Elevated body mass index is associated with executive dysfunction in otherwise healthy adults

Comprehensive Psychiatry 48 (2007) 57 – 61 www.elsevier.com/locate/comppsych

Elevated body mass index is associated with executive dysfunction in otherwise healthy adults John Gunstada,4, Robert H. Paulb, Ronald A. Cohenb, David F. Tateb, Mary Beth Spitznagelc, Evian Gordond,e,f a Department of Psychology, Kent Hall, Kent State University, Kent, OH 44242, USA Department of Psychiatry and Human Behavior, Brown Medical School, Providence, RI 02912, USA c Department of Psychiatry, Summa Health System, Akron, OH 44310, USA d The Brain Resource International Database, Brain Resource Company, Ultimo, NSW 2007, Australia e Department of Psychological Medicine University of Sydney, NSW 2006, Australia f The Brain Dynamics Centre Westmead Hospital, NSW 2145, Australia b

Abstract There is growing evidence that obesity is linked to adverse neurocognitive outcome, including reduced cognitive functioning and Alzheimer disease. However, no study to date has determined whether the relationship between body mass index (BMI) and cognitive performance varies as a function of age. We examined attention and executive function in a cross-section of 408 healthy persons across the adult life span (20-82 years). Bivariate correlation showed that BMI was inversely related to performance on all cognitive tests. After controlling for possible confounding factors, overweight and obese adults (BMI N 25) exhibited poorer executive function test performance than normal weight adults (BMI, 18.5-24.9). No differences emerged in attention test performance, and there was no evidence of a BMI  age interaction for either cognitive domain. These results provide further evidence for the relationship between elevated BMI and reduced cognitive performance and suggest that this relationship does not vary with age. Further research is needed to identify the etiology of these deficits and whether they resolve after weight loss. D 2007 Elsevier Inc. All rights reserved.

1. Introduction Obesity is a leading preventable cause of death and is associated with many medical conditions, including cardiovascular disease and type 2 diabetes [1]. There is growing evidence that obesity is also associated with adverse neurocognitive outcome. It is an independent risk factor for Alzheimer disease and is linked to both temporal lobe atrophy and white matter disease in older adults [2-4]. Half of bariatric surgery candidates show impaired executive functioning [5]. Obese middle-aged and older adult men exhibit difficulties in working memory, verbal fluency, and memory [6]. Although these findings demonstrate a relationship between elevated body mass index (BMI) and cognitive 4 Corresponding author. Tel.: +1 330 672 2589; fax: +1 330 672 3786. E-mail address: [email protected] (J. Gunstad). 0010-440X/$ – see front matter D 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.comppsych.2006.05.001

performance, it is unknown whether this relationship varies with age. No study to date has examined whether the reduced cognitive test performance is limited to older adults or whether a similar pattern can be found across the adult life span. To do so, we examined the cognitive test performance of a cross-section of healthy adults ranging from 21 to 82 years of age. Tests of attention and executive function were selected for comparison, because past studies suggest that younger obese adults may show deficits in these cognitive domains. For example, rates of attention deficit hyperactivity disorder (ADHD) are elevated in obese individuals and impulsivity is linked to BMI in some populations [7,8]. Therefore, we compared the attention and executive function test performance of normal weight adults to overweight and obese individuals. Analyses were conducted to specifically test the effects of BMI and the possible BMI  age interaction. Based on the above findings, we predicted that

58

J. Gunstad et al. / Comprehensive Psychiatry 48 (2007) 57 – 61

Table 1 Demographic and psychosocial characteristics of 408 participants Younger adults

n Age BMI Estimated IQ Education % Female Depression Anxiety Stress

Older adults

Normal weight

Over/obese

Normal weight

Over/obese

178 31.56 22.09 107.21 14.86 55.1 1.61 0.67 2.52

140 32.40 28.41 106.41 14.81 46.4 1.85 0.95 2.32

32 58.34 23.09 108.65 13.31 53.1 1.47 0.53 2.59

58 60.40 29.17 109.07 14.05 55.1 1.22 0.79 2.19

F F F F

8.71 1.71 6.04 2.40

F 2.10 F 0.95 F 2.53

overweight and obese individuals would show poorer attention and executive function performance than normal weight persons.

