This snapshot taken on 16/07/2006, shows web content selected for preservation by The National Archives. External links, forms and search boxes may not work in archived websites.
Appendix 1: PM2.5 and daily mortality and hospital admissions in the West Midlands conurbation
Introduction
1. The epidemiological evidence about the health effects of PM2.5 comes largely from North America, and none is currently available for the UK (see Epidemiology Chapter for literature review). The Panel therefore commissioned, through the Department of Health, an investigation of PM2.5 and daily mortality and hospital admissions in the West Midlands during the period from October 1994 to December 1996. Other available measures of particles were PM10, sulphur dioxide, nitrogen dioxide, ozone and carbon monoxide.
2. The purpose of the investigation was to inform the thinking of EPAQS about the setting of a standard for PM2.5. Two as yet unpublished reports, one on mortality and the other on hospital admissions were discussed by EPAQS and will be made available on the Internet after peer-reviewed papers have appeared in the medical press. This chapter summarises the main results for particles, focussing especially on the following questions:
- Is PM2.5 associated with daily mortality or daily hospital admissions?
- How do the effects of PM2.5 compare with those for PM10?
- How do the effects of PM2.5 compare with those for the coarse fraction (PM10-2.5)?
- Do the results for Black Smoke and sulphate (both of which are predominantly fine particles) give insight into the active component of the particle mixture?
Methods
3. The area of study was the West Midlands conurbation, which has a population of 2.3 million and includes the city of Birmingham. The study period was constrained by the availability of PM2.5 data to October 1994 to December 1996. Counts of daily deaths and emergency hospital admissions were obtained from the Office for National Statistics for those people resident in and dying or admitted in the study area. Mortality series were constructed for all-causes (excluding accidents), respiratory and cardiovascular causes. Hospital admissions series were constructed for respiratory and cardiovascular diagnoses, divided as appropriate by sub-diagnoses and age.
4. Air pollution data were obtained from AEA Technology for all sites with data on at least 75% of days. The particle measures were PM10 (3 sites), PM2.5 (one site), Black Smoke (3 sites), sulphate (2 nearest rural sites), and PM10-2.5 (obtained from subtraction for the coterminous PM2.5 and PM10 site). 18% of PM2.5 measurements were missing, mainly at the end of the study period. The mean of the daily maxima and minima of temperature, and of 6am and 3pm humidity were calculated from data provided by the Meteorological Office and University of Birmingham.
5. Poisson regression was used to model the association of the daily number of deaths or admissions on pollution levels after controlling for seasonal patterns, temperature and humidity, influenza episodes and day of week and holiday effects. Generalised Additive Models, or GAM, were used to allow for the non-linear dependence of the outcome measures on some explanatory variables. Particular attention was paid to the relationships between daily mortality and admissions and the weather factors of temperature and relative humidity. The most appropriate functional form, parametric or non-parametric, of temperature and humidity on the same day and previous few days were investigated and compared. Those which minimised the Akaike's Information Criteria were selected for inclusion in the model. Finally, the pollution measures were added, in turn, and the parameter estimates were adjusted for overdispersion (assumed constant over time) and any remaining serial correlation.
6. The a priori hypothesis concerned an effect of lag 0+1. However, individual lags from day 0 to day 3 were all examined, as well as the other cumulative lags, 0+1+2 and 0+1+2+3, and the results of all these may be found in the main reports. The pollutants were fitted as linear terms.
7. The statistical technique used gives a relative risk estimate for a pollution "effect" together with a standard error, which indicates the precision of the estimate. In this report, the relative risk is presented as a percentage change in daily mortality or admissions associated with a change in pollutant from the 10th to 90th percentile of its range. This method of presentation is appropriate for comparing the effects of various pollutants in one city. With such a large number of analyses, it is important to have an explicit policy for interpreting the results. In addition to using an a priori hypothesis for the lag (day 0 + 1), other yardsticks were: size of estimate, direction of estimate, statistical significance (1% level), and consistency of lag pattern.
8. Summary data for the environmental and health outcome variables are in Table A.1. The mean daily counts of deaths from all-causes, cardiovascular causes and respiratory causes were 61, 28 and 10, respectively. The mean daily counts for cardiovascular and respiratory admissions were 71 and 66 respectively. Concentrations of PM10 were quite low in comparison to the current standard of 50 µg/m3, with a mean of 23 and maximum of 102. PM2.5 concentrations were about two thirds of those of PM10, with a mean of 15 and maximum of 83. The correlation between PM2.5 and PM10 was high (r = 0.94), as might be expected. The correlations between PM2.5 and SO4 and Black Smoke were lower (r = 0.63 and 0.71 respectively) and between PM2.5 and the coarse fraction (PM10-2.5) was low (r = 0.34).
9. The main mortality results are presented in Table A.2 and Figure A.1, for the all-year analysis and by season. For the all-year analysis, there were no significant associations with PM10, PM2.5, PM10-2.5, Black Smoke or sulphate in the all-cause, cardiovascular or respiratory diagnostic groups. There were however, a number of significant seasonal differences. Whilst no associations were significantly positive in the cool season, in the warm season there were significant positive associations between all-cause mortality PM10, PM2.5, Black Smoke and Sulphate, but not with PM10-2.5. There was no convincing evidence of positive associations between any particle measures and cardiovascular or respiratory mortality in either season. There was, however, a clear tendency for small negative associations in the cool season and larger positive associations in the warm season. A notable finding was a highly significant negative association between the coarse fraction and respiratory mortality in the cool season.
