The Functional Rating Index: Twenty Years of Invalid Measurement.

Abstract

The 2001 Functional Rating Index (FRI) was not developed under today's standard psychometric analysis. The original data of 108 cases were re-analyzed using Rasch item response theory. In 2015, 2 alternative forms were administered to an additional 140 patients for establishing and perhaps improving its psychometric characteristics. To evaluate the FRI with item response theory. The 2001 FRI data showed internal agreement among items and weak item-total correlation items. The FRI's true reliability and validity were never established. The original 2001 FRI 108 and two new versions with 140 respondents with back pain were compared by Rasch analysis for unidimensionality, local independence, monotonicity, and differential item functioning. All three versions exhibited more than the requisite single latent construct. The original Form 2001 had two items that were non-monotonic, four with differential item functioning (DIF), five with poor infit, and four with poor outfit. Form B had five nonmonotonic items, five had DIF, three had poor infit, and three had poor outfit. Form C had only monotonic items, one item with DIF, three items with poor infit, and two with poor outfit. The original FRI and alternative forms all fail failed crucial psychometric tests and fail to accurately measure more than one latent construct. It is thus unfit as a pain, function, and disability assessment. Only reducing the number of Likert choices improved the test. Other back pain assessments should be used instead, and all surveys would benefit from periodic item responses to adjust to shifts in grammar and meaning.Level of Evidence: 3.