Rehabilitation and outcome measurement: where is Rasch analysis-going?

Abstract

Outcomes are clinical or functional changes at the whole-person level. Measures are based on questionnaires which summate scores assigned to a series of items representing a person's variable. The metric properties of raw scores are known to be of limited validity. Yet, such variables represent the main target of outcome assessment in Rehabilitation Medicine as far as they may cover performances, perceptions and knowledge. Rasch statistical models, first available in English to a large readesrship in late 70s, allow to transform the arbitary raw scores into true linear measures. Rehabilitation seemed a privileged field for the application of Rasch models, yet these are still far from becoming popular. A bibliometric study was conducted across ten leading digital libraries. A MEDLINE search showed a dramatic increment of published papers covering the intersection between the MESH terms ''Rehabilitation'' and ''Outcome assessment'', which rose in from 5 to 4 302 between the 1981-90 and 2001-07 time frames, respectively. ''Rasch'' paper also rose remarkably from 35 to 539. When the triple intersection was considered, articles only rose from 0 to 12. Results point towards a privileged application of Rasch analysis to build and refine questionnaires, rather than to actually measuring people. Commentaries and suggestions are invited.