ROBERT MACLENNAN
Epidemiology Unit, Queensland Institute of Medical Research, 300 Herston Rd, Brisbane, QLD 4029

PETER SMITH
Department of Haematology-Oncology, Royal Children's Hospital, Flemington Road, Parkville, VIC 3052

TERMS OF REFERENCE

We were asked to comment on the report of the Committee to Review the Health Effects on Vietnam Veterans of Exposure to Herbicides, hereafter referred to as the NAS Report. The Committee was appointed by the Division of Health Promotion and Disease Prevention, Institute of Medicine, National Academy of Sciences, and was published by the National Academy Press, Washington, in 1993. Our terms of reference were to:

• Examine the Report of the National Academy of Sciences and comment on the scientific merit of the Report;
• Comment on the applicability of the findings of the Report to Australian servicemen and servicewomen, and in particular the diseases listed as "limited/suggestive" of an association, and if needed, assist in preparation of Statements of Principle that address these diseases;
• Comment on the relevance of the Recommendations of the Report to Australian veterans.

SCIENTIFIC MERIT OF THE NAS REPORT

In commenting on the scientific merit of the NAS Report we considered the following:

1. The members of the Committee are regarded highly by academic and scientific colleagues, nationally and internationally. They have an appropriately wide range of expertise.

2. The Committee's specific mandate was to determine, if possible,

a. "whether there is a statistical association between the suspect diseases and herbicide use, taking into account the strength of the scientific evidence and the appropriateness of the methods used to detect the association";

b. "the increased risk of disease among individuals exposed to herbicides during service in Vietnam; and

c. "whether there is a plausible biologic rnechanism or other evidence of a causal relationship between herbicide exposure and a disease".

The Committee's first task was to assess statistical association, and in doing so they took into account the strength of the evidence and the appropriateness of the methods used to detect the association. They also considered the consistency of association among studies. Their conclusions are given in the body of the NAS Report.

The Committee's findings regarding associations between exposure to herbicides and specific health outcomes are summarised in table 1-1 of the executive summary (pages 1-5 and 1-6) of the NAS Report. The findings are grouped under four categories. This table has had considerable impact in the United States of America, but it should be noted that the table refers only to 2 (a) above, and not to 2 (b) or (c). Diseases were classified by the Committee in four categories as follows:

l. sufficient evidence of an association;

2. limited/suggestive evidence of an association;

3. inadequate/insufficient evidence to determine whether an association exists;

4. limited/suggestive evidence of no association.

These categories are very similar to those used in International Agency for Research on Cancer (IARC) monographs on the evaluation of carcinogenic risks to humans where the evidence relevant to carcinogenicity from studies in humans is classified into one of the following categories:

APPLICABILITY OF THE FINDINGS OF THE NAS REPORT TO AUSTRALIAN SERVICEMEN AND SERVICEWOMEN

The study of effects of herbicides and Vietnam service among veterans.

In the general absence of an index of personal exposure to herbicides, two approaches have been used to assess possible adverse health effects of service in Vietnam. The first approach assumes that exposure to herbicides occurred but is generally impossible to study directly, and hence the effects of herbicide exposure in other situations are used as surrogates for the effects of service in Vietnam. This approach is the basis of the NAS Report, and classifies diseases according _o the level of association with herbicide exposure in non-Vietnam studies. No attempt is made to take account of the high probability that apart from personnel directly involved in spraying herbicides, the exposure to herbicides of veterans in Vietnam was very much less than in studies of persons exposed in manufacturing herbicides or in applying them.

The second approach pragmatically seeks to determine what adverse health effects are associated with Vietnam service. The issue as to whether or not there was herbicide exposure and consequent health effects is regarded as an insoluble problem beyond the scope of scientific investigation, and thus of little relevance to decision making. In preparatory work for an epidemiological investigation by the Commonwealth Institute of Health and commissioned by the Australian Government, the problem of development of an adequate index of exposure was considered to be insurmountable, and as a consequence it would not be possible to demonstrate direct associations between exposure to herbicides and diseases in veterans. One of the authors of this report (MacLennan) was the principal investigator during the planning and initiation of Australian studies in 1981, and the approach taken in relation to studies of Australian veterans was to attribute any differences in disease outcomes between veteran and control groups to service in Vietnam rather than to herbicide exposure, per se. An example is the study of birth defects in children fathered by Australian veterans (Donovan et al 1984).

