Infant mortality and the Lancet study

http://www.unicef.org/infobycountry/iraq_statistics.html

Unicef clearly indicates here, which year data are for eg "% of population using improved drinking water sources 2000", or "Number per 100 population 2001" (phones).

I have written to them to defend their number for 2002 and so far they haven't replied. Maybe, if others join me, we'll get an answer out of them.

But writing 2002, and meaning 1994-1999, while in loads of other cases, on the same page, clearly stating the time periods data apply for, is pretty substandard, if that is really what Unicef has done.

So, I want to know how they estimated their number for 2002, in detail.

http://www.unicef.org/publications/files/pub_children_of_iraq_en.pdf

There is plenty of information here, including on the causes of infant mortality (70% is due to diarrhea and acute respiratory infections) and trends in malnutrition, which they split into three categories (acute, underweight and chronic).

Interesting enough, while acute malnutrition in 2002 is substantially below the figure for 1999 (4% compared to 9.3%), chronic malnutrition in 2002 is actually listed at 23.4%, and above the 20.4% given for 1999.

So, this measure actually shows deterioration between 1999 and 2002, rather than the improvement you need to explain away the Unicef number for infant mortality.

Furthermore, all their measures for malnutrition in 2002 were worse than in 1991 and their figures indicate that Iraqi infant mortality never was below 50, even at the best of times.

The figure for the year 1999 is from page 21 of their report. The following year chronic malnutrition is listed as 30%.

Also, the figure of 102 for infant mortality in 2002 is, in fact, lower than the figure of 107 given earlier in that report for the period 95-99, or elsewhere on the Unicef site for the year 2001.

I don't see why they shouldn't have taken account of the improvement in acute malnutrition in their estimate. They must have used some method to come up with a number of 102 for 2002, while giving different and slightly higher numbers for earlier years.

How did they do it though?

If you think the study is such a hack job, then go do one yourself.Think this kind of work is easy? Do a study yourself. If you're reallyreally lucky, no one will criticize your numbers [but that will mean no one read your paper, which is unlucky].NO study is perfect. We do not have perfect knowledge. Yet every day we slog on, despite the little critics who must criticize everything, announcing a picnic discovery when stumbling upon the tiniest crumb.Best,D

Hi Dano,

what I am doing here is defending Unicef (a job they've failed to take up themselves so far). They say infant mortality was 102 in 2002, and they say that very clearly. On the same page, they give other numbers and equally clearly state that they apply to other years.

Not only that, they give a different number for infant mortality in 2001 or the period 95-99.

Tim uses the acute malnutrition data to try to rubbish the Unicef estimate for 2002, yet, in 2002 all malnutrition (acute, underweight or chronic) was worse than in 91 and Unicef say that infant mortality 90-94 was 79 and 85-89 47 (sorry by the way for leaving out the word substantially earlier, it should have said infant mortality never substantially fell below 50).

Not to mention that (the rather volatile) chronic malnutrition data for 2002 were actually worse than for 1999.

Let me suggest how this can be resolved:

The Lancet study undercounted infant mortality pre-war substantially, and that makes their excess mortality ON THAT SCORE ONLY dubious.

I am sorry that I used the rhetorical phrase (if you don't believe this number, why believe any other).

Violent death (also when, as I think one should, excluding the Fallujah cluster entirely) clearly has increased, and several ten thousand Iraqis have been killed, with attribution and classification of the deaths (combatants/civilians?, killed by ordinary crime, the coalition or terrorists?) a matter of dispute that the study doesn't really adequately address.

What gets me annoyed is that what's out there is the meme "100,000 civilians, mostly women and children, killed by the coalition", I've seen that kind of summary of the Lancet study aplenty.

What I haven't seen is an anti-war blogger rectify this perception, apparently them plain refusing to acknowledge that 99.99% of people won't read the study and need to hear an accurate and fair summary, which I'd say would be something like:

The study provides very strong evidence, as does, on balance, all the other available evidence, that there have been tens of thousands of violent deaths of Iraqis after the invasion, and that this represents a substantial increase compared to the last few years under Saddam, when there was no large-scale fighting or the kind of mass grave filling murder that occurred in earlier years.

It also indicates that mortality ex violence appears to have been broadly stable, with the possible exception of infant mortality, a topic which deserves further study.

The number of recorded violent deaths was too small to accurately extrapolate a breakdown into civilians/combatants or along lines of responsibility, such as terrorists/ordinary crime/coalition forces, except that is in an outlier sampling location, Fallujah, where a large number of children is reported as having died from coalition bombing. Both extrapolation from this location to the rest of Iraq and the reliability of the interviews in Fallujah are questionable.

