Chronic kidney disease, physical activity and cognitive function in older adults—results from the National Health and Nutrition Examination Survey (2011–2014)

ABSTRACT

Background

Cognitive impairment is common among persons with chronic kidney disease (CKD), due in part to reduced kidney function. Given that physical activity (PA) is known to mitigate cognitive decline, we examined whether associations between CKD stage and global/domain-specific cognitive function differ by PA.

Methods

We leveraged 3223 participants (≥60 years of age) enrolled in National Health and Nutrition Examination Survey (NHANES, 2011–2014), with at least one measure of objective cognitive function [immediate recall (CERAD-WL), delayed recall (CERAD-DR), verbal fluency (AF), executive function/processing speed (DSST), global (average of four tests) or self-perceived memory decline (SCD)]. We quantified the association between CKD stage {no CKD: estimated glomerular filtration rate [eGFR] ≥60 mL/min/1.73 m² and albuminuria [albumin:creatinine ratio (ACR)] <30 mg/g; stages G1–G3: eGFR ≥60 mL/min/1.73 m² and ACR ≥30 mg/g or eGFR 30–59 mL/min/1.73 m²; stages G4 and G5: eGFR <30 mL/min/1.73 m²} and cognitive function using linear regression (objective measures) and logistic regression (SCD), accounting for sampling weights for nationally representative estimates. We tested whether associations differed by PA [Global Physical Activity Questionnaire, high PA ≥600 metabolic equivalent of task (MET) · min/week versus low PA <600 MET · min/week] using a Wald test.

Results

Among NHANES participants, 34.9% had CKD stages G1–G3, 2.6% had stages G4 and G5 and 50.7% had low PA. CKD stages G4 and G5 were associated with lower global cognitive function {difference = −0.38 standard deviation [SD] [95% confidence interval (CI) −0.62 to −0.15]}. This association differed by PA (P_interaction = 0.01). Specifically, among participants with low PA, those with CKD stages G4 and G5 had lower global cognitive function [difference = −0.57 SD (95% CI −0.82 to −0.31)] compared with those without CKD. Among those with high PA, no difference was found [difference = 0.10 SD (95% CI −0.29–0.49)]. Similarly, the CKD stage was only associated with immediate recall, verbal fluency, executive function and processing speed among those with low PA; no associations were observed for delayed recall or self-perceived memory decline.

Conclusions

CKD is associated with lower objective cognitive function among those with low but not high PA. Clinicians should consider screening older patients with CKD who have low PA for cognitive impairment and encourage them to meet PA guidelines.

Graphical Abstract

Graphical Abstract.

KEY LEARNING POINTS.

What is already known about this subject?

• Cognitive impairment is common among persons with chronic kidney disease (CKD) due in part to reduced kidney function. Given that physical activity is known to mitigate cognitive decline, we examined whether associations between CKD stage and global/domain-specific cognitive function differ by physical activity.

What this study adds?

• CKD is associated with worse global cognitive function, immediate recall, verbal fluency, executive functioning and processing speed among those with low, but not high physical activity. No associations were observed for delayed recall or self-perceived memory decline. Additionally, objective measures of cognitive function detected low cognitive performance more frequently than subjective measures.

What impact this may have on practice or policy?

• Clinicians should consider objectively screening older patients with CKD who have low physical activity for cognitive impairment and encourage them to meet physical activity guidelines.

INTRODUCTION

Cognitive impairment is common among people of all ages with chronic kidney disease (CKD), with prevalence ranging from 13 to 58% [1–5]. The association between CKD and cognitive function has been extensively studied [2, 6–11], with collective evidence demonstrating that with each 10 mL/min/1.73 m² decrease in estimated glomerular filtration rate (eGFR) among those with CKD, the odds of cognitive impairment increase by 10–12% [1, 11, 12]. Putative mechanisms underlying the association between CKD and cognitive decline include vascular dysfunction, lymphatic dysfunction, decreased clearance of uremic toxins and hemodynamic changes during dialysis [6, 7]. Nonetheless, prevention and treatment of cognitive impairment in CKD are limited [13]. A more thorough examination of the association between kidney function and cognitive function by specific cognitive domains can help discern potential differences in etiologies; for example, two of the most common types of dementia, Alzheimer's disease and vascular dementia, are often distinguished by disproportionate impairments in episodic memory and executive function, respectively [14].

