The cross-sectional association between mean corpuscular volume level and cognitive function in Chinese over 45 years old: Evidence from the China Health and Retirement Longitudinal Study.

Fewer studies have focused on the independent association between mean corpuscular volume (MCV) and cognitive performance. This study was designed to characterize the cross-sectional association between MCV and cognitive performance in a large sample of Chinese residents (age≥45 years) from the China Health and Retirement Longitudinal Study (CHARLS). A total of 4023 male and 4173 female adults with MCV ≥ 80 fl were included for analysis. By multivariable linear regression analysis, for the total subjects, MCV level was significantly negatively associated with global cognitive function and episodic memory. When adjusted by sex, only in male subjects, higher MCV level was associated with reduced scores for global cognitive function, episodic memory and mental status. Via binary logistic regression analysis, the higher MCV level (MCV>100 fl) was associated with poor global cognitive function (OR = 1.601; 95% CI = 1.198-2.139; p = 0.001), episodic memory (OR = 1.679; 95% CI = 1.281-2.201; p<0.001), and mental status (OR = 1.422; 95% CI = 1.032-1.959; p = 0.031) for the whole participants. When testing this association by sex, the significant relationship between higher MCV level with worse episodic memory was observed both in male (OR = 1.690; 95% CI = 1.211-2.358; p = 0.002) and female (OR = 1.729; 95% CI = 1.079-2.770; p = 0.023) subjects; while the association between higher MCV level and poor global cognitive function (OR = 1.885; 95% CI = 1.329, 2.675; p<0.001) and mental status (OR = 1.544; 95% CI = 1.034, 2.306; p = 0.034) only existed in male subjects. Further studies are warranted to clarify the association between MCV level and cognitive performance by considering sex into consideration both cross-sectionally and longitudinally.

Introduction

Dementia is one of the non-communicable diseases, which has exerted the greatest economic and social burden worldwide [1]. It is estimated by the WHO that 4.6 million new patients are inflicted with dementia each year with numbers affected nearly doubling every 20 years to reach 81.1 million by 2040 [2]. In the past 30 years, China’s economy has developed exponentially and Chinese population has been aging rapidly [3]. A meta-analysis published in 2018 reported that the overall prevalence of dementia in Chinese elderly (>60 years old) was 5.30% (95%CI: 4.30, 6.30) [4]. Especially, as for one of the major form of dementia, i.e., Alzheimer’s disease (AD), China bears a heavy burden of AD costs, which greatly change the evaluation of AD cost around the world [5]. Nevertheless, there are no effective approaches to slow down the rapid rate of dementia incidence. Dementia is the final stage of many years’ development of pathological changes in the cerebral tissue [6]. In order to prevent dementia, it is of great significance to pay more attention to potential risk factors associated with cognitive performance, especially in Chinese population.

Multiple modifiable risk factors including anemia might be associated with cognitive decline or dementia. For example, results from previous studies have suggested that anemia and low hemoglobin concentrations are independent risk factors of cognitive decline [7-13]. Mechanistically, anemia could lead to brain hypo-oxygenation and consequently to cognitive decline [14]. In contrast, accumulating evidence also suggest that there was no association between anemia or low hemoglobin and cognitive performance or impairment after adjusting for a series of factors in different study populations [15-18]. Except for anemia and hemoglobin concentration, mean corpuscular volume (MCV) can reflect the morphology of erythrocyte, which is commonly used as an index of anemia [19]. Anemia is categorized into microcytic anemia (under 80 fl), normocytic anemia (80–100 fl) and macrocytic anemia(over 100 fl) based on the value of MCV [19]. Healthy erythrocyte ensures that oxygen- and nutrient-rich blood is pumped to the tissues especially brain tissue, so they can function normally. However, few studies have explored the independent association between MCV and cognitive function. The earlier study by Danon et al. [20] observed that a significantly negative correlation existed between erythrocyte volume and memory performance [20]. In recent years, Gamaldo et al. [21] found that high MCV level in older adults is associated with poorer cognitive function and this relationship appears not to be explained by anemia and inflammation by analyzing data from the Baltimore Longitudinal Study of Aging (BLSA). In contrast, by using data from AddNeuroMed Study, no significant linear relationship was observed between MCV and MMSE score [12]. Additionally, there is also evidence suggesting that biological sex plays a critical role in cognitive function [22] and that the relationship between abnormal hemoglobin with worse global cognition was greater in women than in men [13]. Therefore, it is of great necessity to further clarify the association between abnormal MCV level and cognitive performance by taking sex into consideration, also in Chinese subjects.

Owing to the fact that larger red blood cells may have more difficulty passing through small capillaries, compromising to deliver adequate amounts of oxygen to cerebral tissues, there is biological plausibility that higher MCV level than the normal level might be associated with worse cognitive performance. Consequently, we aimed to explore the association between high MCV level and cognitive function based on data from China Health and Retirement Longitudinal Study (CHARLS). We hypothesized that higher MCV levels than 100 fl might be associated with worse cognitive performance.

Materials and methods

Study population

The current study was based on the baseline data of CHARLS, which is conducted by the National School of Development at Peking University from 2011 to 2012. The current study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving study participants were approved by the biomedical ethics committee of Peking University. Written informed consent was obtained from all participants. The specified design of sampling methods for CHARLS has been described previously by Zhao et al. [23]. The flowchart of sample selection for this study was shown in Fig 1. A total of 17708 samples at baseline were extracted and 8409 participants were excluded for those lacking MCV (n = 6176) and cognitive score (n = 2233). Then a total of 1103 participants were excluded for those younger than 45 years old (n = 223), MCV<80 fl (n = 783), and subjects with memory-related disease (n = 97). Consequently, our final analysis compromised of 8,196 subjects. The subjects were further stratified into male (n = 4,023) and female (n = 4,173) group.

Fig 1

Flowchart of recruited participants.

Measurement of cognitive function

Two main assessments for assessing mental status and episodic memory were utilized to identify the cognitive function of participants by trained interviewers, which have been described in multiple previous studies [7, 24–27]. The mental status measurement was based on the Telephone Interview of Cognitive Status (TICS) and drawing a figure successfully. The TICS is one of the more practical ways of testing the respondents’ mental status of cognition, which is seen as the telephone version of MMSE. The TICS contains 10 questions about today’s date (year, month and day), the day of the week and season of the year, and to subtract 7 consecutive times (up to 5 times) from 100. A correct answer is equal to one point and the final score ranges from 0 to 10. Additionally, participants were asked to repaint a picture given by the interviewers and he/she could get 1 point if the participant has drawn the figure successfully. The second assessment was designed to assess episodic memory through immediate and delayed recall of 10 Chinese nouns. Ten unrelated Chinese words were read by interviewers only once, and each participant was required to recall of Chinese nouns immediately (immediate word recall) and to recall the same nouns again 4 minutes later (delayed word recall). The memory ability was evaluated by counting the average of how many words they recall correctly in both immediate- and delayed-word recall tests, with the final score ranging from 0 to 10. In order to compare individual scores from different measurements, we used Z-scores to standardize different tests’ scores based on the mean and standard deviation of the domain-specific cognitive score (i.e. scores for mental status and episodic memory, respectively) and global cognitive score (i.e. scores for the summation of mental status and episodic memory) from recruited subjects [28]. The lowest quartiles of Z score was derived for both global and domain specific cognitive score. We defined the respondents have poorer cognitive function if a Z-score < the lowest quartile (i.e. p25) of respective Z score.

