Association between occupational complexity and cognitive function in older adults from Brazil and Mexico

Abstract

More complex occupations during adulthood may be associated with better cognition in later life. We used data from the Brazilian Longitudinal Study of Aging (ELSI) and the Mexican Health and Aging Study (MHAS) to investigate the association between occupational complexity and cognitive function. Using a regression-based approach, participants were classified as cognitively impaired or not. Occupation was categorized into four levels using the 2008 International Standard of Classification of Occupations. We used logistic regression models for separate and pooled analysis. The odds of cognitive impairment decreased with higher occupation skill levels in the ELSI, but this trend was not seen in the MHAS, where the highest skill level was not associated with cognitive impairment. ELSI participants had a lower probability of cognitive impairment compared to MHAS participants with the same occupation skill level. The results of this study suggest that programs addressing the negative consequences of lower-complexity occupations need to be tailored to specific regional contexts, considering their unique socioeconomic, cultural, and occupational landscapes.

Introduction

It is estimated that, in 2019, almost 60 million people were living with dementia worldwide, 58% of whom were from low-and middle-income countries (LMIC).^1,2 Latin America (LA) has one of the highest dementia burdens globally, with a prevalence of 8.5% among people 60 years or older.³ Currently, there is no effective treatment for dementia, highlighting the need for preventive measures.⁴ It has been estimated that 40% of dementia cases worldwide could be prevented with the control of several modifiable risk factors, spanning early to late life.⁵

Employment is a protective factor of critical importance for cognitive function. Occupational environments provide opportunities for developing complex tasks, and social engagement and also determine income and economic resources, all of which have been associated with a lower risk of dementia.^6,7 In addition, occupational activity often occurs in midlife, when neuroplasticity is more intense relative to later life.⁸ Cognitive reserve refers to the flexibility of the cognitive processes to resist age-related disease or injury, and occupational complexity is one of its main proxies in midlife.⁹ The association between higher occupational complexity during middle age and later-life cognitive function has shown mixed results. Some studies have reported that higher occupational complexity was associated with better cognitive performance and a lower risk of dementia in later life, while other studies reported no association between occupational complexity and cognitive function.^10-14

Research on occupational complexity and cognitive function has focused on high-income countries, where there is a large proportion of individuals with complex occupations, such as managers, engineers, and medical doctors.¹⁵ Thus, it is important to investigate this association in the context of low-and middle-income countries where there is a greater variance in occupation complexity and a higher proportion of lower-complexity occupations due to educational and other socioeconomic limitations.¹⁵ Brazil and Mexico are the largest countries in LA, summing up half the population in this region, and these countries have two of the highest prevalences of dementia worldwide.^16,17 In addition, both countries have similar levels of economic development, and similar societal challenges, including a high prevalence of illiteracy, low quality of education, and historical gender inequality and discrimination.^18,19 Therefore, we used nationally representative data from both countries to assess the association between occupation complexity and cognitive function and investigate if gender and country modify this association.

Methods

Study samples

We used data from the Brazilian Longitudinal Study of Aging (ELSI) and the Mexican Health and Aging Study (MHAS), which are partner studies of the Health and Retirement Study (HRS). The HRS sister studies were designed similarly to facilitate comparisons across different countries.²⁰ ELSI is an ongoing longitudinal survey of Brazilian adults aged 50 years or older, and the baseline survey was conducted in 2015-16.²¹ One follow-up wave was conducted in 2019-2021. MHAS is an ongoing longitudinal survey of Mexican adults aged 50 years or older, and Wave 1 was collected in 2001.²² Five follow-up waves were conducted in 2003, 2012, 2015, 2018, and 2021. Data from 2018-2019 (Wave 2 in the ELSI and Wave 5 in the MHAS) were used in this study. In both studies, an in-person interview was conducted, and participants answered questions about demographic characteristics, lifestyle habits, socioeconomic factors, health-related problems, and cognitive function. Both studies conducted direct interviews where participants responded to the questions, and proxy interviews where an informant responded if the participant could not be interviewed due to health or other reasons. Participants gave written informed consent. The analytical sample included adults aged 50 years or older, who participated in the proxy or direct interviews. This study was approved by the University of São Paulo Medical School Ethics Committee (CAAE: 54344721.5.0000.0068). This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Occupational complexity

