Skip to main content
. 2017 Aug 10;117(8):1185–1191. doi: 10.1038/bjc.2017.262

Table 2. Subgroup analysis of diagnostic effect.

      Sensitivity Specificity Diagnostic odds ratio
Subgroup No. studies No. patients Value I2 (%) P Value I2 (%) P Value I2 (%) P
Race
Caucasian 5 651 0.91 (0.84–0.96) 0.0 0.513 0.87 (0.84–0.89) 93.5 0.000 80.01 (27.41–233.57) 45.7 0.118
Others 7 2182 0.83 (0.80–0.86) 66.4 0.007 0.92 (0.90–0.93) 72.7 0.001 59.09 (37.64–92.77) 38.3 0.137
Number of patients
<100 5 282 0.95 (0.85–0.99) 0.0 0.604 0.87 (0.82–0.91) 92.9 0.000 147.25 (33.91–639.50) 27.8 0.236
⩾100 7 2551 0.84 (0.80–0.86) 65.6 0.008 0.91 (0.90–0.92) 82.7 0.000 59.56 (38.97–91.03) 45.0 0.091
Number of slices detected by PE
2 5 702 0.88 (0.81–0.93) 28.6 0.231 0.95 (0.93–0.97) 35.3 0.186 138.70 (71.44–269.2) 0.0 0.814
Others 4 1226 0.88 (0.84–0.91) 0.0 0.641 0.86 (0.83–0.88) 92.9 0.000 52.51 (34.95–78.89) 0.0 0.477
Thicknesses
<1 mm 5 491 0.85 (0.75–0.92) 34.3 0.193 0.86 (0.82–0.89) 94.5 0.000 66.07 (20.10–217.27) 37.4 0.172
⩾1 mm 3 1518 0.82 (0.77–0.86) 83.3 0.003 0.92 (0.90–0.93) 55.4 0.106 28.66 (29.82–115.39) 66.4 0.051
Interval
200 mm 5 1153 0.87 (0.82–0.91) 24.9 0.255 0.90 (0.88–0.92) 86.5 0.000 71.61 (39.00–131.49) 23.5 0.265
Others 4 803 0.89 (0.84–0.93) 0.0 0.614 0.91 (0.88–0.93) 61.3 0.052 96.78 (39.18–239.06) 44.0 0.148
Reference method
H&E 4 2181 0.88 (0.81–0.93) 45.9 0.136 0.95 (0.93–0.97) 0.0 0.550 131.15 (66.62–258.20) 0.0 0.818
H&E and IHC 8 652 0.84 (0.80–0.87) 63.7 0.007 0.89 (0.87–0.91) 90.6 0.000 50.19 (31.89–78.99) 33.3 0.162
IHC for CK19
Yes 3 453 0.91 (0.84–0.96) 0.0 0.452 0.94 (0.91–0.96) 0.0 0.608 160.21 (68.80–373.11) 0.0 0.851
No 9 2380 0.83 (0.80–0.86) 60.1 0.010 0.90 (0.88–0.91) 90.7 0.000 53.37 (34.96–81.50) 31.7 0.164
Excluding TAB
Yes 3 453 0.91 (0.84–0.96) 0.0 0.452 0.94 (0.91–0.96) 0.0 0.608 160.21 (68.80–373.11) 0.0 0.851
No 9 2380 0.83 (0.80–0.86) 60.1 0.010 0.90 (0.88–0.91) 90.7 0.000 53.37 (34.96–81.50) 31.7 0.164

Abbreviations: CK19=cytokeratin 19; H&E=haematoxylin and eosin staining; IHC=immunohistochemistry; PE=pathological examination; TAB=tissue allocation bias.