Investigating the individual and joint effects of socioeconomic status and lifestyle factors on mild cognitive impairment in older Italians living independently in the community: results from the NutBrain study

Abstract

Objectives

Despite extensive research, a clear understanding of the role of the interaction between lifestyle and socioeconomic status (SES) on cognitive health is still lacking. We investigated the joint association of socioeconomic factors in early to midlife and lifestyle in later life and Mild Cognitive Impairment (MCI).

Design

Observational cross-sectional study.

Setting

NutBrain study in northern Italy.

Participants

773 community-dwelling adults aged 65 years and older (73.2 ± 6.0 SD, 58.6% females) participating in the NutBrain study (2019–2023).

Measurements

Three SES indicators (home ownership, educational level, occupation) and five lifestyle factors (adherence to Mediterranean diet, physical activity, smoking habits, social network, leisure activities) were selected. Each factor was scored and summed to calculate SES and healthy lifestyle scores; their joint effect was also examined. The association with MCI was assessed by logistic regression controlling for potential confounders. Sex-stratified analysis was performed.

Results

In total, 24% of the subjects had MCI. The multivariable logistic model showed that a high SES and a high lifestyle score were associated with 81.8% (OR0.182; 95%CI 0.095−0.351), and 44.1% (OR0.559; 95%CI 0.323−0.968) lower odds of having MCI, respectively. When examining the joint effect of SES and lifestyle factors, the cognitive benefits of a healthy lifestyle were most pronounced in participants with low SES. A healthier lifestyle score was found to be significantly associated with lower odds of MCI, only in females.

Conclusions

According to our findings, SES was positively associated with preserved cognitive function, highlighting the importance of active lifestyles in reducing socioeconomic health inequalities, particularly among those with a relatively low SES.

Trial registration

Trial registration number NCT04461951, date of registration July 7, 2020 (retrospectively registered, ClinicalTrials.gov).

1. Introduction

With the increasing life expectancy and the resulting ageing population, the global prevalence of dementia is expected to continually rise, leading to an increase in the burden of illness and healthcare costs [1]. Several factors are hypothesized to affect cognition during the physiological process of ageing, according to a multiple-factor framework [2]. However, although age is the strongest known risk factor for cognitive decline, aging does not necessarily mean the development of dementia and, at the same time, dementia is not an inevitable and natural consequence of ageing [3]. Similarly, dementia is also a complex and multifactorial disease that results from the interaction of genetic susceptibility and environmental factors throughout the life course [4]. It is well known that dementia onset is usually preceded by a long asymptomatic phase and then, as the disease progresses, by an early symptomatic pre-dementia phase [5], which represents the transitional phase between normal ageing and the pathological one, called Mild Cognitive Impairment (MCI) [6].

At present, although there are some new potentially disease-modifying drugs for dementia, their effect is still quite limited to treating symptoms [7], so early prevention still seems particularly important to counteract cognitive decline and/or slow the progression of dementia. Indeed, dementia care should focus not only on the continuum from healthy to pathological aging, but also on the entire life span, to identify which factors may play a crucial role in influencing the quality of ageing and the course of the disease [1,3].

These modifiable factors are mostly related to health status/pre-existing comorbidities (e.g., diabetes, hypertension, dyslipidaemia, obesity, cardiovascular disease, traumatic brain injury, frailty, depression and hearing loss) and lifestyle habits (e.g., diet, smoking habits, sleep disturbances, physical activity, leisure time activities) [4,[8], [9], [10]]. The literature suggests that a proper management of these factors, both in middle and late life, could prevent up to 40% of dementia cases [1]. Moreover, different lifestyle factors appear to share some common pathways, rather than specific mechanisms, in conferring protection against cognitive decline [11]. Therefore, adherence to multiple lifestyle recommendations has a significantly greater impact than implementing a single behaviour. Thus, indices that combine information on these key modifiable health and lifestyle factors should be used to assess and monitor risk status for developing dementia [[12], [13], [14]].

Further evidence indicating that dementia incidence may be modifiable also comes from repeated cohort studies comparing data between birth cohorts, which have observed a decline in dementia incidence in high-income countries, suggesting a role of certain social determinants dependent on the context in which people grows up and live in influencing health even in later life [[15], [16], [17]]. However, while the impact of socioeconomic inequalities has been well described for other diseases, including cancer and cardiovascular disease [18], it has been less explored for dementia. Many studies have suggested that socio-cultural and living conditions may affect cognition both indirectly, by influencing health and lifestyle changes [[19], [20], [21], [22]], and directly, by affecting the level of cognitive reserve and consequently, the level of tolerance to pathological changes in brain structure [23,24]. In addition, some studies have found that socioeconomic status (SES) may act as a modifier of the effect of lifestyle on other health outcomes such as mortality and self-reported health, raising concerns that the same healthy behaviour may have different effects on cognition depending on SES [25,26]. Also for social determinants, as well as for lifestyle factors, the use of composite measures that include multiple SES indicators acting across the life course is recommended [[27], [28], [29]].

Despite extensive research, a clear understanding of the role of lifestyle and SES on cognitive health and their interaction is still lacking [30]. Particularly, Wang [31] found no significant interaction. In contrast, other authors found a protective effect of lifestyle only in participants with low SES [32], especially with low income [33], while others [16] suggested that brain health lifestyle explains a large proportion of dementia risk in individuals with high versus low wealth. Moreover, in an attempt to clarify the role of each proxy of SES (such as education, occupation, income, wealth) over the life course on dementia risk in later life, Krasnova [29] did not find evidence for a critical period where SES exposure had particularly consequence, in contrast to other studies [34,35].

In light of these considerations, we hypothesised that a healthy lifestyle might somehow compensate for a disadvantaged SES. Therefore, in the current study, we aimed to examine the associations of a composite early-to-midlife socioeconomic score and a late-life lifestyle score with cognitive impairment in a community-dwelling Italian sample of older people. We also investigated whether the effect of healthy lifestyle behaviours on cognitive function varied by SES.

2. Methods

2.1. Study design, setting, and participants

Study participants were part of the NutBrain project, a population-based cohort study of community-dwelling older adults aged 65 years and older recruited in the Lombardy region (Italy), between October 2019 and January 2023. The study design and recruitment methods have been described previously [36]. Briefly, at the presentation point subjects were assessed for cognitive function and sociodemographic characteristics, lifestyle and clinical information, and anthropometric measurements were recorded by trained and certified study technicians according to harmonised protocols. A total of 807 participants were recruited; excluding those who did not complete the screening data collection (N = 8), those with psychiatric syndrome (N = 1), and those with dementia (N = 25), 773 subjects were included in the present analysis.

2.2. Outcome: cognitive function assessment

MCI was defined as the presence of subjective cognitive complaints and objective cognitive impairment on one or two neuropsychological tests; the impairment had to be greater than would be expected for the individual’s age and level of education, without impairment in activities of daily living, according to Albert’s criteria [37]. The neuropsychological profile was assessed using a comprehensive battery of neuropsychological tasks selected to assess global cognitive function and domain-specific functions as reported in Supplementary material.

2.3. Exposure: Early-to-midlife socioeconomic factors and late-life lifestyle factors

A set of socioeconomic and lifestyle factors was selected based on empirical evidence from previous epidemiological studies [1,4,29,33,38] and on the opinions of the multidisciplinary team of experts involved in the study, in particular epidemiologists, public health and nutrition researchers, neurologists, neuroscientists, and neuropsychologists.

Socioeconomic factors in early to middle life included:

• -Education level defined as primary school or less (0), middle school [1], high school or university (score = 2).
• -Occupation, grouped by level of complexity according to the longest job held: lower (0 = routine/manual labourer, housewife), middle (1 = craftsman or clerk), higher (2 = managerial/executive positions).
• -Homeownership, as a proxy for income [39], classified as not owning (0) or owning/family property [1].

We calculated the early to midlife socioeconomic score (SES) by summing the individual factors; the total score ranged from 0 to 5, with a higher score indicating a higher SES. SES was classified as low (≤3), moderate [4] and high (5+), corresponding to the first, second, and third tertiles of the distribution, respectively.

Late-life lifestyle factors included:

• -Physical activity, which was assessed using the International Physical Activity Questionnaire (IPAQ) [40]. Participants were classified as inactive if they did no physical activity (0), minimally active if they did only one activity such as slow walking, [1], moderately to vigorously active if they did more than one light activity or moderate activity such as slow swimming, dancing, or cycling to work, or vigorous activities such as hiking, skiing, fast swimming, running, etc. [2].
• -The frequency of engagement in leisure activities or hobbies over a lifetime (in years) was assessed using the Cognitive Reserve Index questionnaire (CRIq) [41]. Each leisure activity/hobby was first assigned a score of 0 (never engaged) or 1 (ever engaged) according to the frequency of engagement. An overall leisure score was calculated by summing each activity, ranging from 1 to 15, and then categorized into tertiles of 0 = low (≤8), 1 = moderate [9,10], and 2 = high engagement (11+).
• --Social network was derived by combining information on marital status (1 = married vs. 0 = single/unmarried/widower/divorced) and number of cohabitants (0 = none; 1 = one; 2 = two or more). A total score was created by summing the two variables and then divided into tertiles based on the distribution of the scores: 0 = very poor social network, 1 = poor social network, 2 = moderate social network, and 3 = rich social network.
• -Smoking habits were categorised as 0 = current smoker, 1 = former smoker and 2 = never smoker.
• -Dietary habits over the past year were collected using a 102-item semi-quantitative food frequency questionnaire (SFFQ) and administered by a certified dietician. To measure adherence to the Mediterranean diet, we used the score proposed by Trichopoulou et al. [42], which was categorized as 0 = low (0−3), 1 = moderate [[4], [5], [6]] and 2 = high [[7], [8], [9]].

We calculated the healthy lifestyle score (HLS) by summing up each factor with the total possible lifestyle score ranging from 0 to 11, with higher scores indicating high adherence. Lifestyle scores were categorized as low (≤4 healthy factors), moderate [5,6], and high (7+), corresponding to the first, second, and third tertiles, respectively.

2.4. Covariates

Covariates were selected based on theoretical knowledge [1,4,29,33,38] and empirical criteria (P-value<0.05 in univariate descriptive analysis), as they could potentially affect both exposures and outcomes. They included sex [43], age, waist circumference (as a proxy for overweight or obesity), total number of daily medications (as a proxy for comorbidities), depressive symptoms (defined using the Centre for Epidemiological Studies Depression Scale [44]; standard cut-off points of 16 or more) and daily energy intake in kcal derived from the Italian Food Composition Databases for Epidemiological Studies in Italy (http://www.bda-ieo.it/).

2.5. Statistical analysis

Characteristics of the sample were described using mean (standard deviation - SD) for continuous variables and frequency (%) for categorical variables. Student t-test and chi-squared test were used to compare the characteristics of participants with respect to cognitive status for continuous and categorical variables, respectively. We performed binary logistic regression models estimating odds ratios (OR) along with 95% confidence intervals (CI) to assess the independent associations of the SES score and the HLS with MCI. The first model was adjusted for age and sex and included SES and HLS separately. In the second model, SES and HLS were included together; and in the third model we adjusted for covariates. To assess whether the effect of a healthy lifestyle on MCI varied by SES, we examined the statistical interactions between the effects of HLS and SES score after adjusting for confounders. We then assessed the joint effect of SES and HLS by creating a dummy variable and classifying participants into nine groups based on three SES classes (low, moderate, high) and three HLS groups (low, moderate, high). The dummy variable was included in a third model with participants with low SES and low HLS as the reference group. We also examined our data for potential effect modification by sex on the association between SES, HLS and their joint effect and MCI by including interaction terms in the models and calculating stratum sex-specific estimates.

