Revised SF-6D scoring programmes

The SF-6D scoring programmes have recently been revised primarily in order to more accurately deal with missing SF-36/SF-12 item level data. The table below summarises the issues raised by the revisions, the decisions implemented in the scoring programmes and the benefit to the user.

Issue	Decision implemented in revised scoring programmes	Benefit to user
Range checking for raw data	An out of range value is generated for missing dimensions	Clear indication of out of range values
Missing SF-36/SF-12 item values	An SF-6D score can be computed with missing values for an item, if the score on that item would have no influence on the SF-6D score	The SF-6D score can be computed with inessential missing values
Scoring inconsistency	Values on the variable defining the worst state should take precedence over values defining better states	Consistent set of weights used
Weighting of domain scores	Weighting of domain scores from Brazier and Roberts (2004)	Most recent published weights used
Different versions of the SF-36 and SF-12	Different scoring programmes are available for the different SF-36 and SF-12 versions with explicit explanation for which programme is relevant	Ease of use of programmes for all versions of SF-36 and SF-12
Recoding of items in different versions of the SF-36 and SF-12	Each item response requiring random recoding is recoded independently	Independent random assignment of SF-6D dimensions

The changes implemented in the revised SF-6D scoring programmes were agreed by all 3 previous providers of the programmes: John Brazier (Sheffield), Quality Metric and Dennis Fryback and Janel Hanmer (University of Wisconsin-Madison).

Refer to PRO newsletter No. 40 Fall Issue 2008 for details on the potential differences in the computed SF-6D score using the original and revised SF-6D scoring programmes for a dataset with seven patient groups.

New programmes available

A new excel programme is now available from the University of Sheffield to convert SF-36 data into the SF-6D utility score estimated using a set of non-parametric Bayesian preference weights. These nonparametric preference weights are an improvement on the parametric preference weights as the nonparametric model has many advantages over the conventional parametric random effects model which is reflected in improvements in the predictive ability of the model. For further details see Kharroubi et al. (2007).

Furthermore a new excel programme is available to convert SF-36 data into the SF-6D utility score estimated using a set of preference weights obtained using an ordinal valuation technique for a sample of the general population. The estimates using ordinal data represent an alternative value set based on a different valuation technique which produces estimates that are comparable to estimates produced using standard gamble data. For further details see McCabe et al. (2006).

References

Brazier, JE, Roberts, JR,. The estimation of a preference-based index from the SF-12. Medical Care, 2004;42(9):851-859

Brazier, JE, Rowen, D, Hanmer, J,. Revised SF-6D scoring programmes: a summary of improvements. PRO newsletter, 2008;40:14-15

Kharroubi SA, Brazier JE, Roberts J, O´Hagan A. Modelling SF-6D health state preference data using a nonparametric Bayesian method Journal of Health Economics. Journal of Health Economics 2007; 26:597-612

Kharroubi S, O'Hagan A, Brazier J. Estimating utilities from individual health preference data: a nonparametric Bayesian method. Applied Statistics 2005; 54:879-895