The main lifetime occupation was categorized according to the Brazilian Classification of Occupations (CBO, Classificação Brasileira de Ocupações) in the ELSI and according to the Mexican Occupation Classification (SINCO, Sistema Nacional de Clasificación de Ocupaciones) in the MHAS.^23,24 Data on occupation for both datasets were harmonized using the International Standard of Classification of Occupations 2008 (ISCO-08).²⁵ The ISCO-08 is a hierarchical classification that groups 436 detailed jobs into 10 major groups that were then grouped into four levels based on similarities of skill levels and skill specializations. Skill level 1 includes occupations with simple and routine physical or manual tasks, such as office cleaners, freight handlers, garden laborers, and kitchen assistants. Skill level 2 includes occupations that involve operating machinery and electronic equipment; driving vehicles; maintenance and repair of electrical and mechanical equipment; and manipulation, ordering, and storage of information, such as butchers, bus drivers, secretaries, shop sales assistants, police officers, hairdressers, and motor vehicle mechanics. Skill level 3 includes occupations that involve the performance of complex technical and practical tasks that require factual, technical, and procedural knowledge in a specialized field, such as shop managers, medical laboratory technicians, legal secretaries, commercial sales representatives, diagnostic medical radiographers, computer support technician, and broadcasting and recording technicians. Skill level 4 includes occupations that involve the performance of tasks that require complex problem-solving, decision-making, and creativity based on an extensive body of theoretical and factual knowledge in a specialized field, such as sales and marketing managers, civil engineers, secondary school teachers, medical practitioners, musicians, nurses, and computer systems analysts.

Participants who reported having never worked for pay were classified as “never worked”. Participants who reported being in the military (n = 33 in the ELSI) were excluded from the analysis since this category included many skill levels and would need finer detail to be included in the ISCO-08 classification. Participants who answered “don’t know” or “other” occupations were also excluded from the analysis (n = 2 in the ELSI and n = 27 in the MHAS). The correspondence between the ELSI and MHAS categories with the ISCO-08 skill level categories can be found in Supplemental Tables 1 and 2.

Cognitive function

The cognitive function of those with direct interviews in both datasets was assessed with different tasks adapted from the cross-cultural cognitive examination (CCCE). The cognitive assessment in the ELSI included episodic memory (10-word list immediate and delayed recall), temporal orientation (day, month, year, and day of the week), semantic verbal fluency (animal naming), prospective memory (remembering to write one’s initials after finishing another test), and semantic memory questions (two questions on naming common objects and two questions about naming the current president and vice-president). The cognitive assessment in the MHAS included episodic memory (8-word list immediate and delayed recall), visual scanning, immediate and delayed constructional praxis, temporal orientation (day, month, year), semantic verbal fluency (animal naming), counting backward from 20 to 11, and serial 7 subtractions. Missing cognitive test data were imputed using multiple imputations by chained equations (MICE).

The detailed descriptions of the tests used in both studies can be found in the Supplementary Material. While the CCCE cognitive battery was translated and culturally adapted, some tests that evaluated the same cognitive domain in each study were applied differently. Additionally, translations of word recall lists may introduce heterogeneity.²⁶ Given the differences in the cognitive test batteries in the ELSI and MHAS, we opted to categorize participants in each study as cognitively impaired or not based on a normative sample specific to each study.

The presence of cognitive impairment was evaluated as described previously.²⁷ Briefly, for the direct interviews, we selected a normative sample with similar characteristics in both studies: participants without (1) a history of stroke; (2) complaints about memory or instrumental activities of daily living; (3) depressive symptoms according to the Center for Epidemiological Scale-Depression (CESD-9, cut-off of 5 points for MHAS, and CESD-8, cut-off of 4 points for ELSI); (4) regression-based classification of dementia in 2015 for MHAS, or history of Alzheimer’s disease for ELSI; (5) missing cognitive, age, education, or gender data. We then performed a regression on the global cognitive z-score of the normative sample adjusted for age, gender, and education, and used the coefficients of the resulting regression to calculate the predicted global z-scores for the whole sample. Then, we subtracted the predicted z-scores from the participants’ actual global cognitive z-scores and divided this result by the root mean square error from the normative sample regression. We categorized participants as cognitively impaired if the resulting z-score was −1.5 or lower and as not impaired if the resulting z-score was higher than −1.5.²⁸ For the proxy interview, cognitive impairment was defined as an Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) score above or equal to the cut-off of 3.4, as previously described.²⁹ We combined the results of the direct and proxy interviews to classify participants into the cognitively impaired or not categories.

