The term 'protein-bound iodine' (PBI) is used to describe that fraction of blood iodine which is nondialysable and precipitates with serum or plasma proteins (Trevorrow, 1939). In the absence of structural damage to, or carcinoma of, the thyroid, about 75-100% of serum PBI consists of L-* Part 8: Birkinshaw & Chaplen (1955).

The effects of physiological and environmental factors, of thyroid and non-thyroid diseases, and of drugs on the serum protein-bound iodine are described and discussed.

No abstract

SYNOPSISThe assessment of accuracy and precision in a routine clinical chemistry laboratory is discussed. A figure and tables are presented which relate the chance of making incorrect clinical decisions to the precision of an analytical method.The question of accuracy and precision in clinical chemistry is frequently discussed (see, for instance, Thompson and Jones, 1965;Mitchell, 1966). It is a waste of time in routine practice to strive for a degree of accuracy or precision which is greater than that required by clinicians. The object of this communication is to enable the performance of a laboratory method to be judged quantitatively in relation to clinical needs.'Accuracy' is the extent to which a method measures what it sets out to measure. Determinations by an accurate method will not show systematic error, i.e., they will not differ consistently and in the same direction from the result which should be obtained. 'Precision' is the reproducibility of a method. A precise method has a small random error, i.e., there is little variation in results when the sample is run repeatedly. A method may be fairly precise but inaccurate, e.g., the determination of blood 'sugar' by the method of Folin and Wu.In practice, systematic error may result either from the use ofinaccurate methods or from technical faults, such as the deterioration of standards or two types (Fig. 1): first, method error may cause the analytical result from a normal sample to fall outside the normal range; secondly, it may cause the analytical result from an abnormal sample to fall within the normal range. The chance of making either of these mistakes depends (1) on the size of the method error compared with that of the normal clinical range and (2) on the difference between the values of the normal limits and the true analytical value of the sample. Mistakes are most likely to occur when the true sample value is close to (either within or outside) the limits. As the true sample value moves away from the limits, the chance ofmistakes decreases at a rate which depends on the precision of the method. The greater the precision, the more rapidly the chance of mistakes decreases on either side of the limits.The standards of precision achieved, or aimed at, in a routine clinical laboratory are thus directly related to the proportion of clinical mistakes which would be expected to occur within areas of uncertainty, defined in terms of the normal ranges for particular constituents. STATISTICAL METHODIn the account which follows, al is the standard deviation ofrepeated determinations on a singlesample and measures the precision of the method, whereas a, is the standard deviation of accurately determined values in a population of healthy subjects, and is related to the normal clinical range. The use of these standard deviations in this context impliesthat both the method errors and the accurate values for healthy subjects are distributed according to the Gaussian (= 'normal' in the statistical sense) distribution.Small deviations from the Gaussian distri...

No abstract