Analysis was based on deaths for which there was an adequately described occupation (1,202,888 men and 408,152 women). Associations between job group and cause of death were summarised by sex-specific proportional mortality ratios (PMRs) standardised for age (in five-year bands) and social class. PMRs were expressed as percentages, with associated 95% confidence intervals (CIs) based on the Poisson distribution.

Table 2 shows mortality in the same job groups from four categories of injury and from suicide. Deaths from falls and from injury by fire, were generally more frequent than expected, the largest excesses being for deaths caused by falls on stairs. On the other hand, PMRs for motor vehicle traffic accidents and suicide were consistently (and often significantly) less than 100.

Table 3 lists job groups with significantly elevated PMRs for HIV/AIDS, and shows also their PMRs for viral hepatitis and the ratios of PMRs for the two diseases. Several of the PMRs for HIV/AIDS were remarkably high, notably in tailors and dressmakers (918, 95%CI 369-1890, in men; 804, 95%CI 219-2060, in women), and in male hairdressers (918, 95%CI 717-1160). Furthermore, most of the job groups with high mortality from HIV/AIDS also had more deaths than expected from viral hepatitis. For most job groups, the ratio of the PMR for HIV/AIDS to that for viral hepatitis was between 0.8 and 1.3, and for several others, the 95%CI for the ratio embraced this range. Exceptions, however, were male hairdressers (ratio = 5.8, 95%CI 1.6-49) and “other service personnel” (ratio = 3.5, 95%CI 1.9-6.6).

Analysis of deaths by place of birth indicated that among cooks and kitchen porters the proportion of men who had been born in South Asia (6.9%) was higher than in all occupations combined (1.8%). Moreover the proportion of deaths from viral hepatitis was much higher in men born in South Asia (0.4%) than in those born in the UK (0.03%).

Our method of analysis was subject to various well-documented limitations. Errors are known to occur in the recording of occupations and causes of death on death certificates,[4,5] and even where accurate, occupational data related only to the decedent’s last full-time job. Also, the PMR of a job group for a specific cause could be spuriously inflated if mortality in the job group from all causes combined was unusually low. However, it is extremely unlikely that biases of this sort could explain the substantially elevated PMRs that we have demonstrated in this report.

The job groups with highest mortality from alcohol-related diseases were mostly in the catering industry, a finding that has been noted before,[6] and which has persisted through to 2005.[7] It is unsurprising, given the ready access to alcoholic drinks in these occupations. Indeed, publicans and bar staff may be directly encouraged to consume alcohol during the course of their work by offers of free drinks from customers. To this extent, their risk of alcohol-related disease could be regarded as a true occupational hazard. In addition, however, there may be a tendency for people who drink more heavily, selectively to seek jobs in establishments where alcohol is served.[8] Moreover, employment in such environments may encourage a culture of drinking outside working hours.

Cultural influences of this sort could also be one reason why male seafarers have high mortality from alcohol-related diseases. Many sailors work in small, closed communities, in which drinking habits can easily be transmitted to others. In addition, the stresses associated with absence from home for prolonged periods may be a further contributing factor. High rates of alcohol-related disease in seamen have been reported also in other countries,[6,9,10] and should be a priority for preventive action. There is scope for employers to control access to alcohol while ships are at sea, as some already do.[11] Furthermore, statutory periodic health checks of a kind that British merchant seamen are obliged to undergo, offer opportunities for health promotion at the individual level.

Interestingly, relative mortality from different alcohol-related diseases differed markedly between job groups. Caterers and cooks and kitchen porters had higher PMRs for liver cancer than publicans and bar staff and seafarers, but lower PMRs for oral cancer, laryngeal cancer, cirrhosis and “other alcohol-related diseases”. Their lower mortality from oral and laryngeal cancer may in part reflect a lower prevalence of smoking, in support of which, they also had their lower PMRs for cancer of the bronchus and COPD (Table 1). In addition, however, it is notable that caterers and cooks and kitchen porters also had unusually high mortality from viral hepatitis (Table 3). Chronic hepatitis B and C infections are a major cause of liver cancer in their own right,[12,13] and the risk of liver cancer from these infections is enhanced by alcohol consumption.[13,14]

Occupations with high mortality from diseases related to alcohol also had high PMRs from injury by falls (especially falls on stairs) and by fire (Table 2). This seems likely to reflect an increased risk of such injuries in people who are intoxicated by alcohol, and is consistent with previous findings on mortality by occupation from falls on stairs.[15] In contrast, however, their mortality from motor vehicle traffic accidents and suicide was lower than expected. In the case of motor vehicle accidents, this may be because people in the occupations concerned drive less than those in other jobs.

Mortality from HIV/AIDS varied widely between jobs, with PMRs in male hairdressers and tailors and dressmakers more than nine times the average for all occupations. A major factor in this is likely to be differences between job groups in the prevalence of homosexuality, rates of HIV infection being higher in male homosexuals.[16] A second influence may be the frequency of intravenous drug abuse, shared use of needles being another important route by which the virus is transmitted.[17] Hepatitis B and C can also be transmitted in this way, and this may explain why many of the job groups with high PMRs for HIV/AIDS also had high mortality from viral hepatitis. In addition, rates of hepatitis are likely to be elevated in occupations that include higher proportions of immigrants from countries where the infection is prevalent in early life. For example, the cooks and kitchen porters in our study included a relatively high proportion of men who had been born in South Asia (6.9% compared with 1.8% for all occupations combined), while the overall proportion of deaths that was from viral hepatitis was much higher in South Asian men (0.4%) than among men born in the UK (0.03%). This additional reason for hepatitis in some occupations may partly explain why the ratio of the PMR for HIV/AIDS to that for viral hepatitis varied between job groups.

Consistent with an increased risk of HIV/AIDS and viral hepatitis from shared use of needles, male literary and artistic occupations were among the job groups with high mortality from drug dependence and accidental poisoning by drugs. Interestingly, however, most of the other job groups with elevated PMRs for drug-related deaths were in the construction industry (painters and decorators, bricklayers and masons, plasterers, and roofers and glaziers). This may be because men who abuse drugs tend to seek work in casual jobs, and it does not necessarily imply a culture of drug abuse in the construction industry. Nevertheless, it could contribute to the high rate of occupational injuries in construction trades.[18] There would be merit, therefore in further research to establish the scale of the problem, and the potential for targeted interventions to reduce it.