Assessment of MCV and covariates

For estimation of MCV, 8mL of venous fasting blood sample was collected and the detailed protocols for blood collection, storage and biomedical measurement have been described previously by Zhao et al. [23] and other studies [7, 24]. Each participant’s MCV level was measured by medically-trained staff on automated analyzers available at the local hospitals or China Center for Disease Control and Prevention on automated analyzers. The calculation of MCV value is by multiplying the percent hematocrit by ten divided by the erythrocyte count [19]. We divided the subjects into normal (80≤MCV≤100 fl) and higher (MCV>100 fl) MCV groups.

Potential covariates including participants’ socio-demographic factors (age, Hukou, and education level) and health-related variables [cigarette smoking and alcohol drinking status, body mass index (BMI), chronic diseases and blood biochemical index] were collected at the baseline survey. Hukou is the registration system in China created in 1955 to restrict internal population movement, especially rural-to-urban migration [29] and stratified Chinese residents into three groups: agriculture, non-agriculture and unified residency. Education level was classified as four mutually exclusive categories: illiterate, elementary school, middle school and high school or above. Cigarette smoking was categorized into current, former and never. Alcohol drinking status was categorized into excessive alcohol consumption or not. Excessive alcohol consumption was defined as ≥210 g alcohol/week for men and ≥140 g alcohol/week for women [30]. BMI was obtained directly from weight and height and calculated by the standard formula: kg/m2. Chronic diseases including diabetes and hypertension were diagnosed by self-reported doctors’ diagnosis and medication use. In addition to MCV, other blood biochemical measurements including hemoglobin and hematocrit were measured on the same device as MCV. The detection procedures of other blood bioassays were reported in multiple related studies [7, 24]. In brief, blood lipids consisting of low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, total cholesterol (TC) and triglycerides (TG) were detected by enzymatic colorimetric test. Glycosylated hemoglobin (HbA1c) was measured by boronate affinity HPLC. Equation for estimated glomerular filtration rate (GRF) was calculated by the CKD-EPI creatinine equation [31]. Dyslipidemia was defined as TC≥240mg/dl or LDL-C≥160mg/dl or TG≥200mg/dl according to the "Guidelines for the Prevention and Treatment of Dyslipidemia in Adults in China (2016 Revised Edition)" [32].

Statistical analysis

Continuous variables were expressed as Mean (SD) for normally distributed data and median (interquartile range, IQR) for non-normally distributed data, respectively. Categorical variables were expressed as number (percentage). Baseline characteristics were compared between the male and female subjects by performing Student’s t-test and Wilcoxon test for normally and non-normally distributed continuous data, respectively. Additionally, Chi-square test was used for the comparison of categorical variables.

Multivariable linear regression models and logistic regression models were conducted to assess the relationship between MCV level and cognitive function. To assess potential confounding factors, covariate adjustments were used to gradually add demographic factors, health-related factors and blood biomedical factors into statistical models. The first statistical crude model (Model 1) was used to examine the raw association between MCV and cognitive function. The second statistical model (Model 2) was additionally included age, sex, Hukou and education level on the basis of Model 1. Model 3 was adjusted for health-related drinking status, smoking status, diabetes, hypertension, BMI and blood biochemical factors, including hemoglobin, dyslipidemia and GFR based on Model 2. To further investigate whether there was heterogeneity between male and females with varied MCV level, we have classified the subjects into 4 groups, i.e. female with 80≤MCV≤100 (low MCV), female with MCV>80 (high MCV), male with 80≤MCV≤100 (low MCV), and male with MCV>80 (high MCV). I2 and Q statistic were used to examine the heterogeneity between studies via meta analysis. I2<25% suggested little or no heterogeneity, 25–50% suggested moderate heterogeneity, and >50% was considered as high heterogeneity, respectively. As for Q statistic, p<0.1 indicated statistically significant [33]. Beta coefficient (β), standard error (SE) and p value for each variable in multiple linear regression models were presented. The analysis result of binary logistic regression is expressed as odds ratio (OR), 95% confidence interval (95% CI) and p value. All statistical processes were carried out using the Statistical Package for the Social Sciences (SPSS), version 22.0 (SPSS Inc, Chicago, IL, U.S.A.). Reported p values were two-tailed with a statistically significant level of p < 0.05.

Results

Baseline characteristics and cognitive function of recruited participants

Table 1 presented baseline characteristics and cognition measurement scores across male and female groups in 2011. A large part of the participants are rural households (80.2%). Only 11.5% of participants had high school or above education level. As for smoking status, men are more prone to consume cigarettes than women because more than 90% of female group members are never smoking. Female subjects have more excessive alcohol consumers than male subjects (p<0.005).The BMI of the female group was significantly higher than the male group (p<0.001). There were 489 diabetic patients and 2133 hypertensive patients, and the female group had significantly more participants afflicted with diabetes and hypertension than the male group (p<0.005). The hemoglobin level of women was significantly lower than men (p<0.001).

Table 1

Basic characteristics of the CHARLS respondents in 2011 stratified by sex.

Variables	Total	Male	Female	p value
Age (year) (Mean ± SD)	58.57 ± 9.05	59.21 ± 8.98	57.95 ± 9.07	<0.001b
Hukou, n (%)				0.003a
Agriculture	6573 (80.2)	3170 (78.8)	3403 (81.6)
Non-Agriculture	1582 (19.3)	832 (20.7)	750 (18.0)
Unified Residency	38 (0.5)	19 (0.5)	19 (0.5)
Educational level, n (%)				<0.001a
Illiterate	2051 (25.0)	459 (11.4)	1592 (38.2)
Elementary school	3428 (41.8)	1874 (46.6)	1554 (37.2)
Middle school	1771 (21.6)	1086 (27.0)	685 (16.4)
High school and above	946 (11.5)	604 (15.0)	342 (8.2)
Smoking status, n (%)				<0.001a
Current	2611 (31.9)	2360 (58.7)	251 (6.0)
Former	793 (9.7)	708 (17.6)	85 (2.0)
Never	4790 (58.5)	954 (23.7)	3836 (91.9)
Drinking Status, n (%)				0.004a
Excessive alcohol consumption	541 (6.6)	233 (5.8)	308 (7.4)
BMI (kg/m2) (Mean ± SD)	23.67 ± 3.96	23.11 ± 3.65	24.2 ± 4.16	<0.001b
Diabetes, n (%)	489 (6.0)	201 (5.0)	288 (6.9)	<0.001a
Hypertension, n (%)	2133 (26.0)	989 (24.6)	1144 (27.4)	0.004a
Dyslipidemia, n (%)	1423 (17.4)	694 (17.3)	729 (17.5)	0.413 a
Blood Biochemical Index (Mean ± SD)
Hemoglobin (g/dl)	14.57 ± 2.11	15.31 ± 2.02	13.86 ± 1.96	<0.001b
GFR (ml/min/1.73m2)	410.23 ± 239.20	411.56 ± 237.68	408.95 ± 240.69	0.690b
Global cognitive function, Median (IQR)	11.5 (8.0 to 14.0)	12.5 (9.5 to 14.5)	10.5 (7.0 to 13.5)	<0.001c
Episodic memory score, Median (IQR)	3.5 (2.5 to 4.5)	3.5 (2.5 to 5.0)	3.5 (2.5 to 4.5)	<0.001c
Mental status score, Median (IQR)	8.0 (5.0 to 10.0)	9.0 (6.0 to 11.0)	7.0 (4.0 to 10.0)	<0.001c
MCV (fl) (Mean ± SD)	92.37 ± 5.99	93.50 ± 6.23	91.29 ± 5.54	<0.001b

a. Chi-square test

b. Student’s t test

c.Wilcoxon test

Hukou is the registration system in China created in 1955 to restrict internal population movement; BMI, body mass index; HbA1c, glycated hemoglobin; GRF, equation for estimated glomerular filtration rate; MCV, Mean Corpuscular Volume.

The median score of global cognitive function, episodic memory and mental status were significantly lower from female participants compared to male participants (p<0.001). The mean MCV of female respondents [mean (SD) = 91.29(5.54)] were significantly lower than the male subjects [mean (SD) = 93.50(6.23)] (p<0.001).

Z-score of cognitive function

Table 2 demonstrated mean and standard deviation of cognitive Z scores for all participants stratified by MCV level and sex. Approximately 90% (n = 7440) of the examined subjects were in the normal MCV level (80≤MCV≤100 fl) and 10% (n = 756) were in the second status (MCV >100 fl). Moreover, more men (n = 521) were in the abnormal MCV level than women (n = 235). We defined Z-score of cognition (Cognition Z) = 0 as the average cognition score for total participants. The mean Z scores of global cognitive function and episodic memory for participants with MCV higher than 100 were significantly lower than those with normal MCV level for the whole subjects. When adjusted by sex, the mean Z scores of global cognitive function, episodic memory and mental status were all significantly lower for those with MCV above 100 fl in comparison with subjects having normal MCV level for both male and female participants.

Table 2

Mean and standard deviation of cognitive Z scores stratified by MCV levels.

Cognition Variable	Global cognitive function		Episodic Memory		Mental status
	Mean ± SD	p value	Mean ± SD	p value	Mean ± SD	p value
Total
80≤MCV≤100 (n = 7440)	0.02 ± 1.67	<0.001	0.02 ± 1.00	<0.001	0.01 ± 1.00	0.159
MCV >100 (n = 756)	-0.22 ± 1.61		-0.16 ± 0.99		-0.06 ± 0.98
Male
80≤MCV≤100 (n = 3502)	0.32 ± 1.53	<0.001	0.05 ± 0.96	<0.001	0.27 ± 0.89	0.001
MCV >100 (n = 521)	-0.01 ± 1.52		-0.13 ± 0.98		0.12 ± 0.91
Female
80≤MCV≤100 (n = 3938)	-0.24 ± 1.75	<0.001	-0.01 ± 1.03	0.001	-0.23 ± 1.04	0.003
MCV >100 (n = 235)	-0.68 ± 1.73		-0.22 ± 1.02		-0.46 ± 1.03

Data are presented as mean ± SD. Z scores of domain-specific episodic memory and mental status, as well global cognitive function were derived based on the mean and standard deviation of respective cognitive scores.

Cross-sectional association between MCV and cognitive function

Multivariable linear regression model was firstly performed to explore the association between MCV level and global cognitive function, as well as domain-specific episodic memory and mental status. As shown in Table 3, for the whole subjects, MCV level was significantly negatively associated with global cognitive function (β = -0.038, -0.040 and -0.038 for model 1, 2 and 3, respectively, p<0.05) and episodic memory (β = -0.055, -0.035 and -0.036 for model 1, 2 and 3, respectively, p<0.05). Additionally, for male participants, higher MCV level was associated with reduced scores for both global cognitive function, episodic memory and mental status even adjusting for all additional covariates (model3) (β = -0.061, -0.051 and -0.049, respectively, p<0.05). After adjusted for all additional covariates, no association was observed between MCV level and mental status for the whole subjects, also there was no association between MCV level and global cognitive function, as well as episodic memory for female subjects.

Table 3

Multiple linear regression model testing the association between MCV and cognitive function.

	Global cognitive function		Episodic memory		Mental status
	β (95% CI)	p value	β (95% CI)	p value	β (95% CI)	p value
Total (n = 8196)
Model 1	-0.038 (-0.020, -0.001)	0.032	-0.055 (-0.015, -0.003)	0.002	-0.009 (-0.007, 0.004)	0.631
Model 2	-0.040 (-0.020, -0.003)	0.010	-0.035 (-0.011, 0.000)	0.036	-0.031 (-0.010, 0.000)	0.047
Model 3	-0.038 (-0.019, -0.002)	0.016	-0.036 (-0.012, 0.000)	0.036	-0.027 (-0.010, 0.001)	0.093
Male (n = 4023)
Model 1	-0.087 (-0.033, -0.009)	0.001	-0.071 (-0.018, -0.003)	0.006	-0.074 (-0.018, -0.003)	0.004
Model 2	-0.062 (-0.026, -0.004)	0.006	-0.047 (-0.014, 0.000)	0.051	-0.056 (-0.015, -0.001)	0.016
Model 3	-0.061 (-0.025, -0.004)	0.008	-0.051 (-0.015, -0.001)	0.033	-0.049 (-0.014, 0.000)	0.037
Female (n = 4173)
Model 1	-0.053 (-0.033, -0.001)	0.036	-0.052 (-0.020, 0.000)	0.040	-0.036 (-0.016, 0.002)	0.149
Model 2	-0.018 (-0.019, 0.007)	0.390	-0.023 (-0.013, 0.004)	0.321	-0.007 (-0.009, 0.007)	0.742
Model 3	-0.016 (-0.019, 0.008)	0.449	-0.020 (-0.013, 0.005)	0.402	-0.007 (-0.010, 0.007)	0.742

Model 1: Crude model. Model 2: adjusted for model 1+ age, sex, Hukou and education level. Model 3: adjusted for model 2+ drinking status, smoking status, diabetes, hypertension, hemoglobin, dyslipidemia, GFR and BMI.

Furthermore, binary logistic regression was conducted to confirm the association between MCV level and cognitive function (Table 4). After adjusting for potential confounding factors and being referenced to the normal MCV level, the higher MCV level (MCV>100 fl) was associated with poor global cognitive function (OR = 1.601; 95% CI = 1.198–2.139; p = 0.001), episodic memory (OR = 1.679; 95% CI = 1.281–2.201; p<0.001), and mental status (OR = 1.422; 95% CI = 1.032–1.959; p = 0.031) in the Model 3. When testing this association by sex after adjusting for all potential covariates, the significant relationship between higher MCV level with worse global cognitive function (OR = 1.885; 95% CI = 1.329–2.675; p<0.001), episodic memory (OR = 1.690; 95% CI = 1.211–2.358; p = 0.002) and mental status (OR = 1.544; 95% CI = 1.034–2.306; p = 0.034) was observed in the male subjects. In contrast, in female subjects, higher MCV level was only associated with poor episodic memory (OR = 1.729; 95% CI = 1.079–2.770; p = 0.023).

Table 4

Binary logistic regression model testing the association between MCV level and cognitive performance.

	Global cognitive function		Episodic memory		Mental status
	OR (95% CI)	p value	OR (95% CI)	p value	OR (95% CI)	p value
Total
Model 1
80≤MCV≤100 (ref.)	1 (Reference)		1 (Reference)		1 (Reference)
MCV>100	1.382 (1.072, 1.782)	0.013	1.682 (1.306, 2.165)	0.000	1.118 (0.850, 1.470)	0.427
Model 2
80≤MCV≤100 (ref.)	1 (Reference)		1 (Reference)		1 (Reference)
MCV>100	1.621 (1.219, 2.157)	0.001	1.674 (1.281, 2.188)	0.000	1.444 (1.052, 1.981)	0.023
Model 3
80≤MCV≤100 (ref.)	1 (Reference)		1 (Reference)		1 (Reference)
MCV>100	1.601 (1.198, 2.139)	0.001	1.679 (1.281, 2.201)	<0.001	1.422 (1.032, 1.959)	0.031
Male
Model 1
80≤MCV≤100 (ref.)	1 (Reference)		1 (Reference)		1 (Reference)
MCV>100	1.884 (1.368, 2.596)	<0.001	1.756 (1.283, 2.404)	<0.001	1.558 (1.076, 2.256)	0.019
Model 2
80≤MCV≤100 (ref.)	1 (Reference)		1 (Reference)		1 (Reference)
MCV>100	1.940 (1.376, 2.735)	<0.001	1.715 (1.235, 2.381)	0.001	1.647 (1.110, 2.445)	0.013
Model 3
80≤MCV≤100 (ref.)	1 (Reference)		1 (Reference)		1 (Reference)
MCV>100	1.885 (1.329, 2.675)	<0.001	1.690 (1.211, 2.358)	0.002	1.544 (1.034, 2.306)	0.034
Female
Model 1
80≤MCV≤100 (ref.)	1 (Reference)		1 (Reference)		1 (Reference)
MCV>100	1.273 (0.821, 1.975)	0.281	1.790 (1.152, 2.780)	0.010	1.295 (0.832, 2.017)	0.252
Model 2
80≤MCV≤100 (ref.)	1 (Reference)		1 (Reference)		1 (Reference)
MCV>100	1.111 (0.674, 1.831)	0.679	1.658 (1.043, 2.636)	0.032	1.143 (0.684, 1.912)	0.610
Model 3
80≤MCV≤100 (ref.)	1 (Reference)		1 (Reference)		1 (Reference)
MCV>100	1.112 (0.667, 1.855)	0.685	1.729 (1.079, 2.770)	0.023	1.160 (0.685, 1.964)	0.581

Model 1: Crude model. Model 2: adjusted for model 1+ age, sex, Hukou and education level. Model 3: adjusted for model 2+ drinking status, smoking status, diabetes, hypertension, hemoglobin, dyslipidemia, GFR and BMI.

Heterogeneity test via meta analysis

Our results suggested high heterogeneity existed for subjects with varied MCV level for both male and female subjects (I2 = 64.8%, p = 0.059, I2 = 60.4%, p = 0.080, and I2 = 63.7%, p = 0.064 for global cognitive function, episodic memory and mental status, respectively), this further suggested that the association between MCV and cognitive function should be explored by sex.

Discussion

In this large cross-sectional study on middle-aged and older Chinese residents, we found that higher MCV level (>100fl) was associated with poor global cognitive function, as well as domain-specific episodic memory and metal status, but, surprisingly, after testing this correlation by sex, the MCV’s main effect on these cognitive domains fully remained in the male group, in female subjects, only a significant association between higher MCV level and episodic memory was observed.

Existing studies demonstrated that erythrocyte volume seemed to be negatively associated with cognitive performance [20, 21]. For instance, Danon et al. [20] reported that that the elderly with smaller erythrocytes performed better in the delayed recall tasks. Additionally, a recent longitudinal study conducted by Gamaldo et al. [21] showed that higher baseline MCV levels were also significantly associated with accelerated rates of decline on tasks of global mental status and long delay memory. Similarly, based on data from UK Biobank dataset, there was a significant negative correlation for MCV with reaction time [12]. Our present findings demonstrated that higher MCV level (>100 fl) was associated with poor global cognitive function, and domain specific episodic memory and metal status, which is partially consistent with these previous studies. It is suggested that in Chinese subjects over 45 years old, higher MCV level might be associated poor cognitive performance.

There are several possible explanations for the association between higher MCV level and poorer cognitive function. Mohanty et al. [34] reported that 15% of erythrocytes in AD subjects are elongated and have altered membrane structure. Morphological changes of erythrocytes might result in decreased deformability, disordered physical state of membrane proteins, and oxidation imbalance [22, 35], consequently affecting cognitive function [22]. In addition, the relationship between larger MCV and the poor performance of cognition could be attributed to pathophysiological changes in carrying oxygen and glucose to cerebral tissues [36], thus adversely impacting on cognitive performance.

Currently, there was no evidence about the relationship between MCV level and cognitive function stratified by sex. When it comes to sex discrepancy for the association between anemia and cognitive function, a previous study conducted in USA found that there was a relationship between low or high hemoglobin with worse global cognition, which was greater in women compared to men [13]. In contrast, by using data from CHARLS, Qin et al. [7] recently demonstrated that there was a negative significant association between anemia and global cognitive function, episodic memory and TICS, which is independently from sex. Our study demonstrated that there might be sex-specific association between higher MCV level and global cognitive function, as well as domain-specific episodic memory and mental status. To be specific, the association between higher MCV level and poor global cognitive function, as well as domain-specific episodic memory and mental status might be more pronounced in male subjects than in female subjects. Generally, in non-demented subjects, women tend to outperform men in verbal-based episodic memory tasks [37], and a steeper age-associated episodic memory decline for males [38]. However, the female superiority in episodic memory would be decreasing with advancing age [39]. This might explain, at least partially the fact that the higher MCV level was associated with poor episodic memory was observed both in male and female subjects. The differences in sample size, participants’ characteristics and cognitive measurements also might contribute to these sex-specific effects. The lack of the association between MCV level and global cognitive function, as well as mental status in female subjects might be explained by the hormone discrepancy and physiological status which leads to estrogen loss induced by menopause, as well as the cross-sectional study design. Nevertheless, our current findings suggest that for middle-aged and elderly Chinese male subjects with higher MCV level than 100 fl, special attention on cognitive function might be warranted to prevent cognitive decline or dementia. Meanwhile, it is also suggested that further studies are still required to clarify the sex-specific association between MCV level and cognitive performance both cross-sectionally and longitudinally.

There are some strengths and limitations deserving to be mentioned. Firstly, to the best of our knowledge, the current study is the very first study to explore the cross-sectional association between normal and higher MCV level with cognition in Chinese population. Secondly, our study has a relatively large sample size, which allows a much greater possibility of making reasonable conclusions and providing support for further investigations on the underlying association between MCV and cognitive performance. Besides, CHARLS offers a broad range of potential confounders including blood biomedical parameters measured by standardized protocols and rigid quality control [23, 40], social-demographic and health-related factors. However, this study also has some limitations. In the first place, the majority of the recruited participants are lowly educated and 82.1% of them hold rural Hukou, which might not be representative of the general Chinese population. For example, Jia et al. [3] reported that a greatly higher prevalence of dementia and AD was found in rural areas than in urban ones and they proposed an explanation for the urban-rural differences is education. Hence, the extrapolation of this study still requires caution. Second of all, a relatively limited number of cognitive domains were measured, the association of MCV level with other cognitive function tests’ scores such as reaction time and prospective memory remains unclear. Thirdly, some other variables including vitamin B12, folate level and thyroid function are also closely associated with MCV level [41], how these factors might affect the association between MCV and cognitive performance remain unanswered. Last but not least, we only explored the association between higher MCV level and cognitive performance cross-sectionally, whether high MCV level might play a causal role in adversely affecting cognitive function remain unclear.

Conclusions

In conclusion, in this large population-based cross-sectional study, we demonstrated that the association between higher MCV level and poor episodic memory existed both in male and female subjects, while the association between higher MCV level and poor global cognitive performance, as well as mental status only existed in male subjects. Our study suggests that for middle-aged and elderly Chinese male subjects with MCV level higher than 100 fl, it is necessary for them to pay special attention to cognitive function, in order to prevent cognitive decline. Meanwhile, further studies are required to evaluate the association between MCV level and cognitive performance by considering sex into consideration both cross-sectionally and longitudinally.

22 Sep 2020

PONE-D-20-25060

The cross-sectional association between mean corpuscular volume level and cognitive function in Chinese over 45 years old: evidence from the China Health and Retirement Longitudinal Study

PLOS ONE

Dear Dr. Wan,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Nov 06 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Claudia K. Suemoto

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: N/A

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: COMMENTS TO AUTHOR(S):

In this well-designed and well-written manuscript, Yao Chen, et al, examined the cross-sectional baseline association between mean corpuscular volume (MCV) and cognitive performance in the China Health and Retirement Longitudinal Study (CHARLS). A few points need clarification and improvement to assist in putting the study in context with current literature.

(1) INTRODUCTION

There are many more evidence suggesting that there was no association between anemia or low hemoglobin and cognitive impairment. You just mentioned 2 studies. I suggest that you also evaluate other studies in order to make it clear that the topic remains unfinished. For example: https://doi.org/10.1111/j.1440-1819.2012.02347.x, https://doi.org/10.1176/appi.neuropsych.19040088, https://doi.org/10.1017/S1041610211001724.

The choice of specifically selecting MCV values > 80 is not clear in the text. I also suggest adding a justification for the study's interest in focusing on high levels of MCV. Example: There is biological plausibility for this hypothesis due to the fact that larger red blood cells may have more difficulty passing through small capillaries, compromising to deliver adequate amounts of oxygen to cerebral tissues.

(2) METHODS

SAMPLE

It is not clear why a total of 2,120 participants were excluded due to missed formation on covariates of minor importance. The default is just to treat them as NA (missing values). So, I would like to recommend redoing the sample excluding only patients with no data on cognition (n=4,171), MCV (n=4,238), and those who didn`t attend the study`s criteria (younger than 45 (n= 223), subjects with memory-related disease (n= 88), MCV less than 80 (n=690)). Thus, I would suggest the use of a larger sample with 8,298 individuals and the management of the absence of data in the other variables such as NA.

VARIABLES

I notice the lack of important variables that are more related to cognitive performance: income, thyroid function, medications that could alter cognitive function (i.e., antipsychotic medications, antiparkinsonian agents, and anticonvulsants), major depressive disorder, cerebrovascular disease (i.e., history of ischemic attack, ateriosclerotic cerebrovascular disease, cerebral arteriosclerosis, and stroke), GFR rates (i.e., using CKD-EPI equation), vitamin B12, folate. If you have information about any of these, I strongly recommend using them. Otherwise, use what you have with some suggestions for improvement below.

I didn`t understand the value of the information about drinking status: current, former and never. In this form, I do not see any advantage to use drinking status in analysis. The most valuable information for cognition is if the individual has excessive alcohol use. It was defined as ≥210 g alcohol/week for men and ≥140 g alcohol/week for women.

It is not clear why a large number of biochemical variables with no specific advantage were chosen. Excessive or unnecessary information tends to weaken your analysis model. I suggest thinking about cleaning these polluting variables. I would highlight the importance of hemoglobin only, but it can also be chosen to be presented as an anemia Yes or No variable (according to WHO criteria). If you have data on thyroid function, folate and vitamin B12, I suggest including it. I highlight the importance of information on folate and vitamin b12 as it is a study that specifically analyzes high levels of MCV. Both vitamin B12 and folate cause macrocytosis and are nutrients that have been linked to neural health with evidence of affecting cognition by itself. That is, they are extremely relevant as potential confounders of your findings.

There is no need for glycemic or glycated hemoglobin data if you have information about diabetes in the population studied. There is no need for BPD or SBP if you have information about hypertension in the population studied. If you insist on presenting lipids in your analysis, I recommend condensing the information into a single variable: dyslipidemia (Yes or No). For renal function, I recommend to use GFR rates (i.e., using CKD-EPI equation). I do not recommend to give so much importance or use any of the others biochemical variable in analytic models.

ANALYSIS

I would like to suggest that you prefer to use the term "crude" instead of "raw" (line 184). I take this opportunity to recommend that Model 1 to be considered a Crude Model should not be adjusted for any variable.

Most important: Your analysis suffers from overadjustment and unnecessary adjustment. It can obscure a true effect or create an apparent effect when none exists.

Almost all respondents (99.3%) has got married. I would recommend to exclude the adjustment for marital status.

My personal recommendations:

- Model 1 (Crude Model): without any adjustment.

- Model 2: adjusted for age, sex, Hukou, education level. (+ income if you have this data)

- Model 3: Model 2 + adjusted for smoking status, BMI, diabetes, hypertension, hemoglobin (or anemia according to WHO criteria), dyslipidemia (according to lipids measures). (+ excessive alcohol use, + thyroid function, + GFR if you have this data).

Further models with your other variables can be presented in text as sensitivity analyzes.

RESULTS

Try to clear table 1 with only the relevant variables (as stated in the previous observations).

For table 3, it is better and more recommended to present the beta-coefficient results without SE, but with 95% CI.

If you don’t have data about important variables (i.e., income, thyroid function, vitamin B12, folate) it is also important to point out this is a limitation in your study.

Reviewer #2: In aging society, the research question is very relevant and worth of investigation. So my comments involves only the study design and the statical analysis.

1. The sample selection criteria reported in Figure 1 lead to the exclusion of all respondents lacking cognitive scores. This is a clear problem if those with severe cognitive decline are excluded from testing (as in HRS), leading to a downward bias (in absolute terms) in the association bettie MCV and cognition. The authors can test it by regressing a dummy which indicates whether the respondent lacks the cognitive scores on the MCV value.

2. The authors do not explain why they are using both OLS and logistic regression. I suspect that they want to test for the presence of non-linearity. Moreover, they do not explain well what outcome variable the use for the logistic regression. I assume from the end of page 6, it is a dummy for poor cognitive score set at the median value. If so, I strongly suggest to set a more reasonable threshold for poor cognition based on the neuropsychological literature. For instance, I would suggest the first quartile or quintile or a cognitive score one SD below the mean.

3. Regarding the MCV, I don't get why the authors excludes values below 80.

4. The selection of the control sounds arbitrary and also potentially endogenous, especially those included in Model 2 and 3. What is the logic behind? What are the determinant of the MCV?

4. As it stands, it not very clear why the results differ by sex. Have the author tried to test the difference using pool data and an interaction term between sex and MCV?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: José Benedito Ramos Valladão Júnior

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

16 Oct 2020

Dear Dr. Suemoto:

In regards to our Manuscript PONE-D-20-25060 entitled The cross-sectional association between mean corpuscular volume level and cognitive function in Chinese over 45 years old: evidence from the China Health and Retirement Longitudinal Study, we would like to thank the reviewers and the Editors for their thorough, insightful and thoughtful review of our manuscript. Below, we have addressed each of the comments in a point-by-point manner. Changes to the text of the manuscript are highlighted in RED in the revised version of the manuscript.

Especially, we have corrected the analytic models per the 1st reviewer’s suggestion, we have also redefined the poor cognitive function as the Z score less than the lowest quartiles of respective Z score per the 2nd reviewer’s suggestion. All results have been updated and the overall trends of the findings remained. We have also made sure that our manuscript meets PLOS ONE's style requirements.

We feel the manuscript is much improved after revising based on the constructive comments from the reviewers and hope that you will find it acceptable for publication in Plos One. We are also willing to make any further revisions if required.

Regards,

Zhongxiao Wan, PhD, Professor

Department of Nutrition and Food Hygiene

School of Public Health, Soochow University

Reviewers’ and editors’ comments are listed below.

Reviewer #1: COMMENTS TO AUTHOR(S):

In this well-designed and well-written manuscript, Yao Chen, et al, examined the cross-sectional baseline association between mean corpuscular volume (MCV) and cognitive performance in the China Health and Retirement Longitudinal Study (CHARLS). A few points need clarification and improvement to assist in putting the study in context with current literature.

(1) INTRODUCTION

There are many more evidence suggesting that there was no association between anemia or low hemoglobin and cognitive impairment. You just mentioned 2 studies. I suggest that you also evaluate other studies in order to make it clear that the topic remains unfinished. For example: https://doi.org/10.1111/j.1440-1819.2012.02347.x, https://doi.org/10.1176/appi.neuropsych.19040088, https://doi.org/10.1017/S1041610211001724.

Response: Thank you very much for providing us more evidence in regards to the association of anemia or low hemoglobin with cognitive impairment. We have put these valuable references into our revised introduction section.

The choice of specifically selecting MCV values > 80 is not clear in the text. I also suggest adding a justification for the study's interest in focusing on high levels of MCV. Example: There is biological plausibility for this hypothesis due to the fact that larger red blood cells may have more difficulty passing through small capillaries, compromising to deliver adequate amounts of oxygen to cerebral tissues.

Response: We greatly appreciate this comment. Sentences have been added in the introduction section.

(2) METHODS

SAMPLE

It is not clear why a total of 2,120 participants were excluded due to missed formation on covariates of minor importance. The default is just to treat them as NA (missing values). So, I would like to recommend redoing the sample excluding only patients with no data on cognition (n=4,171), MCV (n=4,238), and those who didn`t attend the study`s criteria (younger than 45 (n= 223), subjects with memory-related disease (n= 88), MCV less than 80 (n=690)). Thus, I would suggest the use of a larger sample with 8,298 individuals and the management of the absence of data in the other variables such as NA.

Response: Thank you very much! We have followed your suggestion, and we have only excluded those with no data on MCV (6176), cognitive score (2233). We further exlcuded those who didn`t attend the study`s criteria (younger than 45 (n= 223), MCV less than 80 (n=783), and subjects with memory-related disease (n= 97). Eventually, a total of 4023 male subjects and 4173 famale subjects were included in the final analysis. Additionally, we have managed the absence of the other variables as missing values.

VARIABLES

I notice the lack of important variables that are more related to cognitive performance: income, thyroid function, medications that could alter cognitive function (i.e., antipsychotic medications, antiparkinsonian agents, and anticonvulsants), major depressive disorder, cerebrovascular disease (i.e., history of ischemic attack, ateriosclerotic cerebrovascular disease, cerebral arteriosclerosis, and stroke), GFR rates (i.e., using CKD-EPI equation), vitamin B12, folate. If you have information about any of these, I strongly recommend using them. Otherwise, use what you have with some suggestions for improvement below.

Response: Thank you very much! In regards to the important variables mentioned here, we have calculated GFR rates, however, we don’t have other factors including income, thyroid function, medications that could alter cognitive function (i.e., antipsychotic medications, antiparkinsonian agents, and anticonvulsants), major depressive disorder, cerebrovascular disease (i.e., history of ischemic attack, ateriosclerotic cerebrovascular disease, cerebral arteriosclerosis, and stroke), vitamin B12 and folate. We have added GFR in the new model for the improvement, we have also discussed the limitations for the absence of other factors in the revised discussion section.

I didn’t understand the value of the information about drinking status: current, former and never. In this form, I do not see any advantage to use drinking status in analysis. The most valuable information for cognition is if the individual has excessive alcohol use. It was defined as ≥210 g alcohol/week for men and ≥140 g alcohol/week for women.

Response: Thank you very much! We have reclassified the drinking status into excessive alcohol consumption or not based on your suggestion.

It is not clear why a large number of biochemical variables with no specific advantage were chosen. Excessive or unnecessary information tends to weaken your analysis model. I suggest thinking about cleaning these polluting variables. I would highlight the importance of hemoglobin only, but it can also be chosen to be presented as an anemia Yes or No variable (according to WHO criteria). If you have data on thyroid function, folate and vitamin B12, I suggest including it. I highlight the importance of information on folate and vitamin b12 as it is a study that specifically analyzes high levels of MCV. Both vitamin B12 and folate cause macrocytosis and are nutrients that have been linked to neural health with evidence of affecting cognition by itself. That is, they are extremely relevant as potential confounders of your findings.

Response: We greatly appreciate your comment. We totally agree that vitamin B12 and folate are closely associated with MCV level, however, our study don’t have these data. We have discussed this as one of our study limitations. Additionally, we have only included hemaglobin, GFR and dyslipidemia of the blood biochemical index in the adjusted models and excluded all other unrelated factors per your suggestion.

There is no need for glycemic or glycated hemoglobin data if you have information about diabetes in the population studied. There is no need for BPD or SBP if you have information about hypertension in the population studied. If you insist on presenting lipids in your analysis, I recommend condensing the information into a single variable: dyslipidemia (Yes or No). For renal function, I recommend to use GFR rates (i.e., using CKD-EPI equation). I do not recommend to give so much importance or use any of the others biochemical variable in analytic models.

Response: Thank you very much! In our analytic models, we have removed glycemic hemoglobin, DBP and SBP data. We have presented lipids data into dyslipidemia (yes or no) based on your suggestion. We have added GFR rates in our new model. We greatly appreciate your suggestion!

ANALYSIS

I would like to suggest that you prefer to use the term "crude" instead of "raw" (line 184). I take this opportunity to recommend that Model 1 to be considered a Crude Model should not be adjusted for any variable.

Response: Thank you very much! We have made Model 1 as a crude model.

Most important: Your analysis suffers from overadjustment and unnecessary adjustment. It can obscure a true effect or create an apparent effect when none exists.

Almost all respondents (99.3%) has got married. I would recommend to exclude the adjustment for marital status.

Response: Thank you very much! Marital status adjustment has been excluded.

My personal recommendations:

- Model 1 (Crude Model): without any adjustment.

- Model 2: adjusted for age, sex, Hukou, education level. (+ income if you have this data)

- Model 3: Model 2 + adjusted for smoking status, BMI, diabetes, hypertension, hemoglobin (or anemia according to WHO criteria), dyslipidemia (according to lipids measures). (+ excessive alcohol use, + thyroid function, + GFR if you have this data).

Further models with your other variables can be presented in text as sensitivity analyzes.

Response: We greatly appreciate your comments. Followed with your suggestion, model 1 was the crude model. Model 2 was further adjusted for age, sex, Hukou and education level. We had no income information, thus income was not included in the model 2. Model 3 was further adjusted for smoking status, BMI, diabetes, hypertension, hemoglobin, dyslipidemia, excessive alcohol consumption or not, and GFR.

RESULTS

Try to clear table 1 with only the relevant variables (as stated in the previous observations).

Response: We greatly appreciate your comments. We have revised table 1 with only the relevant variables included.

For table 3, it is better and more recommended to present the beta-coefficient results without SE, but with 95% CI.

Response: We have revised table 3, 95% CI data has been provided.

If you don’t have data about important variables (i.e., income, thyroid function, vitamin B12, folate) it is also important to point out this is a limitation in your study.

Response: Thank you! We have discussed this as one of our study limitations.

Reviewer #2: In aging society, the research question is very relevant and worth of investigation. So my comments involves only the study design and the statical analysis.

1. The sample selection criteria reported in Figure 1 lead to the exclusion of all respondents lacking cognitive scores. This is a clear problem if those with severe cognitive decline are excluded from testing (as in HRS), leading to a downward bias (in absolute terms) in the association bettie MCV and cognition. The authors can test it by regressing a dummy which indicates whether the respondent lacks the cognitive scores on the MCV value.

Response: We totally agree with your concern. We have compared the MCV values for those with and without the cognitive score, as shown below. There was no difference for MCV levels between the two groups. Thus, we might not be worried about this issue.

The comparisons of MCV levels for those with and without cognitive score

N Mean ± SD

Without Cognitive score 1898 92.92 ± 6.49

With Cognitive score 8196 92.37 ± 5.99

2. The authors do not explain why they are using both OLS and logistic regression. I suspect that they want to test for the presence of non-linearity. Moreover, they do not explain well what outcome variable the use for the logistic regression. I assume from the end of page 6, it is a dummy for poor cognitive score set at the median value. If so, I strongly suggest to set a more reasonable threshold for poor cognition based on the neuropsychological literature. For instance, I would suggest the first quartile or quintile or a cognitive score one SD below the mean.

Response: Thank you very much! Generally, linear regression gives you a continuous output, but logistic regression provides a constant output. The combination of OLS and logistic regression will compensate with each other, and make the associations much clearer. It is very common for utilizing both OLS and logistic regression analysis to solve questions. In regards to the outcome variable for logistic regression analysis, we totally agree that setting a Z-score <0 might not be the best, we have corrected this. In our revised manuscript, we have set Z score < the lowest quartile (i.e. p25) as as poorer cognitive function per your suggestion.

3. Regarding the MCV, I don't get why the authors excludes values below 80.

Response: Thank you! In the revised introduction section, we have added background information why we only included those with MCV≥80 fl. In brief, larger red blood cells may have more difficulty passing through small capillaries, compromising to deliver adequate amounts of oxygen to cerebral tissues, there is biological plausibility that higher MCV level than the normal level might be asscociated with worse cognitive performance.

4. The selection of the control sounds arbitrary and also potentially endogenous, especially those included in Model 2 and 3. What is the logic behind? What are the determinant of the MCV?

Response: Thank you! As responded to the first reviewer, we have corrected our control variables with those only closely associated with MCV were included (i.e. Model 1: Crude model. Model 2: adjusted for model 1+ age, sex, Hukou and education level. Model 3: adjusted for model 2+ drinking status, smoking status, diabetes, hypertension, hemoglobin, dyslipidemia, GFR and BMI.). It should be realized that some other variables which are also closely related to MCV including vitamin B12, folate and income were not controlled in our analysis because we couldn’t get these data. We have discussed this as one of our study limitation.

5. As it stands, it not very clear why the results differ by sex. Have the author tried to test the difference using pool data and an interaction term between sex and MCV?

Response: We greatly appreciate this comment. We have utilized the multivariate logistic regression model to test the interaction term between sex and MCV. As shown in newly added table 5, there was no interaction between MCV and sex in fully adjusted model, it is suggested that both sex and MCV affected cognitive function independently.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

9 Nov 2020

PONE-D-20-25060R1

The cross-sectional association between mean corpuscular volume level and cognitive function in Chinese over 45 years old: evidence from the China Health and Retirement Longitudinal Study

PLOS ONE

Dear Dr. Wan,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Dec 24 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Claudia K. Suemoto

Academic Editor

PLOS ONE

Additional Editor Comments (if provided):

Please, address the concern from Reviewer 2.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: In my opinion, the revision carried out by Wan and collaborators met the need for clarification, adjustment and improvement of methodology and analysis.

Stylistically, I personally prefer to just emphasize the central results and main tables of the study. Thus, optionally, all the section "Combined effect of MCV and sex on the prediction of cognitive performance" and table 5 can be eliminated. Leaving only described in the text this piece: “As a sensitivity analysis, we have utilized the multivariate logistic regression model to test the interaction term between sex and MCV. No combined effect was observed between MCV and sex. It is suggested that both sex and MCV affected cognitive function independently”.

My compliments to Wan and collaborators to considering our comments, doing a great revision work and an excellent final article.

Reviewer #2: I'm fine with review which almost my comments.

The only final concern is how to reconcile the heterogeneity by sex in T4 with the evidence in T5 which suggest no heterogeneity, unless I missed something. In other words, the fully interacted model (T4) gives different results from the model with the sex interaction term (T5).

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

15 Nov 2020

Dear Dr. Suemoto:

In regards to our Manuscript PONE-D-20-25060R1 entitled The cross-sectional association between mean corpuscular volume level and cognitive function in Chinese over 45 years old: evidence from the China Health and Retirement Longitudinal Study, we would like to thank the reviewers and the Editors for their thorough, insightful and thoughtful review of our manuscript. Below, we have addressed each of the comments in a point-by-point manner. Changes to the text of the manuscript are highlighted in BLUE in the revised version of the manuscript.

Specially, both of the reviewers were concerned about table 5. We have used meta analysis to explore the potential heterogeneity (see below response to #2 reviewer in detail, our results suggested high heterogeneity existed for subjects with varied MCV level for both male and female subjects, this further supported that the association between MCV and cognitive function should be explored by sex. We think it might not be necessary to include this section of results in the final manuscript, and we have also removed table 5 per the first reviewer’s suggestion.

We feel the manuscript is much improved after revising based on the constructive comments from the reviewers and hope that you will find it acceptable for publication in Plos One.

Regards,

Zhongxiao Wan, PhD, Professor

Department of Nutrition and Food Hygiene

School of Public Health, Soochow University

Reviewers’ and editors’ comments are listed below, followed by our responses in bold-faced type.

Reviewer #1: In my opinion, the revision carried out by Wan and collaborators met the need for clarification, adjustment and improvement of methodology and analysis.

Stylistically, I personally prefer to just emphasize the central results and main tables of the study. Thus, optionally, all the section "Combined effect of MCV and sex on the prediction of cognitive performance" and table 5 can be eliminated. Leaving only described in the text this piece: “As a sensitivity analysis, we have utilized the multivariate logistic regression model to test the interaction term between sex and MCV. No combined effect was observed between MCV and sex. It is suggested that both sex and MCV affected cognitive function independently”.

My compliments to Wan and collaborators to considering our comments, doing a great revision work and an excellent final article.

Response: Thank you very much! We have followed your suggestion, and removed table 5, with the results being described in the text. It should be mentioned that we have used meta analysis to explore the potential heterogeneity (see below response to #2 reviewer in detail), with heterogeneity were observed. This further confirmed that we should explore the association between MCV and cognitive function by sex.

Reviewer #2: I'm fine with review which almost my comments.

The only final concern is how to reconcile the heterogeneity by sex in T4 with the evidence in T5 which suggest no heterogeneity, unless I missed something. In other words, the fully interacted model (T4) gives different results from the model with the sex interaction term (T5).

Response: Thank you very much! We totally agree with your confusion, and we apologize for not solving this issue clearly in the first revision. In this revised manuscript, we have classified the subjects into 4 groups, i.e. female with 80≤MCV≤100 (low MCV), female with MCV>80 (high MCV), male with 80≤MCV≤100 (low MCV), and male with MCV>80 (high MCV). We have used meta analysis to explore the potential heterogeneity, I2<25% suggested little or no heterogeneity, 25–50% suggested moderate heterogeneity, and >50% was considered as high heterogeneity, respectively. As for Q statistic, p<0.1 indicated statistically significant. As shown in below figure 1, our results suggested high heterogeneity existed for subjects with varied MCV level for both male and female subjects, this further suggested it is necessary to look at the association between MCV and cognitive function by sex. We have mentioned this part of results in the text and we have also removed table 5 per the first reviewer’s suggestion.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

18 Nov 2020

The cross-sectional association between mean corpuscular volume level and cognitive function in Chinese over 45 years old: evidence from the China Health and Retirement Longitudinal Study

PONE-D-20-25060R2

Dear Dr. Wan,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Claudia K. Suemoto

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

23 Nov 2020

PONE-D-20-25060R2

The cross-sectional association between mean corpuscular volume level and cognitive function in Chinese over 45 years old: evidence from the China Health and Retirement Longitudinal Study

Dear Dr. Wan:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Claudia K. Suemoto

Academic Editor

PLOS ONE