- Posts: 75
The PROMIS HAQ is now part of HealthMeasures. Named the PROMIS Short Form v2.0 - Physical Function 24a (PROMIS HAQ), it includes 24 PROMIS items that are analogous to the HAQ-DI. Scores from the original HAQ cannot be compared to PROMIS scores - they use different metrics. However, there is a table that can convert scores from the HAQ-DI to the PROMIS metric on the PROsetta Stone website ( www.prosettastone.org ). You can access a PDF of the measure from Search & View Measures.
In answering a user's question, Jim Fries provided the HealthMeasures team with this description of the PROMIS HAQ:
The HAQ, in improved form, was incorporated into the PROMIS Physical Function item bank. The PROMIS Physical Function SF 20a short form is the successor of the PROMIS HAQ, so the PROMIS Physical Function SF 20a is recommended to be used in instances where the PROMIS HAQ or HAQ-DI would have been used (except in studies in progress where baseline measurements used the PROMIS HAQ or HAQ-DI).
The PROMIS HAQ is different than the PROMIS SF 20a. There are some overlapping items. All HAQ items have the same response options. PROMIS 20a uses two different sets of response options.
The PROMIS Physical Function 20a short form outperforms the HAQ, is supported by a more robust item bank, is easily used in IRT and CAT based situations, develops and maintains translations, and is endorsed and accepted by the American College of Rheumatology www.rheumatology.org/Portals/0/Files/ACR...20RA%20Guideline.pdf .
Yes and no.
Construct validity refers to how well a test measures what it claims to measure --- in the event of the HAQ, improving items by changing scoring and response options but keeping the same content should not affect construct validity.
It is possible that responsiveness to change or internal consistency or convergent validity or some other type of reliability and validity could change when something changes on a PRO (whether that be changing items, response options, scoring, CTT to IRT, etc.), but likely those effects would be small and close to the original values.
Regardless, getting evidence for reliability and validity is never "done". So if you are going to use a measure--any measure--it is good to report at least some psychometric characteristics within your sample, even if it doesn't get a full separate psychometric publication. Maybe a full separate publication is needed, maybe not. But if the data is there, then glean what evidence you can from it.
Perhaps I can restate my follow-up question: When a CTT-based PRO is "converted" to an IRT-based PRO, does the original psychometric properties of the CTT-based PRO automatically apply? For example, if the HAQ-DI has strong construct validity in people with lupus, is it correct to then say that the PROMIS HAQ also has construct validity in lupus, or does that require another validation study that specifically uses the PROMIS HAQ?