Case-crossover and cohort studies of short-term relationships between air pollution and mortality
| Author | Population | Health variable(number of cases) | Exposure variables(time resolution,space resolution) | SES variables(resolution) | Method to evaluate effect modification | Lags tested (days) | Principal results (95% confidence interval when available) |
|---|---|---|---|---|---|---|---|
| BS, black smoke; HS, high school; O3, ozone; PM10, particulate matter with an aerodynamic diameter of up to 10 μm. | |||||||
| Bateson and Schwartz42 | Residents of Cook County(IL, USA),>65 years and previously hospitalised for cardiac or pulmonary diagnosis,1988–1991 | All-cause mortality(65 180) | PM10(mean: day of death+preceding day, county) | -Median house-hold income-% Adults with bachelor’s degree-% Adults not speaking English at home (ZIP code of residence) | Ratio of log of OR(pollution/mortality)of the population with a given SES characteristic to the log of the OR of 1 for the population with-out this characteristic Linear continuous interaction terms between PM10 and SES variables in the model | 0 | No significant modifying effect Increase in mortality for a 10 μg.m−3 increment in PM10concentration-increase of 0.2% (−0.28%; 0.67%)for an increase of 10% in bachelor’s degrees-increase of 0.002% (−0.393%;0.399%) for an increase of 10% in adults not speaking English at home-decrease of 0.04% (−0.65; 0.55)for an increase of US$10 000 in median household income |
| Zekaet al35 | Residents of 20 US cities,all ages,1989–2000 | Non-trauma mortality(1 896 306),respiratory mortality(190 000),mortality from cardiac disease(625 800),mortality from infarction(493 000),mortality from stroke(132 700) | PM10 (mean:day of death+2 preceding days;city) | Educational attainment:<8 years of schooling 8–12 years of schooling⩾13 years of schooling(individual) | Stratified analysis Trend tests | 0 | Increase in mortality for a 10 μg.m−3increment in PM10 concentration Non-trauma mortality<8 years of schooling 0.62 (0.29; 0.95)8–12 years of schooling 0.36 (0.12; 0.60)>12 years of schooling 0.27 (−0.004; 0.54)(trend not significant, p = 0.29) Respiratory mortality<8 years of schooling 0.82 (−0.32; 1.96)8–12 years of schooling 0.88 (0.12; 1.64)>12 years of schooling 0.88 (−0.04; 1.80)Mortality by cardiac disease<8 years of schooling 0.72 (0.23; 1.21)8–12 years of schooling 0.38 (0.07; 0.69)>12 years of schooling 0.54 (0.13; 0.95)Mortality by infarction<8 years of schooling 0.33 (−0.83; 1.49)8–12 years of schooling 0.79 (0.28; 1.30)>12 years of schooling 0.13 (−0.82; 0.56)Mortality by stroke<8 years of schooling 0.07 (−1.44; 1.58)8–12 years of schooling 0.29 (−0.32; 0.90)>12 years of schooling 0.52 (−0.28; 1.32) |
| Filleulet al37 | Residents of Bordeaux(France),>65 years,1988–1997 | Non-trauma mortality(527),cardio-respiratory mortality(197) | BS (daily mean, city) | Educational attainment:-without primary school diploma-primary school diploma-secondary validated or higher Previous occupation-never worked-white-collar-blue-collar(individual) | Stratified analysis | 3 | Increase in mortality for a 10 μg.m−3increment in BS concentration Non-trauma mortality:only subgroup with significant OR:blue collar OR = 1.41 (1.05–1.90)Cardiorespiratory mortality:only subgroup with significant OR:+high educational level OR = 4.36(1.15–16.54) |
| Filleulet al43 | Residents of Bordeaux(France),>65 years,1988–1997 | Non-trauma mortality(543) | BS (above 90th percentile or below 10th percentile of observed concentrations)(daily mean, city)Adjustments for age, sex,smoking habits and education(or occupation) | Educational level:-no school-primary without diploma-primary with diploma Previous occupation:-domestic employees and women at home-blue-collar workers-craftsmen and shopkeepers-other employees,intellectual occupations(individual) | Stratified analysis | 0 | No modifying effect apparent |
| Romieu33 | Children in Ciudad Juarez,Mexico,aged 1 month to 1 year,1997–2001 | Total mortality(628),respiratory mortality(216) | PM10,O3(mean for the 8 h of highest concentrations)(averaged on 1,2 or 3 days before death,city) model with these 2 pollutants | Composite SES index(3 levels)(ZIP code of residence) | 1) Global analysis with introduction of an interaction term for SES 2) Stratified analysis | 1 and 2 | 1) Global analysis: no significant association between pollutants and mortality, but “marginally significant” interaction between SES and PM10 (p = 0.10)2) Stratified analysis: for respiratory mortality, only the children with the lowest SES present statistically significant or almost significant OR for PM10Averaged concentrations for 1 day,lag 1: OR = 1.61 (0.97; 2.66)Concentrations averaged for the 2 days before death: OR = 2.56(1.06; 6.17)No statistically significant association with O3 |