Article Text
Abstract
Objectives To assess associations between occupation and motor neuron disease (MND).
Methods We conducted a population-based case–control study with cases (n=321) recruited through the New Zealand Motor Neurone Disease Association and hospital discharge data. Controls (n=605) were recruited from the Electoral Roll. Information on personal and demographic details, lifestyle factors and a full occupational history was collected using questionnaires and interviews. Associations with ever/never employed and employment duration were estimated using logistic regression stratified by sex and adjusted for age, ethnicity, socioeconomic deprivation, education and smoking.
Results Elevated risks were observed for field crop and vegetable growers (OR 2.93, 95% CI 1.10 to 7.77); fruit growers (OR 2.03, 95% CI 1.09 to 3.78); gardeners and nursery growers (OR 1.96, 95% CI 1.01 to 3.82); crop and livestock producers (OR 3.61, 95% CI 1.44 to 9.02); fishery workers, hunters and trappers (OR 5.62, 95% CI 1.27 to 24.97); builders (OR 2.90, 95% CI 1.41 to 5.96); electricians (OR 3.61, 95% CI 1.34 to 9.74); caregivers (OR 2.65, 95% CI 1.04 to 6.79); forecourt attendants (OR 8.31, 95% CI 1.79 to 38.54); plant and machine operators and assemblers (OR 1.42, 95% CI 1.01 to 2.01); telecommunications technicians (OR 4.2, 95% CI 1.20 to 14.64); and draughting technicians (OR 3.02, 95% CI 1.07 to 8.53). Industries with increased risks were agriculture (particularly horticulture and fruit growing), construction, non-residential care services, motor vehicle retailing, and sport and recreation. Positive associations between employment duration and MND were shown for the occupations fruit growers, gardeners and nursery growers, and crop and livestock producers, and for the horticulture and fruit growing industry.
Conclusions This study suggests associations between MND and occupations in agriculture and several other occupations.
- epidemiology
- public health
Statistics from Altmetric.com
Key messages
What is already known about this subject?
A number of possible occupational/environmental exposures have been suspected of contributing to the risk of developing motor neuron disease (MND).
What are the new findings?
We observed positive associations between the risk of MND and a range of occupations within agriculture in both men and women.
Positive duration–response associations were also seen in horticultural occupations.
Positive associations were also found for building trades workers, electricians, telecommunication technicians and forecourt attendants.
How might this impact on policy or clinical practice in the foreseeable future?
These results have confirmed previous findings and generated a range of hypotheses for specific occupational risk factors for MND.
If specific causal exposures can be identified, they may provide important opportunities for the prevention of MND.
Introduction
Motor neuron diseases (MND) are progressive and terminal neurodegenerative conditions affecting the motor neuron system, with death usually occurring within 2–5 years after the first symptoms of weakness.1 2 Amyotrophic lateral sclerosis accounts for 70% of cases1; other forms include progressive muscular atrophy, progressive bulbar palsy and primary lateral sclerosis.1
There is some evidence of increasing incidence and mortality rates of MND among high-income countries including New Zealand in the last two decades,2 3 with MND mortality in New Zealand (2.8/100 000) reportedly higher than the estimated mean global mortality (1.7/100 000).4 The reasons for the increased incidence remain unclear but are likely due to environmental and lifestyle factors, since genetic factors vary little over time and familial MND is relatively uncommon (5%–10%).1 2
Several studies have reported increased relative risks for certain occupations and occupational exposures,5 6 suggesting a role for agrichemicals,7 8 extremely low-frequency electromagnetic fields (ELF-EMFs),9 electric shocks,10 some heavy metals,2 welding fumes11 and solvents,12 although the evidence is equivocal.
We report the findings of the first New Zealand population-based case–control study on modifiable risk factors of MND, with a focus on occupational risk factors.
Methods
Study population
A national Motor Neuron Disease Registry was not available at the time of study commencement (a national registry has since been established).13 Incident and prevalent cases (n=295) were invited between 2013 and 2016 through the Motor Neurone Disease Association of New Zealand (MNDANZ). This was supplemented by records contained in the New Zealand National Minimum Dataset (NMDS), a national collection of public and private hospital discharge information including coded clinical data for inpatients and day patients.14 Incident cases were defined based on a primary or secondary diagnosis of MND (International Classification of Diseases, 10th Revision, code G122) for the period 2013–2015, and surviving cases (n=103) in the NMDS but not registered with MNDANZ were invited. Two of these were misclassified and excluded, leaving 396 eligible cases. The inclusion criterion for cases was a diagnosis by a neurologist, with all forms of MND included.
