Printer Friendly

The association of daily diabetes mortality and outdoor air pollution in Shanghai, China.


During the past decade, numerous epidemiological studies have confirmed that ambient air pollution is associated with increases in daily mortality, especially mortality due to cardiovascular and respiratory diseases (Committee of the Environmental and Occupational Health Assembly of the American Thoracic Society, 1996). Subpopulations especially susceptible to ambient air pollution still need to be identified, however, and this issue has been regarded as a key research need (National Research Council, 1998). Also, the issue is obviously of great importance in the exploration of potential biological mechanisms of air pollutants and in setting relevant public health policies.

Recently, elevated exposure to air pollution has been associated with triggering of myocardial infarction (Peters, Dockery, Muller, & Mittleman, 2001), initiation of life-threatening arrhythmias (Peters et al., 2000), changes in cardiac rhythm and autonomic function (Pope et al., 1999), endothelial dysfunction (Brook, Brook, Urch, Vincent, Rajagopalan, & Silverman, 2002), increased plasma viscosity (Peters, Doring, Wichmann, & Koenig, 1997), and increased C-reactive protein (Peters et al., 2001). These findings suggest possible pathways by which air pollutants, especially particulate matter, affect the incidence and death rate of cardiovascular diseases.

Diabetes is known to be a chronic disease characterized by disturbance in the cardiovascular system (Stec et al., 2000). Therefore, diabetics have been suspected to be at higher risk of air pollution-related health events. Recently, the relationship has been investigated and confirmed positive in Canada (Goldberg et al., 2001) and the United States (Zanobetti & Schwartz, 2001, 2002). In Shanghai, the largest city of China, diabetes has become one of the leading causes of death, and the mortality from diabetes has increased sharply from 0.52 per 100,000 in 1966 to 16.95 per 100,000 in 1998 (Ling, Song, & Zhou, 2001). Therefore, it is worthwhile to explore the relationship between air pollution and diabetes mortality; air pollution is a potentially preventable risk factor that does not rely on behavioral changes and genetic characteristics. In the study reported here, the authors used a time-series approach to assess the effects of air pollution on daily diabetes mortality, and also explored the exposure-response patterns for major air pollutants with respect to diabetes mortality in Shanghai.



The authors obtained daily diabetes mortality data (International Classification of Diseases, 9th revision [ICD-9]. Code 250) for the Zhabei District of Shanghai between January 1, 2001, and December 31, 2002. The Shanghai death certificate data should be considered reliable because all data were reported by physicians, not by relatives of the deceased. All mortality data and their accuracy were rechecked by the staffs of the local Center of Disease Control before being entered into the database. Meteorological data (daily average temperature, relative humidity, and dew point), which were measured by one station in central Shanghai, were provided by the Shanghai Meteorological Bureau. Air pollution data (daily average P[M.sub.10], S[O.sub.2], and N[O.sub.2] concentrations) were retrieved from the database of the Shanghai Environmental Monitoring Center. A total of six fixed-site monitoring stations are scattered among the urban districts of Shanghai. To match mortality and air pollution data in this research, the authors selected the results monitored by the station located in the Zhabei District. No data were found missing for the variables described above.

Statistical Analysis

The analysis followed a time-series procedure. The authors used log-linear models to estimate air pollution/diabetes mortality relative risks (RRs), while controlling for longer-term trends, seasonality, weather, and day of the week. The core analysis was a semi-parametric generalized additive model (GAM). The authors first fitted nonparametric smoothing terms (by means of the smoothing spline function) to establish trends covering days 1-730 for temperature, humidity, dew point, and dummy variables for days of the week. After time, weather, and day of the week were controlled for, each pollutant was introduced into the model. The authors fitted models with different combinations of pollutants (up to three pollutants per model) to assess the stability of individual effects. In addition, they considered the lag effects of temperature, humidity, and pollutant concentrations in building the models. To compare the relative quality of the mortality predictions across these models, Akaike's Information Criterion (AIC) was used as a measure of how well the model fit the data (Hastie & Tibshirani, 1990). Smaller AIC values indicate the preferred model. Considering that the assumption of linearity between the log of diabetes mortality and air pollution level may not be accurate, the authors used the smoothing function to graphically analyze the relationship between air pollution and diabetes mortality. Finally, they compared the effect of air pollutants on diabetics and nondiabetics.

