Abstract
Background:
Associations have been reported between the risk of Parkinson’s disease (PD) and employment in certain fields. Most studies have focused on toxic exposures as potential causal explanations. However, PD also has been associated with personality characteristics that may influence occupational choices and patterns.
Objective:
This study evaluates the role of personality as indicated by occupational choices and employment patterns in the risk for PD.
Methods:
In-person interviews were conducted to assess occupational histories and early-adult personality indicators among 89 PD patients and 99 controls.
Results:
PD cases had fewer lifetime jobs than controls (mean for cases = 4.38 ± 2.20; mean for controls = 5.00 ± 2.26; p = 0.03). Among women, PD was positively associated with more complex work with people (OR = 1.45, 95% CI 1.12–1.89), representing a 95% increased risk for PD comparing women with the greatest complexity of work with those requiring the least complexity of work with people. Women PD cases also performed less complex work with things compared with controls (OR = 0.69 (95% CI 0.53–0.90)), translating into a 13-fold increased risk for PD among women whose work involved the least complex work with things compared with the most. The numbers of jobs and job types were associated with taking more activity risks as a young-adult (r = 0.19, p = 0.02; r = 0.26, p = 0.001, respectively).
Conclusions:
Cases with PD held fewer lifetime jobs compared with controls. Occupational complexity was associated with the risk for PD among women, but not men. Further consideration of the possible influence of personality on occupational choices is warranted.
BACKGROUND
Parkinson’s disease (PD) is a neurodegenerative disorder that has been associated with employment in certain occupations. Health care workers, farmers, teachers, lawyers and scientists have been found to be at increased risk for PD [1–16], and individuals employed in the fields of construction, management and service have a reduced risk [11, 17]. Most occupational studies have focused on toxic exposures such as welding [18–21], pesticide exposure [4, 22–25], exposure to magnetic fields [26] and exposure to infectious substances [11] as potential mechanisms associated with the risk for PD.
It has been suggested that PD is associated with a rigid, cautious and introverted personality type [27]. Previous epidemiologic studies reported that PD cases are more cautious, introverted, compulsive, industrious, punctual, and quiet and have reduced novelty- and sensation-seeking behaviors compared with controls [28–30]. We have found that increased risk-taking as a young adult is associated with a reduced risk for PD (OR = 0.78 (95% CI 0.63–0.97)) and higher levels of routinization as a young adult are associated with an increased risk of PD among women (OR = 1.63 (95% CI 1.05–2.53)) [31]. PD also has been associated with reduced novelty-seeking behavior [30, 32].
Jobs involving greater complexity working with people have been associated with a reduced risk of Alzheimer’s disease [33]. However, occupational complexity has only been studied as a risk factor for PD in two other studies [34, 35]. Choice of main lifetime occupation with distinctive complexity characteristics could be influenced by personality dimensions. For example, given the Parkinsonian personality traits of introversion and compulsiveness [28–30], individuals with PD might be more likely to be employed in fields requiring a high level of data involvement such as accounting or engineering rather than those requiring interaction with people such as service or sales.
Given the description of the Parkinsonian premorbid personality as being rigid and cautious [27], we hypothesized that individuals with PD might be less likely to change jobs frequently or to alter the type of work they do during adulthood. The number and variety of jobs has not been studied with regard to the risk for PD. The aim of the present study was to assess the role of occupational characteristics and employment patterns as risk factors for PD and to examine the association of characteristics of occupations with aspects of personality that have been shown to be associated with PD.
METHODS
Subjects
Subjects were identified through review of electronic medical records at an academic-based medical center. Charts of patients who visited the Movement Disorders clinic at the University of South Florida (USF) between January 1, 2007 and May 1, 2010 were reviewed to identify potential cases. A computer-generated list of patients who visited the USF Family Medicine clinic was used to identify potential controls. All potential cases and a random sample of potential controls were contacted to recruit and screen for the study. All subjects were aged 50-80 years, Caucasian, free from memory impairment and able to read and speak English. Cases were further screened and deemed ineligible if they had a diagnosis of atypical PD, a history of neurosurgery for PD, or severe motor fluctuations (greater than 50% of the day “off” or with dyskinesia). In order to increase the sample size, additional cases were recruited from two outlying general neurology clinics. All subjects provided written informed consent according to the protocol approved by the Institutional Review Board.
