Reviews of pollution literature generally conclude that there is at least a weak link between birth outcomes, such as a baby’s weight, and exposure to particulate matter with a diameter less than 2.5 µm (PM2.5). Such pollution consists of elemental carbon, which is emitted by traffic, vanadium, which is generated in the combustion of oil, and various other chemicals.

But individual studies on pollution report inconsistent results. In California, for instance, pollution has been linked to pregnancy term and birth weight, while in Norway no link to birth weight has been found. One potential reason for the inconsistency could be variations in the chemical makeup of the PM2.5 pollution. In the US, such contaminants contain more sulphate in the east of the country, for example, and more nitrate in the west.

An alternative explanation could be a variation in the methods used to judge exposure to pollution. Exposure is often estimated by taking measurements from fixed monitors, and assuming that those measurements hold within a certain radius, or buffer. Scientists would prefer these buffers to be large, to increase their sample sizes and so maximise the strength of their statistics. But within a buffer zone, populations vary: they might include more people of a certain socio-economic status, at the expense of others. What’s more, the pollutants themselves can vary within a buffer zone – some tend to be spread evenly, while others may be patchy.

With this in mind, Keita Ebisu and colleagues at Yale University, US, examined the potential links between PM2.5 pollution and birth weight while taking into account different chemical constituents and buffer sizes. The researchers calculated the distance to a pollution monitor for a large sample of mothers in the state of Connecticut and then performed a traditional analysis but chose seven different buffer sizes.

In general, the team estimated that birth weight fell by 9.6 g per 1.76 µg increase in total PM2.5 pollution. More specifically, the researchers could link 10 chemical constituents to birth weight, although some had stronger links than others. A 0.3 µg rise in elemental carbon, for instance, corresponded to a 16.8 g fall in birth weight.

Yet the choice of buffer also made a difference, at least for some pollutants. Nitrate could be linked to significantly higher birth weights within a 10 km buffer, but significantly lower birth weights within a 30 km buffer.

"This could imply exposure misclassification happened with larger buffers," said Ebisu. "This, however, does not mean that we encourage using small buffers." Small buffers weaken statistics, Ebisu added, and could also bias results towards populations who live in more urban areas where the monitors are located.

Ebisu believes that there could be implications for policymakers. "In the US, PM2.5 total mass is regulated, but not specific PM2.5 chemical constituents," he said. “Our findings indicated that some constituents are more harmful than others. Further studies in other areas are warranted, [and] improving [the] monitoring network is desired.”

The next step is to estimate more exposures with satellite imagery, which is more accurate, Ebisu says.

The team reported the results in Environmental Research Letters (ERL).

Related links

• Keita Ebisu, Yale School of Forestry and Environmental Studies
• Association between airborne PM2.5 chemical constituents and birth weight—implication of buffer exposure assignment Keita Ebisu et al 2014 Environ. Res. Lett. 9 084007

Related stories

• Air pollution lowers birth-weights
• UMass Amherst research finds asthma rate and costs from traffic-related air pollution are much higher than once believed
• Climate change brings small rise in air pollution deaths