Objectives-Recent work using classical test theory (CTT) and item response theory (IRT) has found that the self-report ) and clinician-rated (QIDS-C 16 ) versions of the 16-item Quick Inventory of Depressive Symptomatology were generally comparable in outpatients with nonpsychotic major depressive disorder (MDD). This report extends this comparison to a less welleducated, more treatment-resistant sample that included more ethnic/racial minorities using IRT and selected classical test analyses. 16 and QIDS-C 16 were obtained in a sample of 441 outpatients with nonpsychotic MDD seen in the public sector in the Texas Medication Algorithm Project (TMAP). The Samejima graded response IRT model was used to compare the QIDS-SR 16 and QIDS-C 16 . 16 and QIDS-C 16 related well to overall depression. The slopes of the item response functions a), which index the strength of relationship between overall depression and each symptom, were extremely similar with the two measures. Likewise, the CTT and IRT indices of symptom frequency (item means and locations of the item response functions, b i ) were also similar with these two measures. For example, sad mood and difficulty with concentration/decision making were highly related to the overall depression severity with both the QIDS-C 16 and QIDS-SR 16 . Likewise, sleeping difficulties were commonly reported, even though they were not as strongly related to overall magnitude of depression.
Methods-The QIDS-SR
Results-The nine symptom domains in the QIDS-SR
Conclusion-In this less educated, socially disadvantaged sample, differences between the QIDS-C 16 and QIDS-SR 16 were minor. The QIDS-SR 16 is a satisfactory substitute for the more timeconsuming QIDS-C 16 in a broad range of adult, nonpsychotic, depressed outpatients.
KeywordsQuick Inventory of Depressive Symptomatology; Inventory of Depressive Symptomatology; Item response theory; Samejima graded response model; depressive symptoms
OBJECTIVESThe accurate, rapid, and cost-efficient measurement of depressive symptoms serves both clinical and research purposes. Clinicians can gauge the benefit of treatment and make timely adjustments in the treatment plan. Research, on the other hand, can be made less costly if such measures are available. The Quick Inventory of Depressive Symptomatology (QIDS) is a 16-item scale that measures each of the nine symptom domains that define a major depressive , which defines depression as a latent trait, were employed. The particular IRT model we employed was developed by Samejima (1969Samejima ( , 1997 to examine graded responses. The recent study (Rush et al. in press) focused on a subset of patients with nonpsychotic major depressive disorder (MDD) derived from the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) trial (Fava et al. 2003;Rush et al. 2004a). These patients were reasonably well educated as a group and were drawn from both primary and psychiatric care settings. They were selected to not be treatment resistant. Results were generally comparable between...