Heiko, I've used the acute malnutrition figures because those are the children who are at great risk of dying. I think they should correspond to child mortality rates.
The pre-sanctions figure of 3% is almost the same as the 2002 figure of 4%, so we would expect the infant mortality rate for 2002 to be similar to the pre-war level of about 50. The 95% CI for the Lancet study on prewar infant mortality is 0-64, so it certainly is not contradicted by an estimate of about 50.
I agree that "100,000 civilians killed by the coalition" is an inaccurate summary, but I don't agree with yours. The increase in child mortality is strongly supported by the malnutrition data and it is reasonalbe to conclude that it has approximately doubled.

Heiko: "What gets me annoyed is that what's out there is the meme "100,000 civilians, mostly women and children, killed by the coalition", I've seen that kind of summary of the Lancet study aplenty."

Then shouldn't you be attacking the misrepresentation of the study rather than the study itself? Additionally, if the invasion resulting in an increase in mortality odes it matter whether its due to bombs, bullets or the collapse of medical services?

In fairness to Heiko, the misrepresentation of the "Lancet study" was carried out by the Lancet itself. The actual study made no reference to "civilians", but the editor of the Lancet added an editorial comment which implied that the 98,000 figure referred only to civilian deaths. As far as I'm aware, nobody has tried to make the claim "mostly women and children", which would be at odds with the plainly expressed findings of the study.

The figures for malnutrition mentioned in the Unicef report are for the year 1991, ie not pre-sanctions.

The period 90-94, in which those figures fall, is given by Unicef as having an infant mortality of 79 (of course, that could be dominated by the years 92, 93 and 94).

http://www3.who.int/whosis/country/indicators.cfm?country=irq

http://www.who.int/whr/2003/en/Annex1-en.pdf

"To capture the uncertainty resulting from sampling, indirect estimation technique or projection to 2002, a total of 1000 life tables have been developed for each Member State. Uncertainty bounds are reported in Annex Table 1 by giving key life table values at the 2.5th percentile and the 97.5th percentile. This uncertainty analysis was facilitated by the development of new methods and software tools (5)."

http://www3.who.int/whosis/discussion_papers/discussion_papers.cfm?path…

One can follow the last link to download a number of discussion papers detailing the methods employed by WHO and Unicef.

"Where the evidence is uncertain or incomplete, WHO attempts to make the best possible inferences based on the
knowledge base that is available, and to assess the uncertainty in the resulting estimates."

And even in the latest press release that also mentions nutrition deteriorating, you'll still find the following sentence:

http://www.unicef.org/media/media_24233.html

"She noted that Iraq already had severe problems with malnutrition, water service and sanitation before the war, when 1 in 8 Iraqi children died before the age of five."

The pre-war baseline for infant mortality is a pretty important piece of information. At 29, it'll be so much harder to get an improvement than at 102.

Unicef and WHO either got that figure completely wrong, or other factors apart from malnutrition are more significant in the case of Iraq, and therefore the improved malnutrition was only responsible for a minor improvement in infant mortality (from 107 to 102, rather than to 50, let alone 29).

Finally, the authors of the Lancet study themselves point out that they may have undercounted infant mortality, particularly further in the past.

"In other settings, under-reporting of neonatal and infant deaths in similar surveys has been documented.18,19 In
particular, the further back in time the infant death
occurred, the less likely it was to be reported. The recall
period of this survey, 2.7 years, was longer than most
surveys of crude mortality. Thus, infant deaths from
earlier periods might be under-reported, and recent
infant deaths might be more readily reported, producing
an apparent but spurious increase in infant mortality."

So, what I'd like to have is a reply from Unicef clarifying whether their estimate for 2002 is compatible with the nutrition data, and whether these particular survey data impact their confidence in their estimate.

Hi dsquared,

the following sentence is in the summary of the Lancet study:

"Most individuals reportedly killed by coalition forces were women and children."

I figure that includes Fallujah, and excludes some deaths of men. Otherwise the figures don't add up:

24 children killed in Fallujah, 3 women and 25 men

4 children killed outside Fallujah, 13 men, 2 women and 2 elderly

All the men together add up to 38, while women, children and the elderly only sum to 35. But 12 violent deaths weren't attributed to coalition forces, so it could be 26 men killed by the coalition against 35 women and children (all the elderly could be women).

Of course, to get to that conclusion, you have to accept that the data from the Fallujah cluster can be extrapolated to the rest of Iraq, and that the authors weren't lied to.

dsquared wrote:
nobody has tried to make the claim "mostly women and children", which would be at odds with the plainly expressed findings of the study.
the study said:
"Most individuals reportedly killed by coalition forces were women and children."
oops, but then I keep forgetting; justifying this study is about belief, assumption and saying anything to justify that this was a bad war. Don't imagine that this is science.

