Cognition - India - prevalence - Punjab - type 2 diabetes
Introduction
The alarming prevalence of diabetes in its global perspective has attracted the attention
of researchers to investigate the factors that are associated with it and may worsen
its pathology. Several risk factors of type 2 diabetes (T2DM) such as endothelial
dysfunction, obesity, dyslipidemia, hypertension, hypothalamus-pituitary-adrenal axis
(HPA axis) abnormalities, and underlying inflammation impact negatively on brain.[1] Some studies have suggested that T2DM impinge upon cognitive domains, especially
verbal memory, attention or processing speed, psychomotor ability, and executive functions.[2]
[3]
[4] Duration of diabetes has also been observed to be related with cognitive decrements
such as immediate verbal recall, delayed verbal recall, and abstract reasoning.[5] However, conclusions vary depending on the sampling strategy, study design, tests
for cognition used, diabetes severity, and co-existing morbidities.[6]
[7] Several association pathways of T2DM and cognitive decline have been established
whereby some studies reported the association of diabetes with Alzheimer's[8] and vascular dementia,[9] and other studies reported its association with hypertension[10] and stroke.[11] Another form, i.e., type 3 diabetes results with an insulin resistance in the brain
causing Alzheimer's and has molecular and biochemical features encompassing both type
1 diabetes mellitus and T2DM.[12] Nevertheless, these studies have shown common manifestations of these pathways that
contribute to cognitive impairment.
The possible link of T2DM with cognitive performance is intricate and remains unforeseen
as other factors including age, socio-economic status, marital status, level of education,
and other comorbidities also influence cognitive function. To understand the influence
of diabetes on cognitive impairment, the present study was designed to examine the
prevalence and predictors of neurocognitive impairment in type 2 diabetic population
of Punjab, India.
Materials and Methods
The present cross-sectional study comprised 516 T2DM subjects who attended endocrinology
outpatient department (OPD) of Government Medical College and Hospital (GMCH), Patiala,
Punjab, India. A total of 1098 patients were screened and 1015 qualified for inclusion
according to diagnostic criteria given by American Diabetes Association.[13] After following stringent inclusion-exclusion criteria, 516 diabetic subjects were
included in the present research. The inclusion criterion was consenting T2DM patients
belonging to Punjab. Exclusion criteria was subjects not from Punjab or had any other
neurological conditions such as stroke, Alzheimer's disease, vascular dementia, Parkinson's
disease, epilepsy or head injury, cardiovascular diseases such as hypertension, hypotension,
or atherosclerosis and psychiatric disorders; depression, alcohol dependence, drug
dependence, and use of antidepressant or antipsychotic medications [Figure 1]. All patients gave their written consent prior to participation, and the study was
approved by Institutional Ethical Committee.
Figure 1 Showing data collection protocol after following stringent inclusion and exclusion
criteria
Patients were assessed for cognitive impairment by using the Mini Mental State Examination
(MMSE),[14] a standard version 30-point questionnaire used to screen arithmetic, language use,
memory, and orientation (basic motor skills). It is a validated measure for evaluating
cognitive decline, whereby MMSE score of <23 has sensitivity of 81.3% and specificity
of 60.2%.[15] The scoring by MMSE can be influenced by the effect of age and education level.
To avoid any bias, present data were well matched for age and education status (P > 0.05). Furthermore, other tests were also used to screen subjects for cognitive
decline such as trail making test-A (TMT-A for attention and information processing
speed with visuomotor component) and trail making test-B (TMT-B for visual attention)[16] along with MMSE.
