Relation of a Socioeconomic Index with Cognitive Function and Neuroimaging in Hypertensive Individuals

Abstract

Background:

Socioeconomic factors are important contributors to brain health. However, data from developing countries (where social inequalities are the most prominent) are still scarce, particularly about hypertensive individuals.

Objective:

To evaluate the relationship between socioeconomic index, cognitive function, and cortical brain volume, as well as determine whether white matter hyperintensities are mediators of the association of the socioeconomic index with cognitive function in hypertensive individuals.

Methods:

We assessed 92 hypertensive participants (mean age = 58±8.6 years, 65.2%female). Cognitive evaluation and neuroimaging were performed and clinical and sociodemographic data were collected using questionnaires. A socioeconomic index was created using education, income, occupation (manual or non-manual work), and race. The associations of the socioeconomic index with cognitive performance and brain volume were investigated using linear regression models adjusted for age, sex, time of hypertension since diagnosis, and comorbidities. A causal mediation analysis was also conducted.

Results:

Better socioeconomic status was associated with better visuospatial ability, executive function, and global cognition. We found associations between a better socioeconomic index and a higher parietal lobe volume. White matter hyperintensities were also not mediators in the relationship between the socioeconomic index and cognitive performance.

Conclusion:

Socioeconomic disadvantages are associated with worse cognitive performance and brain volume in individuals with hypertension.

Keywords

Brain volume cognitive dysfunction executive function hypertension neuroimaging social determinants of health white matter hyperintensities

INTRODUCTION

Clinical conditions such as hypertension, diabetes, and stroke have been extensively associated with cognitive impairment and brain changes [1 –4]. However, evidence has shown that socioeconomic conditions may play a significant role in the deterioration of brain health [5, 6]. Social determinants are a group of factors apart from clinical variables that have a powerful role in health outcomes [7]. They are the sociodemographic conditions in which a person is born and lives [8]. These factors include poverty, sanitation, clean water, safe food, income, education, ethnicity, and working and housing conditions [8 –10].

Some social determinants, mainly education, have been extensively associated with cognitive function and brain structure [11, 12]. High levels of education and occupation, as well as higher income, are associated with better cognitive performance [13 –15]. Likewise, fewer years of formal education, lower incomes, and lower occupational status are associated with a higher incidence of Alzheimer’s disease [16] and cognitive impairment [17]. Discrimination associated with race/ethnicity has often been associated with adverse mental health outcomes [6 , 19]. However, the effect of these socioeconomic variables on the brain function and structure of patients with hypertension has not yet been considered.

Another great concern regarding the relationship between social determinants and health is the large amount of evidence available from developed countries, whereas evidence from developing countries is scarce [9, 20]. In addition, the isolated way in which socioeconomic variables have usually been considered in previous analyses is an important issue [11, 12]. Considering a multi-risk socioeconomic score could better reflect the common effect of these variables on brain health. Therefore, the aim of this study was to evaluate the association of a socioeconomic index with cognitive performance and brain volume in hypertensive individuals, and whether this association between the socioeconomic index and cognitive performance can be mediated by white matter hyperintensities, a common condition associated with cognitive impairment [21] and hypertension [22].

METHODS

Study design, participants, and ethical approval

This cross-sectional study included patients from the hypertension unit of the São Paulo University Heart Institute. Participants were contacted and recruited via telephone calls between 2016 and 2019. A total of 394 patients were contacted by phone, and a brief explanation of the study protocol was provided. From this first contact, 102 patients agreed to participate in a face-to-face screening that comprised cognitive assessments, clinical examinations, and a sociodemographic questionnaire. Thereafter, magnetic resonance imaging (MRI) was conducted on a subsample of 37 individuals.

Our inclusion criteria included those aged between 40 and 70 years and who had a previous diagnosis of hypertension (according to a medical diagnosis recorded in the hospital’s database or a report of antihypertensive drug use), while the exclusion criteria included those with less than 4 years of education, severe cognitive or communication impairments, and contraindications for performing MRI. The protocol was approved by the local ethics committee (number 4266/15/093), and all participants provided written informed consent.

Cognitive evaluation

Cognition was evaluated using the following neuropsychological tests: the Mini-Mental State Examination (MMSE), Rey Complex Figure Test (RCF), immediate and delayed recall of Stories A and B from the Wechsler Memory Scale-Revised (WMS-R), Digit Span Forward and Backward Subtests from the WMS-R, Letter-Number Sequencing from the Wechsler Adult Intelligence Scale-Third Edition (WAIS-III), FAS letter fluency from the Controlled Oral Word Association Test, and Frontal Assessment Battery (FAB).

The MMSE is a screening test commonly used to assess global cognition. It is a 30-point scale, with higher scores indicating better performance [23].

For the RCF, the participants were asked to copy the figure that was presented to them. Approximately 20 to 30 min later, they were asked to recall and draw the figure again without looking at the figure presented earlier. The maximum score for this test was 35. The total score was based on the number of correctly recalled pieces. This test measures visuospatial abilities [24].

The immediate and delayed recall of Stories A and B from the WMS-R consisted of two short stories that the participant needed to listen to carefully and then recall. After 20 to 30 min, a delayed recall was requested. The score was based on the number of terms that the participant recalled. The maximum score was 25. This subtest was used to assess the subjects’ memory [25].

The digit span forward and backward subtests from the WMS-R consisted of two lists of numeric strings. For the forward subtest, the participants were asked to listen to a numeric string and repeat it in the same order. For the backward subtest, the participants were asked to listen to a numeric string and repeat it in the opposite order. The score on each subtest consisted of the number of correctly remembered sequences. The maximum score for each test was 14. These subtests were used to assess executive functions, such as working memory and attention [25].

The Letter-Number Sequencing subtest from the WAIS-III consisted of sequences of numbers and letters. The participants were asked to listen to the sequence carefully and then repeat it in an organized manner. Numbers should have come first in an ascending order, and letters should have come last in an alphabetical order. The maximum score was 21. This test was used to assess the executive function [25].

The FAS letters fluency test consisted of asking participants to say as many words as possible, starting with the letter ‘F,’ ‘A,’ and ‘S’ for one minute for each letter. The final score was the sum of the number of correct words for each letter. This test was used to measure verbal fluency, language, and attention [25].