2. Method 2.1. Overview The present study used data from the Brain Resource International Database [9,10]. Six laboratories participated in data acquisition in a total quality-controlled manner (New York, Rhode Island, Holland, London, Adelaide, and Sydney). The database contains demographic, psychiatric, health, and cognitive data on more than 1000 healthy adults. Exclusion criteria for this database include medical and psychiatric conditions with the potential to influence cognitive performance, including history of or present traumatic brain injury, neurologic disorder, and other medical conditions (eg, hypertension, diabetes, cardiac disease, thyroid disease, or sleep apnea). The SPHERE-12 [11] was administered to exclude individuals with history of significant psychiatric and substance use disorders (eg, bipolar disorder, ADHD, or alcohol/drug use disorders). Participants were also excluded if they had a family history of ADHD, schizophrenia, bipolar disorder, or genetic disorder. Participants were also asked to refrain from caffeine and nicotine for at least 2 hours and from alcohol for at least 12 hours before testing. Before onset of testing, participants were also screened for sensory deficits. 2.1.1. Participants A total of 408 adult participants with complete data were extracted from the database. Participants were categorized into normal weight and overweight/obese groups based on established criteria (normal weight, BMI of 18.5-24.9; overweight/obese, BMI z 25.0) [12]. To determine the possible interaction between BMI and age, we categorized the participants into younger (aged 20-50 years) and older (aged 50-82 years) age groups (Table 1). 2.1.2. Procedure After providing informed consent, each participant was assigned an 8-digit identification number for anonymity and

F F F F

9.10 4.42 6.29 2.56

F 2.34 F 1.29 F 2.42

F F F F

6.62 1.59 6.85 3.07

F 1.85 F 0.88 F 2.39

F F F F

7.62 3.54 6.72 2.68

F 1.67 F 1.06 F 2.33

answered questions about medical, psychosocial, personality, and emotional factors. Participants then underwent neurocognitive testing lasting approximately 45 minutes. Tests were administered in a fixed order using prerecorded instructions and a touch-screen computer. The test battery has good validity and reliability [13,14]. To identify the effects of subclinical levels of psychologic conditions, we also had the participants to complete the Depression, Anxiety, and Stress Scale (DASS) [15]. 2.1.3. Instrumentation 2.1.3.1. Estimated intellectual functioning. 2.1.3.1.1. Spot-the-word. This task is a computerized adaptation of the Spot the Real Word test [16]. Participants are presented with 2 words on the touch screen. One of the 2 words is a valid word in the English language and the other a nonword foil. Participants are asked to identify the real word. The total correct score is entered into a regression formula that factored education and age to render an estimated intelligence quotient [17]. 2.1.3.2. Attention. 2.1.3.2.1. Digit span forward. Subjects were presented with a series of digits presented individually for 500 milliseconds and separated by a 1-second interval. The number of digits in each sequence was incrementally increased from 3 to 9. The total number of correct trials served as the dependent variable. 2.1.3.2.2. Choice reaction time. Participants attended to the computer screen as 1 of 4 circles was illuminated. Immediately after presentation, the subject then had to touch the illuminated circle as quickly as possible. Twenty trials were Table 2 Correlation between BMI and cognitive test performance in 408 healthy adults Test Digit span forward Choice reaction time Switching of attention—number Span of visual memory Verbal interference Switching of attention—letter/number Maze errors 4 Denotes 2-tailed significance, P b .01.

BMI 0.174 0.144 0.164 0.154 0.234 0.204 0.114

J. Gunstad et al. / Comprehensive Psychiatry 48 (2007) 57 – 61

59

Table 3 Means and SDs in BMI group by age for attention measures Test

Digit span forward Choice reaction time Switching of attention—number Span of visual memory

Younger adults

Older adults

Normal weight

Over/obese

Normal weight

Over/obese

7.78 0.69 19.22 8.32

7.47 0.71 19.93 7.97

6.84 0.78 25.28 6.88

6.98 0.80 27.95 6.40

F F F F

2.28 0.11 51.43 1.80

F F F F

2.48 0.13 50.48 2.00

F F F F

2.11 0.17 65.33 1.76

F F F F

2.13 0.23 80.90 2.08

BMI4

Age4

BMI  age4

0.45 0.05 0.02 0.04

b 0.01 b 0.01 b 0.01 b 0.01

0.24 0.45 0.09 0.42

4 Denotes univariate 1-tailed P value.

administered with a random delay between trials of 2 to 4 seconds. The dependent variable was the average reaction time across trials. 2.1.3.2.3. Switching of attention—number. This task is a modified version of the Trail Making Test A [18]. Switching of attention—number requires participants to connect numbers in ascending sequence (ie, 1-2-3-, etc). Time to completion served as a dependent variable. 2.1.3.2.4. Span of visual memory. This test is similar to Spatial Span [19]. Participants were presented with squares arranged in a random pattern on the computer screen and asked to repeat the order in which the squares were highlighted. The total number of correct answers was the dependent variable. 2.1.3.3. Executive function. 2.1.3.3.1. Verbal interference. This test was a computerized modification of the Stroop test [20]. The total number of correct items on the color-word condition served as a dependent variable. 2.1.3.3.2. Switching of attention—letter/number. Similar to Trail Making Test B [18], participants are asked to connect numbers and letters in an ascending but alternating sequence (ie, 1-A-2-B, etc). Time to completion served as the dependent variable. 2.1.3.3.3. Maze errors. This task is a computerized adaptation of the Austin Maze [21]. Participants are presented with an 8  8 grid of circles. They are asked to find the hidden path through the grid using the arrow keys and they received feedback after each move. The task ends when the participant is able to complete the maze twice without error or after 10 minutes has elapsed. The dependent variable was the total number of errors made during this task. 2.1.4. Data analysis After screening for outliers and missing data, Pearson correlation between BMI and performance on each cognitive

test was generated. Multivariate analysis of covariance (MANCOVA) was then conducted on attention and executive function test performance using the above tests as dependent variables. Independent variables in each analysis included BMI and age group. To reduce the possible confound of demographic or psychosocial variables, we included in the covariates the estimated IQ, years of education, sex, and selfreported levels of depression, anxiety, and stress from the DASS. One-tailed Holm’s corrected posttests were used to clarify significant omnibus tests. 3. Results 3.1. Correlation between BMI and cognitive test performance Body mass index was significantly related to performance on all cognitive tests (Table 2). Although all relationships were in the expected direction (ie, higher BMI, worse performance), correlations were modest in size. The strongest relationships were found for verbal interference (r = 0.23) and switching of attention—letter/number (r = 0.20). 3.2. Body mass index, age, and attention MANCOVA revealed main effects for age (k = .77, F4,395 = 28.89, P b .001), but not BMI group (k = .98, F4,395 = 1.72, P = .15) (Table 3). No evidence for a BMI  age interaction emerged (k = .99, F 4,395 b 1, P = .60). Posttests showed that age effects were found on all tests, with younger adults outperforming older adults. 3.3. Body mass index, age, and executive function MANCOVA revealed the main effects for BMI (k = .97, F3,396 = 4.41, P = .004) and age group (k = .74, F3,396 = 47.55, P b .001) (Table 4). Again, no evidence emerged for a BMI  age group interaction (k = .99, F3,396 b 1, P = .57). Body mass index effects emerged on verbal interference and

Table 4 Means and SDs in BMI group by age for executive function measures Test Verbal interferencea Switching of attention—letter/number Maze errorsa a

Younger adults

Older adults

Normal weight

Over/obese

Normal weight

Over/obese

12.98 F 3.48 41.62 F 10.47 32.75 F 23.77

12.04 F 3.80 42.38 F 11.00 38.26 F 33.67

9.13 F 3.58 53.36 F 9.11 55.63 F 44.57

7.12 F 3.70 54.22 F 8.24 69.26 F 57.01

Denotes significance after Holm’s correction for multiple comparisons. 4 Denotes univariate 1-tailed P value.

