Relationship between elevated impulsivity and cognitive declines in elderly community-dwelling individuals

Impulse control disorders are recognized as one of the behavioral and psychological symptoms of dementia (BPSD). Majority of studies on the treatment of BPSD related to impulsivity have rather focused on the aggression and agitation. In particular, it has not been investigated how cognitive declines are associated with impulsivity in community-dwelling elderly people. Here, we have measured the cognitive and memory functions and impulsivity of 212 elderly community-dwelling people using a psychometric test battery and analyzed the correlation between their level of impulsivity and cognitive functions by multiple regression analysis. We found an elevation of impulsivity, which was evaluated by the Barratt Impulsiveness Scale-11, closely related to decline of cognitive functions, which were evaluated by the Montreal Cognitive Assessment and the Mini-Mental State Examination, and Logical Memory function, which were evaluated by the Wechsler Memory Scale-Delayed Recall. Then we have divided them into groups based on the severity of cognitive decline and conducted an analysis of each group, the result of which showed that as this tendency was particularly noticeable in the suspected dementia group. Therefore, we have concluded that heightened impulsivity is negatively associated with cognitive and memory functions in community-dwelling elderly people.

Impulsivity is a behavioral trait that could be a source of various social problems such as morbid betting, kleptomania, overeating, and violence [1][2][3][4] . Impulsivity is typically defined as a predisposition toward rapid, unplanned actions responding to internal or external stimuli without regard for the consequences they might bring to oneself or to the others 5 . Impulsivity is also recognized as an intermediate phenotype of various behavioral abnormalities, as well as a phenotype that sometimes accompanies the progression of various types of dementia, including frontotemporal dementia (FTD) [6][7][8] and Alzheimer's disease (AD) [9][10][11][12] . With the progression of dementia, cognitive functions decline in a variety of domains, and behavioral and psychological symptoms of dementia (BPSD), including impaired impulse control, are exhibited [11][12][13][14] . Worsening BPSD, makes it difficult for afflicted individuals to lead a normal social life; more often than not, people who have dementia need medical attention and caregivers' help due to their BPSD, rather than due to a decline in memory or cognitive functions [13][14][15] . In AD patients, first symptoms take the form of mild cognitive impairment, in which memory loss is the main symptom. While their daily lives are still unaffected at the first stage of AD 16,17 , impulsivity may be elevated in a particular percentage (17-40%) of AD patients 18,19 .
Both clinical [20][21][22] and animal studies [23][24][25] have suggested the implication of serotonergic neurons at the dorsal raphe nucleus in the pathophysiology of impulsivity in the series of histochemical analyses showed the degeneration of serotonergic neurons in some of AD patients [26][27][28][29][30][31] . The accumulation of senile plaques and phosphorylated tau has been also reported to be seen in the damaged raphe nucleus in the AD brain 26,31 , and elevated impulsivity was observed in transgenic mouse models of AD 25,[32][33][34][35] . More specifically, alterations in the activity of the dorsal raphe nucleus as well as the degeneration of serotonergic neurons at the dorsal raphe nucleus has been detected in an AD mouse model of APPswe/PSEN1dE9 25,35,36 . Therefore, the serotonergic system is considered to be a potential therapeutic target for AD patients 23,37,38 . On the other hand, changes in the dopaminergic system are also known to affect impulsivity 39  www.nature.com/scientificreports/ on human impulsivity is provided by studies on dopamine-related changes in patients who have Parkinson's disease (PD) such as the observation of behavioral changes that meet the criteria for impulse control disorders in some of PD patients following the administration of dopamine replacement therapies 40 . It has been reported in rat studies as well that the enhanced dopamine neurotransmission increases the number of impulsive actions, which is one of key indicators of impulsivity, while those impulsive responses are suppressed by decreased dopaminergic neurotransmission 41,42 . Enhanced brain dopamine transmission also enhances delay aversion, another indicator of impulsivity, which is suppressed by a decrease in transmission of this neurotransmitter 43 . However, as impulsive actions and delay aversion are not the same type of behavioral abnormalities, the way dopaminergic system affects or triggers each negative behavioral trait may be different, and specific mechanisms of action are proposed for each behavioral symptom that suggests heightened impulsivity. Meanwhile, several personality tests have been developed to assess the degree of impulsivity 20,44 . The Barratt Impulsiveness Scale (BIS) is one of the oldest and most widely used tests of impulsivity 44 , whereas the BIS-11 is the revised version of the original BIS by Patton et al., which includes three additional factors for assessment: motor impulsiveness, non-planning impulsiveness, and attentional impulsiveness 45 . The BIS-11 is a useful personality test that aims to measure the relationship between multidimensional impulsivity and maladaptive behavior and now translated into several languages and used worldwide 46,47 . In Japan, Kobashi and Ida have translated BIS-11 into Japanese 48 , and internal consistency, retest reliability, and construct validity of the Japanese version of the BIS-11 have been confirmed 49 .