Because the health effects observed in non-veteran studies are often done in highly exposed groups of people, eg from occupational exposures, or as a result of an accidental exposure as in the population of Seveso, Italy, application of the conclusions of the NAS Report to veterans who, except for personnel occupationally exposed, had much lower levels of exposure, is giving veterans the benefit of any doubt.

Extrapolation from non-veteran studies to veterans.

The process of extrapolation from studies of highly exposed persons to veterans with generally unknown exposures appears to be a response to the concerns of veterans. It is stated on page 1-4 of the NAS Report that "the Committee felt that considering studies of other groups could help address the issue of whether these compounds might be associated with particular health outcomes, even though these results would have only an indirect bearing on the increased risk of disease in veterans themselves. Some of these studies, especially those of workers in chemical production plants, provide stronger evidence about health effects than studies of veterans because exposure was generally more easily quantified and measured."

The Committee state on page 1-13 that "although there have been numerous health studies of Vietnam veterans, most have been hampered by relatively poor measures of exposure to herbicides or TCDD, in addition to other methodological problems". On page 1-13, the Committee concluded that it is not possible to quantify the degree of risk likely to be experienced by veterans because of their exposure to herbicides in Vietnam "

Applicability of the NAS Report to Australian veterans.

In considering to what extent the findings of the Committee are applicable to Australian veterans, it is noted that the Committee was very conservative in their assessment of statistical associations. They used more stringent criteria than appear to have been required by their terms of reference. On page 8-3, the Committee state that "one must beware of over interpreting an isolated finding of excess risk for a given tumour type within a single study. Consistency across studies, with consideration of dose-response relationships and use of other statistical methods that evaluate plausibility, should be assessed before reaching any conclusions regarding associations between exposures and cancer. Additionally, the confidence intervals around the estimate of association will provide guidance as to the degree of precision and study size." Dose-response and plausibility are criteria also used in making judgements about the causality of association and go beyond simple statistical association.

In our review, we have considered that the NAS Report's category 2 of "limited/suggestive" evidence of an association is unsatisfactory from the point of view of decision making. Hence we have deleted the "limited/suggestive" category. Only two categories - "sufficient evidence of an association" and "inadequate/insufficient evidence to determine whether an association exists" are used by us in our report. We have evaluated the evidence of association for all diseases in the "limited/suggestive" category, and have assigned them to either the "sufficient evidence" or inadequate/insufficient evidence" categories. In addition, we have reviewed diseases in the "inadequate/insufficient" category to assess whether any should be classified in the "sufficient evidence" category.

A. DISEASES WITH SUFFICIENT EVIDENCE OF AN ASSOCIATION

We accept the NAS Report's inclusion of the following diseases in this category: chloracne, soft tissue sarcoma, porphyria cutanea tarda, non--Hodgkin's lymphoma, Hodgkin's disease. To this we would add multiple myeloma, leukaemia and respiratory cancers. The reasons for doing so are given below.

The approach taken by the Committee to categorising evidence of statistical associations was conservative, and included consideration of criteria used to judge the causality of associations. Some of these criteria are measures, in part, of the internal validity of a study and are therefore appropriately considered in the context of assessing associations.

Review of the NAS Report by the US Department of Veterans Affairs Task Force.

Due to the conservative approach taken by the Committee in the NAS Report, we have compared the conclusions of the NAS Committee and those of a US Department of Veterans Affairs (DVA) Task Force in relation to three cancers [respiratory cancers (lung, larynx, trachea), prostate cancer, and multiple myeloma] included in the "limited/suggestive" category of the NAS Report.

The process by which determinations are made regarding compensation for veterans for certain diseases related to service is outlined on page 2-28 of the NAS Report. It is stated that:

"whenever the Secretary determines, on the basis of sound medical and scientific evidence, that a positive association exists between the exposure of humans to an herbicide agent, and the occurrence of a disease in humans, the Secretary prescribes regulations providing that a presumption of service connection is warranted for that disease. The current DVA compensation policy provides that in making determinations, the Secretary shall take into account reports from the National Academy of Sciences and all other sound medical and scientific information and analysis. In evaluating any study for the purpose of making such determinations, the Secretary shall take into consideration whether the results are statistically significant, are capable of replication, and withstand peer review. An association between the occurrence of a disease in humans and exposure to an herbicide agent is considered to be positive if the credible evidence for the association is equal to or outweighs the credible evidence against the association." The NAS Committee "was charged with reviewing the scientific evidence, rather than making recommendations regarding .. policy" (NAS Report page 1-5). Both the NAS Committee and the DVA Task Force considered the evidence. In doing so the DVA Task Force used the NAS Report as a resource, and had access to the same body of scientific literature. We have reviewed the evidence, taking into account both the NAS Report and the DVA Task Force recommendations.