Covariates

Covariates included age as a continuous variable, gender, years of education as a continuous variable, body mass index (BMI) calculated as weight (measured with a Portable digital scale of the brand SECA in the ELSI and self-reported in the MHAS) in kilograms divided by height (measured with a portable vertical stadiometer of the brand Nutri-Vida in the ELSI and self-reported in the MHAS) in meters squared, self-report of a previous diagnosis of hypertension, myocardial infarction, diabetes, smoking (never, former, current), and self-reported current alcohol consumption (yes or no). Missing covariate values were imputed using MICE.

Statistical analyses

Descriptive analyses accounted for the sampling design of the ELSI and MHAS studies and were weighted using the survey weights provided by each study, plus stratum and cluster for ELSI to calculate descriptive statistics that represent the population of Brazil and Mexico. Descriptive analyses were presented as mean and standard deviation (SD) or median and interquartile range (IQR) for continuous variables and percentages for categorical variables. We used logistic regression to evaluate the association between occupational complexity and cognitive impairment separately for each cohort, using the category “never worked” as the reference group. Model 1 was adjusted for age, gender, and education. Model 2 was additionally adjusted for hypertension, cardiovascular disease, diabetes, smoking, alcohol consumption, and BMI. We investigated if gender modified the association between occupational complexity and cognitive impairment by adding an interaction term between occupational complexity and gender in Model 2. We also investigated if country modified the association between occupational complexity and cognitive impairment by performing a pooled analysis, creating a dummy variable for country, and adding an interaction term between occupational complexity and country in Model 2.

We performed three sets of sensitivity analyses. First, we categorized the main lifetime occupation of the ELSI and the MHAS into blue-collar, pink-collar, and white-collar and repeated the regression models to evaluate the association between occupational complexity and cognitive impairment separately for each country.¹⁰ Second, we used only the common cognitive tests in the direct interview for both studies (temporal orientation, immediate and delayed recall, and verbal fluency) to categorize participants as cognitively impaired or not and repeated the logistic regression to evaluate the association between occupational complexity and cognitive impairment separately for each country. Finally, we did an exploratory analysis of the association between occupation skill level and performance in the different cognitive domains. For this, we performed a linear regression using the continuous z-scores from the common tests as the outcome of the regression. We also calculated a global cognitive z-score by calculating the mean of the common tests’ z-scores and standardizing this mean. The alpha level was set at 5% in two-sided tests. All statistical analyses were performed using R version 4.1.2.

Role of the funding source

The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Results

Sample characteristics

Data from 9,914 participants from the ELSI and 16,172 participants from the MHAS were analyzed. The mean (standard deviation, SD) age for the ELSI participants was 64.0 (9.7) years, 50.2% were women, and the median (interquartile range [IQR]) years of education was 4 (2-10) (Table 1). The mean (SD) age of the MHAS participants was 62.8 (10.3) years, 56.4% were women, and the median (IQR) years of education was 6 (3-9) (Table 1). In the ELSI, 15.8% of participants had never worked, 11.3% had main occupations in the ISCO-08 level 1, 62.3% in level 2, and 10.6% in levels 3/4 (Table 1). In the MHAS, 20.0% of participants had never worked, 29.8% had main occupations in the ISCO-08 level 1, 40.2% in level 2, and 10.0% in levels 3/4 (Table 1). In the ELSI, 6% of participants had cognitive impairment and this proportion ranged from 3.4% in participants with occupations in ISCO-08 levels 3/4 to 12.1% in participants who had never worked (Figure 1). In the MHAS, 7.4% of participants had cognitive impairment and this proportion ranged from 5.3% in participants with occupations in ISCO-08 level 1 to 15.6% in participants who had never worked (Figure 1). ELSI and MHAS participants with occupations in skill levels 3/4 were more likely to be younger, men, have more years of education, have higher annual income, consume alcohol, and be current smokers compared to participants who had never worked (Supplemental Table 3). In addition, MHAS participants with occupations in skill levels 3/4 were less likely to have hypertension and diabetes compared to participants who had never worked (Supplemental Table 3).

Table 1.

Sample characteristics of the Brazilian Longitudinal Study of Aging (ELSI, 2019) and the Mexican Health and Aging Study (MHAS, 2018).