Hi Tim,

on your final question. Yes, it matters an enormous deal, whether excess mortality is due to

a) terrorists, criminals and Baathists being killed by coalition forces

b) innocent civilians being killed by coalition forces

c) innocent civilians being killed by terrorists/Baathists

or

d) there is an increase in general mortality due to insecurity

Please note the following sentence:

"Despite the current decline, child malnutrition malnutrition levels are still higher than in 1991, and those were already elevated after a year of sanctions."

Also, the figure of 47 is given for the period 1985-1989, not for the year 1990, as far as I can see.

Nor is the figure of 108 for the year 1999, but rather it is for the period 1994-1999.

You could amend your table accordingly.

http://www.unicef.org/publications/files/29652L01Eng.pdf

The above is an interesting recent Unicef publication. They point to 11 million totally preventable child deaths that still occur each year worldwide.

They also give, unsurprisingly, 2002 under 5 mortality in Iraq as 125, and point to a variety of reasons, malnutrition indeed featuring prominently.

Sorry, I wanted to say, "and point to a variety of underlying causes of infant mortality in general, malnutrition indeed featuring prominently."

They don't say anything about the factors affecting Iraq specifically here.

Heiko,

Please define terrorism for me, will you? According to official U.S. lingo, it is "politically motivated violence, perpetrated against innocents". By this definition, "coalition forces" eminently qualify. In fact, where on the west's media radar is "state terrorism" described, except when our press apply it to ODBG* (*officially designated bad guys). Its never used to describe atrocities committed by ODGG* (*officially designated good guys). The fact is that U.S. and U.K are leading terrorist states, if the definition is applied properly. What is truly Orwellian is that our plutocracies in the west are "waging a war against terror" by "waging wars of terror".

Hi Jeff,

my reply seems to have disappeared (or I made a mistake) when sending it off. My working definition of terrorism is slightly different (politically motivated violance perpetrated by substate groups against non-combatants), but I'd like to stick to the topic of infant mortality and Unicef's/WHO's numbers here.

I agree with d2. If there is an issue with how the story was toot-tooted, then the editor is to blame.

The authors did their jobs.

Now the paper is out in the world, undefended by the authors, ping-ponged about. This speaks deeply to the issues of science in our society, but is OT.

D

Jeff, I think you need to do a bit of background reading on Orwell. He once stated that pacifism is objectively pro-fascist, and also that there is no idea or concept so stupid that you won't get a bunch of academics agreeing to it.

He once stated that pacifism is objectively pro-fascist

If memory serves me correctly, he also later changed his mind on this point.

Heiko, the 47 figure for 1990 is on page 62 of the UNICEF report linked.

I wasn't aware of that before Ken- mind you, I don't think he changed his mind about the second point..... ;-)

I looked at page 19.

You haven't addressed the fact, though, that the report points out that the malnutrition data "were already elevated after a year of sanctions" (not to mention the war itself and the fighting that followed in the South, where Saddam did fill mass graves).

Wilbur,

I've done some googling, and this website has some more information on the Orwell quote.

Heiko, using your definition of terrorism means that the Stalinist, Falangist, etc. exercises of force against all domestic opponents was not terrorism, or to bring it a bit up to date, that the Baathist governmental repression of the opposition was not terrorism. Official death squads such as roamed the land in Argentina and Chile during the military governments would not be terrorists, but the state sponsored, but not official death squads of Central America might be depending on how one defined substate. OTOH, what do you do about freelancers such as the Aum Shinrikyo?

I think your definition needs a bit of work and we are left with the question of what to label the state exercise of terror on the populance.

Heiko: "What gets me annoyed is that what's out there is the meme "100,000 civilians, mostly women and children, killed by the coalition", I've seen that kind of summary of the Lancet study aplenty."

Then shouldn't you be attacking the misrepresentation of the study rather than the study itself? Additionally, if the invasion resulting in an increase in mortality odes it matter whether its due to bombs, bullets or the collapse of medical services?

Heiko: "What gets me annoyed is that what's out there is the meme "100,000 civilians, mostly women and children, killed by the coalition", I've seen that kind of summary of the Lancet study aplenty."

Then shouldn't you be attacking the misrepresentation of the study rather than the study itself? Additionally, if the invasion resulting in an increase in mortality odes it matter whether its due to bombs, bullets or the collapse of medical services?

Heiko: "What gets me annoyed is that what's out there is the meme "100,000 civilians, mostly women and children, killed by the coalition", I've seen that kind of summary of the Lancet study aplenty."

Then shouldn't you be attacking the misrepresentation of the study rather than the study itself? Additionally, if the invasion resulting in an increase in mortality odes it matter whether its due to bombs, bullets or the collapse of medical services?