Regional differences in acute hospitalization risk associated with NO by cause, season, age, sex, and trend: an ecological time series study in Canada.

BACKGROUND

Nitrogen dioxide (NO) is a highly reactive gas produced mainly from burning fossil fuels. Exposure to NO has been shown to impact public health worldwide. However, spatial and temporal variations in its effects by season, age, and sex have been underexamined.

METHODS

We conducted an ecological time-series study based on about 20 million people (52% of Canadians in 2012) in three regions (Western, Central and Eastern Canada) over 17 years (1996-2012). We collected hourly NO concentrations and temperatures, and daily counts of non-accidental all-cause, circulatory-, and respiratory-related hospitalizations, including more specific causes: ischemic heart disease, other heart disease, cerebrovascular disease, influenza/pneumonia, and chronic lower respiratory disease. We first estimated city-specific risks, applying over-dispersed generalized Poisson models, and then regional and national risks for each season, age-group, and sex using Bayesian hierarchical models. We also applied Sen's test to detect linear trends in annual regional and national risks.

RESULTS

We found significant NO effects by cause, season, age, sex, and linear trend. For circulatory hospitalization, only Western Canada showed significant adverse effects for non-seniors (≤ 65) (1.7% with 95% credible interval of 0.3-3.2% per 10 ppb increase in NO), and for males for more specific cause, ischemic heart disease (2.3%, 0.1-4.5%). Regional differences were observed for circulatory but not respiratory hospitalizations. For example, the Western and Eastern regions were at significantly higher risk of circulatory hospitalization but not the Central region: 1.6% (0.2-3.0%) for the Western region; 2.0% (0.6-3.4%) for the Eastern region; and 0.8% (-0.3-2.0%) for the Central region. In particular, the Western region had a much higher risk of cerebrovascular disease hospitalization: 2.8% (1.1-4.6%) for the Western region; 0.1% (-3.0-3.1%) for the Central region; and 0.0% (-3.4-3.5%) for the Eastern region. However, no other regional differences were observed for other causes. Overall, there were noticeable increases in regional differences over time, particularly in the later years.

CONCLUSIONS

This study indicates harmful NO effects on acute hospitalizations year-round: circulatory causes (cold season) and respiratory causes (warm season). Future work is warranted to investigate potential causes of observed regional differences using more community-related information such as socioeconomic status, health-care accessibility, and others.