[An epidemiological study of peripheral blood studies].

A statistical random sample of 600 patients from the population referred to Tygerberg Hospital for peripheral blood investigations was analysed statistically with an IBM 370/158 computer. Suitable transformations were used for normalization of the data. The results of Coulter Model S investigations, platelet counts (coulter F), erythrocyte sedimentation rates, differential counts (with regard to neutrophil, lymphocyte, monocyte and eosinophil counts), age, race and sex of the patients, as well as possible associations between the day of the week and the haematological tests were investigated independently for each of the four race and sex groups, i.e. White men, White women, non-White men and non-White women. Machine sorting was used to obtain graphs representing the mean values of 12 haematological variables versus age in the four race and sex groups. Regression analysis between the different variables was done, and mean values are reported. Normal haematological limits (+/- 2 standard deviations) for the different age groups were used, and with machine processing the results were evaluated as low, normal or high in order to obtain a table for the visualization of a normal or abnormal distribution, skewness to the right or left, and percentages of abnormal values for each variable for each of the different age groups in the four race and sex groups. In this study we investigated a selected group of patients. No conclusions with regard to 'normal' values can be made, but, with the use of statistical computing of routine numerical data, an approach to epidemiology is demonstrated.