Cohort Profile: Health Watch—a 30-year prospective cohort study of Australian petroleum industry workers

Health Watch was established in 1980 at the University of Melbourne to ‘explore possible links between occupational experience in the [petroleum] industry’ and subsequent mortality and cancer incidence. 1 There was concern that occupational exposure to fossil fuels could result in occupational cancers and it was anticipated that ‘well designed epidemiological studies, and carefully planned data bases’ were ‘capable of confirming or refuting such association’ rather than relying on clinician recognition of clusters. 2 Health Watch was ‘designed to operate as a general health surveillance system’ 2 comparable to a contemporary USA study funded by the American Petroleum Institute. 3 Later reports expanded the aim to be ‘a prospective epidemiological study with the capacity to test specific hypotheses regarding the possible health consequences of occupationally determined exposure to hydrocarbon and other chemical substances’. 4 This follows efforts by the industry’s occupational hygienists to link descriptive job categories to an index of hydrocarbon exposure. 4 The main aims of tracking mortality and cancer incidence in the cohort have not changed over time, but sub-analyses have been included e.g. for tanker drivers and for a group of men at a particular refinery. 5 After an excess of lympho-haematopoetic (LH) cancer was identified, 6 a case-control study was instituted to investigate whether benzene exposure was associated with this increased risk. 7–9 This study was followed up in 2008 with a pooled analysis to identify which subtypes of LH cancer might be most strongly associated with benzene exposure. 10–12 The industry has reduced benzene exposure over the period of the cohort. 13 In 1999, the custodianship of the cohort was transferred to the University of Adelaide and then in 2005 to Monash University’s Centre for Occupational and Environmental Health in the Department of Epidemiology and Preventive Medicine, where it currently resides. The Australian Institute of Petroleum (AIP) has funded the relevant universities, including Monash University, to implement the study with funds from their member companies in the Australian petroleum industry. The study has an advisory committee that meets regularly and includes the relevant university investigators and trades union and company representatives. The committee is chaired by a representative of the participating companies on a rotating basis and the AIP provides the secretariat. Health Watch participants are current or ex-petroleum industry employees from participating oil and gas companies at work sites across Australia. Participants come from all industry sectors including refineries, gas plants, distribution terminals including airports, and production sites both onshore and offshore. Employees in the retail sector or head office, and those at sites with fewer than 10 employees were not invited to participate. Participants were enrolled in the study by participating in one or more of the four periodic industry surveys. Most participants joined Health Watch during the first two surveys, during 1981–83 and 1986–87 ( Table 1 ). Additional participants were recruited in the third (1991–93) and fourth (1996–2000) surveys. Breakdown of participant recruitment in each Health Watch Survey Breakdown of participant recruitment in each Health Watch Survey Site rolls were provided to the Health Watch survey interviewers, and eligible employees were then approached at the worksite and invited to participate. The participant completed their personal details, and the remainder of the questionnaire was completed by the researcher during an interview. Refusal to participate was uncommon. The major cause of non-participation was difficulty in locating individuals because of temporary absences such as sick leave or annual leave. The participation rate was 93% in the first two surveys but was lower in the third and fourth surveys at 84% and 73%, respectively. No further information is available about those who did not participate. A participant was eligible to be included in the cohort analysis after completion of a survey and after having served 5 years in the industry. By the end of 2010, the cohort consisted of 18 013 participants, of which only 1374 were women, which reflects the small proportion of women in the industry. The breakdown of the cohort (in November 2010) is illustrated in Figure 1 . A representation of the Health Watch cohort structure as at 30 November 2010. 1 Exclusion due to withdrawal of a company from the AIP in 1994. All-cause SMR, all-cancer SIR for men in Health Watch from 1985 to 2005. Some participants completed more than one questionnaire if they were employed during successive waves of recruitment; 10% ( n = 1987) of cohort members completed all four surveys, 19% ( n = 3454) completed three surveys, 33% ( n = 5939) completed two surveys and 38% ( n = 6833) completed one survey. With the advisory committee’s consent, the employing companies provided information on participants, including job changes, resignations and retirements, to the Health Watch study. The study maintains contact with participants, sending updates on the cohort findings and asking for information on health, smoking and employment status via a regular ‘Health Letter’. Cohort members inform the study custodians of changes