Procedures
Trained interviewers completed private, in-person assessment of subjects at the study site (medical clinic) using highly structured questionnaires.
Exposures
Employment history
Subjects were asked to recall every paid job held for at least one year throughout their lifetime. Job title, industry, job duties and the year the job began and ended were recorded. For jobs in which subjects were currently or recently employed, we used a reference year approach for cases and controls. Each case was assigned his/her own reference year based on age at symptom onset – 1 year. For example, a 65-year old case whose symptoms began at age 60 was assigned a reference year corresponding to the year in which they were 59 years old. We then stratified age by 10-year intervals and calculated the mean reference age for cases within each stratum. Controls were assigned a reference age equal to the mean reference age of cases in each age stratum. These reference ages were used when calculating the number of jobs and number of job categories (described below).
Occupation variables
Occupation variables were defined based on subjects’ employment histories and DOT codes corresponding to each job they reported. Employment data were coded according to the Dictionary of Occupational Titles (DOT) [36] job classification system by the same coder who was blinded to case-control status (KLS).
The DOT provides unique 9-digit codes for occupations based on job title, industry and duties (Table 1). The first digit of each occupational code identifies the occupation as belonging to 1 of 9 categories (see Table 1). The second and third digits of the DOT code provide additional distinctions within each occupational category. The middle 3 digits of the DOT code define the complexity of the occupation with regard to people, data and things with lower codes (closer to 0) for each component indicating more complex work in that area (Table 1).
Number of jobs
The number of jobs reported by each subject up until the reference age was summed. Each job counted individually, even if it had the same title as a previously reported job. For example, a subject who reported employment as a kindergarten teacher 3 separate times or at 3 schools was considered to have worked in 3 jobs.
Number of job categories
We used changes in the first 3 digits of subjects’ DOT codes to indicate changes in job categories with each unique 3-digit code for this field indicating a unique category of job for that subject. The number of job categories in which a subject worked prior to their reference age was summed. For example, a subject who was initially employed as a kindergarten teacher (DOT code beginning with 092), later employed as a special education teacher (DOT code beginning with 094), and finally employed as a university professor (DOT code beginning with 090) was considered to have been employed in 3 job categories.
Duration of employment
Duration of employment at each job was calculated by subtracting the year a job started from the year the job ended, taking reference ages into account. For example, employment after the age of 59 years was not included for a participant with a reference age of 59 years, even if the participant was currently employed.
Duration of longest-held job
The duration of employment at the job with the longest duration was recorded as the duration of the longest-held job, taking reference ages into account, as with duration of employment, above.
Primary lifetime occupation
The job with the longest duration was considered each subject’s primary lifetime occupation.
Covariates
Personality
Personality characteristics related to risk-taking and preference for a routine lifestyle were indirectly assessed through the use of instruments that used subjects’ participation in routine and risk-taking activities as a young adult [29, 37], such as preferences for doing the same activities each day, gambling for small and large sums of money, swimming far from shore, riding a motorcycle or roller coaster, parachuting out of an airplane, parasailing, skiing, flying in a small plane, speeding when driving, flying in airplanes, getting lost in familiar and unfamiliar places, being in a high place, and wearing seatbelts. Latent variable values representing early-adult activity risks and early-adult routinization were calculated using results from factoranalyses [37].
Other covariates
The study interview also included subjects’ current age, sex, and number of years of formal education.
Data analysis
The crude association of demographic and occupational variables with case-control status was examined using Wilcoxon Rank-Sum tests for continuous variables and chi-square tests for discrete variables. The association of occupational characteristics with the risk for PD was analyzed using multiple logistic regression, controlling for age, sex and education. Since lower DOT codes indicate more complex work, the reciprocal of point estimates is reported to facilitate interpretation of results.
We also examined Pearson correlations between occupational variables and early-adult measures of risk-taking and routinization, partialling out the effects of age, sex and education. Additionally, sex was examined as a potential effect-modifier of theseassociations.
P-values of less than 0.05 (2-sided test) were interpreted as being statistically significant. All data were analyzed using SAS version 9.2 [38].
