Relation of daily mortality to air pollution: an analysis of 14 London winters, 1958/59-1971/72.

The relationship between daily deaths and daily concentrations of Smoke and SO2 in London, England for 14 winters during the years 1958-1959 through 1971-1972 has been explored. Three types of analyses were used: (1) year-by-year multiple regression; (2) stratification using nested quartiles of one pollutant within quartiles of the other; and (3) multiple regression of a special subset of high pollution days. An association was found with Smoke, but not with SO2. Whether a linear model with zero threshold or a threshold model best fits could not be determined unambiguously because of a statistical artifact. Reasons for preferring a threshold-type quadratic model are given.