To estimate whether the short-term effect of PM10 has changed over time
Model (1.1) is fitted to the entire dataset for each city and separately to the two
datasets defined by splitting the data for each city at the date 31/12/2000. The
date 31/12/2000 is selected as the “split” point because roughly 50% of the data
occurs either side of this date. We refer to the datasets before and after 31/12/