Page 79 - The diagnostic work-up of women with postmenopausal bleeding

Table 1. Patient characteristics and missing values in validation databases
N 433 559
Age, years; mean (SD)
Diabetes mellitus; n (%)
Hypertension; n (%)
Anticoagulants; n (%)
BMI, kg/m2; mean (SD)
Time since menopause, years; median (IQR)
Nulliparity; n (%)
Endometrial thickness, mm; median (IQR)
Endometrial (pre-) cancer; n (%) Atypical hyperplasia Endometrial cancer
67.4 +/- 11.8 66 (15.2) 176 (40.6) 88 (31.9) 27.8 (6.4) 16 (5-26)
45 (10.6) 6.0 (3.2-13.0)
65 (15.0) 3 (0.69) 62 (14.3)
0
0
0
157 (13.2) 24 (5.5)
6 (1.4)
8 (1.8) 0
0
61.8 +/- 9.95 72 (12.9) 196 (35.1) 94 (16.8) 30.2 (8.2) 5 (2-14)
60 (13.5) 5.7 (2.5-10.0)
57 (10.2) 7 (1.3) 50 (8.9)
0 <0.01
1 (0.2) 0.29 4
0 0.07
0 <0.01 260 (43.4) <0.01 124 (22.2) <0.01
114 (20.4) 0.19 0 0.02
0 0.02
External validation of a prediction model
lower BMI, thicker endometrium, a higher percentage of endometrial cancer and more Swedish women used anticoagulant therapy. For both validation databases, efforts were made to collect data on endometrial cancer in the women with an endometrial thickness below the applied threshold and in none of these women endometrial cancer was diagnosed.
Swedish database
Missing; n (%)
Dutch database
Missing; n (%)
p-value
BMI, body mass index; NA, not applicable.
Calibration
Plots that express the calibration of the two models developed by Opmeer (‘patient characteristics’ model and the ‘patient characteristics and TVS’ model) in the Dutch and Swedish database are presented in Figure 1. In the Dutch database, the calibration slope was better for the ‘patient characteristics and TVS’ model than for the ‘patient characteristics’ model.The predicted probabilities of cancer when using the ‘patient characteristics and TVS’ model were close to the observed frequency of cancer over the whole range of predicted risks. For the lowest risk group, calibration was consistently good across the different models and databases, and some major over and underestimated risk in patients at increased risk for endometrial cancer.
77