All analyses were carried out with S-PLUS 2000 software (Insightful Corporation, Seattle, Washington). Considering that the default Settings in the GAM function of the S-PLUS software package do not ensure convergence of its iterative estimation procedure and can provide biased estimates of regression coefficients and standard errors (Dominici, McDermott, Zeger, & Samet, 2002), the authors analyzed the data with more stringent convergence parameters than those of the default settings when using the GAM function.



Summary statistics of daily mortality counts, air pollutant concentrations, and meteorological measures are presented in Table 1. A total of 434 diabetes deaths were included in the analysis. During the period, there were on average 0.59 deaths from diabetes per day among the total population in the study area.

Table 2 shows the correlations of daily values over the entire period among the air pollutants and weather variables. P[M.sub.10], S[O.sub.2], and N[O.sub.2] all had a strong positive correlation with one another and were negatively correlated with temperature, relative humidity, and dew point.

In the time-series analysis, the relative risks on the best statistical lagged day (the day with minimum AIC) were statistically significant for P[M.sub.10] and N[O.sub.2], but not for S[O.sub.2] (Table 3). In the single-pollutant models, each increase of 10 [micro]g/[m.sup.3] in P[M.sub.10], S[O.sub.2], or N[O.sub.2] corresponded, respectively, to a 1.006 (95 percent CI: 1.000-1.012), 1.011 (95 percent CI: 0.990-1.032), or 1.013 (95 percent CI: 1.000-1.026) relative risk of diabetes mortality. In the multiple-pollutant models, however, the introduction of other pollutants weakened the effect of the single pollutant on diabetes mortality risk.

Figure 1 shows the exposure-response relationships between air pollutant levels and diabetes mortality at the best-lagged day (lag = 1, df = 5) in the single-pollutant models. The associations were essentially linear for most of the air pollution levels, although the risks were not monotonically increasing.

Table 4 shows the results for the effect of P[M.sub.10], S[O.sub.2] and N[O.sub.2] on deaths due to diabetes and nondiabetes causes. These results suggest that air pollutants have a greater effect in diabetics than in nondiabetics.


By design, time-series studies examine the same population repeatedly under varying exposure conditions; thus, time-invariant characteristics, such as age and cigarette smoking, are no longer potential confounders. This feature is a key advantage of the time-series approach. Recently, sophisticated analytical techniques, such as generalized additive models (GAMs), have been introduced into the time-series studies for the adjustment of long-term and seasonal trends, weather variables, and so forth.

To the authors' knowledge, the analysis described here constitutes the first study to assess the acute effects of ambient air pollution on daily diabetes mortality in Asia. During the past decade, Shanghai, the largest city in China, has undergone the most rapid development and urbanization in its history. The traditional coal combustion-related air pollution of Shanghai has improved substantially, although the level of vehicle-originated air pollution is increasing. At the same time, the disease pattern among Shanghai residents has changed considerably. The leading causes of death have shifted from infectious diseases to noncommunicable diseases (NCDs), including tumor, cardiovascular diseases, diabetes, and so forth (Ling, Song, & Zhou, 2001). This background information raises the interest to study the association between ambient air pollution and diabetes. Evidence gained in this study showed that the current levels of P[M.sub.10] and N[O.sub.2] in the Zhabei District of Shanghai are associated with the daily death rates from diabetes. In general, these results are consistent with prior findings in Western countries (Goldberg et al., 2001; Zanobetti & Schwartz, 2001, 2002).

