Key points are not available for this paper at this time.
The definition of anaemia in most published work appears to depend upon the demonstration of either a haemoglobin level or a mean corpuscular haemo- globin concentration estimation, or both, below levels chosen in a largely arbitrary fashion with little reference to the distributions of these variates in the population. Pryce (1960) stressed that "normal" haematolo- gical values derived from series of selected subjects were likely to be misleading, particularly if the issue was pre-judged by the use of subjects considered to be "idea normals". Likewise a true epidemiological picture of anaemia can be based only on data from a sample which is truly representative of the community from which it is drawn. Selection of subjects from institutions, such as hospitals and clinics, may well imply selection by age and parity, both of which may well be associated with anaemia.
P C Elwood (Wed,) studied this question.