There have been other cohort studies of mortality in meat workers, but emphasis has been on the report of excess cancer mortality.30,31,32,33,34,35,36
Information on mortality from non-malignant diseases has been sparse or non-informative because of the practice of combining several distinct causes of death into very broad categories. Thus, excess risks have been reported for diseases of the circulatory and digestive systems, endocrine, nutritional and metabolic diseases, and benign neoplasms and cirrhosis of the liver.22,23,30,33,34,36
We had previously reported on cancer and non-cancer mortality in this cohort,21,22,23
but the present mortality update is the first time ever that cause-specific non-cancer mortality in this or any occupational group has been investigated in such detail. Previous studies have been hampered because available statistical software for analysing cause-specific mortality in occupational studies have all grouped individual causes into very broad groups, resulting in lack of information on specific causes. In this study we examined 134 specific causes of non-cancer deaths. To our knowledge no published cohort mortality study has investigated this many before.
The main findings in this study are the excess occurrences of septicaemia, endocarditis, cardiac dysrhythmias, subarachnoid haemorrhage, nephritis and other kidney diseases in this group of wholesale manufacturing meat workers. These conditions are well known to be clinically related and can all be caused by infection, and could fit into the cardio-mycotic-embolic syndrome which is well described in medical textbooks.37,38,39
The high aerosol and skin exposures to transmissible agents that are known to occur in this group of workers because of their propensity for cuts, lacerations, dermatitis and other injuries40,41,42
put them at high risk of these agents getting into the body and possibly causing an underlying bacteraemia/septicaemia. Once these agents get access to the circulatory system, they may spread to target organs—primarily the heart, brain and kidney. While this is one of other possible explanations for the occurrence of these excesses of deaths in the cohort, it should be emphasised that the excesses could well have arisen by chance because of the small numbers involved. Hence great caution should be exercised in the interpretation of the findings. At this time these results should be regarded as exploratory as it is the first time that they have been reported on, and their validity will depend on confirmation in other studies.
A statistically significant excess of deaths from liver disease was confined to males in meat processing plants, and could be a chance finding or related to alcohol intake or could also be induced by occupationally acquired infection. Similarly, an excess of deaths from Parkinson's disease seem confined to men in abattoirs and meat processing plants, as none was observed in women in these plants or in the corresponding male controls. The cause of Parkinson's disease is unknown, but it is plausible that infection can cause damage to the cells of the substantia nigra in the brain, giving rise to the disease.
A case of bovine spongiform encephalopathy (“mad cow disease”) has been diagnosed in a cow in the US.43
New variant Creutzfeldt–Jakob disease (CJD) in humans is believed to be caused by the prion agent that causes “mad cow disease” in cattle. CJD has been previously linked with occupational and non-occupational exposure to animals, including butchers and other workers in the meat industry.44,45,46
Similarly, multiple sclerosis has been linked with exposure to sheep affected by swayback disease of the central nervous system47
although, to date, no deaths from CJD or multiple sclerosis have been observed in this cohort. However, this and other cohort studies of meat workers may play an important role for monitoring diseases in humans hypothesised to be linked to exposure to zoonotic agents that are found in food animals, as further follow-up of these cohorts is extended.
The studied described above had certain deficiencies. The cause-specific SMRs may not be comparable across the three groups if the individual SMRs vary across age strata and the age distributions of the groups markedly differ. However, all three populations came from the same union and as is seen in table 1, the age distributions were quite similar. Also, in the SMR analyses, information on race was missing for subjects without a death certificate. However, the close agreement between the cause-specific SMR and PMR results predicted by theory when the all-causes SMRs are close to unity evident in these results indicates that any bias resulting from these two potential problems was probably minimal.28,29
Typical of retrospective cohort studies, the ability to control for important occupational and non-occupational confounding factors was limited, as was detailed information on exposure, and this could have affected interpretation of the results. For example, the slight excess of cirrhosis of the liver in males could be alcohol-related.
Secondly, it is possible that underlying risks may have been missed because of insufficient statistical power to detect risk from rarer outcomes. For several causes, elevated SMRs/PMRs that were not statistically significant but were based on one or two deaths were observed (results not shown), and these may well be indicative of potential risk. For example, the SMR for meningitis in all race/sex groups in meatpacking plants was 4.6 based on three deaths, while none was observed in the control group, and higher risks based on single deaths were observed in three of the four race/subgroups in abattoirs (not shown). Similarly, analysis by latency was primarily limited by lack of statistical power, although there is an indication that the increased occurrence of other bacterial diseases (septicaemia) is of late onset. At the end of this follow-up only 20% of the members of this cohort have died, thus further follow-up might well reveal more definite associations.
