Are decisions using cost-utility analyses robust to choice of SF-36/SF-12 preference-based algorithm?

Table 1 Summary of SF-12/SF-36 preference-based algorithms

	Theoretical Range*
Algorithm	Minimum	Maximum	Original source of Preferences	Source of value (country)	Source of sample (country)	Sample Size
Brazier (SF-12)	0.35	1.00	1^st generation – SG	UK	UK	836
Lundberg (SF-12)	0.27	0.97	1^st generation – VAS	Sweden	Sweden	4,180
Franks (SF-12)	-0.24	0.92	2^nd generation – EQ-5D	UK	US	240
Franks (SF-12)	-0.09	0.96	2^nd generation – HUI3	Canada	US	240
Franks (SF-12)	-0.07	0.98	2^nd generation – EQ-5D	UK	US	12,998
Lawrence (SF-12)	0.15	1.01	2^nd generation – EQ-5D	UK	US	14,580
Shmueli (SF-36)	0.23	1.00	1^st generation – VAS	Israel	Israel	2,505
Brazier (SF-36)	0.30	1.00	1^st generation – SG	UK	UK	836
Fryback (SF-36)	0.59	0.84	2^nd generation – QWB	US	US	1,356
Nichol (SF-36)	0.24	1.05	2^nd generation – HUI2	Canada	US	6,921

*Maximum and minimum scores are based on best and worst responses to all items on the SF-36 and SF-12. For the Lundberg algorithm, minimum obtained is based on male, ≥ 80 years of age, while maximum is based on female, <30 years of age. For the Nichol algorithm, the minimum is based on 100 years of age, while maximum is based on 0 years of age.

ISSN: 1477-7525