 |
 |
Mike Addison has raised the old chestnut of significant figures, decimal places and EQA. Though each centre has to exercise judgement as to what is appropriate for each analyte and scheme, UK NEQAS Birmingham policy is to give mean values to one more significant figure/decimal place than is appropriate for individual laboratory results. This does not appear excessive, and if we don't print mean values to sufficient places we get complaints that we've calculated percentage bias from the target value (which is stored with many significant figures) incorrectly! In some cases where the analyte has a wide dynamic range (eg urinary catecholamines) we use the flexibility in UK NEQAS software to work on significant figures rather than a fixed number of decimal places. For TSH we use decimal places appropriate to the concentration range for each specimen (eg 3 for values below 1 mU/L, down to no decimal places above 100 mU/L). Unfortunately, however, many laboratories' computer systems (and personnel?) still seem to suffer from 'fixed decimal place syndrome' and our use of significant figures has been criticised. This appears to be yet another case of being unable to satisfy all of the people all of the time . . . David Bullock > I quite agree with you about reporting oestradiol to 3 significant > figures. But this is not limited to endocrinology and occurs in all > branches of clinical biochemistry. It is not helped by NEQAS giving means > to several places of decimals (I exaggerate only slightly) When I bring > this subject up I get told it is a computer problem and that all results > are reported to the same number of figures. It is about time we told > software suppliers what we want and not just accept what we get. Dr David Bullock Director, Wolfson EQA Laboratory P O Box 3909, Birmingham B15 2UE, U K FAX: 0121 414 1179 [+44 121 414 1179] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%