Incidence and decisions

A lot of medicine involves making decisions. Decision theory is informed by the probability of outcomes given specific information, and the costs associated with those outcomes.

Most people aren't good at probability (and nobody's really that good at it), but without some indication of probabilities, you can't be expected to make good decisions.

So. Given endometrial hyperplasia (thickening of the wall of the uterus) with no particularly visible tumors, how probable is cancer? I Google "incidence of carcinoma given endometrial hyperplasia" and find this: Incidence of endometrial carcinoma in patients with endometrial hyperplasia. (PubMed is our friend.)

Answer: 0.81%. Not a lot, but enough to take seriously. The abstract in full:

OBJECTIVE:

The purpose of this retrospective study was to establish the risk of developing endometrial adenocarcinoma in patients diagnosed with endometrial hyperplasia.

MATERIAL AND METHODS:

The incidence of endometrial hyperplasia and its relation with endometrial adenocarcinoma was evaluated in 1,139 patients who presented with abnormal bleeding between January 2000 and December 2004; D&C was performed in all cases. There were 591 (51.88%) cases of simple endometrial hyperplasia, out of which 110 (18.61% from 51.88%) cases had atypia, 60 (5.26%) cases of complex hyperplasia, out of which 19 (31.66% from 5.26%) had atypia, and the remaining 488 (42.84%) had different forms of mixed hyperplasia.

RESULTS:

The incidence of endometrial adenocarcinoma was 3.87% in atypical hyperplasia and 0.81% in other forms, and was related only to cases with atypia in which the incidence was 0.61%.

CONCLUSIONS:

The most indicated measure to prevent endometrial carcinoma in cases with complex endometria hyperplasia with atypia is hysterectomy, while for other forms of hyperplasia, hormonal treatment is used but only under strict control.

Not sure I agree with the conclusions, but I love those retrospective studies.