Analyses were performed with IBM SPSS Statistics for Windows version 25.0 (IBM Corp., Armonk, NY) and Stata version 15.0 (StataCorp LP, College Station, Texas, USA). A two-sided P-value < 0.05 was considered statistically significant.

3. Results

A total of 773 individuals were analysed (mean age 73.2 ± 6.0 SD, 58.6% females, 61.4% high school or university education, and 68.2% married). One hundred and eighty-six individuals (24%) met the criteria for MCI (46.2% multi-domain), while the remainder were classified as cognitively intact controls. Those with MCI were more likely to be older, less educated, manual workers/housewives, non-homeowners, unmarried or widow/er, living alone, former or never smokers, inactive or minimally active, less engaged in leisure activities over the life course, had a higher waist circumference, used more daily medications, and had a lower MMSE score than cognitively intact controls. Detailed data on participants’ characteristics are shown in Table 1.

Table 1.

Characteristics of participants by cognitive status in the total sample (N = 773).

		Cognitively intact (N = 587, 75.9%)		MCI (N = 186, 24.1%)		Total (N = 773)
		N	%	N	%	N	%
Sex	Males	239	40,7%	81	43,5%	320	41,4%
	Females	348	59,3%	105	56,5%	453	58,6%
Age at baseline*	65-69	241	41,1%	38	20,4%	279	36,1%
	70-74	154	26,2%	38	20,4%	192	24,8%
	75-79	117	19,9%	46	24,7%	163	21,1%
	80-84	56	9,5%	48	25,8%	104	13,5%
	85+	19	3,2%	16	8,6%	35	4,5%
Early-to-midlife socioeconomic status
Education*	Primary or less	46	7,8%	48	25,8%	94	12,2%
	Middle school	141	24,0%	63	33,9%	204	26,4%
	High school or university	400	68,1%	75	40,3%	475	61,4%
Occupation*	Routine/manual labourers/housewives	188	32,0%	111	59,7%	299	38,7%
	Intermediate professionals	254	43,3%	55	29,6%	309	40,0%
	High managerial/executive positions	145	24,7%	20	10,8%	165	21,3%
Homeownership*	No ownership	52	8,9%	26	14,0%	78	10,1%
	Ownership or family property	535	91,1%	160	86,0%	695	89,9%
Social network
Civil status*	Single	26	4,4%	12	6,5%	38	4,9%
	Married	415	70,7%	112	60,2%	527	68,2%
	Divorced/separated	49	8,3%	14	7,5%	63	8,2%
	Widow/er	97	16,5%	48	25,8%	145	18,8%
Living situation*	Alone	124	21,1%	55	29,6%	179	23,2%
	Lives with one other person	420	71,6%	119	64,0%	539	69,7%
	Lives with two or more other persons	43	7,3%	12	6,5%	55	7,1%
Lifestyle factors
Smoking habits*	Current smokers	61	10,4%	17	9,1%	78	10,1%
	Former smokers	255	43,4%	58	31,2%	313	40,5%
	Never smokers	271	46,2%	111	59,7%	382	49,4%
Physical activity*	Inactive or minimally active	438	74,6%	155	83,3%	593	76,7%
	Moderately active	92	15,7%	26	14,0%	118	15,3%
	Highly active	57	9,7%	5	2,7%	62	8,0%
Leisure activities*	Low engagement	192	32,7%	111	59,7%	303	39,2%
	Moderate engagement	189	32,2%	41	22,0%	230	29,8%
	High engagement	206	35,1%	34	18,3%	240	31,0%
Adherence to Mediterranean diet	Low (0−3)	193	32,9	62	33,3	255	33,0
	Moderate (4−5)	250	42,6	82	44,1	332	42,9
	High (6−9)	144	24,5	42	22,6	186	24,1
Energy intake (Kcal)		1982,38	477,85	2001,79	530,60	1987,05	490,77
Health-related conditions
Waist circumference (cm)*		92,85	12,72	95,08	13,17	93,38	12,86
Daily number of drugs*		2,97	2,40	3,69	2,64	3,14	2,48
Depressive symptoms		9,43	8,31	10,01	8,82	9,57	8,44
Mini-mental State Examination*		27,59	1,95	26,32	2,00	27,29	2,04