Thirdly, the classification into abattoirs and meat processing plants used in this study may not be sufficiently detailed enough for accurately depicting the expected exposure gradient of higher exposure to transmissible agents in abattoir workers than in those in meat processing plants, and could well explain why for example the risks of death from acute rheumatic fever and subarachnoid haemorrhage were higher in meat processing plants than in abattoirs. For example, abattoirs and meat processing plants were not separated according to whether they handled pigs, cattle or sheep. It is well known that although high rates of infection with some microbial agents like bacillus anthracis
, brucella spp
, leptospira spp
, staphylococcus aureus
, streptococcus group A–E
, G & L, salmonella spp
and E coli
are commonly encountered in workers in cattle, pig and sheep abattoirs and meat processing plants; high rates of others are found only in particular plants.42
For example, high rates of infection with streptococcus suis type II and yersinia spp are found only in plants that handle pigs; bovine pustular stomatitis and rabies viral infections are associated with workers who handle cattle; infection with the virus that causes orf is seen primarily in workers exposed to sheep.42
Many of the plants in this study handled a combination of these animals, although some handled one type predominantly. Related to this is the inability of the study to take into the account specific tasks performed which may be related to different exposures; for example, wrapping was predominantly a female activity throughout the industry.20,48
Although we focused mainly on transmissible agents, it is possible that some of the excess occurrence of deaths observed may well be related to other exposures such as fumes from the wrapping machine,49
or even exposures such as nitrites and nitrosamines during curing,50
or preservatives such butylated hydroxytoluene or butylated hydroxyanisole which are known to enhance tumour formation in animals,51,52
or smoke during the smoking of meat,53
which some of these workers could have been exposed to. It should be pointed out though that activities such as curing and smoking of meat are usually carried out by only a handful of individuals.
Fourthly, in this update of mortality in the cohort, employment history was not updated because of subsequent loss of data on dues payment in the union database that occurred after the first follow-up while the union was converting from a hard copy based record system to a computerised one. Thus we could not examine risk by duration of employment.
Finally, causes of death reported on in tables 3 and 4 represent the 10 causes out of 134 examined for which a statistically significant increase was observed in any race/sex group within any of the three departments investigated. Therefore, because of multiple comparisons, it is possible that some of the increased risks observed could have been chance occurrences, in spite of the fact that they all could be clinically related to each other from an underlying infective aetiology. Because of these limitations, the interpretation of the findings must be tempered with caution.
In spite of these deficiencies, the study is of importance and has obvious advantages. This is one of the largest studies of meat workers exposed to cattle, pigs and sheep to date, and the only study of this occupational group in the US. No study to date has examined so many specific causes of death in any occupational mortality study. The cohort was uniquely completely defined. Because of the exceptional recordkeeping system of this union, everyone who had ever been employed in these plants—for even just a few days—had a record, making selection bias an unlikely factor that would explain the findings. Similarly, the presence of the control group of non-meat workers from the same union permitted the control for the healthy worker effect and also for any geographic variation in rates between the local general population and the US general population. Finally, the results were consistent in so far as an underlying infective process could potentially give rise to all of them, and the increased risks for these causes were observed irrespective of whether the US general population or unexposed workers of the same union was used as the comparison group. In fact in some cases no deaths were observed in the non-meat group. This pattern strengthens the case for these observations to be real, particularly as lost persons were assumed to be alive in the analyses. Further follow-up of this cohort and other similar cohorts worldwide, and the conduct of case-control studies nested within these cohorts that will permit detailed characterisation of exposures and controlling for occupational and non-occupational confounding factors should be encouraged, because such studies will be important in shedding light on some of the interesting findings in this study. For the moment the findings in this study should be regarded as preliminary and as hypothesis generating, drawing attention to the possibility that zoonotic infections might play a role in the occurrence of these diseases even in the general population.
- Workers who handle food animals or their products have high exposure to transmissible agents present in the animals or their products. The findings of this study indicate that these workers are at increased risks of dying from diseases of the heart and circulatory system, and neurological and kidney diseases, which may be caused by these agents.
- It is possible also that the occurrence of these diseases in the general population may be partly due to these agents.
This study confirms previous suggestions that these workers may be at risk of some of these diseases. There is therefore sufficient evidence for more detailed studies to be embarked on, and for consideration to be given to start adopting steps to minimise exposure to these agents in the workplace to protect workers' health.