HAEMOPOIETIC MALIGNANCY

Multiple myeloma
Multiple myeloma is a malignant disease resulting from a monoclonal proliferation of plasma cells which are the most mature cells of the B lymphocyte lineage. It is a very uncommon malignancy representing approximately lt of haematological malignancies and thus a very small percentage of malignancy overall. Its incidence is age dependant being very uncommon before middle age and with incidence rising with advancing age. It is interesting to note that it is approximately 14 times more common in Afro Americans than in white Americans. Thus in a racially representative sample of the US population one would expect to find approximately twice the cases that one would find in a similarly representative sample of the Australian population.

Epidemiologic studies: Epidemiologic studies of multiple myeloma in populations exposed or likely to have been exposed to herbicides are summarised in table 8-32 of the NAS Report. As the DVA Task Force has commented, these data are at least as compelling in establishing an association as are the data for Hodgkin's disease.

Epidemiological studies of multiple myeloma in Vietnam veterans however have not established a clear association. These studies are difficult to accept as definitive however because of the rarity of the disease under study and the fact that the median age of the population under study is still too low to pick up many cases. If the effect simply causes an increase in age specific incidence this hypothesised effect may not be seen for many more years.

Biological considerations: As pointed out in the introduction multiple myeloma is a malignancy of mature cells of the B lymphocyte lineage. The data strongly support an association between non-Hodgkin's lymphoma and both herbicide exposure and Vietnam service. Non Hodgkin's lymphoma in this age group is usually a malignant proliferation of less mature cells of the lymphocyte lineage. While this cell lineage evidence does not prove an association it does provide plausible support for the hypothesis of association.

The DVA Task Force stated that "The NAS Report is both a valuable contribution to the medical and scientific literature and a valuable resource to DVA in carrying out the Congressional mandate to prescribe regulations for a presumption of service-connection, when DVA determines that the credible evidence for the association (between herbicide use and a disease) is equal to or outweighs the credible evidence against the association.

Conclusion: We would agree with the DVA Task Force reviewers that the data provide sufficient evidence of an association between multiple myeloma and exposure to herbicides.

Leukaemia
Leukaemia results from an abnormal monoclonal proliferation of haemopoietic stem or precursor cells of either myeloid or lymphoid lineage resulting in the recognised forms of this disease, ie acute myeloid leukaemia (AML), acute lymphoblastic leukaemia (ALL), chronic myeloid leukaemia (CML), and chronic lymphocytic leukaemia (CLL). Leukaemia in humans is associated with exposure to ionising radiation and in some cases, a genetic predisposition.

Epidemiological Studies: Results of epidemiological studies of leukaemia in populations exposed or likely to have been exposed to herbicides are given in the NAS report in table 8-33 (production workers), table 8-34 (agricultural workers) and on page 8-137 (Vietnam veterans). These studies are complicated by (i) small numbers as this is an uncommon disorder especially in younger to middle aged persons and (ii) possibly long latency. A positive association was found in studies. of Seveso survivors where there was a relative risk of 1.98 (1.25Ä3.13) for leukaemia (Bertazzi, 1989). There has been an extensive literature over many years of the association between leukaemia and farming, summarised by Burmeister (1982). Studies reported by Alavanja (1988) among agricultural extension workers found an increased risk of leukaemia of 1.92 (1.04-3.54), with significant dose response related to number of years worked. However the amount of exposure of farmers and agricultural workers to herbicides is uncertain.

Smoking: Inclusion of leukaemia as a war service related disease is supported by a recent meta-analysis supporting a causal relationship between cigarette smoking and certain forms of adult leukaemia (Brownson et al 1993). The summary smoking related risk derived from prospective studies was 1.3 (1.3-1.4), and there was evidence of dose-response. In an evaluation of epidemiologic studies and applying the criteria for causal inference, Siegel (1993) concluded that smoking causes myeloid leukaemia.