*P-value ≤0.05.

Table 2 shows the distribution of SES and HLS scores by cognitive status. Compared with cognitively intact controls, people with MCI had lower SES scores (45.1% vs. 72%) and lower HLS (35.4% vs. 48.9%). The distribution of the joint exposure of SES and HLS by cognitive status is shown in Fig. 1. Compared with cognitively intact controls, individuals with MCI were more likely to have low SES combined with a low (41.4% vs. 19.4%) and moderate HLS (25.3% vs. 16.5%). In the logistic model adjusted for age and sex (model 1, Table 3), the highest SES was statistically significantly associated with an 82.1% reduction in the odds of MCI (OR 0,179; 95%CI 0,094−0,341) and the highest HLS with a 56.4% reduction (OR 0,436; 95%CI 0,277−0,741). When SES and HLS were simultaneously included in the model (model 2), the highest SES was statistically significantly associated with an 81.3% reduction in the odds of MCI (OR 0,187; 95%CI 0,098−0,359) and the highest HLS with a 46.7% reduction (OR 0,533; 95%CI 0,309−0,918). In the multivariable logistic model (model 3), the odds of having MCI was reduced by 81.8% with the highest SES (OR 0,182; 95%CI 0,095−0,351) and by 44.1% with the highest HLS (OR 0,559; 95%CI 0,323−0,968). Fig. 2 shows that compared to low SES together with low HLS, low SES and high HLS, moderate SES and low HLS, moderate SES and moderate HLS, moderate SES and high HLS, high SES and low HLS, high SES and moderate HLS, and high SES and high HLS were associated with a statistically significant reduction in the odds of having MCI, controlling for covariates. Compared to those with high SES scores, those with low SES scores had a poorer social network, were more likely to be physically inactive and engaged less in leisure activities (Supplementary Table S1). Individuals with a favourable lifestyle were more likely to be highly educated, in high managerial positions and homeowners than those with an unfavourable lifestyle (Supplementary Table S2). As we found a statistically significant interaction between sex and SES, HLS and their combined exposure (P for interaction ≤0.05), sex-stratified analyses were performed. The highest HLS was statistically significantly associated with reduced odds of MCI in females (OR 0,349, 95%CI 0,145−0,840) but not in males (OR 0,805, 95%CI 0,381−1,697). In addition, moderate SES together with high HLS was significantly inversely associated with cognitive impairment only in females (OR 0,079, 95%CI 0,010−0,630) (Supplementary Table S3).

Table 2.

Distribution of socioeconomic status and lifestyle score and their joint effect by cognitive status in the total sample (N = 773).