	ELSI (n = 9,914)	MHAS (n = 16,172)
Age, mean (SD)	64.0 (9.7)	62.8 (10.3)
Women, %	50.2	56.4
Education, median (IQR)	4 (2-10)	6 (3-9)
Annual family income*, median (IQR)	6,574.80 (4,274.40-11,506.80)	1,761.04 (296.10-5,116.88)
Occupational complexity‡, %
Never worked	15.8	20.0
Level 1	11.3	29.8
Level 2	62.3	40.2
Level 3/4	10.6	10.0
Body mass index, mean (SD)	27.8 (5.3)	34.1 (15.2)
Hypertension, %	49.1	39.8
Diabetes, %	17.8	22.7
Myocardial infarction, %	3.9	3.1
Alcohol consumption, %	14.3	28.1
Smoking, %
Never	65.4	77.6
Former	20.7	4.6
Current	13.9	17.8
Cognitively impaired, %	6.0	7.4

ELSI, Brazilian Longitudinal Study of Aging; MHAS, Mexican Health and Aging Study

^*in US dollars. Conversion rates: 1.00 USD = 3.65 BRL = 19.25 MXN

^‡Classified according to the International Standard Classification of Occupations 2008 (ISCO-08)

Figure 1.

Percentage of cognitive impairment by study according to the occupational complexity in the Brazilian Longitudinal Study of Aging (ELSI) and the Mexican Health and Aging Study (MHAS). Occupational complexity was classified according to the International Standard Classification of Occupations 2008 (ISCO-08).

Association between occupational complexity and cognitive function

ELSI participants who had occupations in ISCO-08 levels 1, 2, and 3/4 had lower odds of cognitive impairment compared to participants who had never worked (level 1: odds ratio [OR]=0.72, 95% confidence interval [CI]=0.56 to 0.94, p=0.016; level 2: OR=0.51, 95% CI=0.42 to 0.61, p<0.001; level 3/4: OR=0.41, 95% CI=0.27 to 0.61, p<0.001) (Table 2). The odds of cognitive impairment also decreased with each level of higher occupation complexity (p<0.001) (Figure 2). MHAS participants who had main occupations in ISCO-08 levels 1 and 2 had lower odds of cognitive impairment compared to participants who had never worked (level 1: OR=0.63, 95% CI=0.52 to 0.76, p<0.001; level 2: OR=0.61, 95%CI=0.52 to 0.72, p<0.001) (Figure 2 and Table 2). There was no association between participants who had main occupations with skill levels 3/4 and cognitive impairment in the MHAS (OR=0.90, 95% CI=0.67 to 1.19, p=0.454) (Figure 2 and Table 2). Similar results were obtained using the blue collar, pink collar, and white collar occupation classification (Supplemental Table 4), and using only the common cognitive tests in both studies for the regression-based categorization (Supplemental Table 5). Moreover, higher occupational complexity was associated with better performance in memory, verbal fluency, temporal orientation, and global cognition in both countries, with the exception of a negative relationship in episodic memory for those with level 1 occupation in the ELSI and in temporal orientation for those with level 3/4 occupation in the MHAS (Supplemental Table 6). Gender did not modify the association between occupation and cognitive impairment in either study (ELSI: p=0.956, MHAS: p=0.535). Pooled analysis showed that there was a significant difference in the association between occupation and cognitive impairment by country (p=0.001) (Figure 3). ELSI participants who had occupations in levels 2 and 3/4 had a lower probability of cognitive impairment compared to MHAS participants with the same occupation skill level (Figure 3). There were no country differences in the probability of cognitive impairment for those who had never worked or with level 1 occupations (Figure 3).

Table 2.

Association between occupational complexity^* and cognitive impairment.

