Page 82 - The diagnostic work-up of women with postmenopausal bleeding
P. 82
Chapter 4
Discriminative performance of the prediction models
Figure 2 shows ROC curves for the two models compared to endometrial thickness as measured byTVS in the two validation datasets. In both the Dutch and the Swedish databases, the AUC for the ‘patient characteristics and TVS’ model (respectively 0.89 (95% CI 0.86 to 0.92) and 0.89 (95% CI 0.86 to 0.91)) was higher than the AUC for the ‘patients characteristics’ model (0.71 (95% CI 0.65 to 0.76) and 0.69 (95% CI 0.64 to 0.73)). The AUC for the ‘patient characteristics and TVS’ model was similar to the AUC for endometrial thickness only: 0.87 (95% CI 0.83 to 0.90) in the Dutch database and 0.90 (95% CI 0.88 to 0.93) in the Swedish database.
Clinical consequence of the three strategies
The estimated clinical consequences of applying the three strategies are reported in Tables 2 and 3.With all three strategies, all cases of endometrial (pre) cancers would be detected in both databases.This is under the assumption that among women with a thin endometrium (below the threshold of 4 or 4.4 mm) was diagnosed with endometrial cancer. This means that with the ‘patient characteristics’ strategy you could skip the measurement of endometrial thickness safely, but then you would have to perform an invasive procedure to get histology in 93% of women in both databases, compared to only 61-63% (respectively in the Dutch and Swedish database) when patients would be selected based on the measurement of endometrial thickness (which is current clinical practice).
When using the patient characteristics in a ‘sequential’ strategy, one could save 7% of women an ultrasound and these could be reassured. In the remaining group of women endometrial thickness has to be measured and then 57-58% would have to undergo an invasive procedure.Thus, this strategy would save 7% of women an ultrasound and in 3-6 % less women an invasive procedure has to be done compared to the current clinical practice:TVS only.
When using the patient characteristics in an ‘integrated’ strategy, all women would need a TVS and the amount of women that would need further invasive procedures to retrieve histology would be the same as in the ‘sequential’ strategy.
80