Background The Patient Reported Outcomes Measurement Information System (PROMIS) initiative uses a mixed methods approach, including item response theory to develop patient reported outcome measures (PROs) for self-assessment of physical, mental, and social health across disease conditions. Item level calibration allows for computer adaptive testing or use of brief “short forms” to reduce respondent burden. PROs are used increasingly in health care. A framework to interpret scores according to degree of clinical severity of a health condition would enhance their practical use. Use of PROs in evaluation of treatment effectiveness over time requires establishment of minimal important differences (MID) in change scores.
Objectives To estimate clinically relevant cut-points for scores (none–mild-moderate - severe) and MIDs for PROMIS pediatric measures we previously examined in a validation study in a JIA cohort.
Methods We identified clinical severity thresholds and MID for measures of mobility, upper extremity (UE) function, fatigue, and pain interference working with JIA patients, parents of JIA patients, and clinicians using standard setting methodology modified from educational testing. Data from PROMIS item bank longitudinal validation collected on 121 JIA patients was used to develop clinical vignettes across a range of symptom severity. Vignettes were created based on most likely item responses at different levels on the T score metric [mean =50; SD =10]. Vignettes were anchored at 5-point intervals (0.5 SDs). Parents, patients, and clinicians participated in separate one-day meetings. Vignettes were ordered and placed on cards. Panelists identified adjacent vignettes considered to represent upper and lower boundaries separating category cut points (i.e., none /mild problems, mild/moderate, moderate/severe). Cut scores were defined as mean score for boundary vignettes. To define MIDs, panelists responded to items to represent “just enough improvement to make a difference”. Average change scores served as estimates of MID.
Results For pain interference, mobility, and UE, patients set higher cut points for severity than parents, typically by 0.5 SD. For function measures cut points were concordant between patients and clinicians for moderate and severe categories, but clinicians saw mild disability where patients reported none. For fatigue and pain, clinicians rated cutoffs for severe were 0.5 SD below patient ratings. Size of MID varied according to severity classification of the symptom. Clinicians consistently estimated MIDs smaller than parents. Patient MID estimates varied in relation but tended to be between clinicians and parents estimates. MIDs estimated by the panelists were typically larger than the MIDs determined using statistical methods.
Conclusions We used a modified educational standard setting method to estimate clinically relevant cut points to classify severity for PROMIS measures of mobility, UE, fatigue, and pain interference. Parallel exercises identified these cut points from the perspectives of JIA patients, parents and clinicians. We explored a novel means of determining MID from the patient/parent/clinician perspective for meaningful interpretation of PROMIS measures in a clinical setting.
Cook KF, Qual Life Res. 2014 Aug 23. Cella D, Qual Life Res. 2014 Dec;23(10):2651-61.
Disclosure of Interest None declared