RASCH MEASUREMENT AND HTA

RASCH MEASUREMENT AND LATENT ATTRIBUTES

Alongside the consistent endorsement of measurement inversion, HTA knowledge-base
interrogations reveal an equally striking absence of interest in Rasch measurement. Without
exception, Rasch receives weak endorsement despite its status as the necessary and sufficient
framework for transforming subjective observations into lawful measures of latent attributes.
This omission is particularly unfortunate for two reasons. First, Rasch demonstrates how patient-
reported outcome instruments can be developed to satisfy the requirements of representational
measurement. Second, it would have prevented the persistent assumption within HTA that
ordinal scores can be treated as though they were interval or ratio measures, a misconception that
lies at the heart of arithmetic chaos.

Any attempt to reconstruct HTA following the closure of the reference case must therefore place
Rasch measurement at its center. For latent attributes there is no alternative measurement
framework. Given the importance attached in HTA to constructs such as symptom burden,
treatment satisfaction, need fulfilment and quality of life, the neglect of Rasch is particularly
striking. Rather than creating ordinal utilities and then the mathematically impossible QALY,

Rasch measurement focuses on the possession of a latent attribute by individuals and patient
groups, providing a lawful basis for evaluating the quantitative impact of therapy interventions.
The lack of understanding and often the complete absence of awareness of Rasch measurement,
first proposed in 1960, remains a major obstacle to HTA reconstruction. Virtually all patient-
reported outcome instruments currently employed in HTA fail to meet Rasch standards. Indeed,
Rasch requirements are absent from the evaluation frameworks promoted by COSMIN and the
Cochrane Collaboration. For this reason, Rasch measurement occupies a central place in the 9-
unit reconstruction program. If HTA is to move beyond ordinal scores, utilities and QALYs, the
development and application of Rasch logit ratio measures for latent attributes will be
indispensable.

REFERENCES

Langley P. UNITED STATES: COSMIN AND MEASUREMENT INVERSION – THE
CERTIFICATION OF NON-MEASURES, Logit Working Paper No 237

Langley P. UNITED STATES: COCHRANE AND MEASUREMENT INVERSION – THE
AGGREGATION OF NON-MEASURES, Logit Working Paper No 238