		Cognitively intact		MCI		Totale
		N	%	N	%	N	%
Socioeconomic status*	Low	265	45,1%	134	72,0%	399	51,6%
	Moderate	188	32,0%	40	21,5%	228	29,5%
	High	134	22,8%	12	6,5%	146	18,9%
Lifestyle score*	Low	208	35,4%	91	48,9%	299	38,7%
	Moderate	239	40,7%	73	39,2%	312	40,4%
	High	140	23,9%	22	11,8%	162	21,0%
Joint exposure of SES with lifestyle factors*
Low SES and low lifestyle score		114	19,4%	77	41,4%	191	24,7%
Low SES and moderate lifestyle score		97	16,5%	47	25,3%	144	18,6%
Low SES and high lifestyle score		54	9,2%	10	5,4%	64	8,3%
Moderate SES and low lifestyle score		59	10,1%	11	5,9%	70	9,1%
Moderate SES and moderate lifestyle score		81	13,8%	19	10,2%	100	12,9%
Moderate SES and high lifestyle score		48	8,2%	10	5,4%	58	7,5%
High SES and low lifestyle score		35	6,0%	3	1,6%	38	4,9%
High SES and moderate lifestyle score		61	10,4%	7	3,8%	68	8,8%
High SES and high lifestyle score		38	6,5%	2	1,1%	40	5,2%

*P-value ≤0.05.

Fig. 1.

Percentages have been calculated within cognitive status. For example, the last bar shows that 41.4% of people with MCI were more likely to have a low SES combined with a low lifestyle score, compared with cognitively intact people (19.4%).

Table 3.

Association of early to midlife socioeconomic status and late life lifestyle factors with mild cognitive impairment in the total sample (N = 773).

		Model 1		Model 2		Model 3
		OR	95%CI	OR	95%CI	OR	95%CI
Socioeconomic status	Low	1	–	1	–	1	–
	Moderate	0,464	0,307−0,702	0,486	0,320−0,740	0,476	0,311−0,726
	High	0,179	0,094−0,341	0,187	0,098−0,359	0,182	0,095−0,351
P for trend		<0,001		<0,001		<0,001
Lifestyle score	Low	1	–	1	–	1	–
	Moderate	0,791	0,545−1,148	0,956	0,649−1,408	0,988	0,668−1,460
	High	0,436	0,277−0,741	0,533	0,309−0,918	0,559	0,323−0,968
P for trend		0,003		0,047		0.076

Model 1 includes age and sex and SES and HLS separately.

Model 2 includes age, sex, and SES and HLS simultaneously.

Model 3 includes age, sex, number of drugs, Kcal, depressive symptoms, waist circumference and SES and HLS simultaneously.

Fig. 2.

Low SES and low lifestyle score as the reference category.

The model includes age, sex, number of drugs, Kcal, depressive symptoms, and waist circumference.

P for trend < 0,001.

4. Discussion

This study investigates the individual and joint effects of SES and lifestyle factors on mild cognitive impairment in a sample of community-dwelling dementia-free adults aged 65 years and older.

Consistent with prevalence in the general population [45], 24% of our sample met the criteria for MCI. They had socioeconomic, lifestyle and health conditions that were more often disadvantageous than intact controls, both when these variables were considered individually and when combined into multidimensional constructs of SES and HLS. Indeed, the literature suggests that the conditions in which individuals are born, grow, live and age also affect their health status, providing evidence of the influence of lifestyle choices, social determinants and social hierarchy [30,38,46,47]. The distribution of SES and HLS and their joint effect by cognitive status also showed that individuals with MCI were more likely to have low SES combined with low/moderate HLS. These findings are in line with previous studies showing that both healthy lifestyle behaviours [[48], [49], [50], [51], [52]] and favourable socioeconomic conditions [[53], [54], [55], [56]] are associated with cognitive function and represent protective factors against cognitive decline and dementia. Indeed, a healthy lifestyle can directly influence cardiovascular risk factors and oxidative stress by promoting neuroprotective factors that trigger neurogenesis and consequently increase cognitive reserve [57,58]. Similarly, SES has also been associated with functional and structural neural differences in a wide range of cortical areas [59]. Therefore, high SES during the life course can also affect cognition both indirectly by influencing the quality of lifestyle and health [20] and directly by affecting cognitive reserve in relation to educational level and occupational complexity. Indeed, research on the modifiable determinants of cognitive reserve has provided consistent evidence, particularly for the role of educational attainment (as SES in young adulthood) and occupational complexity (as SES in midlife) in exerting a protective effect on overall cognition in the face of multiple measures of brain burden [23,29]; therefore, SES level at different life stages may contribute differently to shaping cognitive reserve and preventing cognitive impairment [33,53].

