Validity of information on gynecological operations in the Swedish in-patient registry.

In order to validate information held at the Swedish In-patient Registry on oophorectomy and/or hysterectomy procedures, the register codes were compared with data from the medical records in a random sample of 1,338 women. Only 1% of these codes were missing but 5% were erroneous, which in most cases meant that the oophorectomy had been misclassified. The positive predictive value of operation codes was high, ranging from 86 to 100% of the registered events. The actual procedures among women registered with a code for hysterectomy with or without oophorectomy comprised hysterectomy alone in 47% of the women and hysterectomy with a bilateral or unilateral oophorectomy in 30% and 20%, respectively. The reliability of register codes for major gynecological surgical procedures is good, but when the code for hysterectomy is used, medical record data are needed to ascertain the ovarian status. Revised codes are therefore recommended.