There have been several media reports of apparent excess rates of cancers and birth defects in the town of Fallujah in Iraq, some 50 miles west of Baghdad [1
]. In 2004, one year after the end of the second Persian Gulf War in March 2003 there was heavy fighting between US led occupation troops and Iraqi elements in this town. Little is known about the types of weapons deployed, but reports began to emerge after 2005 of a sudden increase in cancer and leukaemia rates.
Concerns have been expressed for some time about increases in cancer, leukemia and congenital birth anomalies in Iraq. These have been blamed [4
] on mutagenic and carcinogenic agents (like depleted uranium) employed in the wars of 1991 and 2003. Increases in childhood leukaemia in Basrah have recently been investigated [5
] and the findings confirm that there has indeed been a significant increase since 1991. Unfortunately, since many reports from Iraq and Fallujah have been anecdotal, and have rarely been backed up by any population-based epidemiological evidence, it is difficult in these cases to assess the validity of the various assertions. Questionnaire survey studies have a long history of use in areas where there are difficulties obtaining accurate population numbers or illness rates [6
]. Epidemiology in post-conflict areas where official population, cancer and birth data are not available can use questionnaire survey methods developed and used earlier in a number of areas of the UK and Ireland. The method is described fully with a sample questionnaire in Busby 2006 [7
] where breast cancer rates in the town of Burnham on Sea, Somerset were reported. The study was later investigated by the official South West Cancer Intelligence Service and was shown to have given an accurate result for the breast cancer incidence rates.
For these reasons we decided to conduct such a survey study in Fallujah.
3. Results and Discussion
The population base obtained from the questionnaires was 711 households with 4,843 persons. The sex and age breakdown by 5-year groups is given in Table 1
. Reported cancers from Jan 1st 2005 to the end of January 2010 are given in Table 2
. Cancers reported before 2005 were not included. All cancer and infant death cases reported were checked against duplicate sex and age patterns to ensure there was no double reporting; if there was any doubt, data was discarded (one such instance was found). Table 3
shows the infant mortality cases reported from 2004 and includes reports of deaths in the first two months of 2010.
In Table 4
the reported numbers of cancers are compared with expected numbers for 5-year period 2005 to the sample cutoff date in 2010. The expected numbers are calculated by applying the sex and 5-year age group rates obtained from the Middle East Cancer Consortium [8
] for Egypt 1999 and also checked against rates in Jordan [9
shows the mean infant mortality rate per 1000 births for the period 2006–2010 including deaths reported in the first two months of 2010. Also shown are rates for the period from 1st January 2009 and comparisons are made with infant mortality rates in Jordan, Egypt and Kuwait.
The responses show that there is an anomalous sex ratio in the 0–4 age group. There are 860 males to 1000 females, a significant 18% reduction in the male births from the normal expected value of 1,055 (267 boys expected, 234 observed; p < 0.01) Perturbation of the sex ratio is a well known consequence of exposure of mutagenic stress and results from the sensitivity of the male sex chromosome complement to damage (the females have two X chromosomes whereas the males have only one). A number have studies have examined sex-ratio and radiation exposure of mothers and fathers. Of relevance is the study of Muller et al.
] of the offspring of 716 exposed fathers who were Uranium miners. There was a significant reduction in the birth sex ratio (fewer boys). Lejeune et al.
] examined the offspring of fathers who had been treated with pelvic irradiation; at high doses there was an increase in the sex-ratio, but this reversed in the low doses (around 200 mSv). Schull et al.
] found a reduction in the sex ratio in A-Bomb survivor fathers (mothers “unexposed”) for children born 1956–1962 a reversal of an earlier finding by Schull and Neel 1958 [14
] of a positive effect in the 1948–1955 births. It should be noted that there were external and internal irradiation effects in these groups, with the internal effects predominating in the later years. Yoshimoto et al.
] found an overall reduction in the sex ratio for A-Bomb survivors for children born 1946–1984. Thus the evidence suggests that exposure to ionising radiation at low doses and specifically exposure to Uranium may cause a reduction in the sex ratio.