BMI4

Age4

BMI  age4

b 0.01 0.33 0.02

b 0.01 b 0.01 b 0.01

0.12 0.45 0.21

60

J. Gunstad et al. / Comprehensive Psychiatry 48 (2007) 57 – 61

maze errors, with normal weight individuals outperforming overweight/obese persons on these measures. Age effects were found on all executive function measures, with young adults outperforming older adults. 4. Discussion Results from the present study extend past work and demonstrate that BMI is related to cognitive test performance, even in very healthy adults. Though relationships were found between BMI and all cognitive tests, only reduced executive function performance differed between normal weight and overweight/obese adults when adjusting for possible confounds. No evidence for a BMI  age interaction emerged. The finding that persons with elevated BMI have reduced executive function performance is consistent with the growing number of studies linking obesity to poor neurocognitive outcome [2-6]. Although the observed differences were not large enough to be clinically meaningful (b 1.5 SD), the present results likely underestimate the cognitive differences between normal weight and overweight/obese individuals in other settings. Excluding persons with significant medical and psychiatric conditions allowed the present study to determine the independent contribution of elevated BMI to executive function. However, overweight and obese individuals frequently have medical conditions with known cognitive consequences and may show poorer test performance than the present sample. The exact mechanism for the relationship between BMI and reduced cognitive performance remains unknown. Elevated BMI is associated with many pathophysiologic changes with the potential to negatively impact cognitive functioning, including vascular changes, impaired insulin regulation, systemic inflammation, and reduced cardiovascular fitness [22-26]. It is possible that subclinical levels of these processes account for the observed differences in our sample. However, it is also possible that persons with reduced executive function are more likely to become overweight or obese. Many aspects of executive functioning appear to have direct bearing on the ability to maintain energy balance, including impulse control, self-monitoring, and goal-directed behavior [27]. If so, the decline in executive function that is part of normal aging may thus help explain the increased prevalence of overweight and obesity with age [28-30]. Much additional work is needed to clarify the relationship between BMI and reduced cognitive performance, particularly studies examining persons before and after substantial weight loss and studies using an expanded test battery. Contrary to predictions, normal weight and overweight/ obese participants did not differ on attention test performance. Although univariate comparisons were revealed between group differences on several tests (Table 3), this effect was not large enough to achieve omnibus significance. These findings are inconsistent with the high rate of ADHD reported for obese individuals presenting for treatment (27% of patients) [7]. There are several possible

explanations for these inconsistent findings. As noted above, our stringent exclusion criteria may have resulted in a sample that is not fully representative of obese individuals presenting for treatment. Another possibility includes the manner in which ADHD was diagnosed in the earlier study, as the age of onset criteria was relaxed and may have resulted in inflated prevalence rates [7]. A final possibility involves test selection. Though the tests are reliable and are valid measures of attention, they were not specifically developed to be sensitive to ADHD [13,14]. Future studies are needed to clarify the relationship between overweight/obesity and reduced attention abilities. The ability to generalize findings to other populations may be limited in several ways. One such limitation is its reliance upon BMI to determine overweight and obese status. Elevated BMI can result from large amounts of lean muscle mass, an important consideration in our sample of very healthy individuals. It is possible that use of other indices (eg, body composition, waist-to-hip ratio) would result in different findings, as our very healthy sample may underestimate the relationship between BMI and executive function. A second limitation involves the absence of information regarding obesity duration. No study has examined the relationship between duration of obesity and cognitive performance in healthy adults, and it is possible that persons with a long history of obesity show poorer cognitive performance than those with recent onset obesity. In conclusion, the present study provides further evidence that elevated BMI is associated with reduced cognitive function and demonstrates that this pattern is also found in very healthy adults. Additional work is needed to determine the directionality of the relationship between elevated BMI and executive dysfunction, particularly prospective studies assessing cognitive performance before and after significant weight loss. Such studies may also provide important insight for effective weight loss treatment. Acknowledgment We acknowledge the collaboration with the Brain Resource International Database (under the auspices of the Brain Resource Company; www.brainresource.com) for data acquisition and methodology.

References [1] Mokdad A, Marks J, Stroup D, Gerberding J. Actual causes of death in the United States, 2000. JAMA 2004;291:1238 - 45. [2] Gustafson D, Lissner L, Bengtsson C, Bjorkelund C, Skoog I. A 24year follow-up of body mass index and cerebral atrophy. Neurology 2004;63:1876 - 81. [3] Gustafson D, Rothenberg E, Blennow K, Steen B, Skoog I. An 18year follow-up of overweight and risk of Alzheimer disease. Arch Intern Med 2003;163:1524 - 8. [4] Gustafson D, Steen B, Skoog I. Body mass index and white matter lesions in elderly women. An 18-year longitudinal study. Int Psychogeriat 2004;16:327 - 36.

J. Gunstad et al. / Comprehensive Psychiatry 48 (2007) 57 – 61 [5] Chelune G, Ortega D, Linton J, Boustany M. Personality and cognitive findings among patients electing gastroplasty for morbid obesity. Int J Eat Disord 1986;5:701 - 12. [6] Elias M, Elias P, Sullivan L, Wolf P, D’Agostino R. Lower cognitive function in the presence of obesity and hypertension: the Framingham heart study. Int J Obes 2003;27:260 - 8. [7] Altfas J. Prevalence of attention deficit/hyperactivity disorder among adults in obesity treatment. BMC Psychiatry 2002;2:9. [8] Fassino S, Leombruni P, Piero A, Abbate-Daga G, Giacomo Rovera G. Mood, eating attitudes, and anger in obese women with and without binge eating disorder. J Psychosom Res 2003;54:559 - 66. [9] Gordon E. Integrative neuroscience in psychiatry: the role of a standardized database. Australas Psychiat 2003;11:156 - 63. [10] Gordon E, Cooper N, Rennie C, Hermens D, Williams L. Integrative neuroscience: the role of a standardized database. Clin EEG Neurosci 2005;36:64 - 75. [11] Hickie I, Davenport T, Naismith S, Scott E. SPHERE: a national depression project. Med J Aust 2001;175:S2. [12] National Heart, Lung, and Blood Institute: National Institutes of Health. Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults. No. 98-4083; 1998. [13] Paul R, Lawrence J, Williams L, Clark R, Cooper N, Gordon E. Validity of IntegNeuro: a new computerized battery of neurocognitive tests. J Integr Neurosci 2005;115:1549 - 67. [14] Williams L, Simms E, Clark R, Paul R, Rowe D, Gordon E. The test-retest reliability of a standardized neurocognitive and neurophysiological test battery: bneuromarker Q. Int J Neurosci 2005;115: 1605 - 30. [15] Lovibond S, Lovibond P. Manual for the depression anxiety stress scales. 2nd ed. Sydney7 Psychology Foundation; 1995. [16] Baddeley A, Emslie H, Nimmo-Smith I. The Spot-the-Word test: a robust estimate of verbal intelligence based on lexical decision. Br J Clin Psychol 1993;32:55 - 65. [17] Sullivan R, Senior G, Hennessy M. Australian age-education and premorbid cognitive/intellectual estimates for the WAIS-III. Poster

[18] [19] [20] [21] [22] [23]

[24]

[25]

[26]

[27] [28] [29]

[30]

61

presented at the 6th Annual Conference of the APS College of Clinical Neuropsychologists, Hunter Balley, NSW, Australia; 2000. Reitan R. Validity of the Trail Making Test as an indicator of organic brain damage. Percept Mot Skills 1958;8:271 - 6. Wechsler D. Wechsler Memory Scale-III Manual. New York7 The Psychological Co; 1997. Golden C. Stroop color and word task: a manual for clinical and experimental uses. Wood Dale (Ill)7 Stoeling; 1978. Walsh K. Neuropsychology: a clinical approach. 3rd ed. Edinburgh7 Churchill Livingstone; 1994. Colcombe S, Kramer A. Fitness effects on the cognitive function of older adults: a meta-analytic study. Psychol Sci 2003;14:125 - 30. Convit A, Wolf O, Tarshish C, de Leon M. Reduced glucose tolerance is associated with poor memory performance and hippocampal atrophy among normal elderly. Proc Natl Acad Sci U S A 2003;100:2019 - 22. Rahmouni K, Correia M, Haynes W, Mark A. Obesity-associated hypertension: new insights into mechanisms. Hypertension 2005; 45:9 - 14. Teunissen C, van Boxtel M, Bosma H, Bosmans E, Delanghe J, De Bruijn C, et al. Inflammation markers in relation to cognition in a healthy aging population. J Neuroimmunol 2003;134:142 - 50. Ylikoski R, Ylikoski A, Raininko R, Keskivaara P, Sulkava R, Tilvis R, et al. Cardiovascular diseases, health status, brain imaging findings and neuropsychological functioning in neurologically healthy elderly individuals. Arch Gerontol Geriatr 2000;30:115 - 30. Lezak M, Howieson D, Loring D. Neuropsychological assessment. 4th ed. New York7 Oxford University Press; 2004. Flegal K, Carroll M, Ogden C, Johnson C. Prevalence and trends in obesity among US adults, 1999-2000. JAMA 2002;288:1723 - 7. Salthouse T, Atkinson T, Berish D. Executive functioning as a potential mediator of age-related cognitive decline in normal adults. J Exp Psychol 2003;132:566 - 94. Park H, O’Connell J, Thomson R. A systematic review of cognitive decline in the general elderly population. Int J Geriatr Psychiatry 2003;18:1121 - 34.