It is interesting that the study reported here reveals air pollution to have a greater effect on diabetes mortality than on other causes of death (Table 4). As noted in the introduction, recent studies have focused on the health effects of air pollution on the cardiovascular system of susceptible populations. Many studies have been concerned with the potential mechanisms linking air pollution and cardiovascular diseases. The findings of those studies suggest possible pathways by which air pollution affects the cardiovascular system. These cardiovascular diseases are affected by diabetes as well, which makes the observations reported here plausible. The underlying biological mechanisms still need to be further explored, however.

The effect of weather variables, especially temperature, on mortality or morbidity risk is well known. Furthermore, there is an association between air pollution level and weather variables (Table 2) related to the mixing and transport of air pollutants. Therefore, weather-pollution interactions might confound analysis of the effect of air pollution on mortality. In order to estimate the independent effect of air pollution on mortality, the authors used a nonparametric regression model, a generalized additive model, to control for weather.

The limitations of these analyses should be noted. Diabetes is generally underreported as a cause of death, both in China and worldwide. Moreover, compared with other similar studies (Goldberg et al., 2001; Zanobetti et al., 2001; Zanobetti & Schwartz, 2001, 2002), the data collected for this study were limited, both in time and duration and in population numbers enrolled. Data on levels of P[M.sub.2.5] and ozone are not yet available in Shanghai, although studies at other sites have confirmed the association of these two air pollutants and daily mortality changes (Tolbert et al., 2000; Lee et al, 2000). The authors have not addressed in detail how the association between air pollution and diabetes mortality changes by gender, age, and many other factors. For example, it would be useful if diabetes-related deaths could be classified in terms of cardiovascular and noncardiovascular causes. Also, other confounders, such as influenza epidemics, should be further controlled for. Finding solutions to these questions requires more--and more detailed--data.

In reality, people cannot selectively inhale some air pollutants and not others. Because of the problems of colinearity between air pollutants (high correlation among pollutants) (Table 2), it was very difficult to separate the effects of individual pollutants from that of others. The multiple-pollutant models can increase the standard error of the results (Bollen, 1989), which might lead to lower statistical significance. Nevertheless, it seems clear that current air pollution levels in Shanghai are related to increased daily diabetes mortality.

In summary, the study reported here provides new evidence of the association between air pollution and diabetes mortality, and the relationship deserves more attention. Focusing on the preventable aspects of diabetes mortality--for example, air pollution--could significantly reduce diabetes-related health problems.
TABLE 1 Mean, Standard Deviation (SD), and Distribution of Daily
Diabetes Mortality, Air Pollution Levels, and Meteorologic Measures
(n = 730)

 Mean SD 10% 25% 50%

Mortality Counts
 Total 13.86 4.29 6 12 13
 Diabetes 0.59 0.71 0 0 0
Meteorologic Measures
 Temperature ([degrees]C) 17.56 8.49 -1.75 10.30 18.50
 Relative humidity (%) 73.66 11.28 39.75 66.50 74.00
 Dew point ([degrees]C) 12.44 9.18 -13.825 4.94 13.80
Air Pollutants Concentrations
 P[M.sub.10] ([micro]g/[m.sup.3]) 97.01 74.92 12.00 48.00 73.00
 S[O.sub.2] ([micro]g/[m.sup.3]) 48.36 32.57 5.00 28.50 40.00
 N[O.sub.2] ([micro]g/[m.sup.3]) 67.18 24.00 18.40 51.00 64.00

 75% 90%
Mortality Counts
 Total 19 22
 Diabetes 1 2
Meteorologic Measures
 Temperature ([degrees]C) 24.45 32.75
 Relative humidity (%) 81.30 97.00
 Dew point ([degrees]C) 20.19 26.73
Air Pollutants Concentrations
 P[M.sub.10] ([micro]g/[m.sup.3]) 116.50 564.00
 S[O.sub.2] ([micro]g/[m.sup.3]) 59.50 302.00
 N[O.sub.2] ([micro]g/[m.sup.3]) 80.00 199.00

TABLE 2 Pearson Correlation Coefficients Among Daily Weather and Air
Pollution Variables

