Subsequent to the release of the Update 2000 report, AIHW announced that it had discovered an error in its derivation of the Australian community standard for AML in veterans’ children. This affected the estimate of the expected number of cases and the associated confidence interval. Based on community rates, the previously reported expected number of cases (3 with a 95% confidence interval of 0–6) was corrected to 9 (95% CI 3–15 cases). The number of cases of AML among the offspring of Australian veterans who served in Vietnam and whose diagnoses were validated in the study was unchanged (n = 9). The predicted numbers of cases under various assumptions regarding non-respondents and cases not able to be validated changed slightly, yielding a total of 12 cases in veterans’ children under the assumptions adopted by the authors. Two leukemia cases were also reclassified after further investigation: one case of CLL in veterans’ children was reclassified to AML and one case of AML was reclassified to ALL. These changes had a small effect on the predicted number of cases.

A revised report correcting this error was published in October 2001 (AIHW, 2001). Under the assumptions favored by the study’s authors⁶ regarding cases they could not validate, the predicted number of cases (12) remained higher than the new expected number of cases (9) but the difference was not statistically significant (RR = 1.3; range = 0.8–4.0).

In Update 2000, three studies were found to provide evidence regarding an association between exposure to the herbicides used in Vietnam and acute myelogenous leukemia in the children of veterans. The first was a case-control study of AML and parental occupational exposures conducted by the Children’s Cancer Study Group (Buckley et al., 1989). Use of pesticides by either the mother or father, as reported in detailed interviews, was associated with an elevated risk. However, because of a high correlation among exposures in the three time periods studied (before, during, and after pregnancy), it was not possible to determine whether exposure uniquely prior to the pregnancy was associated with increased risk of AML in the children. The strongest associations were for children diagnosed before 5 years of age and for children with M4/M5⁷ morphology.

In a second case-control study of AML conducted by the Children’s Cancer

Group (Wen et al., 2000), self-reported service in Vietnam or Cambodia was associated with an elevated risk of AML (OR = 1.7, 1.0–2.9), after adjusting for potential confounders including education, race, income, smoking, X-ray exposure, and paternal marijuana use. Since service in Vietnam or Cambodia would be an extremely memorable event, under-reporting by controls or over-reporting by cases seems unlikely. Also arguing against recall bias was the lack of association with ALL in this study, as well as the lack of association of AML with general paternal military service. When stratified by time spent in Vietnam or Cambodia, those with one year or less of service there showed a stronger risk than those with more than one year; additionally, self-reported exposure to Agent Orange was not associated with AML. However, these results are not particularly convincing evidence against a causal association since neither length of service in Vietnam or Cambodia nor self-reported exposure are known to be strongly related to the actual level of herbicide exposure. Two or more tours of duty in Vietnam showed a stronger association than a single tour, although the number of cases and controls were small. This study showed the strongest association to be with childhood AML diagnosed before the age of two years (OR = 4.6, 1.3–16.1). One concern was the apparent lack of adjustment for maternal marijuana use, which has been shown to be related to AML (Robison et al., 1989). Additionally, the authors point out that an unexplained increase in risk with longer time since service in Vietnam or Cambodia might have been due to randomness in the data, but could also have been due to an unmeasured post-war exposure that was different from those who did not serve in the military, or who served elsewhere.

A third study was that of the Australian Vietnam veterans. Investigators surveyed veterans regarding their medical conditions and the health of their children (AIHW, 1998), with a follow-up validation of the self-reported conditions and a calculation of the expected number of cases based on Australian community standards (AIHW, 1999; AIHW, 2000). The results were adjusted for age and gender, but not for other potential confounding factors. Among respondents, 9 cases of AML were successfully validated. The expected number of cases had been originally reported in error as 3, with a range of 0–6. The corrected calculations indicate 9 expected cases, with a range of 3–15 (AIHW, 2001). Taking into account possible additional cases among non-respondents and cases that might have been validated had the information been obtainable, the authors estimated that there were 13 cases of AML among the children of the surveyed veterans, representing a 1.4-fold increased risk. Therefore, though elevated, the number of cases fell within the range that might be expected in the community. Sensitivity analyses were conducted using a variety of strategies for assignment of non-respondent cases. This finding was not outside the range of values consistent with random fluctuations.

There are two other analyses not previously reviewed by the Update 2000 or previous committees that are pertinent to the issue of childhood AML and paternal preconceptional exposure to herbicides used in Vietnam or their contaminants.