Table 1 Description of study population from each screening center
From: Impact of using artificial intelligence as a second reader in breast screening including arbitration
Characteristic | Center 1 (n = 23,019) | Center 2 (n = 22,583) | ||||
|---|---|---|---|---|---|---|
Age, years | ||||||
Mean, years (s.d.) | 58.7 (5.9) | 58.7 (6.0) | ||||
Range (years) | 50.0–70.0 | 50.0–70.0 | ||||
50–54, n (%) | 7,063 (30.7) | 6,963 (30.8) | ||||
55–59, n (%) | 6,053 (26.3) | 5,863 (26.0) | ||||
60–64, n (%) | 4,995 (21.7) | 4,751 (21.0) | ||||
65+, n (%) | 4,908 (21.3) | 5,006 (22.2) | ||||
Ethnicity, n (%)a | ||||||
White | 13,781 (59.9) | 15,262 (67.6) | ||||
Black | 2,139 (9.3) | 1,965 (8.7) | ||||
Asian | 4,274 (18.6) | 1,990 (8.8) | ||||
Mixed | 646 (2.8) | 460 (2.0) | ||||
Other | 1,021 (4.4) | 608 (2.7) | ||||
Not specified | 1,158 (5.0) | 2,298 (10.2) | ||||
Manufacturer | ||||||
Hologic, n (%) | 23,010 (100.0) | 19,974 (88.4) | ||||
GE Healthcare, n (%) | 9 (0) | 262 (1.2) | ||||
Siemens, n (%) | 0 (0) | 2,347 (10.4) | ||||
Clinical workflow | ||||||
Arbitration protocol | Discordant recalls | Discordant and concordant recalls | ||||
Cancer type, n | ||||||
SD | 169 | 183 | ||||
IC | 68 | 68 | ||||
NSD | 112 | 123 | ||||
Cancer grade | SD | IC | NSD | SD | IC | NSD |
Invasive, n | 138 | 58 | 87 | 144 | 53 | 93 |
G1 | 30 | 4 | 20 | 37 | 5 | 20 |
G2 | 86 | 20 | 52 | 67 | 26 | 52 |
G3 | 21 | 9 | 15 | 37 | 20 | 21 |
No grade | 1 | 25 | – | 3 | 2 | – |
In situ | 30 | 5 | 25 | 39 | 6 | 30 |
High | 16 | 3 | 14 | 16 | 3 | 16 |
Intermediate | 9 | – | 8 | 18 | 1 | 11 |
Low | 5 | – | 3 | 2 | 1 | 3 |
No grade | – | 2 | – | 3 | 1 | – |
Not specified | 1 | 5 | – | – | 9 | – |