The FAB is a questionnaire composed of six subtests about similarities, verbal fluency (letter ‘S’ score from FAS letters fluency test), go/no-go tasks, and gripping behavior. These subtests were summed to obtain a total score of 18. The final score was the sum of correct answers. This test was used to measure executive function [26, 27].

To make more standardized comparisons of the cognitive tests, a Z-score was calculated for each test by subtracting the participant’s test score from the mean sample score and dividing the difference by the sample standard deviation (SD). Thus, a Z-score of -1 represents a cognitive performance that is 1 SD below the mean sample score for each test [28, 29]. A composite memory Z-score was calculated by averaging the Z-scores of the two parts of the stories’ subtests from the WMS-R and then standardizing this mean [28]. The same was made for the two parts of the RCF to obtain a composite Z-score for visuospatial ability. The forward and backward digit span tests, Letter-Number Sequencing, FAS letters fluency test, and the FAB were taken together in a composite executive function Z-score.

Socioeconomic index

Sociodemographic and clinical variables were self-reported. Socioeconomic variables included education, monthly household income in minimum wages, race, and occupation. The occupation was reported by the participant and later classified as manual (mainly requiring physical effort) or non-manual work (mainly requiring mental effort) by the researchers. These variables have been reported as important social determinants of health [9]. We assigned a score of 1 or 0 to each socioeconomic variable according to their positive or negative association with cognitive performance respectively [13 –17]. For the quantitative variables, which were education and income, the cut-off was the median value, with 1 attributed to those equal or above the median and 0 to those below the median. We also attributed a score of 1 to patients classified as white or other and 0 to those classified as black or brown. We assigned a score of 1 for non-manual work and 0 for manual work. The social index variable was equal to the sum of the scores attributed to each of the four socioeconomic variables (Fig. 1), with higher values indicating better socioeconomic status.

Fig. 1

Variables and score system used to build the socioeconomic index.

MRI

A subsample of 37 participants was scanned using a 3.0 T MR system (Philips Achieva, Eindhoven, the Netherlands) that was equipped with 80 mT/m gradients and a 32-channel head coil. T1-weighted SENSE 3D images (repetition time/echo time of 7.0 / 3.2 ms, flip angle 8°, Sense 1.5, a field of view of 240×240, matrix 240×240, 180 slices of 1 mm each with no gap, voxel size of 1 mm³, and acquisition time 6.01 min) were acquired for morphometric brain analysis. Axial fluid-attenuated inversion recovery images were collected (TR/TE, 11000/130 ms; inversion time, 2800 ms; FOV, 230×183; matrix, 356×210; and 28 slices, 4.5 mm each with a 0.5 gap) to identify incidental findings, such as lacunar infarcts and white matter hyperintensities. All images were visually inspected by a board-certified neuroradiologist.

T1-weighted MRI data were analyzed using FreeSurfer version 6.0 (https://surfer.nmr.mgh.harvard.edu/). The software provides a standard full processing pipeline for structural MRI data, including skull stripping, B1 bias field correction, gray-white matter and cerebrospinal fluid (CSF) segmentation, reconstruction of cortical surface models, and labeling of regions on the cortical surface and subcortical brain structures. Cortical parcellation was performed using the Destrieux atlas. Cortical thickness and volume were calculated, and their structures were measured bilaterally in all segments. For total brain volume, we considered the volume of all voxels of the brain segmentation without the ventricles, CSF, choroid plexus, and brain stem. For the hippocampus, frontal lobe, parietal lobe, temporal lobe, and occipital lobe volumes, we considered the sum of the regional volumes to produce global measures. All volumes were corrected for estimated total intracranial volume [30].

White matter hyperintensities (WMH) were defined according to the Fazekas scale [31, 32]. Periventricular and deep WMHs were rated separately. Periventricular hyperintensities were scored as follows: 0 = absence, 1 = ‘caps’ or pencil-thin lining, 2 = smooth ‘halo,’ and 3 = irregular periventricular hyperintensities extending into the deep white matter. Deep WMH were scored as follows: 0 = absence, 1 = punctate foci, 2 = beginning confluence of foci, and 3 = large confluent areas. The total Fazekas score used in the analysis, ranging from 0 to 6, was obtained by summing the periventricular and deep white matter scores [33].

Other variables

Possible confounding variables of the relationship between the socioeconomic index and cognitive performance included age, sex, the clinical variables that were grouped in the comorbidity index, and the duration of hypertension in years since the initial diagnosis.

The clinical variables grouped in the comorbidity index were body mass index (BMI), previous diagnosis of stroke or myocardial infarction, presence of diabetes or dyslipidemia defined by hypoglycemic and lipid-lowering drug use, and smoking status (never, current, or previous smoker). These variables were previously associated with cognitive impairment [2 , 34–36]. Thus, for the numeric BMI variable, the cut-off was the median value, with 0 attributed to those below the median and 1 to those equal or above the median. For dichotomous categorical variables (stroke, myocardial infarction, diabetes, and dyslipidemia), we assigned a score of 0 if the variable was absent and 1 if it was present. For smoking, we attributed a 0 to those defined as never smokers or those who had quit smoking for more than one year, and 1 to those who were current smokers or who had quit smoking for less than one year. The comorbidity index variable was equal to the sum of each index attributed to these six clinical variables (Fig. 2), with higher values indicating higher cardiovascular comorbidity.

Fig. 2

Variables and score system used to build the clinical comorbidity index.

Statistical analysis

To describe the quantitative variables, we used the mean and SD. Categorical variables are described as relative frequencies. Data distribution was determined using the Shapiro–Wilk test. The independent samples t-test, Mann–Whitney U test, and chi-square test were used to compare groups. To investigate the association of the socioeconomic index or their individual variables with cognitive performance and brain volume, we used linear regression models adjusted for age, sex, comorbidity index, and time since hypertension diagnosis. Normal quantile-quantile (Q-Q) plots of the residuals and plots of the residuals versus the predicted values, as well as histograms of the residuals, were used to assess the assumptions of linear regression. The absence of multicollinearity was verified using variance inflation factor. The dependent variables were the composite Z-scores of the cognitive tests or cortical structure volumes (in mm³). We investigated whether WMH are mediators of the association between socioeconomic index and cognitive performance. The total Fazekas score for WMHs was used as a mediating factor. The mediation model comprised the direct effect of the socioeconomic index in the composite Z-scores of the cognitive tests and the indirect effect of the total Fazekas score in the composite Z-scores of the cognitive tests. Causal mediation analysis was performed with the “mediation” R package for causal mediation analysis [37, 38]. Statistical analyses were performed using R software (version 4.0.0). The statistical significance was set at 5%.

