Letter: TSH level and thyroid function.

SIR,-Dr. E. G. M. D'Haene and others (21 September. p. 708) have compared the free thyroxine index (F.T.I.) and effective thyroxine ratio (E.T.R.) using correlation coefficients (Spearman rank correlation) and a minimum of experimental evidence. We fec 1 that a number of noints and the final conclusions may mislead potential users of one-tube free thyroxine tests. (1) The correlation coefficient, however, derived, between F.T.I. and E.T.R. does not reflect the ability of these tests to sort out the patients into the three categories, hypohyper-, and euthvroid. Thus the fact that there is no sipnificant correlation between F.T.I. and E.T.R. values falling insi:de the normal range is not relevant in any way to the final diagnosis. (2) While the use of comparison of "objective measurements" with "subjective clinical impressions" is certainly dangerous and difficult in practice, it is nevertheless essential for tests to be judged at some stage by their practical usefulness. Thus the dbservations of Thorson et al.,' Murray et al.,2 and ourselves (in an unpublished retrospective survey) that the diagnostic accuracy of the E.T.R. was in practice about 99%, 94 5%, and 94%/. retrospectively are more relevant than the judgement of a test by its correlation with another, different, "objective measurement." (3) The statement that "correction for binding-protein abnormalities . .. when the free T4 index is calculated, is more trustworthy than the chemically regulated correction which is more or less automatically and uncontrollably applied when measuring the E.T.R." (our italics) is based on a claim that at high ooncentrations of thyroxine (T4) a greater percentage of labelled TA binds to the thvroxine -binding proteins than is found if triiodothyronine (T-3) is used in-stead. A reduction of the TA concentration results in binding "more or less identical" with that of T-3. They appear to have overlooked two points: (a) The non-labelled TA acts similarly to labelled TA and so their addition of 0-5 ml normal serum to each 3-ml phial (instead of 05 n/100 mi in the E.T.R.) will increase the final concentration of TA fromn 1-03 to about 15 8 nmol/l. Even after their reduction of labelled T-4 to 0-21 nnmol/l. their total concentration is about 14 9 nmmol/l., while the E.T.R. no¶mally contains about 3 5 nmol T4/1. (b) The equivalent oDncentration of T-3 used in the experimnents is about one-fifth that of TA4 and so their comparisons between the binding of these substances to thyroxine-binding globulin were not valid. Dr. D'Haene and his colleagues conclude that "at present there are no convincing reasons" for using E.T.R. instead of F.T.I. The time required to carry out both T-3 resin uptake and total TA estimation and their cost are about do-uble those of singletube tests such as E.T.R. It would seem to us that a reduction in cost and time of trained personnel would be sufficient rea.sons for preferring a single test procedure of comparable diagnostic reliability. Add to this a smaller coefficient of variation (even found by Dr. D'Haene and his colleagues) and a roved value for both diagnosis and monitoring treatment and it would need a oonvincing reason indeed to use F.T.I. in-stead of E.T.R. as an index of thyroid function.-We are, etc., M. G. WELLS C. E. ANDREW Radioche-nistry Department, Edgwa-e General Hospital, Edgware, Middlesex