Realism and construct validity of novel pelvic models of common gynecologic conditions

To test a novel set of pelvic trainers for realism and construct validity.

Seven models of the female pelvis were studied. Participants performed speculum and bimanual examinations, documented their findings, and recorded opinions of the models in a structured questionnaire. Results were analyzed by participant grade (inexperienced and experienced).

Twenty-six inexperienced and 24 experienced gynecologists participated. Experienced doctors were more likely to correctly identify the uterus corresponding to 16 weeks of gestation (P<0.001), the large fibroid uterus (P=0.01), and uterine anteversion (P=0.04). Identification of the uterus containing a small fibroid, the uterus corresponding to 10 weeks of gestation, and an adnexal cyst was low overall (<35%) and not significantly different between the groups. Correct identification of cervical findings (ectropion and polyp) was high in both groups (65%-88%). Experienced doctors were more consistent-with 17 (71%) reporting the same correct finding on a repeated model, compared with 8 (31%) inexperienced doctors. Forty-nine (98%) doctors completed the structured questionnaire, 36 (73%) of whom felt the models were realistic.

The models were found to be realistic and have construct validity. Senior participants were consistent at correctly identifying most abnormalities. The models may be useful for pelvic examination training; further testing is required regarding their ability to aid learning of clinical and communication skills.