High occurrence of transportation and logistics occupations among vascular dementia patients: an observational study

Background Growing evidence suggests a role of occupation in the emergence and manifestation of dementia. Occupations are often defined by complexity level, although working environments and activities differ in several other important ways. We aimed to capture the multi-faceted nature of occupation through its measurement as a qualitative (instead of a quantitative) variable and explored its relationship with different types of dementia. Methods We collected occupational information of 2121 dementia patients with various suspected etiologies from the Amsterdam Dementia Cohort (age 67 ± 8, 57% male; MMSE 21 ± 5). Our final sample included individuals with Alzheimer’s disease (AD) dementia (n = 1467), frontotemporal dementia (n = 281), vascular dementia (n = 98), Lewy body disease (n = 174), and progressive supranuclear palsy/corticobasal degeneration (n = 101). Within the AD group, we used neuropsychological data to further characterize patients by clinical phenotypes. All participants were categorized into 1 of 11 occupational classes, across which we evaluated the distribution of dementia (sub)types with χ2 analyses. We gained further insight into occupation-dementia relationships through post hoc logistic regressions that included various demographic and health characteristics as explanatory variables. Results There were significant differences in the distribution of dementia types across occupation groups (χ2 = 85.87, p < .001). Vascular dementia was relatively common in the Transportation/Logistics sector, and higher vascular risk factors partly explained this relationship. AD occurred less in Transportation/Logistics and more in Health Care/Welfare occupations, which related to a higher/lower percentage of males. We found no relationships between occupational classes and clinical phenotypes of AD (χ2 = 53.65, n.s.). Conclusions Relationships between occupation and dementia seem to exist beyond the complexity level, which offers new opportunities for disease prevention and improvement of occupational health policy.


4
To overcome this scarcity in the literature, we took a different approach by i) measuring 0 4 [42,43] and ran these models in all participants with complete data (n=1,562/2,121). Finally, we carried out Chi 2 analyses to investigate whether occupation related to cognitive 2 0 7 profiles within the AD subsample, following the same approach as described above. In a 2 0 8 second analysis, we reran the same model using a different threshold for the dichotomization

1 4
Participants 2 1 5 The mean age across diagnostic groups was 67±8 years (Table 1). Individuals with FTD 2 1 6 All rights reserved. No reuse allowed without permission. certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not this version posted January 8, 2020. .

0
were younger compared to other groups (63±7), and DLB participants were significantly older 2 1 7 than the AD group (69±7). Overall, our sample included a somewhat larger proportion of 2 1 8 males (57%), presumably reflecting the historically higher prevalence of males in the labor 2 1 9 market (by comparison, the excluded "no occupation" sample was 82% female). The 2 2 0 proportion of males was even higher among DLB participants (82%) in comparison with most 2 2 1 other groups, while this was significantly lower for AD (52%) in contrast to all groups except 2 2 2 PSP/CBD. Educational level was largely similar across dementia types in our sample (except 2 2 3 for a DLB>VaD difference). Global cognitive impairment was most severe among AD 2 2 4 participants (mean MMSE=20±5). Finally, VaD participants had a higher VRF score than 2 2 5 most other diagnostic groups (p<.05; except FTD). Table 2 provides an overview of the 2 2 6 number of participants in each occupational class for the total sample and according to 2 2 7 dementia type. The largest occupation groups were Technical (n=429, 20%) and 2 2 8 Business/Administrative (n=388, 18%), while the Agricultural class had the lowest number of 2 2 9 participants (n=31, 1%).  Relationship between occupation and dementia types 2 3 4 There were significant differences in the distribution of dementia types across occupation 2 3 5 groups (Chi 2 =85.87, p<.001, Table 2). The adjusted residuals revealed three effects 2 3 6 significant at p<.001. First, individuals from the Transportation/Logistics sector were more 2 3 7 often diagnosed with VaD (adjusted residual: 4.0). Second, this occupational class had fewer 2 3 8 AD participants (adjusted residual: -3.7). Third, in the Health Care/Welfare group, AD was 2 3 9 relatively common (adjusted residual: 3.6). These results are displayed in Figure 2 Figure 3 gives an overview of the number of VRFs and percentage of males for all dementia 2 6 2 types and occupational classes. Relationship between occupation and cognitive profiles of AD 2 6 5 We found only trend-significant differences between occupations in the distribution of 2 6 6 cognitive profiles (Chi 2 =53.65, p=.07; Table 3). Supplementary Figure 2 provides an overview 2 6 7 of these findings. In a secondary analysis, we repeated the same analysis using cognitive 1). Similar to the original findings, these results were not significant (Chi 2 =38.24, p=.55).

7 0
All rights reserved. No reuse allowed without permission. certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not this version posted January 8, 2020. . Our study has several limitations. First, although we collected occupational data with a 3 2 8 refined method that involved a semi-structured interview with the patient and caregiver, we 3 2 9 did not have detailed enough information to capture a person's entire occupational timeline.
For individuals who had multiple occupations listed, we often could not retrieve their 3 3 1 sequence and respective durations. It is therefore possible that the first-mentioned job not 3 3 2 always reflect the person's foremost/primary occupation, but rather the most recent 3 3 3 occupation. This is non-optimal because of potential reverse causation: we cannot rule out 3 3 4 that some patients switched jobs as a consequence of their dementia diagnosis (e.g. that reverse causation is more prominent for some dementia (sub)types than others, and 3 3 7 thus the degree of bias seems limited. Moreover, our results did not change after exclusion individuals with dementia as a whole. As not every person with dementia is referred to a 3 4 9 memory clinic or seeks medical care in an academic expertise center such as the Alzheimer 3 5 0 Center Amsterdam, the generalizability of our results is somewhat limited. Fifth, although 3 5 1 All rights reserved. No reuse allowed without permission. certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not this version posted January 8, 2020. . neuroimaging biomarkers, we did not have neuropathological confirmation of the underlying 3 5 3 etiologies. It is therefore possible that some participants were categorized under a dementia 3 5 4 type that would not completely correspond with their neuropathological diagnosis. Finally, 3 5 5 although our total sample was large (N=2,121), some occupational classes (e.g. Agricultural) 3 5 6 contained a low number of observations. It is possible that we have failed to find certain 3 5 7 relationships with dementia types due to limited power in these classes. To conclude, we observed relationships between occupation and dementia types. Our 3 6 1 findings suggest that these relationships emerged beyond occupational complexity level.