ELSI
	Unadjusted		Model 1		Model 2		Model 3
Occupational complexity	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p
Never worked (n = 1,786)	1 (Reference)		1 (Reference)		1 (Reference)		1 (Reference)
Level 1 (n = 1,168)	0.60 (0.46; 0.77)	<0.001	0.67 (0.52; 0.86)	0.002	0.71 (0.55; 0.92)	0.010	0.72 (0.56; 0.94)	0.016
Level 2 (n = 5,952)	0.42 (0.36; 0.51)	<0.001	0.47 (0.39; 0.56)	<0.001	0.49 (0.41; 0.59)	<0.001	0.51 (0.42; 0.61)	<0.001
Level 3/4 (n = 1,008)	0.25 (0.17; 0.35)	<0.001	0.42 (0.28; 0.62)	<0.001	0.40 (0.26; 0.58)	<0.001	0.41 (0.27; 0.61)	<0.001
MHAS
	Unadjusted		Model 1		Model 2		Model 3
Occupational complexity	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p
Never worked (n = 3,698)	1 (Reference)		1 (Reference)		1 (Reference)		1 (Reference)
Level 1 (n = 4,459)	0.51 (0.44; 0.59)	<0.001	0.36 (0.21; 0.59)	<0.001	0.63 (0.52; 0.76)	<0.001	0.63 (0.52; 0.76)	<0.001
Level 2 (n = 6,234)	0.40 (0.34; 0.46)	<0.001	0.53 (0.47; 0.60)	<0.001	0.61 (0.52; 0.72)	<0.001	0.61 (0.52; 0.72)	<0.001
Level 3/4 (n = 1,781)	0.35 (0.28; 0.44)	<0.001	0.81 (0.61; 1.07)	0.156	0.90 (0.67; 1.19)	0.456	0.90 (0.67; 1.19)	0.454

ELSI, Brazilian Longitudinal Study of Aging; MHAS, Mexican Health and Aging Study

^*Classified according to the International Standard Classification of Occupations 2008 (ISCO-08)

Model 1: Logistic regression adjusted for education.

Model 2: Logistic regression adjusted for age, sex, education, and income.

Model 3: Logistic regression adjusted for age, sex, education, income, alcohol consumption, smoking, body mass index, diabetes, hypertension, and myocardial infarction.

Figure 2.

Probability of cognitive impairment according to occupational complexity in the (A) Brazilian Longitudinal Study of Aging (ELSI) and (B) Mexican Health and Aging Study (MHAS). Occupational complexity was classified according to the International Standard Classification of Occupations 2008 (ISCO-08). Logistic regression was adjusted for age, gender, education, income, alcohol consumption, smoking, body mass index, diabetes, hypertension, and myocardial infarction. P-values represent the p for trend for the categories of occupational complexity, calculated by a two-sided Wald test.

Figure 3.

Probability of cognitive impairment according to occupational complexity considering the interaction between occupational complexity and cohort. ELSI, Brazilian Longitudinal Study of Aging. MHAS, Mexican Health and Aging Study. Occupational complexity was classified according to the International Standard Classification of Occupations 2008 (ISCO-08). Logistic regression was adjusted for age, gender, education, income, alcohol consumption, smoking, body mass index, diabetes, hypertension, and myocardial infarction. P-value for the interaction between occupational complexity and cohort, calculated by a two-sided Wald test.

Discussion

In nationally representative samples of 9,914 participants from Brazil and 16,172 participants from Mexico, more complex occupations were associated with lower odds of cognitive impairment in the ELSI, but not in the MHAS. The odds of cognitive impairment decreased with higher occupation skill levels in the ELSI, but this trend was not seen in the MHAS, where the highest skill level was not associated with cognitive impairment. In pooled analysis, ELSI participants had a lower cognitive impairment probability than MHAS participants with the same occupation skill level.

Our results are in line with previous evidence from high-income countries that showed that higher occupational skill levels were associated with better cognitive performance in later life.^10,11,30 Previous studies have shown mixed results in this regard. A previous study from Brazil using data from the São Paulo Health and Aging Study (SPAH) reported higher odds of dementia in participants who had a non-skilled main occupation.³¹ On the other hand, a study that used data from The Sao Paulo Autopsy Service from the University of Sao Paulo (SPAS-USP) and used the 2010 Standard Occupational Classification of the O^*Net database to create a detailed profile of occupational complexity reported no association between occupation and cognitive abilities.¹⁴ Similarly, another study that investigated the association between occupation and incidence of dementia in Cuba, the Dominican Republic, Venezuela, Peru, Mexico, and China found that occupation was not associated with incident dementia.³² Although gender did not modify the association between occupation and cognitive impairment in our study, a recent study that used data from the Harmonized Cognitive Assessment Protocol (HCAP) in the United States, Mexico, India, and South Africa reported different associations between occupational complexity and cognitive performance for men and women in India.¹⁵ In that study, higher occupational complexity was associated with better cognitive performance for men and women in the United States and Mexico and for men in India.¹⁵ There was no association between occupation and cognitive performance in South Africa.¹⁵

We found a difference between ELSI and MHAS participants in the association between occupational complexity and cognitive impairment, with ELSI participants having a lower probability of cognitive impairment compared to MHAS participants with the same occupation skill level. Differences in the degree of detail of the occupation variables in the ELSI and the MHAS may partly explain this finding.³³ Because the occupation variable we used for the ELSI was more detailed than the one for MHAS, we were able to perform a more consistent skill level classification using the ISCO-08 for the ELSI compared to the occupation variable in the MHAS. Another possible explanation is the length of time participants performed their main occupations during life. Unfortunately, due to information about the length of time in the main occupation not being available in the ELSI, we could not control for this variation and test this hypothesis.