to their addresses, jobs, smoking and health status. The next of kin also report deaths to the study custodians. In recent years, as a higher proportion of the participants have retired, this reporting has been less comprehensive than in the early years of the study and a higher proportion of Health Letters are returned marked ‘not known at this address’. Of the 15 771 Health Letters sent to eligible participants in 2007, 6225 (40%) returned the questionnaire. Of the remaining 9546 non-responders, there were 1040 (11%) returned ‘not known at this address’. We have had updated information from 29 282 Health Letters between 1994 and 2010; this included 5212 changes of smoking status. A participant is considered lost to follow-up when reliable contact data are not available. If a ‘return to sender’ is received on mail sent by Health Watch, the provided telephone number is disconnected and/or the e-mail address is not functioning, then the participant is considered lost. In 2010, there were 696 participants who are considered lost to follow-up in the Health Watch cohort. These participants are still assumed to be resident in Australia and remain part of the cohort and are included in the cancer and death linkages and in the analyses. No active tracing of those who are lost to follow-up has taken place since 2003 when a substantial effort was made to locate Health Watch participants with whom contact was lost. This included a search of the Australian Electoral Commission Roll (registering for this Roll is compulsory in Australia), a telephone survey and a search of the Health Insurance Commission, the New Zealand National Health Index and the Department of Immigration, Multiculturalism and Indigenous Affairs. Participants remain in the Health Watch cohort for the length of the study unless they ask to be withdrawn ( n = 20). One company withdrew from the study in 1994 and the follow-up of the 239 employees ceased at that time. The questionnaires obtained information as detailed in Table 2 . The demographic details were completed by the participants and the workplace details entered by an interviewer. Data collected on participants by phase Demographic information: Name, previous names, address, date of birth, sex, county of birth, year of arrival in Australia (where applicable) Current employment information: Company and work site Date of first employment in the petroleum industry Current job: hours per week, shift work (Yes/No) Work areas and activities Hours per week by area Month and year started Previous petroleum industry jobs Petroleum industry jobs in previous 5 years: Hours per week, shift work (Yes/No) Work areas and activities Hours per week by area Month and year started and finished Working environment Details of concerns about current or previous workplace conditions Previous non-petroleum industry jobs Industry sector, job and employment period in calendar years Lifestyle information Current smoking, smoking history Alcohol consumption Health information Illness that required more than 1 week of medical treatment, year Cancer diagnosis, year Demographic information: Name, previous names, address, date of birth, sex, county of birth, year of arrival in Australia (where applicable) Current employment information: Company and work site Date of first employment in the petroleum industry Current job: hours per week, shift work (Yes/No) Work areas and activities Hours per week by area Month and year started Previous petroleum industry jobs Petroleum industry jobs in previous 5 years: Hours per week, shift work (Yes/No) Work areas and activities Hours per week by area Month and year started and finished Working environment Details of concerns about current or previous workplace conditions Previous non-petroleum industry jobs Industry sector, job and employment period in calendar years Lifestyle information Current smoking, smoking history Alcohol consumption Health information Illness that required more than 1 week of medical treatment, year Cancer diagnosis, year Data collected on participants by phase Demographic information: Name, previous names, address, date of birth, sex, county of birth, year of arrival in Australia (where applicable) Current employment information: Company and work site Date of first employment in the petroleum industry Current job: hours per week, shift work (Yes/No) Work areas and activities Hours per week by area Month and year started Previous petroleum industry jobs Petroleum industry jobs in previous 5 years: Hours per week, shift work (Yes/No) Work areas and activities Hours per week by area Month and year started and finished Working environment Details of concerns about current or previous workplace conditions Previous non-petroleum industry jobs Industry sector, job and employment period in calendar years Lifestyle information Current smoking, smoking history Alcohol consumption Health information Illness that required more than 1 week of medical treatment, year Cancer diagnosis, year Demographic information: Name, previous names, address, date of birth, sex, county of birth, year of arrival in Australia (where applicable) Current employment information: Company and work site Date of first employment in the petroleum industry Current job: hours per week, shift work (Yes/No) Work areas and activities Hours per week by area Month and year started Previous petroleum industry jobs Petroleum industry jobs in previous 5 years: Hours per week, shift work (Yes/No) Work areas and activities Hours per week by area Month and year started and finished Working environment Details of concerns about current or previous workplace conditions Previous non-petroleum industry jobs Industry sector, job and employment period in calendar years Lifestyle information Current smoking, smoking history Alcohol consumption Health information Illness that required more than 1 week of medical treatment, year Cancer diagnosis, year Mortality and cancer incidence in the cohort are obtained by linking the Health Watch data set with Australian National registries. The Australian Institute of Health and Welfare (AIHW) compiles the National Death Index and the Australian Cancer Database (ACD) on behalf of all State Death and Cancer Registries. The cohort was linked to these registries 10 times between 1984 and 2007. Initially the linkages were executed annually, then every 3 years and most recently every 5 years. Analyses are carried out by comparing the mortality and cancer rates in the Health Watch cohort with the expected rates calculated from the age- and sex-specific national rates. For some mortality causes and specific cancer types, within-cohort analyses are also carried out to examine the risk of different job groups, calendar period of employment and duration of industry employment. Lifestyle risks are also examined by analysing specific cancers and mortality rates, with self-reported smoking and drinking status and state of residence (for melanoma). The study has found a strong and sustained healthy worker effect in this industry for men and women over the 30 years of follow-up. 14 A common finding with the healthy worker effect is that it decreases as the cohort ages, that is, the standardized mortality ratio (SMR) tends to increase with time, approaching the mortality rates of the national population. There is less effect on the cancer standardized incidence ratio (SIR). This tendency is becoming evident for men, as shown by the trend lines in Figure 2 . In the 13th report, the overall SMR for men was 0.72 [95% confidence interval (CI) 0.68–0.76] ( n = 1473) and for women was 0.65 (95% CI 0.45–0.91) ( n = 34). The overall SIR for men was 0.99 (95% CI 0.94–1.04) ( n = 1467) and for women was 0.89 (95% CI 0.68–1.15) ( n = 58). 14 Cohort participants are less likely to smoke compared with the general Australian population. Within the cohort, the differences in mortality and cancer incidence between smokers and non-smokers have become more pronounced as the cohort ages. 14 For the male cohort members, the incidence of mesothelioma (SIR 1.76, 95% CI 1.12–2.65) and melanoma (SIR 1.29, 95% CI 1.13–1.48) were in excess, lung cancer was reduced (SIR 0.74, 95% CI 0.62–0.87). 14 There was a higher than expected incidence of leukaemia (SIR 2.8, 95% CI 1.7–4.5) identified in the 1992 Health Watch report. 5 However, the leukaemia incidence reported in 2007 was no longer in excess (SIR 0.92, 95% CI 0.65–1.27). 14 The initial excess of leukaemia led to two follow-up studies. The first of these was a case-control study nested in the cohort, which found a strong association between benzene exposure and risk of acute non-lymphocytic leukaemias but no association for non-Hodgkin lymphoma or multiple myeloma. 7 An increased risk of leukaemia was found for men exposed above 16 ppm-years (parts per million years) of benzene ( Table 3 ) compared with age- and sex-matched controls from the cohort. 8 When possible occasional high exposures were included in the exposure estimates, e.g. spillages during tanker or drum filling, the odds ratios were reduced ( Table 3 ) because these exposures are more likely to occur among higher-exposed individuals. 15 Analyses suggest that exposure to benzene within 15 years of diagnosis are more closely associated with increased risk of leukaemia than exposure more than 15 years previously. 16 Conditional (fixed-effects) logistic regression, leukaemia among men in Health Watch by cumulative exposure (ppm-years) [8] OR, odds ratio. Conditional (fixed-effects) logistic regression, leukaemia among men in Health Watch by cumulative exposure (ppm-years) [8] OR, odds ratio. The second follow-up study involved updating the Australian case-control study and then pooling the cases and controls with those of two similar cohort studies 17,18 of petroleum industry workers in Canada and the UK, respectively. This study examined the association between benzene exposure and subtypes of leukaemia categorized according to the recent WHO scheme. 19 Before the pooled analyses were carried out, data quality-checking procedures examined the data to ensure that it could be pooled. There was a review by haemopathologists of the available diagnostic data for cases 10 and the benzene exposure estimates were compared and some realignment carried out. 13 The study also included certainty scores for the diagnoses and the job exposure estimates. 10 Findings from this pooled study have been published and showed that there was an excess of myelodysplastic syndrome but not of acute myeloid leukaemia associated with low levels of benzene exposure. 10–13 Results from the Health Watch cohort study have been published in periodic reports and in scientific journals. 