Controls were randomly selected from the New Zealand Electoral Roll (2008) with two controls for each case, frequency matched by age (5-year categories, based on the age distribution of the UK MND incidence distribution)15 and sex. Controls with a neurodegenerative disease were excluded.
Of the 396 eligible cases, 390 responded to invitation letters. Of these 44 were not eligible (27 deceased and 17 in intensive care), 25 (6%) refused to participate, leaving 321 participants equating to a 92% response rate.
Of the 2400 potential controls, 333 (14%) could not be contacted, 230 (10%) were returned to sender and 587 (24%) were not eligible. Of the remaining 1250 controls, 645 declined. Thus, 605 participated in the study, equating to a 48% response rate.
All study participants gave written informed consent.
Data collection
Identical data collection methods were used for cases and controls. These included a face-to-face (59% of cases and 16% of controls) or telephone interview by research nurses (23% of cases and 66% of controls) or a postal questionnaire (18% in cases and 18% in controls). Three cases used a proxy (family member) for the face-to-face interview and six used proxy assistance for reading and writing.
We used a European questionnaire16 with modifications to adapt it to New Zealand (with particular emphasis on agriculture) to collect information on demographic and personal data, lifestyle factors and lifetime occupational history.
Classification of occupational histories
Participants listed all jobs ever held for 6 months or more, and for each job provided information on job title, employer’s name, industry, the year and month in which the job began and ended and a detailed description of tasks performed and work processes undertaken.
Each job was classified according to the New Zealand Standard Classification of Occupations (NZSCO99)17; industries were coded according to the Australian and New Zealand Standard Industrial Classification.18 The occupational coding was based on the full job description, rather than on job title alone. Response outside scope was used for responses, such as ‘housewife’, ‘pensioner’ or ‘student’, which are not covered by NZSCO99. The industry code was based on information provided on the activity of the employer. All coding was done blind to case–control status.
Statistical analyses
Analyses were conducted using SAS V.9.3. Differences in general characteristics between cases and controls were tested using χ2 tests. Unconditional logistic regression was used to estimate ORs and 95% CIs, for ever-employed compared with never employed/self-employed in a particular occupation or industry.
Analyses were also stratified by sex, because men and women have different occupational profiles. Therefore, the specific occupational risk factors contributing to MND may differ between men and women. Analyses were adjusted for age (5-year categories), ethnicity (European/Pakeha, Maori, Pacific and others), highest education level (primary school or secondary school, technical or trade school diploma, undergraduate university degree and postgraduate university degree), smoking (never, ex-smokers and current) and for socioeconomic deprivation status using the New Zealand Deprivation Index (NZDep2006).19 NZDep is census-based with a relative deprivation score assigned to geographical meshblocks based on place of residence recorded on the Electoral Roll (with 1 representing the least and 10 representing the most deprived areas).
In order to establish the role of duration of employment, categorical variables were constructed for each job/industry using cut-points of <2, 2–10 and >10 years. These cut-points, which we have previously used in studies on occupational risk factors and cancer,20–22 ensured that sufficient numbers of cases and controls were available in each category. These categorical variables were included in the logistic regression using never employed in the occupation/industry as the reference. A test for trend was performed by fitting it as a continuous variable in the model.
Lag time analyses to take into account potential disease latency were conducted, in which employment 5, 10, 15 and 20 years prior to the interview date was disregarded. Analyses were repeated while adjusting for the mode of interview.
To reduce the number of associations presented, tables 2 and 3 only include results for broad occupation and industry categories (one-digit codes), irrespective of statistical significance, as well as results for specific occupations and industries (2–5 digits) if the association was statistically significant (p<0.05) and based on at least 10 subjects (cases plus controls). Results for all 2755 occupations and 3149 industries are available in supplementary tables (see online supplementary tables 1 and 2), as well as results by sex (see online supplementary tables 7 and 8).