It is clear that the 0–4 population, born in 2004–2008, after the fighting, is significantly 30% smaller than the 5–9, 10–14 and 15–19 populations. This could be a result of lower fertility or early foetal losses in this cohort. It has been pointed out by a referee that it might also in principle be a result of the deaths of men in the 2004 fighting but this does not seem to be supported by the sex ratios in the men and women aged 25 and over. The infant mortality numbers reported by year point to sudden increase in deaths in 2006 (Table 3
). There was only one death reported for the two years 2004 and 2005 in the sample population. For the period from 2006 to the end of the survey there was a mean death rate of 80 per 1,000 births, more than 4 times the rate in Egypt and in Jordan (p < 0.00001) and some 9 times the rate in Kuwait. The rate seems to have increases markedly after 2009 to a rate of 136 per 1,000 births. These results support the many reports of congenital illness and birth defects in Fallujah and suggest that there is evidence of genetic stress which appeared around 2004, one year before the effects began to show.
The results for cancer show some alarming rates in the 5-year period. Relative Risk based on the Egypt and Jordan cancer rates are significantly higher for all malignancy, leukaemia, lymphoma, brain tumours and female breast cancer. Between January 2005 and the survey end date there were 62 cases of cancer (all malignancies) reported (RR = 4.22; CI: 2.8, 6.6; p < 0.00000001) including 16 cases of childhood cancer 0–14 (RR = 12.6; CI: 4.9, 32; p < 0.00000001). Highest risks were found in all leukaemias in the age groups 0–34 (20 cases RR = 38.5; CI: 19.2, 77; p < 0.00000001), all lymphomas 0–34 (8 cases, RR = 9.24;CI: 4.12, 20.8; p < 0.00000001), female breast cancer 0–44 (12 cases RR = 9.7;CI: 3.6, 25.6; p < 0.00000001) and brain tumours all ages (4 cases, RR = 7.4;CI: 2.4, 23.1; P < 0.004). These results for cancer also support the idea that there has been exposure to some mutagenic agent at some time in the past. Could this have been around 2004 when the fighting occurred? The answer depends upon whether it is plausible to accept such a short time lag between exposure and clinical expression of the cancer, leukaemia or lymphoma. It is commonly believed that the lag between initiation and expression of cancer is a significant period: for exposure to acute external low LET radiation the onset of leukaemia is stated to be about 5 to 7 years, and for breast cancer and solid tumours as high as 15 to 20 years. However, genetic damage expansion models for cancer [16
] hold that it is the acquisition of a key number of mutations which lead to final clinical expression. This is then seen as purely probabilistic so long as the mutagenic stresses are constant; in this way the exponential increases in cancer rates with age are explained as are cancer rates and initiation expression lags in cell populations with different natural replication rates. However, such an explanation makes it also clear that the sudden (spike) introduction of a mutagenic stress could supply a final key mutation in those individuals who already carry almost the full necessary complement of mutations for the specific cancer [18
]. This idea explains many observations of increases in cancer shortly (a few years) after an exposure. For example, there seems to have been a rapid increase in lymphoma in Italian peacekeepers potentially exposed to depleted uranium in the Balkans [19
]. Tondel et al.
have reported increased cancer risk in Northern Sweden peaking less than 5 years after the Chernobyl contamination and significantly associated with the levels of Caesium-137 fallout in municipalities [20
]. Despite the assertions of the studies of the Japanese A-Bombs (which did not begin until 1952) that the first increases in leukaemia in the study group appeared more than 5 years after the bomb, leukemia in victims of Hiroshima and Nagasaki was reported beginning only months after the explosion, and even in those who had not been exposed to the prompt radiation but to fallout and uranium in the bombed city debris [21
]. Furthermore, the onset lag for internal exposure to high LET radiation (e.g., Uranium) has not been determined and it can be argued that this lag cannot be deduced from the external low LET studies that make up the current radiation risk model. On the other hand, it may be that the increases in cancer found here for some individuals are the result of some earlier exposure, perhaps during the 1991 Gulf War. The origin and time of introduction of the carcinogenic agent causing the effects found here will be the subject of a separate report. However it does not seem unreasonable to conclude that the causes of the infant deaths and the cancer increases are one and the same.
We must finally address the earlier listed shortcomings of the interview questionnaire survey method. These might have been of concern had the findings been less clear but since the Relative Risks for the various indicators were extremely high, it can hardly be possible that these results could have occurred through errors introduced through any of the potential problems outlined in the Methods Section. A 100% error in the population would only halve the relative risks. The levels of cancer and infant mortality which have been found are too great to be accommodated by any hypothesis except that a significant proportion of those interviewed completely invented the results, and for the reasons already given i.e., that they had given names, addresses and identities and the names of the doctors and clinics involved in an area where the consequences of giving misleading responses to questions are severe, this seems highly unlikely.