RESULTS

Sociodemographic and clinical characteristics of the sample

Of the 102 participants screened, five participants who did not meet the educational level criteria, four that had incomplete cognitive test data, and one that did not have a hypertension diagnosis were excluded. Ninety-two participants met the eligibility criteria and were included in the analysis of this cross-sectional study (Fig. 3). The sociodemographic and clinical characteristics of the patients are presented in Table 1. The mean age of the sample was 58.0±8.6 years, and 65.2%were women. The mean duration devoted to education was 11.3±3.8 years. All participants had hypertension, with a mean time of diagnosis of 19.9±11.9 years, and diuretics were the most frequently prescribed drugs. The comparison of cognitive performance and sociodemographic and clinical characteristics between participants with and without MRI data is shown in Table 2. Most characteristics were similar, except for the participants’ monthly household income, which was higher among participants who underwent MRI. Previous diagnosis of stroke was more common among participants who were not evaluated with MRI. In this subsample with MRI data, 72.9%(n = 27) of the participants ranged from 0 to 2 on the Fazekas scale, and no evidence of brain lesions as lacunes in their cortical or cortico/subcortical regions, nor WMH lesions in their subcortical regions that resembled CSF signal intensity were observed.

Fig. 3

Flowchart of the study participants.

Table 1

Sociodemographic and clinical characteristics of the sample (n = 92)

Variables	Mean (SD) or n (%)
Age (y), mean (SD)	58.07 (8.6)
Female, n (%)	60 (65.2)
Race, n (%)
White	43 (46.7)
Black	24 (26.1)
Brown	22 (23.9)
Others	3 (3.3)
Schooling (y), mean (SD)	11.29 (3.8)
Monthly income (USD^*), mean (SD)	840 (761)
Diabetes, n (%)	22 (23.9)
Dyslipidemia, n (%)	46 (50)
Stroke, n (%)	8 (8.7)
Myocardial infarction, n (%)	5 (5.4)
BMI (Kg/m²), mean (SD)	30.10 (5.5)
Smoking, n (%)
Never	52 (56.5)
Current	9 (9.8)
Previous	31 (33.7)
Years since Hypertension diagnosis, mean (SD)	20 (12)
Number of drugs, mean (SD)	5.02 (2.6)
Most-used drugs, n (%)
ARB	48 (52.2)
ACEI	38 (41.3)
Diuretics	64 (69.6)
CCB	47 (51.1)
BB	43 (46.7)
Others	55 (59.8)

^*1 USD, 5.59 Real; BMI, body mass index; ARB, angiotensin II receptor blockers; ACEI, angiotensin-converting enzyme inhibitors; CCB, calcium-channel blockers; BB, beta-blockers.

Table 2

Comparison between samples with and without MRI

Variables	Without MRI (n = 55)	With MRI (n = 37)	p
Age (y), median (IQR)	60 (13.5)	59 (14)	0.77
Female, n (%)	34 (36.96)	26 (28.26)	0.54
Race, n (%)			0.19
White	27 (29.4)	16 (17.4)
Black	17 (18.5)	7 (7.6)
Brown	9 (9.8)	13 (14.1)
Others	2 (2.2)	1 (1.1)
Schooling (y), median (IQR)	12 (3)	12 (5)	0.76
Monthly income (USD^*, median (IQR)	536 (469)	894 (626)	0.02
Diabetes, n (%)	16 (17.4)	6 (6.5)	0.24
Dyslipidemia, n (%)	29 (31.5)	17 (18.5)	0.67
Stroke, n (%)	8 (8.7)	0 (0)	0.04
Myocardial infarction, n (%)	4 (4.4)	1 (1.1)	0.63
BMI (Kg/m²), median (IQR)	29.8 (7.1)	31.2 (8.2)	0.40
Smoking, n (%)			0.86
Never	30 (32.6)	22 (23.9)
Current	6 (6.5)	3 (3.3)
Previous	19 (20.7)	12 (13.0)
Years since Hypertension diagnosis, mean (SD)	21.73 (12.7)	17.31 (10.5)	0.08
Number of drugs, median (IQR)	5 (3)	4 (3)	0.08
Most-used drugs, n (%)
ARB	28 (30.4)	20 (21.7)	0.93
ACEI	24 (26.1)	14 (15.2)	0.73
Diuretics	37 (40.2)	27 (29.4)	0.72
CCB	30 (32.6)	17 (18.5)	0.55
BB	25 (27.2)	18 (19.6)	0.92
Others	34 (37.0)	21 (22.8)	0.78
Memory, mean (SD)	–0.06 (1.0)	0.1 (1.0)	0.43
Visuospatial ability, mean (SD)	–0.01 (1.1)	0.02 (0.8)	0.84
Executive Function, mean (SD)	0.03 (1.1)	–0.05 (0.8)	0.62
MEEM, median (IQR)	0.18 (1.3)	0.18 (1.3)	0.17

^*1 USD, 5.59 Real; IQR, Interquartile range; BMI, body mass index; ARB, angiotensin II receptor blockers; ACEI, angiotensin-converting enzyme inhibitors; CCB, calcium-channel blockers; BB, beta-blockers; MMSE, Mini-Mental State Examination.

Socioeconomic index and cognitive performance

Higher scores in the socioeconomic index were associated with better visuospatial ability (β= 0.29, 95%CI = 0.13 to 0.46, p < 0.001), higher executive function (β= 0.35, 95%CI = 0.20 to 0.51, p < 0.001), and higher global cognition z-scores (β= 0.33, 95%CI = 0.18 to 0.49, p < 0.001) in the adjusted analyses for age, sex, time since hypertension diagnosis, and clinical comorbidities. In contrast, the socioeconomic index was not associated with memory performance (β= 0.15, 95%CI = –0.02 to 0.32, p = 0.09) (Table 3).

