Prevalence and pattern of cognitive dysfunction in young adults and middle-aged patients with type-2 diabetes

ABSTRACT

Background and Aims:

With the world’s aging population, twin epidemics of type-2 diabetes (T2D) and dementia take a great toll on the healthcare burden. T2D carries a 2–3 times greater risk of developing cognitive impairment than controls. Early identification of cognitive impairment is important as it impairs diabetes self-management, making patients prone to complications. However, data about the assessment of cognitive impairment in T2D using a comprehensive cognitive battery is sparse in India. This study was undertaken to estimate the prevalence and pattern of cognitive impairment among young and middle-aged patients with T2D.

Materials and Methods:

A cross-sectional observational study was conducted in a tertiary care teaching hospital in Kolkata (2022–2024) among 125 Bengali-speaking T2D patients with formal education > class IV, aged between 20 and 60 years. The cognitive evaluation was done using the clinical dementia rating scale, mini-mental status examination (MMSE), Montreal cognitive assessment (MoCA), and Addenbrooke’s cognitive examination (ACE)-III. Statistical analyses were done by Jeffrey’s Amazing Statistics Program version 0.19 with appropriate tests (Chi-squared test, Mann–Whitney U test, Spearman correlation statistics, and logistic regression). P value < 0.05 was considered significant.

Results:

T2D patients reported a more subjective sensation of forgetfulness compared to the control group (P = 0.001). MMSE was an insufficient screening tool to distinguish between these two groups. On MoCA and ACE-III, there was a significant difference in total scores between case and control groups (MoCA, P = 0.012 and ACE-III, P < 0.001). Based on ACE-III, 59.20% of T2D patients had cognitive impairment (P < 0.001). The odds of having cognitive impairment in T2D were 3.72 times higher than in the control group (P < 0.001). There was significant impairment of memory (P < 0.001), fluency (P = 0.020), and visuospatial ability (P = 0.032). Females (P = 0.010), less education (P < 0.001), lower socioeconomic status (P < 0.001), BMI < 23 kg/m² (P = 0.049), peripheral neuropathy (P = 0.001), hypothyroidism (P = 0.007), anxiety (P < 0.001), and depression (P < 0.001) were significantly associated with cognitive impairment in diabetes.

Conclusion:

This is the first study from Eastern India to use a comprehensive cognitive scale validated in the local vernacular. Cognitive impairment is prevalent among a significant portion of middle-aged, educated individuals with T2D. Cognitive evaluation should be incorporated into diabetes management from the onset, with a focus on addressing modifiable factors.

Introduction

As the global population ages, the secular trends of diabetes and dementia follow parallel paths, leading to an increased co-occurrence of these diseases. Cognitive impairment and type 2 diabetes (T2D) share multiple pathogenetic pathways, including insulin resistance, chronic low-grade inflammation, oxidative stress, advanced glycation end-products, vascular dysfunction, and common genetic predispositions, with a well-established bidirectional relationship between them.[1] The prevalence of cognitive dysfunction among individuals with T2D varies widely across studies due to differences in patient selection and the tools used to assess cognitive function. Albai et al.[2] reported that 42.03% of individuals with diabetes suffer from mild cognitive impairment (MCI). Epidemiological studies and systematic reviews show that individuals with T2D have a 2–3 times higher risk of developing MCI and dementia compared to non-diabetics.[3,4,5,6] The relative risk of progression from MCI to dementia is also nearly twice as high in individuals with T2D.[7,8]

Interestingly, even prediabetes and metabolic syndrome without diabetes are associated with an elevated risk of cognitive dysfunction compared to controls.[9,10] Various studies have shown that both prediabetes and T2D are linked to significant declines in cognitive domains, particularly in attention, information processing speed, executive function, memory, and fluency.[10,11] Early identification of the cognitive impairment is crucial, as it affects daily living and diabetes self-management and increases vulnerability to diabetes-related complications.[11,12] Unfortunately, cognitive dysfunction often goes undiagnosed in busy diabetes clinics due to the lack of simple, rapid screening tools and limited awareness among caregivers that diabetes-related cognitive impairment is as critical as other complications.[13]

The cognitive impairment pattern in patients with diabetes shares similarities with both Alzheimer’s disease (AD) and vascular dementia (VaD). Most previous studies have focused on older patients with diabetes, where AD, VaD, or mixed forms of dementia are highly prevalent,[11] potentially confounding their findings. Therefore, it is crucial to document the cognitive impairment patterns in middle-aged diabetics, ensuring that other reversible and irreversible forms of dementia are carefully excluded. Moreover, since no effective treatment exists to halt the progression of cognitive impairment in diabetes, managing risk factors is essential. This approach will likely have a significant impact if initiated early and sustained throughout life.[1]

Against this background, the present study was conducted at a tertiary care hospital in Eastern India using a culturally and linguistically adaptable comprehensive cognitive battery to estimate the prevalence and patterns of cognitive dysfunction, along with its clinical correlates, in young and middle-aged patients with T2D.

Material and Methods

Study design and population

This study, part of the Cognition In Diabetes (CID) project, was a quantitative, analytical, hospital-based cross-sectional observational study conducted at the diabetic clinic in the Department of Endocrinology, Medical College and Hospital, Kolkata, from September 2022 to September 2024. Bengali-speaking individuals with T2D (diagnosed according to American Diabetes Association [ADA] criteria),[14] aged 20–60 years, with at least a class IV education, were included. A comprehensive list of exclusion criteria was developed to form a cohort of individuals with T2D without additional risk factors for cognitive impairment [Box 1].[15]

Box 1.

Exclusion criteria

1.Any types of diabetes other than T2D including secondary DM

2.Subjects whose vernacular language other than Bengali

3.History of severe (level-3) hypoglycemia

4.Diagnosed neuropsychiatric condition with/without history of use of psychotropic medications

5.History of any central nervous system disorders, including, but not limited to, stroke, epilepsy, demyelination, space-occupying lesion, or identification of the above diseases by neuroimaging (preferably by >1.5 Tesla MRI and by CT brain, if MRI was contraindicated)

6.Patients having vascular dementia (VaD) and degenerative dementias such as Alzheimer’s disease (AD), Parkinson’s disease dementia (PDD), dementia with Lewy bodies (DLB), frontotemporal dementia (FTD)

7.Any end-organ failure, including chronic liver disease, grade 4 or 5 chronic kidney disease

8.History of malignancy

9.Coexisting vitamin B12 deficiency

10.Other comorbidities having independent potential to hamper cognitive functioning (hyperthyroidism or uncontrolled hypothyroidism, etc.)

11.Any clinical condition requiring admission

12.Impairment of vision, hearing, speech, or other bodily disabilities (such as impaired motor functioning of the dominant hand affecting writing capability), which preclude performing the cognitive tests reliably

13.Pregnant mothers

Sampling design and sample size

For the present study, convenience non-probability sampling was adopted. Sample size (N) was calculated by using the following formula: N = [Zα/2/L]² × p q [where level of significance, α = 5%, Zα/2 = Z-value (standard normal variate) at α level of significance = 1.96 at α = 0.05 (α/2 = 0.025), L = allowable error (assume 10% = 0.1), P = prevalence rate, and q = complementary probability = (1- p)]. Considering a 50% prevalence of cognitive dysfunction in diabetes reported in various studies,[16,17,18,19,20] P = 50% = 0.5, q = 1 -0.5 = 0.5. So, sample size (N) = [(1.96/0.1) 2 × 0.5 × 0.5 = 96.04 = Approx. 96. Taking design effect and non-response, the revised sample size became 125 [(96 × 1.2) + (96 × 10%) = 124.8 = Approx. 125].

Every 10^th patient presented to the diabetic clinic was evaluated for suitability to be included in the present study based on the inclusion and exclusion criteria mentioned above. Thus, over the 18 months of data collection, 125 T2DM patients fulfilling the criteria were finally recruited into one group.

Non-diabetic individuals matched for age (±5 years), sex (with near-equal representation of both sexes), socioeconomic status (±1 class), and education (±2 years) and who met all other inclusion and exclusion criteria were recruited to form the control group, aiming for at least one-third the size of the test group. Ultimately, we recruited 57 individuals in the control group.

Upon completion of the study, the achieved power of the study was computed post-hoc with an α error probability of 0.05. Taking sample size in case group and control 125 and 57, respectively, and the difference of median Addenbrooke’s cognitive examination (ACE)-III scores (median ± IQR) between case and control, the power of the current study came out to be 98.9% (with a critical t of 1.97), indicating a very high probability of detecting an effect.

Parameters studied

Demographic variables (age, sex, years of formal education, socioeconomic status,[21] marital status), clinical factors (duration of diabetes, current drug regimens, history of level-1 and level-2 hypoglycemia, history of comorbidities, addiction, body mass index [BMI]), and biochemical parameters (fasting and postprandial plasma glucose [FPG, PPPG], glycated hemoglobin [HbA1c], capillary blood glucose [CBG], lipid profile [serum low-density lipoprotein cholesterol [LDL-C], high-density lipoprotein cholesterol [HDL-C], and triglycerides], estimated glomerular filtration rate [eGFR] using the 2021 CKD-EPI creatinine equation, urine albumin-creatinine ratio [ACR], liver function tests, and abdominal ultrasonography to assess fatty liver) were recorded. The history of diabetic microvascular complications (retinopathy, neuropathy, nephropathy) and macrovascular complications (coronary artery disease, peripheral arterial disease) was retrieved from past medical records. The cognitive profiles of both case and control groups were evaluated using multiple cognitive batteries.

Cognitive evaluation was done by employing the following battery of tests:

1. Clinical Dementia Rating Scale – Sum of Boxes (CDR-SOB) score: Cognitive functioning was assessed through a semi-structured interview with the patient and an informant, with scoring across six domains: i) memory, ii) orientation, iii) judgment and problem-solving, iv) community affairs, v) home and hobbies, and vi) personal care. Each domain, except personal care, was rated on a 5-point scale: 0 (no impairment), 0.5 (questionable impairment), 1 (mild impairment), 2 (moderate impairment), and 3 (severe impairment). Personal care was rated on a 4-point scale without a 0.5 rating. The total score was calculated by summing the domain scores, yielding a range from 0 to 18, and was further stratified into six clinical stages of dementia.[22] While the CDR provides a subjective assessment, it has proven to be an effective screening tool for detecting Alzheimer ‘s-type dementia, with strong interobserver reliability.[23,24]

2. The Folstein Mini-Mental Status Examination (MMSE): The MMSE is a widely recognized and perhaps the most commonly used dementia screening tool in clinical practice.[25] It evaluates five cognitive domains: orientation (10 points), registration (3 points), attention and calculation (5 points), recall (3 points), and language and praxis (9 points).[26] For diagnosing MCI in this study, the following cut-offs were used based on educational level: 17–21 for those educated up to class VIII, 17–23 for classes IX to XII, and 17–24 for education beyond class XII. A total score below 17 was considered indicative of dementia, regardless of formal education.[27]

3. Montreal cognitive assessment (MoCA): MoCA is one of the most used cognitive batteries in both dementia practice and research, balancing comprehensiveness with practicality for use in busy dementia clinics. To enhance its reliability and accuracy, MoCA has been validated in several Indian languages, including Bengali.[28] The ADA has recommended MoCA for screening cognitive impairment in elderly patients with diabetes.[29] Its validation among Indian patients with T2D makes MoCA an ideal tool for diagnosing dementia.[30] MoCA assesses various cognitive domains, including visuospatial/executive function (5 points), naming (3 points), attention (6 points), language (3 points), abstraction (2 points), delayed recall (5 points), and orientation (6 points), for a total score of 30. Since MoCA scores are not significantly influenced by educational achievement in Indian settings, a single cut-off score has been established. A total score of 22–23 indicates MCI, while a score below 21 is used to diagnose dementia.[28]

4. Addenbrooke’s Cognitive Examination-III (ACE-III): The ACE-III is a highly detailed and widely used cognitive battery for diagnosing MCI and dementia in clinical and research settings.[31] It has been translated, adapted, and validated in several Indian languages, including Bengali.[32,33] The ACE-III assesses five cognitive domains: attention (18 points), memory (26 points), fluency (14 points), language (26 points), and visuospatial skills (16 points), with a total score of 100. Since performance on this test is significantly influenced by the level of formal education, different cut-off scores have been established. For individuals who have completed at least the 10^th standard, scores below 88 indicate MCI, and scores below 85 suggest dementia. For those with education below the 10^th grade, the cut-off scores are < 86 for MCI and < 83 for dementia.[32]

Neuropsychiatric evaluation

The prevalence of depression and anxiety is notably high among individuals living with diabetes.[34,35] Given the inverse relationship between cognitive status and anxiety, depression, and neuroticism[36,37,38] we also explored these associations in patients with T2D.

1. Hamilton Anxiety Rating Scale (HAM-A): The HAM-A is a clinician-administered questionnaire consisting of 14 symptom-defined elements, each rated from 0 to 4 based on severity, with a maximum composite score of 56.[39] Anxiety severity was categorized into four groups: no/minimal anxiety (0–7), mild (8–14), moderate (15–23), and severe (>23).[40]

2. Hamilton Depression Rating Scale (HAM-D): The HAM-D includes 17 clinician-administered questions assessing depression symptoms experienced over the past week. Each response is rated on a scale from 0 to 4, though some items have a maximum score of 2 or 3. The total possible score is 53.[41] Scores of 0–7 are considered normal, 8–16 indicate mild depression, 17–23 moderate depression, and scores 24 or above signify severe depression.[42]

We started collecting data after obtaining the approval of the Institutional Ethics Committee of Medical College and Hospital, Kolkata (Ref No. MC/KOL/IEC/NON-SPON/1747/01/2023 dated 012/01/2023) and getting written informed consent from each participant.

Statistical framework

Statistical analyses were conducted using the Jeffreys’ Amazing Statistics Program (JASP) version 0.19 (2024, Netherlands). A P value of < 0.05 was considered statistically significant. The normality of continuous variables was assessed using the Shapiro–Wilk test, with a P value of < 0.05 indicating non-parametric distribution. Non-parametric continuous variables were presented as median ± interquartile range (IQR), while parametric data were expressed as mean ± standard deviation (SD). Categorical variables were reported as numbers and percentages. Intergroup comparisons of categorical variables were performed using Pearson’s Chi-squared test, with Yates’ correction applied when appropriate. For non-normally distributed continuous variables, the Mann–Whitney U test was used. Spearman’s correlation statistics were applied to assess linear relationships between two continuous variables, reporting Spearman’s rank correlation coefficient (Rho, r), along with the P value and 95% confidence interval (CI). Univariate and multivariate logistic regression analyses were performed to adjust for potential confounders, with goodness of fit evaluated using Nagelkerke’s R² method.

Results

Demographic variables

The patients had a median age of 47 years (±15) with an age range of 20 to 60 and an almost equal representation of males and females. Most patients (91.2%) were married, 4.8% were widowed, and the remainder were unmarried. The majority (47.2%) belonged to the middle socioeconomic class, while 36%, 9.6%, and 7.2% were from the upper-middle, lower-middle, and upper classes, respectively. Additionally, 52.8% of the patients resided in urban areas, with the remainder from rural localities. Both the case and control groups were well-matched in terms of age, sex, education, and socioeconomic status [Table 1].

Table 1.

Comparison of demographic variables between case and control

Variables	Case	Control	P
Age (Median±IQR)	47±15 years	45±17 years	0.749
Sex
Male	50.40%	51.85%	0.873
Female	49.60%	49.15%
Education status (Median ± IQR)	9±5	9±5	0.499
Socioeconomic status (Median ± IQR)	3±1	3±1	0.668

Diabetes related variables

The mean age at diagnosis was 36.91 ± 8.85 years, with a median disease duration of 8 ± 9 years (range: 0–35 years). Glycemic parameters are summarized in Table 2. At the time of the study, 71.2% of patients were on oral anti-diabetic drugs (OAD) only, 0.8% on insulin alone, and 26.4% were on both OAD and insulin, while 1.6% were managing their condition through lifestyle modifications alone. Among OADs, 96.8% of patients were on metformin, 59.2% on teneligliptin, 56.8% on glimepiride, 55.2% on dapagliflozin, 19.2% on pioglitazone, and 10.4% on voglibose. The mean insulin dose was 46.81 ± 24.06 IU/day (range: 4–108 IU/day). Additionally, 10.4% of patients had a history of hospitalization for glycemic control, and 30.4% had experienced level-1 or level-2 hypoglycemia. Among those who had at least one hypoglycemic event in the past year, the median number of events was 3 ± 6.

Table 2.

Glycemic parameters of T2D patients

Glycemic parameters	Value (median ± IQR)	Range
CBG on day of interview (mg/dl)	179±108	81–500
Most recent FPG (mg/dl)	120±64	58–307
Most recent PPPG (mg/dl)	176±88.75	81–487
Most recent HbA1c (NGSP %)	7.30±2.30	4–15.50

Comorbidities, complications, and addictions

The median BMI of T2D patients was 24.42 ± 4.45 kg/m², with 28.8% categorized as lean diabetics (BMI <23 kg/m²).[43] Among the patients, 14.4% were hypothyroid (maintained in a euthyroid state with levothyroxine), and 39.2% had hypertension. Additionally, 33.6% were on angiotensin receptor blockers (ARBs) or angiotensin-converting enzyme inhibitors (ACE-I), and 18.4% were on calcium channel blockers (CCBs). The median serum concentrations of LDL-C, HDL-C, and triglycerides were 85 ± 52 mg/dL, 42 ± 10 mg/dL, and 134 ± 95 mg/dL, respectively, with 76.8% on statin therapy.

Peripheral neuropathy symptoms were present in 32% of T2D patients who were being treated with amitriptyline, gabapentin, pregabalin, or a combination of these medications. Based on Kidney Disease: Improving Global Outcomes (KDIGO) criteria, 35.2% had reduced eGFR (G2 to G3b), and 47.2% had A2 or A3 microalbuminuria. Additional complications included sexual dysfunction (59.2%), fatty liver (52%), retinopathy (32.8%), cardiac dysfunction (15.2%), and foot problems (20.8%).

Regarding lifestyle factors, 10.4% had a history of smoking, 19.2% consumed alcohol, and 22.4% had other addictions.

Global cognitive assessment

Patients with T2D reported more subjective memory issues compared to the control group, as evidenced by a significantly higher median CDR score (P = 0.001). However, the MMSE proved to be an inadequate screening tool for differentiating between the two groups. More detailed cognitive assessments, such as the MoCA and ACE-III, revealed significant differences in total scores between the case and control groups (MoCA, P = 0.012; ACE-III, P < 0.001) [Table 3].

Table 3.

Comparison between case and control based on total score obtained in different cognitive batteries

Cognitive battery (maximum allotted score)	Total score (Median ± IQR)		P
	Case	Control
CDR (18)	0.5±1	0±0.5	0.001*
MMSE (30)	28±3	28±2	0.065
MoCA (30)	25±4	26±5	0.012*
ACE-III (100)	85±11.5	92±8.5	<0.001*

*Mann–Whitney U test, P<0.05 has been taken as statistically significant

According to the CDR, T2D patients had significantly more subjective complaints of cognitive impairment than controls (P = 0.036) [Table 4]. Using the MMSE scale, 6.40% of T2D patients were identified as having MCI, while no MCI cases were detected in the control group. Additionally, no dementia cases were found in either group based on MMSE. The prevalence of MCI and dementia did not significantly differ between T2D patients and controls when assessed by MMSE.

Table 4.

Comparison of the prevalence of cognitive dysfunction between cases and controls in the CDR scale

Diagnosis	Case (n=125)	Control (n=57)	P
Normal (%)	40.80	63.16	0.036*
Questionable impairment (%)	56.80	36.84
Very mild dementia (%)	1.60	0
Mild dementia (%)	0.80	0

*Chi square with Yates’ continuity correction. P<0.05 has been taken as statistically significant

According to MoCA results, 16% of T2D patients were diagnosed with MCI, and 22.40% had dementia. However, there was no statistically significant difference in the prevalence of cognitive impairment compared to the control group. In contrast, the ACE-III, the most comprehensive cognitive assessment used in this study, found MCI and dementia prevalence rates of 16% and 43.20% in T2D patients, respectively, both of which were significantly higher than in the control group (P < 0.001) [Table 5]. Based on ACE-III results, the odds of having cognitive impairment (MCI and dementia) in T2D patients were 3.72 times higher than in non-diabetic individuals (95% CI: 1.89–7.33, P < 0.001).

Table 5.

Comparison between case and control in prevalence of MCI and dementia diagnosed on the basis of three different cognitive batteries

Diagnosis	Cognitive battery	Case (n=125)	Control (n=57)	P
MCI (%)	MMSE	6.40	0	0.118$
	MoCA	16	22.81	0.128
	ACE-III	16	10.53	<0.001*
Dementia (%)	MMSE	0	0	1$
	MoCA	22.40	10.53	0.128
	ACE-III	43.20	17.54	<0.001*

^$Chi Square test with Yates continuity correction. *Chi-square test has been employed. P<0.05 has been taken as statistically significant

Given its comprehensiveness, ACE-III was selected for further analysis of cognitive domains and to explore the association between cognition in T2D and various other factors.

Analysis of cognitive domains (ACE-III)

Among the five cognitive domains assessed in ACE-III, there was no significant difference in attention and language between T2D and the control group. T2D patients did significantly worse in all the subdomains of memory. While lexical fluency was comparable between case and control groups, there was significant impairment in categorical fluency among T2D (P = 0.004), resulting in reduced scores in the overall fluency domain (P = 0.020). Although T2D patients did not perform significantly worse in individual tests of visuospatial function, the total visuospatial score was significantly lower among them than the control group (P = 0.032) [Table 6].

Table 6.

Comparison between case and control based on score obtained in different cognitive domains of ACE-III

Domains	Subdomains (maximum allotted score)	Score (Median ± IQR)		P
		Case	Control
Attention	Orientation (10)	10±1	10±0	0.104
	Registration (3)	3±0	3±0	1
	Calculation (5)	4±2	4±2	0.869
	Total score (18)	17±3	17±2	0.263
Memory	Immediate recall (3)	2±1	3±1	0.031*
	Learning (7)	7±1	7±0	0.033*
	Delayed recall
	Uncued (7)	2±5	5±2	<0.001*
	Cued (5)	5±1	5±0	0.002*
	Semantic memory (4)	4±1	4±0	0.002*
	Total score (26)	20±7	23±2	<0.001*
Language	Naming (12)	12±0	12±0	0.619
	Object semantics (4)	4±0	4±0	1
	Repetition of words (2)	2±0	2±0	1
	Repetition of sentences (2)	2±0	2±0	1
	3-stage command (3)	3±0	3±0	1
	Reading (1)	1±1	1±0	0.054
	Writing (2)	2±1	1±1	0.441
	Total score (26)	25±2	25±2	0.454
Fluency	Lexical fluency (7)	4±2	4±1	0.227
	Categorical fluency (7)	7±1	7±0	0.004*
	Total score (14)	10±3	11±2	0.020*
Visuospatial	Infinity diagram (1)	1±1	1±0	0.186
	Wire’s cube (2)	1±2	1±2	0.711
	Clock drawing test (5)	5±2	5±1	0.088
	Visuoperceptual (8)	8±0	8±0	0.870
	Total score (16)	14±4	15±3	0.032*

*Mann-Whitney U test, P<0.05 has been taken as statistically significant

Neuropsychiatric evaluation

Although T2D patients had numerically higher median HAM-A and HAM-D scores than controls (7 ± 9 vs. 5 ± 5 and 4 ± 5 vs. 3 ± 3, respectively), it did not reach statistical significance (P = 0.069 and 0.582, respectively). Similarly, although the prevalence of both anxiety and depression was numerically higher among T2D than in control (45.60% vs. 28.07% and 23.20% vs. 19.30%, respectively), it was not statistically significant (P = 0.071 and 0.502, respectively).

Clinical correlates of cognitive dysfunction in T2D

Age was not significantly associated with cognitive impairment. Females had significantly lower ACE-III scores than males (87 ± 10 vs 83 ± 12.75, P = 0.010). Socioeconomic status (P < 0.001) was significantly associated with reduced cognitive performance. ACE-III scores had a significant positive correlation with years of formal education received (P < 0.001, Spearman’s rho = 0.612).

Cognitive status was not significantly associated with age at diabetes diagnosis, duration of diabetes, type of OAD intake, the total dose of insulin, history of hospital admission for glycemic control, or frequency of hypoglycemia (level -1 and 2) in the last 1 year. There was no significant correlation between the ACE-III total score and various glycemic parameters [CBG (P = 0.163), FPG (P = 0.494), PPPG (P = 0.845), and HbA1c (P = 0.631)].

Although the correlation between BMI and ACE-III total score did not achieve statistical significance in the present study, non-obese T2D (BMI <23 kg/m²) had significantly lower ACE-III total score than overweight and obese patients (83 ± 13.50 vs. 86 ± 10, P = 0.049). Among the comorbid conditions studied, only hypothyroidism (currently in euthyroid status with levothyroxine therapy) was associated with significantly lower total ACE-III scores than those without a history of hypothyroidism (80 ± 12.25 vs. 86 ± 10, P = 0.007). Among the various diabetic angiopathic complications, the presence of symptomatic peripheral neuropathy was significantly associated with cognitive impairment (88 ± 10 vs. 82 ± 8.50, P = 0.001). There was no significant correlation between ACE-III score and lipid profile, any drug intake (except those for painful peripheral neuropathy), and addiction.

ACE-III total scores were significantly negatively correlated with the HAM-A score (P < 0.001, Spearman’s rho= −0.424) and HAM-D score (P < 0.001, Spearman’s rho = −0.379).

On univariate logistic regression analysis, fewer years of formal education, lower socioeconomic status, hypothyroidism, painful diabetic neuropathy, anxiety, and depression were predictors of cognitive impairment in T2D [Table 7]. However, on the multivariate logistic regression model, only educational achievement (socioeconomic status was not included in the analysis) reached statistical significance (P < 0.001).

Table 7.

Univariate logistic regression to identify the potential predictors of cognitive impairment in T2D

Variable	Nagelkarke’s R2	P	Odds Ratio	CI (95%)
Female sex	0.040	0.053	2.034	-0.016 – 1.436
Years of formal education	0.309	<0.001*	0.694	-0.514 – -0.216
Socioeconomic status	0.225	<0.001*	0.275	-1.889 – -0.696
BMI <23 kg/m2	0.024	0.134	1.855	-0.205 – 1.441
Symptomatic peripheral neuropathy	0.115	<0.001*	4.095	0.526 – 2.294
Hypothyroidism	0.059	0.018*	4.068	0.106 – 2.700
HAM-A score	0.167	<0.001*	1.126	0.050 – 0.188
HAM-D score	0.144	<0.001*	1.188	0.064 – 0.281

*P<0.05 has been taken as statistically significant

Discussion

There is ample evidence from Indian studies demonstrating a significant association between diabetes and cognitive impairment.[16,17,18,19,20,44,45,46,47,48,49,50,51,52,53,54,55] In the current study, patients with T2D reported significantly more cognitive dysfunction, as evidenced by the CDR scores. This suggests that the CDR, a rapid and sensitive tool for dementia screening in a busy clinical setting, could serve as a useful first step in cognitive assessment.[56] Patients exhibiting impairments on the CDR should undergo further evaluation using additional cognitive tools.[57,58] Although the MMSE has been widely used in previous studies,[18,47,53,58,59] it did not show a significant difference between subjects with diabetes and non-diabetic subjects in our cohort. This is likely because the MMSE is not sensitive enough to detect cognitive impairment in younger patients with MCI.[60,61]

Various Indian studies have utilized the MoCA to evaluate cognitive status in T2D patients.[16,17,19,46,48,49,51,62,63] Gupta et al.[30] validated the MoCA for its use in elderly Indian T2D patients. While our study found that MoCA scores were significantly higher in T2D patients than in controls, it did not show a statistically significant difference in the prevalence of cognitive dysfunction between individuals with and without T2D. Discrepancies with prior studies may stem from differences in diagnostic cut-offs, lack of validation in younger populations, and the absence of cognitive tests validated in local languages.

The ACE-III, being the most comprehensive tool used in this study, proved useful for clinical application in Indian settings.[59] Our findings showed a significantly higher prevalence of cognitive impairment in T2D patients compared to controls (59.20% vs. 28.07%, P < 0.001). Similarly, Verma et al.[55] observed cognitive impairment in 63% of T2D patients in a North Indian cohort of elderly individuals (mean age: 64.5 ± 5.3 years). A South Indian study conducted among younger literate subjects (ages 41–60) found that cognitive impairment was prevalent in 63.8% of T2D patients, compared to only 10.8% in the control group, with an odds ratio of 8.78 (CI: 4.47–17.22).[20] In contrast, a study from Western India, which had a relatively small sample size, reported a lower prevalence of dementia (30%) and did not find a statistically significant difference in the prevalence of cognitive dysfunction between T2D and non-diabetic individuals.[50]

Among the five domains, the present study identified significant differences in memory, fluency, and visuospatial ability among T2D compared to controls. Compared to control, domain-wise affliction of cognitive functioning differs across studies among patients with T2D. Verma et al.[55] found impairment in all domains of ACE-III, while Varghese et al.[20] observed sparing of fluency. In a study by Kinattingal et al.,[16] all domains were affected except language and orientation on the MoCA, whereas Lalithambika et al.[48] found no significant problem in abstraction and orientation among individuals with diabetes compared to non-diabetics. Utilizing the same battery, Mohan et al.[49] found significant problems only in visuospatial, attention, and delayed recall. However, the underlying reason behind the differential affliction of cognitive domains across studies needs further research.

In the present study, females had significantly lower ACE-III composite scores. Similar associations have also been reported by other authors,[17,55,64] but not all.[20,46,48,49] Significantly lower educational attainment (P = 0.020), increased prevalence of hypothyroidism (P < 0.001), greater anxiety (P = 0.011), and depression (P = 0.021) among females might have resulted in such an association in our cohort.

Our findings indicate that lean T2D is significantly associated with higher rates of cognitive impairment, and we observed a weak, non-significant positive correlation between BMI and ACE-III scores. The literature presents conflicting evidence on the association between BMI and cognitive performance in T2D patients.[65,66,67] Understanding how lower BMI, or leanness, may contribute to dementia risk in T2D is an important question for future research.

Numerous studies have reported a significant association between diabetes duration and cognitive decline, with Chakraborty et al.[17] finding that a duration exceeding 20 years was independently linked to poorer cognitive outcomes. Other researchers have also observed similar trends.[19,20,49] In our study, however, we noted only a weak, non-significant negative correlation between diabetes duration and cognitive performance, aligning with findings from several other studies.[16,48,50,55,68] Additionally, we observed a statistically non-significant negative association between HbA1c levels and cognitive performance, in contrast to most prior research that found poorly controlled glycemia significantly associated with cognitive impairment.[17,19,48,49,51,55,69,70] Some studies among older age groups, including octogenarians, have also failed to demonstrate this link.[71,72] The lack of a significant association between HbA1c and cognitive decline in our younger T2D cohort highlights an area for further research.

There is a growing body of evidence linking cognitive performance with the use of antidiabetic drugs;[73,74,75] however, our study did not observe a significant association between cognitive function and antidiabetic medication use. Hypoglycemia, widely recognized as a risk factor for cognitive dysfunction in diabetes,[76] also showed no significant association with cognitive impairment in our cohort. This result may be attributed to our deliberate exclusion of patients with a history of level-3 hypoglycemia, as this more severe hypoglycemic event is a known contributor to cognitive decline.[77]

In our study, we did not observe a significant association between cognitive decline and either dyslipidemia or hypertension. There is ongoing debate regarding the optimal serum lipid levels necessary for brain health, especially given that very low LDL-C levels have been linked to cognitive impairment in some research.[78] While diabetic dyslipidemia has been identified as a risk factor for dementia in certain studies, others have not corroborated this association.[79] For instance, recent findings by Ma et al.[80] indicated no significant association between total cholesterol, LDL-C, HDL-C, triglycerides, and cognitive impairment in diabetes. Similarly, although hypertension has been identified as a cognitive decline risk factor in diabetes,[81] our study did not confirm this relationship. This lack of significance may be due to the limited power of our study to detect these specific associations.

No angiopathic complication, except symptomatic peripheral neuropathy, had a significant association with cognitive impairment. This might be because of the exclusion of patients with advanced complications sufficient to cause cognitive impairment from this study. However, painful diabetic peripheral neuropathy and the usage of drugs to tame the same were significantly associated with cognitive decline. Similar observations have been documented by other authors.[82,83]

Given the established impact of thyroid hormones on cognitive function,[84] all T2D patients with abnormal thyroid-stimulating hormone (TSH) or free thyroxine levels were excluded from our study. At the study’s conclusion, 14.4% of patients had primary hypothyroidism, though they were in a euthyroid state with levothyroxine replacement. Notably, T2D patients with a history of hypothyroidism demonstrated significantly greater cognitive impairment despite achieving euthyroid status, aligning with findings by Wekking et al.[85] Moreover, as highlighted in prior research,[86] we observed an increased risk of cognitive impairment among patients with both diabetes and comorbid anxiety or depression, underscoring the importance of mental health management in diabetes care.

The study stands out due to its rigorous patient selection criteria, which excluded potential confounders such as advanced age (over 60 years), lack of formal education, coexisting brain diseases (ruled out via neuroimaging and medical history), and known psychiatric disorders.[15] In addition to its robust methodology, using multiple cognitive batteries enhanced the reliability and accuracy of the results. Moreover, the study employed culturally and linguistically adaptable cognitive tools, such as the MoCA and ACE-III, both validated for the Bengali population, which further distinguishes this research. To our knowledge, no similar effort have been undertaken in any prior studies conducted in Eastern India.

However, the current study was not devoid of limitations. This was a single-center study with a relatively smaller sample size. A longitudinal study would be preferable to follow the trajectory of cognitive performance and further the knowledge regarding the causal relationship between various diabetes-related correlates and dementia.

Conclusion

This is the first study from Eastern India using a comprehensive cognitive scale validated in local vernacular. This study underscores the fact that cognitive impairment is present in a sizable portion of middle-aged, educated T2D patients without advanced angiopathic complications. However, MMSE and MoCA may be insensitive tools to pick up cognitive impairment in this group of patients, and we advocate for ACE-III for this purpose. Memory, visuospatial, and fluency domains were particularly affected in these patients. So, cognitive assessment should be an integral part of diabetes care from the beginning, and the modifiable factors should be taken care of accordingly.

Authors’ contribution

SC, RB, and SD generated the idea and formulated the plan of study. Field work and data collection were done by SC and SD. SC and RB did the statistical analysis. SC wrote the first draft, which was further edited by all the coauthors who agreed upon the final version of the manuscript.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.