Interactions (no. of studies)† | |||||
| Female (7) | 9.22 | 3.87 | 21.97 | < .0001 | |
| Male (7) | 1.83 | 0.64 | 5.23 | .291 | |
| Females vs males | — | — | — | .020 | |
| Surgical (4) | 13.93 | 4.33 | 44.76 | < .0001 | |
| Medical (3) | 1.75 | 0.73 | 4.22 | .233 | |
| Surgical vs medical | — | — | — | .005 | |
| Female surgical (4) | 17.39 | 4.22 | 71.70 | < .0001 | |
| Female medical (3) | 3.75 | 1.16 | 12.17 | .025 | |
| Female surgical vs female medical | — | — | — | .103 | |
All comparisons showed homogeneity among the respective studies (Breslow-Day, P > .15), except for overall effect of heparin type: UFH vs LMWH (7) (Breslow-Day, P = .009). The inhomogeneity resulted from inclusion of one prospective before-and-after cohort study.34,35 When this study was removed from analysis, the resulting Breslow-Day statistic (P = .614) indicated homogeneity. Analysis of the remaining 6 studies28-33 showed even greater overall risk for HIT with UFH than with LMWH (common OR, 15.63; 95% CI, 4.86-50.24; P < .001). In addition, the remaining studies showed even greater overall risk for HIT in females than in males (common OR, 4.00; 95% CI, 1.89-8.48; P = .0001). Similar results were seen when the randomized controlled trial28 analyzed in study 2 was excluded from this analysis. For example, when analyzing only the remaining studies,29-35 significant overall effects of heparin type (common OR, 4.56; 95% CI, 2.29-9.10; P < .0001), patient type (common OR, 2.77; 95% CI, 1.58-4.87; P < .001), and gender (common OR, 2.17; 95% CI, 1.16-4.07; P = .013) were seen. Similar results were observed in analyses for interactions in heparin type, patient type, and gender (data not shown).
Studies were pooled across patient type to produce a simple 2 × 2 table. Surgical, 42 of 1999; medical, 25 of 3811. Fisher exact test (2-sided) P value.
Male surgical and male medical comparisons were not considered because of lack of events. Interactions included all other parameters regarding risk for HIT comparing treatment with UFH and with LMWH.