The experience of depression, anxiety, and mania among perinatal women.

We assessed differential item functioning (DIF) based on computerized adaptive testing (CAT) to examine how perinatal mood disorders differ from adult psychiatric disorders. The CAT-Mental Health (CAT-MH) was administered to 1614 adult psychiatric outpatients and 419 perinatal women with IRB approval. We examined individual item-level differences using logistic regression and overall score differences by scoring the perinatal data using the original bifactor model calibration based on the psychiatric sample data and a new bifactor model calibration based on the perinatal data and computing their correlation. To examine convergent validity, we computed correlations of the CAT-MH with contemporaneously administered Edinburgh Postnatal Depression Scales (EPDS). The rate of major depression in the perinatal sample was 13 %. Rates of anxiety, mania, and suicide risk were 5, 6, and 0.4 %, respectively. One of 66 depression items, one of 69 anxiety items, and 15 of 53 mania items exhibited DIF (i.e., failure to discriminate between high and low levels of the disorder) in the perinatal sample based on the psychiatric sample calibration. Removal of these items resulted in correlations of the original and perinatal calibrations of r = 0.983 for depression, r = 0.986 for anxiety, and r = 0.932 for mania. The 91.3 % of cases were concordantly categorized as either "at-risk" or "low-risk" between the EPDS and the perinatal calibration of the CAT-MH. There was little evidence of DIF for depression and anxiety symptoms in perinatal women. This was not true for mania. Now calibrated for perinatal women, the CAT-MH can be evaluated for longitudinal symptom monitoring.