Supplemental material
Results
Population characteristics
Population characteristics are described in table 1. MND was more common in men (64%) than women (36%), and most cases occurred over 60 years of age. While the 70+ year age group was over-represented in the controls, there was little difference between cases and controls in terms of smoking, ethnicity and education. However, there was a difference in socioeconomic deprivation status for men, with cases being less deprived compared with controls. There was no difference in the number of occupations held by cases and controls (mean=6.8 for cases and controls). The median and IQR of age was 64 and 13 for cases and 68 and 15 for controls. There were 225 incident and 96 prevalent cases, and the time between diagnosis and interview was 6–18 months (median=238 days, IQR=269 days).
Broad occupation and industry categories
Tables 2 and 3 present the findings for MND risk associated with occupations and industries overall and by the duration of employment.
Ever-employment in the following broad occupation categories (one digit, table 2) showed an increased risk: service and sales workers; agriculture and fishery workers; plant and machine operators and assemblers; and elementary occupations. A reduced risk was observed for clerks.
Statistically significant increased risks for ever-employed in the broad industry categories (one digit, table 3) were observed for: agriculture, forestry and fishing; mining; and construction.
Specific occupations within the broad occupation and industry categories
Agriculture and fishery workers
Significantly elevated risks were found for field crop and vegetable growers; fruit growers; gardeners and nursery growers; crop and livestock producers (table 2), with similar risks for both men and women (see online supplementary table 7). Positive and statistically significant associations between employment duration and MND were shown for most of these groups (table 2). A significant increased risk was also found for fishery workers, hunters and trappers although based on small numbers (table 2). By contrast, no increased risk was observed for livestock producers, the largest 4-digit group within agricultural workers (see online supplementary table 1).
Similar results were observed in analyses by industry category, with significantly elevated risks in agriculture, in particular, horticulture and fruit growing, with ORs increasing by longer duration (table 3) and with similar risks for both men and women (see online supplementary table 8). For grain, sheep and beef cattle farming and dairy cattle farming, there was no statistically significant risk (see online supplementary table 2).
Building trades workers
Employment as building trades worker was associated with elevated risk (table 2), particularly for builders and electricians. These associations were only found in males as there were very few women in these occupations. Risks did not increase with duration of employment.
Analysis by industry also showed a statistically significant increase in risk for construction, particularly in general construction, non-building construction and road and bridge construction (table 3), but notably not in painting and decorating services (see online supplementary table 2).
Service and sales workers
An increased risk was observed among service and sales workers (table 2). Within this heterogeneous category, women who had ever worked as caregiver had a statistically significant increased risk (see online supplementary table 7), and a similar result was observed for women who had worked in non-residential care services industry (see online supplementary table 8). However, increased risks were not observed for other healthcare related occupations or industries.
A particular high risk was found for working as a forecourt attendant (table 2), and similar results were also found for employment in both car retailing and automotive fuel retailing industry (table 3). None of the other retail trade sectors was associated with a statistically significant increased risk (see online supplementary table 2).
Other occupations and industries
Occupations in white-collar categories were generally associated with a lower risk, with a significant inverse association found for clerks (table 2). While male finance and administration managers showed a decreased risk, in contrast, women in this job showed a significantly elevated risk (see online supplementary table 7). However, within white-collar occupations, an elevated risk overall was found for men who worked as physical science and engineering technicians (see online supplementary table 7). Within this occupation group, telecommunications technicians and draughting technicians both had increased risks (table 2).
An elevated risk was observed for plant and machine operators and assemblers (table 2); this risk did not increase with duration.
Analyses by industry also showed that men having worked in the sport and recreation industry were associated with an increased risk (see online supplementary table 8) but not for women. A similar excess was observed in mining especially other mining (table 3).
Neither latency analyses (see online supplementary table 3) nor adjustment for mode of interview (see online supplementary table 4) made any appreciable difference.
Discussion
This study found that certain occupations in agriculture and construction were associated with an increased risk of MND, which are consistent with prior studies,8 thus further supporting that occupation may be an important aetiological factor for MND. This study also identified other occupations associated with increased risk including building trades workers, electricians (electrical occupations), telecommunications technicians, draughting technicians, forecourt attendants, caregivers, and plant and machine operators and assemblers.
