Pregnant women in India, Pakistan, Bangladesh, who are exposed to poor air quality, may be at higher risk of stillbirths and miscarriages, according to a modelling study published in The Lancet Planetary Health journal.
Researchers found that an estimated 349,681 pregnancy losses per year in south Asia were associated with exposure to PM2.5 concentrations that exceeded India's air quality standard of 40 micrograms per cubic metre of small particulate matter (PM2.5).
These account for 7 per cent of annual pregnancy loss in the region from 2000-2016, they said.
For air pollution above WHO air quality guideline of of 10 micrograms per cubic metre, exposure may have contributed to 29 per cent of pregnancy losses, according to the study.
"South Asia has the highest burden of pregnancy loss globally and is one of the most PM2.5 polluted regions in the world," said study lead author Tao Xue, from Peking University, China.
"Our findings suggest that poor air quality could be responsible for a considerable burden of pregnancy loss in the region, providing further justification for urgent action to tackle dangerous levels of pollution," Tao Xue said.
Tianjia Guan, from the Chinese Academy of Medical Sciences said that losing a pregnancy can have knock-on mental, physical and economic effects on women.
These include increased risk of postnatal depressive disorders, infant mortality during subsequent pregnancy, and increase the costs related to pregnancy, such as loss of labour.
"Therefore, reducing pregnancy loss may also lead to knock-on improvements in gender equality," Tianjia Guan, one of the authors of the study said.
The researchers noted that pregnancy loss associated with air pollution was more common in the Northern plains region in India and Pakistan.
Although the total burden of pregnancy loss was predominantly borne by rural women aged under 30 years old in recent years, the burden attributable to PM2.5 also affected older mothers, aged 30 years or over, in rural areas because of their high susceptibility to the adverse effects of pollution, they said.
The team combined data from household surveys on health from 1998-2016 and estimated exposure to PM2.5 during pregnancy through combining satellite with atmospheric modelling outputs.
They created a model to calculate the risk of pregnancy loss for each 10 micrograms per cubic metre increase in PM2.5 after adjusting for maternal age, temperature and humidity, seasonal variation, and long-term trends in pregnancy loss.
The researchers calculated the number of pregnancy losses that may have been caused by PM2.5 in the whole region for the period 2000-16 and looked at how many pregnancy losses might have been prevented under India's and WHO's air quality standard.
In the study, they included 34,197 women who had lost a pregnancy, including 27,480 miscarriages and 6,717 stillbirths, which were compared to livebirth controls.
Of the pregnancy loss cases, 77 per cent were from India, 12 per cent from Pakistan, and 11 per cent from Bangladesh.
The researchers said that gestational exposure to PM2.5 was associated with an increased likelihood of pregnancy loss, and this remained significant after adjusting for other factors.
Each increase in 10 micrograms per cubic metre was estimated to increase a mother's risk of pregnancy loss by 3 per cent, they said.
The increase in risk was greater for mothers from rural areas or those who became pregnant at an older age, compared to younger mothers from urban areas.
Although WHO's guidelines aims for a safer level of air pollution, the researchers note that India''s standard is a more realistic target level, given the high average levels of air pollution in the region and the need to balance practical governance and public health.
The researchers noted several limitations of their study.
In the surveys, they were not able to distinguish between natural pregnancy loss and abortions and there was under-reporting of pregnancy losses because of stigma or ignoring very early pregnancy losses.
They also note that the survey data is subject to recall bias, therefore, recommending the causality of the association should be further examined in longitudinal studies.