Time-series studies of short-term relationships between air pollution and mortality
| Author | Population | Health variables(number of cases) | Exposure variables(time resolution,spatial resolution) | SES variables(resolution) | Method of evaluation of effect modification | Lags tested(days) | Principal results (95% confidence interval when available) |
|---|---|---|---|---|---|---|---|
| *No further details available. | |||||||
| BS, black smoke; CO, carbon monoxide; CoH, coefficient of haze; HS, high school; NO2, nitrogen dioxide; O3, ozone; PM, particulate matter; PM10, particulate matter with an aerodynamic diameter of up to 10 μm; PM2.5, particulate matter with an aerodynamic diameter of up to 2.5 μm; SO2, sulfur dioxide; TSP, total suspended particulates. | |||||||
| Sametet al36 | Residents of 20 US cities,all ages,50 000 000 inhabitants,1987–1994 | Non-trauma mortality(3 000 000),cardio-vascular and respiratory mortality(1 600 000) | PM10 adjusted for O3, SO2,NO2, CO(daily mean,city) | % High school graduates% Annual income<US$12 675% Annual income>US$100 000(city) | Hierarchical bayesian model:First level, log-linear regression of mortality rate according to pollutants and confounders in each city Second level, linear regression of pollutant effects in all cities according to SES characteristics | 1 | PM10 - mortality associations in cities not associated with city SES characteristics (95% posterior interval for these SES variables,all include 0*) |
| Schwartz38 | Residents of 10 US cities,all ages,14 000 000 inhabitants,1986–1993 | Non-trauma mortality(1 100 000) | PM10 (mean:day of death and preceding day, city)excluding days where it exceeded 150 μg.m−3 | Unemployment rate% Population below poverty level% Population with a college degree(city) | Hierarchical model meta-regression Measurement of the variation of the effect of PM10 for a 5% increase in the SES variable | 0 | No modifying effect Results in graphic form Variation of effects of PM on mortality for a 5% increase in the SES variables:% poverty 0 (−0.05; 0.05)% college degree 0 (−0.03; 0.03)unemployment rate 0.02 (−0.06–0.1) |
| O’Neillet al39 | Residents of Mexico City,>65 years,20 000 000 inhabitants,1996–1998 | Non-trauma mortality(206 510) | PM10, O3(mean: day of death and preceding day, city) | Sociospatial development index(3 classes)% Homes with electricity% Homes with piped water% Homes with drainage% Literacy% Indigenous language speakers(county) | Stratified analysis | 0 | PM10 not associated with mortality O3:Sociospatial development index High/medium high 2.76 (1.14; 4.40)Medium/medium low 0.64 (−0.44; 1.72)Low/very low 3.78 (0.76; 6.89)% Homes with electricity>99.8, 1.63 (0.50; 2.78)93–99.8, 0.92 (−0.93; 2.81)82–93, 1.50 (−0.94; 4.00)60–82, 2.04 (−0.64; 4.80)% Homes with piped water>99.6, 2.09 (0.56; 3.65)75–99.6, 0.78 (−0.85; 2.44)35–75, 1.58 (−0.28; 3.48)21–35, 1.79 (−0.03; 3.64)% Homes with drainage>99.3, 1.45 (−0.07; 3.01)80–99.3, 1.47 (−0.22; 3.18)45–80, 1.54 (−0.29; 3.40)21–45, 1.88 (0.14; 3.66)% Literacy>97.5, 2.89 (1.34; 4.46)96–97.5, 0.43 (−1.15; 2.04)88–96, 0.92 (−1.09; 2.97)80–88, 1.59 (−0.16; 3.36)% Indigenous language speakers 0.3–1, 1.58 (−0.25; 3.44)1–1.5, 1.20 (−0.26; 2.69)1.5–2, 3.54 (1.56; 5.57)2–6, 0.59 (−1.34; 2.55)6–10.4, 1.18 (−1.52; 3.96) |
| Martins et al32 | Residents of six zones of Sao Paulo (Brazil),>60 years,992 018 inhabitants,1997–1999 | Respiratory mortality(1991) | PM10 (3-day moving average,city subdivision(called regions)included in 2-km radius around traffic pollution monitors) | % People with college education% Families with monthly income>US$3500% People living in slums (city subdivisions) | Spearman rank correlations between associations of PM10with respiratory mortality and SES variables | 0 | Effect of PM10 on respiratory mortality:negatively correlated with:% college education (−0.94, p<0.01)% family income >US$3500(−0.94, p<0.01)positively correlated with % people living in slums (0.71, “not significant”*) |