 P[M.sub.10] S[O.sub.2] N[O.sub.2] Temperature

P[M.sub.10] 1.00
S[O.sub.2] 0.67 1.00
N[O.sub.2] 0.65 0.73 1.00
Temperature -0.33 -0.30 -0.40 1.00
Relative humidity -0.39 -0.47 -0.22 0.25
Dew point -0.41 -0.40 -0.42 0.96

 Relative Dew
 Humidity Point
Relative humidity 1.00
Dew point 0.50 1.00

TABLE 3 Relative Risk of Diabetes Mortality for an Increase of 10
[micro]g/[m.sup.3] in Air Pollutant Concentration Under Single- and
Multiple-Pollutant Models

 Mean 95% CI

P[M.sub.10] 1.006 1.000-1.012
 Adjusted for S[O.sub.2] 1.003 0.997-1.009
 Adjusted for N[O.sub.2] 1.002 0.995-1.009
 Adjusted for both S[O.sub.2] and N[O.sub.2] 1.000 0.991-1.009
S[O.sub.2] 1.011 0.990-1.032
 Adjusted for P[M.sub.10] 1.008 0.987-1.029
 Adjusted for N[O.sub.2] 1.001 0.990-1.012
 Adjusted for both P[M.sub.10] and N[O.sub.2] 0.986 0.923-1.050
N[O.sub.2] 1.013 1.000-1.026
 Adjusted for P[M.sub.10] 1.009 0.991-1.027
 Adjusted for S[O.sub.2] 1.008 0.989-1.027
 Adjusted for both P[M.sub.10] and S[O.sub.2] 1.006 0.990-1.022

TABLE 4 Percentage Increase in Deaths Due to Diabetes and Other Causes
for Increases of 10 [micro]g/[m.sup.3] in P[M.sub.10], S[O.sub.2], and
N[O.sub.2] in Shanghai

 With Diabetes Without Diabetes
 % 95% CI % 95% CI

P[M.sub.10] 0.6 0.0 ~ 1.2 0.4 0.0 ~ 0.8
S[O.sub.2] 1.1 -1.0 ~ 3.2 0.7 -0.2 ~ 1.6
N[O.sub.2] 1.3 0.0 ~ 2.6 1.1 -0.2 ~ 2.4


Bollen, K.A. (1989). Structural equations with latent variables. New York: John Wiley & Sons.

Brook, R.D., Brook, J.R., Urch, B., Vincent, R., Rajagopalan, S., & Silverman, F. (2002). Inhalation of fine particulate air pollution and ozone causes acute arterial vaso-constriction in healthy adults. Circulation, 105, 1534-1536.

Committee of the Environmental and Occupational Health Assembly of the American Thoracic Society (CEOHA-ATS). (1996). Health effects of outdoor air pollution. American Journal of Respiratory and Critical Care Medicine, 153, 3-50.

Dominici, F., McDermott, A., Zeger, S.L., & Samet, J.M. (2002). On the use of generalized additive models in time-series studies of air pollution and health. American Journal of Epidemiology, 156, 193-203.

Goldberg, M.S., Burnett, R.T., Brook, J., Bailar, J.C., Valois, M.F., & Vincent, R. (2001). Associations between daily cause-specific mortality and concentrations of ground-level ozone in Montreal, Quebec. American Journal of Epidemiology, 154, 817-826.

Hastie, T.J., & Tibshirani, R.J. (1990). Generalized additive models. London: Chapman and Hall Press.

Lee, J.T., Kim, H., Hong, Y.C., Kwon, H.J., Schwartz, J., & Christiani, D.C. (2000). Air pollution and daily mortality in seven major cities of Korea, 1991-1997. Environmental Research, 84(3) 247-254.

Ling, S., Song, G., & Zhou, F. (2001). Mortality study of major noncommunicable diseases in Shanghai, from 1951 to 1998. Chinese Journal of Epidemiology, 22, 265-268.