RESULTS
Eighty-nine cases (56% of 125 who were eligible at the main study site plus 19 from outlying sites) and 99 controls (48% of 207 who were eligible) completed study assessments. Full participation data including reasons for refusal have been detailed previously [37]. For both cases and controls, there were no significant differences in age or sex between those who participated and those who refused. Cases and controls did not differ by mean age (Table 2). However, cases were more likely to be men (p = 0.005) and to have fewer years of formal education compared with controls (p = 0.003).Educational differences were particularly pronounced among women.
Occupational patterns
All subjects reported employment in at least 1 job. Cases held an average of 4.38 ± 2.20 jobs compared with 5.00 ± 2.26 jobs for controls (p = 0.03) and worked in an average of 3.40 ± 1.74 types of jobs compared with 3.63 ± 1.92 types of jobs among controls (p = 0.47). Among men, there were no significant differences between cases and controls in the mean number of jobs, number of job categories, or duration of primary occupation, while women with PD worked in fewer jobs (mean = 3.81 ± 1.70) compared with women without PD (mean = 4.85 ± 2.43) (p = 0.04) (Table 2).
Occupational categories
In our data, cases’ and controls’ primary occupation was most often in the category of “professional, technical and managerial” (cases 60% , controls 63% ). This occupational category is quite broad and comprises occupations in architecture, engineering, surveying, mathematics, sciences, medicine and health, education, law, religion, art, entertainment, administration (accounting, human resources, purchasing) and management. Within this category, the most common field for lifetime occupation of both cases and controls was education (DOT codes beginning with “09”), with 11.24% of cases and 11.11% of controls reporting main occupations in this category. There was no association between the first DOT digit and case-control status.
Occupational complexity
None of the occupational complexity characteristics showed a statistically significant difference between cases and controls in the total sample.
Occupation and Risk of PD
In logistic regression models adjusted for age, sex, and education, PD was associated with the number of jobs held (OR = 0.87 (95% CI 0.75–1.00) p = 0.048), but not with the number of job categories (OR = 0.88 (95% CI 0.74–1.05)) or the duration of the longest held job (OR = 1.00 (95% CI 0.97–1.04)) in the total sample (Table 3). PD was also not associated with complexity of work with data (OR = 1.01 (95% CI 0.81–1.25)) or things (OR = 0.98 (95% CI 0.86–1.11)). However, a borderline significant association was seen between higher complexity of work with people and greater risk of PD (OR = 1.15 (95% CI 1.00–1.33), p = 0.053). Among women, PD was associated with greater complexity of work with people (OR = 1.45 (95% CI 1.12–1.89)) and less complex work with things (OR = 0.69 (95% CI 0.53–0.90)).
Occupation and personality
Taking more activity risks in early adulthood was positively associated with the number of jobs (r = 0.19, p = 0.02) and with the number of types of jobs (r = 0.26, p = 0.001), partialling out the effects of age, sex and education (Table 4). Routinization was not associated with either the number of jobs or the number of types of jobs. Complexity of work with people, data and things was not associated with early-adult activity risks or routinization.
DISCUSSION
In this case-control study, we found associations among women between higher risk of PD and greater complexity of work with people as well as less complexity of work with things. Women whose jobs required the greatest possible complexity of work with people were approximately 95% more likely to have PD compared with those whose work required the least possible complexity of work with people. In addition, women with the least complex work with things were over thirteen times more likely to have PD compared with women with the most complex work with things. There was a marginally significant association between risk for PD and higher complexity of work with people in the entire sample as well as an association between PD and the number of jobs held over the lifetime. These findings provide additional support for the hypothesis that behaviors and personality characteristics associated with the PD personality may exist many years prior to the presentation of motor symptoms of the disease.
While the association of PD and greater complexity of work with people might seem to contradict previous descriptions of the Parkinsonian personality as introverted [28], the DOT coding for this characteristic does not reflect the frequency of work with people. Individuals who are unlikely to take risks and who enjoy predictable routines would seem to be best suited for jobs involving supervision, instruction and mentoring, which all have DOT codes representing the greatest amounts of complexity of work with people (Table 1). Jobs with less complexity of work with people such as serving and helping are jobs that might offer less dependability and require greater risks, thereby having less appeal to people with personality characteristics such as those present in PD.
Effect-modification by sex was evident with associations between PD and complexity of work with people and things, which were present among women but not men. This finding differs from results from a study of Swedish twins which reported an increased risk of PD among men with higher occupational complexity of work with data (hazard ratio (HR) = 1.08, 95% CI 1.01–1.16) and work with people (HR = 1.15, 95% CI 1.03–1.28) (35). However, these sex-specific associations were attenuated when analyses were stratified by twin pair, suggesting a partial explanation by familial factors (35). Given the prevailing societal expectations and gender roles that were more strongly held in past decades, it is likely that women in our study had fewer occupational choices that were easily obtainable. It is possible that women who had more assertive or determined personalities would have been more likely to work in a broader range of occupations compared with women who were content with more traditional occupational choices. Therefore, personality might have differentially influenced occupational choices more in women than in men in our study.
Another recent study examined personality aspects of occupation as risk factors for PD by evaluating the demands, skills and aptitudes required by the participants’ longest held job using US Census Occupational Codes [34]. Occupations that required more adaptability, ability to make generalizations or preference for abstract activities were marginally inversely associated with the risk for PD. Although the DOT coding system we used does not provide insight into these aspects of occupation, the finding of decreased flexibility among cases with PD is consistent with our previous report of an inverse association between PD and a preference for routine during early-adulthood [37].
Other studies have used DOT codes to evaluate the association of occupation with PD. A nested case-control study evaluated occupational histories of 144 cases with PD and 464 control subjects [17]. Subjects who reported ever working in the “service” category had a lower risk for PD (OR = 0.56, p = 0.01 adjusted for age, sex and race). Another population-based case-control study examined occupations and the risk for PD in 404 incident cases and 526 controls in Washington State [39]. There was no association between PD and any DOT category among men or women, adjusting for age, ethnicity and smoking. Finally, the largest study that used DOT coding in examining risk factors for PD was a multi-center case-control study conducted in Scotland, Sweden, Italy and Romania [40] that evaluated 649 cases with PD and 1,587 controls. A history of ever working in “processing occupations” was associated with a reduced risk of PD (OR = 0.69 (95% CI 0.50–0.95) adjusting for age, sex, ever-use of tobacco and family history of PD). Similarly, there was no association between risk for PD and a primary occupation in any DOT category in our study, either crudely (p = 0.38) or adjusted for age, sex and history of ever smoking (p = 0.85). This difference could be due to the high educational level obtained by controls relative to cases or to the urban setting of the center from which subjects were recruited. Also, few subjects reported working in certain occupational categories, which limited the statistical power to obtain significant findings if they existed.
A strength of our study is that all assessments were conducted via in-person interviews, which helped ensure completeness of data collection and reduce the frequency of invalid responses and missing data. The personality assessments used in this study were unique in that they used activities and events as historical indicators of personality traits and full occupational histories. Our findings that higher complexity of work with people is associated with increased PD risk among women warrants further study.
Several limitations were present in this study due to sample selection. Since our study was based in an urban academic medical center, referral bias may be present and personality characteristics could result in certain individuals requesting referrals to our center. This potential bias would apply equally to cases and controls. There was also a high percentage of educators in our sample and it is possible that individuals with teaching backgrounds were more likely to be patients or to participate in research studies. Such restriction may have resulted in increased homogeneity of the sample that would drive the measures of association toward the null. Another limitation is the possibility of imprecision in DOT coding. Although the DOT categories provide a structured system for classifying and grouping similar occupations, this aggregation may obscure associations between specific occupations and PD. Such misclassification would have occurred independently of case-control status, biasing the OR toward the null. We considered only the subjects’ primary occupation for all analyses rather than their complete job histories. It is possible that exposures occurred during employment of shorter duration and were not captured by this analysis. Additionally, since our aims included several exploratory analyses, we did not adjust for multiple comparisons, and results should be interpreted with this in mind. The generalizability of our results is limited by the restriction of enrollment to Caucasians. Differences in personality among races has been reported (41, 42) and PD is more prevalent among Caucasians than in Asians and Blacks (43, 44). Finally, it is possible that the primary lifetime occupation may have been influenced by very early disease symptoms. However, given the eligibility criteria requiring cases to have been diagnosed with PD no more than 10 years prior to study entry, it is expected that their choice of primary occupation occurred well before the onset of PD symptoms (as indicated by duration of primary occupation of approximately 20 years).
CONFLICTS OF INTEREST
The authors declare that they have no conflicts of interest to report.