The cognitive functions of elderly people decrease as they age 50 . However, no studies have been yet conducted on how the decline in cognitive functions are connected with high levels of impulsivity. Therefore, in this study, we aimed to investigate the relationship between cognitive decline and impulsivity using a psychometric test battery in community-dwelling elderly people.
Multiple linear regression analyses. The skewness of the total BIS-11 scores was 0.68, while the kurtosis was 4.70, which with the test of normality performed has shown that the BIS-11 scores were not normally distributed (p < 0.001). Therefore, we have converted the BIS-11 scores to a logarithmic scale for multiple regression analysis and found the normal distribution of the BIS-11 scores (p = 0.060). A set of multiple regression analyses was performed to examine the relationship between cognitive functioning and the degree of impulsivity (Tables 2, 3 and 4). Multiple linear regression was used to examine the relation between Montreal Cognitive Assessment (MoCA) score and the BIS-11 score with several covariates. A significant regression equation that has the MoCA score as the dependable variable was found with the three explanatory variables which are as follows: (F (3, 212) = 16.13, p < 0.001, R 2 = 0.19). Age, sex, and the BIS-11 score were incorporated into the multiple regression model, and these three factors turned out to be the ones that showed the most significant correlations  www.nature.com/scientificreports/ with the MoCA score (Table 2). Table 3 shows the results of a multiple linear regression that examines the relation between Mini-Mental State Examination (MMSE) score, the BIS-11 score and covariates. Here, a significant regression equation was found with the three explanatory variables (F (3, 212) = 8.70, p < 0.001, R 2 = 0.11). Among the factors incorporated into the multiple regression model, age, sex, and BIS-11 score were again significantly associated with MMSE score. Of these three factors, age and sex have shown significant linkage to the scores of both MoCA and MMSE. Table 4 shows the results of a multiple linear regression for the evaluation of the relation between WMS-DR (Wechsler Memory Scale-Delayed Recall) score, the BIS-11 score and covariates. This time, a significant regression equation was found with the five explanatory variables (F (5, 212) = 11.45, p < 0.001, R 2 = 0.22). Among the factors incorporated into the multivariable model, age, years of education and BIS-11 score were significantly associated with the WMS-DR score ( As a result of analysis of covariance (ANCOVA) using the total score of BIS-11 as the objective variable, group as the fixed factor, and age, sex, and education as covariates, a significant regression equation was found (p = 0.004; significance test for linear regression). The remarkable difference in the total BIS-11 score between groups was quite significant (p = 0.001), and a post-hoc test revealed that Group 3 had scored much higher than the other two groups (Table 5).

Discussion
This study aims to investigate the relationship between the level of impulsivity and the decline in cognitive functions in the elderly population. In the community-dwelling elderly people, we found a strong correlation between elevated impulsivity and cognitive declines. Levels of impulsivity was measured by BIS-11 questionnaire, and cognitive abilities were assessed by three psychiatric tests which are as follows: MMSE, MoCA , and WMS-DR.
In this study, we found a clear negative correlation between high levels of impulsivity and cognitive functions by analyzing the scores in MoCA, MMSE, and BIS-11. We must also discuss correlation coefficients between three cognitive test we have used. The correlation coefficient was 0.62 for MoCA and MMSE, 0.42 for MoCA and WMS-DR, and 0.35 for MMSE and WMS-DR. Guilford defines, -the magnitude of a correlation coefficient as follows:-Less than 0.20……. Slight; 0.20-0.40…….  53 . Since both MMSE and MoCA were developed as a screening tests for cognitive impairment, one might expect their correlation number should be higher. However, MoCA is a more rigorous test for MCI and focuses more on language and execution functioning skills, which require higher cognitive abilities than MMSE; therefore, the correlation coefficient between MoCA and MMSE turns out to be rather moderate.
As discussed, the comparison results of MoCA and MMSE, and BIS-11 suggest a strong correlation between cognitive severity stages and the degree of impulsivity. Besides, WMS-DR is a sub-category of WMS, which is a test that evaluates the episodic memory function by recalling a meaningful story presented verbally after a while 54 . Therefore, the results regarding the relationship between BIS-11 and WMS-DR in this study might indicate relationship between impulsivity and episodic memory rather than general cognitive functions. This is the first study that focused on the negative correlation of the BIS-11 score with cognitive functions of the elderly, but the idea that the stages of dementia is closely associated with the level of impulsivity is not new; in fact, in other personality tests that assess the impulsivity-agitation domain such as Urgency Premeditation Perseverance Sensation Seeking (UPPS) Impulsive Behavior Scale 11,12 or Cohen-Mansfield Agitation Inventory (CMAI) 9,13 , it is already shown that high levels of impulsivity are seen in some of the AD patients and it itself is one of, the symptoms of BPSD. Impulsivity-agitation domain is one of the most troublesome BPSD 13 and measuring BIS-11 in the elderly may lead to a supplementary diagnosis with individuals with a rapid decline in cognitive functions.
We observed gender differences in cognitive decline. A multiple regression analysis of the scores of the MMSE and MoCA tests has shown that female subjects scored much higher than the male counterpart. This tendency has been already observed in MMSE test scores 55 . Although it may sound rather obvious participants' age was also negatively associated with both cognitive and memory functions. This supports previous findings that among the clinically normal elderly, cognitive and memory functions decline with aging [56][57][58] .
Our study has some limitations. First, the sample size may not have been sufficient to detect all the changes in impulsivity levels that occur along with the progression of dementia. Since the sample population in this study was not patients with dementia but the community-dwelling elderly, the number of subjects with suspected dementia was relatively few. Second, since it is sometimes hard to distinguish AD symptoms from other types of cognitive impairment without a definitive diagnostic tool such as magnetic resonance imaging (MRI) scanner, which we were unable to use in this study, we cannot entirely rule out the possibility that subjects we regarded as having AD might have had other types of dementia. However, we did our best by collecting a medical history, if available, of subjects showing the signs of cognitive impairment, or by a perusal of their medical interviews with physicians to know the symptoms subjects had, before determining the type of dementia they were most likely to have. Here are the four classifications of dementia we examined: FTD, vascular dementia (VaD), and dementia with Lewy bodies (DLB). AD is the most common type of dementia, and 70% of patients with dementia is said to have AD. VaD comes next and account for 20%, whereas FTD and DLB take up only a small fraction of people with dementia-5% for each 59 . VaD comes from abnormalities formed in cerebral blood vessels and having had cerebral infarction or cerebral hemorrhage is a huge risk factor for VaD, which often takess the form of sudden dementia 60 . We have to note here that in this study, none of the subjects with suspected dementia had cerebral infarction or cerebral hemorrhage before participating in our research. On the other hand, DLB is a type of dementia caused by accumulation of Lewy bodies and characterized by a combination of cognitive decline and core symptoms that include Parkinson's syndrome, REM sleep disorder and visual hallucinations 61 , however, in none the subjects with suspected dementia were seen such characteristic symptoms of DLB. FTD is a type of dementia in which another characteristic protein which is called (Pick body) accumulate in the prefrontal cortex and temporal lobes, and its main symptoms consist of personality changes and behavioral disorders 62 . The proportion is 5% of all dementia patients, which is considerably lower than that of AD 59 . Based on these facts, we determined that the majority of the subjects with cognitive decline in this study were likely to have mild AD. The last limitation we must admit is that we did not explore other possible causes of high levels of impulsivity, mainly previously undiagnosed psychiatric disorders such as depression or adult attention deficit hyperactivity disorder (ADHD); as our focus was on the aspects and symptoms of cognitive impairment senior citizen may have, we did not conduct any test for psychiatric disorders.
Impulsivity is an intermediate phenotype that leads to various behavioral abnormalities, and impulse control disorder is recognized as one of BPSD in dementia including AD. Symptoms of BPSD vary greatly depending on Table 5. Sub-group analysis based on the severity of cognitive dysfunction. *Significant difference from the value of G1 group, p < 0.05 (Bonferroni post-hoc-test). # Significant difference from the value of G2 group, p < 0.05 (Bonferroni post-hoc-test). † Data are the number of people. p-value for the significant difference between groups was calculated by chi-squared test. § Data are mean ± standard deviation. p-value for the significant difference between groups was calculated by one-way analysis of variance (ANOVA). ¶ Data are mean ± standard deviation. p-value for the significant difference between groups was calculated by analysis of covariance (ANCOVA). www.nature.com/scientificreports/ the type of dementia and the severity of cognitive decline and each symptom is a subject of study on its own 63,64 . However, certain symptoms of BPSD have a tendency to develop in clusters which are called domains 65 , and they may have the same etiology, suggesting that correlated symptoms better be studied as a group 66 . Clustering approach yields following domains in BPSD: Affective, Apathy, Psychosis, Euphoria, and Hyperactivity-Impulsivity-Irritability-Disinhibition-Aggression-Agitation (HIDA) Domain 67 . As its name suggests, impulsivity belongs to the HIDA domain, which as a whole is difficult to manage and known to impose a heavy burden on caregivers 18,68 . However, as aggression and agitation are more serious problems that threaten patients' relatives and healthcare workers, studies on the treatment of HIDA Domain usually focus on the reduction of aggression and agitation 13 , and seldom do they touch upon problems that arise from impulsive behaviors or the way to handle it. We must mention that what we call impulsivity here and we have measured using BIS-11 in this study are purely personal traits and do not have physical components in it, such as the act of aggression or agitation itself; hence it could be considered to be independent of those two traits in the HIDA Domain. What is unique about our study is that although there are studies that have focused on impulsivity as one of symptoms of dementia, none had investigated the direct connection between impulsivity and cognitive functions in general population.
In conclusion, the present study demonstrated that high levels of impulsivity are associated with poor cognitive and memory functions in the community-dwelling elderly because the elevated impulsivity measured by BIS-11 was a predictor of cognitive decline measured by MoCA, MMSE, and WMS-DR.

Methods
Participants. Based on the United Nations agreed cut-off for elderly age 69 , we recruited community-dwelling elderly individuals over the age of 60 who have not been diagnosed with dementia. This study was approved by the Ethics Committee of the University of Tokyo. All participants gave their written informed consent before participating in this study, which was conducted in accordance with the Declaration of Helsinki.
Outcome measures. Cognitive functions were evaluated using the MMSE and the MoCA. Logical memory function was evaluated using the WMS-DR, which belongs to a sub-category of the WMS-Revised 70 . High scores in the MMSE and the MoCA indicate better cognitive function, and high score in the WMS-DR indicate better logical memory function. MMSE is a cognitive assessment test published in 1975 and is currently the most common screening test for dementia 51,71 . Therefore, in this study, we adopted dementia cut-off values of the MMSE for classification criteria for the group with suspected dementia. However, the diagnostic accuracy of the MMSE in detecting MCI is suggested to be rather modest 72,73 , and the MoCA was developed as a more challenging one to address this issue the MMSE had 52,74 ; in fact, the MoCA is known for the improved sensitivity of detection for MCI 75,76 . For that reason, in this study, we adopted the MCI cut-off criteria of the MoCA for classification criteria for a group with suspected MCI. We divided subjects into three groups based on their cognitive scores, using suspected dementia cut-off points of MMSE (23/24) 51 and suspected MCI cut-off points of MoCA (25/26) 52 . Group 1 (MoCA ≥ 26 and MMSE ≥ 24) corresponded to cognitive normal elderly, Group 2 (MoCA ≤ 25 and MMSE ≥ 24) corresponded to suspected MCI stage, and Group 3 (MoCA ≤ 25 and MMSE ≤ 23) corresponded to suspected dementia. Impulsivity was evaluated using the Japanese version of the BIS-11 49 . Higher scores reflect higher degrees of impulsivity. Statistical analysis. This study was designed assuming a Cohen's f 2 for an effect size of 0.08 with a type 1 error protection of 0.05 two-sided and 90% of the power. The number of subjects needed was calculated to be 205. To investigate the relationship between impulsivity and cognitive/memory function, we performed a multiple linear regression analysis with MoCA, MMSE, and WMS-DR scores as objective variables, and BIS-11 score as explanatory variables. A Variance Inflation Factor (VIF) value over 2.5 is considered problematic for multicollinearity 77 . We determined that there was no multicollinearity between the explanatory variables since no factor had a VIF above 2.5. The items used as candidates for covariates were age, sex, body mass index, and years of education. Cognitive functions generally decline with age 59 , and gender differences are reportedly observed in broad areas of cognitive functions even in the elderly who do not yet show obvious symptoms of dementia 58 . It is said that years of education the elderly had also affect the scores of cognitive assessment tests. Previous studies on the association between cognitive functions and Body Mass Index (BMI) of the elderly have produced conflicting results, but as some studies have reported a positive association between them 78 , we adopted BMI as well for a candidate covariate. We calculated the skewness and kurtosis of the score of BIS-11 and tested for normality. The test of normality was performed by the Shapiro-Wilk test. As a result, the BIS-11 score in the linear scale was not normally distributed. Therefore, we converted the scores of BIS-11 to a logarithmic scale. As a result, the BIS-11 score converted to the logarithmic scale was normally distributed. In the multiple regression analyses, all variables including covariates were standardized by Z score. The combinations of candidates for covariates with the lowest Akaike Information Criterion were adopted as the final multiple regression model 79 . In the analyses of severity of cognitive dysfunction, all variables were standardized by Z score, and ANCOVA was performed with total BIS-11 score as the objective variable, the severity of cognitive dysfunction level as the fixed factor, and age, sex, and years of education as covariates. A p-value of less than 0.05 was defined as statistically significant. Microsoft Excel Add-in for multiple linear regression analysis was used as a data analytic tool in this study.

Data availability
Data and materials can be obtained by contacting the corresponding author. www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.