Varying the item format improved the range of measurement in patient-reported outcome measures assessing physical function

Table 3 PROMIS PF full bank and short form scoring characteristics and agreement with simulated “true” scores

General health groups	True PF T-score^a mean (SD)	Full bank (124 items)^b			Format A^e (5-item short form)			Format B (5-item short form)			Format C (5-item short form)
General health groups	True PF T-score^a mean (SD)	T-score mean (SD)	RMSE^c	Percentage floor/percentage ceiling^d	T-score mean (SD)	RMSE	Percentage floor/percentage ceiling	T-score mean (SD)	RMSE	Percentage floor/percentage ceiling	T-score mean (SD)	RMSE	Percentage floor/percentage ceiling
Poor	35.6 (6.5)	35.7 (6.4)	0.7	0.0/0.0	36.6 (6.3)	3.0	3.9/0.2	36.4 (6.3)	2.7	1.5/0.2	36.3 (6.3)	2.7	0.5/0.0
Fair	41.9 (7.6)	41.9 (7.7)	0.8	0.0/0.0	42.5 (7.9)	2.9	1.3/3.8	42.3 (790)	2.6	0.5/4.6	42.3 (7.7)	2.5	0.2/0.7
Good	48.9 (7.8)	49.0 (7.9)	1.1	0.0/0.1	49.4 (8.0)	3.2	0.1/17.0	49.4 (8.0)	3.1	0.0/21.5	49.1 (7.9)	2.6	0.0/6.2
Very good	54.4 (7.2)	54.5 (7.3)	1.5	0.0/0.2	54.5 (6.9)	3.8	0.0/37.9	54.5 (6.8)	3.7	0.0/45.4	54.4 (7.3)	2.9	0.0/16.8
Excellent	58.8 (6.5)	58.7 (6.4)	1.9	0.0/0.7	57.8 (5.3)	4.3	0.0/59.0	57.8 (5.0)	4.4	0.0/67.1	58.4 (6.3)	3.3	0.0/32.6
Full sample	47.9 (11.0)	48.0 (11.0)	1.3	0.0/0.2	48.1 (10.4)	3.5	1.1/23.6	48.1 (10.5)	3.4	0.4/27.8	48.1 (10.5)	2.8	0.1/11.3

^a T-scores have a mean of 50 and standard deviation of 10 in the analytic sample. ^bFinal Patient Reported Outcomes Measurement Information System Physical Function (PROMIS PF) item bank version 1.0. ^cRSME = root mean square error between estimated T-scores and simulated “true” T-scores. ^dPercentage of the simulated sample who reached the lowest (“floor”) or highest (“ceiling”) possible score. ^eFormat A: “Are you able to …” (five-category response scale from “Without any difficulty” to “Unable to do”); Format B: “Does your health now limit you in …” (five-category response scale from “Not at all” to “Cannot do”); format C: “How difficult is it for you to…” (six-category response scale from “Very easy” to “Impossible”)

ISSN: 1478-6362