Citing a PROMIS Measure
If you have used a specific PROMIS short form or CAT, cite the article describing the development of that domain item bank. You can find it by searching PubMed for “PROMIS development” for the domain of interest (e.g., anxiety, depression, or anger). Suggestions for recommended citations and a list of publications are are on PROMIS Publications. If you want a generic PROMIS citation, use the 2007 Medical Care supplement article.
Measure Development & Research
PROMIS® utilizes rigorous methodology for developing its measures, also known as item banks, and testing their validity.
- Comprehensive literature searches of existing measures, yielding hundreds of potential items, are performed to ensure content validity (i.e., the assurance that each measure represents all facets of a domain)
- To ensure comprehensive coverage of the conceptual area, focus groups are conducted with relevant participants and thematic analyses are performed of the topics discussed. Following the focus groups, an initial item-review process is completed involving elimination of items that are redundant, confusing, or poorly written.
- Cognitive interviews are also performed so that each candidate item is reviewed by multiple individuals with diverse characteristics (i.e., both genders, members of minority groups, participants with modest reading levels) for feedback on the language and clarity of items and the relevance of the content.
- Responses to candidate items are collected from relevant participants, usually via computer administration to both community and clinical samples, during the psychometric testing stage of item bank development. Data from large samples confirm the factor structure of the domains and allow for analyses at the item and bank level. This approach is described by Reeve et al with an update from Hansen et al. Techniques from both classical test theory (CTT) and item response theory (IRT) were used.
- Validity studies are conducted to determine the degree to which an instrument measures what it is intended to measure.
- Translations result from a process of forward and back-translation, multiple expert reviews, harmonization across languages, and cognitive debriefing with a sample of native speakers of the target language (linguistic validation). A universal approach to translation ensures that, whenever possible, one language version is created for multiple countries instead of country-specific versions of the same language.
Read about the PROMIS Instrument Development and Validation Scientific Standards.
For additional information on PROMIS instrument development, see a presentation on PROMIS Instrument Development and Psychometric Evaluation Scientific Standards. This presentation describes a set of standards that serve as the scientific foundation and guidance for the development and evaluation of PROMIS item banks and instruments.
Also, the PROMIS Instruments Maturity Model provides information to assist developers in meeting the scientific standard criteria, from item pool or scale development to fully validated instruments ready for use in clinical research and practice.
Interested in modifying a PROMIS item or measure? Read our guidance here. Learn more>>
Recent PROMIS research focuses on
- Validation of measures such as Alcohol Use, Family Relationships, PROMIS-29 Profile, and PROMIS Global Scale.
- Validation of Pediatric and Parent-Proxy measures in children with 11 chronic conditions. Learn more>>
- Development of new item banks (e.g., urinary incontinence, itch) using PROMIS methods
- Use of PROMIS methodology to develop item pools for mental health disorders
- Use of PROMIS item banks in specific populations such as sickle cell disease, people with special healthcare needs, glomerular disease, and post-acute care
- Identify symptom trajectories experienced by patients with specific conditions including survivors of childhood brain tumors and patients undergoing ankle fusion
- New data collection platforms that enable use of PROMIS measures
A two part series published in Psychological Test and Assessment Modeling 2016 v.1 and v.2 provide strong evidence supporting the measurement equivalence of the PROMIS short form measures in ethnically, socio-demographically diverse groups. This is a step towards meeting the international call for further study of their performance in such groups. Learn more>>
PROMIS research is also active outside of the United States (see PROMIS International).
PROMIS is Expanding Instruments to Include Early Childhood
As a part of the National Institutes of Health Environmental influences on Child Health Outcomes (ECHO) research program, the Person Reported Outcomes Core at Northwestern University is extending certain PROMIS domains to include early childhood (ages 1-5 years) parent-report instruments. Learn more about ECHO>>
Does Recall Period Matter?
PROMIS Physical Function measures do not include a recall period. PROMIS investigators decided to focus items on patient report of current capability rather than recall of function over a specified time frame. This was done to make it possible to answer items that refer to functions that might not have been actually conducted during the recall period. In 2020, Condon and colleagues tested the wisdom of this approach. They compared the original “no recall” items with identical item stems, but using 24-hour and 7-day recall periods. They report “Compared to no recall, the use of a recall period has little to no effect upon PROMIS physical function responses or scores” (p. 745). Consequently, this supports the original “no recall” condition. Use of a recall period is not recommended for physical function.
Cella and colleagues (2019) published a comprehensive validation of the popular 7-domain PROMIS health profiles: PROMIS-29; PROMIS-43; and PROMIS-57. This is the first paper reporting on the development, reliability, and validity of these profiles. Results provide relative precision and correlation of 4-, 6-, and 8-item short forms with their parent banks. Readers can learn estimated power of each short form to detect small effects (d=0.2) for general population and clinical samples, and they can evaluate relative validity of each short form with regard to differentiating known groups.