Agricultural workers
A major finding was the strong association between agricultural employment and MND, with several horticultural occupations within this group showing increased risks. Similar results were observed for analysis by industry. When the duration of employment was considered, the risk increased monotonically for market farmers and crop growers, fruit grower and gardeners/nursery growers. The presence of an increased risk for multiple non-overlapping occupational groups, the presence of positive duration–response associations and the presence of increased risks for both men and women in these occupations, strongly suggests these are not chance findings.
We found no significant difference in urban/rural residency between cases and controls (see online supplementary table 5), suggesting it is unlikely that risk factors associated with urban/rural residency could be responsible for the observed increased MND risks for agricultural workers. To test whether these associations could be explained by differences in urban/rural residency between participating and non-participating controls, the geographical meshblock for place of residence for all potential controls were linked to New Zealand geographic concordance files to obtain their urban/rural classification,23 which was then compared between participants and non-participants (see online supplementary table 5). This showed that participating controls were slightly more likely to live rurally (18%) compared with non-participating controls (14%), suggesting that participation bias could not explain the observed increased MND risks for agricultural workers.
Our findings are consistent with prior studies that observed increased MND risk among farmers and agricultural workers24–26 and workers exposure to herbicides/pesticides.27 28 Also, several meta-analyses6 8 29 have shown that previous exposure to agricultural chemicals, especially to pesticides, is associated with MND. Pesticide exposure is also a plausible explanation for the risk patterns observed in this study, given that risks were mainly elevated for agricultural occupations and industries in fruit and crop growing, while agricultural occupations and industries primarily in livestock production did not show an increased risk.
Construction workers
A strong association was observed with construction workers, particularly building trades workers and general labourers. The analysis by industry category confirmed this, and results are also consistent with earlier studies in construction workers,12 30 heavy labour and blue-collar occupations.31 Associated exposures to dusts, heavy metals2 and repetitive and strenuous work have also previously been shown to be a risk factor. As blue-collar workers have been related to lower socioeconomic deprivation status and higher smoking rates,32 these confounders were considered in our study. Although male cases were more deprived on average compared with controls, and there were no differences in education and smoking status between cases and controls in our study, we also adjusted for socioeconomic deprivation status, education and smoking status. Therefore, the general pattern of increased MND risk for blue-collar occupations is unlikely due to confounding.
This study showed an elevated risk for electricians and telecommunications technicians, which is consistent with previous studies showing associations with electrical occupations.33 34 Exposure to ELF-EMFs or electric shocks have been suggested as an explanation for these findings.6 9 35
Other occupations
An increased risk was observed among forecourt attendants and in the automotive fuel retailing industry. While this association has not previously been reported (possibly due to the absence of a specific code for forecourt attendants in occupational classifications used in other studies), increased risks associated with exposures experienced by forecourt attendants have been shown, in particular exposure to lead, which was used as a fuel additive until 1996 in New Zealand.36 A Spanish study37 found that MND mortality was associated with higher air lead levels, and a recent Australian study38 showed a 1% increase in lifetime petrol lead exposure increased the MND death rate by approximately one-third of a percent.
Other significant associations were observed in plant and machine operators and assemblers. This is a heterogeneous occupational group including stationary machine operators as well as vehicle drivers, but none of the specific occupations within this group showed an increased risk. The increased risk may, therefore, be associated with non-specific exposures such as cutting, cooling, or lubricating oils,12 diesel exhaust emissions39 and ELF-EMFs.9
We also observed an elevated risk for women caregivers but not for other healthcare related occupations, although two mortality studies25 40 showed that female nurses and medical services workers had an increased risk for MND.
Strengths and limitations
Using the MNDANZ national register, the NMDS and the New Zealand Electoral Roll to identify cases and controls was an important strength of this study. In particular, the MNDANZ national register and NMDS provided a reliable source for all patients with MND in New Zealand, and the Electoral Roll records virtually all New Zealand citizens and permanent residents in the age of particular relevance to this study (ie, >40 years).41 These sources are representative of the general population that generated the cases. Misclassification of disease status was also minimised as cases were diagnosed by a neurologist, and diagnosis details and neurologists’ contact details were provided by all cases. The use of both prevalent and incident cases was necessary to achieve an adequate sample size, but as the time between diagnosis and interview (6–18 months) was short and within the normal survival time for all cases, this was considered unlikely to introduce a bias. Additional analyses excluding prevalent cases did not alter our main findings, apart from wider confidence intervals due to lower numbers. We also ran additional models adjusting for sports and alcohol consumption, but this made little difference to the risk estimates and did not alter our findings. Another important strength of the study was that full occupational histories were collected from all cases and controls without the use of proxies to answer the questionnaire, a particular advantage compared with studies based on mortality and cause of death data. The study is also relatively large in comparison with many other case–control studies focusing on occupation31 42 and particularly compared with small clinic-based samples.43 44
The limitations include the reliance on self-reporting, which could introduce recall bias. To minimise this, the lifetime work history questionnaire was provided to every participant a few weeks before the interview to allow sufficient time to recall their work history, and the interviewers were trained to probe for the full occupational history without any gaps. There was no difference in the number of occupations held by cases and controls (mean=6.8), and there was, therefore, no indication of recall bias in the occupational histories (ie, cases recalling particular jobs more often than controls), although this cannot be fully excluded.
