A report of the 1996 retrospective cohort study of Australian Vietnam veterans
This work is copyright. Apart from any use permitted under the Copyright Act 1968, no part may be reproduced by any process without written permission from the Department of Veterans’ Affairs. Requests and inquiries concerning reproduction and rights should be directed to the Department of Veterans’ Affairs, PO Box 21, Woden ACT 2601.
The Department of Veterans’ Affairs and the study team in particular are grateful to:
- the members of the Study Advisory Committee for their guidance;
- Dr Graham Giles and Dr Richard Madden for their comments;
- INTSTAT Australia Pty Ltd for the statistical analysis;
- the staff at the Australian Institute of Health and Welfare who did the bulk of the data matching; and
- the staff at the Health Insurance Commission who also did data matching.
National Library of Australia Cataloguing-in-Publication data
Crane PJ, Barnard DL, Horsley KD, Adena MA. Mortality of Vietnam veterans: the veteran cohort study. A report of the 1996 retrospective cohort study of Australian Vietnam veterans. Canberra: Department of Veterans’ Affairs, 1997.
Produced by the Department of Veterans’ Affairs, Canberra
ISBN 0 642 27098 8
- Design of this study
- The nominal roll of Australian Vietnam veterans
- Determination of vital status
- Geographical profile of living Vietnam veterans
- Statistical methods used in the analysis
- Standardised Mortality Ratios
- ‘Healthy worker effect’
- Standardised Relative Mortality Ratios
- Strengths and weaknesses of the study
- Summary of findings
This is a study of the mortality of male Australian Vietnam veterans.
There has been one previous study of mortality – the 1984 Australian Veterans Health Study – of Australian men who were conscripted to serve in Vietnam. However, this is the first mortality study of all Australian male personnel – navy, army, air force, the Citizen Military Forces and civilian – who served in Vietnam, and covers the period from the end of the Vietnam War until 31 December 1994.
Before going to Vietnam, all the army, navy and air force personnel were medically examined to ensure that they were healthy and fit. Men selected as being fit for work usually have a lower mortality rate than the national population. This survival advantage is known as the ‘healthy worker effect’. The national servicemen studied in 1984 showed this effect.
The aims of the Vietnam Veterans Mortality Study were to:
- develop a nominal roll of all Vietnam veterans (male and female);
- determine mortality rates of male Vietnam veterans;
- compare the mortality rates of male Vietnam veterans with those of Australian males; and
- develop a geographic profile of living Vietnam veterans (male and female).
To conduct this study it was necessary to compile a listing (or nominal roll) of all Vietnam veterans and then determine who was still alive and who had died during the period from the last date of service until 31 December 1994. Mortality rates were then estimated. Age and calendar year of death were taken into account.
The nominal roll is a listing of all those who conform to this study’s definition of ‘Vietnam veterans’. This definition broadly covers any member of the army, navy and air force and some civilian personnel who served on land or in Vietnamese waters for at least one day during the period of the Vietnam War. The roll lists each veteran’s surname, up to two given names, service number, date of birth, one or more unit/ship/squadron and period/s of service. Compilation of the nominal roll was a major goal of the study and meant that the study was not restricted to those Vietnam veterans known to the Department of Veterans’ Affairs.
Data for the nominal roll were provided by the Department of Defence and appropriate civilian agencies.
Table 1 shows the total numbers of male and female Vietnam veterans, categorised by the first group in which they served.
|Army||41 388||47||41 435|
|Navy||12 376||0||12 376|
|Air force||4 438||106||4 544|
|Philanthropic organisations a||28||23||51|
|Civilians b||806||308||1 114|
|Total||59 036||484||59 520|
a. Includes personnel from the Red Cross, Everyman’s Welfare Service, Salvation Army, Young Men’s Christian Association and Australian Forces Overseas Fund.
b. Includes war correspondents, merchant seamen, entertainers and South-East Asia Treaty Organization aid program staff.
After the nominal roll had been compiled the names of male veterans were then matched with a number of registers to determine whether the veterans had died since the end of their service in Vietnam or were still alive on 31 December 1994. The sources of data on vital status were: Department of Defence records; Department of Veterans’ Affairs records; the National Death Index; Electoral Commission rolls; and, for those not found on the preceding registers, the Health Insurance Commission Medicare database. Only deaths recorded after 1 January 1980 are contained in the National Death Index. Thus, Vietnam veteran deaths that occurred before this date were only known from Department of Veterans’ Affairs records and those supplied to it by the Department of Defence.
|Army a||37 769||2 756b||891||41 416|
|Navy||11 309||583c||484||12 376|
|Air force||3 928||409d||101||4 438|
|Total||53 391||3 840||1 805||59 036|
a. Army group includes 28 members of philanthropic organisations.
b. Includes 477 combat deaths.
c. Includes 8 combat deaths.
d. Includes 14 combat deaths.
Row totals may not add up to 100 per cent due to rounding.
Table 2 shows the number and percentage of veterans who were determined to be alive or dead in the study.
From a total cohort of 59 036 male Vietnam veterans followed up after approximately 22 to 32 years, the vital status of 3.1 per cent remained unknown. This percentage is very low for a study of this type. It was estimated that about two-thirds of veterans with unknown vital status may have emigrated from Australia since the end of the Vietnam War.
The current place of residence of Vietnam veterans was determined by the postcode recorded on the relevant State or Territory electoral roll. As shown in Table 3 and the maps in Chapter 11, surviving male Vietnam veterans live predominantly on the coast of the eastern states and approximately 60 per cent are urban metropolitan dwellers – living in the State and Territory capitals.
|Capital city||Rest of State/Territory||Total|
|New South Wales||7 293||6 028||13 321|
|Queensland||8 351||4 266||12 617|
|Victoria||5 396||3 050||8 446|
|West Australia||4 045||1 311||5 356|
|South Australia||2 745||1 171||3 916|
|Australian Capital Territory||2 104||20||2 124|
|Total||30 805||16 967||47 772|
A particular problem encountered in this study was that not all deaths could be ascertained by linkage with the various databases – some deaths remain hidden within the 3.1 per cent of ‘unknown’ vital status. Under-ascertainment in the National Death Index varied by branch of service and calendar year of death, while under-ascertainment in the Veterans’ Affairs databases varied with branch of service and cause of death. After taking into account under-ascertainment using statistical models developed for this study death rates called standardised mortality ratios (SMRs) were estimated. These compare the death rates of Vietnam veterans with the rates for the Australian male population.
However, the adjustment for under-ascertainment was incomplete. This meant that the SMRs were inaccurately estimated and that the absolute values, especially for navy Vietnam veterans, were an artefact of the way vital status had been inferred. The absolute values of the SMRs should not be taken as being definitive because they may contain bias that was introduced by the adjustment. Nevertheless, their values relative to each other are valid. This is equivalent to an analysis of standardised relative mortality ratios (SRMRs). Being ratios of SMRs, SRMRs are much less affected by adjustments for under-ascertainment than the component SMRs. SRMRs describe proportional differences in death rates within the Vietnam veteran cohort not absolute differences compared with the Australian male population.
Few deaths were expected before 1980 because most Vietnam veterans were between 20 and 40 years old during this period. Based on 673 deaths, the SMR for all causes before 1980 was 0.68 (95% CI* 0.63 to 0.74). Low SMRs were obtained in each branch of service and for most specific causes. However, it is likely that SMRs before 1980 were underestimated because the sole source of death data for this period was lists held by Veterans’ Affairs and the corresponding lists after 1980 were known to be incomplete.
The 2607 deaths recorded between 1980-94 comprised 1713 army, 529 navy, 311 air force and 54 civilian Vietnam veteran deaths. Because vital status could not be determined for 41 per cent of civilians, their death rates could not be determined reliably.
The estimated SMR for all causes of death for military Vietnam veterans between 1980-94, which also takes account of under-ascertainment of deaths, was 1.07 (95% CI 1.02 to 1.12). It varied by branch of service, being 1.00 (95% CI 0.94 to 1.05) for army Vietnam veterans, 1.12 (95% CI 0.97 to 1.27) for air force Vietnam veterans, and 1.37 (95% CI 1.23 to 1.52) for navy Vietnam veterans.
There were statistically significantly more deaths of military Vietnam veterans between 1980-94 than were expected given death rates in the Australian male population for:
|Cause of death||SMR||95% CI|
|All causes||1.07||(1.02, 1.12)|
|Neoplasms (cancer)||1.21||(1.11, 1.31)|
|Prostate cancer||1.53||(1.07, 2.12)|
|Lung cancer||1.29||(1.12, 1.49)|
|Ischaemic heart disease||1.10||(1.01, 1.21)|
There were statistically significantly fewer deaths of military Vietnam veterans between 1980-94 than were expected given death rates in the Australian male population for:
|Cause of death||SMR||95% CI|
|Endocrine, nutritional and metabolic diseases||0.71||(0.53, 0.93)|
|Mental disorders||0.50||(0.27, 0.85)|
|Congenital anomalies||0.15||(0.00, 0.81)|
The AVH Study grouped army corps according to the assumed ‘stress and danger to which men in the corps would have been exposed in Vietnam’. In this study, death rates were determined for the same corps groupings; however, it did not replicate the finding from the AVH Study that engineers within the army Vietnam veterans had higher death rates than other corps groupings. Analogous groupings were developed for navy and air force Vietnam veterans but there was no statistically significant difference in the all causes death rate between the corps groupings.
Only corps grouping, duration and year of first service in Vietnam were available as surrogate measures for ‘exposure’ to Vietnam service. These measures are crude summary measures and may not necessarily be related to increased risk of death. In addition, differential under-ascertainment of deaths meant that the analysis was at best tentative. Accordingly, no conclusions can be drawn from either the general absence of statistically significant results, or from the few scattered nominally statistically significant trends.
For some causes of death, particularly cancers, there may be a latency period or delay between the exposure and the development of the disease and death. In addition, if a cause of death shows a ‘healthy worker effect’ that becomes less over time, the SMR should increase with increasing latency period. For all military Vietnam veterans, the SMRs for cancers, circulatory diseases, external causes and for all other causes combined showed no statistically significant trend with latency period.
The SMRs for 1980-94 are probably artefacts of the way in which vital status had to be inferred and the estimates calculated. This is especially so for the navy Vietnam veterans. Therefore, the SMRs need to be interpreted with caution.
The ‘healthy worker effect’ describes the effect of a selection bias which results in workers having lower death rates than the general population. The selection bias arises because the general population includes persons who are not fit enough to work or to remain in work. Death rates in employed populations are typically 70 to 90 per cent of those in the general population, but the effect usually diminishes with increasing age of the study cohort or time from entry into the cohort.
As for other workers, Vietnam veterans would be expected to have lower death rates than other Australian males because they were medically screened at enlistment and unfit men were rejected. Medical screening prior to departure for service in Vietnam may magnify this bias further.
The ‘healthy worker effect’ typically varies by cause of death. Several grouped causes of death include deaths whose antecedents would be sufficiently obvious in young men for them to be rejected for military service. For example, presence of childhood diabetes, current alcohol or drug abuse or physical or mental defects would be likely to be recorded or detected at an enlistment medical examination, and would result in rejection of the young man for military service. Since these factors may lead to an early death, the Vietnam veteran cohort excludes some men at high risk of an early death.
Conversely, it is difficult to predict risk of subsequent long-term cancer in young men, and the military selection process is unlikely to select recruits at a risk of subsequent long-term cancer death that is substantially different from the Australian male population. Therefore, the ‘healthy worker effect’ for cancers is usually less obvious.
If a specific factor is responsible for a high SMR in a group of Vietnam veterans, this factor may be associated with specific components (for example, causes of death or subgroups of Vietnam veterans) within that group. Analysis of the ratio of SMRs (that is, of standardised relative mortality ratios, or SRMRs) can assist in identifying these components and may suggest the reason for the elevation. Another advantage of using SRMRs is that they are much less affected by adjustments for ascertainment than SMRs – the bias is in effect cancelled out.
In the Vietnam veteran cohort, for the period 1980-94 there were large and statistically significant differences in the SRMRs between the major groupings of cause of death (Table 4).
As expected, low SRMRs occurred for causes of death such as endocrine, nutritional and metabolic diseases, mental disorders and congenital anomalies and the highest SRMR was for cancers. The SRMR for cancer deaths compared with other causes of death between 1980-94 for the military Vietnam veterans was 119 (95% CI 109 to 130). The SRMRs were similar for the three branches of service, and did not differ greatly over time.
|Major group of causes of death||Standardised relative mortality ratio (SRMR) a|
|(ICD-9 Chapter)||All||(95% CI)||Army||Navy||Air force|
|External causes||XVII||107||( 96, 119)||106||110||109|
|Circulatory system||VII||96||( 88, 105)||95||89||111|
|Digestive system||IX||95||( 78, 116)||96||98||88|
|Nervous system/sense organs||VI||93||( 66, 130)||94||99||74|
|Symptoms and ill-defined||XVI||87||( 43, 177)||95||51||119|
|Respiratory system||VIII||83||( 67, 103)||86||76||76|
|Blood/blood forming organs||IV||76||–||56||94||158|
|Infective and parasitic diseases||I||67||( 37, 119)||57||101||55|
|Endocrine, nutritional||III||66||( 50, 86)||62||83||49|
|Genito-urinary system||X||55||( 26, 118)||59||79||0|
|Mental disorders||V||47||( 27, 80)||39||43||111|
|Skin and subcutaneous tissue||XII||0||–||0||0||0|
a. For the ICD-9 chapters each SRMR is the ratio between the SMR for the chapter and the SMR for all other chapters combined.
There was little evidence for elevated death rates for the individual causes of death specified in the protocol (Table 5).
|Protocol specified cause of death||Standardised relative mortality ratio (SRMR) a|
|All||(95% CI)||Army||Navy||Air force|
|Chronic Obstructive Airways Disease||85||( 64, 115)||83||92||87|
|Cerebral haemorrhage||80||( 53, 122)||99||34||54|
|Neurological disorders||93||( 51, 171)||125||42||0|
|Ischaemic heart disease||104||( 94, 114)||108||83||114|
|Cirrhosis of the liver||94||( 75, 119)||96||92||89|
|Pancreatitis||105||( 52, 213)||133||66||0|
|Motor vehicle accidents||99||( 82, 120)||101||95||94|
|Suicide||114||( 97, 133)||120||100||98|
|Other external causes||103||( 88, 122)||94||129||128|
a. For the protocol identified causes each SRMR is the ratio between the SMR for the cause of death and the SMR for all other causes of death combined.
When specific cancer sites were compared with all other cancer sites combined, there was little evidence for elevated death rates for the individual cancer sites specified in the protocol (Table 6, column a). However, the SRMR for cancers of the head and neck + , which had not been identified in the protocol, was statistically significantly elevated, being 150 (95% CI 113 to 199). Lung cancer and prostate cancer were statistically significantly elevated when the alternative comparison, each cancer site compared with all other causes of death combined, was analysed (Table 6, column b).
|Protocol specified cancer site||All military|
|SRMR a||(95% CI)||SRMR b||(95% CI)|
|Non-Hodgkin’s lymphoma||85||(60, 131)||97||(68, 137)|
|Liver||49||(24, 100)||56||(24, 111)|
|Soft tissue, other sarcomas||83||(42, 162)||94||(43, 178)|
|Hodgkin’s disease||87||(36, 214)||99||(32, 230)|
|Prostate||128||(91, 179)||143||(102, 200)|
|Leukaemia||105||(73, 149)||118||(83, 168)|
|Multiple myeloma||52||(23, 118)||59||(22, 129)|
|Lung||109||(93, 128)||123||(106, 142)|
a. The SRMR is the ratio between the SMR for the specific site and the SMR for all other cancer sites combined.
b. The SRMR is the ratio between the SMR for the specific site and the SMR for all other causes of death combined.
The study has various strengths, and given the constraints of the study design, unavoidable weaknesses that affect its interpretation.
The strengths of this study include:
- identification of the veteran population was close to complete;
- follow-up or tracing was from the end of service until 31 December 1994 – the length of time that elapsed since service ranged from approximately 22 to 32 years;
- vital status was established for 96.9 per cent of veterans in the study; and
- there was racial homogeneity among veterans studied – virtually all were Caucasian and all were male.
The weaknesses of this study include:
- under-ascertainment of deaths when using linkage to the various databases. Although methods to adjust for this were used to estimate Standardised Mortality Ratios (SMRs) the bias could not be completely eliminated;
- lack of measurement of exposure to risk factors that might confound the observed associations, such as cigarette smoking, alcohol intake, Hepatitis B, and herbicide and dioxin exposures in Vietnam; and
- the only exposure measures available were corps groupings and length of service in Vietnam. These could not reflect occupational variations or the variation in exposure to Vietnam service within and between the army, navy and air force.
The death rate from all causes for Vietnam veterans relative to other Australian males was estimated to be 0.68 (95% CI 0.63 to 0.74) for 1964-79 and 1.07 (95% CI 1.02 to 1.12) for 1980-94. These estimated SMRs take account of the known variation in death rates by age group and calendar year. There was some evidence of excess mortality compared with the Australian male population. However, the overall level of excess mortality was difficult to estimate because of the continuation of a ‘healthy worker effect’ which would lower the death rates for some causes, and the bias in the estimation method due to the underestimation of deaths.
The ‘healthy worker effect’ varied by grouped cause of death and was particularly pronounced forcongenital anomalies, mental disorders, and endocrine, nutritional and metabolic diseases. This was expected because men likely to die from these causes would not have been accepted for military service.
Vietnam veterans had a higher SRMR for death from all cancers than for death from other causes of deaths before 1980 and between 1980-94. The higher SRMRs for cancer deaths compared with other causes of death are likely to reflect, at least in part, an expected differential ‘healthy worker effect’ by cause of death. This differential is due to a reduction in the number of deaths from other causes because of the rejection at enlistment of unfit men. The ‘healthy worker effect’ may have become less over time, particularly for respiratory diseases, but was apparent in death rates between 1980-94. This study did not find statistically significant differences in the ‘healthy worker effect’ between the three branches of service.
The pattern of cause of death between 1980-94 was similar for army, navy and air force Vietnam veterans. The AVH Study finding of differences in death rate between corps groupings within the army was not found in this study. Veterans from different corps groupings within each branch of service had similar overall death rates. However, national service army Vietnam veterans had lower overall death rates than regular army Vietnam veterans.
Compared with deaths from other sites of cancer, death rates from soft tissue and other sarcomas, non-Hodgkin’s lymphoma, and Hodgkin’s disease, which are consistently associated with exposure to herbicides or dioxin, were not statistically significantly elevated. This study suggests there may be a slight excess risk of death from lung cancer and prostate cancer. This elevation may be related to some exposure that occurred during service in Vietnam however, there are multiple other and more usual risk factors associated with both cancers that should also be considered. Although not hypothesised in the protocol for this study, Vietnam veterans may have been at increased risk of death from head and neck cancers between 1980-94.