Biological considerations: Leukaemia is a disorder affecting a haemopoietic stem or precursor cell. There is evidence of a clear association between herbicide exposure and disorders of cells of haemopoietic lineage (NHL and HD) and a situation where we believe the evidence for such an association outweighs the evidence against the association (multiple myeloma). Thus, it is plausible to argue that an insult suffered by cells of a particular lineage resulting in malignant proliferation of more mature precursor cells might, after an appropriate latency, produce a similar effect on less mature precursors of the same lineage. Indeed, there is a clear association between occurrence of NHL or HD and acute leukaemia, in that patients who have had HD or NHL are at much higher risk than the general population to develop leukaemia. We believe that these biological arguments provide plausible support for the argument of an association.

Conclusions: In contrast to the NAS report and DVA Task Force findings, we believe that the evidence for an association between herbicide exposure and leukaemia at least balances the evidence against such an association. Such an argument is partly supported by biological considerations. If the association between smoking and leukaemia is accepted, the number of cases of leukaemia in non-smoking veterans would be small, and these should be given the benefit of the doubt.

RESPIRITORY CANCERS
Tobacco smoking causes a very high proportion of respiratory cancer (of the larynx, trachea and bronchus) in Western populations. Smoking varies by occupation, the proportion of smokers generally being higher in "blue--collar" workers who because of their occupation may also be exposed to a number of chemicals. Smoking may thus be associated with both respiratory cancer and occupational chemical exposure. Hence, in an epidemiological study, a suspected occupational exposure may be found to be associated with lung cancer because the exposure is associated with smoking which in turn is associated with lung cancer. This type of association is termed secondary association, and the relationship between exposure and disease is said to be confounded. The effects of confounding can be controlled in analysis, but only if smoking information is available for the individuals whose chemical exposure and disease have been documented. It appears likely that the assignment in the NAS Report of respiratory cancers to the 'limited/suggestive evidence of association" category was based mainly on the inadequate control of potential confounding by smoking in the analysis of the studies reviewed.

In addition to the overall summary in chapter 1 of the NAS Report, we have reviewed four studies of particular relevance to the issue of association of herbicide exposure and respiratory cancers.

Lynge (1985) included a high proportion of all persons employed in the manufacture of phenoxy herbicides in Denmark before 1982 when the predominant product was MCPA and only a very limited amount of 2,3,5-T was processed in one of the two factories included in the study. Cancer cases were identified by linkage to the Danish National Cancer Registry. Eleven cases of lung cancer were observed in men compared with 5.33 expected, a relative risk of 2.06 (95 percent confidence interval 1.03-3.69). Lange (sic) considered this nevertheless to be due to chance because of the large number of diagnostic groups tested. The plants were located near provincial towns and workers were previously recruited from the countryside where tobacco consumption was relatively low in the 1950s. Lynge concluded that "based on the data presented here it is not possible to draw a conclusion concerning the lung cancer risk following exposure to phenoxy herbicides".

In their study of 5,172 workers exposed to TCDD when working for 12 companies in the US, Fingerhut et al (l991), from the National Institute for Occupational Safety and Health, found that "excess mortality from all may result from exposure to TCDD", although they could not exclude the possible contribution of factors such as smoking and occupational exposure to other chemicals". The total cohort of workers "had a non-significant increase in mortality from cancers of the trachea, bronchus and lung (ICD code ]62). Mortality from cancers of the respiratory system (ICD codes 160 to 165) was significantly higher in the high exposure subcohort (SMR 142; 95 percent confidence interval 103-192).

To estimate the effect of smoking on the increase in lung cancer, the expected number of lung cancers was adjusted according to the smoking prevalence found in lifetime histories obtained in 1987 by interviewing 223 workers in two plants. This adjustment increased the expected number of lung cancers in the overall cohort by 5 percent and in the high exposure subcohort by 1 percent, and reduced the SMR in the full cohort to 105 (95 percent confidence interval 85-130) and in the high exposure subcohort to 137 (95 percent confidence interval 98-187). There was no significant linear trend in mortality from lung cancer with increasing duration of exposure to products contaminated with TCDD. In their discussion the authors stated that:

the increased number of lung cancers in the high-exposure subcohort was probably not due to confounding by smoking for several reasons.

First, other diseases related to smoking were not more common than expected in this subcohort.

Second, in the exposed population with 20 years of latency, whose members presumably shared similar smoking habits, the increase was confined to the high exposure subcohort.

Third, on the basis of empirical evidence from other studies, Siemiatycki et al have shown that between a blue-collar population and the general US population, confounding by smoking is unlikely to account for an excess risk of more than l0 to 20 percent.

Finally, a limited adjustment in the risk of lung cancer based on the smoking prevalence of surviving workers at only two plants, did not substantially change our results.".

However, they stated that "it remains possible that the increase was due to confounding by occupational exposure other than TCDD. For example, asbestos may have contributed to mortality from lung cancer in the cohort, since two deaths were due to mesotheliomas". The small but significant increase from all cancers combined was considered to be "consistent with a carcinogenic effect of TCDD.

After excluding lung cancers, Fingerhut et al (l991) found increased cancer mortality with an overall SMR of 117 (95 percent confidence interval 100--136); the high-exposure subconort had an SMR of 150 (95 percent confidence interval 118-189) Thus the increased mortality could not be explained by smoking, and the authors considered it biologically plausible that TCDD may produce tumours in more than one organ in humans. In an editorial in the New England Journal of Medicine, Bailar (1991) commented that "there is some weakening of the position of those who believe that low levels of exposure to TCDD are entirely safe for humans". He further stated despite the problems, which Fingerhut et al carefully note, this work is a model of its kind. Occupational cohort studies are inherently difficult, and we are likely to wait a long time for appreciably better evidence of the effects of TCDD on human health.

Manz et al (1991) reported a mortality follow-up of 1184 men and 399 women employed in a Boehringer chemical plant in Germany that produced herbicides including processes contaminated with TCDD. The deaths from all cancers had an SMR of 124 when compared with national data, similar to the SMR of 115 in the NIOSH study. They concluded that their findings point to TCDD as a human carcinogen, and that their results, showed that the increase in cancer mortality is not directed at special sites. However, when compared with a cohort of 3417 male workers at the Hamburg gas company, who had similar smoking histories, the SMR for lung cancer among the TCDD exposed workers in the Boehringer cohort was 167 (95 percent confidence interval 109-244). This SMR was based on 26 cases observed versus 15.6 expected. They concluded that "substantial confounding due to smoking seems unlikely".

Saracci et al (1991) reported results from an international register of 18,910 production workers or sprayers from ten countries. Using cause -specific national death rates as reference, no excess mortality was observed for all causes, for all neoplasms, for the most common epithelial cancers, or for lymphomas. Among 13,482 workers regarded as exposed to chlorophenoxy herbicides, the SMR for trachea, bronchus and lung cancers was 102 (95 percent confidence interval 87-118); but in a subcohort of 416 workers regarded as probably exposed (no job titles were available but it was judged that most workers would have been exposed) the SMR was 221 (95 percent confidence interval 110-395); and in 3951 non-exposed workers the SMR was 140 (95 percent confidence interval 100-190). The statistical significance of these increased SMRs must be considered with caution due to the large number of sites and exposure groups assessed. Furthermore, it was not possible to adjust for confounding by smoking due to the absence of information. The authors do not comment on lung cancer in the discussion of their results.

Williams (1991) stated in a letter to the Lancet (December 21/28, 1991, page 1592) that exposure to TCDD has been clearly associated with depression of the immune system in animals; the weaker evidence for man is due largely to methodological shortcomings. . . Secondly, if we assume that TCDDs are carcinogenic promoters and that they have the ability to stimulate initiated cells to produce a tumour, it is plausible that TCDD could promote overt cancer in many different types and at different sites. Biologically the precedence for this can be seen in organ transplant surgery which has been causally linked with the development of several de novo cancers - for example, lymphoma (notably reticulum cell sarcoma), liver cancer, and skin cancer. This variety of de novc cancers is caused bv then immunocompromised state of the individual, rendering them more susceptible to the impact of carcinogenic stimulus".

In its report (page 8-2) the NAS Committee said that based on its effects in animal studies, TCDD is considered a tumour promoter, not a tumour initiator". Although it is biologically plausible to have cancers of more than one type following TCDD exposure, lung cancer has not been recognised to be increased following organ transplantation and immunosuppression.

Conclusions: There is evidence both .or and against herbicide-lung cancer association, with some studies providing evidence of statistically significant association (despite absence of formal control of confounding) and others being unable to replicate the finding of association. In an Australian context where smoking related diseases have already been determined to be Service related, veterans with the very few respiratory cancels not presumptively related to exposure to tobacco should be given the benefit of any doubt, and it is recommended that respiratory cancers should be included in the "Sufficient evidence" category.

