Association of magnesium depletion score with cognitive function in older adults: An analysis of US National Health and Nutrition Examination Survey (NHANES) 2011 to 2014

Abstract

This study aims to explore the association between the magnesium depletion score (MDS), a newly developed indicator of magnesium levels, and cognitive function in older adults residing in the United States. We analyzed data from 768 participants aged ≥ 60 years in the 2011 to 2014 National Health and Nutrition Examination Survey. Participants were stratified by MDS levels: none-to-low (0–1), moderate (2), and high (3–5). Cognitive function was assessed through digit symbol substitution test (DSST), animal fluency test (AFT), and Consortium to Establish a Registry for Alzheimer’s Disease Word Learning subtest. Using National Health and Nutrition Examination Survey mobile examination center weights, sample-weighted multivariable linear regression models calculated β coefficients (95% confidence intervals) for MDS–cognition associations, adjusting for age, sex, race, education, income, body mass index, smoking, and hypertension. In fully-adjusted models, high MDS group showed significantly lower cognitive scores versus none-to-low group: DSST: β = −4.91 (−7.73, −2.08), P = .0007, AFT: β = −2.09 (−3.25, −0.93), P = .0004. No significant association with Consortium to Establish a Registry for Alzheimer Disease scores (β = −0.96 [−2.25, 0.33], P = .1445). Stratified analyses revealed stronger associations in obese (body mass index > 30 kg/m²) and current smoking subgroups. Participants meeting magnesium recommended dietary allowance showed attenuated cognitive risks in DSST (interaction P = .022) and AFT (interaction P = .015). Higher MDS is independently associated with poorer processing speed and executive function in older Americans, particularly among obese individuals and smokers. Achieving dietary magnesium recommended dietary allowance may mitigate these low cognitive function. In our cross-sectional study, MDS was found to be associated with cognitive vulnerability.

1. Introduction

The aging population in the United States is growing rapidly By 2060, the U.S. population aged ≥ 65 years is projected to grow from 52 million (16% of the total population) in 2018 to 95 million (23%), reflecting a significant demographic shift.^[1] The World Alzheimer Report 2022 projects approximately 152 million global dementia cases by 2050, driven by population aging. Identifying modifiable risk factors is essential to delay cognitive impairment onset and mitigate dementia progression.^[2] Cognitive decline is viewed as an important predictor of dementia. Hence, it is crucial to investigate the controllable factors which may conduce to the early decline.^[3] Aging, socioeconomic status, physical activity, genetics, and nutrition are established risk factors for cognitive decline.^[3] Magnesium, the second most prevalent intracellular cation, is essential for energy metabolism, protein synthesis, and nucleic acid regulation,^[4–8] maintaining muscle and nerve membrane integrity, as well as facilitating neurochemical and synaptic transmission.^[9] Low magnesium consumption has been linked to increased risks of metabolic syndrome, type 2 diabetes mellitus, cardiovascular disorders,^[4] and cognitive decline.^[10] Prior observational studies indicate that magnesium deficiency correlates with cognitive impairment and dementia risk.^[5–9] While the magnesium tolerance test (MTT) remains the gold standard for assessing magnesium status, its clinical utility is constrained by requiring pre- and post-infusion 24-hour urine collections following intravenous administration, creating procedural complexity.^[11–13] Fan et al recently introduced the magnesium depletion score (MDS), a novel assessment tool whose efficacy in detecting magnesium deficiency was confirmed using the MTT.^[14] The MDS calculation integrates 4 key factors: diuretic use, proton-pump inhibitor (PPI) use, renal function decline, and alcohol consumption. Its diagnostic accuracy, as measured by the ROC curve, surpasses that of serum and urine magnesium levels. Magnesium deficiency contributes to cognitive decline, especially in hippocampal-dependent memory processes, via neuroinflammation and oxidative stress pathways. By modulating synaptic plasticity and neurotransmitter activity, magnesium deficiency exacerbates cognitive dysfunction when these mechanisms are impaired.^[15,16] Given the established connections between magnesium deficiency and disruptions in synaptic plasticity, neuroinflammation, and oxidative stress, we propose that elevated MDS correlate with diminished cognitive performance. Notably, prior research has not examined this specific relationship. Using data from the 2011 to 2014 National Health and Nutrition Examination Survey (NHANES), we assessed the link between MDS and cognitive impairment in older adults.

2. Methods

2.1. Study population

This study utilized data from the 2011 to 2014 NHANES cycles, which included cognitive assessments for older adults. As a nationally representative survey of noninstitutionalized U.S. residents, NHANES was designed to assess nutritional and health status. Detailed methodology has been published elsewhere.^[17] This study is overseen by the National Center for Health Statistics, a division of the Centers for Disease Control and Prevention. Participants aged 60 years or older provided serum creatinine data, completed 24-hour dietary recalls, and underwent cognitive assessments (Consortium to Establish a Registry for Alzheimer’s Disease [CERAD], animal fluency test [AFT], digit symbol substitution test [DSST]), as detailed in prior publications,^[18] completed alcohol use consumption questionnaire and completed prescription medications use consumption questionnaire were involved in the research populace. Participators with missing data for those information or possible confounders were removed from the analyses. Hence, 768 subjects in total were involved in this survey. The recruitment process is illustrated in Figure 1.

Figure 1.

Participant selection flowchart. Note: The number of participants excluded due to missing data: 16,944 participants missing cognitive tests; 2054 participants missing MDS data, including 761 participants missing serum creatinine, 427 participants missing alcohol use consumption questionnaire, 691 participants missing prescription medications use consumption questionnaire, 175 participants missing 24-h dietary recall, and 165 participants missing covariates. MDS = magnesium depletion score, NHANES = National Health and Nutrition Examination Survey.

Written informed consent was obtained from all participants, and the survey protocol received approval from the Institutional Review Board of the NCHS,#2011-17 for the 2011 to 2012 cycle and its continuation for the 2013 to 2014 cycle.

3. Assessment of magnesium depletion score and dietary magnesium intake

Recently, the MDS as a novel approach is put forward to evaluate individuals’ total magnesium status.^[14] Unlike the MTT, the MDS is easily calculated and it could be extensively applied to scientific and clinical research. Furthermore,the MDS is derived from 4 key factors:

1. Current diuretic use, identified by self-report within the previous 30 days, was given a score of one point.

2. A score of one point was given for self-reported PPI consumption in the preceding 30 days.

3. Renal function was assessed using the chronic kidney disease (CKD)-EPI equation, with scoring as follows: one point for moderate impairment (eGFR 60–89 mL/min/1.73 m², CKD-EPI stage G2) and 2 points for chronic kidney disease (eGFR < 60 mL/min/1.73 m², stages G3a-G5). Normal function was defined as eGFR ≥ 90 mL/min/1.73 m².^[19]

4. One point was assigned to heavy drinkers, defined as consuming more than 2 daily drinks for men or exceeding one drink per day for women, per the 2015 to 2020 Dietary Guidelines for Americans.^[20] Participants were stratified into 3 groups based on their MDS: low (MDS 0–1), moderate (MDS 2), and high (MDS 3–5). NHANES collected dietary data through two 24-hour recalls, the first conducted in person at the mobile examination center and the second via telephone 3 to 10 days later. For each participant, the mean magnesium intake over these 2 days was used to estimate dietary magnesium consumption.Participants were further stratified based on daily magnesium intake, assessed against the estimated average requirement (EAR) and recommended dietary allowance (RDA). For individuals aged ≥ 60 years, the EAR thresholds were set at 350 mg/day (males) and 265 mg/day (females), with corresponding RDA values of 420 mg/day (males) and 320 mg/day (females). Magnesium consumption levels were categorized into 3 groups: intake below the EAR (<EAR); meeting or exceeding the EAR but below the RDA (≥EAR but < RDA); and intake at or above the RDA (≥RDA).^[21]

4. Assessment of cognitive function

During the 2011 to 2014 NHANES cycles, cognitive performance was assessed using 4 standardized modules: the CERAD Word Learning (CERAD-WL) and Delayed Recall (CERAD-DR) tests,^[22] the AFT,^[23] and the DSST.^[24]

The CERAD-WL and CERAD-DR evaluated episodic and declarative memory through a 3-trial learning task and a delayed recall test. Participants were instructed to memorize and recall 10 words per trial, with scores ranging from 0 to 10. For the AFT, categorical verbal fluency was measured by asking participants to list as many animals as possible within a 60-second interval, awarding one point per correct response. The DSST, adapted from the Wechsler Adult Intelligence Scale, quantified processing speed, sustained attention, and working memory. Participants had 2 minutes to pair symbols with numbers across 133 boxes, scoring one point per correct match (maximum score: 133).

