Evaluation of proxy measures of impulsivity in relationship to obesity

(1)

Självständigt arbete (30 hp) HT-17

Institutionen för Neurovetenskap

Läkarprogrammet, Uppsala Universitet

Evaluation of proxy measures of

impulsivity in relationship to obesity

Author:

Supervisors:

Jonathan Andersson Helgi Schiöth, PhD Gaia Olivo, MD

Department of Neuroscience

(2)

Abstract

Background: Obesity is rapidly becoming a global epidemic, and previous research has shown that

impulsivity seems to have a role to play in the development of the disease.

Purpose: The aim of this study was to investigate the association between impulsivity and obesity in

a population.

Methods: Data from 496 269 individuals participating in the UK Biobank project was used to build a

proxy model for impulsivity using impulsivity-related variables. Regression analysis was then used to estimate the relationship between this measure of impulsivity and two important indicators of general obesity and abdominal adiposity, body mass index (BMI) and waist-to-hip ratio (WHR).

Results: Highly impulsive individuals were found to be 59,4% more likely to be obese (BMI ≥ 30) and

21,4% more likely to have an unhealthy waist-to-hip ratio (WHR > 0,9 for men, WHR > 0,85 for women) than non-impulsive individuals. This relationship was much stronger for men, with a 103,5% increased risk of obesity and 33,3% increased risk of having an unhealthy waist-to-hip ratio. Highly impulsive women, on the other hand, only showed a 15,6% increased risk of obesity and an 8,6% increased risk of having an unhealthy waist-to-hip ratio.

Conclusions: A significant association between impulsivity and obesity could be seen in a large

sample of the general UK population, and this association was much stronger for men. This could have implications for future approaches to the prevention and treatment of obesity.

Sammanfattning

Fetma är ett växande problem, och ett globalt sådant (WHO, 2000). Man uppskattar att över 700 miljoner människor på jorden har ett BMI över 30, vilket är definitionen av fetma (Afshin, et al., 2017). Det är en stark riskfaktor för en rad olika sjukdomar, allt från diabetes och hjärt- och

kärlsjukdom till artros och många olika typer av cancer (Haslam & James, 2005). Impulsivitet har som personlighetsdrag kopplats till fetma och övervikt i flertalet olika studier (Stice, 2002).

Syftet med den här studien har varit att undersöka om detta samband går att påvisa i ett stort urval av en befolkning, och vilka slutsatser som går att dra av resultatet av den statistiska analysen av det sambandet. För att uppnå detta har uppmätta variabler från 496 269 människor studerats, insamlade och organiserade av biobanksprojektet UK Biobank (Collins, 2012).

Eftersom ett mer direkt mått på impulsivitet inte har varit tillgängligt, har en modell för impulsivitet baserat på impulsivititetsrelaterade variabler konstruerats istället. Dessa variabler, bestående av svar på frågor som ställts i webbaserade frågeformulär, kretsar kring deltagarnas personliga erfarenheter av rökning, fortkörning och riskfyllt sexuellt beteende, som alla är exempel på fenomen som i tidigare

(4)

studier har kopplats till impulsivitet och bristande impulskontroll (Lopez-Torrecillas, et al., 2014) (Bachoo, et al., 2013) (Deckman & DeWall, 2011).

Resultaten från denna studie visar att individer som uppvisar höga nivåer av impulsivitet hade 59,4% större risk att lida av fetma, mätt i kroppsmasseindex (BMI), jämfört med individer med låga nivåer av impulsivitet, och 21,4% större risk att ha en midjemåttkvot (WHR) som av WHO anses vara ohälsosam (WHO, 2008), och att båda dessa associationer var starkare för män än för kvinnor. Dessa resultat bekräftar det samband mellan impulsivitet och fetma som påvisats i tidigare studier, samt att det finns en könsskillnad när det gäller detta samband, men visar också att sambandet även gäller bukfetma och inte bara fetma generellt. Detta är av betydelse eftersom bukfetma har visat sig vara en bättre indikator än fetma när det gäller risken för hjärt- och kärlsjukdom (Huxley, et al., 2010) (Després, 2006), och eftersom bukfetma utgör en riskfaktor för hjärt- och kärlsjukdom oberoende av kroppsmasseindex (Cameron & Zimmet, 2008).

Mer forskning behövs för att ta ställning till vilken roll olika personlighetsdrag, som impulsivitet, har att spela vid utformning av framtida strategier för prevention och behandling av övervikt och fetma, och huruvida könsskillnader i de psykologiska mekanismer som ligger till grund för hur personlighet interagerar med vikt är något som borde tas större hänsyn till vid utformning av sådana strategier.

