Statistical derivation of a standardised procedure for counting expectorated alveolar macrophages as indicators of occupational air pollution.

A biological test in which counts of expectorated alveolar macrophages are used to evaluate lung irritation attributable to occupational air pollution has been applied to workers exposed to dust and gas pollution. To determine the optimum allocation of resources when performing the test, the sampling and counting procedures have been evaluated statistically. Alveolar macrophages (AM) were recovered from workers at an iron works, an aluminium plant (one large and one small group), and from a small group of non-exposed smokers; the number of subjects was 213. Sampling was repeated with the small group at the aluminium plant (27 subjects) on three consecutive days and the small group of non-exposed smokers on five (3 subjects). AM were counted from smear slides using a light microscope. A standard random effects model was used as the basis for estimating the expectation of the log-transformed overall means (mu) and the variances (var(mu)) of the two categories at the aluminium plant and of the non-exposed smokers. The costs connected with finding mu consist of the financial expenses and the working efforts and time used to perform the test. The precision and reproducibility of the test are closely related to the variability in the results, var (mu). To optimise the costs but still obtain reproducible results, AM should be counted in four drops of expectorate from each of three samples from at least 10 to 20 subjects.