The authors report on the development and calibration of item banks for depression, anxiety, and anger as part of the Patient-Reported Outcomes Measurement Information System (PROMIS®). Comprehensive literature searches yielded an initial bank of 1,404 items from 305 instruments. After qualitative item analysis (including focus groups and cognitive interviewing), 168 items (56 for each construct) were written in a first person, past tense format with a 7-day time frame and five response options reflecting frequency. The calibration sample included nearly 15,000 respondents. Final banks of 28, 29, and 29 items were calibrated for depression, anxiety, and anger, respectively, using item response theory. Test information curves showed that the PROMIS item banks provided more information than conventional measures in a range of severity from approximately −1 to +3 standard deviations (with higher scores indicating greater distress). Short forms consisting of seven to eight items provided information comparable to legacy measures containing more items.

This article reports on the development of short forms from the Patient-Reported Outcomes Measurement Information System (PROMIS™) Sleep Disturbance (SD) and Sleep-Related Impairment (SRI) item banks. Results from post-hoc computerized adaptive testing (CAT) simulations, item discrimination parameters, item means, and clinical judgments were used to select the best-performing 8 items for SD and SRI. The final 8-item short forms provided less test information than the corresponding full banks, but correlated strongly with the longer forms. The short forms had greater measurement precision than the Pittsburgh Sleep Quality Index (PSQI) and the Epworth Sleepiness Scale (ESS), as indicated by larger test information values across the continuum of severity, despite having fewer total items—a major advantage for both research and clinical settings.

To develop an archive of self-report questions assessing sleep disturbance and sleep-related impairments (SRI), to develop item banks from this archive, and to validate and calibrate the item banks using classic validation techniques and item response theory analyses in a sample of clinical and community participants. Cross-sectional self-report study. Academic medical center and participant homes. One thousand nine hundred ninety-three adults recruited from an Internet polling sample and 259 adults recruited from medical, psychiatric, and sleep clinics. None This study was part of PROMIS (Patient-Reported Outcomes Information System), a National Institutes of Health Roadmap initiative. Self-report item banks were developed through an iterative process of literature searches, collecting and sorting items, expert content review, qualitative patient research, and pilot testing. Internal consistency, convergent validity, and exploratory and confirmatory factor analysis were examined in the resulting item banks. Factor analyses identified 2 preliminary item banks, sleep disturbance and SRI. Item response theory analyses and expert content review narrowed the item banks to 27 and 16 items, respectively. Validity of the item banks was supported by moderate to high correlations with existing scales and by significant differences in sleep disturbance and SRI scores between participants with and without sleep disorders. The PROMIS sleep disturbance and SRI item banks have excellent measurement properties and may prove to be useful for assessing general aspects of sleep and SRI with various groups of patients and interventions.

Purpose: Patient-reported outcome and experience measures (PROMs/PREMs) are well established in research for many health conditions, but barriers persist for implementing them in routine care. Implementation science (IS) offers a potential way forward, but its application has been limited for PROMs/PREMs. Methods: We compare similarities and differences for widely used IS frameworks and their applicability for implementing PROMs/PREMs through case studies. Three case studies implemented PROMs: (1) pain clinics in Canada; (2) oncology clinics in Australia; and (3) pediatric/adult clinics for chronic conditions in the Netherlands. The fourth case study is planning PREMs implementation in Canadian primary care clinics. We compare case studies on barriers, enablers, implementation strategies, and evaluation. Results: Case studies used IS frameworks to systematize barriers, to develop implementation strategies for clinics, and to evaluate implementation effectiveness. Across case studies, consistent PROM/PREM implementation barriers were technology, uncertainty about how or why to use PROMs/PREMs, and competing demands from established clinical workflows. Enabling factors in clinics were context specific. Implementation support strategies changed during pre-implementation, implementation, and post-implementation stages. Evaluation approaches were inconsistent across case studies, and thus, we present example evaluation metrics specific to PROMs/PREMs. Conclusion: Multilevel IS frameworks are necessary for PROM/PREM implementation given the complexity. In cross-study comparisons, barriers to PROM/PREM implementation were consistent across patient populations and care settings, but enablers were context specific, suggesting the need for tailored implementation strategies based on clinic resources. Theoretically guided studies are needed to clarify how, why, and in what circumstances IS principles lead to successful PROM/PREM integration and sustainability.

A plausible factorial structure for many types of psychological and educational tests exhibits a general factor and one or more group or method factors. This structure can be represented by a bifactor model. The bifactor structure results from the constraint that each item has a nonzero loading on the primary dimension and, at most, one of the group factors. The authors develop estimation procedures for fitting the graded response model when the data follow the bifactor structure. Using maximum marginal likelihood estimation of item parameters, the bifactor restriction leads to a major simplification of the likelihood equations and (a) permits analysis of models with large numbers of group factors, (b) permits conditional dependence within identified subsets of items, and (c) provides more parsimonious factor solutions than an unrestricted full-information item factor analysis in some cases. Analysis of data obtained from 586 chronically mentally ill patients revealed a clear bifactor structure.