SUMMARY Results for biochemical and haematological variables have been correlated with data on atmospheric temperature in order to identfy possible mechanisms through which low environmental temperature may increase mortality from myocardial infarction and cerebrovascular disease. With the exception of cholesterol, there were no associations in the case of several clinical chemistry variables, or of haemoglobin and related indices. With varying degrees of consistency among the sex and age groups studied, temperature was positively correlated with factor VII, antithrombin III, and cholesterol, and negatively correlated with fibrinolytic activity. The correlations were all low but may offer some clues to mechanisms whereby air temperature influences ischaemic heart and cerebrovascular disease mortality.In previous publications we reported that there is a negative correlation between atmospheric temperature and the death rates from a number of conditions, in particular, myocardial infarction and cerebrovascular accidents (Bull, 1969;Bull, 1973; Bull and Morton, 1975a, b). This inverse relationship between temperature and death rates was confined to or found to be greatest in older subjects (Bull, 1973;Bull and Morton, 1975a).In these earlier publications we speculated on a possible causal relationship between temperature and cerebrovascular and ischaemic heart disease. If the relationship between environmental temperature and mortality rates from arterial diseases is one of cause and effect, it is presumably mediated by changes in one or more body systems; these changes might be detectable in groups of people (mostly without clinical arterial disease) for whom biochemical, haematological, and other data can be related to environmental temperature. We have therefore studied the associations between the levels of various blood constituents and atmospheric temperature.
MaterialThere are four sources of data. March 1978 in an archival file on a computer together with identifying particulars and the date of the sample. The file for 1973 was used to prepare a table of the mean blood levels of 10 blood constituents by days in each sex and in two age groups (< 55 and > 55 years). All readings outside the limits of 2 standard deviations (SD) from the quality control mean of the laboratory were rejected before calculation of the daily means. Thus the mean values should approximate to the 'normal' mean values for this hospital's population. In all, data from 103 370 individual blood samples were available for analysis, and on most of these five or more blood constituents were analysed.The number of individual readings making up the daily mean varied, and on certain days, at weekends or on public holidays, none was available. The missing values were filled by inserting the average of the readings straddling the vacant cell.