Article Text

Death stroked by dusty air: more mysteries to be solved
  1. J-C Chen
  1. Correspondence to:
 Dr J-C Chen
 Department of Epidemiology, UNC-CH, School of Public Health, 2104G, McGavran-Greenberg, CB#7435, Chapel Hill, NC 27599-7435, USA; jcchen{at}

Statistics from

Commentary on the paper by Yamazaki et al (see page 17)

After several reports in the early 1990s showing excessive mortality associated with unexpectedly low concentrations of particulate matter, substantial research efforts have been devoted to investigating particulate matter health effects and elucidating the related pathophysiological mechanism. Despite remaining knowledge gaps, epidemiological and toxicological studies have provided reasonably coherent evidence supporting the significance of particulate matter-mediated acute cardiovascular effects.1 Among the mounting volume of literature on air pollution epidemiology, multi-city ecological studies using time-series or case-crossover analyses have reported fairly consistent associations between increased concentrations of particulate matter over short periods of one or several days and increased cardiovascular mortality and morbidity (eg, hospitalisations and emergency visits). However, scientific data linking air pollution to stroke, including ischaemic stroke and intracerebral haemorrhage (ICH), remain relatively scant and inconclusive.2

In this issue, Yamazaki et al3 present a case-crossover study examining the acute effect of particles on mortality from stroke among elderly people (aged ⩾65 years) across 13 Japanese cities in 1990–1994. They found that high hourly particulate matter (PM7 >200 μg/m3), measured 2 h before death and independent of the 24-h particulate matter, was associated with increased mortality from ICH (ICD-9: 431). Compared with previous air pollution time-series studies or case-crossover analyses relying only on daily exposure data, although all ecological in nature, this study had two major strengths. Firstly, given data on “timing” of the presumed acute event, Yamazaki et al prudently applied time-stratified selection of control periods matched to the clockhours. This additional matching effectively adjusted for potential confounding by circadian variations of unmeasured physical activities, which may determine exposures to particulate matter and temporally correlate with the onset of either cardiovascular4 or cerebrovascular events.5 Secondly, except few panel or individual-based case-crossover studies,6,7 the rich literature on air pollution epidemiology provides only limited corroborative insights for experimental studies attempting to delineate the short-term course of particulate matter effects.8,9 By contrast, the availability of data on hourly pollutants matched to time of death allowed the authors to explore temporal profile analyses of mortality from stroke associated with both daily and hourly exposures to particulate matter.

Interestingly, the illustrated time dependence of particulate matter effects on mortality from stroke seemed to be present only in warmer months (April–September), especially for ICH. Daily particulate matter was associated with statistically non-significant increases in both ICH and ischaemic stroke mortalities, and such associations disappeared with exposures lagged for >12–35 h (fig 1). Hourly concentrations of particulate matter with various lags of up to 30 h were associated acutely with increased mortality from stroke, with many reaching statistical significance for ICH cases (fig 3). The variation of hourly particulate matter effect within the 24-h before mortality from ICH calls for a distributed lag structure10 to jointly model this dynamic temporal profile, which speaks to the probably invalid assumption of “equal weighting” (equation 1) implicated for estimating 24-h (lag0–23) particulate matter effect.

On the other hand, both the shown temporal dynamics and the magnitude of effect estimate (adjusted odds ratio of 2.4 associated with PM7 >200 μg/m3) are intriguing and appropriate interpretation is challenging. Clearly, the time of death recorded on death certificates as used in this study should not be equated to the onset of either ICH or ischaemic stroke. Although arguably high exposure to particulate matter may trigger abrupt changes to sympathetic tone which cause rupture of large cerebral aneurysms and consequently massive subarachnoid haemorrhage (SAH) leading to sudden death in hours, it cannot explain the described particulate matter–ICH association because victims with SAH (ICD-9: 430) were excluded. One related hypothesis is that high exposure to particulate matter may cause rapid-accumulating infratentorial haemorrhage, with subsequent brain-stem herniation and sudden death. However, these severe forms of ICH account only for a small portion of clinical cases, hardly explaining the statistically significant and large effect size observed in this study. Therefore, the reported acute effect of hourly particulate matter on increased mortality from stroke should not be interpreted as the onset of fatal stroke triggered by short-term increases in ambient particulate matter. Although growing evidence on particulate matter toxicity may provide putative pathways (eg, vascular inflammation, endothelial dysfunction and hypercoagulation) linking particulate matter to ischaemic stroke, the biological plausibility for the alleged direct effect of particulate matter on ICH is still lacking. Alternatively, findings of Yamazaki et al may suggest that high exposure to particulate matter triggers other acute deadly events in subjects with existing ICH (or ischaemic stroke) but without the real causes of death being identified in their death certificate. This assumed indication of susceptibility to particulate matter-mediated excessive mortality in patients with stroke is consistent with recent findings11 and is also supported by the propensity to cerebrogenic sudden death in stroke.12

The report by Yamazaki et al also left us other scientific uncertainties and unanswered questions: spatial heterogeneity, the roles of gaseous pollutants and the interaction with temperature for the air pollution-mediated effect on cerebrovascular diseases, to only name a few. Interested readers shall wish to see more uncertainties regarding air pollution and stroke demystified in the years to come.

Commentary on the paper by Yamazaki et al (see page 17)


View Abstract


  • Competing interests: None.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles