PROMIS-25 Reliability and Validity Among Children Living with Burn Injury: A Burn Model System National Database Study

Abstract

This study examined the reliability and validity of the Patient Reported Outcomes Measurement System (PROMIS)-25, a profile instrument consisting of four-item fixed short forms for six health domains, in children living with burn injury. Data were provided by children participating in a multi-center longitudinal study of outcomes after burn injury. Floor and ceiling effects, unidimensionality, internal consistency, reliability, and differential item functioning (DIF) of the PROMIS-25 Profile v.2.0 were examined. Correlations with other established measures were calculated to assess concurrent validity. Children (n = 256) between the ages of 8-18 years with moderate to severe injury provided responses on PROMIS-25 domains. All PROMIS-25 domains showed high internal consistency. Substantial portions of the sample reported no symptoms (anxiety [58.2%], depressive symptoms [54.6%], fatigue [50.8%], pain [60.1%]). There was a large ceiling effect on peer relationships (46.8%) and physical function mobility (57.5%). One-factor confirmatory factor analyses supported unidimensionality for all domains. Reliability was sufficient for group mean comparisons (>0.8) across at least some trait levels for most domains except fatigue and anxiety. No DIF with respect to burn status was detected when comparing the burn sample to the PROMIS pediatric general U.S. population testing sample. These results provide evidence of reliability and validity of PROMIS-25 scores among children living with burn injury. Reliability of domains was low to moderate and would likely be improved, and ceiling effects reduced for some domains, by administering the PROMIS-37, which includes six items per domain.

Figure 1.

Reliability of PROMIS-25 short form scores plotted in conjunction with a histogram of the study sample scores on the T-metric. Solid lines show reliability of the PROMIS-25 four-item short forms along the T-scale metric (reference population mean=50; SD=10). Scores with reliability above 0.8 are considered adequate for group comparisons. The histogram represents the frequency of each score in the study sample.

Figure 2.

Reliability of mobility and peer relationship short forms from the PROMIS-25, −37, and −49 profiles in addition to the full item bank. Scores with reliability above 0.8 or 0.9 are considered adequate for group and individual comparisons, respectively.