| Gouveia and Fletcher40 | Residents of Sao Paulo(Brazil),>65 years,9 500 000 inhabitants,1991–1993 | Non-trauma mortality(151 756) | PM10(daily mean,city) | Composite index(4 classes)(58 districts in Sao Paulo) | Stratified analysis then interaction term in a model Significance of interaction tested by log-likelihood ratio test | 0, 1 and 2 | Results in graphic form Relative risks slightly stronger in more advantaged neighbourhoods(but p>0.50*) |
| Jerrettet al26 | Residents of Hamilton(Canada),all ages,320 000 inhabitants,1985–1994 | Non-trauma mortality(27 458) | CoH, SO2,(reciprocal adjustments for these pollutants)(concentrations averaged for 1–3 days before death,5 city subdivisions) | Mean household income% Unemployment% Poverty% HS or less% <grade 9% Manufacturing employment(5 city subdivisions) | Stratified analysis 1) Maximum likelihood random effects model(evaluation of differences between relative risks of city subdivisions and relative risk for the entire city)2) Regression of mean % change in mortality associated with exposure for SES characteristics in each area | 0–3 | Overall, stronger and more significant relative risks and in zones with unfavourable SES characteristics 1) Random effects model: no significant differences in relative risks between zone and entire city 2) Manufacturing employment and educational level significantly correlated with the effect of CoH on mortality(p value not given)* |
| Cifuenteset al28 | Residents of Santiago(Chile),25–64 years,5 000 000 inhabitants,1988–1996 | Non-trauma mortality(43 400) | PM2.5, CO(reciprocal adjustments)(mean: day of death+preceding day, city) | Educational attainment:-no education-elementary school-HS-university(individual) | Stratified analysis | 0 | PM2.5After stratification, relative risks were significant (or very nearly so) only in the group with elementary education CO After stratification, no significant relative risk |
| Villeneuveet al34 | Vancouver(Canada)residents in British Columbia Health datasets,>65 years,550 000 subjects,1986–1999 | Non-trauma mortality(93 612)Cardio-vascular mortality(40 840)Respiratory mortality(11 650) | TSP, CO, NO2,SO2, O3, PM10,PM2.5 (mean concentrations for the 3 days before death,city) | Mean family income(3 classes)(enumeration area) | Stratified analysis | 0 | Results in graphic form After stratification, the only statistically significant relative risks concerned non-trauma mortality:-NO2 in low and middle income-SO2 in low income (borderline significance)-TSP in high and low income |
| Zanobetti and Schwartz41 | Residents of Chicago,Detroit,Minneapolis-St Paul and Pittsburgh (USA),all ages,10 000 000 inhabitants,1986–1993 | Non-trauma mortality(782 502) | PM10 (mean:day of death+preceding day,city), excluded days where it exceeded 150 μg.m−3 | Educational attainment:<HS⩾HS(individual) | Stratified analysis | 0 | % Increase in death for 10 μg.m−3increase in PM10<HS 0.92 (0.66–1.18)⩾HS 0.71 (0.19–1.23)Difference judged not statistically significant (because of overlapping confidence intervals) |
| Wojtyniaket al27 | Residents of Cracow, Lodz,Poznan and Wroclaw(Poland), 0–70 years or>70 years,2 000 000 inhabitants,1990–1996 | Non-trauma mortality(unknown)Cardio-vascular mortality(unknown) | BS, NO2 and SO2 (mean:day of death+preceding day,city) | Educational attainment:-below secondary(primary and vocational)-secondary and above (post secondary and university)(individual) | Stratified analysis | 0–1 | Non-trauma mortality BS: significant effects only for less than secondary education(< and ⩾70 years)NO2: significant in ⩾70 years regard-less of education: in ⩾70 years for below secondary education only SO2: significant effects only in those ⩾70 years with less than a secondary education Cardiovascular mortality BS: significant effects only for those with less than a secondary education (< and ⩾70 years)NO2: in <70 years, significant for secondary education and above only (similar but not significant for below secondary education)In ⩾70 years, significant regardless of education (but larger in lower education)SO2: significant effects only for those⩾70 years with below secondary education |