Table 3

Association between socioeconomic index and cognitive performance (n = 92)

Cognitive domains	Socioeconomic index
	Model 1			Model 2			Model 3
	β	95%CI	p	β	95%CI	p	β	95%CI	p
Memory	0.17	0.01 to 0.33	0.04	0.16	–0.00 to 0.33	0.05	0.15	–0.02 to 0.32	0.09
Visuospatial ability	0.26	0.10 to 0.42	0.001	0.29	0.14 to 0.45	< 0.001	0.29	0.13 to 0.46	< 0.001
Executive Function	0.36	0.22 to 0.51	< 0.001	0.37	0.22 to 0.51	< 0.001	0.35	0.20 to 0.51	< 0.001
Global Cognition	0.35	0.21 to 0.50	< 0.001	0.35	0.20 to 0.50	< 0.001	0.33	0.18 to 0.49	< 0.001

Model 1: Unadjusted linear regression model. Model 2: Linear regression model adjusted for age and sex. Model 3: Linear regression model adjusted for age, sex, time since hypertension diagnosis, and clinical comorbidity index.

Individual socioeconomic variables were considered in a multivariate model adjusted for age, sex, time since hypertension diagnosis, and clinical comorbidity index (Table 4). Higher education was an independent predictor for higher visuospatial ability (β= 0.11, 95%CI = 0.04 to 0.18, p < 0.01), executive function (β= 0.09, 95%CI = 0.04 to 0.16, p < 0.001), and global cognition z-scores (β= 0.10, 95%CI = 0.04 to 0.17, p < 0.01). Higher income was an independent predictor of better executive function z-scores (β= 0.06, 95%CI = 0.01 to 0.11, p = 0.02). Occupation status and race were not independently associated with any cognitive tests (Table 4). Furthermore, the associations between each isolated socioeconomic variable and cognitive functions were analyzed in separate models adjusted for age, sex, time since hypertension diagnosis, and the clinical comorbidity index, and are presented in Supplementary Table 1. Higher education was associated with better performance in all cognitive functions. The higher the income, the better the performance in terms of visuospatial ability, executive function, and global cognition. The non-manual work category of occupation was associated with better memory performance, visuospatial ability, and executive function. Finally, non-black participants had higher global cognitive performance than black participants (p = 0.04) (Supplementary Table 1). The interaction between the socioeconomic index and race was also analyzed, but there was no significant interaction of this term with any cognitive test (Supplementary Table 2).

Table 4

Association between socioeconomic index variables and cognitive performance (n = 92)

Cognitive domains	Education			Income			Occupation			Race
	β	95%CI	p	β	95%CI	p	β	95%CI	p	β	95%CI	p
Memory	0.05	–0.02 to 0.12	0.15	–0.02	–0.08 to 0.04	0.43	0.36	–0.13 to 0.84	0.15	–0.18	–0.60 to 0.25	0.40
Visuospatial ability	0.11	0.04 to 0.18	<0.01	–0.01	–0.06 to 0.04	0.61	0.17	–0.28 to 0.63	0.45	0.12	–0.28 to 0.52	0.55
Executive Function	0.09	0.04 to 0.16	<0.01	0.06	0.01 to 0.11	0.02	0.11	–0.31 to 0.52	0.61	–0.01	–0.38 to 0.35	0.95
Global Cognition	0.10	0.04 to 0.17	<0.01	0.01	–0.05 to 0.06	0.82	0.01	–0.43 to 0.45	0.95	0.33	–0.06 to 0.71	0.09

All socioeconomic variables in the table were considered predictors in a multivariate linear regression model adjusted for age, sex, time since hypertension diagnosis, and clinical comorbidity index. Reference category for occupation: manual work; reference category for race: black.

Socioeconomic index and brain volumes

In the linear regression analysis adjusted for age, sex, time since hypertension diagnosis, and the clinical comorbidity index, we found that better socioeconomic index was associated with higher parietal lobe volume (β= 5557, 95%CI = 679 to 10434, p =0.03). However, we did not find an association between socioeconomic index and total brain (β=38777, 95%CI = –10743 to 88297, p = 0.12), temporal lobe (β= 3010, 95%CI = –769 to 6791, p = 0.11), occipital lobe (β= 1749, 95%CI = –1256 to 4755, p = 0.24), hippocampus (β= 253, 95%CI = –63 to 569, p = 0.11), and gray (β= 21063, 95%CI = –6872 to 48999, p = 0.13), and white matter volumes (β= 17456, 95%CI = –5394 to 40308, p = 0.13) (Table 5).

Table 5

Association between socioeconomic index and brain volumes (n = 37)

	Socioeconomic Index			Comorbidity Index
	β	95%CI	p	β	95%CI	p
Total brain (mm³)	38777	–10743 to 88297	0.12	34540	–19342 to 88422	0.20
Frontal lobe (mm³)	2852	–2342 to 8047	0.27	1289	–4362 to 6941	0.64
Parietal lobe (mm³)	5557	679 to 10434	0.03	2728	–2578 to 8035	0.30
Temporal lobe (mm³)	3010	–769 to 6791	0.11	1634	–2478 to 5747	0.42
Occipital lobe (mm³)	1749	–1256 to 4755	0.24	2241	–1028 to 5512	0.17
Hippocampus (mm³)	253	–63 to 569	0.11	154	–190 to 498	0.37
Gray matter (mm³)	21063	–6872 to 48999	0.13	14533	–15863 to 44930	0.34
White matter (mm³)	17456	–5394 to 40308	0.13	20223	–4640 to 45088	0.11

Linear regression model adjusted for age, sex, time since hypertension diagnosis, and clinical comorbidity index.

Mediation effects

In a mediation model that comprises the socioeconomic index, WMH evaluated by the Fazekas score, and cognitive performance adjusted for age, sex, time since hypertension diagnosis, and clinical comorbidity index, we found that WMH was not a mediator in the association of the socioeconomic index with visuospatial ability (β= –0.04, 95%CI = –0.14 to 0.00, p = 0.16), executive function performance (β= –0.002, 95%CI = –0.04 to 0.03, p = 0.96), or global cognition (β= 0.02, 95%CI = –0.03 to 0.11, p = 0.44). We observed a direct effect of the socioeconomic index on visuospatial ability and executive function (Supplementary Table 3).

DISCUSSION

In this cross-sectional study with hypertensive participants, we found that higher scores in the socioeconomic index (composed of education, income, occupation, and race) were independently associated with better performance in visuospatial ability, executive function, and global cognition. A higher socioeconomic index was independently associated with a higher parietal lobe volume. However, WMH was not a mediator of the association between socioeconomic index and cognition.

Several studies have shown that hypertension is an important risk factor for cognitive impairment and is associated with lower processing speed and executive function [1 , 39]. However, the impact of social factors on cognition in this population has not yet been considered. The idea that social determinants play an important role in health has been gaining traction in recent decades. In 2005, the World Health Organization launched a Commission on Social Determinants of Health to provide evidence and encourage public policies to decrease the impact of social inequalities on people’s health worldwide [9]. In addition, the identification of social determinants in developing countries was particularly encouraged [9], since social inequalities are great in these settings, with rampant poverty affecting millions of people [9, 20]. Thus, understanding the relationship between socioeconomic factors and cognition, especially in large developing countries like Brazil, which have the worst social inequalities in Latin America [40], is of paramount importance.

Socioeconomic status, usually measured by income, education or occupation, is among the most important determinants of health in almost every society worldwide (WHO Health Commission 2008) [41], and they are also the most common social factors associated with cognitive performance [13–15 , 42–44]. Both better education and occupation were associated with better global cognitive performance (composed of memory, executive function, visuospatial ability, and language) in participants without dementia [13]. In a large sample of Chinese middle-aged individuals, individuals with higher self-reported financial status and whose parents had a higher education and engaged in non-agricultural work showed better baseline cognitive performance [45]. Other studies have also demonstrated the relationship between better socioeconomic status in childhood and better performance of executive functions later [46]. Similarly, we found an association between the socioeconomic index and visuospatial ability, executive function, and global cognition, but not memory. Memory impairments are commonly associated with neurodegenerative dementia, such as Alzheimer’s disease [47], which can explain the lack of association in the adjusted analysis. Our multivariate analyses also showed an independent relationship between education and visuospatial ability, executive function, and global cognition, and an independent relationship between income and executive function. These results have already been demonstrated in previous studies [14, 15], with education being the individual socioeconomic factor that is most associated with cognitive impairment and dementia [14, 16]. In contrast, we need to highlight that no single socioeconomic variable reflects the complexity of social disadvantages. In addition, the effect of a single socioeconomic variable can often be overestimated when considered alone [48]. Therefore, combining socioeconomic factors may better reflect the cumulative effects of risk on health [49].

Racial and ethnic disparities are associated with higher rates of morbidity and mortality in the United States and globally [18 , 50]. According to data from the Brazilian Institute of Geography and Statistics (IBGE), the illiteracy rate and frequency of extreme poverty were higher among black and brown adults. This condition represents a barrier in accessing good education, better jobs, and sufficient income. Thus, understanding the complex ways in which race and socioeconomic status uniquely and in combination contribute to health outcomes is critical [41]. In this context, our results contribute to showing the association of these factors combined in an index and cognitive performance in individuals with hypertension, a disease that affects 32 million Brazilians [51] and is one of the main risk factors for dementia [52]. Poor access to healthcare and low quality of life likely contribute to the association between lower socioeconomic index and worse cognitive performance in our findings, increasing the burden for this vulnerable population. It is important to note that race has no causal effect on brain function. Racism and discrimination greatly contribute to social disadvantages. Another important issue is the non-equivalence of socioeconomic indicators that were combined in the socioeconomic index (e.g., income and education) across racial groups [19 , 53]. Considering the above, race can interact with other socioeconomic factors and lead to different effects on health [44, 54].

Many studies have shown associations between socioeconomic status and brain structure [55 –57]. However, due to the diversity of these populations and the methodologies used, it is not possible to establish a specific relationship between socioeconomic status and certain brain regions [44]. In our study, we found an association between a higher socioeconomic index and a greater volume of the parietal lobe, a cortical region involved in visuospatial ability and attention [58, 59], as well as some executive functions [60]. However, we did not find an association between the socioeconomic index and the volumes of the other brain lobes. In contrast, Waldstein et al. found that non-Hispanic whites with higher socioeconomic status had greater total brain, white matter, and gray matter volumes compared to white individuals with low socioeconomic status, and African Americans with low and high socioeconomic status [49]. Brain atrophy is usually associated with more severe deterioration, such as neuronal loss and dementia [61], but educational, occupational, and socioeconomic factors have contributed to minimizing cognitive decline, preclinical neurodegeneration, and slowing down the reduction in brain volume due to aging [62]. In particular, neuronal loss in the hippocampus has been linked to Alzheimer’s disease [63, 64], but functional changes in this structure precede structural changes (atrophy) [65], which may possibly explain the lack of association between the socioeconomic index and hippocampal volumes in our sample of individuals without dementia.

The pattern of cognitive changes associated with hypertension involves impairments in processing speed and executive function, a pattern that is more frequently found in cerebrovascular diseases [66]. The pathophysiological mechanisms of hypertension behind this pattern are not fully understood, but they are commonly associated with white matter lesions [67], small vessel diseases [68, 69], and vascular lesions that are responsible for the detrimental effects on brain structure and function [70, 71]. Many studies have shown that WMH are associated with impaired cognitive function [72], especially in individuals with hypertension [73]. Thus, we investigated whether WMH could mediate the relationship between socioeconomic index and cognitive performance. However, we did not find a mediation effect in our sample. Social risk factors seem to be independently associated with brain dysfunction, or vascular lesions other than WMH could indirectly affect this relationship. Another explanation may be that most participants in this subsample had mild effects on the Fazekas scale (72.9%ranged from 0 to 2 on the Fazekas scale), and we did not find evidence of lacunar infarcts in their cortical or cortico/subcortical regions. Although the Fazekas score is widely used to classify WMH [33 , 74–76], we did not compare this score with the WHM volume measurements using a computational quantitative program; this needs to be explored in future analyses.

Finally, we must highlight the limitations of the study. First, this was an observational study based on a cross-sectional analysis and it did not establish causation between the socioeconomic index and brain function. In addition, the small size of the sample is an important issue that will be improved with larger future samples. Although we adjusted our analyses for demographic and clinical variables, it is possible that unmeasured confounders were present. The clinical variables needed to be grouped in an index because of the limited sample size compared to the number of adjustments required in the statistical analysis. Thus, we were unable to include each clinical variable in the model. In addition, it was not possible to describe how controlled hypertension was because we did not measure the subjects’ current blood pressures. However, the use of antihypertensive drugs was described, and diuretics were the most frequently prescribed drugs. Finally, only a small subset of the sample underwent MRI evaluation. However, the characteristics of participants who underwent MRI were similar to those who did not undergo MRI, except for household income, which was higher among participants who received MRI, and previous diagnosis of stroke, which was more common among participants who were not evaluated with MRI. Although the Fazekas score is widely used to classify WMH, we did not compare this score with the WHM volume measurements using a computational quantitative program.