2,4,5,20–26 Periodic Health Watch reports are published on the Australian Institute of Petroleum website [ www.aip.com.au ] and on the Monash University website [ www.coeh.monash.org/healthwatch.html ]. Summary reports are distributed to all members of the Health Watch cohort for whom contact details are available. A major strength of Health Watch is that there is at least one personal interview record for every participant in the cohort. The interview-based data provide considerable detail about jobs and tasks performed during employment in the industry. Details of smoking history and alcohol intake are available for each participant, although much of these data were collected many years ago and smoking rates may have changed over time. Some of this information has been updated for some participants through recent Health Letters. Written consent was obtained from members of the cohort to search for their names in the ACD. Such informed consent was unusual at that time and showed foresight by the study’s founders in 1980. Identification of cancer incidence is a strength of the study, because cancer registration is mandatory in all Australian states and territories and is considered to be complete. However, the probabilistic matching cannot be guaranteed to identify all cases, but the study utilizes several cross-checking and validation procedures to ensure maximum ascertainment of possible cancers. Participation in Health Watch is voluntary and this could cause bias if those motivated to participate had a different health status from non-participants. The high participation rates (93%) in the first two surveys make volunteer bias very unlikely. The lower participation rate in the fourth survey (74%) resulted from a lack of new recruits from offshore production. 26 This did not greatly alter the overall composition of the cohort: 4.0% of the cohort were in the offshore production sector before the fourth survey and 3.7% afterwards. The high participation rate in the first three surveys is a reflection of the local management support and personal approaches by researchers mediated by site nurses who were known and trusted by employees. The lower participation rate in the fourth survey was probably because offshore employees were surveyed by postal questionnaire. A potential weakness of the study is that the date of first employment is self-reported. This could affect analyses by time-related variables, i.e. era of employment, duration of employment and time since employment. The personnel records of most companies have been overhauled in recent years and these data could not be verified. However, the error is likely to be random and hence unlikely to lead to bias. Moreover, errors from imperfect recollection of the year of hire are likely to be small in relation to the size of time-related categories (e.g. period of employment categories are pre-1954, 1955–64, 1965–74, 1975–84 and post-1985). Date of termination is obtained from participating companies but is not always complete. A check of company employment records and other follow-up measures minimized the errors from this source. 26 In the past 10 years, company update information has not been as complete and a much higher proportion of the participants have since retired. The Health Watch data are owned by the AIP, although Monash University has custodianship of the cohort. Applicants to use the data should approach Monash University in the first instance. Monash will discuss the matter with the AIP, the Health Watch advisory committee and the Monash University human research ethics committee (MUHREC). There may be restrictions by the ethics committee on provision of cancer incidence data outside Australia. In addition, ethics approval would be required from State and Territory Cancer Data Custodians and Registries and the AIHW. Data have been used for an international combined case-control study investigating the association with LH cancer and benzene exposure. To find out more contact: Malcolm R Sim; e-mail: [malcolm.sim@monash.edu]. Profile in a nutshell Health Watch, established in 1980, is a prospective cohort to investigate mortality and cancer incidence in the Australian petroleum industry. It includes individuals across Australia who worked for at least 5 years in any petroleum industry sector except retail. Employees were enrolled by participating in one or more surveys carried out in 1981–83, 1986–87, 1991–93 and 1996–2000. In 1983 there were 8353 men and 373 women in the cohort. In 2010 the cohort had 16 639 male participants and 1374 women. Participants are contacted periodically to update their details; 20 individuals have withdrawn from the cohort. Participants completed questionnaires on employment, smoking, alcohol and health status. The cohort is linked to national death and cancer registries and mortality and cancer incidence are compared with those expected from Australian population rates. The data are owned by the Australian Institute of Petroleum (AIP), although Monash University has custodianship of the cohort. Applicants who wish to use the data should approach Monash University in the first instance: [ www.coeh.monash.org/healthwatch.html ]. This work was supported by the Australian Institute of Petroleum (AIP) which funded the relevant universities, including Monash University, to implement the study with funds from their member companies in the Australian petroleum industry. Conflict of Health Watch is funded by the Australian Institute of Petroleum through a to Monash The have no other

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.