10. The results for hospital admissions are shown in Table A.3 and Figure A.2. In the all-year analysis of cardiovascular admissions, there were no associations with particles. In the seasonal analysis, the only significant association was with sulphate. Analysis of cardiac disease, ischaemic heart disease and stroke as separate series provided little further useful information.
11. For respiratory admissions, analysed over the whole year, there were no significant associations with particles in the all-ages group. However, in the 0-14 age group, PM10, PM2.5, and Black Smoke all showed positive associations. In the 15-64 and 65+ age groups no associations were observed. When respiratory admissions were examined by season, a contradictory pattern emerged. Amongst the 0-14 age group, a number of significant associations were observed in the cool season only, whereas in the elderly, there were large and strongly significant positive associations in the warm season alongside significant negative, but smaller, associations in the cool season. It therefore appears that the positive associations observed in the cool season among the all-ages group are being driven by cold season associations in children whilst the significant positive associations in the warm season in the all-ages group are being driven by warm season associations in the 65+ group.
12. It was not possible to discern an obvious difference between the effects of PM2.5 and PM10, which is not surprising since they are not independent of one another. PM2.5 also tended to have effects of similar size and direction to those of Black Smoke and sulphate. There were however indications that the PM2.5 associations did not parallel those of the coarse fraction, PM10-2.5, and in some instances were in opposite directions. In two pollutant models, the most notable result was the stability of the Black Smoke estimate together with increased statistical significance when other particle measures were included in the Black Smoke model. When the pattern of single day lags was examined, there was a clear tendency for the PM2.5 pattern to resemble that of PM10, and Black Smoke, whereas the lag patterns of the PM10-2.5 and sulphate were somewhat different.
13. No strong and consistent associations between with PM2.5 and daily mortality or admissions were observed in the main all-year analyses. Similar findings applied to PM10, Black Smoke and sulphate. These findings contrast with the majority of published studies but it should be noted that the results for PM10 and Black Smoke are within the bounds of results obtained in a previous study of Birmingham (Wordley et al 1997) and of London (Bremner et al 1999). Possibly, the relative shortness of the PM2.5 series in particular, (700 days) contributed to a lack of statistical power and analysis of a longer series would help clarify this possibility. Associations did, however, emerge when the data were analysed by season and by diagnosis and age. Because a large number of analyses were conducted, small and isolated significant effects must be regarded cautiously and more credence should be placed on strong consistent and highly significant findings. On the other hand, the policy to use a cumulative lag a priori will have led to more conservative results than one which selected the most significant single day lag post hoc.
14. It was not possible to identify convincingly the most important component of the particle mixture because the estimates were similar in scale and could not be disentangled with multi-pollutant models. There were indications however that Black Smoke was the most consistently positive and most stable in multi-pollutant models. Also, there were quite clear indications that the coarse fraction was less important because it did not follow the pattern of the other particle measures and was less likely to be significantly positive. The coarse fraction was sometimes found to have significantly negative associations. This was not predicted but an explanation may lie in an understanding of the origin of this fraction. The coarse fraction arises mainly from resuspension of dust from roads and other crustal sources. The sources of fine particles of primary or secondary origin are very different, as are the conditions for their accumulation. Thus it is not surprising that the correlation between the coarse fraction and PM2.5 is low (r = 0.34). There is some evidence that the coarse fraction is positively associated with wind speed, which in turn could plausibly be associated with a reduction in pollution associated with stagnant conditions. This could explain the negative associations with PM10-2.5, but is speculative.
1. In the all-age, all-year analyses of mortality (all-cause, cardiovascular and respiratory) and emergency hospital admissions (cardiovascular and respiratory), the associations with PM2.5 were small and not statistically significant.
2. Seasonal analysis of these outcome series revealed associations between PM2.5 and all-cause mortality, and admissions for respiratory disease in the warm season but no associations in the cool season.
3. Similar findings applied to other measures of particles (PM10, sulphate, Black Smoke)
4. When divided by age and diagnosis, significant associations were observed between PM2.5 and all respiratory admissions and asthma in the 0-14 age group, and between PM2.5 and respiratory admissions in the elderly.
5. Of all the particle measures, Black Smoke appeared to be the one most consistently associated with health effects, and which was most stable in models which included other pollutants.
6. The results for PM2.5 tended to parallel those for PM10, of which it is one component. The associations with PM2.5 could not be separated from those of PM10.
7. The associations with the coarse fraction (PM10-2.5) did not parallel those of PM2.5 or the other particles and none were significantly positive.
8. If PM10 does have short-term effects on health in this population, it is likely that this is due to toxicity contained mainly in the PM2.5 fraction.
Bremner SA, Anderson HR, Atkinson RW, et al. Short-term associations between outdoor air pollution and mortality in London 1992-94. Occup Environ Med 1999; 56: 237-244.
Wordley J, Walters S, Ayres JG. Short-term variations in hospital admissions and mortality and particulate air pollution. Occup Environ Med 1997; 54: 108-116.
Figure A.2. Particles and daily hospital admissions in Birmingham 1994-96. All relative risks are expressed in terms of the 10th-90th percentiles of concentrations of the metrics indicated.
Published 17 May 2001
Expert Panel on Air Quality Standards
Air and Envirnonmental Quality
Environmental Protection Index
Defra Home Page