In conclusion, we found that higher socioeconomic index scores were associated with better visuospatial ability, executive function, and global cognition. WMH were not mediators of this association. Higher socioeconomic index scores were also associated with a greater parietal lobe volume. These findings suggest that social disadvantages are associated with brain dysfunction in individuals with hypertension or are associated with hypertensive vascular lesions other than WMH.

Footnotes

ACKNOWLEDGMENTS

This work was supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior –Brasil (CAPES) –88882.179915/2018-01, São Paulo Research Foundation (FAPESP) grant 2018/19006-2, and Conselho Nacional de Desenvolvimento Científico e Tecnologico (CNPq) grant 307138/2015-1.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Iadecola

, Yaffe

, Biller

, Bratzke

, Faraci

, Gorelick

, Gulati

, Kamel

, Knopman

, Launer

, Saczynski

, Seshadri

, Zeki Al Hazzouri

, American Heart Association Council on Hypertension; Council on Clinical Cardiology; Council on Cardiovascular Disease in the Young; Council on Cardiovascular and Stroke Nursing; Council on Quality of Care and Outcomes Research; and Stroke Council (2016) Impact of hypertension on cognitive function: A scientific statement from the American Heart Association. Hypertension 68, e67–e94.

Zilliox

, Chadrasekaran

, Kwan

, Russell

(2016) Diabetes and cognitive impairment. Curr Diab Rep 16, 87.

Kalaria

, Akinyemi

, Ihara

(2016) Stroke injury, cognitive impairment and vascular dementia. Biochim Biophys Acta 1862, 915–925.

Lane

, Barnes

, Nicholas

, Sudre

, Cash

, Parker

, Malone

, Lu

, James

S-N

, Keshavan

, Murray-Smith

, Wong

, Buchanan

, Keuss

, Gordon

, Coath

, Barnes

, Dickson

, Modat

, Thomas

, Crutch

, Hardy

, Richards

, Fox

, Schott

(2019) Associations between blood pressure across adulthood and late-life brain structure and pathology in the neuroscience substudy of the 1946 British birth cohort (Insight 46): An epidemiological study. Lancet Neurol 18, 942–952.

Aggarwal

, Everson-Rose

, Evans

(2015) Social determinants, race, and brain health outcomes: Findings from the Chicago Health and Aging Project. Curr Alzheimer Res 12, 622–631.

Alegría

, NeMoyer

, Falgàs Bagué

, Wang

, Alvarez

(2018) Social determinants of mental health: Where we are and where we need to go. Curr Psychiatry Rep 20, 95.

Braveman

, Gottlieb

(2014) The social determinants of health: It’s time to consider the causes of the causes. Public Health Rep 129(Suppl 2), 19–31.

Spruce

(2019) Back to basics: Social determinants of health. AORN J 110, 60–69.

Marmot

(2005) Social determinants of health inequalities. Lancet 365, 1099–1104.

10.

Marmot

, Friel

, Bell

, Houweling

, Taylor

(2008) Closing the gap in a generation: Health equity through action on the social determinants of health. Lancet 372, 1661–1669.

11.

Prencipe

, Casini

, Ferretti

, Lattanzio

, Fiorelli

, Culasso

(1996) Prevalence of dementia in an elderly rural population: Effects of age, sex, and education. J Neurol Neurosurg Psychiatry 60, 628–633.

12.

Foubert-Samier

, Catheline

, Amieva

, Dilharreguy

, Helmer

, Allard

, Dartigues

J-F

(2012) Education, occupation, leisure activities, and brain reserve: A population-based study. Neurobiol Aging 33, 423.e15–25.

13.

Vemuri

, Lesnick

, Przybelski

, Machulda

, Knopman

, Mielke

, Roberts

, Geda

, Rocca

, Petersen

, Jack

(2014) Association of lifetime intellectual enrichment with cognitive decline in the older population. JAMA Neurol 71, 1017–1024.

14.

Contador

, Del Ser

, Llamas

, Villarejo

, Benito-León

, Bermejo-Pareja

(2017) Impact of literacy and years of education on the diagnosis of dementia: A population-based study. J Clin Exp Neuropsychol 39, 112–119.

15.

Shen

, Cox

, Adams

, Howard

, Lawrie

, Ritchie

, Bastin

, Deary

, McIntosh

, Whalley

(2018) Resting-state connectivity and its association with cognitive performance, educational attainment, and household income in the UK Biobank. Biol Psychiatry Cogn Neurosci Neuroimaging 3, 878–886.

16.

Evans

, Hebert

, Beckett

, Scherr

, Albert

, Chown

, Pilgrim

, Taylor

(1997) Education and other measures of socioeconomic status and risk of incident Alzheimer disease in a defined population of older persons. Arch Neurol 54, 1399–1405.

17.

Brody

, Kramarow

, Taylor

, McGuire

(2019 Cognitive performance in adults aged 60 and over: National Health and Nutrition Examination Survey, 2011-2014. Natl Health Stat Rep, pp. 1-23.

18.

Berger

, Sarnyai

(2015) “More than skin deep”: Stress neurobiology and mental health consequences of racial discrimination. Stress 18, 1–10.

19.

Pachter

, Caldwell

, Jackson

, Bernstein

(2018) Discrimination and mental health in a representative sample of African-American and Afro-Caribbean youth. J Racial Ethn Health Disparities 5, 831–837.

20.

Eshetu

, Woldesenbet

(2011) Are there particular social determinants of health for the world’s poorest countries? Afr Health Sci 11, 108–115.

21.

H-Y

, Ou

Y-N

, Shen

X-N

, Qu

, Ma

Y-H

, Wang

Z-T

, Dong

, Tan

, Yu

J-T

(2021) White matter hyperintensities and risks of cognitive impairment and dementia: A systematic review and meta-analysis of 36 prospective studies. Neurosci Biobehav Rev 120, 16–27.

22.

Zhao

, Ke

, He

, Cai

(2019) Volume of white matter hyperintensities increases with blood pressure in patients with hypertension. J Int Med Res 47, 3681–3689.

23.

Muela

HCS

, Costa-Hong

, Yassuda

, Moraes

, Memória

, Machado

, Macedo

, Shu

EBS

, Massaro

, Nitrini

, Mansur

, Bortolotto

(2017) Hypertension severity is associated with impaired cognitive performance. J Am Heart Assoc 6, e004579.

24.

Dwan

, Ownsworth

, Chambers

, Walker

, Shum

DHK

(2015) Neuropsychological assessment of individuals with brain tumor: Comparison of approaches used in the classification of impairment. Front Oncol 5, 56.

25.

Strauss

, Sherman

EMS

, Spreen

(2006), A Compendium of Neuropsychological Tests, Oxford University Press, New York.

26.

Cunha

, Nicastri

, de Andrade

, Bolla

(2010) The frontal assessment battery (FAB) reveals neurocognitive dysfunction in substance-dependent individuals in distinct executive domains: Abstract reasoning, motor programming, and cognitive flexibility. Addict Behav 35, 875–881.

27.

Yamashiro

, Tanaka

, Okuma

, Shimura

, Ueno

, Miyamoto

, Urabe

, Hattori

(2014) Cerebral microbleeds are associated with worse cognitive function in the nondemented elderly with small vessel disease. Cerebrovasc Dis Extra 4, 212–220.

28.

Rawlings

, Sharrett

, Schneider

ALC

, Coresh

, Albert

, Couper

, Griswold

, Gottesman

, Wagenknecht

, Windham

, Selvin

(2014) Diabetes in midlife and cognitive change over 20 years: A cohort study. Ann Intern Med 161, 785–793.

29.

de Souza-Talarico

, Suemoto

, Santos

, Griep

, Yamaguti

STF

, Lotufo

, Bensenõr

IJM

(2020) Work-related stress and cognitive performance among middle-aged adults: The Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Stress Health 36, 19–30.

30.

Yasar

, Moored

, Adam

, Zabel

, Chuang

Y-F

, Varma

, Carlson

(2020) Angiotensin II blood levels are associated with smaller hippocampal and cortical volumes in cognitively normal older adults. J Alzheimers Dis 75, 521–529.

31.

Fazekas

, Niederkorn

, Schmidt

, Offenbacher

, Horner

, Bertha

, Lechner

(1988) White matter signal abnormalities in normal individuals: Correlation with carotid ultrasonography, cerebral blood flow measurements, and cerebrovascular risk factors. Stroke 19, 1285–1288.

32.

Fazekas

, Chawluk

, Alavi

, Hurtig

, Zimmerman

(1987) MR signal abnormalities at 1.5 T in Alzheimer’s dementia and normal aging. AJR Am J Roentgenol 149, 351–356.

33.

Valdés Hernández M del

, Morris

, Dickie

, Royle

, Muñoz Maniega

, Aribisala

, Bastin

, Deary

, Wardlaw

(2013) Close correlation between quantitative and qualitative assessments of white matter lesions. Neuroepidemiology 40, 13–22.

34.

Concha-Cisternas

, Lanuza

, Waddell

, Sillars

, Leiva

, Troncoso

, Martinez

, Villagrán

, Mardones

, Martorell

, Nazar

, Ulloa

, Labraña

, Diaz-Martinez

, Sadarangani

, Alvarez

, Ramirez-Campillo

, Garrido-Mendez

, Luarte

, Ho

, Gray

, Petermann-Rocha

, Celis-Morales

(2019) Association between adiposity levels and cognitive impairment in the Chilean older adult population. J Nutr Sci 8, e33.

35.

Levine

, Davydow

, Hough

, Langa

, Rogers

MAM

, Iwashyna

(2014) Functional disability and cognitive impairment after hospitalization for myocardial infarction and stroke. Circ Cardiovasc Qual Outcomes 7, 863–871.

36.

, Yin

, Zhu

, Luo

, Shi

, Gao

(2017) Blood cholesterol in late-life and cognitive decline: A longitudinal study of the Chinese elderly. Mol Neurodegener 12, 24.

37.

Imai

, Keele

, Tingley

(2010) A general approach to causal mediation analysis. Psychol Methods 15, 309–334.

38.

Tingley

, Yamamoto

, Hirose

, Keele

, Imai

(2014) mediation: R Package for Causal Mediation Analysis. J Stat Softw 59, 1–38.

39.

Launer

, Masaki

, Petrovitch

, Foley

, Havlik

(1995) The association between midlife blood pressure levels and late-life cognitive function. The Honolulu-Asia Aging Study. JAMA 274, 1846–1851.

40.

Tramujas Vasconcellos Neumann

, Albert

(2018) Aging in Brazil. Gerontologist 58, 611–617.

41.

Williams

, Priest

, Anderson

(2016) Understanding associations between race, socioeconomic status and health: Patterns and prospects. Health Psychol 35, 407–411.

42.

Rodriguez

, Zheng

, Chui

; Aging Brain: Vasculature, Ischemia, and Behavior Study (2019) Psychometric characteristics of cognitive reserve: How high education might improve certain cognitive abilities in aging. Dement Geriatr Cogn Disord 47, 335–344.

43.

Lee

, Kawachi

, Berkman

, Grodstein

(2003) Education, other socioeconomic indicators, and cognitive function. Am J Epidemiol 157, 712–720.

44.

Farah

(2017) The neuroscience of socioeconomic status: Correlates, causes, and consequences. Neuron 96, 56–71.

45.

Sha

, Yan

, Cheng

(2018) Associations of childhood socioeconomic status with mid-life and late-life cognition in Chinese middle-aged and older population based on a 5-year period cohort study. Int J Geriatr Psychiatry 33, 1335–1345.

46.

Last

, Lawson

, Breiner

, Steinberg

, Farah

(2018) Childhood socioeconomic status and executive function in childhood and beyond. PloS One 13, e0202964.

47.

Jahn

(2013) Memory loss in Alzheimer’s disease. Dialogues Clin Neurosci 15, 445–454.

48.

Adler

, Bush

, Pantell

(2012) Rigor, vigor, and the study of health disparities. Proc Natl Acad Sci U S A 109(Suppl 2), 17154–17159.

49.

Waldstein

, Dore

, Davatzikos

, Katzel

, Gullapalli

, Seliger

, Kouo

, Rosenberger

, Erus

, Evans

, Zonderman

(2017) Differential associations of socioeconomic status with global brain volumes and white matter lesions in African American and white adults: The HANDLS SCAN Study. Psychosom Med 79, 327–335.

50.

Williams

(2012) Miles to go before we sleep: Racial inequities in health. J Health Soc Behav 53, 279–295.