National Research Council. (1998). Research priorities for airborne particulate matter. Washington, DC: National Academic Press.

Peters, A., Dockery, D.W., Muller, J.E., & Mittleman, M.A. (2001). Increased particulate air pollution and the triggering of myocardial infarction. Circulation, 103, 2810-2815.

Peters, A., Doring, A., Wichmann, H.E., & Koenig, W. (1997). Increased plasma viscosity during an air pollution episode: A link to mortality? Lancet, 349, 1582-1587.

Peters, A., Frohlich, M., Doring, A., Immervoll, T., Wichmann, H.E., Hutchinson, W.L., Pepys, M.B., & Koenig, W. (2001). Particulate air pollution is associated with an acute phase response in men: Results from the MONICA-Augsburg Study. European Heart Journal, 22, 1198-1204.

Peters, A., Liu, E., Verrier, R.L., Schwartz, J., Gold, D.R., Mittleman, M., Baliff, J., Oh, J.A., Allen, G., Monahan, K., & Dockery, D.W. (2000). Air pollution and incidence of cardiac arrhythmia. Epidemiology, 11, 11-17.

Pope, C.A., Verrier, R.L., Lovett, E.G., Larson, A.C., Raizenne, M.E., Kanner, R.E., Schwartz, J., Villegas, G.M., Gold, D.R., & Dockery, D.W. (1999). Heart rate variability associated with particulate air pollution. American Heart Journal, 138, 890-99.

Stec, J.J., Silbershatz, H., Tofler, G.H., Matheney, T.H., Sutherland, P., Lipinska, I., Massaro, J.M., Wilson, P.F., Muller, J.E., & D'Agostino, R.B. (2000). Association of fibrinogen with cardiovascular risk factors and cardiovascular disease in the Framingham Offspring Population. Circulation, 102, 1634-1638.

Tolbert, P.E., Klein, M., Metzger, K.B., Peel, J., Flanders, W.D., Todd, K., Mulholland, J.A., Ryan, P.B., & Frumkin, H. (2000). Interim results of the study of particulates and health in Atlanta (SOPHIA). Journal of Exposure Analysis and Environmental Epidemiology, 10(5), 446-460.

Zanobetti, A., & Schwartz, J. (2001). Are diabetics more susceptible to the health effects of airborne particles? American Journal of Respiratory and Critical Care Medicine, 164, 831-833.

Zanobetti, A., & Schwartz, J. (2002). Cardiovascular damage by airborne particles: Are diabetics more susceptible? Epidemiology, 13, 588-592.

Haidong Kan, Ph.D

Jian Jia, M.P.H., M.D.

Bingheng Chen, M.P.H., M.D.

Corresponding Author: Haidong Kan, Assistant Professor, Department of Environmental Health, School of Public Health, Fudan University, Box 249, 138 Yixueyuan Road, Shanghai 200032, China P.R. E-mail:
COPYRIGHT 2004 National Environmental Health Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2004, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:International Perspectives
Author:Chen, Bingheng
Publication:Journal of Environmental Health
Geographic Code:9CHIN
Date:Oct 1, 2004
Previous Article:Critical thinking among environmental health undergraduates and implications for the profession.
Next Article:Critical thinking among environmental health undergraduates and implications for the profession.

Related Articles
The association of daily diabetes mortality and outdoor air pollution in Shanghai, China.
Air pollution is a serious cardiovascular risk.
Air pollution and adverse pregnancy outcomes: response.
Does particulate air pollution contribute to infant death? A systematic review.
Nitrogen dioxide increases cardiorespiratory admissions in Torrelavega (Spain).
Estimating the independent effects of multiple pollutants in the presence of measurement error: an application of a measurement-error-resistant...
Fine particulate air pollution and mortality in nine California counties: results from CALFINE.
Diabetes, obesity, and hypertension may enhance associations between air pollution and markers of systemic inflammation.
Gaseous air pollutants and hospitalization for respiratory disease in the neonatal period.
Ozone and daily mortality in Shanghai, China.

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |