Clinical correlates of pathological gambling symptoms in patients with epilepsy


Address correspondence to Andrea E. Cavanna, M.D., Department of Neuropsychiatry, Institute of Neurology, Queen Square, London WC1N3BG, U.K. E-mail:


Pathological gambling symptoms (PGS), that is, the subjective urge to gamble and the actual gambling behaviors, are currently acknowledged as relatively common symptoms among Western countries, with an estimated point prevalence of 0.6–1.1% in the general population. Converging evidence suggests that PGS are overrepresented in patients with neurological conditions affecting dopaminergic reward pathways, and can be expressed in both impulse control disorders and obsessive-compulsive spectrum disorders. This study explored the clinical correlates of PGS in patients with epilepsy. Eighty-eight consecutive adult outpatients recruited at three epilepsy clinics in northern Italy were assessed using the Gambling-Symptom Assessment Scale (G-SAS), along with a battery of psychometric instruments to index depression (Beck Depression Inventory [BDI]), anxiety (Spielberger State-Trait Anxiety Inventory [STAI]), and obsessionality (Yale-Brown Obsessive Compulsive Scale [YBOCS]) symptoms. On the G-SAS, patients with a diagnosis of temporal lobe epilepsy (TLE) reported a mean [sd] G-SAS score of 2.0 [5.7], significantly higher than patients with frontal lobe epilepsy (FLE) (0.6 [1.7]) and idiopathic generalized epilepsy (IGE) (0.4 [1.4]). Moreover, multiple regression analysis showed that G-SAS scores were selectively predicted by YBOCS scores, thus suggesting an association between the expression of obsessional spectrum symptoms and PGS in patients with TLE. Alterations in the mesolimbic reward system could represent the putative neuropathological substrate for this multifaceted clinical picture.

“Of course, I am living in constant trepidation,

playing for the smallest of stakes,

and always looking out for something

– calculating, standing whole days

by the gaming tables to watch the play –

even seeing that play in my dreams …”

Fjodor Dostoevskij, The Gambler, XVII

Pathological gambling symptoms (PGS) are behavioral problems characterized by a compelling urge to gamble and inappropriate, persistent, and maladaptive gambling behaviors that have repercussions on family, personal, and professional life (Goudriaan et al., 2004). Subjects with PGS can fulfill DSM-IV diagnostic criteria for impulse control disorders (American Psychiatric Association, 2000), and PGS are also widely understood as part of the obsessive-compulsive spectrum (Hollander & Wong, 1995; Blaszczynski, 1999). Epidemiological studies conducted in Western countries showed that PGS are not rare, the estimated point prevalence being 0.6–1.1% in the general population (Shaffer et al., 1999). Moreover, recent research demonstrated that PGS can be overrepresented in several neurological conditions, including Parkinson's disease (Voon et al., 2006), restless legs syndrome (Tippmann-Peikert et al., 2007), frontotemporal dementia (Nakaaki et al., 2007), and Huntington's disease (De Marchi et al., 1998), possibly due to dysregulation of dopaminergic reward pathways following neurodegenerative or iatrogenic processes (Grosset et al., 2006).

Overall, behavioral problems are common in epilepsy, and encompass a wide spectrum of conditions ranging from those that are a direct consequence of the epileptogenic activity to others that are simply comorbid (Krishnamoorthy et al., 2007). Over the past decade several studies have consistently shown a higher prevalence of psychiatric disorders in epilepsy than in general population (Kanner et al., 2004; Jones et al., 2007; Tellez-Zenteno et al., 2007). Moreover, the link between affective/anxiety symptomatology and limbic dysfunction in temporal lobe epilepsy (TLE) seems to be compelling (Swinkels et al., 2005). Specifically, according to two recent studies (Isaacs et al., 2004; Monaco et al., 2005), obsessive-compulsive disorder (OCD) could be associated with TLE per se. However, to date no study has been conducted to investigate the clinical correlates of PGS in epilepsy, let alone the historical case report of the Russian writer Fjodor Dostoevskij, who was a compulsive gambler and probably suffered from TLE (Baumann et al., 2005). Thus, in this study, we sought to investigate the clinical correlates of PGS in patients with epilepsy, in order to gain further insights into the relationship among impulse control disorders, obsessive-compulsive spectrum disorders, and epilepsy.



A total of 88 subjects were enrolled in this cross-sectional, multicenter study, over a 18-month period (January 2006 to June 2007). The subjects were adult consecutive outpatients recruited at three secondary referral centers for the diagnosis and management of epilepsy in northern Italy, namely the Epilepsy Clinic of the Department of Neurology, Amedeo Avogadro University, Novara (n = 70), the Epilepsy Clinic of the Department of Neurology, St Andrea Hospital, Vercelli (n = 10), and the Epilepsy Clinic of the Department of Neurology, Infermi Hospital, Biella (n = 8). Epilepsy was diagnosed according to the ILAE criteria (Commission on Classification and Terminology of the International League Against Epilepsy, 1989)—clinical features, EEG or video/EEG, and neuroimaging—by at least two different neurologists not involved in the present study. We excluded patients younger than 18 years, patients with an uncertain diagnosis of epilepsy, patients with a reading level less than sixth grade, patients with learning disabilities or mental retardation and, in general, patients with a Mini Mental State Examination <24. Informed consent was provided by each subject prior to enrolment in the study, for which ethics approval was granted by the local Ethics Committee.

Quantification of symptom phenomenology

All subjects underwent a comprehensive assessment by fully trained neurologists, under the supervision of a neuropsychiatrist (MM). The Structured Clinical Interview for DSM-IV Axis I Disorders, Clinician Version (SCID-CV) was used to diagnose psychiatric disorders according to DSM-IV criteria (First et al., 1996). Moreover, depression, anxiety, and obsessional symptoms were indexed in each subject using the following standardized psychometric instruments: the Beck Depression Inventory (BDI) (Beck et al., 1988), the Spielberger State-Trait Anxiety Inventory (STAI-Y1 and Y2) (Spielberger, 1983), and the Yale-Brown Obsessive Compulsive Scale (YBOCS–Obs and Comp) (Goodman et al., 1989), respectively. The presence of PGS was investigated using the Gambling-Symptom Assessment Scale (G-SAS), a reliable and valid 12-item self-report measure of gambling urges, thoughts, and behaviors, experienced during the week preceding the scale administration (Kim et al., 2001). The psychometric scales were administered by the neurologists who performed the assessment in a standardized way and in the same sequence to all subjects.


First, patients with TLE, frontal lobe epilepsy (FLE), and idiopathic generalized epilepsy (IGE) were compared with respect to G-SAS and other psychometric scores using ANOVA. Subsequently, the association between clinical variables and G-SAS scores was explored using nonparametric tests and linear regression analysis, which allows adjustment for general confounders; the variables which correlated significantly to G-SAS scores after adjustment for general confounding entered a multiple variable regression model. All statistical analyses were two-sided with a probability level set at p = 0.05 and were conducted using the Statistical Package for Social Sciences (Version 12 for Windows, SPSS, Inc. Chicago, IL, U.S.A.).


We analyzed data from 88 patients with epilepsy. The clinical characteristics of the sample, including DSM-IV-TR-validated psychiatric diagnoses, are presented in Table 1. With respect to antiepileptic medication, the most commonly used drug was carbamazepine (44.3%), followed by valproate (25.0%), barbiturates (14.8%), lamotrigine (11.4%), oxcarbazepine (9.1%), and topiramate (6.8%). We identified three groups of patients based on epilepsy diagnosis (TLE, n = 58; FLE, n = 8; IGE, n = 22). There were no statistically significant differences between the groups on clinical seizure variables that might be related to PGS. Two patients (2.3%), both diagnosed with TLE, displayed severe gambling behaviors (bingo and national lottery) and met DSM-IV-TR criteria for pathological gambling disorder. One of them also met diagnostic criteria for OCD, whereas the other one had no additional comorbidity. G-SAS, BDI, STAI, and YBOCS scores for each group are shown in Table 2. Patients with TLE had higher scores than patients with FLE or IGE for G-SAS, BDI, and YBOCS scores (all p < 0.05). Overall, we found no relationship between G-SAS scores and disease duration, etiology, EEG laterality, seizure frequency, MRI findings, and antiepileptic medication. Table 3 shows the results of the multiple regression analysis including BDI, STAI, and YBOCS scores. According to this model, the strongest predictors of G-SAS scores were YBOCS scores (p < 0.05 for both Obs and Comp subscales). Importantly, the possibility of a duplicated symptom reporting in the G-SAS and the YBOCS was ruled out by controlling for overlap between G-SAS and YBOCS principal components.

Table 1.  Demographic and clinical features of the sample
 n = 88
  1. TLE, temporal lobe epilepsy; FLE, frontal lobe epilepsy; IGE, idiopathic generalized epilepsy.

Gender (%)
Age, years (SD)43.1 (15.2)
Disease duration, years (SD)15.6 (15.5)
Epilepsy diagnosis (%)
 FLE 9.1
Etiology (%)
 Mesial temporal sclerosis 6.8
 Other idiopathic68.2
Seizure frequency (%)
 >20/month 4.5
EEG (%)
MRI lesion (%)
 Right-sided 8.0
 Bilateral 3.4
Epileptic aura (%)36.4
Psychiatric disorders (%)
 Major depression disorder19.3
 Generalized anxiety disorder12.5
 Obsessive compulsive disorder 4.5
 Pathological gambling disorder 2.3
 Other disorders 4.5
Therapy (%)
 No therapy 4.5
 Polytherapy 3.4
Table 2.  Mean (SD) scores of the psychometric scales in the three patient groups
 TLE (n = 58)FLE (n = 8)IGE (n = 22)Fp
  1. TLE, temporal lobe epilepsy; FLE, frontal lobe epilepsy; IGE, idiopathic generalized epilepsy; G-SAS, Gambling-Symptom Assessment Scale; BDI, Beck Depression Inventory; STAI, Spielberger State (Y1) and Trait (Y2) Anxiety Inventory; YBOCS, Yale Brown Obsessive (Obs) Compulsive (Comp) Scale. Values of p < 0.05 are shown in bold.

G-SAS2.0 (5.7)0.6 (1.7)0.4 (1.4)3.240.04
BDI10.6 (10.1)4.5 (5.5)5.8 (5.6)3.320.04
STAI-Y133.7 (18.9)35.7 (22.0)30.0 (17.4)2.590.08
STAI-Y235.0 (18.8)34.7 (13.3)33.1 (17.2)3.060.06
YBOCS-Obs6.4 (5.2)2.3 (1.9)1.8 (1.6)3.290.04
YBOCS-Comp5.8 (4.1)2.8 (2.8)0.9 (0.6)3.440.03
Table 3.  Best fit multiple variable regression model predicting G-SAS scores
 t (β)P95% CI
  1. G-SAS, Gambling-Symptom Assessment Scale; BDI, Beck Depression Inventory; STAI, Spielberger State (Y1) and Trait (Y2) Anxiety Inventory; YBOCS, Yale Brown Obsessive (Obs) Compulsive (Comp) Scale. Values of p < 0.05 are shown in bold.

BDI0.990.34−0.03 to 0.10
STAI-Y11.860.07−0.02 to 0.19
STAI-Y21.550.13−0.03 to 0.35
YBOCS-Obs2.130.03   0.04 to 0.82
YBOCS-Comp2.190.02   0.05 to 0.91


To the best of our knowledge, the present study is the first one to investigate the clinical correlates of PGS in patients with epilepsy using a validated psychometric instrument such as the G-SAS. First, our results suggest that PGS are not rare in adult patients with epilepsy, particularly when compared with normative samples of healthy subjects (Shaffer et al., 1999). Specifically, TLE as a group seems to be associated with higher G-SAS scores compared to other patients group (FLE, IGE). The association between depression, anxiety, and obsessionality symptoms and TLE is in line with previous literature (reviewed in Swinkels et al., 2005), whereas the finding of a higher expression of PGS among patients with TLE has not been documented previously. It has to be pointed out that we found a relatively low rate of DSM-IV-TR-defined pathological gambling disorder in the context of subjective reports of significantly increased gambling urges and behaviors. However, it has been shown that increased self report gambling scores identify the higher end of the population officially considered diagnostically subthreshold for pathological gambling disorder, and may be important from both a clinical and public health perspective (Cox et al., 2004). Furthermore, in our population G-SAS scores were selectively predicted by YBOCS scores, thus supporting existing evidence toward a clinical and neurobiological overlap between impulse disorders and obsessive compulsive spectrum symptoms (McElroy et al., 1993; Grant & Potenza, 2006). Taken together with the recent clinical investigations of OCD in TLE (Isaacs et al., 2004; Monaco et al., 2005), these findings suggest that temporolimbic dysfunction affecting reward pathways (via the nucleus accumbens) could represent a shared neuropathological substrate for the expression of a wider range of obsessive-compulsive spectrum symptoms in patients with epilepsy (Reuter et al., 2005). This view is supported by the observation that dopaminergic pathways originating in Papez circuit deeply affect the cortico-basal-thalamo-cortical system involved in OCD, thus leading to possible additional dysfunction in serotonergic neurotransmission (Heimer et al., 2007).

Limitations of our study need to be acknowledged. First, the small sample size reduces the statistical power of our study. Specifically, the size of the FLE group could have led to results of uncertain significance. Second, our findings may be not representative for epilepsy patients in general because our population represents a selected sample coming mainly from a university-based referral center. It is possible that individuals with more severe epilepsy and behavioral problems are more likely to be referred to our clinics and correspondingly patients in our clinics are subject to referral bias. Furthermore, an intrinsic limitation of this study is the incorporation of psychometric scales covering obsessive-compulsive spectrum symptoms, such as the YBOCS, as predictors of G-SAS scores. Finally, psychosocial problems that arise from living with, and adapting to, a still stigmatized disorder such as epilepsy, need to be taken into account because they may contribute to the severity of associated psychopathology.

In conclusion, this study suggests that PGS may occur in adult patients with TLE and obsessive-compulsive spectrum symptoms, further complicating the clinical picture. At a theoretical level, it is plausible that TLE and PGS may share a common pathoaetiology (temporolimbic dysfunction affecting dopaminergic reward pathways), and our results are in line with previous publications showing a complex relationship between PGS and obsessive-compulsive spectrum disorders. Further research—including controlled studies on larger samples—is needed to replicate our findings and to shed light on the interlinked neurobiology of PGS, obsessive-compulsive spectrum disorders, and TLE.


We wish to thank Dr. C. Civardi, Dr. E. Terazzi, and Dr. C. Varrasi for their contribution.

Conflict of interest: We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. We have no financial interests or conflicts of interest to declare.