SPECIAL ARTICLEjanuary 10, 2009 EPW Economic & Political Weekly56New Evidence on Child Mortality in IraqTim DysonThis paper examines new evidence on the level and trend of child mortality in Iraq. Until recently, it has generally been thought that there was a sharp rise in the level of child mortality in the country during the early 1990s as a result of the first Gulf war and the accompanying United Nations economic sanctions. The main basis for this view was a survey conducted in 1999. However, estimates of the level and trend in child mortality are now available from two additional surveys. Neither of the new sets of estimates show any sign of a sharp increase in child mortality in the early 1990s. Therefore it seems probable that, as was suggested by a report in 2005, the 1999 survey data were deliberately manipulated by the then government of Iraq.The level and trend of child mortality in Iraq has been a subject of controversy. The key issue relates to the longer-run effects of the events of 1990 – in particular, the conse-quences of the economic sanctions imposed by the United Nations Security Council in August of that year, following the Iraqi inva-sion of Kuwait. Few doubt that the United Nations (UN) economic sanctions contributed to the very hard living conditions that were experienced by the Iraqi people. Moreover, the effects ofthe sanctions came on top of damage done by the first Gulf war which led to the removal of the Iraqi army from Kuwait. The economic sanctions affected all aspects of life – e g, food, health, water, sewage, employment – and they were accompanied by reports of widespread suffering. As a result, there was world-wide debate about who was responsible for the distress – Saddam Hussein’s regime, which clamoured for the sanctions to be lifted, or governments (e g, those of the United States (US) and United Kingdom (UK)) that most supported the sanctions. Anyhow, the international debate eventually led to the estab-lishment of the United Nations Oil for Food Programme (OFFP). The OFFP was designed to limit suffering among the Iraqi people. And from about 1998 onwards the programme began to deliver significant supplies of food and medicine to Iraq, and living conditions began to improve as a result. Also, as the 1990s progressed so the government of Saddam Hussein became increasingly adept at circumventing the sanctions. The economic sanctions were eventually lifted in 2003, following theUS/UK invasion of Iraq.A Sketch of the ControversyIn large measure, the controversy about child mortality has arisen from a survey that was conducted by United Nations Children’s Fund (UNICEF) in 1999, with cooperation from the government of Iraq (GOI). The survey was conducted in two parts. The first part was undertaken in the south and centre of the country containing about 80% of the population – this was the area where Saddam Hussein’s regime retained control. The second part was undertaken in the then autonomous northern region containing about 20% of the population. The 1999 survey collected “birth histories” from women aged 15-49 living in a representative total sample of about 38,000 households. Collecting a birth history involves asking a woman to give detailed information about each of her live births separately – in particular, the child’s date of birth, and if the child has died its date/age at death. Provided that the sample is sufficiently large, and theinterviewersareprop-erlytrained,thenthe collection of birth histories allows analysts toconstructafairlydetailedpicture of the past level and trend of child mortality in a population. I thank Mike Murphy for advice. The usual disclaimer applies.Tim Dyson (t.dyson@lse.ac.uk) is at the Development Studies Institute, The London School of Economics and Political Science, London. 
SPECIAL ARTICLEEconomic & Political Weekly EPW january 10, 200957The birth history data collected by the UNICEF/GOI survey in 1999 were examined by an expert review panel, and were gener-ally thought to be of reasonable quality. Therefore UNICEF stated that it was satisfied with the basic integrity of the data (UNICEF 1999). An initial study of the data appeared in the Lancet(Ali and Shah 2000) and a more detailed analysis was produced by Ali, Blacker and Jones (2003). The key conclusion from this research was that the level of child mortality – i e, the probability of a child dying before reaching its fifth birthday – had risen very sharply between 1990 and 1991. Indeed, the data suggested that whereas during 1986-90 the child death rate had averaged about 63 child deaths per thousand live births, in 1991-95 it had increased to about 114 per thousand and had then stayed at roughly that level. In short, the survey data suggested that the under-five death rate in Iraq had almost doubled. This doubling pertained chiefly to the south and centre of the country. In the autonomous northern region, while the under-five death rose in 1991, by 1993 it had reverted to its previous level.2005 WorkingGroupIn 2005, however, the report of a working group appeared which questioned whether the general level of child mortality had risen sharply in 1991 (Working Group 2005). The working group had been established by the Independent Inquiry Committee set up by the thenUN Secretary General, Kofi Annan, to investigate the affairs of the OFFP. The working group report stated that it could not conclude anything with much confidence. But, using data that the working group had been able to get from the 1997 Iraqi census, the report questioned whether there had been a sharp rise in child mortality between 1990 and 1991, and it tentatively raised the possibility that the government of Saddam Hussein might have tampered with the data collected by the 1999 UNICEF/GOI survey in the south and centre of the country. The working group estimated that the under-five mortality rate in Iraq during 1986-90 was about 85 per thousand, and that during 1991-96 it might have risen to around 95 per thousand. The data that the working group had been able to get from the 1997 Iraqi census were those arising from so-called children ever born/children surviving (CEB/CS) questions. These questions involve asking adult women to state the number of children they have ever borne, and the number of these children that are still surviving. The questions provide no information on the dates of birth of the children, or on their date/age at death. The results fromCEB/CS questions can be cross-classified by the ages of the women respondents in order to obtain estimates of the level and trend of child mortality. However, analysing CEB/CS data is not straightforward, and it involves making some important assump-tions. In addition, the 1997 Iraqi CEB/CS census data posed special problems for analysis, because there was a substantial level of non-response to the question on the number of surviving children. Nevertheless, in arriving at its estimates of child mortality, and the suggestion that the 1999UNICEF/GOI survey data might have been tampered with, the working group placed considerable weight on its own analysis of the 1997 CEB/CS census data.About a year after the appearance of the working group report, I published an article on the subject in this journal (Dyson 2006). A key point of the article was that because CEB/CSdataarenecessarilyaggregatedgreatly over both age and time, and because the data collected by the 1997 Census had special problems, the working group’s analysis of these data constituted a weak basis on which to dismiss the occurrence ofasharprisein the level of child mortality in Iraq starting from 1991. Accordingly, I regarded the working group’s suggestion that the 1999 data had been tampered with as fanciful. I also made use of data collected by the Iraqi Living Conditions Survey (ILCS) of 2004 – data that were largely disregarded by the working group. The ILCS was another large and nationally representative survey. It collected birth histories from women living in about 21,000 households. The survey was financed by United Nations Development Programme (UNDP), and run jointly by Iraq’s Central Organisation for Statistics and Information Technology (COSIT) and the Norwegian research institute FAFO (UNDP 2005a,b,c). The ILCS was a general socio-economic survey in which the collection of birth histories formed only one part. However, I noted that for years 1991-98 there was some similar-ity in the trend and the pattern of annual variation in the child death rates produced by both the ILCS (in 2004) and the UNICEF/GOI survey (in 1999). Therefore, because the 1997 CEB/CS data were a weak basis from which to dismiss the sharp rise in child mortality indicated by the 1999 survey, I prorated the ILCS under-five death rates for 1991-98 upwards to the level indi-cated by the results of the UNICEF survey for these same years. A plausible explanation for why women interviewed in the ILCS had apparently failed to mention the deaths of children who had died in the years since 1991 seemed to be that the names of many of these dead children had remained on ration cards and were still being used to obtain supplies of food and other basic items.The working group report also produced a response from Blacker, Ali and Jones – i e, the researchers who had published the original detailed annual estimates of child mortality using the 1999 UNICEF/GOI survey data (Blacker, Ali and Jones 2007). Among other things, Blacker and colleagues also emphasised the major weaknesses of using the 1997CEB/CScensus data as the basis for dismissing the occurrence of a sharp rise in the level of child mortality. And in this context they noted that the particular fram-ing of the CEB/CS questions used by the 1997 Census was both outmoded and unsatisfactory. More importantly, however, Blacker and colleagues showed that if the dates of birth and death contained in the birth history data obtained from the 1999 survey were ignored for purposes of analysis – so as to make these data comparable to the 1997 census data – then the resulting estimates of the level and trend of child mortality in Iraq fell within the range of estimates pro-vided by the working group’s own analysis of the 1997 Census data (Blacker, Ali and Jones 2007: 8-10). In essence, the point Blacker et al were making is that the 1997 CEB/CS census data are not obviously inconsistent with the indications of a sharp rise in child mortality indicated by the 1999 UNICEF/GOI survey data. These analysts also questioned whether Iraq had the tech-nical expertise to tamper with the UNICEF survey data and escape detection.
UNICEF/GOI MICS 3 
-30 -20 -10 0 10 20 30 40 50 1980 1985 1990 1995 2000 60
-40