We found that the odds of cognitive impairment decreased with higher occupation skill levels in the ELSI, but this trend was not seen in the MHAS, where the highest skill level was not associated with cognitive impairment. This result may be due to the low number of participants in the highest occupational skill level in the MHAS. Indeed, only 10% of participants from the MHAS had a main occupation with skill levels 3/4, and 6% of them had cognitive impairment, while 20% of MHAS participants had never worked for pay and, of these, 15.6% had cognitive impairment. In fact, compared to high-income countries, LMICs have a greater variance in occupation complexity and there is a higher proportion of lower-complexity occupations.¹⁵ Previous evidence that included nationally representative samples from the United States, Mexico, and India and a regionally representative sample from South Africa showed that 31% of participants of the HRS had main occupations with skill levels 3/4, while only 11% of MHAS participants, 6% of LASI (Longitudinal Aging Study in India) participants, and 4% of HAALSI (Health and Ageing in Africa: A Longitudinal Study in South Africa) participants had these occupation skill levels.¹⁵ Additionally, a high proportion of participants (50%) from the MHAS, LASI, and HAALSI had never worked for pay or had occupations with skill level 1, while only 10% of participants of the HRS were in these occupation categories.¹⁵ Another possible explanation is that education may contribute to cognitive function differently in both countries according to increasing occupation levels. ELSI participants had lower levels of education with increasing occupation levels compared to MHAS participants, particularly for skill levels 3/4. Therefore, for ELSI participants, higher occupation skill levels may contribute to cognitive performance in addition to education, while for MHAS participants with the highest occupation skill levels, education may be a major contributor to cognitive function and occupational complexity may be a minor contributor. Indeed, in models adjusted only for education, the association between occupational complexity and cognitive impairment remains significant for the skill level 3/4 in the ELSI, but not the MHAS.

The increment in cognitive reserve is a possible mechanism that may explain the association between occupational complexity and cognitive function. Cognitive reserve is a property of the brain to withstand insult while maintaining the same level of cognitive performance.⁹ Occupational complexity is an adulthood contributor to cognitive reserve and may be influenced by educational attainment, an early-life contributor to cognitive reserve.⁹ In fact, previous evidence suggests that occupational complexity was a mediator in the association between educational attainment and cognitive performance, explaining 11% to 22% of the protective effect of education on cognitive function.³⁰ On the other hand, work-related stress and increased cortisol levels have been associated with worse cognitive performance.^34,35 This may partly explain why some studies have not found an association between occupational complexity and cognitive function.^10,12,14,15

Our study has strengths. First, it adds important evidence on the association between occupation complexity during adulthood and cognitive impairment in later life in two LMICs. Second, we used nationally representative samples from Brazil and Mexico. Finally, we used data collected in the same year for both countries, which reduces cohort bias. However, the results of this study should be considered in light of study limitations. First, we used cross-sectional data, which does not allow causal inference. We also cannot exclude the possibility of reverse causation, as participants without cognitive impairment may be able to perform more complex occupations. In addition, the level of detail of the occupation variables in the ELSI and the MHAS differed, which may have led to misclassification of occupation skill level. Our harmonization method may not identify cognitive impairment homogeneously in both cohorts since their cognitive test batteries evaluate different cognitive domains. Nevertheless, sensitivity analysis using only the studies’ common tests for the cognitive impairment classification did not change the results. Finally, although we adjusted our model for relevant covariates, we cannot exclude the possibility of bias from unmeasured confounders.

In conclusion, higher occupational complexity was associated with lower odds of cognitive impairment in Brazil, while in Mexico the highest occupational skill level was not associated with cognitive impairment. The results of this study suggest that programs addressing the negative consequences of lower-complexity occupations need to be tailored to specific regional contexts, considering their unique socioeconomic, cultural, and occupational landscapes. Future studies are needed to understand the association of occupational complexity with different cognitive domains and longitudinal studies in LMICs are needed to confirm our findings and examine if occupational demands are associated with slower cognitive decline.