Maternal death stat questioned

via CAAI News Media

Wednesday, 21 April 2010 15:04 Irwin Loy

HOW do you measure maternal deaths? That is the question health experts in Cambodia are facing in light of new research indicating that the rate of women dying during pregnancy and childbirth may be far lower than previously thought.

A study published in the medical journal The Lancet last week suggests that the maternal mortality rate (MMR) has dropped significantly, both globally and in Cambodia. It pegged Cambodia’s MMR at 266 deaths for every 100,000 live births in 2008, far lower than the government’s figure of 461, which is based on 2008 census data.

This number appears to contradict the widely held belief that little statistical progress has been made in curbing the preventable deaths. And domestic health experts, though buoyed by what appears to be positive news about a problem that has long proved intractable, are debating whether it reflects the true nature of the challenge on the ground.

The issue of how to estimate maternal deaths – particularly in developing countries where accurate data may not consistently be recorded – has long been the subject of debate. In this case, the study’s researchers appear to have differed from the government in their view of what should be included in maternal death totals, which provides at least a partial explanation for the lower MMR, experts say.

The study’s researchers examined data on women who died during pregnancy, childbirth or in the 42 days following delivery. They excluded “late maternal deaths”, or women who died between 42 days and one year after delivery.

The government’s figures, on the other hand, include maternal deaths that occurred within two months, or 60 days, after birth. They also differ from the Lancet figures in that they include all women who died during this timeframe, even if their deaths weren’t directly related to their pregnancies.

“The methodology is completely different,” said Chan Theary, executive director of the Reproductive and Child Health Alliance. “They only include specific categories. In Cambodia, we include them all.”

However, she acknowledged it was unclear how these definitions alone would have affected the differing results.

An explanation accompanying Cambodia’s 2005 national health survey – the last major collection of health data in the Kingdom – asserts that the decision to include deaths not directly related to pregnancy was unlikely to inflate maternal death statistics “because most deaths to women in the specified period are due to maternal causes and maternal deaths in general are more likely to be under-reported than over-reported”.

Beyond these differences, researchers who worked on the Lancet study said new statistical modelling allowed them to produce more accurate data. While vital statistics records are seen as the most useful resource for measuring maternal deaths, many developing countries lack complete records. Cambodia’s MMR figure, for example, is based on survey data for which women are asked to give a history of deceased siblings and to identify sisters who died during pregnancy or after childbirth.

That method produces errors that need to be compensated for, the Lancet study’s co-author, Dr Christopher Murray, told Time magazine.

“You can only ask living women about their sisters, which creates survivor bias,” said Murray, the director of the Institute for Health Metrics and Evaluation at the University of Washington.

“Statisticians figured out a way of solving the problem, and we figured out a way to apply that to these surveys, which also enables us to get measurements going back in time.”

The researchers then extrapolated Cambodia’s data along with that of 180 other countries to produce MMR estimates going back to 1980.

Even if the lower rate recorded in the study could be attributed to methodological differences, the estimates still suggest an overall drop in Cambodia’s MMR, from 511 in 2000 to 266 in 2008 – a downward trend that isn’t found in previous estimates.

Chan Theary said authorities producing the next major health survey, scheduled this year, should take into account the Lancet methodology.

However, she also wants officials to temper expectations that may be raised by the study’s findings.

“I hope people will not be so happy with the figure that they do not pay any more attention to the causes of maternal deaths,” she said. “Whatever the result is [of the pending health survey] in 2010, we have to work hard to make it improve to the level of developed countries. These countries have [MMRs] of only 50 or 20 or even less. We would like to reach that figure.”