6 2
Differences between occupational classes in the number of VRFs and sex distributions 3 6 3 partially explained our results, but other -currently unknown -factors likely play a role as 3 6 4 well. It is important to note that associations between occupation and dementia could exist 3 6 5 for multiple reasons. Whilst it seems plausible that some occupational characteristics 3 6 6 causally relate to the occurrence of specific dementia types, another possibility is that certain of job hazards and more targeted health monitoring by occupational physicians. Ethics approval and consent to participate 3 8 6 Participants gave written informed consent to use their medical data for scientific purposes.

8 7
This procedure was approved by the local Medical Ethics Committee. Availability of data and material 3 9 3 The datasets used and/or analyzed during the current study are available from the 3 9 4 corresponding author on reasonable request. Competing interests 3 9 7 The authors declare that they have no competing interests. Onderzoek (ISAO) (to R.O.). 4 0 5 All rights reserved. No reuse allowed without permission. certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not this version posted January 8, 2020. . Not applicable 4 2 0 All rights reserved. No reuse allowed without permission. certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not this version posted January 8, 2020. . certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not this version posted January 8, 2020. certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not this version posted January 8, 2020.  11 cells (20.0%) had an expected count less than 5; the minimum expected count was 1.53. Each column represents a cognitive profile in which one particular domain is predominantly affected (except the multi-domain). There were no significant differences between groups (Chi 2 n.s.). All rights reserved. No reuse allowed without permission. certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

7
Supplementary Table 1 For each dementia type, the clinical criteria used are provided in chronological order. AD=Alzheimer's disease dementia, FTD=frontotemporal dementia, VaD=vascular dementia, DLB=Lewy Body disease, PSP=progressive supranuclear palsy, CBD=corticobasal degeneration. All rights reserved. No reuse allowed without permission. certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not this version posted January 8, 2020. .   Within the AD subsample (n=1,071), we dichotomized each cognitive domain score (i.e. memory, attention/executive functions, language and visuospatial functions), based on whether or not a domain's W-score was considerably lower compared to an individual's global cognition score. We used an optimal threshold between .250 and .500 for this dichotomization, and defined "optimality" in two different ways: 1) the lowest number of All rights reserved. No reuse allowed without permission.
certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not this version posted January 8, 2020. . https://doi.org/10.1101/19005512 doi: medRxiv preprint 3 2 participants in the multi-domain profile and the lowest sum of squared group sizes (threshold=.255), 2) the highest number of atypical variant AD cases (n=85) categorized into the language (i.e. for logopenic aphasia) or visuospatial cognitive profile (i.e. for PCA; n=.442). Note that neither thresholds (i.e. .255 or .442) resulted in all atypical variants being categorized into these two cognitive profiles, presumably because in advanced disease stages, several cognitive domains become affected. For many participants with an initial logopenic aphasia or PCA diagnosis, visuospatial/language may no longer have been predominant at study inclusion, causing them to be assigned to the multi-domain instead.

Supplementary Figure 2. Proportions of AD-related cognitive profiles for each occupational class.
There were no significant differences between groups (Chi 2 n.s.).

Supplementary Figure 3. Schematic representation of protective and harmful effects and resulting cells in the contingency table.
Each color (red, blue, green) represents a different occupational class. In scenario A, the red occupational class is relatively protected against dementia type 1, while the distribution of the remaining dementia types across occupations is equal. When the healthy control group is included in the analysis, this protective effect will be reflected in a higher proportion of individuals in red in the healthy group (4/6=67% versus 50% for blue/green) and a lower percentage of red persons in the type 1 dementia group (0% versus 1/6=17% for blue/green).
All other proportions will be identical across occupational classes. However, when the healthy control group is not taken into account, the protective effect of the red occupation for dementia type 1 will create also an apparent "harmful" effect of the red occupation for dementia types 2 and 3 (i.e. 1/2=50%, compared to 1/3=33% for blue/green). In scenario B, the green occupational class shows a harmful effect on the development of dementia type 3, while no other differences between groups exist. Again, when the healthy control group is All rights reserved. No reuse allowed without permission. certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not this version posted January 8, 2020. .

3
included, this only results in a lower percentage of healthy green individuals (1/6=17% versus 50% for red/blue) and a greater proportion of persons with green occupations in dementia type 3 (i.e. 50% versus 1/6=17% for red/blue). In the absence of a healthy control group, however, the green occupational class' harmful effect for dementia type 3 leads to an apparent "protective" effect for dementia types 1 and 2 (1/5=20%, compared to 1/3=33% for red/blue). All rights reserved. No reuse allowed without permission. certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not this version posted January 8, 2020.