Definition of risk variables
Socio-economic status was evaluated according to the updated version of Kuppuswamy
and Pareek scale[17] and categorized according to per capita per month income in rupees, ≤10,000 (low-income
group), 10,000-50,000 (middle-income group), and >50,000 (high-income group). Physical
activity was determined on the basis; if subject was doing at least 30 min of aerobic
exercise/walk, he/she was considered active otherwise sedentary. Aerobic walking means
consistent walk of 1 km in 13 min or less. Lipid levels, duration of diabetes, statin
use, and glucose levels were noted down from their medical records. Information regarding
marital status, education level, smoking, and drinking alcohol was recorded by interviewing
them. Anthropometric measurements such as height and weight were measured and body
mass index (BMI) was calculated according to Quetelet equation (BMI = weight in kilograms/height
in meters squared). For waist–hip ratio, waist circumference was measured at the midpoint
between the lower margin of ribs and the superior border of the iliac crest. Hip circumference
was measured around the widest portion of the buttocks. Systolic blood pressure (SBP)
and diastolic blood pressure were noted down as a mean of two tests conducted after
an interval of 3 min in sitting position after 15 min of rest.
Statistical analysis
Data are presented as mean ± standard deviation, numbers or percentages, and interquartile
range is given for skewed data. The difference between the groups was examined using
Chi-square test for categorical variables and Student's t-test for continuous variables.
The internal consistency reliability for MMSE, TMT-A, and TMT-B was checked using
Cronbach's alpha. A linear regression was applied to investigate the association between
cognitive decline and risk variables (general linear model procedure). Those variables
which showed linear relationship with the dependent variable in univariate testing
were further included in the multivariate logistic regression analysis (backward stepwise)
to identify independent association of the significant variables. The significance
was checked at 5% level, but for multiple comparisons, Bonferroni correction was applied
accordingly.
Results
The present study involved 516 adult T2DM subjects of Punjab. Out of which, 289 (56.01%)
were men and 227 (43.99%) were women. One hundred and seventy-four (33.73%) met the
criteria for impaired cognition, whereas 342 (66.27%) were having normal cognition.
The mean age of diabetic subjects was 47.79 ± 10.18. It is noteworthy that out of
174, 89.65% of the diabetic subjects were undiagnosed for neurocognitive impairment
before participation in the present study. Univariate analysis of the variables [Table 1] showed that women were approximately 2 times more vulnerable for cognitive impairment
than men (odds ratio [OR] = 2.13, 95% confidence interval [95% CI]: 1.47-3.09, P< 0.001). Subjects with sedentary life style (OR = 2.81, 95% CI: 1.87-4.21, P = 0.001), divorced, or separated women (OR = 3.11, 95% CI: 1.42-6.80, P = 0.007), smoking (OR = 2.73, 95% CI: 1.75-4.26, P < 0.001), drinking alcohol (OR = 6.00, 95% CI: 3.62-9.93, P< 0.001), SBP (>120 mmHg) (OR = 2.64, 95% CI: 1.81-3.85, P = 0.001), diastolic blood pressure (>80 mmHg) (OR = 1.77, 95% CI: 1.22-2.57, P = 0.003), triglycerides (>150 mg/dl) (OR = 2.02, 95% CI: 1.36-2.99, P = 0.001), high-density lipoprotein (<40 mg/dl) (OR = 2.41, 95% CI: 1.59-3.64, P = 0.001), statin use (OR = 1.66, 95% CI: 1.15-2.34, P = 0.009), glucose levels (>125 mg/dl) (OR = 1.73, 95% CI: 1.20-2.50, P = 0.005), and duration of diabetes starting from 2 to 10 years and more than 10 years
(P = 0.001) were having higher chances of cognitive impairment.
Table 1
Prevalence and determinants of cognitive impairment in type 2 diabetic subjects
|
Diabetic subjects (N = 516)
|
|
Variables
|
Impaired cognition N(%) 174 (33.73)
|
Normal cognition N(%) 342 (66.27)
|
Odds ratio
|
95% Cl
|
P value
|
|
Gender
|
|
|
|
|
|
|
Men
|
76 (43.67)
|
213 (62.28)
|
Referent
|
|
|
|
Women
|
98(56.33)
|
129 (37.72)
|
2.13
|
1.47-3.09
|
<0.001
|
|
Age
|
|
|
|
|
|
|
35-45 years
|
53(30.46)
|
99 (28.25)
|
Referent
|
|
|
|
46-55 years
|
68 (39.08)
|
169 (49.41)
|
0.75
|
0.49-1.16
|
0.24
|
|
56-65 years
|
53 (30.46)
|
74(21.64)
|
1.34
|
0.82-2.17
|
0.29
|
|
Marital Status
|
|
|
|
|
|
|
Single
|
27 (15.53)
|
56 (16.37)
|
Referent
|
|
|
|
Married
|
93 (53.44)
|
223 (65.22)
|
0.86
|
0.51-1.45
|
0.68
|
|
Widow
|
30 (17.24)
|
47 (13.74)
|
1.32
|
0.69-2.53
|
0.50
|
|
Divorced / Separated
|
24(13.29)
|
16 (4.67)
|
3.11
|
1.42-6.80
|
0.007
|
|
Education Level
|
|
|
|
|
|
|
Matriculation
|
49 (28.16)
|
101 (29.53)
|
Referent
|
|
|
|
Secondary
|
87(50.00)
|
142 (41.52)
|
1.26
|
0.82-1.95
|
0.34
|
|
Graduation and above
|
38 (21.84)
|
99 (28.95)
|
0.79
|
0.48-1.31
|
0.44
|
|
Socio-economic Status
|
|
|
|
|
|
|
High income
|
57 (32.75)
|
124 (36.25)
|
Referent
|
|
|
|
Middle income
|
79 (45.42)
|
121 (35.38)
|
1.42
|
0.93-2.17
|
0.128
|
|
Low income
|
38(21.83)
|
97 (28.37)
|
0.85
|
0.52-1.39
|
0.605
|
|
Physical Activity
|
|
|
|
|
|
|
Active
|
43 (24.73)
|
164 (47.95)
|
Referent
|
|
|
|
Sedentary
|
131 (75.28)
|
178 (52.04)
|
2.81
|
1.87-4.21
|
0.001
|
|
Smoking
|
|
|
|
|
|
|
Non smoking
|
51 (29.31)
|
158 (46.19)
|
Referent
|
|
|
|
Smoking
|
74 (42.52)
|
84(24.56)
|
2.73
|
1.75-4.26
|
<0.001
|
|
Ex-smoking
|
49 (28.16)
|
100 (29.25)
|
1.52
|
0.95-2.42
|
0.100
|
|
Alcohol Drinking
|
|
|
|
|
|
|
Non Drinkers
|
28 (16.09)
|
100 (29.25)
|
Referent
|
|
|
|
Drinkers
|
131 (75.28)
|
178 (52.04)
|
6.00
|
3.62-9.93
|
<0.001
|
|
Ex-drinkers
|
15 (8.63)
|
64 (18.71)
|
0.84
|
0.42-1.68
|
0.75
|
|
Blood Pressure: SBP
|
|
|
|
|
|
|
<120 mmEIg
|
62 (35.64)
|
203 (59.35)
|
Referent
|
|
|
|
>120 mmEIg
|
112 (64.36)
|
139 (40.65)
|
2.64
|
1.81-3.85
|
0.001
|
|
Blood Pressure: DBP
|
|
|
|
|
|
|
<80 mmEIg
|
69 (39.65)
|
184 (53.81)
|
Referent
|
|
|
|
>80 mmEIg
|
105 (60.35)
|
158 (46.19)
|
1.77
|
1.22-2.57
|
0.003
|
|
Total Cholesterol
|
|
|
|
|
|
|
<200 mg/dl
|
119 (68.39)
|
232 (67.84)
|
Referent
|
|
|
|
>200 mg/dl
|
55 (31.61)
|
110 (32.16)
|
0.97
|
0.66-1.44
|
0.98
|
|
Low Density Lipoprotein
|
|
|
|
|
|
|
<100 mg/dl
|
123 (70.68)
|
219 (64.04)
|
Referent
|
|
|
|
>100 mg/dl
|
51 (29.32)
|
123 (35.96)
|
0.74
|
0.50-1.09
|
0.16
|
|
Triglycerides
|
|
|
|
|
|
|
<150 mg/dl
|
49 (28.16)
|
151 (44.16)
|
Referent
|
|
|
|
>150 mg/dl
|
125 (71.84)
|
191 (55.84)
|
2.02
|
1.36-2.99
|
0.001
|
|
High Density Lipoprotein
|
|
|
|
|
|
|
>40 mg/dl
|
40 (22.99)
|
143 (41.82)
|
Referent
|
|
|
|
<40 mg/dl
|
134 (77.01)
|
199 (58.18)
|
2.41
|
1.59-3.64
|
0.001
|
|
Statin Use
|
|
|
|
|
|
|
Non-users
|
81 (46.55)
|
202 (59.06)
|
Referent
|
|
|
|
Users
|
93 (53.45)
|
140 (40.94)
|
1.66
|
1.15-2.34
|
0.009
|
|
BMI (kg.m2)
|
|
|
|
|
|
|
<18.4
|
8 (4.59)
|
14 (4.09)
|
Referent
|
|
|
|
18.5-24.9
|
16 (9.19)
|
59 (17.26)
|
0.47
|
0.17-1.33
|
0.25
|
|
25-29.9
|
74(42.52)
|
211 (61.69)
|
0.61
|
0.25-1.52
|
0.42
|
|
30-34.9
|
42 (24.16)
|
33 (9.66)
|
2.23
|
0.84-5.92
|
0.17
|
|
35-39.9
|
25 (14.36)
|
19 (5.55)
|
2.30
|
0.80-6.61
|
0.19
|
|
>40
|
9 (5.18)
|
6 (1.75)
|
2.63
|
0.68-10.12
|
0.28
|
|
WHR in Males (N=289)
|
(n=76)
|
(n=213)
|
|
|
|
|
<90 cm
|
22 (12.64))
|
123 (35.96)
|
Referent
|
|
|
|
>90 cm
|
54(31.03)
|
90 (26.31)
|
3.35
|
1.91-5.90
|
<0.001
|
|
WHR in Females (N=227)
|
(n=98)
|
(n= 129)
|
|
|
|
|
<80 cm
|
25 (17.52)
|
75 (21.92)
|
Referent
|
|
|
|
>80 cm
|
73 (41.95)
|
54 (15.78)
|
4.06
|
2.29-7.19
|
<0.001
|
|
Glucose Level
|
|
|
|
|
|
|
<125 mg/dl
|
75 (43.11)
|
194 (56.72)
|
Referent
|
|
|
|
>125 mg/dl
|
99 (56.89)
|
148 (43.28)
|
1.73
|
1.20-2.50
|
0.005
|
|
Duration of Diabetes
|
|
|
|
|
|
|
<2 years
|
17 (9.77)
|
96 (28.07)
|
Referent
|
|
|
|
2-5 years
|
73 (41.96)
|
194 (56.72)
|
2.12
|
1.19-3.80
|
0.001
|
|
6-9 years
|
52 (29.88)
|
41 (11.98)
|
7.16
|
3.71-13.84
|
0.001
|
|
>10 years
|
32 (18.39)
|
11 (3.23)
|
16.43
|
6.97-38.72
|
0.001
|
Multivariate backward stepwise regression analysis was done to identify variables
which were independently associated with cognitive decline [Table 2]. Being a woman (OR = 2.00, 95% CI: 1.25-3.20, P = 0.004), glucose levels >125 mg/dl (OR = 1.73, 95% CI: 1.15-2.61, P = 0.008), and SBP >120 mmHg (OR = 3.70, 95% CI: 2.40-5.60, P < 0.001) emerged as independent predictors of cognitive decline. Sedentary life style
(OR = 2.32, 95% CI: 1.07-5.08, P = 0.034) and duration of diabetes >10 years (OR = 4.34, 95% CI: 2.57-7.29, P< 0.001) also conferred substantial risk of cognitive impairment independently.
Table 2
Multivariable backward stepwise regression analysis to determine factors which independently
associated with cognitive impairment
|
Variables
|
β ± SE
|
OR
|
95% Cl
|
P
|
|
Being a diabetic woman
|
0.69±2.38
|
2.00
|
1.25-3.20
|
0.004
|
|
Glucose levels > 125 mg/dl
|
0.55±0.21
|
1.73
|
1.15-2.61
|
0.008
|
|
Systolic Blood Pressure
|
1.31± 0.21
|
3.70
|
2.40-5.60
|
<0.001
|
|
> 120 mmHg
|
|
|
|
|
|
Sedentary Life Style
|
0.84±0.40
|
2.32
|
1.07-5.08
|
0.034
|
|
Duration of Diabetes
|
1.47±0.26
|
4.34
|
2.57-7.29
|
<0.001
|
|
> 10 years
|
|
|
|
|
Discussion
The present study examined the prevalence and predictors of neurocognitive impairment
in T2DM patients of Punjab. This is the first study from this region which revealed
that the prevalence of cognitive impairment is 33.73% in diabetic population. Being
a diabetic woman is an independent risk factor for neurocognitive impairment which
doubles the risk (OR = 2.13, 95% CI: 1.47–3.09, P< 0.001) as compared to man. This inference has been corroborated by some studies,[18]
[19] and one study reported the risk of cognitive impairment even higher (3.75 times)
in women.[20] The prevalence of neurocognitive impairment in T2DM has been scarcely examined in
India. One study from Jaipur[21] has shown that the prevalence of neurocognitive impairment in T2DM patients is 48%,
however, data from other regions have shown that cognitive decline in T2DM patients
ranges from 3% to 23%.[22]
[23] The prevalence rates of cognitive impairment may vary depending on the sampling
procedure and diagnostic criteria.[24] Some of the reasons for higher prevalence of cognitive impairment in T2DM patients
in the present study are, first of all, the diabetic OPDs of GMCH, Patiala, primarily
caters to referred cases of severe form of diabetes. About 18% of the subjects in
the present study have more than 10 years of diabetes, which may impinge upon the
prevalence of cognitive impairment. Second, MMSE <23 criteria used in the present
study also included subjects having borderline cognitive impairment (19-23 scores).
Such borderline subjects may also have supplemented the prevalence in the present
study. Other tests (TMT-A and TMT-B) also identify neurocognition in those patients,
and Cronbach's statistics in the present study (α = 0.82) ruled out any internal inconsistency.
The relationship of T2DM and cognitive decline is very intricate as some unforeseen
underlying mechanism targeting neurodegeneration and vascular factors may also participate.
Moreover, some studies have largely ignored the adjustment of the effect of risk variables
in the analysis, however after adjusting the confounding effect in the backward step
wise regression analyses in the present study, subjects having glucose levels >125
mmHg are at 1.73 times higher risk of developing impairment in neurocognition. This
result is in agreement with several studies,[3]
[22]
[25]
[26]
[27]
[28]
[29]
[30] however, few reports including Rancho Bernardo cohort study found no association
of diabetes with neurocognitive decline.[31] Coker and Shumaker[25] compiled 32 studies which investigated the performance of neuropsychological tests
in T2DM patients. Out of these, 67% (twenty studies) reported a positive association
of diabetes with poor performance on neuropsychological tests. Another report has
examined the relationship of impaired fasting glucose and neurocognition as the determinant
of dementia in older women.[26] In a 4-year randomized trial of raloxifene use, it has been observed that after
adjusting the effect of race, education, and co-existing depression, women show severe
form of cognitive decline (P = 0.001). Similarly, another cross-sectional study on older men highlighted that
higher insulin levels and glucose intolerance are significantly associated with lower
MMSE scores.[27] Another study compiling the results of ten population-based reports on the association
of T2DM and domain-specific neurocognition has revealed a substantial influence of
diabetes on neurocognition impairment which may lead to the development of dementia.[22] In a longitudinal cohort study by Luchsinger et al.,[3] diabetes is found to be a significant risk factor for all causes of mild cognitive
impairment after adjustment for all risk covariates. The association of diabetes with
nonamnestic mild cognitive impairment is attenuated after adjustment with vascular
risk factors and socio-economic status. The age, gene/environment susceptibility in
Reykjavik study has demonstrated that after adjusting the effects of demographic and
health factors, persons with diabetes show poor performance in processing speed and
executive functions.[4] The action to control cardiovascular risk in diabetes-memory in diabetes trial showed
statistically significant association between A1C levels and four cognitive tests
after adjusting age and other risk determinants.[28] In the investigation for the relationship of diabetes and cognitive impairment in
Iranian population, Ebady et al.[29] documented that diabetes is significantly associated with lower MMSE scores (OR
= 1.9, 95% CI: 1.01-3.6, P = 0.001), which is significantly correlated with duration and quality of diabetes.
To understand the modifiers of cognitive function and brain structure in T2DM, it
has been observed that subjects with T2DM have reduced hippocampal and prefrontal
volumes, verbal declarative memory deficit, and impaired HPA axis feedback regulation.[30]
A 20-year follow-up atherosclerosis risk in communities for cognitive change[32] has shown that midlife hypertension, especially higher SBP is highly associated
with cognitive decline. Another report concluded that SBP and heart rate independently
influence the development of cognitive decline, incident cognitive dysfunction, and
cognitive deterioration in high-risk patients with atherosclerosis and diabetes.[33] In the present study also, diabetic subjects having >120 mmHg of SBP had 3.70 times
higher risk of neurocognitive decline in comparison to nondiabetic subjects.
Another risk factor that independently influenced cognition in the present study is
sedentary life style. T2DM patients living sedentary life style have 2-fold higher
risk of neurocognitive impairment than their active counterparts. This finding is
in agreement with other studies which have shown that sedentary life style is the
hallmark for neurocognitive impairment.[34]
[35]
Duration of diabetes has been observed to influence neurocognition very strongly.
Considerable derangements of neurocognition in T2DM patients, especially among older
women have been examined.[36] In this longitudinal study, unfavorable cognitive performance has been shown in
diabetic subjects over 4 years compared with subjects who had normal glucose levels.[37] It has been revealed that patients diagnosed for diabetes from the last 5 years
perform poor for domain-specific cognition such as logical memory and word fluency
in comparison to those who are diagnosed lately.[5] A report from Hyderabad, Telangana, India, shows that subjects having duration of
diabetes >5 years show increased P300 latencies (cognitive dysfunction), a marker
of speed of neural events related to attention and short-term memory.[38] Similarly, another study reported that negative correlation exists (r = −0.408, P = 0.001) between MMSE scores and duration of diabetes in Iranian population.[32] In the present study, diabetic subjects who have duration of diabetes >10 years
are at 4.34 times higher risk of cognitive impairment.
Limitations
Some potential limitations of the present study may have influenced the results. First,
MMSE scoring to identify neurocognition can be influenced by age and education. Although
present study was well matched for these two variables (P > 0.05) and other tests involved for diagnosing cognitive decline show significant
matching of reliability index (Cronbach's α = 0.82) with MMSE, residual confounding
for timed elements of memory and executive functions may have remained to be exposed
as MMSE is considered to be less sensitive for them. Second, notwithstanding our study
attained considerable power to detect differences, the number of diabetic subjects
according to gender in the present study are relatively less, so the implications
need caution; more studies involving large cohorts are required to substantiate our
key findings.
Conclusions
The present study shows that majority of the T2DM patients remain undiagnosed for
neurocognition during the course of their life. Cognitive decline in diabetes is independently
influenced by risk variables such as being a women, SBP >120 mmHg, sedentary life
style, glucose levels >125 mg/dl, and duration of diabetes >10 years. Higher prevalence
of cognitive decline in this region is alarming, and results of the present study
suggest that every diabetic subject should be examined for cognition so that future
imperative sequels of T2DM, especially neurodegeneration and vascular dementia may
be tackled effectively.
Financial support and sponsorship
The grant for major research project sanctioned to PPS by University Grants Commission
(F. No. 42-48/2013) is gratefully acknowledged.