Background

The problem of obesity

The rapid increase of obesity is becoming a global problem of epidemic proportions (WHO, 2000). In 2015, there were over 100 million obese children and over 600 million obese adults in the world (Afshin, et al., 2017). In 2010, overweight and obesity was estimated to cause 3-4 million deaths worldwide (Lim, et al., 2012). In the last 33 years, the prevalence of obesity has risen significantly in both the developed and the developing world, by an average of 27% for adults and 47% for children (Ng, et al., 2014). This is a potentially devastating development, since obesity is a major risk factor for a number of diseases, including cardiovascular disease, type 2 diabetes, hypertension, chronic kidney disease, osteoarthritis, sleep apnea, Alzheimer’s disease, depression and many types of cancer, and has been shown to significantly increase mortality and decrease life expectancy (Haslam & James, 2005) (Razay, et al., 2006).

Today, some countries have obesity rates that reach, or even exceed, 50% of the adult population. The discussion regarding the explanation for this global increase in obesity has revolved around several different potentially contributing factors, like increases in energy intake, decreases in physical activity, and changes in the gut biome (Ng, et al., 2014).

There are many different ways to measure obesity, but the most common ones have traditionally been body mass index, waist circumference and waist-to-hip ratio, the former being a general measure of obesity and the latter two being measures of abdominal obesity. There has been much discussion about which of these is the best predictor of morbidity and mortality (WHO, 2008).

(5)

Abdominal obesity is the buildup of fat in the abdominal cavity, and has been strongly linked to cardiovascular disease, which is the main cause of all obesity-related deaths (Yusuf, et al., 2004). There is evidence that abdominal obesity is a risk factor for cardiovascular disease independently of body mass index (Després, 2006) (Cameron & Zimmet, 2008).

One review looked at all measures of obesity and found that measures of abdominal obesity (waist circumference and waist-to-hip ratio) were better than general measures of obesity (body mass index) at predicting cardiovascular disease, although using a combination of both types of measures might be even better (Huxley, et al., 2010).

The connection to impulsivity

Impulsivity can be defined as a predisposition toward rapid and unplanned reactions to either internal or external stimuli, without proper regard to the consequences of these reactions (Moeller, et al., 2001). As such, impulsivity, or lack of impulse control, has been widely linked to obesity (Stice, 2002) (Chamberlain, et al., 2015) (Lange, et al., 2014). A study of 50 000 French participants from 2017 showed a strong connection between impulsivity, as measured through the Barratt

Impulsiveness Scale, and body mass index. This association was stronger in men, and those in the highest range of impulsivity were more likely to be Class III obese (Bénard, et al., 2017). Another study, which looked at the influence of different personality traits on body mass index and other measures of adiposity, found that impulsivity had, by far, the strongest association with these measures. For example, those who scored in the top 10% of impulsivity weighed, on average, 11 kg more than those who scored in the bottom 10% (Sutin, et al., 2011).

Using a structural model of personality to understand impulsivity, one study administered a self-report inventory designed to measure different personality aspects, along with various impulsivity measures, to 500 participants, and was thus able to identify four distinct personality facets that were closely associated with impulsive behavior. These were urgency (the tendency to commit rash actions), lack of premeditation (the tendency to not plan actions beforehand), lack of perseverance (the tendency to not complete tasks) and sensation seeking (the tendency to seek excitement and adventure). A scale based on answers to self-report questionnaires, the so-called UPSS Impulsive Behavior Scale, was created to measure each of these facets (Whiteside & Lynam, 2001). Through statistical analysis of questionnaires administered to 47 obese participants compared to 47 normal-weight controls, three out of four of these facets, all except lack of premeditation, were found to be associated with overweight and obesity (Mobbs, et al., 2010).

One of the most commonly used measures of impulsivity in a research setting is the Barratt Impulsiveness Scale (BIS-11), another self-report questionnaire consisting of 30 items where the answers are scored on a 4-point scale, divided into three sections, measuring inattention and cognitive instability, motor impulsivity and lack of perseverance, and lack of self-control and cognitive complexity (Swann, et al., 2002).

Other widely used measures of impulsivity, beside self-report questionnaires, are behavioral tasks such as Delay discounting tests, where subjects are given a choice between a small immediate

(6)

reward or a bigger delayed reward, and Go/no-go tests, where subjects are instructed to press a button when presented with a “go” signal, unless a “no-go” signal is also given, in which case they must resist the impulse to press the button, thus measuring inhibitory control (Winstanley, et al., 2006).

There is a strong, but not perfect, correlation between tests results from self-report instruments and behavioral tasks, and it is assumed that this fact reflects their tendency to capture slightly different aspects of impulsivity, but nonetheless, both measures have been found to be associated with behaviors such as binge eating or over-eating (Guerrieri, et al., 2008).

Impulsivity and gender

Research suggests that there are gender differences on a trait level when it comes to how impulsivity is exhibited (Lundahl, et al., 2015). According to one study, men display higher rates of sensation seeking and urgency, and possible also lack of perseverance (Cyders, 2013). A meta analysis of the gender differences in impulsivity looked at data from 277 different studies and found that men exhibit significantly higher levels of sensation seeking and risk taking, but that no gender differences could be found when it came to reward sensitivity, delay discounting or executive function tasks (Cross, et al., 2011). Furthermore, there is evidence to suggest that although impulsivity is associated with obesity in both genders, it is possible that the mechanism behind this interaction could be different in men and women (Lange, et al., 2014).

The reason behind gender differences in impulsivity has been the topic of much discussion. Sensation seeking can be described as the need for varied and novel sensations and experiences, and the willingness to take physical and social risks in the pursuit of them (Zuckerberg, 1979). Testosterone levels have been found to have a strong correlation with sensation seeking, as well as prioritization of short-term goals, competition and impulsivity in general (Archer, 2006). From a neurobiological perspective, levels of sensation seeking in the personality of an individual is the result of the balance between the attraction of excitement, mediated by the effects of dopamine on reward and

approach behavior, and the avoidance of danger, mediated by serotonin through its effects on inhibition and restraint. Thus, the higher levels of sensation seeking seen in men have been proposed to be the result of a more reactive dopaminergic system (Zuckerman & Kuhlman, 2000).

Impulsive behaviors

Impulsivity has been found to be associated with a number of psychiatric disorders, like ADHD, Borderline, Bipolar and Antisocial Personality Disorders (Moeller, et al., 2001), but is also an important factor in many types of everyday habits and behaviors seen in the general population. Tobacco consumption, for example, has been linked to various aspects of impulsivity. Research has shown that sensation seeking is related to the initiation of the smoking habit, while dysfunctional inhibitory control is related to the continuation of it (Balevich, et al., 2013). Furthermore, impulsivity

(7)

as a personality trait has been shown to be an important predictor for the outcomes of smoking cessation attempts (Lopez-Torrecillas, et al., 2014) (Celma-Merola, et al., 2017).

There is also an association between impulsivity and many different types of risk taking. Multiple studies, for example, have shown a positive connection between impulsivity and risky sexual behavior (Deckman & DeWall, 2011) (Charnigo, et al., 2013) (Derefinko, et al., 2014) (Dir, et al., 2014). Examples of what could be considered as risky sexual behavior include unprotected sex, sex while intoxicated and sex with multiple partners (Fortunato, et al., 2010) (Beadnell, et al., 2007). Impulsivity as a personality trait has been shown to be an important predictor both for the lifetime number of sexual partners (Muchimba, et al., 2014) and for the age of sexual debut (Khurana, et al., 2012) (Hipwell, et al., 2010).

Another form of risk taking where research has found an association with impulsivity is speeding. One study found that risky driving was associated with all aspects of impulsivity; urgency, lack of premeditation and perseverance, and sensation seeking (Bachoo, et al., 2013). In simulated driving tests, unsafe driving (of which speeding is one example) has been shown to be associated with poor impulse control in young drivers (Hatfield, et al., 2017). Two different studies compared people who had been fined by the police for exceeding the speed limit to control groups. One of them found that they were more likely to score highly on both the Barratt Impulsivity Scale and the Maladaptive Impulsivity Scale (Eensoo, et al., 2010), while the other found that they scored lower on a Go/no-go task measuring inhibitory control (Fearghal & Gormley, 2013), indicating in different ways that impulsivity is an important factor in the type of driving behavior that speeding comprises.

Aims of the study

The aims of this study were to take advantage of the enormous amount of variables contained in the large-scale repository of the UK Biobank, consisting of data from 502 000 individuals, to build a proxy model for impulsivity based on impulsivity-related variables, and investigate whether or not any associations could be found between this model and various adiposity measures, like body mass index and waist-to-hip ratio, seeking to build on earlier research by providing further proof of the connection between impulsivity and obesity, and also showing whether or not it is possible to investigate this association in the statistical analysis of a large population sample even when there is no direct measure of impulsivity available.

Methods

The UK Biobank project is a long-term British study following a total of around 500 000 volunteers, all aged 40-69 at the time of enrollment, for a minimum of 25 years, in order to investigate the complex interaction between genetic predisposition and environment that occurs in the development of disease, with the aim of improving the prevention, diagnosis and treatment of a wide range of diseases (Collins, 2012) (Ollier, et al., 2005). Data recorded includes measurements of physiological

(8)

parameters, blood and urine samples, radiological examinations and answers to extensive web-based questionnaires on various topics ranging from medical history and cognitive function to nutritional habits and lifestyle choices (Ollier, et al., 2005).

496 269 individuals participating in the UK Biobank project were selected for statistical analysis. Excluded were only those with current or previous psychiatric illnesses. Outliers with a Z-score exceeding ± 3,29 were discarded for all continuous variables. Five different parameters were then selected that, with basis in previous research on the subject, could be thought to correlate with impulsivity; “Number of unsuccessful smoking cessation attempts”, “Lifetime number of sexual partners”, “Age at first sexual intercourse”, “Frequency of driving faster than the legal speed limit” and “Risk taking behavior”.

When answering the question “Number of unsuccessful smoking cessation attempts”, participants were free to provide any number of their choosing. Discarding outliers, the median number provided was 2,46 and the range 0-27. For the questions “Age at first sexual intercourse” and “Lifetime number of sexual partners” respondents were free to provide any number, and after any outliers were discarded, the means were 18,96 (range 11-33) for age of sexual debut and 6,23 (range 1-150) for number of sexual partners.

As for the question “Frequency of driving faster than the legal speed limit”, respondents were asked “How often do you drive faster than the legal speed limit on the motorway?” and if applicable, were able to choose from four answers: “Never/rarely”, “Sometimes”, “Often” or “Most of the time”. A majority of people, 85,2%, answered that they do so “Never/rarely” or “Sometimes”, whereas only 14,8% of respondents answered “Often” or “Most of the time”.

When assessing “Risk taking behavior”, respondents were asked “Would you describe yourself as someone who takes risks?” and were able to answer either “Yes” or “No”, or to decline to answer. Out of those who provided an answer, 26,9%

answered yes and 73,1% answered no. A Principal Component Analysis (PCA) was performed on these selected variables, and a principal component was then extracted from them. This component was able to explain 29,7% of the variance in the data, and

correlated in an expected way with each of the variables. The individuals who expressed high values of this proxy measure of

impulsivity had their sexual debut at an earlier age, had an increased number of lifetime sexual partners, had unsuccessfully tried to quit smoking more times, were more prone to speed on the highway, and, when asked, were more likely to describe their own behavior as “risk taking”.

(9)

The population sample was then divided into three categories. Those in the top quartile of this impulsivity factor extracted from the PCA were deemed to exhibit “High impulsivity”, those in the middle two quartiles “Medium impulsivity” and those in the lowest quartile “Low impulsivity”. With regards to measures of obesity, the population was first grouped according to their body mass index into categories of “Underweight” (BMI < 18,5). “Normal weight” (18,5 ≤ BMI < 25),

“Overweight” (25 ≤ BMI < 30) and “Obese” (BMI > 30). Those in the “Obese” category were also further divided into “Class I Obese” (30 ≤ BMI < 35), “Class II Obese” (35 ≤ BMI < 40) and “Class III Obese” (BMI > 40) according to official WHO classifications of weight categories (WHO, 2000). With regards to abdominal adiposity, the population was divided into groups with either an “Unhealthy” (WHR > 0,9 for men and WHR > 0,85 for women) or a “Healthy” (WHR ≤ 0,9 for men and WHR ≤ 0,85 for women) waist-to-hip ratio (WHO, 2008).

Multinomial regression analysis was then performed in order to ascertain the relationship between impulsivity and each of these two measures. Age, sex, education level, employment status, total household income, ethnic background, smoking status and alcohol intake frequency were all controlled for as potential confounders (Granö, et al., 2004) (Chamorro, et al., 2012) by including them as independent variable. Odds Ratios and 95% Confidence Intervals were then calculated. Body mass index (as a 4-level variable first and then again as a 6-level variable to estimate the associations with the different subcategories of obesity) and waist-to-hip ratio (as a 2-level variable) were set as dependent variables in their respective regression analyses, with normal weight and healthy waist-to-hip ratio used as reference categories. A significant interaction between impulsivity and gender was found, and because of this, all regression analyses were also performed separately for men and women, and sex was excluded as an independent variable in these. All statistical analysis was performed in SPSS (version 23).

As for the ethical considerations of the work presented here, the UK Biobank study itself has been approved by the North West – Haydock Research Ethics Committee, and all projects at the

Department of Neuroscience at Uppsala University making use of the repository provided by the UK Biobank have been approved by the Regional Ethical Review Board in Uppsala.

Results

Population sample statistics

Out of the population of 496 269 individuals, 270 396 (54,5%) were women and 225 873 (45,5%) were men. All subjects were between the ages of 40 and 69 and the median age was 58.

With the regards to education, the sample was well-educated, with 32,8% having a college or university degree. Ethnically, the sample was partially mixed, but predominantly white. The most common ethnicities were “British” (88,6%), “Other white” (3,4%) and Indian (1,2%). In terms of current employment status, 57,8% of the subjects were employed, 4,7% were unemployed or on disability pension, and 33,4% were retired. Income-wise, 22,5% of the sample reported having a total

(10)

household income of less than 18 000 £ per year, 25,5% answered 18 000 £ to 30 999 £, 26,2% answered 31 000 £ to 51 999 £ and 20,4% had an income over 100 000 £.

With regards to tobacco consumption, a slim majority, 54,9%, were never-smokers and 10,4% were current smokers. As for alcohol consumption, only 8% of the subjects were non-drinkers. 49,0% reported an alcohol intake frequency of 1-4 times per week, and 20,3% of the sample a “daily or almost daily” alcohol intake.

Overall, excess weight was very prevalent in the population sample (see Figure 1). Only 32,8% of the subjects were considered to have a normal weight. Obesity in general was slightly more common in men, 25% compared to 22,7% of the women. However, extreme cases of obesity was actually slightly more

common in women, of whom 6,9% were considered to be Obese Class II-III (BMI > 35) compared to 6,1% of men.

A gender imbalance was also noted in the waist-to-hip ratio classification (see Figure 2), where 70,6% of women fulfilled the criterion for a waist-to-hip ratio measurement considered by the WHO to be healthy, compared to only 29,4% of men. In total, 23,8% of the subjects were considered obese and 48,9% had an unhealthy waist-to-hip ratio.

In terms of the impulsivity measure, there was also a clear difference between the genders (see

Figure 3). In general, men were more likely to be considered to be highly impulsive than women. For example, in this model 31,2% of all men were deemed to be highly impulsive, compared to only 17,0% of women. At the other end of the spectrum, 30,6% of women were classified as having low impulsivity, compared to only 20,6% of men.

(11)

Multinomial regression analysis

In the regression analysis, highly impulsive individuals were found to be 59,4% more likely to be obese (OR = 1,59, 95% CI: 1,50-1,70, p < 0,001) and 21,4% more likely to have an unhealthy waist-to-hip ratio (OR = 1,21, 95% CI: 1,15-1,28, p < 0,001) than non-impulsive individuals, after controlling for potential confounders (see Figures 4 & 5).

However, it must be noted that this relationship was much stronger for men, with high levels of impulsivity giving a 103,5% increased risk of obesity (OR = 2,14, 95% CI: 1,96-2,33, p < 0,001) and a 33,3% increased risk of having an unhealthy

waist-to-hip ratio (OR = 1,33, 95% CI: 1,24-1,43, p < 0,001). Highly impulsive women, on the other hand, only showed a 15,6% increased risk of obesity (OR = 1,16, 95% CI: 1,05-1,27, p = 0,003) and a 8,6% increased risk of having an unhealthy waist-to-hip ratio (OR = 1,09, 95% CI: 1,01 – 1,17, p = 0,034).

For the different subcategories of obesity, impulsivity was more strongly associated with milder, so-called Class I Obesity (30 ≤ BMI < 35), in women and more severe Class II-III Obesity (BMI > 35) in men, though it must be noted the results for Class II-III Obesity in women were not statistically significant. The strongest risk increases found in the regression analyses were for Class I Obesity in women (OR = 1,17, 95% CI: 1,05-1,30, p = 0,003) and Class II Obesity in men (OR = 2,12, 95% CI: 1,98-2,55, p < 0,001).

(12)

Discussion

Using a proxy model for impulsivity based on impulsivity-related variables, a statistically significant association between impulsivity and measures of obesity could be seen in a large sample of the UK population. This builds on much earlier research providing evidence of a connection between impulsivity and weight (Sutin, et al., 2011) (Mobbs, et al., 2010) (Chamberlain, et al., 2015).

From the results of the regression analysis, there is also evidence of a clear gender difference when it comes to this connection. This is in line with previous findings that indicate not only that impulsivity may tend to follow different patterns depending on gender, but also that the interaction of

impulsivity with the psychological mechanisms underlying eating habits and weight might work differently in men and women (Cyders, 2013) (Lundahl, et al., 2015) (Lange, et al., 2014).

In terms of the proxy measure itself, men were much more likely than women to be considered to exhibit high levels of impulsivity. In order to explain this, it is important to look at gender differences in the underlying foundation of the model. Although impulsivity may have a slightly stronger

correlation with risky sexual behavior for women (Dir, et al., 2014), men are not just more prone to drug use (including tobacco), but are also much more likely to engage in not only reckless driving, like speeding on the highway, but many other types of risk taking behavior (Cross, et al., 2011). This might indicate that the proxy model reflects sensation seeking and risk taking, two related concepts which are both overrepresented in men, possibly because of higher testosterone levels (Archer, 2006) and a more reactive dopaminergic system (Zuckerman & Kuhlman, 2000), to a greater extent than other facets of impulsivity.

One study aimed to examine the association between impulsivity and weight in a large sample of the general French population and found, just like this study did, that those with high impulsivity levels were more likely to be obese, and that this association was stronger for men (Bénard, et al., 2017). In comparison, three major advantages of this study was its much larger sample size (496 269

compared to 51 043), the inclusion of a different measure of adiposity besides body mass index, and the fact that weight and height were actually measured, not self-reported.

On the other hand, in the French study the widely accepted Barratt Impulsiveness Scale (BIS-11) was used to measure impulsivity, whereas this study relied on a proxy model to estimate impulsivity. Interesting to note, though, is how strong the correlation between impulsivity and obesity was in the regression analysis using the proxy model, showing that impulsivity does not necessarily need to be estimated though the use of a dedicated measure. The fact that the variables used to construct the proxy model (“Number of unsuccessful smoking cessation attempts”, “Lifetime number of sexual partners”, “Age at first sexual intercourse”, “Frequency of driving faster than the legal speed limit” and “Risk taking behavior”) appeared to function so well as proxy measures of impulsivity can also give valuable guidance in terms of what types of questions might be valuable to include when it comes to the design of questionnaires for future population studies.

Besides the association between impulsivity and body mass index that was investigated in both studies, in this study, regression analysis was also performed to estimate the association between impulsivity and waist-to-hip ratio. This is a major advantage, since measures of abdominal obesity

(13)

(waist circumference and waist-to-hip ratio) have, in studies, proven to be better than body mass index at predicting the risk of cardiovascular disease (Huxley, et al., 2010), and since it has been shown that abdominal obesity increases the risk of cardiovascular disease regardless of body mass index (Després, 2006) (Cameron & Zimmet, 2008).

As for potential confounders controlled for in the regression analysis, both studies included age, education level, total household income, smoking status, alcohol intake and occupational status. This study also included ethnic background as a potential confounder, while Bénard, et al., on the other hand, included physical activity and matrimonial status. They did not, however, exclude subjects with current or previous psychiatric illness from the sample, something this study did. This is important because impulsivity is a key feature of many types psychiatric illness (Jakuszkowiak-Wojten, et al., 2015), and not only Bipolar Disorders, Anxiety Disorders and Substance Abuse Disorders but also chronic exposure to antidepressants have been found to have a strong association with both impulsivity and obesity (Galez, et al., 2014). Thus, failure to exclude these patients could result in a bias and skew the results of the analysis.

In summary, both studies have advantages and disadvantages when compared to each other, but the results show the same patterns – a clear and significant association between impulsivity and obesity in a population sample, and a distinct gender difference when it comes to the strength of that association.

In previous research, very few other studies have explored the gender differences of this association in a similar fashion. One American study looked at psychological, neurophysiological and genetic measures of impulsivity and compared them to body mass index in a population sample, and found a positive correlation for women, but no significant association could be found for men (Bauer, et al., 2012). That study, however, was limited by its small sample size (74 men and 78 women), the homogenous composition of the sample, and, as in the study previously discussed, the lack of inclusion of other measures of adiposity in addition to body mass index, such as waist-to-hip ratio. From the results of the analysis performed here, it was shown that impulsivity is indeed associated with a waist-to-hip ratio that is more likely to be considered unhealthy, but this association is

relatively weaker than that of impulsivity and body mass index. As in the case of impulsivity and body mass index, though, not only was this association statistically significant, but it was also clear that it was much stronger for men than for women. This shows that, in the general population but in men especially, high levels of impulsivity are not only associated with a higher risk of obesity, but also a higher risk of abdominal obesity, and by extension, a higher risk of cardiovascular disease.

Even though Obesity Class II-III (BMI > 35) was statistically more common in women than men (6,9% compared to 6,1%), impulsivity actually had the strongest association with mild obesity (Obesity Class I) in women and with severe obesity (Obesity Class II-III) in men. These finding could possibly suggest that other factors than impulsivity might be more likely to be at the core of the problem when it comes to severe obesity in women, though the association between impulsivity and Obesity Class II-III in women was found to be non-significant, making it harder to draw conclusions about gender differences in the relationship between impulsivity and the different subcategories of obesity. Some of the key strengths of this study are, as discussed previously, its extremely large sample size, increasing the external validity of the results and decreasing the risk of type II errors, and the vast

(14)

number of recorded variables available for each of the subjects in the sample, making it possible not only to construct a proxy model for impulsivity based on these variables to use in the logistic regression analysis of the association between impulsivity and obesity, but also to control for a large amount of potential confounders in this analysis. Furthermore, the inclusion in the analysis of waist-to-hip ratio, in addition to body mass index, as a measure of obesity is something that has not been done previously, and the fact that these parameters are measured and not self-reported increases the internal validity of the results.

One of the main limitations of this study is the fact the individuals in the population sample were not randomly chosen but rather volunteers, and that all of them were aged between 40 and 69. This could be a negative factor in terms of the external validity of the results, since it could mean that the sample is not fully representative of the general population. The cross-sectional design of the study also makes it difficult to draw any conclusions about the causality behind the results presented. The lack of a widely accepted measure of impulsivity with a foundation of scientific research to support it, such as a self-report instrument or behavioral task, could be seen as a limitation and potentially as a factor decreasing the internal validity of the results. However, the proxy model used to estimate impulsivity in the analysis is given credibility not only by the large amount of previous research supporting the relationship between the each variables and impulsivity, but also how well the results correlate with those of similar studies.

Nevertheless, more research is required to confirm the results presented here by using a randomly selected population sample that with greater certainty can be said to be representative of the general population, and by using a more widely accepted measure of impulsivity, such as the Barratt Impulsiveness Scale, in the analysis of the data. Furthermore, in order to draw any meaningful conclusions about causality, studies using a longitudinal approach when investigating the associations discussed here need to be carried out.

Further studies are also needed to determine if the assessement of individual personality traits such as impulsivity has a bigger role to play in prevention and treatment strategies for obesity, and to what extent gender differences in the psychological mechanisms behind the interaction of personality and weight should be taken into account when formulating these strategies.

References

Afshin, A. et al., 2017. Health Effects of Overweight and Obesity in 195 Countries over 25 Years. New

England Journal of Medicine, 377(1), pp. 13-27.

Archer, J., 2006. Testosterone and human aggression: an evaluation of the challenge hypothesis.

(15)

Bachoo, S., Bhagwanjee, A. & Govender, K., 2013. The influence of anger, impulsivity, sensation seeking and driver attitudes on risky driving behaviour among post-graduate university students in Durban, South Africa. Accident Analysis and Prevention, Volume 55, pp. 67-76.

Balevich, E., Wein, N. & Flory, J., 2013. Cigarette Smoking and Measures of Impulsivity in a College Sample. Substance Abuse, Volume 34, pp. 256-262.

Bauer, L. et al., 2012. GABRA2 Genotype, Impulsivity, and Body Mass. The American Journal on

Addictions, Volume 21, pp. 404-410.

Beadnell, A., Wells, E., Morison, D. & Gillmore, M., 2007. Intrapersonal and Interpersonal Factors Influencing Adolescents’ Decisions About Having Sex: A Test of Sufficiency of the Theory of Planned Behavior. Journal of Applied Social Psychology, 37(12), pp. 2840-2876.

Bénard, M. et al., 2017. Association between Impulsivity and Weight Status in a General Population.

Nutrients, 9(3), p. 217.

Cameron, A. & Zimmet, P., 2008. Expanding Evidence for the Multiple Dangers of Epidemic Abdominal Obesity. Circulation, Volume 117, pp. 1624-1626.

Celma-Merola, J. et al., 2017. Self-changing behaviour in smoking cessation linked to trait and cognitive impulsivity. Addiction, Volume 13, pp. 107-112.

Chamberlain, S., Derbyshire, K., Leppink, E. & Grant, J., 2015. Obesity and dissociable forms of impulsivity in young adults. CNS Spectrums, Issue 20, pp. 500-507.

Chamorro, J. et al., 2012. Impulsivity in the general population: A national study. Journal of

Psychiatric Research, Volume 46, pp. 994-1001.

Charnigo, R. et al., 2013. Sensation Seeking and Impulsivity: Combined Associations with Risky Sexual Behavior in a Large Sample of Young Adults. J Sex Res, 50(5), pp. 480-488.

Collins, R., 2012. What makes UK Biobank special?. The Lancet, 379(9822), p. 1173–1174. Cross, C., Copping, L. & Campbell, A., 2011. Sex Differences in Impulsivity: A Meta-Analysis.

Psychological Bulletin, Volume 137, pp. 97-130.

Cyders, M., 2013. Impulsivity and the Sexes: Measurement and Structural Invariance of the UPPS-P Impulsive Behavior Scale. Assessment, Volume 1, pp. 86-97.

Deckman, T. & DeWall, N., 2011. Negative urgency and risky sexual behaviors: A clarification of the relationship between impulsivity and risky sexual behavior. Personality and Individual Differences, Volume 51, pp. 674-678.

Derefinko, K. et al., 2014. Relations Between Trait Impulsivity, Behavioral Impulsivity Physiological Arousal, and Risky Sexual Behavior among Young Men. Arch Sex Behav., 43(6), pp. 1149-1158. Després, J.-P., 2006. Abdominal obesity: the most prevalent cause of the metabolic syndrome and related cardiometabolic risk. European Heart Journal Supplements, 8(8), pp. B4-B12.

(16)

Dir, A., Coskunpinar, A. & Cyders, M., 2014. A meta-analytic review of the relationship between adolescent risky sexual behavior and impulsivity across gender, age, and race. Clinical Psychology

Review, Volume 34, pp. 551-562.

Eensoo, D., Paaver, M. & Harro, J., 2010. Factors Associated with Speeding Penalties in Novice Drivers. Ann Adv Automot Med., Volume 54, pp. 287-294.

Fearghal, O. & Gormley, M., 2013. The contribution of inhibitory deficits to dangerous driving among young people. Accident Analysis and Prevention, Volume 51, pp. 238-242.

Fortunato, L., Young, A., Boyd, C. & Fons, C., 2010. Hook-up Sexual Experiences and Problem Behaviors Among Adolescents. J Child Adolesc Subst Abuse, 19(3), pp. 261-278.

Galez, J. et al., 2014. Shared clinical associations between obesity and impulsivity in rapid cycling bipolar disorder: A systematic review. Journal of Affective Disorders, Volume 168, pp. 306-313. Granö, N. et al., 2004. Impulsivity as a predictor of smoking and alcohol consumption. Personality

and Individual Differences, Volume 37, pp. 1693-1700.

Guerrieri, R., Nederkoorn, C. & Jansen, A., 2008. The effect of an impulsive personality on overeating.

Psychological Topics, 17(2), pp. 265-286.

Haslam, D. & James, P., 2005. Obesity. The Lancet, Volume 366, pp. 1197-1209.

Hatfield, J., Williamson, A., Kehoe, J. & Prabhakharan, P., 2017. An examination of the relationship between measures of impulsivity and risky simulated driving amongst young drivers. Accident

Analysis and Prevention, Volume 103, pp. 37-43.

Hipwell, A., Keenan, K., Loeber, R. & Battista, D., 2010. Early Predictors of Sexually Intimate Behaviors in an Urban Sample of Young Girls. Dev Psychol, 46(2), pp. 366-378.

Huxley, R. et al., 2010. Body mass index, waist circumference and waist:hip ratio as predictors of cardiovascular risk—a review of the literature. European Journal of Clinical Nutrition, Volume 64, pp. 16-22.

Jakuszkowiak-Wojten, K., Landowski, J., Wiglusz, M. & Jerzy Cubala, W., 2015. Impulsivity in Anxiety Disorders: A Critical Review. Psychiatria Danubina, 27(1), pp. 452-455.

Khurana, A. et al., 2012. Early Adolescent Sexual Debut: The Mediating Role of Working Memory Ability, Sensation Seeking, and Impulsivity. Dev Psychol., 48(5), pp. 1416-1428.

Lange, K. et al., 2014. Effects of weight and gender on a task of inattention. Eating Behaviors, Volume 15, pp. 574-577.

Lim, S. et al., 2012. A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. The Lancet, Volume 380, pp. 2224-60.

Lopez-Torrecillas, F. et al., 2014. The role of impulsivity in dropout from treatment for cigarette smoking. Comprehensive Psyhiatry, Volume 55, pp. 1609-1613.

(17)

Lopez-Torrecillas, F., Perales, J., Nieto-Ruiz, A. & Verdejo-Garcia, A., 2014. Temperament and Impulsivity Predictors of Smoking Cessation Outcomes. Plos One, 9(12).

Lundahl, A., Wahlstrom, L., Christ, C. & Stoltenberg, S., 2015. Gender differences in the relationship between impulsivity and disordered eating behaviors and attitudes. Eating Behaviors, Volume 18, pp. 120-124.

Mobbs, O. et al., 2010. Obesity and the four facets of impulsivity. Patient Education and Counseling, Volume 79, pp. 372-377.

Moeller, G. et al., 2001. Psychiatric Aspects of Impulsivity. Am J Psychiatry, 158(11), pp. 1783-1793. Muchimba, M. et al., 2014. Behavioral Disinhibition and Sexual Risk Behavioramong Adolescents and Young Adults in Malawi. Plos One, 8(9).

Ng, M. et al., 2014. Global, regional, and national prevalence of overweight and obesity in children and adults during 1980–2013: a systematic analysis for the Global Burden of Disease Study 2013.

Lancet, Issue 384, p. 766–81.

Ollier, W., Sprosen, T. & Peakman, T., 2005. UK Biobank: from concept to reality. Pharmacogenomics, 6(6), pp. 639-46.

Razay, G., Vreugdenhil, A. & Wilcock, G., 2006. Obesity, Abdominal Obesity and Alzheimer Disease.

Dement Geriatr Cogn Disord, Volume 22, pp. 173-176.

Stice, E., 2002. Risk and Maintenance Factors for Eating Pathology: A Meta-Analytic Review.

Psychological Bulletin, 128(5), pp. 825-848.

Sutin, A., Ferucci, L., Zonderman, A. & Terracciano, A., 2011. Personality and Obesity across the Adult Lifespan. J Pers Soc Psychol., 101(3), pp. 579-592.

Swann, A., Bjork, J., Moeller, G. & Dougherty, D., 2002. Two Models of Impulsivity: Relationship to Personality Traits and Psychopathology. Society of Biological Psychiatry, Volume 51, pp. 988-994. Whiteside, S. & Lynam, D., 2001. The Five Factor Model and impulsivity: using a structural model of personality to understand impulsivity. Personality and Individual Differences, Volume 30, pp. 669-689.

WHO, 2000. Obesity: Preventing and Managing the Global Epidemic, Geneva: World Health Organization - Technical Report Series.

WHO, 2008. Waist Circumference and Waist-Hip Ratio, Geneva: Report of a WHO Expert Consultation.

Winstanley, C., Eagle, D. & Robbins, T., 2006. Behavioral models of impulsivity in relation to ADHD: Translation between clinical and preclinical studies. Clinical Psychology Review, Volume 26, pp. 379-395.

(18)

Yusuf, S. et al., 2004. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. The Lancet, Volume 364, pp. 937-952.

Zuckerberg, M., 1979. Sensation seeking: Beyond the optimal level of arousal. Hillsdale, N.J: L. Erlbaum Associates.

Zuckerman, M. & Kuhlman, M., 2000. Personality and Risk-Taking: Common Biosocial Factors. Journal