Compared with other causes of death, death rates from external causes were not statistically significantly elevated among Australian military Vietnam veterans between 1980-94. This study suggests there may be a slight excess in risk of death from suicide. However, the number of deaths among Vietnam veterans was not statistically significantly different from the expected number. Compared with other causes of death, death rates from motor vehicle accidents were not statistically significantly elevated. There were few deaths expected or observed as being attributed to legal intervention.
- There is evidence of excess mortality among Vietnam veterans compared with the rest of the Australian male population. However, the overall level of excess mortality is difficult to estimate accurately because of the continuation of a ‘healthy worker effect’ which would lower the death rates for some causes, and the bias in the estimation method due to the underestimation of deaths.
- There is evidence that lung cancer and cancers of the head and neck play a part in the excess mortality among Vietnam veterans.
- There is evidence that prostate cancer plays a part in the excess mortality among Vietnam veterans.
- Although the estimated excess risk is not statistically significant, this study does not preclude a slight excess in risk of death from suicide.
- The key factor limiting the conclusions from this study was the proportion of deaths that could not be definitely ascertained. Thus, any future study should aim to reduce this proportion and to more satisfactorily adjust for it. More generally, the study was constrained by the sometimes incomplete data in the internal and external databases used. This prevented the correct identification of vital status for some Vietnam veterans. Veterans’ Affairs should investigate how it may update and better validate data on the vital status of Vietnam veterans.
- The information from the current study, including the date when each veteran was last known to be alive or dead, should be stored safely together with the nominal roll. Updated vital status data should be stored separately. This would facilitate the conduct of future studies. Future studies should include an assessment of the effect of updated follow-up on the conclusions of this study.
- The future mortality of Vietnam veterans should be monitored. This suggests repeating this study after 2000. Also, if records of vital status are regularly maintained, the mortality experience of Vietnam veterans could be periodically monitored to detect any emerging trends.
- The findings for lung cancer and cancers of the head and neck suggest that there may be a link between these and the rates of smoking and alcohol consumption among the Vietnam veterans. Veterans’ Affairs should explore and implement preventative strategies
- A series of smaller studies of Vietnam veterans should be considered with the aim of ascertaining the risk factors associated with Vietnam service for suicide, lung cancer and prostate cancer.
- Compilation of the nominal roll was fundamental to this study. However, it was a lengthy task and its difficulty was exacerbated by the time that has elapsed since the Vietnam War. So that the health of veterans from more recent wars, conflicts and peace-keeping missions may be easily monitored and researched, nominal rolls should be compiled for each occasion Australian military personnel have served since the Vietnam War. In the future, a nominal roll should be compiled during each war, conflict or peace-keeping mission involving Australian military personnel.
- For some individuals, it was evident that their record of service was inaccurate because data were missing or incorrect. Data appeared to be missing particularly when the period of service was relatively short or temporary. To compile accurate nominal rolls it is vital to be able to identify every person who has ever served in the war, conflict or peace-keeping mission regardless of the length or type of service. Therefore, in the future, details of service should be more accurately recorded and permanently kept.
This study has updated the 1984 Australian Veterans Health Study of the mortality of national service Vietnam veterans by determining their current vital status. In addition it has extended the Australian Veterans Health Study by identifying another 40 000 Vietnam veterans and determining their vital status.
The names and service details of over 59 000 Vietnam veterans have been recorded in the nominal roll compiled by this study. The roll will be a valuable resource for all future research into the health of Vietnam veterans. Indeed, this study is only the first of several that the Department of Veterans’ Affairs will conduct. Already under way is a study which will compare the mortality of national servicemen who served in Vietnam with the mortality of a group of national servicemen who served only in Australia. Also, the health of surviving Vietnam veterans will be the subject of a study that will be conducted in 1997.
* CI = confidence interval; 95% CI is the range in which, allowing for variability in study populations, there is a 95% chance of the true result falling.
+ Six cancer sites: tongue; gum and mouth; oropharynx; other lip, oral cavity; and larynx are classified as cancers of the head and neck.
Department of Veterans' Affairs, Canberra 1997 | dva.gov.au