51.

Nascimento

, Brant

LCC

, Yadgir

, Oliveira

GMM

, Roth

, Glenn

, Mooney

, Naghavi

, Passos

VMA

, Duncan

, Silva

DAS

, Malta

, Ribeiro

ALP

(2020) Trends in prevalence, mortality, and morbidity associated with high systolic blood pressure in Brazil from 1990 to 2017: Estimates from the “Global Burden of Disease 2017” (GBD 2017) study. Popul Health Metr 18, 17.

52.

Walker

, Power

, Gottesman

(2017) Defining the relationship between hypertension, cognitive decline, and dementia: A review. Curr Hypertens Rep 19, 24.

53.

Williams

, Mohammed

, Leavell

, Collins

(2010) Race, socioeconomic status, and health: Complexities, ongoing challenges, and research opportunities. Ann N Y Acad Sci 1186, 69–101.

54.

Braveman

, Cubbin

, Egerter

, Chideya

, Marchi

, Metzler

, Posner

(2005) Socioeconomic status in health research: One size does not fit all. JAMA 294, 2879–2888.

55.

Blair

, Raver

(2016) Poverty, stress, and brain development: New directions for prevention and intervention. Acad Pediatr 16, S30–S36.

56.

Holz

, Boecker

, Hohm

, Zohsel

, Buchmann

, Blomeyer

, Jennen-Steinmetz

, Baumeister

, Hohmann

, Wolf

, Plichta

, Esser

, Schmidt

, Meyer-Lindenberg

, Banaschewski

, Brandeis

, Laucht

(2015) The long-term impact of early life poverty on orbitofrontal cortex volume in adulthood: Results from a prospective study over 25 years. Neuropsychopharmacology 40, 996–1004.

57.

Brito

, Noble

(2014) Socioeconomic status and structural brain development. Front Neurosci 8, 276.

58.

Ghanavati

, Salehinejad

, Nejati

, Nitsche

(2019) Differential role of prefrontal, temporal and parietal cortices in verbal and figural fluency: Implications for the supramodal contribution of executive functions. Sci Rep 9, 3700.

59.

Behrmann

, Geng

, Shomstein

(2004) Parietal cortex and attention. Curr Opin Neurobiol 14, 212–217.

60.

Curtis

(2006) Prefrontal and parietal contributions to spatial working memory. Neuroscience 139, 173–180.

61.

Dickerson

, Wolk

, Alzheimer’s Disease Neuroimaging Initiative (2012) MRI cortical thickness biomarker predicts AD-like CSF and cognitive decline in normal adults. Neurology 78, 84–90.

62.

Fotenos

, Mintun

, Snyder

, Morris

, Buckner

(2008) Brain volume decline in aging: Evidence for a relation between socioeconomic status, preclinical Alzheimer disease, and reserve. Arch Neurol 65, 113–120.

63.

Padurariu

, Ciobica

, Mavroudis

, Fotiou

, Baloyannis

(2012) Hippocampal neuronal loss in the CA1 and CA3 areas of Alzheimer’s disease patients. Psychiatr Danub 24, 152–158.

64.

, Gage

(2011) Adult hippocampal neurogenesis and its role in Alzheimer’s disease.. Mol Neurodegener 6, 85.

65.

Ferrari

, Neto de

G CC

, Nucci

, Mamani

, Lacerda

, Felício

, Amaro

, Gamarra

(2019) The accuracy of hippocampal volumetry and glucose metabolism for the diagnosis of patients with suspected Alzheimer’s disease, using automatic quantitative clinical tools. Medicine (Baltimore) 98, e17824.

66.

O’Brien

, Erkinjuntti

, Reisberg

, Roman

, Sawada

, Pantoni

, Bowler

, Ballard

, DeCarli

, Gorelick

, Rockwood

, Burns

, Gauthier

, DeKosky

(2003) Vascular cognitive impairment. Lancet Neurol 2, 89–98.

67.

Sierra

(2014) Essential hypertension, cerebral white matter pathology and ischemic stroke. Curr Med Chem 21, 2156–2164.

68.

Pantoni

(2010) Cerebral small vessel disease: From pathogenesis and clinical characteristics to therapeutic challenges. Lancet Neurol 9, 689–701.

69.

Liu

, Dong

Y-H

, Lyu

P-Y

, Chen

W-H

, Li

(2018) Hypertension-induced cerebral small vessel disease leading to cognitive impairment. Chin Med J (Engl) 131, 615–619.

70.

Alosco

, Gunstad

, Xu

, Clark

, Labbe

, Riskin-Jones

, Terrero

, Schwarz

, Walsh

, Poppas

, Cohen

, Sweet

(2014) The impact of hypertension on cerebral perfusion and cortical thickness in older adults. J Am Soc Hypertens 8, 561–570.

71.

Meissner

(2016) Hypertension and the brain:Arisk factor for more than heart disease. Cerebrovasc Dis 42, 255–262.

72.

Prins

, Scheltens

(2015) White matter hyperintensities, cognitive impairment and dementia: An update. Nat Rev Neurol 11, 157–165.

73.

Kearney-Schwartz

, Rossignol

, Bracard

, Felblinger

, Fay

, Boivin

J-M

, Lecompte

, Lacolley

, Benetos

, Zannad

(2009) Vascular structure and function is correlated to cognitive performance and white matter hyperintensities in older hypertensive patients with subjective memory complaints. Stroke 40, 1229–1236.

74.

Aeschbacher

, Blum

, Meyre

, Coslovsky

, Vischer

, Sinnecker

, Rodondi

, Beer

, Moschovitis

, Moutzouri

, Hunkeler

, Burkard

, Eken

, Roten

, Zuern

, Sticherling

, Wuerfel

, Bonati

, Conen

, Osswald

, Kühne

, Swiss-AF Investigators (2021) Blood pressure and brain lesions in patients with atrial fibrillation. Hypertension 77, 662–671.

75.

Wang

, Yang

, Wang

, Nie

, Yin

, Liu

(2020) Correlation between white matter hyperintensities related gray matter volume and cognition in cerebral small vessel disease. J Stroke Cerebrovasc Dis 29, 105275.

76.

Honningsvåg

L-M

, Håberg

, Hagen

, Kvistad

, Stovner

, Linde

(2018) White matter hyperintensities and headache: A population-based imaging study (HUNT MRI). Cephalalgia 38, 1927–1939.