5. Covariates

The analysis adjusted for potential confounders, including demographic characteristics (age, sex, race), socioeconomic indicators (education level, family income-poverty ratio), and health-related behaviors (body mass index [BMI], smoking, alcohol consumption, and physical activity). Education was categorized into 5 levels: less than 9th grade, 9th to 11th grade, high school graduate or GED equivalent, some college or an associate degree, and college graduate or higher. BMI groupings comprised < 25 (normal/underweight), 25 to 30 (overweight), and > 30 (obese). Smoking status was determined based on a lifetime consumption of ≥ 100 cigarettes. Alcohol use was defined as ≥ 12 drinks per year. For physical activity, participants were classified as meeting the threshold if they reported ≥ 150 minutes of moderate or vigorous work-related exercise weekly.

6. Statistical analysis

To account for the complex survey design, we followed NHANES analytic guidelines by incorporating appropriate sample weights, specifically using the mobile examination center weights for the 2011 to 2014 cycles. Data for each cycle were collected through a single interview. Baseline characteristics of the study population were summarized as weighted means ± standard deviation for continuous variables and weighted proportions for categorical variables.To assess the relationship between MDS and cognitive test scores, weighted linear regression models generated β coefficients with 95% confidence intervals (CIs).The initial model included no covariates (crude model). Model 1 adjusted for sex, age, and race, while Model 2 further adjusted for BMI, education level, family income-poverty ratio, physical activity, smoking status, dietary magnesium intake, and alcohol consumption. Subgroup analyses were conducted by smoking status, gender, BMI, and dietary magnesium intake. Statistical analyses were performed using IBM SPSS (version 24.0) and R (version 4.2.2), with a 2-sided P-value < .05 considered statistically significant.

7. Results

7.1. Characteristics of study population

Our study included 768 older adults from the United States, with a mean age of 68.47 ± 6.43 years. The cohort comprised 51.81% males and 83.72% non-Hispanic White individuals. Participants were stratified into 3 groups based on MDS: 392 in the none-to-low group, 255 in the middle group, and 121 in the high group. Comparative analysis revealed significant differences across MDS groups (all P < .001). The high MDS group exhibited older age, higher proportions of smokers and alcohol consumers, and elevated BMI levels compared to the none-to-low group. Additionally, dietary magnesium intake was significantly lower in the high MDS group. Cognitive performance, assessed via the DSST, AFT, and CERAD tests, demonstrated lower scores in the high MDS group (DSST: 48.92 ± 16.01; AFT: 16.14 ± 5.22; CERAD: 25.36 ± 6.94) relative to the none-to-low group (DSST: 55.64 ± 16.08; AFT: 19.04 ± 5.92; CERAD: 26.71 ± 6.20). Weighted baseline characteristics stratified by MDS levels are detailed in Table 1.

Table 1.

Participant characteristics by cognitive MDS, NHANES 2011 to 2014 (n = 768).

Characteristics	Overall (n = 768)	MDS (n = 392) None to no	MDS (n = 255) Middle	MDS (n = 121) High	P-value
Age (yr)‡	68.47 ± 6.43	67.61 ± 6.13	69.02 ± 6.53	70.37 ± 6.74	<.0001**
Sex, male (n [%])†	51.81	53.42	51.88	45.53	.3505
PIR‡	3.35 ± 1.50	3.53 ± 1.47	3.27 ± 1.50	2.88 ± 1.48	.0002**
BMI (kg/m2)‡	29.0 ± 6.1	27.84 ± 5.20	29.25 ± 6.42	32.45 ± 6.77	<.0001**
<25 (n [%])†	26.00	30.44	24.73	12.25
≥25, <30 (n [%])†	37.15	39.36	36.83	29.57
≥30 (n [%])†	36.85	30.20	38.43	58.18
Dietary magnesium (mg)‡	300.0 ± 134.0	339.68 ± 156.08	287.36 ± 119.15	275.56 ± 95.27	<.0001**
<EAR (n [%])†	28.75	46.83	61.28	59.40
EAR-RDA (n [%])†	33.12	19.94	12.35	25.68
>RDA (n [%])†	38.14	33.23	26.37	14.92
DSST score‡	55.64 ± 16.08	58.44 ± 15.58	54.05 ± 15.84	48.92 ± 16.01	<.0001**
AFT score‡	19.04 ± 5.92	19.99 ± 6.03	18.76 ± 5.58	16.14 ± 5.22	<.0001**
CERAD score‡	26.71 ± 6.20	27.27 ± 6.01	26.40 ± 6.08	25.36 ± 6.94	.0111**
Race (n [%])†					.0971
Mexican American	3.27	4.61	1.63	2.14
Other Hispanic	3.15	3.53	2.21	3.98
Non-Hispanic White	83.72	83.77	84.84	80.76
Non-Hispanic Black	6.69	4.58	8.11	11.22
Other race including multi-racial	3.18	3.51	3.20	1.90
Education (n [%])†					.1966
Less than 9th grade	3.47	3.20	3.36	4.81
9–11th grade	7.40	6.23	8.21	9.84
High school graduate or equivalent	20.38	18.75	21.52	23.79
Some college or AA degree	30.31	30.21	27.77	36.86
College graduate or above	38.41	41.60	39.14	24.69
Smoking status (n [%])†					.0086**
Never	46.90	52.93	41.65	36.82
Fomer	10.30	8.44	12.30	12.44
Now	42.80	38.62	46.05	50.74
Drinking status (n [%])†					.0438*
Yes	88.67	86.13	90.54	93.78
No	11.33	13.87	9.46	6.22
Physical activity (n [%])†					.0699
Yes	9.82	12.04	8.17	5.45
No	90.18	87.96	91.83	94.55

AFT = animal fluency test, BMI = body mass index, CERAD = Consortium to Establish a Registry for Alzheimer’s Disease, DSST = digit symbol substitution test, EAR = estimated average requirement, MDS = magnesium depletion score, NHANES = National Health and Nutrition Examination Survey, PIR = poverty income ratio, RDA = recommended dietary allowance.

^†Unweighted frequency counts and weighted percentages shown.

^‡Values shown are median and (standard deviation).

^*P < .05.

^**P < .01.

8. Association between the MDS and cognition function

After full adjustment for potential confounders, higher MDS levels were significantly associated with lower performance on both the DSST test (β = −4.91, 95% CI: −7.73 to −2.08, P < .01) and the AFT test (β = −2.09, 95% CI: −3.25 to −0.93, P < .01). However, no significant association was observed between MDS and CERAD test scores (Table 2).

Table 2.

Linear and logistic regression models for associations of MDS with the score of DSST, AFT, and CERAD in NHANES 2011 to 2014.

	DSST score			AFT score			CERAD score
	Crude	Model 1	Model 2	Crude	Model 1	Model 2	Crude	Model 1	Model 2
MDS
None to no	Reference			Reference			Reference
Midle	−4.39 (−6.85, −1.93), .0005**	−3.52 (−5.64, −1.39), .0012**	−2.48 (−4.47, −0.49), .0149*	−1.23 (−2.13, −0.33), .0078**	−0.89 (−1.73, −0.04), .0406*	−0.39 (−1.21, 0.42), .3452	−0.87 (−1.84, 0.09), .0761	−0.57 (−1.47, 0.32), .2098	−0.36 (−1.27, 0.54), .4330
High	−9.52 (−12.89, −6.15), <.0001**	−7.07 (−10.00, −4.14), <.0001**	−4.91 (−7.73, −2.08), .0007**	−3.86 (−5.09, −2.62), <.0001**	−2.99 (−4.16, −1.83), <.0001**	−2.09 (−3.25, −0.93), .0004**	−1.91 (−3.23, −0.59), .0047	−1.28 (−2.52, −0.05), .0419	−0.96 (−2.25, 0.33), .1435

Crude model adjust for: none. Model 1 adjust for: gender, age, and race. Model 2 adjust for: gender, age, race, dietary magnesium, smoking status, drinking status, physical activity, education, BMI, and PIR.

AFT = animal fluency test, BMI = body mass index, CERAD = Consortium to Establish a Registry for Alzheimer’s Disease, DSST = digit symbol substitution test, MDS = magnesium depletion score, NHANES = National Health and Nutrition Examination Survey.

^*P < .05.

^**P < .01.

9. Subgroup analyses

Subgroup analyses stratified by gender, dietary magnesium intake, BMI, and smoking status revealed differential associations with cognitive performance. For the DSST test, significant negative associations were observed among participants with BMI > 30 (β = −6.22, 95% CI: −10.23 to −2.21; P < .01) and current smokers (β = −4.76, 95% CI: −8.51 to −1.01; P < .05). Similarly, those with dietary magnesium intake below the RDA showed poorer cognitive scores (P < .05), while no significant association was found in those meeting RDA requirements (β = −3.20, 95% CI: −12.05 to 5.65; P > .05). No gender-based differences were detected. These patterns were consistent in AFT test analyses (Table 3).

Table 3.

Linear associations of MDS with cognitive function stratifed by selected factors, NHANES, 2011 to 2014.

	DSST test	AFT test
Gender
Male
None to no	Reference	Reference
Middle	−2.38 (−4.89, 0.13), .0633	0.06 (−1.02, 1.13), .9175
High	−4.44 (−8.21, −0.67), .0214*	−2.63 (−4.24, −1.02), .0015**
Female
None to no	Reference	Reference
Middle	−2.34 (−5.79, 1.11), .1853	−0.89 (−2.15, 0.37), .1688
High	−5.64 (−10.32, −0.96), .0187*	−1.92 (−3.64, −0.19), .0300*
BMI
<25
None to no	Reference	Reference
Middle	−2.19 (−6.32, 1.95), .3010	0.01 (−1.51, 1.53), .9914
High	−6.78 (−14.67, 1.11), .0940	−1.77 (−4.71, 1.16), .2385
25–30
None to no	Reference	Reference
Middle	0.56 (−2.84, 3.96), .7485	−1.20 (−2.58, 0.17), .0879
High	−4.11 (−9.42, 1.20), .1307	−1.78 (−3.94, 0.39), .1085
>30
None to no	Reference	Reference
Middle	−5.60 (−9.07, −2.12), .0018**	0.09 (−1.30, 1.49), .8953
High	−6.22 (−10.23, −2.21), .0026**	−2.16 (−3.78, −0.55), .0092**
Smoking
Never
None to no	Reference	Reference
Middle	−2.35 (−5.70, 1.00), .1694	−0.49 (−1.87, 0.88), .4822
High	−5.56 (−10.95, 5.68), .5357	−5.21 (−11.04, 0.61), .0808
Former
None to no	Reference	Reference
Middle	−0.79 (−6.49, 4.90), .7858	2.09 (−0.10, 4.29), .0645
High	−6.78 (−15.10, 1.54), .1134	−0.56 (−3.79, 2.67), .7337
Now
None to no	Reference	Reference
Middle	−3.13 (−5.94, −0.32), .0297*	−0.42 (−1.57, 0.73), .4709
High	−4.76 (−8.51, −1.01), .0134*	−1.93 (−3.47, − 0.40), .0141*
Magnesium intake
<EAR
None to no	Reference	Reference
Middle	−4.65 (−7.10, −2.19), .0002**	−0.61 (−1.67, 0.45), .2617
High	−3.53 (−6.92, −0.15), .0416*	−2.08 (−3.54, −0.61), .0057**
EAR-RDA
None to no	Reference	Reference
Middle	−0.27 (−5.75, 5.20), .9220	0.21 (−1.98, 2.40), .8513
High	−10.43 (−16.36, −4.49), .0008**	−3.72 (−6.09, −1.34), .0027**
>RDA
None to no	Reference	Reference
Middle	0.15 (−4.25, 4.55), .9466	−0.69 (−2.39, 1.02), .4330
High	−3.20 (−12.05, 5.65), .4800	0.85 (−2.59, 4.28), .6305

Crude model adjust for: None. Model 1 adjust for: gender, age, and race. Model 2 adjust for: gender, age, race, dietary magnesium, smoking status, drinking status, physical activity, education, BMI, and PIR.

BMI = body mass index, DSST = digit symbol substitution test, EAR = estimated average requirement, MDS = magnesium depletion score, NHANES = National Health and Nutrition Examination Survey, RDA = recommended dietary allowance.

^*P < .05.

^**P < .01.

10. Discussion

This study analyzed data from 768 older U.S. adults in NHANES 2011 to 2014 to investigate associations between MDS and cognitive performance. After comprehensive adjustment for confounders, higher MDS levels showed significant negative correlations with performance on both the DSST (assessing processing speed, attention, and working memory) and AFT (evaluating verbal fluency) tests. However, no association was observed between MDS and CERAD test scores (measuring episodic memory), suggesting MDS may specifically influence frontal lobe-mediated cognitive functions rather than hippocampal-dependent memory processes. These findings highlight magnesium’s potential selective role in executive function and verbal fluency among older adults.^[25] In contrast, tests like DSST and AFT evaluate processing speed and executive functions tied to prefrontal cortex activity, where magnesium’s effects on synaptic plasticity are less significant.^[26] Aging alters magnesium metabolism, affecting its neuroprotective role across cognitive domains.^[27] In other words,a higher MDS is associated with lower cognitive scores in moderation. After stratified through gender, smoking, BMI, and dietary magnesium intake, this study found that this relationship was significant in BMI > 30 kg/m²,smokers and dieary megsim intake < RAD subgroups.Notably, this association disappeared when dietary magnesium intake met the RDA, implying that adequate magnesium consumption, potentially through magnesium-rich foods, may mitigate cognitive decline in magnesium-deficient individuals. To our knowledge, this represents the first study examining the link between MDS and cognitive performance.

Magnesium plays a critical role in neuronal health by supporting myelination, synaptic plasticity, intracellular signaling, and neurochemical regulation.^[28] Deficiency impairs mitochondrial adenosine triphosphate production, essential for maintaining membrane stability and synaptic function, while also exacerbating glutamatergic neurotransmission. This can lead to excitotoxicity, oxidative stress, and neuronal apoptosis,^[29] mechanisms implicated in numerous neurological disorders.^[30] These findings underscore magnesium’s potential as a therapeutic target for preventing or mitigating such conditionsWhile serum magnesium levels are commonly used to diagnose deficiency clinically, they may not fully represent total body magnesium status due to tight renal regulation, the kidneys reabsorb over 80% of filtered magnesium to maintain homeostasis. To address this limitation, Fan et al^[14] developed the MDS, incorporating 4 key factors that chronically deplete magnesium stores: diuretic use (e.g., furosemide disrupts paracellular transport in the loop of Henle^[31]), PPIs (impair duodenal absorption via pH elevation^[32]), renal dysfunction (CKD increases urinary losses^[33]), and heavy alcohol consumption (promotes urinary excretion of magnesium^[14]). These mechanisms collectively contribute to neurological risks.^[34] Our findings demonstrate that higher MDS values correlate with poorer performance on the DSST and AFT tests, but only in individuals with suboptimal magnesium intake (<RDA). This threshold effect suggests achieving adequate dietary magnesium may offset the cognitive risks associated with magnesium-depleting conditions. Magnesium has demonstrated significant influence on cognitive function through several critical pathways. Firstly, magnesium deficiency leads to neuroinflammation by activating the NLRP3 inflammasome, which disrupts neuronal function and synaptic plasticity, thus contributing to cognitive decline.^[35] Secondly, magnesium plays a vital role in reducing oxidative stress; it is a necessary cofactor for antioxidant enzymes like superoxide dismutase. A deficiency of magnesium increases oxidative damage to neurons and glial cells, further impairing cognitive health.^[36] Lastly, magnesium is essential for mitochondrial function, particularly in adenosine triphosphate production and energy metabolism within neurons. Insufficient levels of magnesium disrupt synaptic transmission as well as cognitive processes such as memory.^[37,38] A recent study revealed that increased total magnesium intake correlated with significantly improved DSST performance and lower cognitive impairment risk.^[39]

Obesity has been consistently associated with lower cognitive function, notably with poor executive function, intellectual functioning, psychomotor performance and speed, and visual construction.^[40] Research indicates that obesity can lead to changes in brain structure and function, such as reduced hippocampal volume and white matter damage in the brain. The biological mechanisms through which BMI affects cognitive function may involve multiple factors. Obesity and being overweight can lead to metabolic syndrome, diabetes, and cardiovascular diseases, all of which are associated with cognitive decline.^[41] Whats more, obesity is related to a chronic inflammatory state, which can cause neuronal damage and cognitive decline^[42]. Additionally, malnutrition and metabolic abnormalities may affect normal brain development and function, thereby impacting cognitive function.^[41] However, some other studies have found that underweight individuals may have issues with malnutrition, which can impact normal brain development and function, particularly in older adults.^[43] In our study,we found that among the olders with BMI > 30,higher MDS was associated with lower cognitive function. This may be due to magnesium deficiency is associated with obesity and metabolic syndrome. Magnesium possesses anti-inflammatory properties. Low magnesium levels can exacerbate inflammation, which is a common feature of obesity and metabolic syndrome. Chronic inflammation can further worsen insulin resistance and weight gain. Further studies are necessary to better understand potential modification of associations by BMI.

Smoking adversely affects multiple cognitive domains, such as memory, attention, executive function, and processing speed.The biological mechanisms underlying these effects are multifaceted, involving neurotoxicity, oxidative stress, inflammation, and vascular damage. Recent studies have provided more detailed insights into the relationship between smoking and cognitive impairment. A study found that smoking is associated with accelerated decline in cognitive abilities in middle-aged and older adults, emphasizing the cumulative negative impact of long-term smoking on brain health.^[44] Another research has identified specific biomarkers that indicate smoking-related decline in cognitive abilities, providing potential targets for early intervention and treatment.^[45] Smoking can affect the absorption of magnesium. Some chemicals in cigarettes may interfere with the absorption of magnesium in the digestive tract, leading to lower levels of magnesium in the body. Smoking can increase the loss of magnesium, especially through urine excretion. Nicotine and other tobacco components may stimulate the kidneys, increasing the excretion of magnesium.^[46] Consistent with previous studies, in this experiment, we found that the olders who is smoking now exhibited a relationship between higher MDS and lower cognitive function. Therefore, avoiding obesity, increasing magnesium intake, and quitting smoking may help reduce the risk of cognitive impairment in patients with high MDS. But in our study, subgroup analyses has limitations. Subgroup analyses (e.g., by BMI, smoking, and magnesium intake) were not adjusted for multiple testing, which may increase the risk of Type I error. These analyses were exploratory, aimed at generating hypotheses rather than definitive conclusions. Future studies with larger samples and specific hypotheses should apply stricter corrections to reduce Type I error risk.

11. Study strengths and limitations

This study offers clinical evidence for an inverse relationship between MDS and cognitive performance, supported by a large, nationally representative U.S. cohort with stringent quality control. However, several limitations warrant consideration. First, the cross-sectional design limits causal interpretation. Second, although multiple covariates were adjusted for, residual bias from unmeasured confounders (e.g., depression, diabetes, neurodegenerative diseases) may persist, affecting the observed associations. Future studies should address these factors to enhance robustness. Third, although the MDS was validated in a population including older adults (Fan et al, 2021), its performance may differ in geriatric-specific cohorts due to age-related physiological changes (e.g., renal function decline or polypharmacy). Future studies should validate the MDS in exclusively older populations to confirm its predictive utility for cognitive outcomes. Forth, reliance on self-reported data for certain variables may introduce recall bias, necessitating cautious interpretation of those measures.

12. Conclusions

Higher MDS levels were significantly linked to poorer performance on the DSST and AFT cognitive tests in older U.S. adults, though no association emerged with CERAD test scores. Notably, participants meeting the RDA for dietary magnesium intake appeared protected against cognitive decline. These findings suggest MDS could serve as a predictive marker for cognitive impairment, highlighting its potential utility in clinical assessments and preventive strategies for aging populations. Further multi-center studies are warranted to validate these observations and refine risk prediction models.

Author contributions

Conceptualization: Chao Wang.

Data curation: Guangling Li, Chao Wang, Ning Zhou, HaiLang Wang.

Formal analysis: Guangling Li, Chao Wang.

Funding acquisition: Chao Wang.

Supervision: Chao Wang.

Software: Guangling Li, Chao Wang, Ning Zhou, HaiLang Wang.

Validation: Guangling Li, Chao Wang, HaiLang Wang.

Visualization: Chao Wang.

Writing – original draft: Guangling Li, Chao Wang.

Writing – review & editing: Guangling Li, Chao Wang.