Physical activity (PA) is likely protective against cognitive decline among the general population and persons with chronic health conditions [15–21]. After PA, older adults experience improved regulation of hippocampal function, neurogenesis, synaptic plasticity, cerebral blood flow and a reduction in cardiovascular risk and proinflammatory activity [15, 22, 23]. Given the high prevalence of cardiovascular disease in individuals with CKD, it is likely that the association between kidney function and cognitive function is modified by PA level [24, 25].

To test whether the association between CKD stage and cognitive function differs by PA level, we leveraged the National Health and Nutrition Examination Survey (NHANES, 2011–2014), a nationally representative cross-sectional study designed to assess the health and nutritional status of the US civilian noninstitutionalized resident population [26]. Our goals were to quantify the association between CKD stage and global and domain-specific cognitive function and test whether these associations differed between those with low and high PA levels.

MATERIALS AND METHODS

Study design

We conducted a cross-sectional study of 3223 participants ≥60 years of age from NHANES (2011–2014). Participants with measurements of serum creatinine for eGFR calculation and at least one assessment of cognitive function, as described below, were included in the study. Participants’ demographic information and health status were collected either through direct measurement or via self-report at each study cycle, including body mass index (BMI; weight divided by height squared), depressive symptoms [Patient Health Questionnaire (PHQ-9) ≥10] [27–29], hypertension [systolic blood pressure (SBP) ≥130 mmHg, diastolic blood pressure (DBP) ≥80 mmHg or reported current use of antihypertensive medication], diabetes (fasting blood glucose level ≥126 mg/dL, nonfasting glucose level ≥200 mg/dL, reported history of diabetes or current use of medications for diabetes or high blood sugar) and anemia (hemoglobin <12 g/dL in males, hemoglobin <11 g/dL in females or reported taking treatment for anemia in the past 3 months). History of coronary heart disease (CHD), myocardial infarction (MI), stroke and smoking (smoked at least 100 cigarettes in life) were collected by self-report using a questionnaire.

CKD stage

Serum creatinine and albuminuria [albumin:creratinine ratio (ACR)] were measured at each study cycle. Serum creatinine was used to calculate eGFR with the Chronic Kidney Diease Epidemiology Collaboration (CKD-EPI) equation [30]. We defined categories of CKD stage based on previously published guidelines: no CKD (eGFR ≥60 mL/min/1.73 m² and ACR <30 mg/g), CKD stages G1–G3 (eGFR 30–59 mL/min/1.73 m² or eGFR ≥60 mL/min/1.73 m² and ACR ≥30 mg/g) and CKD stages G4 and G5 (eGFR < 30 mL/min/1.73 m²) [31].

Cognitive function

We defined global and domain-specific cognitive function using the objective and subjective measures available in NHANES at each cycle (2011–2014). Objective measures included four separate tests: word list learning trials from the Consortium to Establish a Registry for Alzheimer's Disease battery for immediate recall (CERAD-WL) and delayed recall (CERAD-DR) [32], the Animal Fluency (AF) test for verbal fluency [33, 34] and the Digit Symbol Substitution Test (DSST) for executive function and processing speed [35–37]. The word list learning task for immediate recall (CERAD-WL) and delayed recall (CERAD-DR) examines the ability to recall newly learned information and delayed memory and has been validated as a tool designed to discriminate those with potential cognitive impairment from those with healthy cognitive function [38, 39]. Specifically, 10 words were read aloud by the participant, immediately followed by three consecutive recalls [33, 40, 41]. The delayed recall of all 10 words occurred after the AF and DSST assessments, ∼8–10 min from the start of the word learning trials [33, 40, 41]. The AF is a verbal fluency test that assesses semantic memory, where participants are asked to name aloud as many animals as possible in 1 min [33, 34]. AF test scores have also been shown to discriminate those with normal cognitive function from those who have mild cognitive impairment and more severe forms of cognitive impairment, such as Alzheimer's disease and related dementia [42, 43]. The DSST is a highly sensitive, validated [44–46], paper-and-pencil cognitive test that primarily assesses attention and processing speed [35], but it is also linked to executive functioning [35–37]. The examination was conducted using a paper form that has a key at the top containing nine numbers paired with symbols. Participants copy the matching symbol in boxes that adjoin the numbers in 2 min. For each of the cognitive tests, 1 point was given for each correctly recalled, named, or matched response, with a higher score reflecting better cognitive function [33, 40, 41]. All objective tests were then standardized to a mean of 0 and standard deviation (SD) of 1 and averaged into a global cognitive composite score, as described in prior studies [47].

Subjective cognitive function, assessing self-perceived memory through at-home interviews, was also considered as a separate outcome. Specifically, participants reported whether or not they experienced worsening or more frequent confusion or memory loss during the past 12 months.

Physical activity

PA was collected via questionnaire based on the Global Physical Activity Questionnaire [48]. Participants reported their frequency of and time spent in vigorous and moderate work activity, walking or bicycling and vigorous or moderate recreational activities. Metabolic equivalent of task (MET) scores were assigned for each activity based on prior NHANES guidelines [49]: 8 for vigorous work-related activity, 4 for moderate work-related activity, 4 for walking or bicycling, 8 for vigorous leisure-time PA and 4 for moderate leisure-time PA. The total amount of PA was assessed by summing minutes of activity per week multiplied by the MET score of each activity. A high level of PA was defined as ≥600 MET·min/week and a low level of PA was defined as <600 MET·min/week based on the US PA guidelines [50].

Descriptive statistics

We summarized the distributions of characteristics in participants and presented the distribution in participants with no CKD, with CKD stages G1–G3 and CKD stages G4 and G5. We generated means and SDs for normally distributed continuous variables, medians and interquartile ranges (IQRs) for nonnormally distributed continuous variables and proportions for binary or categorical variables. Mobile examination center sample weights were accounted for in the analysis to produce estimates that are representative of the US non-institutionalized civilian resident population [51]. Descriptive statistics of the study population without imposed sampling weights are also available in the Supplementary data, Table S1.

PA, CKD and objective cognitive function

For the standardized objective cognitive tests, we used adjusted linear regression models to assess the difference in cognitive function by CKD stage to generate Cohen's d [52]; Cohen's d refers to the standardized difference between two means, where effect sizes of 0.2 SD are generally considered small, 0.5 SD medium and 0.8 SD large [53]. To assess whether associations differed between high and low PA levels, we tested the interaction between CKD stage and PA on cognitive function using a Wald test. Models were adjusted for potential confounders of CKD and cognitive decline, including age, sex, race, education, BMI, depressive symptoms, hypertension, diabetes, CHD, stroke, Myocardial Infarction (MI), anemia and smoking. The mobile examination center sample weights were accounted for to generate nationally representative estimates.

PA, CKD and subjective cognitive function

We used adjusted logistic regression models to quantify the association between CKD stage and subjective cognitive function [odds ratios (ORs)]. To test whether the association differed between high and low PA levels, we used an approach similar to that outlined above for objective measures.

Sensitivity analysis

To test whether inferences remained robust overall and by PA level, we first removed race from the CKD-EPI equation to adhere to the National Kidney Foundation and American Society of Nephrology's mission to move past race-based medicine [54]. Additionally, we excluded those with severe depressive symptoms, given that it is highly correlated with CKD, cognitive impairment and PA [7, 55–58]; the PHQ-9 has high validity and reliability against the gold standard criteria of a diagnosis of depression among long-term dialysis patients, as well as in primary care populations [28, 29]. We also tested whether inferences remained robust after more rigid control of anemia using continuous hemoglobin levels. Finally, given that smoking is a risk factor for CKD progression as well as cardiovascular morbidity [59, 60], we assessed whether smoking status (ever versus never smokers) significantly modifies associations between PA, CKD and cognitive function.

Statistical analysis

All statistical analysis was conducted using Stata version 16 (StataCorp, College Station, TX, USA) and we used a statistical significance cutoff of α < 0.05.

RESULTS

Participant characteristics

Among the study population, 16.8% were ≥80 years of age, 78.4% were non-Hispanic White and 54.2% were female (Table 1). Overall, 62.5% had no CKD, 34.9% had CKD stages G1–G3 and 2.6% had CKD stages G4 and G5. Additionally, 14.7% reported self-perceived memory impairment and median scores for objective cognitive tests were 20 (IQR 17–23) for immediate recall (CERAD-WL), 6 (IQR 5–8) for delayed recall (CERAD-DR), 18 (IQR 14–21) for verbal fluency (AF) and 53 (IQR 41–64) for executive function and processing speed (DSST). Additionally, 50.7% met criteria for low PA. Before adjustment, participants without CKD were more likely to meet PA guidelines compared with those with CKD stages G1–G3 (55.4% versus 38.6%) and CKD stages G4 and G5 (54.4% versus 21.5%) (Table 1).

Table 1.

Characteristics of participants ≥60 years of age from the NHANES (2011–2014) (n = 3223) with recorded cognitive function

Characteristics	Overall (N = 3223)	No CKD (n = 2015)	CKD stages G1–G3 (n = 1124)	CKD stages G4 and G5 (n = 84)
Age (years)
60–69	53.7	64.2	33.6	32.1
70–79	29.5	26.6	34.9	36.3
≥80	16.8	9.2	31.5	31.6
Race
Mexican American	3.8	3.9	3.4	5.9
Other Hispanic	3.8	3.9	3.5	2.7
Non-Hispanic White	78.4	78.6	78.7	66.8
Non-Hispanic Black	8.4	7.8	9.0	20.4
Non-Hispanic Asian	3.9	4.3	3.3	1.2
Other	1.7	1.4	2.1	3.0
Female	54.2	53.0	56.7	57.0
BMI (kg/m2), mean (SE)	29.0 (0.2)	28.7 (0.3)	29.5 (0.3)	29.7 (0.9)
Education ≥12 years	81.5	84.4	77.2	57.3
Severe depressive symptoms	7.6	6.8	8.5	17.7
Hypertension	75.5	70.0	86.0	88.8
Treated hypertension	93.7	92.6	95.2	93.5
Diabetes	23.3	17.8	32.2	59.7
CHD	10.0	7.1	15.4	19.0
MI	8.8	6.3	13.4	18.7
Stroke	7.7	5.0	12.6	16.8
Anemia	7.0	3.7	11.4	45.8
Treated anemia	4.3	2.8	6.6	16.5
Smoking	50.1	48.7	53.1	47.5
Physical activity, median (IQR)	540 (2000)	840 (2400)	180 (1440)	0 (300)
Low physical activity	50.7	44.6	61.4	78.5

Proportions (%) are presented unless stated otherwise, accounting for NHANES sampling weights. Serum creatinine was measured and used to calculate eGFR with the CKD-EPI equation. CKD stage was then defined by the following cut-points: no CKD: eGFR ≥60 mL/min/1.73 m² and ACR <30 mg/g; CKD stages G1–G3: eGFR 30–59 mL/min/1.73 m² or eGFR ≥60 mL/min/1.73 m² and ACR ≥30 mg/g; CKD stages G4 and G5: eGFR <30 mL/min/1.73 m². Severe depressive symptoms were defined as a Patient Health Questionnaire [PHQ-9] score ≥10. Physical activity was collected using the Global Physical Activity Questionnaire and converted to MET·min/week. High physical activity was defined as ≥600 MET·min/week. SE, standard error.

CKD and cognitive function

Before adjustment for risk factors, participants with CKD stages G1–G3 {Cohen's d = −0.32 SD [95% confidence interval (CI) −0.41 to −0.23]} and CKD stages G4 and G5 [Cohen's d = −0.81 SD (95% CI −1.00 to −0.62)] had significantly lower global cognitive function compared with those without CKD (P < 0.01) (Figure 1); similar results were found for domain-specific tests (Supplementary data, Table S2). After adjustment for risk factors, only participants with CKD stages G4 and G5 had lower global cognitive function compared with those without CKD [Cohen's d = −0.38 SD (95% CI −0.62 to −0.15)] (Table 2). This finding was consistent for domain-specific tests, particularly for executive function and processing speed [Cohen's d = −0.54 SD (95% CI −0.93 to −0.16)], immediate recall [Cohen's d = −0.38 SD (95% CI −0.56 to −0.21)] and verbal fluency [Cohen's d = −0.29 (95% CI −0.58 to −0.01)]. Those with CKD stages G1–G3 had poorer cognitive performance in executive function and processing speed [Cohen's d = −0.12 (95% CI −0.21 to −0.03)] compared with those without CKD. No associations were observed between CKD stage and delayed recall [CKD stages G1–G3 versus no CKD: Cohen's d = −0.03 (95% CI −0.14–0.08); CKD stages G4 and G5 versus no CKD: Cohen's d = −0.23 (95% CI −0.57–0.11)]. Additionally, CKD stages G1–G3 were associated with poorer self-perceived memory compared with those without CKD [OR 1.41 (95% CI 1.06–1.88)]; however, there was no association with self-perceived memory for those with CKD stages G4 and G5 compared with those without CKD [OR 1.16 (95% CI 0.51–2.62)] (Table 3).

FIGURE 1:

Objective cognitive test scores by CKD stage and physical activity level among participants ≥60 years of age from the NHANES (2011–2014) (N = 3223). Medians and IQRs are presented for cognitive test scores (standardized to a mean of 0 and an SD of 1). NHANES sampling weights were accounted for to obtain nationally representative estimates. Serum creatinine was measured and used to calculate eGFR with the CKD-EPI equation. CKD stage was then defined by the following cut-points: no CKD: eGFR ≥60 mL/min/1.73 m² and ACR <30 mg/g; CKD stages G1–G3: eGFR 30–59 mL/min/1.73 m² or eGFR ≥60 mL/min/1.73 m² and ACR ≥30 mg/g; CKD stages G4 and G5: eGFR <30 mL/min/1.73 m². Physical activity was collected using the Global Physical Activity Questionnaire and converted to MET·min/week, where ≥600 MET·min/week represents high physical activity.

Table 2.

CKD stage and objective assessments of cognitive function by levels of PA among participants ≥60 years of age from NHANES (2011–2014) (N = 3223).

	Overall,	Low physical activity,	High physical activity,
Stage and assessment	mean (95% CI)	mean (95% CI)	mean (95% CI)	P for interaction
Global cognitive function
No CKD	0 (ref)	0 (ref)	0 (ref)
CKD stages G1–G3	−0.07 (-0.15–0.00)	−0.13 (−0.21 to −0.05)	−0.01 (−0.11–0.08)	0.03
CKD stages G4 and G5	−0.38 (−0.62 to −0.15)	−0.57 (−0.82 to −0.31)	0.10 (−0.29–0.48)	0.01
Immediate recall (CERAD-WL)
No CKD	0 (ref)	0 (ref)	0 (ref)
CKD stages G1–G3	−0.10 (−0.19 to −0.00)	−0.17 (−0.31 to −0.02)	−0.03 (−0.19–0.14)	0.24
CKD stages G4 and G5	−0.38 (−0.56 to −0.21)	−0.50 (−0.77 to −0.23)	−0.11 (−0.52–0.31)	0.18
Delayed recall (CERAD-DR)
No CKD	0 (ref)	0 (ref)	0 (ref)
CKD stages G1–G3	−0.03 (−0.14–0.08)	−0.10 (−0.25–0.04)	0.05 (−0.11–0.21)	0.13
CKD stages G4 and G5	−0.23 (−0.57–0.11)	−0.37 (−0.80–0.06)	0.08 (−0.36–0.53)	0.15
Verbal fluency (AF)
No CKD	0 (ref)	0 (ref)	0 (ref)
CKD stages G1–G3	−0.09 (−0.19–0.01)	−0.13 (−0.25 to −0.00)	−0.05 (−0.18–0.07)	0.34
CKD stages G4 and G5	−0.29 (−0.58 to −0.01)	−0.48 (−0.78 to −0.18)	0.19 (−0.31–0.68)	0.02
Executive function and processing speed (DSST)
No CKD	0 (ref)	0 (ref)	0 (ref)
CKD stages G1–G3	−0.12 (−0.21 to −0.03)	−0.19 (−0.30 to −0.08)	−0.05 (−0.20–0.09)	0.15
CKD stages G4 and G5	−0.54 (−0.93 to −0.16)	−0.75 (−1.18 to −0.32)	−0.01 (−0.51–0.50)	0.02

Significant values (P < 0.05) are in bold.

Cognitive test scores are standardized to a mean of 0 and an SD of 1. Models are adjusted for age, sex, race, education, BMI, depressive symptoms, hypertension, diabetes, CHD, MI, stroke, anemia and smoking. NHANES sampling weights are accounted for in linear regression analyses to obtain nationally representative estimates. Serum creatinine was measured and used to calculate eGFR with the CKD-EPI equation. CKD stage was then defined by the following cut-points: no CKD: eGFR ≥60 mL/min/1.73 m² and ACR <30 mg/g; CKD stages G1–G3: eGFR 30–59 mL/min/1.73 m² or eGFR ≥60 mL/min/1.73 m² and ACR ≥30 mg/g; CKD stages G4 and G5: eGFR <30 mL/min/1.73 m². Physical activity was collected using the Global Physical Activity Questionnaire and converted to MET·min/week, where ≥600 MET·min/week represents high physical activity.

Table 3.

