Iron in flour.

  • Great Britain. Ministry of Health
Date:
1968
Catalogue details

Licence: Public Domain Mark

Credit: Iron in flour. Source: Wellcome Collection.

    16/78 page 6
    Estimations of absorption In this report the ‘absorption’ of an iron preparation represents the estimation of the proportion (expressed as a percentage) of an oral dose which has been absorbed into the body. The statistical uncertainties of individual estimations of absorption cannot easily be determined, so each has been used as an original variate. Estimates of confidence, such as standard errors of mean absorptions, have been based simply on the variation in individual absorptions, and have ignored the statistical errors in the absorptions themselves. It must be appreciated that because of this the tests of statistical significance are not strictly valid. It is unfortunate that more data are not available, but in work such as that presented here small numbers are inevitable and conclusions have often to be based on reasonable inferences without the support of tests of statistical significance. In fact, although pairs of estimations of absorptions are available for 75 subjects, certain subgroups are represented by very few subjects. Transformation Estimations of absorptions in different subjects showed great variation. The use of means of estimates in small subgroups of subjects is likely therefore to be misleading and tests of significance involving such means invalid. In an attempt to increase the stability of means the original data have been transformed to the angle whose sine is the square root of the percentage (Snedecor, 1956). The mean angles obtained in this way have then been transformed back and expressed as mean percentage absorptions. In almost every subgroup used in the analyses of these data a very marked reduction in the variance proportionate to the mean followed the use of this transformation. However, the original data are presented in full (Tables VI-XHII, Appendix 1) and most of the mean absorptions are also presented without transformation (Table III, p. 12). Tests of statistical significance The term ‘significant’ is only used in this section of the report in a statistical context. Tests of significance have only been done on means obtained by trans- formation of the original data as described above. For convenience and simplicity the conventional level of statistical significance P < 0:05 is used. However, the limitations mentioned above must be appreciated throughout. Because of the transformation used, standard errors are not quoted after the mean estimates, but where appropriate the result of a test of significance on the transposed data is indicated. General plan of the studies Experiments were designed which were as ‘realistic’ as possible. Bread was prepared in a normal commercial way (p. 32), and given to subjects, many of whom were haematologically ‘normal’ (p. 21), as part of a breakfast which was similar to that eaten by many people in this country (p. 8). A small pilot experiment was done followed by a series of five main experi- ments. In each of the main experiments the subjects were given iron baked into bread as part of the breakfast. Each subject received 0-7—1-0 mg of iron in each of two iron preparations, one of which was ferrous sulphate as a ‘control’. One preparation had been labelled with 1-0-5-0 wCi of Fe®*, the other with 15-25 wCi of Fe°*. The three subjects in the pilot experiment received no ferrous sulphate.
    digitised image 1
    digitised image 2
    digitised image 3