Another limitation was the lower response rate in controls (48%) compared with cases (92%). We assessed whether participation was associated with occupation by comparing the occupation, as recorded on the Electoral Roll, between participating and non-participating controls. The frequency of digit 1 and 2 job codes showed no difference between participating and non-participating controls for the occupations for which we found an increased risk, for example, 61: market-oriented agricultural and fishery workers, 4.29% non-participating controls versus 4.63% participating controls (see online supplementary table 6). Although these comparisons were based on one occupation and not the full occupational history, they provide no indication that the increased risks observed in this study are explained by non-response bias.
There were nine cases with proxy, all of whom were proxy assisted for the interview only. Given that this represents only 2.8% of the total case population, we consider that any bias resulting from this would be negligible.
There were also differences in the interview method used between cases and controls. For cases, it was often difficult to engage in a long telephone interview or to complete the full postal questionnaire. As a result, 62% of cases preferred a face-to-face interview, with only 18% interviewed over the phone and 20% completing a postal questionnaire. In controls, 65% preferred a telephone interview, 17% chose a face-to-face interview and 18% completed a postal questionnaire. To minimise potential bias, the completeness of questionnaires was checked, and follow-up interviews by telephone were made for all cases and controls where there was missing or incomplete data. We also did an additional analysis by repeating all analyses controlling for the interview method in the model, which made little difference and did not alter our findings.
Genetic data were not available as genetic testing is not routinely offered to patients in New Zealand, unless there is a clear family history, and then often only at the request of the patient.13 However, familial MND only accounts for 5%–10% of all MND cases, and genetic differences are therefore unlikely to explain our findings.
The other limitation was that the age distribution between cases and controls was different between men and women. This is likely due to age matching controls using the age distribution of MND incidence in the UK, which may be different from that in New Zealand (equivalent New Zealand data was not available at the time of participant recruitment).
Conclusions
The findings of this study suggest that MND risks may be associated with certain occupations and industries in New Zealand. In particular, several agricultural occupations were associated with an increased risk. Agriculture also represented the largest occupational group for which an increased risk was observed (ie, 33% of cases and 24% of controls had worked in agriculture), illustrating that occupational risk factors for MND may have a high prevalence in the New Zealand population. If specific causal exposures can be identified, this may provide important opportunities for the prevention of MND. We also observed increased MND risk for other large occupational groups such as building trades workers, plant and machine operators and assemblers and unspecified labourers and also for smaller more specific occupational groups including care workers, forecourt attendants, telecommunications technicians, draughting technicians and electricians. These results have suggested specific occupational risk factors for MND (eg, agricultural chemicals, organic solvents, metals, ELF-EMFs and electric shocks) that merit further scrutiny in future analyses.
Acknowledgments
We would also like to thank Michelle Gray, Di Marshal, Heather Duckett, Deirdre Thurston and Aly Timmings for their help with conducting interviews.
References
Footnotes
Contributors We are grateful to the study participants, the Motor Neurone Disease Association New Zealand and their field staff for their generous support and contribution to the study.
Funding The study was funded by a grant from the Health Research Council (HRC) of New Zealand (Part of 11/1041 HRC Programme Grant – Building Research in Occupational Health in New Zealand).
Competing interests None declared.
Ethics approval Ethical approval was granted by the New Zealand Multi-region Ethics Committee (ref: MEC/12/01/005).
Provenance and peer review Not commissioned; externally peer reviewed.
Patient consent for publication Obtained.