Are Cardiopulmonary Patients More Susceptible to Acute Effects of Air Pollutants?

Abstract

P-725 Introduction: People with existing respiratory and cardiac conditions may be more susceptible to the acute effects of air pollutants. We tested this hypothesis using a novel approach: a cohort design with nested case-control analysis was used to measure the association between black smoke (BS) and mortality rates among subjects previously admitted to hospital for an emergency respiratory condition, an emergency cardiac condition or a digestive condition. Patients were followed for up to 15 years but attention was restricted to acute effects of pollution. Methods: Three cohorts of patients aged < 90 years admitted to hospitals in Edinburgh, Scotland between 1981 and 1996 were identified: emergency admissions for respiratory conditions, emergency cardiac admissions and any admission for digestive disease. To be eligible, patients also had to be discharged alive and to survive at least 30 days. Follow-up was from day 30 until death or 30/9/1996. For the nested case-control analyses, cases were members of the cohorts who died. For each case, up to 5 controls were chosen at random from patients in the same cohort of the same age and sex, who left hospital on the same day of the week and survived longer after hospitalisation than the case. For cases and matching controls, point source exposure to black smoke (BS) was averaged across the 30 days preceding the case's death. Analysis was by conditional logistic regression with adjustment for calendar year, month and temperature lagged to 30 days. Results were compared with other ‘time series’ estimates of risk for the Edinburgh population. Results There were 34,204 cardiac admissions, 12,422 respiratory admissions and 92,865 digestive admissions of which 45%, 39% and 19% respectively died during follow-up. For the three cohorts combined, the percentage increase in death rate associated with a 10μg/m3 increase in BS was 5.6% (95% CI: 2.7, 8.6), compared with 3.7% for the whole population estimated using time series methods. The increases were 9.5% (95% CI: 5.6, 13.4) for cardiac patients, 7.9% (95% CI: 2.4, 13.7) for respiratory patients and 1.6% (95% CI: -1.8, 5.1) for digestive admissions; differences were statistically significant (p=0.0004) With BS modelled as a categorical variable, dose-response relationships were evident in the cardiac and respiratory groups. Conclusions and discussion The results support the original hypothesis but it should be remembered that there was no adjustment for traditional risk factors. Further work is under way to estimate degree of life shortening and risk of readmission to hospital.