B. DISEASES RECOMMENDED TO BE ADDED TO THE GROUP WITH INDEQUATE/INSUFFICIENT EVIDENCE OF ASSOCIATION

PROSTATE CANCER
The only disease in the "Limited/suggestive" category not so far discussed is prostate cancer. Regarding the applicability to Australian veterans, we have considered the major publications reviewed by the NAS Committee.

Alavanja et al 1988 studied mortality among agricultural extension agents employed by the US Department of Agriculture. They were responsible for disseminating information from the agricultural research community to individual farmers. Although a proportionate mortality analysis showed an increase for prostate cancer (PMR of 1.5 with 95% confidence interval l.13-1.99), a case-control analysis of their mortality data found the odds ratio for ever versus never having been an extension agent was 1.02 (95% confidence interval 0.69-1.49). Approximately one third of these men had been farmers prior to their employment as extension agents.

Blair et al 1985 reviewed the literature and found that despite a generally favourable experience overall, farmers were at an increased risk from several cancers including prostate, although in a previous study of licensed pesticide applicators, Blair et al 1983 had 2 observed cases of prostate cancer compared with 3.8 expected; and in a geographical analysis of prostate cancer in the United States, Blair and Fraumeni (1978) found the highest rates for prostate cancer in the Midwest and north-central region of the United States.

Burmeister (1981) used Iowa death certificate information to compare the mortality rates of white rnale farmers and non-farmers, and found a higher proportional mortality rate for prostate cancer among farmers (PMR = l.l0, p<0.0l). AgeÄadjusted death rates among Iowa farmers from 1971 to 1978 were 206.48 per 100,000 compared with 116.20 among non-farmers. Burmeister et al (1983) in a subsequent case-control analysis of death certificates found an association between farming and prostate cancer (OR 1.19, p < 0.05), but no association with any agricultural practice was found.

Morrison et al (1992) in their review of herbicides and cancer summarised epidemiologic studies of herbicide exposure and the relative risk of prostate cancer. They concluded that there was limited evidence that herbicide exposure may increase the risk of prostate cancer. The studies reviewed included a cohort study of western Canadian farmers (Morrison et al 1993) which found a significant dose-response relationship between risk of dying of prostate cancer and the number of acres sprayed. Morrison et al (1992) concluded that "9 of the l0 studies reviewed noted an increased risk of prostate cancer with herbicide exposure; however, only the Canadian study observed a statistically significant trend in risk.

In the NAS Report, the Committee place considerable weight on the 1993 paper of Morrison et al. They note the increased risk of prostate cancer associated with herbicide spraying, and the increasing risk found with increasing number of acres sprayed. For the entire cohort, the relative risk for prostate cancer and spraying at least 250 acres was 1.2 (CI 1.0--1.5). Adjustment for potential confounders in the analysis showed no evidence of confounding for the association." Analysis was restricted to groups of farmers most likely to be exposed to phenoxy herbicides, and the Committee reported that for each of these restricted comparisons, a statistical test for trend over increasing number of acres sprayed was significant.

In contrast to studies of farmers, Breslin et al (1988) report on patterns of mortality among 24,235 US Army and Marine Corps Vietnam veterans compared with that of 26,685 non-Vietnam veterans using standardised proportional mortality ratios. They found that when all the malignancies were grouped together, Vietnam veterans did not exhibit an excess of cancer when compared to their counterparts who did not serve in Vietnam. They found statistically significant increased risk among Marines for lung cancer (PMR 1.58, p<0.025), and non-Hodgkin's lymphoma (PMR 2.10, p<0.025). The risk for soft tissue sarcoma was not elevated among Vietnam veterans as a whole or in any subgroup of these veterans. There were 30 deaths from prostate cancer among Army Vietnam veterans (PMR 0.92, 95% confidence interval 0.55-1.23) and 5 deaths among Marine Vietnam veterans (PMR 1.29, 95% confidence interval 0.16-10.3). The NAS Committee comment that "prostate cancer is generally a disease of older men, and the risk among Vietnam veterans would not yet be detectable in epidemiologic studies".

Conclusions: The studies reviewed can be grouped into those of farmers and related occupations where some chemical exposure (albeit poorly defined) is assumed to have occurred, and a study of mortality in Vietnam veterans. Neither the first group of studies provides sufficient evidence of statistical association, nor does the study of veterans. Hence prostate cancer should be included in the inadequate/insufficient category. Although studies of Vietnam veterans have not shown an increased risk for prostate cancer, veterans have not yet reached the age where prostate cancer is common, and this outcome should continue to be monitored and kept under review.

RELEVANCE OF THE RESEARCH RECOMMENDATIONS OF THE NAS REPORT TO AUSTRALIAN VETERANS
The research recommendations in chapter 12 of the NAS Report are specifically directed to the USA. The recommendations are abbreviated in numbers 1 to 6 below, and comments are then made as to their relevance to Australian veterans.

1. CONTINUED FOLLOW-UP OP THE AIR FORCE RANCH HAND COHORT
Although small in numbers compared with the Ranch Hands, the Field Hygiene Corps (sic) of the Australian Army used herbicides and pesticides and potentially had higher levels of exposure than other Service personnel. Members of this corps should be identified and offered periodic screening for cancer.

2. IDENTIFICATION OF VIETNAM SERVICE IN COMPUTERISED INDEXES
As a results of previous investigations, a computerised database exists in Australia of the 49,000 persons allotted to service in Vietnam. To these, attempts should be made to identify and add the approximate 3000 persons including members of the Australian Services and civilian employees who were also sent to Vietnam, albeit in many instances for brief periods. The database of National Servicemen who were drafted but who were not subsequently allocated to Vietnam service has been shown in previous studies, such as that of mortality, to be a unique and highly valuable research resource.

Both these databases should be maintained, be held by more than one institution, and, with suitable safeguards, be made available for future research on the effects of Vietnam service.

3. DEVELOPMENT OF BIOMARKERS OF HERBICIDE EXPOSURE
Developments elsewhere should be monitored, although this is a low research priority within Australia.

4. HISTORICAL EXPOSURE RECONSTRUCTION TO DEVELOP MODELS OF HERBICIDE EXPOSURE
Developments elsewhere should be monitored, and if a valid model were to be developed its use in future Australian studies would be informative. Previous consideration of the feasibility of developing an exposure model for Australian veterans concluded that it would not be feasible to develop a valid model because the only objective evidence would be to link information on aerial spraying (from the so called HERBS tapes) with the location of troops on the ground, together with an estimate of exposure that might result taking distance, weather, vegetation, clothing etc into account. A major factor against the feasibility of an exposure index in Australian service personnel was that up to 30 percent of the personnel in units at any given time were misclassified regarding their presence or absence.

5. EVALUATION OF EXPOSURE RECONSTRUCTION MODELS
Developments elsewhere should be monitored, but this is a low research priority for Australia.

6. ADDITIONAL EPIDEMIOLOGIC STUDIES OF VETERANS IF AN EXPOSURE RECONSTRUCTION MODEL IS FOUND TO BE FEASIBLE AND VALID
This conditional recommendation is supported, although it is noted that because it appears unlikely that a valid model of exposure will be developed, epidemiologic research on Australian veterans should focus on Vietnam Service as the relevant exposure.

It is recommended that the occurrence of diseases in veterans should be monitored by periodic linkage of the computerised indexes (recommendation 2 above) of persons who served in Vietnam and of National Service draftees who did not, to the National Cancer Incidence Clearing House and to the National Death Index. Such investigations could be assisted by the Australian Institute of Health and Welfare which maintains the National Clearing House for Cancer and the National Death Index.

REFERENCES
Alavanja MCR, Blair A, Merkle S, Teske J, Eaton B. Mortality among agricultural extension agents. American Journal of Industrial Medicine 1988;14:167-176.
Bailar JC. How dangerous is dioxin? New England Journal of Medicine 1991;324:260-262.
Bertazzi PA, Zocchetti C, Pesatori AC, Guercilena S, Sanarico M & Radice L. Ten year mortality study of the population involved in the Seveso Incident in 1976. American Journal of Epidemiology 1989:129:1187
Bertazzi PA, Pesatori AC, Consonni D, Tironi A, Landi MT' & Zocchetti C. Cancer incidence in a population accidentally exposed to 2,3,7,8-Tetrachlorodibenzo-para-dioxin. Epidemiology l993;4:398-406
Blair A, Frauman DJ, Lubin JH, F'raumeni JF. Lung cancer and other causes of death among licensed pesticide applicators. Journal of the National Cancer Institute 1983:71:31-37
Blair A, White DW. Leukemia cell types and agricultural practices in Nebraska. Archives of Environmental Health 1985;40:211-214.
Blair A, Malker H, Cantor KP, Burmeister L, Wiklund K. Cancer among farmers. Scandinavian Journal of Work and Environmental Health. l085;11:397-407
Breslin P, Kang HK, Lee Y, Burt V, Shepard BM. Proportionate mortality study of US army and US marine corps veterans of the Vietnam war. Journal of Occupational Medicine 1988;30:412-419.
Brown LM, Blair A, Gibson R, Everett GD, Cantor KP, Schuman LM, Burmeister LF, Van Lier SF, Dick F. Pesticide exposures and other agricultural risk factors for leukaemia. Cancer Research 1990:50:6585-6591
Brownson RC, Novotny PE, Perry MC. Cigarette smoking and adult leukemia -meta-analysis. Archives of Internal Medicine 1993;153:469-475.
Bueno de Mesquita HB, Coornbos G, Van der Kuip DAM, Kogevinas M, Winkelmann R. Occupational exposure to phenoxy herbicides and cancer mortality in The Netherlands. American Journal of Industrial Medicine 1993;23:289-300.
Burmeister LF. Cancer mortality in Iowa farmers, 1971-78. Journal of the National Cancer Institute 1981;66:461-464
Burmeister LF, Van Lier SF, Isacson P. Leukemia and farm practices in Iowa. American Journal of Epidemiology 1982;115:720-728
Burmeister LF, Everett GD, Van Lier S, Isacson P. Selected cancer mortality and farm practices in Iowa. American Journal of Epidemiology 1983;118:72-77.
Donovan J, MacLennan R, Adena M. Vietnam service and the risk of congenital anomalies - a case-control study. Medical Journal of Australia 1984;140:394-397
Doe JE:, Paddle CM 'The evaluation of carcinogenic risk to humans: occupational exposures in the spraying and application of insecticides. Regulatory Toxicology and Pharmacology 1994;19:297-308.
Fingerhut MA, Halperin WE, Marlow DA, Piacitelli LA, Honchar PA, Sweeney MH, Greife AL, Dill PA, Streenland K. Cancer mortality in workers exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin. New England Journal of Medicine 1991:324:212-218
Fingerhut MA, Sweeney MH, Halperin WE, Schnorr TM. The epidemiology of populations exposed to dioxin. In: Rappe C, Buser liR, Dodet B, O'Neill IK (Ed), Environmental carcinogens - methods. of analysis and exposure measurement, IARC Scientific Publications No 108. International Agency for Research on Cancer. Lyon 1991
Hansen ES, Hasle H, Lander F. A cohort study on cancer incidence among Danish gardeners. American Journal of Industrial Medicine 1992;21:651-660
Lynge E. A follow-up study of cancer incidence among workers in manufacture of phenoxy herbicides in Denmark. British Journal of Cancer 1985:52:259-270.
Manz A, Berger J, Dwyer JH, Flesch-Janys D, Nagel S, Waltsgott H. Cancer mortality among workers in chemical plant contaminated with dioxin. Lancet 1991;338:959-964.
Morrison HJ, Wilkins K, Semenciw R, Mao Y, Wigle D. Herbicides and cancer (review). Journal of the National Cancer Institute 1992;84:1866-1874
Morrison H, Savitz D, Semenciw R, Hulka B, Mao Y, Morison D, Wigle D. Farming and prostate cancer mortality. American Journal of Epidemiology 1993;137:270-280.
Pesatori AC, Consonni D, Tironi A, Landi MT, Zocchetti C, Bertazzl Cancer morbidity in the Seveso area 1975-1986. Chemosphere 1992:25:209-212
Ronco G, Costa G, Lynge E. Cancer risk among Danish and Italian farmers. British Journal of Industrial Medicine 1992;49:220-225
Saracci R, Kogevinas M, Bertazzi P-A et. al. Cancer mortality in workers exposed to chlorophenoxy herbicides and chlorophenols. Lancet 1991;338:1027-1032.
Siegel M. Smoking and leukemia: evaluation of a causal hypothesis. American Journal of Epidemiology 1993;138:1-9.
Wigle DT, Semenciw RM, Wilkins K, Riedel D, Ritter L, Morrison HI, Mao Y. Mortality study of Canadian male farm operators: non-Hodgkin's lymphoma mortality and agricultural practices in Saskatchewan. Journal of the National Cancer Institute 1990;8:575-582.