Moreover, according to previous studies [31,33], our results also suggest that SES and HLS were independently inversely associated with cognitive status in our sample, supporting evidence about their role in reducing the risk of developing cognitive impairment. However, we found that SES had a stronger role than HLS (81.8% reduction in the odds of MCI for high SES vs. 44.1% for high HLS) even after including both scores simultaneously in the model. Therefore, our result differ from some previous studies supporting the hypothesis that the effect of SES on health is mediated and explained by the different lifestyle that results from it [20,22]. Rather, our data outlined a direct effect of socioeconomic status on cognitive function. A possible explanation for this finding could be that the negative impact of an unhealthy lifestyle may be counterbalanced by high SES, which as a proxy for better cognitive reserve may compensate for neuropathological changes associated with ageing and cognitive decline [23].

We also observed that the cognitive benefits of a healthy lifestyle were only evident among participants with low SES, and this compensatory effect diminished as SES improved, similar to what has been observed in some other studies [[31], [32], [33]]. This could be explained by the fact that the cumulative disadvantages over the life course of individuals low SES (e.g. psychological stress and limited medical resources) may make them more vulnerable to the adverse cognitive effects of unhealthy behaviours (such as poor diet, physical inactivity, smoking, alcohol consumption, etc.), and at the same time more sensitive to the positive effects of healthy behaviours [29,32,33]. In other words, if SES is considered as a proxy for cognitive reserve and thus as an indicator of resilience to the effects of ageing and disease, high SES could nullify/mitigate the negative impact of unhealthy lifestyles [23].

Finally, we found a significant sex-lifestyle interaction in the association with cognitive function. Our results suggest that a healthier lifestyle score was significantly associated with cognitive function in females but not in males consistent with previous studies reporting that females sex may enhance the protective effect of healthy lifestyle behaviours on cognition [60].

In conclusion, despite the abundance of evidence for the existence of a close relationship between lifestyle and SES, how these factors interact and their overall effect on brain health is still under debate. These heterogeneous results may be because previous studies have focused on the effect of single lifestyle and SES variables on cognition (e.g. 32), rather than on composite measures that combine multiple factors acting synergistically over the life course. Overall, our analyses suggest some important implications: (i) SES plays an important role in the maintenance of cognitive function; (ii) a high lifestyle, even in late life, can compensate for lower SES in early to midlife and provide a reasonable degree of protection against cognitive impairment; (iii) individuals with high SES are equally likely to have MCI regardless of lifestyle.

Studies investigating the specific mechanisms underlying the relationship between lifestyle and cognitive impairment using a life‐course approach should be implemented in the near future.

4.1. Limitations and strengths

A major limitation of the study is the cross-sectional design, which precludes causal inference for the association between SES, lifestyle factors, and cognitive outcomes. In particular, the study design does not allow us to determine whether the behaviours preceded or followed cognitive impairment. In addition, the reliability of data collected from people with cognitive impairment may be compromised. This can lead to recall bias, and possible misclassification of exposure may have affected the results, leading to either over- or underestimation of the relationship between behaviours and cognitive function. Although we excluded participants with moderate to severe dementia, we cannot rule out the possibility that cognitive impairment may have influenced behaviours. Further longitudinal prospective studies, including cognitively intact participants at baseline, are needed to confirm and strengthen the association between healthy lifestyle scores and MCI in older people. In addition, the high proportion of highly educated individuals in the enrolled sample relative to the general population leads to sample self-selection, limiting the generalisability of our results to other cohorts of older adults. However, this selection bias is likely to bias our results towards the null, underestimating the observed associations. Furthermore, because the educational level of our population was higher than average, neuropsychological tests may also suffer from a ceiling effect, although we tried to increase the discriminatory power by using 11 cognitive tests to assess performance in different cognitive domains. We also recognised that reducing SES and lifestyle factors to a single score may not reflect the effect of lifelong exposure on ageing. Finally, although we controlled for several potential confounders, we cannot completely rule out the possibility of residual confounding due to unmeasured factors (e.g., APOE genotype).