Association between renal function and self-perceived memory decline by physical activity (N = 3223)

	Overall,	Low physical activity,	High physical activity,
CKD stage	OR (95% CI)	OR (95% CI)	OR (95% CI)	P for interaction
No CKD	1 (ref)	1 (ref)	1 (ref)
CKD stages G1–G3	1.41 (1.06, 1.88)	1.60 (1.09, 2.34)	1.15 (0.76, 1.73)	0.24
CKD stages G4 and G5	1.16 (0.51, 2.62)	1.19 (0.47, 3.03)	1.24 (0.13, 11.47)	0.97

Models are adjusted for age, sex, race, education, BMI, depressive symptoms, hypertension, diabetes, CHD, MI, stroke, anemia and smoking. NHANES sampling weights are accounted for in logistic regression analysis to obtain nationally representative estimates. Serum creatinine was measured and used to calculate eGFR with the CKD-EPI equation. CKD stage was then defined by the following cut-points: no CKD: eGFR ≥60 mL/min/1.73 m² and ACR <30 mg/g; CKD stages G1–G3: eGFR 30–59 mL/min/1.73 m² or eGFR ≥60 mL/min/1.73 m² and ACR ≥30 mg/g; CKD stages G4 and G5: eGFR <30 mL/min/1.73 m². Physical activity was collected using the Global Physical Activity Questionnaire and converted to MET·min/week, where ≥600 MET·min/week represents high physical activity.

PA, CKD stage and global cognitive function

The association between CKD and global cognitive function differed by PA level when comparing those with CKD stages G4 and G5 with those without CKD (P for interaction = 0.01). Before adjustment (Supplementary data, Table S2), among participants with low PA, those with CKD stages G4 and G5 had poorer global cognitive function compared with those without CKD [Cohen's d = −0.99 (95% CI −1.15 to −0.83)]; however, there was no association between CKD and global cognitive function among those with high PA [Cohen's d = −0.27 (95% CI −0.84–0.30)]. The association between CKD and global cognitive function did not differ by PA level when comparing those with CKD stages G1–G3 with those without CKD (P for interaction = 0.05).

After adjustment (Table 2), the association between CKD and global cognitive function differed by PA level for those with CKD stages G4 and G5 compared with those without CKD (P for interaction = 0.01). Among participants with low physical activity, those with CKD stages G4 and G5 had poorer global cognitive function compared with those without CKD [Cohen's d = −0.57 (95% CI −0.82 to −0.31)]; however, there was no association between CKD and global cognitive function among those with high PA [Cohen's d = 0.10 (95% CI −0.29–0.48)]. Similar results were observed by PA level when comparing those with CKD stages G1–G3 with those without CKD [low PA: Cohen's d = −0.13 (95% CI −0.21 to −0.05); high PA: Cohen's d = −0.01 (95% CI −0.11–0.08); P for interaction = 0.03].

PA, CKD stage and cognitive function by cognitive domain

Before adjustment (Supplementary data, Table S2) and after adjustment (Table 2), differences by PA level were consistent for verbal fluency (P for interaction = 0.02) and executive function and processing speed (P for interaction = 0.02). Specifically, among those with low PA, CKD stages G4 and G5 were associated with lower performance in verbal fluency compared with those without CKD [Cohen's d = −0.48 (95% CI −0.78 to −0.18)], but among those with high PA, there was no association [Cohen's d = 0.19 (95% CI −0.31–0.68)]. Similar findings were observed for executive function and processing speed when comparing CKD stages G4 and G5 with those without CKD [low PA: −0.75 (95% CI −1.18 to −0.32); high PA: −0.01 (95% CI −0.51–0.50)]. The association between CKD stages G1–G3 and domain-specific cognitive function did not differ by PA level for any of the cognitive domains (all P for interaction > 0.05) (Table 2). Similarly, the association between CKD and self-perceived memory impairment did not differ by PA (CKD stages G4 and G5 versus no CKD: P for interaction = 0.97; CKD stages G1–G3 versus no CKD: P for interaction = 0.24) (Table 3).

Sensitivity analyses

When we removed race from the CKD-EPI equation to classify CKD stage (Supplementary data, Tables S4 and S5), excluded participants with severe depressive symptoms (Supplementary data, Tables S6 and S7) or controlled more rigidly for anemia using continuous hemoglobin levels (Supplementary data, Tables S8 and S9), inferences remained robust across all global and domain-specific cognitive tests, overall and by PA level. Additionally, smoking status did not significantly modify associations between PA, CKD and cognitive function for all global and domain-specific cognitive tests (P > 0.05); nevertheless, subgroup analyses restricted to ever versus never smokers are presented in the Supplementary data, Tables S10 and S11.

DISCUSSION

In this nationally representative study of 3223 participants, we found that compared to participants without CKD in NHANES, only those with CKD stages G4 and G5 were associated with poorer global cognitive performance [CKD stages G1–G3: difference = −0.07 SD (95% CI −0.15–0.00); CKD stages G4 and G5: difference = −0.38 SD (95% CI −0.62 to −0.15)]. Similar findings were observed for domain-specific tests. Compared with those without CKD, CKD stages G4 and G5 were most strongly associated with poorer performance in executive function and processing speed [Cohen's d = −0.54 SD (95% CI −0.93 to −0.16)], immediate recall [Cohen's d = −0.38 SD (95% CI −0.56 to −0.21)] and verbal fluency [Cohen's d = −0.29 (95% CI −0.58 to −0.01)]. No associations were observed for delayed recall. Interestingly, self-perceived memory decline was only associated with CKD stages G1–G3 [OR 1.41 (95% CI 1.06–1.88)], but not with CKD stages G4 and G5 [OR 1.16 (95% CI 0.51–2.62)]. Nonetheless, associations with global cognitive performance differed by PA level for both CKD stages G4 and G5 (P for interaction = 0.01) and CKD stages G1–G3 (P for interaction = 0.03) compared with those without CKD, with greatest differences found among those in later CKD stages. Specifically, among participants with low PA, those with CKD stages G4 and G5 [Cohen's d = −0.57 (95% CI −0.82 to −0.31)] and CKD stages G1–G3 [(Cohen's d = –0.13 (95% CI −0.21 to −0.05)] had lower global cognitive function compared with those without CKD; however, among those with high PA, no differences were observed [CKD stages G1–G3 versus no CKD: Cohen's d = −0.01 (95% CI −0.11–0.08); CKD stages G4 and G5 versus no CKD: Cohen's d = 0.10 (95% CI −0.29–0.48)]. Consistent with findings for global cognitive performance, associations differed by PA for verbal fluency (P = 0.02) and executive function and processing speed (P = 0.02).

Previous studies of cognitive impairment and kidney disease report similar results [61–66]. Despite the different objective measures used to assess cognitive function, as well as the different algorithms used to assess kidney function, studies have consistently demonstrated that worse kidney function is associated with worse cognitive function. Our findings corroborate prior results in a diverse, nationally representative sample of adults ≥60 years of age. In this study, worse CKD was associated with moderately lower cognitive function [CKD stages G4 and G5 versus no CKD: difference = −0.38 SD (95% CI −0.62 to −0.15)] based on the Cohen's d scale. Interestingly, findings were not consistent when using a subjective measure of self-perceived memory function [CKD stages G1–G3: 1.41 (95% CI 1.06–1.88); CKD stages G4 and G5: 1.16 (95% CI 0.51–2.62)]. This is especially concerning given the lack of standardization in cognitive screening [67]; clinicians often rely on subjective, self-perceived screening rather than more thorough, objective testing [68]. Considering these findings in parallel with current clinical practice, nephrologists should consider using objective measures when available to detect lower cognitive performance related to reduced kidney function.

This study extends prior findings on the relationship between cognition and kidney function by investigating how physical activity modifies this association. Similar to the cognitive benefits observed among healthy older adults generally [15, 17–20], individuals with chronic health conditions [16] and among patients with end-stage kidney disease (ESKD) [21], our nationally representative study of persons ≥60 years of age demonstrates that the correlation between CKD and global cognitive performance was mitigated for those who met PA guidelines for both CKD stages G1–G3 [high PA: difference = −0.01 (95% CI −0.11–0.08); low PA: difference = −0.13 (95% CI −0.21 to −0.05)] and CKD stages G4 and G5 [high PA: difference = 0.10 SD (95% CI −0.29–0.48); low PA: difference = −0.57 SD (95% CI −0.82 to −0.31)] compared with those without CKD. Notably, these results are cross-sectional in nature; however, there is evidence from other studies that walking, pedaling and cycling can preserve cognition among patients with ESKD undergoing dialysis [69–73]. Additionally, other studies have found that impaired physical performance, such as slower gait speed, was an early marker of cognitive dysfunction [74]. It is likely that exercise and improvements in PA or physical performance can reduce cardiovascular risks [75, 76], improve cerebral blood flow and, in turn, mitigate the impact of cerebral infarcts on cognitive decline in the CKD population [77, 78].

Indeed, our results, though cross-sectional in nature, may reflect the potential vascular mechanisms underlying the relationship between kidney function, physical activity and cognitive function by investigating associations by specific cognitive domain. Findings were consistent for verbal fluency (CKD stages G4 and G5 versus no CKD: P for interaction = 0.02), as well as executive function and processing speed measured using the DSST, where the greatest differences were observed by activity level (CKD stages G4 and G5 versus no CKD: P for interaction = 0.02). Other objective tests linked more closely to memory, including the CERAD-WL (immediate recall), CERAD-DR (delayed recall) [32] and SCD (self-perceived memory decline), were less consistent. These findings support prior studies that have highlighted the implications of reduced kidney function on executive function, such as among those with ESKD [79, 80]. However, our findings extend those results to a novel, nationally representative population of older patients with CKD and observed that those with low PA are most vulnerable. Impairments in executive function in patients with CKD may suggest early signs of dementia, particularly vascular dementia [14, 81], and in many cases Alzheimer's disease [82, 83], likely as a result of highly prevalent traditional cardiovascular disease risk factors and inflammation that may predispose those with low PA to abnormal brain function [84]. Further prospective studies designed to assess whether programs to improve PA can mitigate a decline in cognitive function in patients with CKD are warranted given the lack of prevention and treatment strategies available for this vulnerable population [13].

There are several notable limitations that deserve comment. Most importantly, the cross-sectional nature of the data does not allow any causal inference; the relationship between CKD, cognitive function and physical activity should be studied prospectively to corroborate other studies in non-CKD populations that have found bidirectional associations between physical and cognitive function [15, 85–87]. Additionally, as an observational study, NHANES is subject to unmeasured confounding. Though we adjusted for many known correlates of kidney impairment, physical activity and cognitive function, for other unmeasured reasons, the same individuals who tend to have low activity and greater CKD severity may be at greater risk of reduced cognitive function. Finally, though we found no significant differences in the relationship between PA, CKD and cognitive function by smoking status (Supplementary data, Tables S10 and S11), further investigations designed to unpack the role of smoking behavior, CKD progression and PA on cognitive function are needed.

Despite these limitations, this study has many strengths. It extends prior findings to a large, novel, nationally representative sample of individuals ≥60 years of age. It additionally presents both subjective and objective measures of cognitive function, providing the opportunity to examine multiple cognitive domains. Finally, it includes a measure of PA that allows direct calculation of MET scores for comparison with current PA level standards.

In conclusion, while kidney function is associated with reduced global cognitive performance and executive function, meeting PA guidelines may mitigate those relationships. Additionally, objective measures of cognitive function detected low cognitive performance in those with CKD more frequently than subjective measures, particularly among those with low PA. Therefore clinicians should consider objectively screening older patients with CKD who have low PA for cognitive impairment and encourage them to meet PA guidelines.

AUTHORS’ CONTRIBUTIONS

N.M.C. and M.M.D. participated in concept design, interpretation, drafting, critical revision and approval of the article. J.H. participated in data analysis, drafting and approval of the article. O.H., X.C., K.J.F., I.D. and C.B. participated in critical revision and approval of the article. D.L.S. participated in concept design, critical revision and approval of the article.

FUNDING

Study investigators were funded by the National Institute of Diabetes and Digestive and Kidney Disease, the National Institute of Allergy and Infectious Disease and the National Institute on Aging {grant numbers K01AG064040 [principal investigator (PI) N.M.C.], K24AI144954 [PI D.S.], R01AG055781 [PI M.A.M.-D.], R01DK120518 [PI M.A.M.-D.] and R01DK114074 [PI M.A.M.-D.], as well as by the Manitoba Medical Services Foundation (to C.B.). Funders had no role in the study design, data collection, analysis, reporting or decision to submit for publication.

CONFLICT OF INTEREST STATEMENT

C.B. has received research funding from Hope Pharmaceuticals and has ownership interest in Precision Advanced Digital Manufacturing. K.J.F. receives a stipend as the Patient Voice Editor of the Clinical Journal of the American Society of Nephrology, receives compensation as an advisor to Responsum CKD and the Kidney Research Institute and has ongoing consulting contracts with Bayer, Gilead, Natera, Hansa Biopharma, Veloxis, Talaris, Travere Therapeutics, Palladio Biosciences, Otsuka and eGenesis.