Concurrence of monthly variations of mortality related to underlying cause in Europe.

STUDY OBJECTIVE The study aimed to examine the concurrence in the variation of monthly numbers of deaths in summer and winter from the four main underlying causes - respiratory, circulatory, neoplastic, and all others - in four countries. In particular, the hypothesis that most non-respiratory concurrent deaths are miscoded respiratory deaths and that a large proportion of the winter mortality currently attributed to circulatory disorders should be attributed to respiratory causes was considered. DESIGN Mortality data were analysed graphically in relation to cause. Each of the four series of monthly data underwent time series analysis to remove auto-correlation, seasonality, and secular trends. Associations between paired causes of death and between multiple series (using Kendall's coefficient of concordance) were then examined after modelling. SETTING Monthly deaths (65 years and over) related to underlying cause were examined for England and Wales (nine years), The Netherlands (nine years), Denmark (10 years), and Portugal (10 years - all ages). Weekly data for England and Wales (51 weeks) were also analysed. MAIN RESULTS All combinations of monthly deaths related to underlying cause were strongly associated in all four countries. This concurrence was evident down to the lowest monthly values so that all seasonally related deaths above the minimum monthly value can be used as an estimate of the "concurrent" proportion. Associations involving deaths from neoplasm were weakest. Concurrence was evident even on a weekly analysis (England and Wales). Concurrent deaths in England and Wales accounted for 31.1% of respiratory, 16.0% of circulatory, 3.5% of neoplastic, 14.1% of deaths from other causes and 14.2% for all deaths combined. The equivalent percentages for concurrent deaths from all causes were 8.4% in the Netherlands, 9.3% in Denmark, and 16.8% in Portugal. CONCLUSIONS Concurrence, which was present in each of the underlying causal groups in each of the four national data sets examined, suggests a common cause separate from the underlying cause that has been used in the presentation of mortality statistics. If the person concerned had not died at that time, as a result of this cause, he would not have died from the recorded underlying cause. Most of these non-respiratory concurrent deaths are miscoded. As a consequence, a large proportion of winter mortality currently attributed to circulatory disorders should be attributed to other causes, probably respiratory. More intensive research into the contribution made by acute respiratory diseases is proposed. The proportion of concurrent deaths varied in the four countries thereby limiting the validity of simple comparisons of national mortality statistics.