The current work also has several strengths. The study provides data from a large community-based sample, which is more representative of the general population than a clinical sample of older people. The study is also characterized by a comprehensive assessment of participant's cognitive function and clinical and behavioural factors, using instruments that have been validated and widely used in previous epidemiological studies. Furthermore, lifestyle factors and SES were assessed exhaustively, taking into account their mutual components, as suggested in the literature. Another strength relates to the fact that we used a comprehensive neuropsychological assessment (according to diagnostic guidelines) to identify the possible presence of a MCI profile, rather than a single screening tool (such as the MMSE), to avoid the risk of false negatives and to strengthen our results. In addition, the final diagnosis of MCI was made by a multidisciplinary consensus expert team based on the uniform application of widely accepted criteria, which increases the sensitivity of the diagnosis. Finally, this study examined easily measurable socioeconomic and lifestyle factors that are largely assessable and therefore replicable in other studies.

5. Conclusions

Our study suggests that SES plays an important role in the maintenance of cognitive function and also highlights the importance of healthy lifestyles in reducing socioeconomic health inequalities, particularly among the economically disadvantaged individuals. This prospect is certainly encouraging and is likely to have both individual and public health implications. Indeed, given that lifestyle is not a random individual behaviour but a pattern of aggregated routines based on choices among a range of options, responsibility for unhealthy behaviours should be attributed not only to individual context and responsibility but also to public health interventions. Therefore, targeted health campaigns aimed at effectively and equitably reducing the risk of cognitive decline are desirable. In the future, we recommend designing additional longitudinal and intervention studies to explore the mechanisms underlying the interactions between SES, lifestyle factors and cognition. This will provide further evidence to support recommendations for reducing health inequalities.

Funding statement

The NutBrain study is supported by the Ministero della Salute (Bando di Ricerca Finalizzata Giovani Ricercatori 2016, GR -2016-02361730). The study funders had no role in the design and conduct of the study, in the data collection, analysis, and interpretation; in the preparation of the manuscript; or in the review or approval of the manuscript.

Data availability

Data will be made available on request.

CRediT authorship contribution statement

Sara Bernini – conceptualization, methodology, data collection, writing original draft, review & editing.

Elena Perdixi – conceptualization, data collection, review & editing.

Silvia Conti - conceptualization, data collection, review & editing.

Nithiya Jesuthasan - conceptualization, data collection, review & editing.

Alfredo Costa – conceptualization, data collection, review & editing.

Marco Severgnini – conceptualization, methodology, review & editing.

Matteo Cotta Ramusino – conceptualization, data collection, review & editing.

Federica Prinelli – project administration, conceptualization, methodology, data curation, formal analysis, writing original draft, review & editing, supervision.

All authors saw and approved the final version and no other person made a substantial contribution to the paper.

Conflict of interest

The authors report no conflict of interest.

Ethical standard

All participants provided formal written informed consent to participate in the original study, and for those who were found to be unable to give full written informed consent, a caregiver or guardian was identified and their advice was sought regarding participation. The study protocol and amendments were reviewed and approved by the Medical Ethics Committee of Pavia, Italy. The study was designed and conducted in accordance with the tenets of the Declaration of Helsinki, as revised in 2013. Data were handled and stored in accordance with the European Union General Data Protection Regulation (EU GDPR) 2016/679, and data transmission was protected by encryption/decryption and password protection.

Appendix A. Supplementary data

The following is Supplementary data to this article: