Personality Disorders

Families and schools often pressure clinicians to “do something” when children or adolescents persistently bully, threaten, or injure others. This demand poses a treatment dilemma when psychosocial and educational interventions have failed to manage pediatric aggression.

Aggression is the main reason for drug therapy in youths with conduct disorder, but very little safety and efficacy data exist to help us choose medications. This places young patients at risk for polypharmacy, unmanaged symptoms, short-term side effects, and unknown long-term consequences of exposure to psychotropics.

Table 1

4 precautions when prescribing for pediatric aggression

This article reviews the limited data on using medications to reduce aggression in children and adolescents, focusing on double-blind, placebo-controlled trials in conduct disorder. Based on this evidence and our clinical experience, we offer a sample case and treatment recommendations.

Prescribing principles

Precautions. When prescribing drugs to treat aggressive youth, remember the American Academy of Child and Adolescent Psychiatry’s precautions (Table 1)¹ Recently published recommendations prepared by expert consensus are also valuable treatment guides.²

Linking treatment to diagnosis. Should we attempt to manage aggression as a manifestation of an underlying psychiatric disorder? Or should we treat it the same across all disorders? The latter approach is akin to the “fever model.”

Fever—regardless of cause—may be treated with a nonsteroidal anti-inflammatory drug. However, evidence from drug studies suggests that underlying psychiatric disorders should help determine the choice of aggression treatment. For example, a recent study in adults found that divalproex was effective for aggressive patients only within a specific diagnostic subgroup (in this case, cluster B personality disorders).³

Clinical experience also links aggression treatment with underlying diagnoses. For example, aggression secondary to agitated depression is treated with an antidepressant, whereas aggression secondary to command hallucinations in schizophrenia is treated with antipsychotics.

In treating aggression in conduct disorder (Table 2), first treat comorbid disorders—such as attention deficit/hyperactivity disorder (ADHD) or bipolar disorder—and address the child’s psychosocial and educational needs. Then if medication is appropriate, consider drugs with evidence of safety and efficacy, such as antipsychotics, lithium, and stimulants.

Antipsychotics

Three conventional antipsychotics—chlorpromazine, haloperidol, and thioridazine—are FDA-approved for controlling disruptive behaviors in children.⁴ No atypical antipsychotics are so indicated, but atypicals are preferred in children and adolescents because of lower risks for tardive dyskinesia, neuroleptic malignant syndrome, and extrapyramidal symptoms.²

Risperidone is the most-studied atypical antipsychotic for treating pediatric aggression, particularly in patients with low intellectual functioning or mental retardation. In a 6-week, double-blind, placebo-controlled trial, 118 children ages 5 to 12 with severely disruptive behavior and IQs of 36 to 84 were given low-dose risperidone (mean 1.16 mg/d). Risperidone reduced conduct problems significantly more than placebo, although aggression was not measured directly.⁵ Adverse events included somnolence, headache, vomiting, weight gain, and elevated serum prolactin. Similar results have been reported in other studies.⁶

Table 2

Diagnostic criteria for conduct disorder

Box 1

Risperidone also reduced aggression in children with normal intelligence in one small study.⁷ As a cautionary note, however, long-term risperidone treatment has been associated with withdrawal dyskinesias.⁸

Olanzapine, quetiapine, ziprasidone, and aripiprazole are less well-studied for treating pediatric aggression but are preferable to conventional agents when antipsychotics are considered.

Recommendation. Expert consensus opinion² recommends using atypicals when psychosocial treatments and first-line medications for primary conditions have failed. Start with low dosages, and titrate up slowly while monitoring symptoms and side effects. Because no studies have compared any atypical’s efficacy over others for aggressive behavior, base your choices on:

• discussions with the patient and family (Box 1)
• medical comorbidities
• how the patient responded to antipsychotics in the past
• side-effect profile
• long-term treatment planning.²

If the patient cannot tolerate the medication or does not respond after 4 to 6 weeks, try switching atypicals. To improve partial response, consider adding a mood stabilizer such as lithium or divalproex. If aggressive symptoms remit for 6 months or longer, attempt to taper or discontinue the antipsychotic.²

Lithium

In placebo-controlled trials, lithium reduced aggression in:

• male prisoners ages 16 to 24.⁹
• children ages 7 and 12 with conduct disorder¹⁰
• children and adolescents ages 10 to 17 with conduct disorder.¹¹

Among these studies, only ours¹¹ specifically measured aggression. We randomly assigned 40 children to receive 4 weeks of lithium, 900 to 2,100 mg/d (mean 1,425 ± 321 mg/d), or placebo. Serum lithium levels were 0.78 to 1.55 mEq/L (mean 1.07 ± 0.19 mEq/L). We used the Overt Aggression Scale (OAS)^12,13 (see Related resources) to track frequency and severity of verbal aggression, aggression against objects, aggression against others, and self-aggression.

Lithium reduced aggression more than did placebo, as measured by the clinician-rated Clinical Global Impressions (CGI) scale and staff-rated Global Clinical Judgments (Consensus) Scale (GCJCS). The CGI showed a 70% response rate with lithium and 20% with placebo. Similarly, the GCJCS scale showed 80% response with lithium and 30% with placebo.

The aggression reduction with lithium was statistically significant and clinically evident. Most subjects (37 of 40) experienced at least one adverse event, however, whether receiving lithium or placebo. Nausea, vomiting, and urinary frequency were significantly more common in the lithium-treated group than with placebo. Fewer adverse events were reported in a similar outpatient study,¹⁴ probably because of less-frequent monitoring.

Lithium did not reduce aggression in adolescent girls treated for 2 weeks¹⁵ or in an outpatient study of children with ADHD.¹⁶

Recommendation. Lithium has shown efficacy for reducing severe aggression in hospitalized children with conduct disorder but not in similar outpatients. Consider this drug to reduce severe aggression in children with conduct disorder, especially if they have failed other treatments.

Anticonvulsants

Anticonvulsants have been used to decrease aggression for more than 50 years, and epidemiologic data show their use is increasing markedly.¹⁷ Few controlled studies support this prescribing trend, however.¹⁸

Initial reports suggested that anticonvulsants reduce disruptive behaviors, but more-critically designed studies have not supported this finding. For example, phenytoin sodium (diphenylhydantoin) demonstrated efficacy in open trials, but controlled trials found this anticonvulsant no more effective than placebo. In fact, placebo may have reduced aggression more than the active drug. Likewise, earlier controlled trials of carbamazepine indicated efficacy, but more-carefully designed trials using specific measures of aggression did not.

Divalproex is the anticonvulsant most commonly used for aggression in children and adolescents. Only one small, placebo-controlled study has found it effective in reducing aggression in children.¹⁹

Twenty children ages 10 to 18 with conduct disorder or oppositional defiant disorder were randomized to divalproex, 750 to 1,500 mg/d, or placebo. Eighteen completed the first phase, and 15 crossed over to the other treatment. Concomitant drug treatment, including stimulants, was allowed. The authors reported that 12 of 15 subjects showed some response to divalproex.

A 7-week study compared divalproex in high dosages (up to 1,500 mg/d) versus low dosages (up to 250 mg/d). This study was not placebo-controlled, but aggression was reduced more in the high-dosage than in the low-dosage group.²⁰

Recommendation. If you use an anticonvulsant, first obtain informed consent from the patient and parent. Divalproex causes weight gain and has been associated with increased risk of polycystic ovary syndrome with masculinizing effects.²¹

Double-blind, placebo-controlled studies of divalproex and other anticonvulsants in treating aggression are needed, particularly as prescriptions for these agents are rising.

Stimulants

Some small controlled studies suggest that stimulants can reduce aggression in children with ADHD, but their effects on aggression in conduct disorder have not been well studied. Aggression was not measured directly in the National Institute of Mental Health Multimodal Treatment Study of Children with ADHD.²¹ Most other studies have been small and included children with ADHD but not necessarily conduct disorder.

Recommendation. Stimulants may help reduce aggression in children with ADHD, but studies gauging their effects in conduct disorder are needed.

Alpha agonists

Alpha agonists such as clonidine and guanfacine are increasingly being used to treat children with disruptive disorders, despite limited evidence. The small controlled studies that examined alpha agonists as monotherapy or add-ons in this population did not directly measure aggression.^22,23

Recommendation. Little data support alpha agonists for reducing aggression. They should probably be considered second-line treatment.

SSRIs

No double-blind, placebo-controlled studies have tested any selective serotonin reuptake inhibitor (SSRIs) for reducing aggression in conduct disorder. In a 6-week open study, citalopram (mean 27 mg/d) reduced impulsive aggression in 12 children with mixed diagnoses, as measured by the modified OAS,¹³ Child Behavior Checklist, and CGI.²⁴

Recommendation. Use caution when prescribing SSRIs to aggressive youth, as these drugs may contribute to aggression in some mood-disordered children. More evidence of SSRIs’ safety and efficacy in this population is needed.

Related resources

• Overt Aggression Scale. In: Coccaro EF, Harvey PD, Kupsaw-Lawrence E, et al. Development of neuropharmacologically-based behavioral assessments of impulsiveaggressive behavior. J Neuropsychiatry 1991;3(2):S44-S51.

Drug brand names

• Aripiprazole • Abilify
• Carbamazepine • Tegretol
• Citalopram • Celexa
• Chlorpromazine • Thorazine
• Clonidine • Catapres
• Phenytoin sodium • Dilantin
• Divalproex • Depakote
• Guanfacine • Tenex
• Haloperidol • Haldol
• Olanzapine • Zyprexa
• Quetiapine • Seroquel
• Risperidone • Risperdal
• Thioridazine • Mellaril
• Ziprasidone • Geodon

Disclosure

Dr. Malone receives research support from Pfizer Inc. and Eli Lilly and Co. and is a consultant to Janssen Pharmaceutica.

Dr. Delaney is a consultant to Shire Pharmaceuticals.

Dr. Sheikh reports no financial relationship with any company whose products are mentioned in this article or with manufacturers of competing products.

Families and schools often pressure clinicians to “do something” when children or adolescents persistently bully, threaten, or injure others. This demand poses a treatment dilemma when psychosocial and educational interventions have failed to manage pediatric aggression.

Aggression is the main reason for drug therapy in youths with conduct disorder, but very little safety and efficacy data exist to help us choose medications. This places young patients at risk for polypharmacy, unmanaged symptoms, short-term side effects, and unknown long-term consequences of exposure to psychotropics.

Table 1

4 precautions when prescribing for pediatric aggression

This article reviews the limited data on using medications to reduce aggression in children and adolescents, focusing on double-blind, placebo-controlled trials in conduct disorder. Based on this evidence and our clinical experience, we offer a sample case and treatment recommendations.

Prescribing principles

Precautions. When prescribing drugs to treat aggressive youth, remember the American Academy of Child and Adolescent Psychiatry’s precautions (Table 1)¹ Recently published recommendations prepared by expert consensus are also valuable treatment guides.²

Linking treatment to diagnosis. Should we attempt to manage aggression as a manifestation of an underlying psychiatric disorder? Or should we treat it the same across all disorders? The latter approach is akin to the “fever model.”

Fever—regardless of cause—may be treated with a nonsteroidal anti-inflammatory drug. However, evidence from drug studies suggests that underlying psychiatric disorders should help determine the choice of aggression treatment. For example, a recent study in adults found that divalproex was effective for aggressive patients only within a specific diagnostic subgroup (in this case, cluster B personality disorders).³

Clinical experience also links aggression treatment with underlying diagnoses. For example, aggression secondary to agitated depression is treated with an antidepressant, whereas aggression secondary to command hallucinations in schizophrenia is treated with antipsychotics.

In treating aggression in conduct disorder (Table 2), first treat comorbid disorders—such as attention deficit/hyperactivity disorder (ADHD) or bipolar disorder—and address the child’s psychosocial and educational needs. Then if medication is appropriate, consider drugs with evidence of safety and efficacy, such as antipsychotics, lithium, and stimulants.

Antipsychotics

Three conventional antipsychotics—chlorpromazine, haloperidol, and thioridazine—are FDA-approved for controlling disruptive behaviors in children.⁴ No atypical antipsychotics are so indicated, but atypicals are preferred in children and adolescents because of lower risks for tardive dyskinesia, neuroleptic malignant syndrome, and extrapyramidal symptoms.²

Risperidone is the most-studied atypical antipsychotic for treating pediatric aggression, particularly in patients with low intellectual functioning or mental retardation. In a 6-week, double-blind, placebo-controlled trial, 118 children ages 5 to 12 with severely disruptive behavior and IQs of 36 to 84 were given low-dose risperidone (mean 1.16 mg/d). Risperidone reduced conduct problems significantly more than placebo, although aggression was not measured directly.⁵ Adverse events included somnolence, headache, vomiting, weight gain, and elevated serum prolactin. Similar results have been reported in other studies.⁶

Table 2

Diagnostic criteria for conduct disorder

Box 1

Risperidone also reduced aggression in children with normal intelligence in one small study.⁷ As a cautionary note, however, long-term risperidone treatment has been associated with withdrawal dyskinesias.⁸

Olanzapine, quetiapine, ziprasidone, and aripiprazole are less well-studied for treating pediatric aggression but are preferable to conventional agents when antipsychotics are considered.

Recommendation. Expert consensus opinion² recommends using atypicals when psychosocial treatments and first-line medications for primary conditions have failed. Start with low dosages, and titrate up slowly while monitoring symptoms and side effects. Because no studies have compared any atypical’s efficacy over others for aggressive behavior, base your choices on:

• discussions with the patient and family (Box 1)
• medical comorbidities
• how the patient responded to antipsychotics in the past
• side-effect profile
• long-term treatment planning.²

If the patient cannot tolerate the medication or does not respond after 4 to 6 weeks, try switching atypicals. To improve partial response, consider adding a mood stabilizer such as lithium or divalproex. If aggressive symptoms remit for 6 months or longer, attempt to taper or discontinue the antipsychotic.²

Lithium

In placebo-controlled trials, lithium reduced aggression in:

• male prisoners ages 16 to 24.⁹
• children ages 7 and 12 with conduct disorder¹⁰
• children and adolescents ages 10 to 17 with conduct disorder.¹¹

Among these studies, only ours¹¹ specifically measured aggression. We randomly assigned 40 children to receive 4 weeks of lithium, 900 to 2,100 mg/d (mean 1,425 ± 321 mg/d), or placebo. Serum lithium levels were 0.78 to 1.55 mEq/L (mean 1.07 ± 0.19 mEq/L). We used the Overt Aggression Scale (OAS)^12,13 (see Related resources) to track frequency and severity of verbal aggression, aggression against objects, aggression against others, and self-aggression.

Lithium reduced aggression more than did placebo, as measured by the clinician-rated Clinical Global Impressions (CGI) scale and staff-rated Global Clinical Judgments (Consensus) Scale (GCJCS). The CGI showed a 70% response rate with lithium and 20% with placebo. Similarly, the GCJCS scale showed 80% response with lithium and 30% with placebo.

The aggression reduction with lithium was statistically significant and clinically evident. Most subjects (37 of 40) experienced at least one adverse event, however, whether receiving lithium or placebo. Nausea, vomiting, and urinary frequency were significantly more common in the lithium-treated group than with placebo. Fewer adverse events were reported in a similar outpatient study,¹⁴ probably because of less-frequent monitoring.

Lithium did not reduce aggression in adolescent girls treated for 2 weeks¹⁵ or in an outpatient study of children with ADHD.¹⁶

Recommendation. Lithium has shown efficacy for reducing severe aggression in hospitalized children with conduct disorder but not in similar outpatients. Consider this drug to reduce severe aggression in children with conduct disorder, especially if they have failed other treatments.

Anticonvulsants

Anticonvulsants have been used to decrease aggression for more than 50 years, and epidemiologic data show their use is increasing markedly.¹⁷ Few controlled studies support this prescribing trend, however.¹⁸

Initial reports suggested that anticonvulsants reduce disruptive behaviors, but more-critically designed studies have not supported this finding. For example, phenytoin sodium (diphenylhydantoin) demonstrated efficacy in open trials, but controlled trials found this anticonvulsant no more effective than placebo. In fact, placebo may have reduced aggression more than the active drug. Likewise, earlier controlled trials of carbamazepine indicated efficacy, but more-carefully designed trials using specific measures of aggression did not.

Divalproex is the anticonvulsant most commonly used for aggression in children and adolescents. Only one small, placebo-controlled study has found it effective in reducing aggression in children.¹⁹

Twenty children ages 10 to 18 with conduct disorder or oppositional defiant disorder were randomized to divalproex, 750 to 1,500 mg/d, or placebo. Eighteen completed the first phase, and 15 crossed over to the other treatment. Concomitant drug treatment, including stimulants, was allowed. The authors reported that 12 of 15 subjects showed some response to divalproex.

A 7-week study compared divalproex in high dosages (up to 1,500 mg/d) versus low dosages (up to 250 mg/d). This study was not placebo-controlled, but aggression was reduced more in the high-dosage than in the low-dosage group.²⁰

Recommendation. If you use an anticonvulsant, first obtain informed consent from the patient and parent. Divalproex causes weight gain and has been associated with increased risk of polycystic ovary syndrome with masculinizing effects.²¹

Double-blind, placebo-controlled studies of divalproex and other anticonvulsants in treating aggression are needed, particularly as prescriptions for these agents are rising.

Stimulants

Some small controlled studies suggest that stimulants can reduce aggression in children with ADHD, but their effects on aggression in conduct disorder have not been well studied. Aggression was not measured directly in the National Institute of Mental Health Multimodal Treatment Study of Children with ADHD.²¹ Most other studies have been small and included children with ADHD but not necessarily conduct disorder.

Recommendation. Stimulants may help reduce aggression in children with ADHD, but studies gauging their effects in conduct disorder are needed.

Alpha agonists

Alpha agonists such as clonidine and guanfacine are increasingly being used to treat children with disruptive disorders, despite limited evidence. The small controlled studies that examined alpha agonists as monotherapy or add-ons in this population did not directly measure aggression.^22,23

Recommendation. Little data support alpha agonists for reducing aggression. They should probably be considered second-line treatment.

SSRIs

No double-blind, placebo-controlled studies have tested any selective serotonin reuptake inhibitor (SSRIs) for reducing aggression in conduct disorder. In a 6-week open study, citalopram (mean 27 mg/d) reduced impulsive aggression in 12 children with mixed diagnoses, as measured by the modified OAS,¹³ Child Behavior Checklist, and CGI.²⁴

Recommendation. Use caution when prescribing SSRIs to aggressive youth, as these drugs may contribute to aggression in some mood-disordered children. More evidence of SSRIs’ safety and efficacy in this population is needed.

Related resources

• Overt Aggression Scale. In: Coccaro EF, Harvey PD, Kupsaw-Lawrence E, et al. Development of neuropharmacologically-based behavioral assessments of impulsiveaggressive behavior. J Neuropsychiatry 1991;3(2):S44-S51.

Drug brand names

• Aripiprazole • Abilify
• Carbamazepine • Tegretol
• Citalopram • Celexa
• Chlorpromazine • Thorazine
• Clonidine • Catapres
• Phenytoin sodium • Dilantin
• Divalproex • Depakote
• Guanfacine • Tenex
• Haloperidol • Haldol
• Olanzapine • Zyprexa
• Quetiapine • Seroquel
• Risperidone • Risperdal
• Thioridazine • Mellaril
• Ziprasidone • Geodon

Disclosure

Dr. Malone receives research support from Pfizer Inc. and Eli Lilly and Co. and is a consultant to Janssen Pharmaceutica.

Dr. Delaney is a consultant to Shire Pharmaceuticals.

Dr. Sheikh reports no financial relationship with any company whose products are mentioned in this article or with manufacturers of competing products.

Families and schools often pressure clinicians to “do something” when children or adolescents persistently bully, threaten, or injure others. This demand poses a treatment dilemma when psychosocial and educational interventions have failed to manage pediatric aggression.

Aggression is the main reason for drug therapy in youths with conduct disorder, but very little safety and efficacy data exist to help us choose medications. This places young patients at risk for polypharmacy, unmanaged symptoms, short-term side effects, and unknown long-term consequences of exposure to psychotropics.

Table 1

4 precautions when prescribing for pediatric aggression

This article reviews the limited data on using medications to reduce aggression in children and adolescents, focusing on double-blind, placebo-controlled trials in conduct disorder. Based on this evidence and our clinical experience, we offer a sample case and treatment recommendations.

Prescribing principles

Precautions. When prescribing drugs to treat aggressive youth, remember the American Academy of Child and Adolescent Psychiatry’s precautions (Table 1)¹ Recently published recommendations prepared by expert consensus are also valuable treatment guides.²

Linking treatment to diagnosis. Should we attempt to manage aggression as a manifestation of an underlying psychiatric disorder? Or should we treat it the same across all disorders? The latter approach is akin to the “fever model.”

Fever—regardless of cause—may be treated with a nonsteroidal anti-inflammatory drug. However, evidence from drug studies suggests that underlying psychiatric disorders should help determine the choice of aggression treatment. For example, a recent study in adults found that divalproex was effective for aggressive patients only within a specific diagnostic subgroup (in this case, cluster B personality disorders).³

Clinical experience also links aggression treatment with underlying diagnoses. For example, aggression secondary to agitated depression is treated with an antidepressant, whereas aggression secondary to command hallucinations in schizophrenia is treated with antipsychotics.

In treating aggression in conduct disorder (Table 2), first treat comorbid disorders—such as attention deficit/hyperactivity disorder (ADHD) or bipolar disorder—and address the child’s psychosocial and educational needs. Then if medication is appropriate, consider drugs with evidence of safety and efficacy, such as antipsychotics, lithium, and stimulants.

Antipsychotics

Three conventional antipsychotics—chlorpromazine, haloperidol, and thioridazine—are FDA-approved for controlling disruptive behaviors in children.⁴ No atypical antipsychotics are so indicated, but atypicals are preferred in children and adolescents because of lower risks for tardive dyskinesia, neuroleptic malignant syndrome, and extrapyramidal symptoms.²

Risperidone is the most-studied atypical antipsychotic for treating pediatric aggression, particularly in patients with low intellectual functioning or mental retardation. In a 6-week, double-blind, placebo-controlled trial, 118 children ages 5 to 12 with severely disruptive behavior and IQs of 36 to 84 were given low-dose risperidone (mean 1.16 mg/d). Risperidone reduced conduct problems significantly more than placebo, although aggression was not measured directly.⁵ Adverse events included somnolence, headache, vomiting, weight gain, and elevated serum prolactin. Similar results have been reported in other studies.⁶

Table 2

Diagnostic criteria for conduct disorder

Box 1

Risperidone also reduced aggression in children with normal intelligence in one small study.⁷ As a cautionary note, however, long-term risperidone treatment has been associated with withdrawal dyskinesias.⁸

Olanzapine, quetiapine, ziprasidone, and aripiprazole are less well-studied for treating pediatric aggression but are preferable to conventional agents when antipsychotics are considered.

Recommendation. Expert consensus opinion² recommends using atypicals when psychosocial treatments and first-line medications for primary conditions have failed. Start with low dosages, and titrate up slowly while monitoring symptoms and side effects. Because no studies have compared any atypical’s efficacy over others for aggressive behavior, base your choices on:

• discussions with the patient and family (Box 1)
• medical comorbidities
• how the patient responded to antipsychotics in the past
• side-effect profile
• long-term treatment planning.²

If the patient cannot tolerate the medication or does not respond after 4 to 6 weeks, try switching atypicals. To improve partial response, consider adding a mood stabilizer such as lithium or divalproex. If aggressive symptoms remit for 6 months or longer, attempt to taper or discontinue the antipsychotic.²

Lithium

In placebo-controlled trials, lithium reduced aggression in:

• male prisoners ages 16 to 24.⁹
• children ages 7 and 12 with conduct disorder¹⁰
• children and adolescents ages 10 to 17 with conduct disorder.¹¹

Among these studies, only ours¹¹ specifically measured aggression. We randomly assigned 40 children to receive 4 weeks of lithium, 900 to 2,100 mg/d (mean 1,425 ± 321 mg/d), or placebo. Serum lithium levels were 0.78 to 1.55 mEq/L (mean 1.07 ± 0.19 mEq/L). We used the Overt Aggression Scale (OAS)^12,13 (see Related resources) to track frequency and severity of verbal aggression, aggression against objects, aggression against others, and self-aggression.

Lithium reduced aggression more than did placebo, as measured by the clinician-rated Clinical Global Impressions (CGI) scale and staff-rated Global Clinical Judgments (Consensus) Scale (GCJCS). The CGI showed a 70% response rate with lithium and 20% with placebo. Similarly, the GCJCS scale showed 80% response with lithium and 30% with placebo.

The aggression reduction with lithium was statistically significant and clinically evident. Most subjects (37 of 40) experienced at least one adverse event, however, whether receiving lithium or placebo. Nausea, vomiting, and urinary frequency were significantly more common in the lithium-treated group than with placebo. Fewer adverse events were reported in a similar outpatient study,¹⁴ probably because of less-frequent monitoring.

Lithium did not reduce aggression in adolescent girls treated for 2 weeks¹⁵ or in an outpatient study of children with ADHD.¹⁶

Recommendation. Lithium has shown efficacy for reducing severe aggression in hospitalized children with conduct disorder but not in similar outpatients. Consider this drug to reduce severe aggression in children with conduct disorder, especially if they have failed other treatments.

Anticonvulsants

Anticonvulsants have been used to decrease aggression for more than 50 years, and epidemiologic data show their use is increasing markedly.¹⁷ Few controlled studies support this prescribing trend, however.¹⁸

Initial reports suggested that anticonvulsants reduce disruptive behaviors, but more-critically designed studies have not supported this finding. For example, phenytoin sodium (diphenylhydantoin) demonstrated efficacy in open trials, but controlled trials found this anticonvulsant no more effective than placebo. In fact, placebo may have reduced aggression more than the active drug. Likewise, earlier controlled trials of carbamazepine indicated efficacy, but more-carefully designed trials using specific measures of aggression did not.

Divalproex is the anticonvulsant most commonly used for aggression in children and adolescents. Only one small, placebo-controlled study has found it effective in reducing aggression in children.¹⁹

Twenty children ages 10 to 18 with conduct disorder or oppositional defiant disorder were randomized to divalproex, 750 to 1,500 mg/d, or placebo. Eighteen completed the first phase, and 15 crossed over to the other treatment. Concomitant drug treatment, including stimulants, was allowed. The authors reported that 12 of 15 subjects showed some response to divalproex.

A 7-week study compared divalproex in high dosages (up to 1,500 mg/d) versus low dosages (up to 250 mg/d). This study was not placebo-controlled, but aggression was reduced more in the high-dosage than in the low-dosage group.²⁰

Recommendation. If you use an anticonvulsant, first obtain informed consent from the patient and parent. Divalproex causes weight gain and has been associated with increased risk of polycystic ovary syndrome with masculinizing effects.²¹

Double-blind, placebo-controlled studies of divalproex and other anticonvulsants in treating aggression are needed, particularly as prescriptions for these agents are rising.

Stimulants

Some small controlled studies suggest that stimulants can reduce aggression in children with ADHD, but their effects on aggression in conduct disorder have not been well studied. Aggression was not measured directly in the National Institute of Mental Health Multimodal Treatment Study of Children with ADHD.²¹ Most other studies have been small and included children with ADHD but not necessarily conduct disorder.

Recommendation. Stimulants may help reduce aggression in children with ADHD, but studies gauging their effects in conduct disorder are needed.

Alpha agonists

Alpha agonists such as clonidine and guanfacine are increasingly being used to treat children with disruptive disorders, despite limited evidence. The small controlled studies that examined alpha agonists as monotherapy or add-ons in this population did not directly measure aggression.^22,23

Recommendation. Little data support alpha agonists for reducing aggression. They should probably be considered second-line treatment.

SSRIs

No double-blind, placebo-controlled studies have tested any selective serotonin reuptake inhibitor (SSRIs) for reducing aggression in conduct disorder. In a 6-week open study, citalopram (mean 27 mg/d) reduced impulsive aggression in 12 children with mixed diagnoses, as measured by the modified OAS,¹³ Child Behavior Checklist, and CGI.²⁴

Recommendation. Use caution when prescribing SSRIs to aggressive youth, as these drugs may contribute to aggression in some mood-disordered children. More evidence of SSRIs’ safety and efficacy in this population is needed.

Related resources

• Overt Aggression Scale. In: Coccaro EF, Harvey PD, Kupsaw-Lawrence E, et al. Development of neuropharmacologically-based behavioral assessments of impulsiveaggressive behavior. J Neuropsychiatry 1991;3(2):S44-S51.

Drug brand names

• Aripiprazole • Abilify
• Carbamazepine • Tegretol
• Citalopram • Celexa
• Chlorpromazine • Thorazine
• Clonidine • Catapres
• Phenytoin sodium • Dilantin
• Divalproex • Depakote
• Guanfacine • Tenex
• Haloperidol • Haldol
• Olanzapine • Zyprexa
• Quetiapine • Seroquel
• Risperidone • Risperdal
• Thioridazine • Mellaril
• Ziprasidone • Geodon

Disclosure

Dr. Malone receives research support from Pfizer Inc. and Eli Lilly and Co. and is a consultant to Janssen Pharmaceutica.

Dr. Delaney is a consultant to Shire Pharmaceuticals.

Dr. Sheikh reports no financial relationship with any company whose products are mentioned in this article or with manufacturers of competing products.

Not everyone agrees that borderline personality disorder (BPD) should be a diagnostic category. BPD became “official” with DSM-III in 1980, although the term had been used for 40 years to describe various patient groups. Being listed in DSM-III legitimized BPD, which was thought to represent a specific—though not necessarily distinct—diagnostic category.

The history of the BPD diagnosis and opinions as to its usefulness can be viewed as a microcosm of psychiatric diagnosis in general. Before DSM-III, diagnoses were broadly defined and did not contain specific inclusion or exclusion criteria.¹ For the 5 to 10 years prior to DSM-III, however, two assumptions developed:

• distinct diagnostic categories did, in fact, exist
• by rigorously defining and studying those categories we could develop more specific and effective treatments for our patients.²

The specificity and exclusivity that we assumed we could achieve by categorical diagnoses, however, remain a distant wish. Comorbidity appears more common in psychiatry than was originally thought and confounds both treatment and outcome.³ Also, many patients appear treatment-resistant, despite fitting neatly into diagnostic categories.⁴

As psychiatry considers DSM-V, questions linger as to whether BPD (and personality disorder in general) should remain as a categorical diagnosis or if dimensional measures may be more appropriate. Dimensions imply that no one ever fits into a given box because no specific box exists. Rather, patients are described as being closer to or more distant from a prototypic model of the diagnosis. In personality disorders, the dimensions most often mentioned are cognition, impulsivity, emotional lability, environmental hyperreactivity, and anxiety. The case report (above) illustrates the interplay of these dimensions in a typical patient with presumed BPD.

What’s in a name?

The symptom complex or syndrome that bears the name borderline personality disorder has probably existed for as long as people have thought about patients in psychopathologic terms.⁵ Before 1980, the term “borderline” applied primarily to two separate but overlapping concepts:

• Patients thought to reside on the “border” with psychosis, such as the patient in our case example. They seemed to have an underlying psychotic disorder, but the psychosis—if it surfaced—appeared briefly, was not exceptionally deep or firmly held, and was not regularly evident or immediately accessible to the clinician.
• Patients who appeared to occupy the space between neurosis and psychosis. This concept evolved into the idea of a character or personality disorder distinguished primarily by unstable interpersonal relationships, a confused or inconsistent sense of identity, and emotional instability.

How DSM is changing. Comparing the disorders listed in DSM-IV (1994)⁶ versus DSM-II (1968)¹ suggests that psychiatry has become enamored of the naming process. For example, DSM-II lists anxiety neurosis (300.0), phobic neurosis (300.2), and obsessive-compulsive neurosis (300.3), whereas DSM-IV lists 11 different categories of anxiety disorders.

But beyond naming, subsequent DSMs have differed even more dramatically from DSM-II. We have seen a shift from describing a diagnostic category with a simple explanatory paragraph to lists of specific inclusion and exclusion criteria. These more-specific lists imply that they define categories closer to some reality or authenticity than did previous definitions.

Before DSM-III, the borderline concept was conceived in broad object relational and psychodynamic terms. In contrast, DSM-III produced a definitive set of criteria and required that a subset be met before the diagnosis could be made.⁷ An example of this criteria-based model is shown in Box 1, which lists the DSM-IV-TR criteria for BPD.

Some psychiatrists objected that BPD was solely a psychoanalytic construct and too theoretical for inclusion in DSM-III. Others argued that if BPD were not defined, it would be difficult to study the clinical usefulness of that definition or any other. Nonetheless, many have argued that BPD does not exist, though to what category BPD patients should belong has changed over the years:

• Is BPD nothing more than a milder or unusual presentation of an affective disorder⁸ or actually bipolar II disorder?⁹
• Is it a presentation of posttraumatic stress disorder (PTSD) called “complex PTSD,”^10-11 or an adult presentation of attention-deficit/ hyperactivity or other brain disorder?¹²
• Is it a stigmatizing diagnosis that we apply to patients whom we do not like?¹³

In truth, the diagnosis of BPD reflects a particular clinical presentation no more or less accurately than many of the well-accepted axis I disorders. Despite recent advances in the neurosciences, the dilemma we face as psychiatrists is that we make a diagnosis based upon what we see in the clinical setting (i.e., a phenotype). Yet in labeling what we believe is a specific psychiatric disorder, we make assumptions—for better or for worse, consciously or unconsciously—about pathophysiology and indirectly about genotype.

Box 1

Defining the borderline personality

Stern first used the term “borderline” in 1938 to describe patients who appeared to occupy the border between neurosis and psychosis.¹⁴ In 1942, Deutsch described the “as if” personality in patients who seemed chameleon-like. They could adapt or play the role demanded of them in specific situations, yet elsewhere—as in the analyst’s office—they had little sense of themselves and were thought to be internally disorganized and probably psychotic.¹⁵

Border to psychosis. The idea that borderline-type patients were psychotic continued in Hoch and Polatin’s description of the “pseudoneurotic schizophrenic,”¹⁶ a patient who appeared severely neurotic but was thought to employ many defenses and interpersonal styles to ward off a fundamental inner psychosis. Knight used the label “borderline states”¹⁷ to describe severely ill patients who were not frankly psychotic but fell within the realm of psychosis without qualifying for a diagnosis of schizophrenia. Knight was the first person to use the term “borderline” as a diagnostic entity, though simultaneously he argued against its use as a label because the term lacked precision.

Psychotic character. About the same time, Schmideberg characterized a group of patients whose emotional lability or affective reactivity seemed to be a consistent aspect of their clinical presentation. She believed this appearance of “stable instability”¹⁸ represented the patient’s characterologic adaptation to the world.

Frosch coined the term “psychotic character”¹⁹ that aptly captured both the characterologic and the border-to-psychosis aspects of these patients’ clinical picture. According to Frosch, these patients appeared to regress readily into psychotic thinking, yet they did not lose their ability to test reality.

Affective and emotional instability. Thus until the 1960s, the term borderline was applied primarily to patients who appeared to occupy the border between neurosis and psychosis but were thought to be closer to psychotic than neurotic. And this sitting close to the edge of psychosis appeared to be a stable condition.

Most of the attention up until this point had been paid to how these patients thought—with little attention to their affective lability or emotional instability, save for Schmideberg’s comments. In the 1960s, however, the term borderline was applied somewhat differently—not completely divorced from previous concepts but with greater emphasis on borderline as a stable but psychopathologic functioning of the personality that included affective and emotional instability and an impaired sense of self.

Box 2

Impaired personality organization. In 1967, Kernberg published a seminal article in the history of BPD diagnosis. Although he did not discuss the diagnosis of BPD, Kernberg did develop a concept concerning a specific organization of the personality based upon impaired object relations. This impaired organization could apply across several personality disorders. The construct, named borderline personality organization (BPO),²⁰ was defined by:

• an impaired sense of identity and lack of integration of one’s own identity
• use of primitive defenses, including splitting, rage, and regression
• ability to test reality.

Kernberg’s theory is too complex to summarize here, but he—along with Roy Grinker—is responsible for placing BPD on the diagnostic map. He was the first to describe BPO (and by extension BPD) in terms of a personality disorder.

Grinker’s ‘core’ group. Almost simultaneously (in 1968), Grinker published a careful study of 50 hospitalized patients. His work on the “borderline syndrome”²¹ revealed four subgroups to which the label of borderline had been applied:

• those occupying the border with psychosis
• those occupying the border with neurosis
• those similar to Deutsch’s “as if” group
• the “core” borderline group.

The core group—with its symptoms of anger and loneliness, a nonintegrated sense of self, and labile and oscillating interpersonal relationships— defined patients closest to our current definition.

Six criteria for BPD. In 1975, Gunderson and Singer published an article that greatly influenced our definition of BPD. They reviewed major descriptive accounts of BPD or BPD-like syndromes²² and proposed six diagnostic criteria (Box 2), though they did not identify a specific number or subset of the criteria as needing to be met for the diagnosis. (It is important to note that the term BPD did not become official for 5 more years.)

DSM-IV’s definition of BPD retains four of Gunderson and Singer‘s criteria among the nine it lists (five being necessary for a diagnosis of BPD). Missing are:

• social adaptiveness—though DSM-IV does say that social adaptiveness may be superficial (as in the “as if” personality) and may hide a disturbed identity⁶
• and the criterion relating to psychological test performance (this omission reflects a movement since 1980 away from listing “psychological” or psychodynamic criteria in the DSM).

DSM-III. BPD was included in DSM-III⁷ following an important study that tried to determine whether the term “borderline” refers to patients at the border of psychosis or to a stable group with mood instability and affective lability as part of a personality disorder. Spitzer et al²³ asked 808 clinicians to describe patients they would label as borderline and to use 22 items gleaned from the literature to score two of their own patients:

• one patient who the clinician felt truly had borderline personality, borderline personality organization, or borderline schizophrenia
• and a control patient who was not diagnosed as psychotic and did not fall into any borderline category.

Box 3

Their responses showed that BPD and schizotypal personality disorder (SPD) were separate, independent (though not mutually exclusive) disorders. Spitzer et al preserved the “schizotypal” label in DSM-III to describe the personality disorder that closely matched the border-to-psychosis subset. The other criteria set, which they labeled the “unstable personality item set,” was renamed “borderline” in DSM-III to describe the personality disorder that closely matched the emotional lability subset.

Box 4

Not everyone agrees with renaming the unstable set “borderline” because the word:

• has always been ambiguous
• does not connote or denote any specific criteria or characteristic of patients who bear the label
• brands the patient as untreatable, defiant, or just “bad.”

Post-DSM-III: Where are we now?

From DSM-III evolved the hope that psychiatry could describe valid, well-defined diagnostic categories. Lost in the DSM-III enthusiasm was the fact that the categories were based upon theoretic constructs—theories no more or less valid than other theories that had preceded them. Because some of these categories were based upon empiric data— such as the Spitzer et al study—these diagnoses were perceived as more valid and more related to pathophysiology and perhaps genotype than prior constructs and definitions.

In the 1980s and early 1990s, a proliferation of studies attempted to examine the validity and reliability of DSM-III definitions, and BPD became the most studied of the personality disorders. The BPD concept took hold, even though several studies did not support it and despite refinements in subsequent DSM editions (Box 3).

One refinement in the BPD construct applied to transient psychotic or psychotic-like experiences, including dissociative phenomena. Yet questions remain about the duration of these transient episodes (Box 4).

Categorical versus dimensional

The categorical concept of BPD is facing new scrutiny,^26,27 as recent studies have implied that biological disturbances may be spread across a number of personality disorders.²⁸ If biological findings are found to be more closely allied with genotypic variations (alleles),²⁹ then perhaps a dimensional classification system is needed for BPD and personality disorders in general.

On the other hand, categories provide a well-defined population that we can study and try to delineate from other populations, whereas dimensions—while perhaps closer to the reality of clinical presentation—may allow too much variability for research to proceed without confounding restraints.

BPD will continue to evolve, as will all psychiatric diagnostic categories, but the need to modify its definition does not negate its usefulness and clinical applicability. Most of our patients do not read the DSM before coming to us. They present with symptom complexes and problems that demand that we listen to what they say and understand who they are while we also try to fit them—as best we can—into categories or dimensions³⁰ that help us choose the most appropriate interventions.

Related resources

• BPD Sanctuary (borderline personality disorder education, communities, support, books, and resources) http://www.mhsanctuary.com/borderline/ Borderline Central (resources for people who care about someone with borderline personality disorder) http://www.bpdcentral.com/
• Gunderson JG. Borderline personality disorder: a clinical guide. Washington, DC: American Psychiatric Publishing, Inc., 200l.
• Paris J (ed). Borderline personality disorder. Psych Clin N Am 2000;23:1 (entire volume devoted to BPD).
• Silk KR (ed). Biological and neurobehavioral studies of borderline personality disorder. Washington, DC: American Psychiatric Press, 1994.

Not everyone agrees that borderline personality disorder (BPD) should be a diagnostic category. BPD became “official” with DSM-III in 1980, although the term had been used for 40 years to describe various patient groups. Being listed in DSM-III legitimized BPD, which was thought to represent a specific—though not necessarily distinct—diagnostic category.

The history of the BPD diagnosis and opinions as to its usefulness can be viewed as a microcosm of psychiatric diagnosis in general. Before DSM-III, diagnoses were broadly defined and did not contain specific inclusion or exclusion criteria.¹ For the 5 to 10 years prior to DSM-III, however, two assumptions developed:

• distinct diagnostic categories did, in fact, exist
• by rigorously defining and studying those categories we could develop more specific and effective treatments for our patients.²

The specificity and exclusivity that we assumed we could achieve by categorical diagnoses, however, remain a distant wish. Comorbidity appears more common in psychiatry than was originally thought and confounds both treatment and outcome.³ Also, many patients appear treatment-resistant, despite fitting neatly into diagnostic categories.⁴

As psychiatry considers DSM-V, questions linger as to whether BPD (and personality disorder in general) should remain as a categorical diagnosis or if dimensional measures may be more appropriate. Dimensions imply that no one ever fits into a given box because no specific box exists. Rather, patients are described as being closer to or more distant from a prototypic model of the diagnosis. In personality disorders, the dimensions most often mentioned are cognition, impulsivity, emotional lability, environmental hyperreactivity, and anxiety. The case report (above) illustrates the interplay of these dimensions in a typical patient with presumed BPD.

What’s in a name?

The symptom complex or syndrome that bears the name borderline personality disorder has probably existed for as long as people have thought about patients in psychopathologic terms.⁵ Before 1980, the term “borderline” applied primarily to two separate but overlapping concepts:

• Patients thought to reside on the “border” with psychosis, such as the patient in our case example. They seemed to have an underlying psychotic disorder, but the psychosis—if it surfaced—appeared briefly, was not exceptionally deep or firmly held, and was not regularly evident or immediately accessible to the clinician.
• Patients who appeared to occupy the space between neurosis and psychosis. This concept evolved into the idea of a character or personality disorder distinguished primarily by unstable interpersonal relationships, a confused or inconsistent sense of identity, and emotional instability.

How DSM is changing. Comparing the disorders listed in DSM-IV (1994)⁶ versus DSM-II (1968)¹ suggests that psychiatry has become enamored of the naming process. For example, DSM-II lists anxiety neurosis (300.0), phobic neurosis (300.2), and obsessive-compulsive neurosis (300.3), whereas DSM-IV lists 11 different categories of anxiety disorders.

But beyond naming, subsequent DSMs have differed even more dramatically from DSM-II. We have seen a shift from describing a diagnostic category with a simple explanatory paragraph to lists of specific inclusion and exclusion criteria. These more-specific lists imply that they define categories closer to some reality or authenticity than did previous definitions.

Before DSM-III, the borderline concept was conceived in broad object relational and psychodynamic terms. In contrast, DSM-III produced a definitive set of criteria and required that a subset be met before the diagnosis could be made.⁷ An example of this criteria-based model is shown in Box 1, which lists the DSM-IV-TR criteria for BPD.

Some psychiatrists objected that BPD was solely a psychoanalytic construct and too theoretical for inclusion in DSM-III. Others argued that if BPD were not defined, it would be difficult to study the clinical usefulness of that definition or any other. Nonetheless, many have argued that BPD does not exist, though to what category BPD patients should belong has changed over the years:

• Is BPD nothing more than a milder or unusual presentation of an affective disorder⁸ or actually bipolar II disorder?⁹
• Is it a presentation of posttraumatic stress disorder (PTSD) called “complex PTSD,”^10-11 or an adult presentation of attention-deficit/ hyperactivity or other brain disorder?¹²
• Is it a stigmatizing diagnosis that we apply to patients whom we do not like?¹³

In truth, the diagnosis of BPD reflects a particular clinical presentation no more or less accurately than many of the well-accepted axis I disorders. Despite recent advances in the neurosciences, the dilemma we face as psychiatrists is that we make a diagnosis based upon what we see in the clinical setting (i.e., a phenotype). Yet in labeling what we believe is a specific psychiatric disorder, we make assumptions—for better or for worse, consciously or unconsciously—about pathophysiology and indirectly about genotype.

Box 1

Defining the borderline personality

Stern first used the term “borderline” in 1938 to describe patients who appeared to occupy the border between neurosis and psychosis.¹⁴ In 1942, Deutsch described the “as if” personality in patients who seemed chameleon-like. They could adapt or play the role demanded of them in specific situations, yet elsewhere—as in the analyst’s office—they had little sense of themselves and were thought to be internally disorganized and probably psychotic.¹⁵

Border to psychosis. The idea that borderline-type patients were psychotic continued in Hoch and Polatin’s description of the “pseudoneurotic schizophrenic,”¹⁶ a patient who appeared severely neurotic but was thought to employ many defenses and interpersonal styles to ward off a fundamental inner psychosis. Knight used the label “borderline states”¹⁷ to describe severely ill patients who were not frankly psychotic but fell within the realm of psychosis without qualifying for a diagnosis of schizophrenia. Knight was the first person to use the term “borderline” as a diagnostic entity, though simultaneously he argued against its use as a label because the term lacked precision.

Psychotic character. About the same time, Schmideberg characterized a group of patients whose emotional lability or affective reactivity seemed to be a consistent aspect of their clinical presentation. She believed this appearance of “stable instability”¹⁸ represented the patient’s characterologic adaptation to the world.

Frosch coined the term “psychotic character”¹⁹ that aptly captured both the characterologic and the border-to-psychosis aspects of these patients’ clinical picture. According to Frosch, these patients appeared to regress readily into psychotic thinking, yet they did not lose their ability to test reality.

Affective and emotional instability. Thus until the 1960s, the term borderline was applied primarily to patients who appeared to occupy the border between neurosis and psychosis but were thought to be closer to psychotic than neurotic. And this sitting close to the edge of psychosis appeared to be a stable condition.

Most of the attention up until this point had been paid to how these patients thought—with little attention to their affective lability or emotional instability, save for Schmideberg’s comments. In the 1960s, however, the term borderline was applied somewhat differently—not completely divorced from previous concepts but with greater emphasis on borderline as a stable but psychopathologic functioning of the personality that included affective and emotional instability and an impaired sense of self.

Box 2

Impaired personality organization. In 1967, Kernberg published a seminal article in the history of BPD diagnosis. Although he did not discuss the diagnosis of BPD, Kernberg did develop a concept concerning a specific organization of the personality based upon impaired object relations. This impaired organization could apply across several personality disorders. The construct, named borderline personality organization (BPO),²⁰ was defined by:

• an impaired sense of identity and lack of integration of one’s own identity
• use of primitive defenses, including splitting, rage, and regression
• ability to test reality.

Kernberg’s theory is too complex to summarize here, but he—along with Roy Grinker—is responsible for placing BPD on the diagnostic map. He was the first to describe BPO (and by extension BPD) in terms of a personality disorder.

Grinker’s ‘core’ group. Almost simultaneously (in 1968), Grinker published a careful study of 50 hospitalized patients. His work on the “borderline syndrome”²¹ revealed four subgroups to which the label of borderline had been applied:

• those occupying the border with psychosis
• those occupying the border with neurosis
• those similar to Deutsch’s “as if” group
• the “core” borderline group.

The core group—with its symptoms of anger and loneliness, a nonintegrated sense of self, and labile and oscillating interpersonal relationships— defined patients closest to our current definition.

Six criteria for BPD. In 1975, Gunderson and Singer published an article that greatly influenced our definition of BPD. They reviewed major descriptive accounts of BPD or BPD-like syndromes²² and proposed six diagnostic criteria (Box 2), though they did not identify a specific number or subset of the criteria as needing to be met for the diagnosis. (It is important to note that the term BPD did not become official for 5 more years.)

DSM-IV’s definition of BPD retains four of Gunderson and Singer‘s criteria among the nine it lists (five being necessary for a diagnosis of BPD). Missing are:

• social adaptiveness—though DSM-IV does say that social adaptiveness may be superficial (as in the “as if” personality) and may hide a disturbed identity⁶
• and the criterion relating to psychological test performance (this omission reflects a movement since 1980 away from listing “psychological” or psychodynamic criteria in the DSM).

DSM-III. BPD was included in DSM-III⁷ following an important study that tried to determine whether the term “borderline” refers to patients at the border of psychosis or to a stable group with mood instability and affective lability as part of a personality disorder. Spitzer et al²³ asked 808 clinicians to describe patients they would label as borderline and to use 22 items gleaned from the literature to score two of their own patients:

• one patient who the clinician felt truly had borderline personality, borderline personality organization, or borderline schizophrenia
• and a control patient who was not diagnosed as psychotic and did not fall into any borderline category.

Box 3

Their responses showed that BPD and schizotypal personality disorder (SPD) were separate, independent (though not mutually exclusive) disorders. Spitzer et al preserved the “schizotypal” label in DSM-III to describe the personality disorder that closely matched the border-to-psychosis subset. The other criteria set, which they labeled the “unstable personality item set,” was renamed “borderline” in DSM-III to describe the personality disorder that closely matched the emotional lability subset.

Box 4

Not everyone agrees with renaming the unstable set “borderline” because the word:

• has always been ambiguous
• does not connote or denote any specific criteria or characteristic of patients who bear the label
• brands the patient as untreatable, defiant, or just “bad.”

Post-DSM-III: Where are we now?

From DSM-III evolved the hope that psychiatry could describe valid, well-defined diagnostic categories. Lost in the DSM-III enthusiasm was the fact that the categories were based upon theoretic constructs—theories no more or less valid than other theories that had preceded them. Because some of these categories were based upon empiric data— such as the Spitzer et al study—these diagnoses were perceived as more valid and more related to pathophysiology and perhaps genotype than prior constructs and definitions.

In the 1980s and early 1990s, a proliferation of studies attempted to examine the validity and reliability of DSM-III definitions, and BPD became the most studied of the personality disorders. The BPD concept took hold, even though several studies did not support it and despite refinements in subsequent DSM editions (Box 3).

One refinement in the BPD construct applied to transient psychotic or psychotic-like experiences, including dissociative phenomena. Yet questions remain about the duration of these transient episodes (Box 4).

Categorical versus dimensional

The categorical concept of BPD is facing new scrutiny,^26,27 as recent studies have implied that biological disturbances may be spread across a number of personality disorders.²⁸ If biological findings are found to be more closely allied with genotypic variations (alleles),²⁹ then perhaps a dimensional classification system is needed for BPD and personality disorders in general.

On the other hand, categories provide a well-defined population that we can study and try to delineate from other populations, whereas dimensions—while perhaps closer to the reality of clinical presentation—may allow too much variability for research to proceed without confounding restraints.

BPD will continue to evolve, as will all psychiatric diagnostic categories, but the need to modify its definition does not negate its usefulness and clinical applicability. Most of our patients do not read the DSM before coming to us. They present with symptom complexes and problems that demand that we listen to what they say and understand who they are while we also try to fit them—as best we can—into categories or dimensions³⁰ that help us choose the most appropriate interventions.

Related resources

• BPD Sanctuary (borderline personality disorder education, communities, support, books, and resources) http://www.mhsanctuary.com/borderline/ Borderline Central (resources for people who care about someone with borderline personality disorder) http://www.bpdcentral.com/
• Gunderson JG. Borderline personality disorder: a clinical guide. Washington, DC: American Psychiatric Publishing, Inc., 200l.
• Paris J (ed). Borderline personality disorder. Psych Clin N Am 2000;23:1 (entire volume devoted to BPD).
• Silk KR (ed). Biological and neurobehavioral studies of borderline personality disorder. Washington, DC: American Psychiatric Press, 1994.

Not everyone agrees that borderline personality disorder (BPD) should be a diagnostic category. BPD became “official” with DSM-III in 1980, although the term had been used for 40 years to describe various patient groups. Being listed in DSM-III legitimized BPD, which was thought to represent a specific—though not necessarily distinct—diagnostic category.

The history of the BPD diagnosis and opinions as to its usefulness can be viewed as a microcosm of psychiatric diagnosis in general. Before DSM-III, diagnoses were broadly defined and did not contain specific inclusion or exclusion criteria.¹ For the 5 to 10 years prior to DSM-III, however, two assumptions developed:

• distinct diagnostic categories did, in fact, exist
• by rigorously defining and studying those categories we could develop more specific and effective treatments for our patients.²

The specificity and exclusivity that we assumed we could achieve by categorical diagnoses, however, remain a distant wish. Comorbidity appears more common in psychiatry than was originally thought and confounds both treatment and outcome.³ Also, many patients appear treatment-resistant, despite fitting neatly into diagnostic categories.⁴

As psychiatry considers DSM-V, questions linger as to whether BPD (and personality disorder in general) should remain as a categorical diagnosis or if dimensional measures may be more appropriate. Dimensions imply that no one ever fits into a given box because no specific box exists. Rather, patients are described as being closer to or more distant from a prototypic model of the diagnosis. In personality disorders, the dimensions most often mentioned are cognition, impulsivity, emotional lability, environmental hyperreactivity, and anxiety. The case report (above) illustrates the interplay of these dimensions in a typical patient with presumed BPD.

What’s in a name?

The symptom complex or syndrome that bears the name borderline personality disorder has probably existed for as long as people have thought about patients in psychopathologic terms.⁵ Before 1980, the term “borderline” applied primarily to two separate but overlapping concepts:

• Patients thought to reside on the “border” with psychosis, such as the patient in our case example. They seemed to have an underlying psychotic disorder, but the psychosis—if it surfaced—appeared briefly, was not exceptionally deep or firmly held, and was not regularly evident or immediately accessible to the clinician.
• Patients who appeared to occupy the space between neurosis and psychosis. This concept evolved into the idea of a character or personality disorder distinguished primarily by unstable interpersonal relationships, a confused or inconsistent sense of identity, and emotional instability.

How DSM is changing. Comparing the disorders listed in DSM-IV (1994)⁶ versus DSM-II (1968)¹ suggests that psychiatry has become enamored of the naming process. For example, DSM-II lists anxiety neurosis (300.0), phobic neurosis (300.2), and obsessive-compulsive neurosis (300.3), whereas DSM-IV lists 11 different categories of anxiety disorders.

But beyond naming, subsequent DSMs have differed even more dramatically from DSM-II. We have seen a shift from describing a diagnostic category with a simple explanatory paragraph to lists of specific inclusion and exclusion criteria. These more-specific lists imply that they define categories closer to some reality or authenticity than did previous definitions.

Before DSM-III, the borderline concept was conceived in broad object relational and psychodynamic terms. In contrast, DSM-III produced a definitive set of criteria and required that a subset be met before the diagnosis could be made.⁷ An example of this criteria-based model is shown in Box 1, which lists the DSM-IV-TR criteria for BPD.

Some psychiatrists objected that BPD was solely a psychoanalytic construct and too theoretical for inclusion in DSM-III. Others argued that if BPD were not defined, it would be difficult to study the clinical usefulness of that definition or any other. Nonetheless, many have argued that BPD does not exist, though to what category BPD patients should belong has changed over the years:

• Is BPD nothing more than a milder or unusual presentation of an affective disorder⁸ or actually bipolar II disorder?⁹
• Is it a presentation of posttraumatic stress disorder (PTSD) called “complex PTSD,”^10-11 or an adult presentation of attention-deficit/ hyperactivity or other brain disorder?¹²
• Is it a stigmatizing diagnosis that we apply to patients whom we do not like?¹³

In truth, the diagnosis of BPD reflects a particular clinical presentation no more or less accurately than many of the well-accepted axis I disorders. Despite recent advances in the neurosciences, the dilemma we face as psychiatrists is that we make a diagnosis based upon what we see in the clinical setting (i.e., a phenotype). Yet in labeling what we believe is a specific psychiatric disorder, we make assumptions—for better or for worse, consciously or unconsciously—about pathophysiology and indirectly about genotype.

Box 1

Defining the borderline personality

Stern first used the term “borderline” in 1938 to describe patients who appeared to occupy the border between neurosis and psychosis.¹⁴ In 1942, Deutsch described the “as if” personality in patients who seemed chameleon-like. They could adapt or play the role demanded of them in specific situations, yet elsewhere—as in the analyst’s office—they had little sense of themselves and were thought to be internally disorganized and probably psychotic.¹⁵

Border to psychosis. The idea that borderline-type patients were psychotic continued in Hoch and Polatin’s description of the “pseudoneurotic schizophrenic,”¹⁶ a patient who appeared severely neurotic but was thought to employ many defenses and interpersonal styles to ward off a fundamental inner psychosis. Knight used the label “borderline states”¹⁷ to describe severely ill patients who were not frankly psychotic but fell within the realm of psychosis without qualifying for a diagnosis of schizophrenia. Knight was the first person to use the term “borderline” as a diagnostic entity, though simultaneously he argued against its use as a label because the term lacked precision.

Psychotic character. About the same time, Schmideberg characterized a group of patients whose emotional lability or affective reactivity seemed to be a consistent aspect of their clinical presentation. She believed this appearance of “stable instability”¹⁸ represented the patient’s characterologic adaptation to the world.

Frosch coined the term “psychotic character”¹⁹ that aptly captured both the characterologic and the border-to-psychosis aspects of these patients’ clinical picture. According to Frosch, these patients appeared to regress readily into psychotic thinking, yet they did not lose their ability to test reality.

Affective and emotional instability. Thus until the 1960s, the term borderline was applied primarily to patients who appeared to occupy the border between neurosis and psychosis but were thought to be closer to psychotic than neurotic. And this sitting close to the edge of psychosis appeared to be a stable condition.

Most of the attention up until this point had been paid to how these patients thought—with little attention to their affective lability or emotional instability, save for Schmideberg’s comments. In the 1960s, however, the term borderline was applied somewhat differently—not completely divorced from previous concepts but with greater emphasis on borderline as a stable but psychopathologic functioning of the personality that included affective and emotional instability and an impaired sense of self.

Box 2

Impaired personality organization. In 1967, Kernberg published a seminal article in the history of BPD diagnosis. Although he did not discuss the diagnosis of BPD, Kernberg did develop a concept concerning a specific organization of the personality based upon impaired object relations. This impaired organization could apply across several personality disorders. The construct, named borderline personality organization (BPO),²⁰ was defined by:

• an impaired sense of identity and lack of integration of one’s own identity
• use of primitive defenses, including splitting, rage, and regression
• ability to test reality.

Kernberg’s theory is too complex to summarize here, but he—along with Roy Grinker—is responsible for placing BPD on the diagnostic map. He was the first to describe BPO (and by extension BPD) in terms of a personality disorder.

Grinker’s ‘core’ group. Almost simultaneously (in 1968), Grinker published a careful study of 50 hospitalized patients. His work on the “borderline syndrome”²¹ revealed four subgroups to which the label of borderline had been applied:

• those occupying the border with psychosis
• those occupying the border with neurosis
• those similar to Deutsch’s “as if” group
• the “core” borderline group.

The core group—with its symptoms of anger and loneliness, a nonintegrated sense of self, and labile and oscillating interpersonal relationships— defined patients closest to our current definition.

Six criteria for BPD. In 1975, Gunderson and Singer published an article that greatly influenced our definition of BPD. They reviewed major descriptive accounts of BPD or BPD-like syndromes²² and proposed six diagnostic criteria (Box 2), though they did not identify a specific number or subset of the criteria as needing to be met for the diagnosis. (It is important to note that the term BPD did not become official for 5 more years.)

DSM-IV’s definition of BPD retains four of Gunderson and Singer‘s criteria among the nine it lists (five being necessary for a diagnosis of BPD). Missing are:

• social adaptiveness—though DSM-IV does say that social adaptiveness may be superficial (as in the “as if” personality) and may hide a disturbed identity⁶
• and the criterion relating to psychological test performance (this omission reflects a movement since 1980 away from listing “psychological” or psychodynamic criteria in the DSM).

DSM-III. BPD was included in DSM-III⁷ following an important study that tried to determine whether the term “borderline” refers to patients at the border of psychosis or to a stable group with mood instability and affective lability as part of a personality disorder. Spitzer et al²³ asked 808 clinicians to describe patients they would label as borderline and to use 22 items gleaned from the literature to score two of their own patients:

• one patient who the clinician felt truly had borderline personality, borderline personality organization, or borderline schizophrenia
• and a control patient who was not diagnosed as psychotic and did not fall into any borderline category.

Box 3

Their responses showed that BPD and schizotypal personality disorder (SPD) were separate, independent (though not mutually exclusive) disorders. Spitzer et al preserved the “schizotypal” label in DSM-III to describe the personality disorder that closely matched the border-to-psychosis subset. The other criteria set, which they labeled the “unstable personality item set,” was renamed “borderline” in DSM-III to describe the personality disorder that closely matched the emotional lability subset.

Box 4

Not everyone agrees with renaming the unstable set “borderline” because the word:

• has always been ambiguous
• does not connote or denote any specific criteria or characteristic of patients who bear the label
• brands the patient as untreatable, defiant, or just “bad.”

Post-DSM-III: Where are we now?

From DSM-III evolved the hope that psychiatry could describe valid, well-defined diagnostic categories. Lost in the DSM-III enthusiasm was the fact that the categories were based upon theoretic constructs—theories no more or less valid than other theories that had preceded them. Because some of these categories were based upon empiric data— such as the Spitzer et al study—these diagnoses were perceived as more valid and more related to pathophysiology and perhaps genotype than prior constructs and definitions.

In the 1980s and early 1990s, a proliferation of studies attempted to examine the validity and reliability of DSM-III definitions, and BPD became the most studied of the personality disorders. The BPD concept took hold, even though several studies did not support it and despite refinements in subsequent DSM editions (Box 3).

One refinement in the BPD construct applied to transient psychotic or psychotic-like experiences, including dissociative phenomena. Yet questions remain about the duration of these transient episodes (Box 4).

Categorical versus dimensional

The categorical concept of BPD is facing new scrutiny,^26,27 as recent studies have implied that biological disturbances may be spread across a number of personality disorders.²⁸ If biological findings are found to be more closely allied with genotypic variations (alleles),²⁹ then perhaps a dimensional classification system is needed for BPD and personality disorders in general.

On the other hand, categories provide a well-defined population that we can study and try to delineate from other populations, whereas dimensions—while perhaps closer to the reality of clinical presentation—may allow too much variability for research to proceed without confounding restraints.

BPD will continue to evolve, as will all psychiatric diagnostic categories, but the need to modify its definition does not negate its usefulness and clinical applicability. Most of our patients do not read the DSM before coming to us. They present with symptom complexes and problems that demand that we listen to what they say and understand who they are while we also try to fit them—as best we can—into categories or dimensions³⁰ that help us choose the most appropriate interventions.

Related resources

• BPD Sanctuary (borderline personality disorder education, communities, support, books, and resources) http://www.mhsanctuary.com/borderline/ Borderline Central (resources for people who care about someone with borderline personality disorder) http://www.bpdcentral.com/
• Gunderson JG. Borderline personality disorder: a clinical guide. Washington, DC: American Psychiatric Publishing, Inc., 200l.
• Paris J (ed). Borderline personality disorder. Psych Clin N Am 2000;23:1 (entire volume devoted to BPD).
• Silk KR (ed). Biological and neurobehavioral studies of borderline personality disorder. Washington, DC: American Psychiatric Press, 1994.

The two articles in your July issue on sexual addiction address two different subjects.

Steven L. Mahorney, MD, writes about a problem that troubles patients and for which they seek help. These people are troubled by how their sexual activities disturb their lives. Since we physicians may be able to help these troubled persons, we call it a “disorder.”

Neal W. Dunsieth, MD, writes about people whose sexual acts disturb others and thus create a problem for society. Society tries to cope with these disturbers of our peace in several ways, one of which is to lock them up in hospitals and in jails. If we confine these offenders and try to treat them, we are more or less obligated to give them a medical diagnosis. But it should not bear the same name as the disorder that applies to the people Dr. Mahorney writes about.

The distinction is similar to the difference between “obsessive-compulsive disorder” and “obsessive-compulsive personality.” The person with the “disorder” is troubled by it, but the one with the “personality” sees himself as orderly, neat, and thinks that “if a task is worth doing, it is worth doing correctly.” Others may find that person to be a problem.

Unfortunately, some lawyers think that a client with a medical diagnosis should not be held responsible for his or her behavior, which may be seen as part of the diagnosis. This type of thinking would dictate that a person with a smoking addiction is not responsible for quitting smoking.

Yehuda Sherman, MD
Lafayette, CA

Dr. Dunsieth responds

Although I agree there may be an analogy between degrees of problem sexual behavior and obsessive-compulsive disorder versus obsessive-compulsive personality, I view the distinction somewhat differently.

We probably have a “neurotic/organic” split that sorts self-identified patients from felons. This would follow somewhat logically because the felons have lost control of their impulses to the point that their infractions are severe and harmful to society. Therefore, their underlying problems probably are more severe. This may have implications for treatment.

Overall, though, there is too much heterogeneity among individuals with problem sexual behavior to make many generalizations. I will hold to only one generalization with these patients as a whole: Defensiveness surrounding problem sexual behavior runs higher among individuals than it does for virtually any other problem in psychiatry. When treating sexual behavior, the clinician must employ a high degree of structure and a somewhat skeptical ear.

Neal W. Dunsieth, MD
University of Cincinnat
College of Medicine

Dr. Mahorney responds

I appreciate Dr. Sherman’s letter. I had the same reaction to the juxtaposition of my article with that of Dr. Dunsieth.

The combination of DSM labeling, pharmaceutical marketing, and managed care has left many of us feeling that the best a psychiatrist can hope for is to sign the prescriptions and orders of a multidisciplinary team in a prison or jail, give the prisoners structured interviews, and pass the statistical analysis of those interviews off as “research.” The “research” can be published and the authors declared “experts” whose testimony can then be peddled to lawyers and their clients. I didn’t get the idea from Dr. Hillard’s first editorial, justifying the creation of a new journal, that that was the only vision out there. It certainly isn’t what my article was about.

I still think that as doctors, we should try to find ways to help our patients—before they go to jail.

I also want to disavow the automatic connection some have seen in my article between the usefulness of the sexual addiction paradigm and 12-step programs as treatment modalities. An exciting debate on this subject is emerging in clinical psychology circles. Psychiatrists should be part of that debate, and we can be if we could just tear ourselves away from our DSM bean counting.

Steven L. Mahorney, MD
University of North Carolina
School of Medicine, Chapel Hill

The two articles in your July issue on sexual addiction address two different subjects.

Steven L. Mahorney, MD, writes about a problem that troubles patients and for which they seek help. These people are troubled by how their sexual activities disturb their lives. Since we physicians may be able to help these troubled persons, we call it a “disorder.”

Neal W. Dunsieth, MD, writes about people whose sexual acts disturb others and thus create a problem for society. Society tries to cope with these disturbers of our peace in several ways, one of which is to lock them up in hospitals and in jails. If we confine these offenders and try to treat them, we are more or less obligated to give them a medical diagnosis. But it should not bear the same name as the disorder that applies to the people Dr. Mahorney writes about.

The distinction is similar to the difference between “obsessive-compulsive disorder” and “obsessive-compulsive personality.” The person with the “disorder” is troubled by it, but the one with the “personality” sees himself as orderly, neat, and thinks that “if a task is worth doing, it is worth doing correctly.” Others may find that person to be a problem.

Unfortunately, some lawyers think that a client with a medical diagnosis should not be held responsible for his or her behavior, which may be seen as part of the diagnosis. This type of thinking would dictate that a person with a smoking addiction is not responsible for quitting smoking.

Yehuda Sherman, MD
Lafayette, CA

Dr. Dunsieth responds

Although I agree there may be an analogy between degrees of problem sexual behavior and obsessive-compulsive disorder versus obsessive-compulsive personality, I view the distinction somewhat differently.

We probably have a “neurotic/organic” split that sorts self-identified patients from felons. This would follow somewhat logically because the felons have lost control of their impulses to the point that their infractions are severe and harmful to society. Therefore, their underlying problems probably are more severe. This may have implications for treatment.

Overall, though, there is too much heterogeneity among individuals with problem sexual behavior to make many generalizations. I will hold to only one generalization with these patients as a whole: Defensiveness surrounding problem sexual behavior runs higher among individuals than it does for virtually any other problem in psychiatry. When treating sexual behavior, the clinician must employ a high degree of structure and a somewhat skeptical ear.

Neal W. Dunsieth, MD
University of Cincinnat
College of Medicine

Dr. Mahorney responds

I appreciate Dr. Sherman’s letter. I had the same reaction to the juxtaposition of my article with that of Dr. Dunsieth.

The combination of DSM labeling, pharmaceutical marketing, and managed care has left many of us feeling that the best a psychiatrist can hope for is to sign the prescriptions and orders of a multidisciplinary team in a prison or jail, give the prisoners structured interviews, and pass the statistical analysis of those interviews off as “research.” The “research” can be published and the authors declared “experts” whose testimony can then be peddled to lawyers and their clients. I didn’t get the idea from Dr. Hillard’s first editorial, justifying the creation of a new journal, that that was the only vision out there. It certainly isn’t what my article was about.

I still think that as doctors, we should try to find ways to help our patients—before they go to jail.

I also want to disavow the automatic connection some have seen in my article between the usefulness of the sexual addiction paradigm and 12-step programs as treatment modalities. An exciting debate on this subject is emerging in clinical psychology circles. Psychiatrists should be part of that debate, and we can be if we could just tear ourselves away from our DSM bean counting.

Steven L. Mahorney, MD
University of North Carolina
School of Medicine, Chapel Hill

The two articles in your July issue on sexual addiction address two different subjects.

Steven L. Mahorney, MD, writes about a problem that troubles patients and for which they seek help. These people are troubled by how their sexual activities disturb their lives. Since we physicians may be able to help these troubled persons, we call it a “disorder.”

Neal W. Dunsieth, MD, writes about people whose sexual acts disturb others and thus create a problem for society. Society tries to cope with these disturbers of our peace in several ways, one of which is to lock them up in hospitals and in jails. If we confine these offenders and try to treat them, we are more or less obligated to give them a medical diagnosis. But it should not bear the same name as the disorder that applies to the people Dr. Mahorney writes about.

The distinction is similar to the difference between “obsessive-compulsive disorder” and “obsessive-compulsive personality.” The person with the “disorder” is troubled by it, but the one with the “personality” sees himself as orderly, neat, and thinks that “if a task is worth doing, it is worth doing correctly.” Others may find that person to be a problem.

Unfortunately, some lawyers think that a client with a medical diagnosis should not be held responsible for his or her behavior, which may be seen as part of the diagnosis. This type of thinking would dictate that a person with a smoking addiction is not responsible for quitting smoking.

Yehuda Sherman, MD
Lafayette, CA

Dr. Dunsieth responds

Although I agree there may be an analogy between degrees of problem sexual behavior and obsessive-compulsive disorder versus obsessive-compulsive personality, I view the distinction somewhat differently.

We probably have a “neurotic/organic” split that sorts self-identified patients from felons. This would follow somewhat logically because the felons have lost control of their impulses to the point that their infractions are severe and harmful to society. Therefore, their underlying problems probably are more severe. This may have implications for treatment.

Overall, though, there is too much heterogeneity among individuals with problem sexual behavior to make many generalizations. I will hold to only one generalization with these patients as a whole: Defensiveness surrounding problem sexual behavior runs higher among individuals than it does for virtually any other problem in psychiatry. When treating sexual behavior, the clinician must employ a high degree of structure and a somewhat skeptical ear.

Neal W. Dunsieth, MD
University of Cincinnat
College of Medicine

Dr. Mahorney responds

I appreciate Dr. Sherman’s letter. I had the same reaction to the juxtaposition of my article with that of Dr. Dunsieth.

The combination of DSM labeling, pharmaceutical marketing, and managed care has left many of us feeling that the best a psychiatrist can hope for is to sign the prescriptions and orders of a multidisciplinary team in a prison or jail, give the prisoners structured interviews, and pass the statistical analysis of those interviews off as “research.” The “research” can be published and the authors declared “experts” whose testimony can then be peddled to lawyers and their clients. I didn’t get the idea from Dr. Hillard’s first editorial, justifying the creation of a new journal, that that was the only vision out there. It certainly isn’t what my article was about.

I still think that as doctors, we should try to find ways to help our patients—before they go to jail.

I also want to disavow the automatic connection some have seen in my article between the usefulness of the sexual addiction paradigm and 12-step programs as treatment modalities. An exciting debate on this subject is emerging in clinical psychology circles. Psychiatrists should be part of that debate, and we can be if we could just tear ourselves away from our DSM bean counting.

Steven L. Mahorney, MD
University of North Carolina
School of Medicine, Chapel Hill

The two articles in your July issue on sexual addiction address two different subjects.

Steven L. Mahorney, MD, writes about a problem that troubles patients and for which they seek help. These people are troubled by how their sexual activities disturb their lives. Since we physicians may be able to help these troubled persons, we call it a “disorder.”

Neal W. Dunsieth, MD, writes about people whose sexual acts disturb others and thus create a problem for society. Society tries to cope with these disturbers of our peace in several ways, one of which is to lock them up in hospitals and in jails. If we confine these offenders and try to treat them, we are more or less obligated to give them a medical diagnosis. But it should not bear the same name as the disorder that applies to the people Dr. Mahorney writes about.

The distinction is similar to the difference between “obsessive-compulsive disorder” and “obsessive-compulsive personality.” The person with the “disorder” is troubled by it, but the one with the “personality” sees himself as orderly, neat, and thinks that “if a task is worth doing, it is worth doing correctly.” Others may find that person to be a problem.

Unfortunately, some lawyers think that a client with a medical diagnosis should not be held responsible for his or her behavior, which may be seen as part of the diagnosis. This type of thinking would dictate that a person with a smoking addiction is not responsible for quitting smoking.

Yehuda Sherman, MD
Lafayette, CA

Dr. Dunsieth responds

Although I agree there may be an analogy between degrees of problem sexual behavior and obsessive-compulsive disorder versus obsessive-compulsive personality, I view the distinction somewhat differently.

We probably have a “neurotic/organic” split that sorts self-identified patients from felons. This would follow somewhat logically because the felons have lost control of their impulses to the point that their infractions are severe and harmful to society. Therefore, their underlying problems probably are more severe. This may have implications for treatment.

Overall, though, there is too much heterogeneity among individuals with problem sexual behavior to make many generalizations. I will hold to only one generalization with these patients as a whole: Defensiveness surrounding problem sexual behavior runs higher among individuals than it does for virtually any other problem in psychiatry. When treating sexual behavior, the clinician must employ a high degree of structure and a somewhat skeptical ear.

Neal W. Dunsieth, MD
University of Cincinnat
College of Medicine

Dr. Mahorney responds

I appreciate Dr. Sherman’s letter. I had the same reaction to the juxtaposition of my article with that of Dr. Dunsieth.

The combination of DSM labeling, pharmaceutical marketing, and managed care has left many of us feeling that the best a psychiatrist can hope for is to sign the prescriptions and orders of a multidisciplinary team in a prison or jail, give the prisoners structured interviews, and pass the statistical analysis of those interviews off as “research.” The “research” can be published and the authors declared “experts” whose testimony can then be peddled to lawyers and their clients. I didn’t get the idea from Dr. Hillard’s first editorial, justifying the creation of a new journal, that that was the only vision out there. It certainly isn’t what my article was about.

I still think that as doctors, we should try to find ways to help our patients—before they go to jail.

I also want to disavow the automatic connection some have seen in my article between the usefulness of the sexual addiction paradigm and 12-step programs as treatment modalities. An exciting debate on this subject is emerging in clinical psychology circles. Psychiatrists should be part of that debate, and we can be if we could just tear ourselves away from our DSM bean counting.

Steven L. Mahorney, MD
University of North Carolina
School of Medicine, Chapel Hill

The two articles in your July issue on sexual addiction address two different subjects.

Steven L. Mahorney, MD, writes about a problem that troubles patients and for which they seek help. These people are troubled by how their sexual activities disturb their lives. Since we physicians may be able to help these troubled persons, we call it a “disorder.”

Neal W. Dunsieth, MD, writes about people whose sexual acts disturb others and thus create a problem for society. Society tries to cope with these disturbers of our peace in several ways, one of which is to lock them up in hospitals and in jails. If we confine these offenders and try to treat them, we are more or less obligated to give them a medical diagnosis. But it should not bear the same name as the disorder that applies to the people Dr. Mahorney writes about.

The distinction is similar to the difference between “obsessive-compulsive disorder” and “obsessive-compulsive personality.” The person with the “disorder” is troubled by it, but the one with the “personality” sees himself as orderly, neat, and thinks that “if a task is worth doing, it is worth doing correctly.” Others may find that person to be a problem.

Unfortunately, some lawyers think that a client with a medical diagnosis should not be held responsible for his or her behavior, which may be seen as part of the diagnosis. This type of thinking would dictate that a person with a smoking addiction is not responsible for quitting smoking.

Yehuda Sherman, MD
Lafayette, CA

Dr. Dunsieth responds

Although I agree there may be an analogy between degrees of problem sexual behavior and obsessive-compulsive disorder versus obsessive-compulsive personality, I view the distinction somewhat differently.

We probably have a “neurotic/organic” split that sorts self-identified patients from felons. This would follow somewhat logically because the felons have lost control of their impulses to the point that their infractions are severe and harmful to society. Therefore, their underlying problems probably are more severe. This may have implications for treatment.

Overall, though, there is too much heterogeneity among individuals with problem sexual behavior to make many generalizations. I will hold to only one generalization with these patients as a whole: Defensiveness surrounding problem sexual behavior runs higher among individuals than it does for virtually any other problem in psychiatry. When treating sexual behavior, the clinician must employ a high degree of structure and a somewhat skeptical ear.

Neal W. Dunsieth, MD
University of Cincinnat
College of Medicine

Dr. Mahorney responds

I appreciate Dr. Sherman’s letter. I had the same reaction to the juxtaposition of my article with that of Dr. Dunsieth.

The combination of DSM labeling, pharmaceutical marketing, and managed care has left many of us feeling that the best a psychiatrist can hope for is to sign the prescriptions and orders of a multidisciplinary team in a prison or jail, give the prisoners structured interviews, and pass the statistical analysis of those interviews off as “research.” The “research” can be published and the authors declared “experts” whose testimony can then be peddled to lawyers and their clients. I didn’t get the idea from Dr. Hillard’s first editorial, justifying the creation of a new journal, that that was the only vision out there. It certainly isn’t what my article was about.

I still think that as doctors, we should try to find ways to help our patients—before they go to jail.

I also want to disavow the automatic connection some have seen in my article between the usefulness of the sexual addiction paradigm and 12-step programs as treatment modalities. An exciting debate on this subject is emerging in clinical psychology circles. Psychiatrists should be part of that debate, and we can be if we could just tear ourselves away from our DSM bean counting.

Steven L. Mahorney, MD
University of North Carolina
School of Medicine, Chapel Hill

The two articles in your July issue on sexual addiction address two different subjects.

Steven L. Mahorney, MD, writes about a problem that troubles patients and for which they seek help. These people are troubled by how their sexual activities disturb their lives. Since we physicians may be able to help these troubled persons, we call it a “disorder.”

Neal W. Dunsieth, MD, writes about people whose sexual acts disturb others and thus create a problem for society. Society tries to cope with these disturbers of our peace in several ways, one of which is to lock them up in hospitals and in jails. If we confine these offenders and try to treat them, we are more or less obligated to give them a medical diagnosis. But it should not bear the same name as the disorder that applies to the people Dr. Mahorney writes about.

The distinction is similar to the difference between “obsessive-compulsive disorder” and “obsessive-compulsive personality.” The person with the “disorder” is troubled by it, but the one with the “personality” sees himself as orderly, neat, and thinks that “if a task is worth doing, it is worth doing correctly.” Others may find that person to be a problem.

Unfortunately, some lawyers think that a client with a medical diagnosis should not be held responsible for his or her behavior, which may be seen as part of the diagnosis. This type of thinking would dictate that a person with a smoking addiction is not responsible for quitting smoking.

Yehuda Sherman, MD
Lafayette, CA

Dr. Dunsieth responds

Although I agree there may be an analogy between degrees of problem sexual behavior and obsessive-compulsive disorder versus obsessive-compulsive personality, I view the distinction somewhat differently.

We probably have a “neurotic/organic” split that sorts self-identified patients from felons. This would follow somewhat logically because the felons have lost control of their impulses to the point that their infractions are severe and harmful to society. Therefore, their underlying problems probably are more severe. This may have implications for treatment.

Overall, though, there is too much heterogeneity among individuals with problem sexual behavior to make many generalizations. I will hold to only one generalization with these patients as a whole: Defensiveness surrounding problem sexual behavior runs higher among individuals than it does for virtually any other problem in psychiatry. When treating sexual behavior, the clinician must employ a high degree of structure and a somewhat skeptical ear.

Neal W. Dunsieth, MD
University of Cincinnat
College of Medicine

Dr. Mahorney responds

I appreciate Dr. Sherman’s letter. I had the same reaction to the juxtaposition of my article with that of Dr. Dunsieth.

The combination of DSM labeling, pharmaceutical marketing, and managed care has left many of us feeling that the best a psychiatrist can hope for is to sign the prescriptions and orders of a multidisciplinary team in a prison or jail, give the prisoners structured interviews, and pass the statistical analysis of those interviews off as “research.” The “research” can be published and the authors declared “experts” whose testimony can then be peddled to lawyers and their clients. I didn’t get the idea from Dr. Hillard’s first editorial, justifying the creation of a new journal, that that was the only vision out there. It certainly isn’t what my article was about.

I still think that as doctors, we should try to find ways to help our patients—before they go to jail.

I also want to disavow the automatic connection some have seen in my article between the usefulness of the sexual addiction paradigm and 12-step programs as treatment modalities. An exciting debate on this subject is emerging in clinical psychology circles. Psychiatrists should be part of that debate, and we can be if we could just tear ourselves away from our DSM bean counting.

Steven L. Mahorney, MD
University of North Carolina
School of Medicine, Chapel Hill

What is kleptomania? An independent illness, a symptom of other psychiatric disorders, or merely criminal behavior? Kleptomania—a disorder defined by an inability to resist the impulse to steal—is one of psychiatry’s most poorly understood diagnoses, even though it has been recognized in the literature for almost 200 years.

Kleptomania causes notable distress and impaired functioning.¹ People with kleptomania often suffer from comorbid mood, anxiety, substance use, and other impulse-control disorders.¹⁴ They experience the humiliation of repeated arrests, which leads to guilt, depression, and even suicide.^1,5 Yet kleptomania usually goes undiagnosed and untreated, despite a lifetime prevalence as high as 0.6%.⁶

Case report: ‘I’m a thief’

“I’m a thief,” began Susan, age 39. “I steal something four or five times every week. I steal from grocery stores and clothing stores. Sometimes I might steal something like vanilla extract; other times an expensive men’s tie. I probably steal $200 worth of items every week.

“You probably won’t believe this, but I don’t want or need the stuff I take. I have plenty of money. I have no idea why I take the things I do. That’s why I’m so depressed. What kind of person does something like this?

The urge to steal “I was probably 14 when I started stealing. I would go to stores with my mother. When I saw certain objects, I would get urges to steal them. The odd thing was that the items I stole were so ridiculous. I remember stealing key chains for several months, maybe three or four times a week. When I got older, things got worse. I was having urges more often, and so I needed to steal more often.

Box

“My entire life has been torment. Each day I worry about having the urges, and then I worry about being caught stealing. I can’t relax. I’ve been married for 17 years, and I haven’t told my husband. My secrecy is tearing our marriage apart. My husband thinks I’m having an affair because I’ve distanced myself emotionally from him.”

Table 1

SCREENING TEST FOR KLEPTOMANIA

Susan described urges to steal almost every day. When the urges were mild, she could resist them. Other days they were severe, and Susan felt unable to control her behavior. At work, her urges distracted her from completing projects, and her performance suffered. The urge to steal would often compel Susan to leave work early so she could get to a store.

Calm, then guilt “Every time I steal something I feel both a thrill and a great sense of calm,” she said. “It feels good. The problem is that almost immediately after each theft, I feel guilty and ashamed. After I steal, I usually donate the items to the Salvation Army, throw them away, or give them away as gifts.”

Drug trials We started treating Susan with the selective serotonin reuptake inhibitor (SSRI), citalopram. She reported notable improvement in her mood after 3 weeks on a dosage of 60 mg/d and she had been attending weekly psychotherapy, although her stealing continued unchanged. The addition of naltrexone, 200 mg/d for 2 weeks, decreased the frequency of Susan’s stealing and reduced her urges to steal, but her symptoms continued to interfere significantly with her overall functioning.

We then added the atypical antipsychotic quetiapine, 100 mg bid, and Susan’s urges to steal and stealing behavior went into remission within 3 weeks. She has refrained from stealing for the last 9 months.

Making the diagnosis

In our clinic, we have treated more than 50 patients with kleptomania. Rather than coming to us through the criminal justice system, they are usually self-referred. Often they contact us after discovering on the Internet that we specialize in treating persons with this disorder.

In our experience, kleptomania typically goes undiagnosed in clinical settings, in part because patients are ashamed and embarrassed to discuss their symptoms with physicians unless specifically asked.¹ If left untreated, however, kleptomania frequently becomes chronic.⁴ If persons with kleptomania are to seek treatment, it is important that family, friends, and mental health professionals understand the myths and facts about this disorder (Box).

To make the diagnosis, we use the simple screening instrument shown in Table 1.⁷ In general, because of high comorbidity with certain disorders, we screen every patient presenting to our clinic with a mood, substance use, anxiety, or eating disorder, or who has a problem with impulse control. Kleptomania is likely if the patient answers “yes” to questions 1 through 4 and to question 5 or 6. Stealing may be a symptom of several other psychiatric disorders, however, and misdiagnosis is fairly common (Table 2).^6-8

Data suggest that the female-to-male ratio in kleptomania is approximately 2:1, with onset in adolescence. Typical individuals with kleptomania steal because they have urges to steal, often triggered by specific stimuli such as the sights and sounds of stores or feelings of loneliness or stress.¹ Most patients with kleptomania are fairly specific about the types of stores from which they steal and the items they steal, and most hoard stolen items.¹

Although many patients with kleptomania function quite well, others are severely debilitated in social and occupational realms. In a series of 22 patients with kleptomania:

• 64% had been apprehended
• 23% had served jail time
• 27% had been hospitalized because of their kleptomania symptoms
• 18% had considered or attempted suicide because of the distress associated with their kleptomania.¹

Treatment recommendations

Patient history With patients who steal, we begin by identifying the motivation behind the stealing. Most patients with kleptomania report urges to steal. Some of these patients may have comorbid depression; for them, stealing makes them feel less depressed. Anger or irritability may point to borderline personality disorder. Stealing for the enjoyment of risk may suggest bipolar disorder.

Table 2

DIFFERENTIAL DIAGNOSIS: DISORDERS THAT MAY INVOLVE STEALING

Many patients with kleptomania have comorbid mood, substance, or anxiety disorders. Treating these other symptoms while ignoring the symptoms of kleptomania may be unsuccessful. Comorbidity also may influence the choice of medication.

Medical assessment Case reports have associated the onset of kleptomania with a variety of medical conditions, including presenile cortical atrophy in a 25-year-old, a parietal tumor that caused blackouts and obliterated any memory of stealing episodes, narcolepsy, and an insulinoma that caused severe hypoglycemia.⁹ The relationship of these conditions with the onset of kleptomania is unclear, but the reports suggest that medical causes—although unlikely—should be ruled out before you consider kleptomania as a psychiatric illness.

Patient education Persons with kleptomania often feel that no one else has the same problem. They do not think of their behavior as being an illness. It is helpful to explain that kleptomania is treatable and to connect patients with educational books, self-help groups, and Web sites providing information and support (see “Related resources”).

Cognitive-behavioral therapy (CBT) Although the evidence is quite limited, covert sensitization, exposure and response prevention, and imaginal desensitization have all been shown effective in case reports.¹⁰

What medications are effective?

Only case reports, a case series of five subjects, and a single open-label treatment study involving 10 subjects with kleptomania have been done.

So far, uses of tricyclic antidepressants (imipramine, nortriptyline), SSRIs (fluoxetine, fluvoxamine, paroxetine), the opioid antagonist naltrexone, and mood stabilizers (lithium, valproate) have met with varying degrees of success. Strategies targeting urge and behavior reduction and mechanisms for coping with urges and behavior (e.g., cognitive-behavioral therapies) may represent important adjunctive components.^2,11-17

No medications are FDA-approved for treating kleptomania. Therefore, it is important to inform patients of any off-label use of medications for this disorder, as well as the empirical basis for considering pharmacologic treatment.

SSRIs Only case reports exist on the use of SSRIs in treating kleptomania. The disorder may share a common pathology with pathologic gambling, and in our clinical experience appears to respond to similar treatments.¹⁸ We draw on research of pathologic gambling as well as our clinical experience in choosing SSRIs as first-line treatment, especially for patients with significant mood symptoms.¹⁹

We suggest titrating SSRIs to the maximum recommended dosage. As in the treatment of pathologic gambling, dosages of SSRIs required to treat kleptomania symptoms appear to be higher than average dosages required to treat depressive disorders. An SSRI should not be considered ineffective unless it has been tried for at least 10 to 12 weeks and the highest dosage tolerated or recommended by the manufacturer has been reached.

Response to SSRIs usually is characterized by decreased thoughts about stealing, decreased stealing behavior, and improvement in social and occupational functioning. If an SSRI is only partially effective, we consider augmentation with naltrexone, buspirone, or a mood stabilizer.

Naltrexone Patients taking naltrexone often report less-intense urges to steal. The urges may not disappear but are often sufficiently reduced so that the patient can resist them more easily. Patients also report that the thrill associated with stealing is reduced or eliminated.

Naltrexone was used in the first medication study of kleptomania and showed a significant decline in the intensity of urges to steal, stealing thoughts, and stealing behavior. Average dosage was 150 mg/d;¹¹ a reduced dosage (e.g., 50 mg/d) may work in adolescents with kleptomania.²⁰

Nausea as a side effect can be reduced by starting patients on 25 mg/d for the first 3 or 4 days and possibly adding ondansetron, 4 to 8 mg/d. Nausea and diarrhea are usually mild and resolve within the first week. Clinically, most patients respond to naltrexone within 2 weeks. After that, the dosage usually needs to be adjusted.

In patients with comorbid depression, augmentation with an SSRI may prevent worsening of untreated depressive symptoms. It is prudent to obtain liver function tests prior to naltrexone administration and again 3 to 4 weeks after starting the drug.²¹ Repeat testing should be performed at 2-to 4-week intervals for the next 2 months, then once a month for the following 3 months. After 6 months, testing three to four times a year is usually sufficient.

Nonsteroidal analgesics should not be used with high dosages of naltrexone (>50 mg/d), as concurrent use may increase the risk of hepatic transaminase elevation.²¹

Mood stabilizers Responses to lithium and valproate have been described in two case reports of patients with kleptomania.^14,15 In the case of valproate, the effective dosage was 2,000 mg/d, whereas lithium reduced stealing urges at a serum level of 0.5 mEq/L.

Although it would be premature to recommend the use of mood stabilizers, their possible benefit may be related to their efficacy in bipolar disorder treatment and the existence of features (e.g., impulsivity) shared by kleptomania and bipolar disorder.

Atypical antipsychotics Although there is no evidence that atypical antipsychotics are useful in kleptomania, augmenting an SSRI with an atypical neuroleptic may be beneficial. Atypical antipsychotics have been explored as augmenting agents in the treatment of nonpsychotic disorders and behaviors, including pathologic gambling and obsessive-compulsive disorder.

The role of psychotherapy

Cognitive-behavioral therapy Based on the evidence of its effectiveness in treating pathologic gambling, CBT may hold promise as monotherapy for mild cases of kleptomania.

Combination therapy Combined pharmacologic and behavioral therapy may be the optimal treatment strategy for kleptomania. In our experience, patients who respond only partially or fail to respond to pharmacotherapy alone are more likely to find relief with a combination of drug and cognitive-behavioral therapies.

Related resource

• Goldman MJ. Kleptomania: the compulsion to steal—what can be done? Far Hills, N New Horizon Press, 1998.
• Shoplifters Alternative http://www.shoplifters.org
• Impulse Control Disorders Clinic, University of Minnesota http://www.med.umn.edu/psychiatry/research/impulse.htm

Drug brand names

• Citalopram • Celexa
• Fluvoxamine • Luvox
• Imipramine • Tofranil
• Naltrexone • Revia
• Nortriptyline • Aventyl, Pamelor
• Paroxetine • Paxil
• Quetiapine • Seroquel
• Valproic acid • Depakote

Disclosure

The authors report no affiliation or financial arrangement with any of the companies whose products are mentioned in this article.

What is kleptomania? An independent illness, a symptom of other psychiatric disorders, or merely criminal behavior? Kleptomania—a disorder defined by an inability to resist the impulse to steal—is one of psychiatry’s most poorly understood diagnoses, even though it has been recognized in the literature for almost 200 years.

Kleptomania causes notable distress and impaired functioning.¹ People with kleptomania often suffer from comorbid mood, anxiety, substance use, and other impulse-control disorders.¹⁴ They experience the humiliation of repeated arrests, which leads to guilt, depression, and even suicide.^1,5 Yet kleptomania usually goes undiagnosed and untreated, despite a lifetime prevalence as high as 0.6%.⁶

Case report: ‘I’m a thief’

“I’m a thief,” began Susan, age 39. “I steal something four or five times every week. I steal from grocery stores and clothing stores. Sometimes I might steal something like vanilla extract; other times an expensive men’s tie. I probably steal $200 worth of items every week.

“You probably won’t believe this, but I don’t want or need the stuff I take. I have plenty of money. I have no idea why I take the things I do. That’s why I’m so depressed. What kind of person does something like this?

The urge to steal “I was probably 14 when I started stealing. I would go to stores with my mother. When I saw certain objects, I would get urges to steal them. The odd thing was that the items I stole were so ridiculous. I remember stealing key chains for several months, maybe three or four times a week. When I got older, things got worse. I was having urges more often, and so I needed to steal more often.

Box

“My entire life has been torment. Each day I worry about having the urges, and then I worry about being caught stealing. I can’t relax. I’ve been married for 17 years, and I haven’t told my husband. My secrecy is tearing our marriage apart. My husband thinks I’m having an affair because I’ve distanced myself emotionally from him.”

Table 1

SCREENING TEST FOR KLEPTOMANIA

Susan described urges to steal almost every day. When the urges were mild, she could resist them. Other days they were severe, and Susan felt unable to control her behavior. At work, her urges distracted her from completing projects, and her performance suffered. The urge to steal would often compel Susan to leave work early so she could get to a store.

Calm, then guilt “Every time I steal something I feel both a thrill and a great sense of calm,” she said. “It feels good. The problem is that almost immediately after each theft, I feel guilty and ashamed. After I steal, I usually donate the items to the Salvation Army, throw them away, or give them away as gifts.”

Drug trials We started treating Susan with the selective serotonin reuptake inhibitor (SSRI), citalopram. She reported notable improvement in her mood after 3 weeks on a dosage of 60 mg/d and she had been attending weekly psychotherapy, although her stealing continued unchanged. The addition of naltrexone, 200 mg/d for 2 weeks, decreased the frequency of Susan’s stealing and reduced her urges to steal, but her symptoms continued to interfere significantly with her overall functioning.

We then added the atypical antipsychotic quetiapine, 100 mg bid, and Susan’s urges to steal and stealing behavior went into remission within 3 weeks. She has refrained from stealing for the last 9 months.

Making the diagnosis

In our clinic, we have treated more than 50 patients with kleptomania. Rather than coming to us through the criminal justice system, they are usually self-referred. Often they contact us after discovering on the Internet that we specialize in treating persons with this disorder.

In our experience, kleptomania typically goes undiagnosed in clinical settings, in part because patients are ashamed and embarrassed to discuss their symptoms with physicians unless specifically asked.¹ If left untreated, however, kleptomania frequently becomes chronic.⁴ If persons with kleptomania are to seek treatment, it is important that family, friends, and mental health professionals understand the myths and facts about this disorder (Box).

To make the diagnosis, we use the simple screening instrument shown in Table 1.⁷ In general, because of high comorbidity with certain disorders, we screen every patient presenting to our clinic with a mood, substance use, anxiety, or eating disorder, or who has a problem with impulse control. Kleptomania is likely if the patient answers “yes” to questions 1 through 4 and to question 5 or 6. Stealing may be a symptom of several other psychiatric disorders, however, and misdiagnosis is fairly common (Table 2).^6-8

Data suggest that the female-to-male ratio in kleptomania is approximately 2:1, with onset in adolescence. Typical individuals with kleptomania steal because they have urges to steal, often triggered by specific stimuli such as the sights and sounds of stores or feelings of loneliness or stress.¹ Most patients with kleptomania are fairly specific about the types of stores from which they steal and the items they steal, and most hoard stolen items.¹

Although many patients with kleptomania function quite well, others are severely debilitated in social and occupational realms. In a series of 22 patients with kleptomania:

• 64% had been apprehended
• 23% had served jail time
• 27% had been hospitalized because of their kleptomania symptoms
• 18% had considered or attempted suicide because of the distress associated with their kleptomania.¹

Treatment recommendations

Patient history With patients who steal, we begin by identifying the motivation behind the stealing. Most patients with kleptomania report urges to steal. Some of these patients may have comorbid depression; for them, stealing makes them feel less depressed. Anger or irritability may point to borderline personality disorder. Stealing for the enjoyment of risk may suggest bipolar disorder.

Table 2

DIFFERENTIAL DIAGNOSIS: DISORDERS THAT MAY INVOLVE STEALING

Many patients with kleptomania have comorbid mood, substance, or anxiety disorders. Treating these other symptoms while ignoring the symptoms of kleptomania may be unsuccessful. Comorbidity also may influence the choice of medication.

Medical assessment Case reports have associated the onset of kleptomania with a variety of medical conditions, including presenile cortical atrophy in a 25-year-old, a parietal tumor that caused blackouts and obliterated any memory of stealing episodes, narcolepsy, and an insulinoma that caused severe hypoglycemia.⁹ The relationship of these conditions with the onset of kleptomania is unclear, but the reports suggest that medical causes—although unlikely—should be ruled out before you consider kleptomania as a psychiatric illness.

Patient education Persons with kleptomania often feel that no one else has the same problem. They do not think of their behavior as being an illness. It is helpful to explain that kleptomania is treatable and to connect patients with educational books, self-help groups, and Web sites providing information and support (see “Related resources”).

Cognitive-behavioral therapy (CBT) Although the evidence is quite limited, covert sensitization, exposure and response prevention, and imaginal desensitization have all been shown effective in case reports.¹⁰

What medications are effective?

Only case reports, a case series of five subjects, and a single open-label treatment study involving 10 subjects with kleptomania have been done.

So far, uses of tricyclic antidepressants (imipramine, nortriptyline), SSRIs (fluoxetine, fluvoxamine, paroxetine), the opioid antagonist naltrexone, and mood stabilizers (lithium, valproate) have met with varying degrees of success. Strategies targeting urge and behavior reduction and mechanisms for coping with urges and behavior (e.g., cognitive-behavioral therapies) may represent important adjunctive components.^2,11-17

No medications are FDA-approved for treating kleptomania. Therefore, it is important to inform patients of any off-label use of medications for this disorder, as well as the empirical basis for considering pharmacologic treatment.

SSRIs Only case reports exist on the use of SSRIs in treating kleptomania. The disorder may share a common pathology with pathologic gambling, and in our clinical experience appears to respond to similar treatments.¹⁸ We draw on research of pathologic gambling as well as our clinical experience in choosing SSRIs as first-line treatment, especially for patients with significant mood symptoms.¹⁹

We suggest titrating SSRIs to the maximum recommended dosage. As in the treatment of pathologic gambling, dosages of SSRIs required to treat kleptomania symptoms appear to be higher than average dosages required to treat depressive disorders. An SSRI should not be considered ineffective unless it has been tried for at least 10 to 12 weeks and the highest dosage tolerated or recommended by the manufacturer has been reached.

Response to SSRIs usually is characterized by decreased thoughts about stealing, decreased stealing behavior, and improvement in social and occupational functioning. If an SSRI is only partially effective, we consider augmentation with naltrexone, buspirone, or a mood stabilizer.

Naltrexone Patients taking naltrexone often report less-intense urges to steal. The urges may not disappear but are often sufficiently reduced so that the patient can resist them more easily. Patients also report that the thrill associated with stealing is reduced or eliminated.

Naltrexone was used in the first medication study of kleptomania and showed a significant decline in the intensity of urges to steal, stealing thoughts, and stealing behavior. Average dosage was 150 mg/d;¹¹ a reduced dosage (e.g., 50 mg/d) may work in adolescents with kleptomania.²⁰

Nausea as a side effect can be reduced by starting patients on 25 mg/d for the first 3 or 4 days and possibly adding ondansetron, 4 to 8 mg/d. Nausea and diarrhea are usually mild and resolve within the first week. Clinically, most patients respond to naltrexone within 2 weeks. After that, the dosage usually needs to be adjusted.

In patients with comorbid depression, augmentation with an SSRI may prevent worsening of untreated depressive symptoms. It is prudent to obtain liver function tests prior to naltrexone administration and again 3 to 4 weeks after starting the drug.²¹ Repeat testing should be performed at 2-to 4-week intervals for the next 2 months, then once a month for the following 3 months. After 6 months, testing three to four times a year is usually sufficient.

Nonsteroidal analgesics should not be used with high dosages of naltrexone (>50 mg/d), as concurrent use may increase the risk of hepatic transaminase elevation.²¹

Mood stabilizers Responses to lithium and valproate have been described in two case reports of patients with kleptomania.^14,15 In the case of valproate, the effective dosage was 2,000 mg/d, whereas lithium reduced stealing urges at a serum level of 0.5 mEq/L.

Although it would be premature to recommend the use of mood stabilizers, their possible benefit may be related to their efficacy in bipolar disorder treatment and the existence of features (e.g., impulsivity) shared by kleptomania and bipolar disorder.

Atypical antipsychotics Although there is no evidence that atypical antipsychotics are useful in kleptomania, augmenting an SSRI with an atypical neuroleptic may be beneficial. Atypical antipsychotics have been explored as augmenting agents in the treatment of nonpsychotic disorders and behaviors, including pathologic gambling and obsessive-compulsive disorder.

The role of psychotherapy

Cognitive-behavioral therapy Based on the evidence of its effectiveness in treating pathologic gambling, CBT may hold promise as monotherapy for mild cases of kleptomania.

Combination therapy Combined pharmacologic and behavioral therapy may be the optimal treatment strategy for kleptomania. In our experience, patients who respond only partially or fail to respond to pharmacotherapy alone are more likely to find relief with a combination of drug and cognitive-behavioral therapies.

Related resource

• Goldman MJ. Kleptomania: the compulsion to steal—what can be done? Far Hills, N New Horizon Press, 1998.
• Shoplifters Alternative http://www.shoplifters.org
• Impulse Control Disorders Clinic, University of Minnesota http://www.med.umn.edu/psychiatry/research/impulse.htm

Drug brand names

• Citalopram • Celexa
• Fluvoxamine • Luvox
• Imipramine • Tofranil
• Naltrexone • Revia
• Nortriptyline • Aventyl, Pamelor
• Paroxetine • Paxil
• Quetiapine • Seroquel
• Valproic acid • Depakote

Disclosure

The authors report no affiliation or financial arrangement with any of the companies whose products are mentioned in this article.

What is kleptomania? An independent illness, a symptom of other psychiatric disorders, or merely criminal behavior? Kleptomania—a disorder defined by an inability to resist the impulse to steal—is one of psychiatry’s most poorly understood diagnoses, even though it has been recognized in the literature for almost 200 years.

Kleptomania causes notable distress and impaired functioning.¹ People with kleptomania often suffer from comorbid mood, anxiety, substance use, and other impulse-control disorders.¹⁴ They experience the humiliation of repeated arrests, which leads to guilt, depression, and even suicide.^1,5 Yet kleptomania usually goes undiagnosed and untreated, despite a lifetime prevalence as high as 0.6%.⁶

Case report: ‘I’m a thief’

“I’m a thief,” began Susan, age 39. “I steal something four or five times every week. I steal from grocery stores and clothing stores. Sometimes I might steal something like vanilla extract; other times an expensive men’s tie. I probably steal $200 worth of items every week.

“You probably won’t believe this, but I don’t want or need the stuff I take. I have plenty of money. I have no idea why I take the things I do. That’s why I’m so depressed. What kind of person does something like this?

The urge to steal “I was probably 14 when I started stealing. I would go to stores with my mother. When I saw certain objects, I would get urges to steal them. The odd thing was that the items I stole were so ridiculous. I remember stealing key chains for several months, maybe three or four times a week. When I got older, things got worse. I was having urges more often, and so I needed to steal more often.

Box

“My entire life has been torment. Each day I worry about having the urges, and then I worry about being caught stealing. I can’t relax. I’ve been married for 17 years, and I haven’t told my husband. My secrecy is tearing our marriage apart. My husband thinks I’m having an affair because I’ve distanced myself emotionally from him.”

Table 1

SCREENING TEST FOR KLEPTOMANIA

Susan described urges to steal almost every day. When the urges were mild, she could resist them. Other days they were severe, and Susan felt unable to control her behavior. At work, her urges distracted her from completing projects, and her performance suffered. The urge to steal would often compel Susan to leave work early so she could get to a store.

Calm, then guilt “Every time I steal something I feel both a thrill and a great sense of calm,” she said. “It feels good. The problem is that almost immediately after each theft, I feel guilty and ashamed. After I steal, I usually donate the items to the Salvation Army, throw them away, or give them away as gifts.”

Drug trials We started treating Susan with the selective serotonin reuptake inhibitor (SSRI), citalopram. She reported notable improvement in her mood after 3 weeks on a dosage of 60 mg/d and she had been attending weekly psychotherapy, although her stealing continued unchanged. The addition of naltrexone, 200 mg/d for 2 weeks, decreased the frequency of Susan’s stealing and reduced her urges to steal, but her symptoms continued to interfere significantly with her overall functioning.

We then added the atypical antipsychotic quetiapine, 100 mg bid, and Susan’s urges to steal and stealing behavior went into remission within 3 weeks. She has refrained from stealing for the last 9 months.

Making the diagnosis

In our clinic, we have treated more than 50 patients with kleptomania. Rather than coming to us through the criminal justice system, they are usually self-referred. Often they contact us after discovering on the Internet that we specialize in treating persons with this disorder.

In our experience, kleptomania typically goes undiagnosed in clinical settings, in part because patients are ashamed and embarrassed to discuss their symptoms with physicians unless specifically asked.¹ If left untreated, however, kleptomania frequently becomes chronic.⁴ If persons with kleptomania are to seek treatment, it is important that family, friends, and mental health professionals understand the myths and facts about this disorder (Box).

To make the diagnosis, we use the simple screening instrument shown in Table 1.⁷ In general, because of high comorbidity with certain disorders, we screen every patient presenting to our clinic with a mood, substance use, anxiety, or eating disorder, or who has a problem with impulse control. Kleptomania is likely if the patient answers “yes” to questions 1 through 4 and to question 5 or 6. Stealing may be a symptom of several other psychiatric disorders, however, and misdiagnosis is fairly common (Table 2).^6-8

Data suggest that the female-to-male ratio in kleptomania is approximately 2:1, with onset in adolescence. Typical individuals with kleptomania steal because they have urges to steal, often triggered by specific stimuli such as the sights and sounds of stores or feelings of loneliness or stress.¹ Most patients with kleptomania are fairly specific about the types of stores from which they steal and the items they steal, and most hoard stolen items.¹

Although many patients with kleptomania function quite well, others are severely debilitated in social and occupational realms. In a series of 22 patients with kleptomania:

• 64% had been apprehended
• 23% had served jail time
• 27% had been hospitalized because of their kleptomania symptoms
• 18% had considered or attempted suicide because of the distress associated with their kleptomania.¹

Treatment recommendations

Patient history With patients who steal, we begin by identifying the motivation behind the stealing. Most patients with kleptomania report urges to steal. Some of these patients may have comorbid depression; for them, stealing makes them feel less depressed. Anger or irritability may point to borderline personality disorder. Stealing for the enjoyment of risk may suggest bipolar disorder.

Table 2

DIFFERENTIAL DIAGNOSIS: DISORDERS THAT MAY INVOLVE STEALING

Many patients with kleptomania have comorbid mood, substance, or anxiety disorders. Treating these other symptoms while ignoring the symptoms of kleptomania may be unsuccessful. Comorbidity also may influence the choice of medication.

Medical assessment Case reports have associated the onset of kleptomania with a variety of medical conditions, including presenile cortical atrophy in a 25-year-old, a parietal tumor that caused blackouts and obliterated any memory of stealing episodes, narcolepsy, and an insulinoma that caused severe hypoglycemia.⁹ The relationship of these conditions with the onset of kleptomania is unclear, but the reports suggest that medical causes—although unlikely—should be ruled out before you consider kleptomania as a psychiatric illness.

Patient education Persons with kleptomania often feel that no one else has the same problem. They do not think of their behavior as being an illness. It is helpful to explain that kleptomania is treatable and to connect patients with educational books, self-help groups, and Web sites providing information and support (see “Related resources”).

Cognitive-behavioral therapy (CBT) Although the evidence is quite limited, covert sensitization, exposure and response prevention, and imaginal desensitization have all been shown effective in case reports.¹⁰

What medications are effective?

Only case reports, a case series of five subjects, and a single open-label treatment study involving 10 subjects with kleptomania have been done.

So far, uses of tricyclic antidepressants (imipramine, nortriptyline), SSRIs (fluoxetine, fluvoxamine, paroxetine), the opioid antagonist naltrexone, and mood stabilizers (lithium, valproate) have met with varying degrees of success. Strategies targeting urge and behavior reduction and mechanisms for coping with urges and behavior (e.g., cognitive-behavioral therapies) may represent important adjunctive components.^2,11-17

No medications are FDA-approved for treating kleptomania. Therefore, it is important to inform patients of any off-label use of medications for this disorder, as well as the empirical basis for considering pharmacologic treatment.

SSRIs Only case reports exist on the use of SSRIs in treating kleptomania. The disorder may share a common pathology with pathologic gambling, and in our clinical experience appears to respond to similar treatments.¹⁸ We draw on research of pathologic gambling as well as our clinical experience in choosing SSRIs as first-line treatment, especially for patients with significant mood symptoms.¹⁹

We suggest titrating SSRIs to the maximum recommended dosage. As in the treatment of pathologic gambling, dosages of SSRIs required to treat kleptomania symptoms appear to be higher than average dosages required to treat depressive disorders. An SSRI should not be considered ineffective unless it has been tried for at least 10 to 12 weeks and the highest dosage tolerated or recommended by the manufacturer has been reached.

Response to SSRIs usually is characterized by decreased thoughts about stealing, decreased stealing behavior, and improvement in social and occupational functioning. If an SSRI is only partially effective, we consider augmentation with naltrexone, buspirone, or a mood stabilizer.

Naltrexone Patients taking naltrexone often report less-intense urges to steal. The urges may not disappear but are often sufficiently reduced so that the patient can resist them more easily. Patients also report that the thrill associated with stealing is reduced or eliminated.

Naltrexone was used in the first medication study of kleptomania and showed a significant decline in the intensity of urges to steal, stealing thoughts, and stealing behavior. Average dosage was 150 mg/d;¹¹ a reduced dosage (e.g., 50 mg/d) may work in adolescents with kleptomania.²⁰

Nausea as a side effect can be reduced by starting patients on 25 mg/d for the first 3 or 4 days and possibly adding ondansetron, 4 to 8 mg/d. Nausea and diarrhea are usually mild and resolve within the first week. Clinically, most patients respond to naltrexone within 2 weeks. After that, the dosage usually needs to be adjusted.

In patients with comorbid depression, augmentation with an SSRI may prevent worsening of untreated depressive symptoms. It is prudent to obtain liver function tests prior to naltrexone administration and again 3 to 4 weeks after starting the drug.²¹ Repeat testing should be performed at 2-to 4-week intervals for the next 2 months, then once a month for the following 3 months. After 6 months, testing three to four times a year is usually sufficient.

Nonsteroidal analgesics should not be used with high dosages of naltrexone (>50 mg/d), as concurrent use may increase the risk of hepatic transaminase elevation.²¹

Mood stabilizers Responses to lithium and valproate have been described in two case reports of patients with kleptomania.^14,15 In the case of valproate, the effective dosage was 2,000 mg/d, whereas lithium reduced stealing urges at a serum level of 0.5 mEq/L.

Although it would be premature to recommend the use of mood stabilizers, their possible benefit may be related to their efficacy in bipolar disorder treatment and the existence of features (e.g., impulsivity) shared by kleptomania and bipolar disorder.

Atypical antipsychotics Although there is no evidence that atypical antipsychotics are useful in kleptomania, augmenting an SSRI with an atypical neuroleptic may be beneficial. Atypical antipsychotics have been explored as augmenting agents in the treatment of nonpsychotic disorders and behaviors, including pathologic gambling and obsessive-compulsive disorder.

The role of psychotherapy

Cognitive-behavioral therapy Based on the evidence of its effectiveness in treating pathologic gambling, CBT may hold promise as monotherapy for mild cases of kleptomania.

Combination therapy Combined pharmacologic and behavioral therapy may be the optimal treatment strategy for kleptomania. In our experience, patients who respond only partially or fail to respond to pharmacotherapy alone are more likely to find relief with a combination of drug and cognitive-behavioral therapies.

Related resource

• Goldman MJ. Kleptomania: the compulsion to steal—what can be done? Far Hills, N New Horizon Press, 1998.
• Shoplifters Alternative http://www.shoplifters.org
• Impulse Control Disorders Clinic, University of Minnesota http://www.med.umn.edu/psychiatry/research/impulse.htm

Drug brand names

• Citalopram • Celexa
• Fluvoxamine • Luvox
• Imipramine • Tofranil
• Naltrexone • Revia
• Nortriptyline • Aventyl, Pamelor
• Paroxetine • Paxil
• Quetiapine • Seroquel
• Valproic acid • Depakote

Disclosure

The authors report no affiliation or financial arrangement with any of the companies whose products are mentioned in this article.

What is kleptomania? An independent illness, a symptom of other psychiatric disorders, or merely criminal behavior? Kleptomania—a disorder defined by an inability to resist the impulse to steal—is one of psychiatry’s most poorly understood diagnoses, even though it has been recognized in the literature for almost 200 years.

Kleptomania causes notable distress and impaired functioning.¹ People with kleptomania often suffer from comorbid mood, anxiety, substance use, and other impulse-control disorders.¹⁴ They experience the humiliation of repeated arrests, which leads to guilt, depression, and even suicide.^1,5 Yet kleptomania usually goes undiagnosed and untreated, despite a lifetime prevalence as high as 0.6%.⁶

Case report: ‘I’m a thief’

“I’m a thief,” began Susan, age 39. “I steal something four or five times every week. I steal from grocery stores and clothing stores. Sometimes I might steal something like vanilla extract; other times an expensive men’s tie. I probably steal $200 worth of items every week.

“You probably won’t believe this, but I don’t want or need the stuff I take. I have plenty of money. I have no idea why I take the things I do. That’s why I’m so depressed. What kind of person does something like this?

The urge to steal “I was probably 14 when I started stealing. I would go to stores with my mother. When I saw certain objects, I would get urges to steal them. The odd thing was that the items I stole were so ridiculous. I remember stealing key chains for several months, maybe three or four times a week. When I got older, things got worse. I was having urges more often, and so I needed to steal more often.

Box

“My entire life has been torment. Each day I worry about having the urges, and then I worry about being caught stealing. I can’t relax. I’ve been married for 17 years, and I haven’t told my husband. My secrecy is tearing our marriage apart. My husband thinks I’m having an affair because I’ve distanced myself emotionally from him.”

Table 1

SCREENING TEST FOR KLEPTOMANIA

Susan described urges to steal almost every day. When the urges were mild, she could resist them. Other days they were severe, and Susan felt unable to control her behavior. At work, her urges distracted her from completing projects, and her performance suffered. The urge to steal would often compel Susan to leave work early so she could get to a store.

Calm, then guilt “Every time I steal something I feel both a thrill and a great sense of calm,” she said. “It feels good. The problem is that almost immediately after each theft, I feel guilty and ashamed. After I steal, I usually donate the items to the Salvation Army, throw them away, or give them away as gifts.”

Drug trials We started treating Susan with the selective serotonin reuptake inhibitor (SSRI), citalopram. She reported notable improvement in her mood after 3 weeks on a dosage of 60 mg/d and she had been attending weekly psychotherapy, although her stealing continued unchanged. The addition of naltrexone, 200 mg/d for 2 weeks, decreased the frequency of Susan’s stealing and reduced her urges to steal, but her symptoms continued to interfere significantly with her overall functioning.

We then added the atypical antipsychotic quetiapine, 100 mg bid, and Susan’s urges to steal and stealing behavior went into remission within 3 weeks. She has refrained from stealing for the last 9 months.

Making the diagnosis

In our clinic, we have treated more than 50 patients with kleptomania. Rather than coming to us through the criminal justice system, they are usually self-referred. Often they contact us after discovering on the Internet that we specialize in treating persons with this disorder.

In our experience, kleptomania typically goes undiagnosed in clinical settings, in part because patients are ashamed and embarrassed to discuss their symptoms with physicians unless specifically asked.¹ If left untreated, however, kleptomania frequently becomes chronic.⁴ If persons with kleptomania are to seek treatment, it is important that family, friends, and mental health professionals understand the myths and facts about this disorder (Box).

To make the diagnosis, we use the simple screening instrument shown in Table 1.⁷ In general, because of high comorbidity with certain disorders, we screen every patient presenting to our clinic with a mood, substance use, anxiety, or eating disorder, or who has a problem with impulse control. Kleptomania is likely if the patient answers “yes” to questions 1 through 4 and to question 5 or 6. Stealing may be a symptom of several other psychiatric disorders, however, and misdiagnosis is fairly common (Table 2).^6-8

Data suggest that the female-to-male ratio in kleptomania is approximately 2:1, with onset in adolescence. Typical individuals with kleptomania steal because they have urges to steal, often triggered by specific stimuli such as the sights and sounds of stores or feelings of loneliness or stress.¹ Most patients with kleptomania are fairly specific about the types of stores from which they steal and the items they steal, and most hoard stolen items.¹

Although many patients with kleptomania function quite well, others are severely debilitated in social and occupational realms. In a series of 22 patients with kleptomania:

• 64% had been apprehended
• 23% had served jail time
• 27% had been hospitalized because of their kleptomania symptoms
• 18% had considered or attempted suicide because of the distress associated with their kleptomania.¹

Treatment recommendations

Patient history With patients who steal, we begin by identifying the motivation behind the stealing. Most patients with kleptomania report urges to steal. Some of these patients may have comorbid depression; for them, stealing makes them feel less depressed. Anger or irritability may point to borderline personality disorder. Stealing for the enjoyment of risk may suggest bipolar disorder.

Table 2

DIFFERENTIAL DIAGNOSIS: DISORDERS THAT MAY INVOLVE STEALING

Many patients with kleptomania have comorbid mood, substance, or anxiety disorders. Treating these other symptoms while ignoring the symptoms of kleptomania may be unsuccessful. Comorbidity also may influence the choice of medication.

Medical assessment Case reports have associated the onset of kleptomania with a variety of medical conditions, including presenile cortical atrophy in a 25-year-old, a parietal tumor that caused blackouts and obliterated any memory of stealing episodes, narcolepsy, and an insulinoma that caused severe hypoglycemia.⁹ The relationship of these conditions with the onset of kleptomania is unclear, but the reports suggest that medical causes—although unlikely—should be ruled out before you consider kleptomania as a psychiatric illness.

Patient education Persons with kleptomania often feel that no one else has the same problem. They do not think of their behavior as being an illness. It is helpful to explain that kleptomania is treatable and to connect patients with educational books, self-help groups, and Web sites providing information and support (see “Related resources”).

Cognitive-behavioral therapy (CBT) Although the evidence is quite limited, covert sensitization, exposure and response prevention, and imaginal desensitization have all been shown effective in case reports.¹⁰

What medications are effective?

Only case reports, a case series of five subjects, and a single open-label treatment study involving 10 subjects with kleptomania have been done.

So far, uses of tricyclic antidepressants (imipramine, nortriptyline), SSRIs (fluoxetine, fluvoxamine, paroxetine), the opioid antagonist naltrexone, and mood stabilizers (lithium, valproate) have met with varying degrees of success. Strategies targeting urge and behavior reduction and mechanisms for coping with urges and behavior (e.g., cognitive-behavioral therapies) may represent important adjunctive components.^2,11-17

No medications are FDA-approved for treating kleptomania. Therefore, it is important to inform patients of any off-label use of medications for this disorder, as well as the empirical basis for considering pharmacologic treatment.

SSRIs Only case reports exist on the use of SSRIs in treating kleptomania. The disorder may share a common pathology with pathologic gambling, and in our clinical experience appears to respond to similar treatments.¹⁸ We draw on research of pathologic gambling as well as our clinical experience in choosing SSRIs as first-line treatment, especially for patients with significant mood symptoms.¹⁹

We suggest titrating SSRIs to the maximum recommended dosage. As in the treatment of pathologic gambling, dosages of SSRIs required to treat kleptomania symptoms appear to be higher than average dosages required to treat depressive disorders. An SSRI should not be considered ineffective unless it has been tried for at least 10 to 12 weeks and the highest dosage tolerated or recommended by the manufacturer has been reached.

Response to SSRIs usually is characterized by decreased thoughts about stealing, decreased stealing behavior, and improvement in social and occupational functioning. If an SSRI is only partially effective, we consider augmentation with naltrexone, buspirone, or a mood stabilizer.

Naltrexone Patients taking naltrexone often report less-intense urges to steal. The urges may not disappear but are often sufficiently reduced so that the patient can resist them more easily. Patients also report that the thrill associated with stealing is reduced or eliminated.

Naltrexone was used in the first medication study of kleptomania and showed a significant decline in the intensity of urges to steal, stealing thoughts, and stealing behavior. Average dosage was 150 mg/d;¹¹ a reduced dosage (e.g., 50 mg/d) may work in adolescents with kleptomania.²⁰

Nausea as a side effect can be reduced by starting patients on 25 mg/d for the first 3 or 4 days and possibly adding ondansetron, 4 to 8 mg/d. Nausea and diarrhea are usually mild and resolve within the first week. Clinically, most patients respond to naltrexone within 2 weeks. After that, the dosage usually needs to be adjusted.

In patients with comorbid depression, augmentation with an SSRI may prevent worsening of untreated depressive symptoms. It is prudent to obtain liver function tests prior to naltrexone administration and again 3 to 4 weeks after starting the drug.²¹ Repeat testing should be performed at 2-to 4-week intervals for the next 2 months, then once a month for the following 3 months. After 6 months, testing three to four times a year is usually sufficient.

Nonsteroidal analgesics should not be used with high dosages of naltrexone (>50 mg/d), as concurrent use may increase the risk of hepatic transaminase elevation.²¹

Mood stabilizers Responses to lithium and valproate have been described in two case reports of patients with kleptomania.^14,15 In the case of valproate, the effective dosage was 2,000 mg/d, whereas lithium reduced stealing urges at a serum level of 0.5 mEq/L.

Although it would be premature to recommend the use of mood stabilizers, their possible benefit may be related to their efficacy in bipolar disorder treatment and the existence of features (e.g., impulsivity) shared by kleptomania and bipolar disorder.

Atypical antipsychotics Although there is no evidence that atypical antipsychotics are useful in kleptomania, augmenting an SSRI with an atypical neuroleptic may be beneficial. Atypical antipsychotics have been explored as augmenting agents in the treatment of nonpsychotic disorders and behaviors, including pathologic gambling and obsessive-compulsive disorder.

The role of psychotherapy

Cognitive-behavioral therapy Based on the evidence of its effectiveness in treating pathologic gambling, CBT may hold promise as monotherapy for mild cases of kleptomania.

Combination therapy Combined pharmacologic and behavioral therapy may be the optimal treatment strategy for kleptomania. In our experience, patients who respond only partially or fail to respond to pharmacotherapy alone are more likely to find relief with a combination of drug and cognitive-behavioral therapies.

Related resource

• Goldman MJ. Kleptomania: the compulsion to steal—what can be done? Far Hills, N New Horizon Press, 1998.
• Shoplifters Alternative http://www.shoplifters.org
• Impulse Control Disorders Clinic, University of Minnesota http://www.med.umn.edu/psychiatry/research/impulse.htm

Drug brand names

• Citalopram • Celexa
• Fluvoxamine • Luvox
• Imipramine • Tofranil
• Naltrexone • Revia
• Nortriptyline • Aventyl, Pamelor
• Paroxetine • Paxil
• Quetiapine • Seroquel
• Valproic acid • Depakote

Disclosure

The authors report no affiliation or financial arrangement with any of the companies whose products are mentioned in this article.

What is kleptomania? An independent illness, a symptom of other psychiatric disorders, or merely criminal behavior? Kleptomania—a disorder defined by an inability to resist the impulse to steal—is one of psychiatry’s most poorly understood diagnoses, even though it has been recognized in the literature for almost 200 years.

Kleptomania causes notable distress and impaired functioning.¹ People with kleptomania often suffer from comorbid mood, anxiety, substance use, and other impulse-control disorders.¹⁴ They experience the humiliation of repeated arrests, which leads to guilt, depression, and even suicide.^1,5 Yet kleptomania usually goes undiagnosed and untreated, despite a lifetime prevalence as high as 0.6%.⁶

Case report: ‘I’m a thief’

“I’m a thief,” began Susan, age 39. “I steal something four or five times every week. I steal from grocery stores and clothing stores. Sometimes I might steal something like vanilla extract; other times an expensive men’s tie. I probably steal $200 worth of items every week.

“You probably won’t believe this, but I don’t want or need the stuff I take. I have plenty of money. I have no idea why I take the things I do. That’s why I’m so depressed. What kind of person does something like this?

The urge to steal “I was probably 14 when I started stealing. I would go to stores with my mother. When I saw certain objects, I would get urges to steal them. The odd thing was that the items I stole were so ridiculous. I remember stealing key chains for several months, maybe three or four times a week. When I got older, things got worse. I was having urges more often, and so I needed to steal more often.

Box

“My entire life has been torment. Each day I worry about having the urges, and then I worry about being caught stealing. I can’t relax. I’ve been married for 17 years, and I haven’t told my husband. My secrecy is tearing our marriage apart. My husband thinks I’m having an affair because I’ve distanced myself emotionally from him.”

Table 1

SCREENING TEST FOR KLEPTOMANIA

Susan described urges to steal almost every day. When the urges were mild, she could resist them. Other days they were severe, and Susan felt unable to control her behavior. At work, her urges distracted her from completing projects, and her performance suffered. The urge to steal would often compel Susan to leave work early so she could get to a store.

Calm, then guilt “Every time I steal something I feel both a thrill and a great sense of calm,” she said. “It feels good. The problem is that almost immediately after each theft, I feel guilty and ashamed. After I steal, I usually donate the items to the Salvation Army, throw them away, or give them away as gifts.”

Drug trials We started treating Susan with the selective serotonin reuptake inhibitor (SSRI), citalopram. She reported notable improvement in her mood after 3 weeks on a dosage of 60 mg/d and she had been attending weekly psychotherapy, although her stealing continued unchanged. The addition of naltrexone, 200 mg/d for 2 weeks, decreased the frequency of Susan’s stealing and reduced her urges to steal, but her symptoms continued to interfere significantly with her overall functioning.

We then added the atypical antipsychotic quetiapine, 100 mg bid, and Susan’s urges to steal and stealing behavior went into remission within 3 weeks. She has refrained from stealing for the last 9 months.

Making the diagnosis

In our clinic, we have treated more than 50 patients with kleptomania. Rather than coming to us through the criminal justice system, they are usually self-referred. Often they contact us after discovering on the Internet that we specialize in treating persons with this disorder.

In our experience, kleptomania typically goes undiagnosed in clinical settings, in part because patients are ashamed and embarrassed to discuss their symptoms with physicians unless specifically asked.¹ If left untreated, however, kleptomania frequently becomes chronic.⁴ If persons with kleptomania are to seek treatment, it is important that family, friends, and mental health professionals understand the myths and facts about this disorder (Box).

To make the diagnosis, we use the simple screening instrument shown in Table 1.⁷ In general, because of high comorbidity with certain disorders, we screen every patient presenting to our clinic with a mood, substance use, anxiety, or eating disorder, or who has a problem with impulse control. Kleptomania is likely if the patient answers “yes” to questions 1 through 4 and to question 5 or 6. Stealing may be a symptom of several other psychiatric disorders, however, and misdiagnosis is fairly common (Table 2).^6-8

Data suggest that the female-to-male ratio in kleptomania is approximately 2:1, with onset in adolescence. Typical individuals with kleptomania steal because they have urges to steal, often triggered by specific stimuli such as the sights and sounds of stores or feelings of loneliness or stress.¹ Most patients with kleptomania are fairly specific about the types of stores from which they steal and the items they steal, and most hoard stolen items.¹

Although many patients with kleptomania function quite well, others are severely debilitated in social and occupational realms. In a series of 22 patients with kleptomania:

• 64% had been apprehended
• 23% had served jail time
• 27% had been hospitalized because of their kleptomania symptoms
• 18% had considered or attempted suicide because of the distress associated with their kleptomania.¹

Treatment recommendations

Patient history With patients who steal, we begin by identifying the motivation behind the stealing. Most patients with kleptomania report urges to steal. Some of these patients may have comorbid depression; for them, stealing makes them feel less depressed. Anger or irritability may point to borderline personality disorder. Stealing for the enjoyment of risk may suggest bipolar disorder.

Table 2

DIFFERENTIAL DIAGNOSIS: DISORDERS THAT MAY INVOLVE STEALING

Many patients with kleptomania have comorbid mood, substance, or anxiety disorders. Treating these other symptoms while ignoring the symptoms of kleptomania may be unsuccessful. Comorbidity also may influence the choice of medication.

Medical assessment Case reports have associated the onset of kleptomania with a variety of medical conditions, including presenile cortical atrophy in a 25-year-old, a parietal tumor that caused blackouts and obliterated any memory of stealing episodes, narcolepsy, and an insulinoma that caused severe hypoglycemia.⁹ The relationship of these conditions with the onset of kleptomania is unclear, but the reports suggest that medical causes—although unlikely—should be ruled out before you consider kleptomania as a psychiatric illness.

Patient education Persons with kleptomania often feel that no one else has the same problem. They do not think of their behavior as being an illness. It is helpful to explain that kleptomania is treatable and to connect patients with educational books, self-help groups, and Web sites providing information and support (see “Related resources”).

Cognitive-behavioral therapy (CBT) Although the evidence is quite limited, covert sensitization, exposure and response prevention, and imaginal desensitization have all been shown effective in case reports.¹⁰

What medications are effective?

Only case reports, a case series of five subjects, and a single open-label treatment study involving 10 subjects with kleptomania have been done.

So far, uses of tricyclic antidepressants (imipramine, nortriptyline), SSRIs (fluoxetine, fluvoxamine, paroxetine), the opioid antagonist naltrexone, and mood stabilizers (lithium, valproate) have met with varying degrees of success. Strategies targeting urge and behavior reduction and mechanisms for coping with urges and behavior (e.g., cognitive-behavioral therapies) may represent important adjunctive components.^2,11-17

No medications are FDA-approved for treating kleptomania. Therefore, it is important to inform patients of any off-label use of medications for this disorder, as well as the empirical basis for considering pharmacologic treatment.

SSRIs Only case reports exist on the use of SSRIs in treating kleptomania. The disorder may share a common pathology with pathologic gambling, and in our clinical experience appears to respond to similar treatments.¹⁸ We draw on research of pathologic gambling as well as our clinical experience in choosing SSRIs as first-line treatment, especially for patients with significant mood symptoms.¹⁹

We suggest titrating SSRIs to the maximum recommended dosage. As in the treatment of pathologic gambling, dosages of SSRIs required to treat kleptomania symptoms appear to be higher than average dosages required to treat depressive disorders. An SSRI should not be considered ineffective unless it has been tried for at least 10 to 12 weeks and the highest dosage tolerated or recommended by the manufacturer has been reached.

Response to SSRIs usually is characterized by decreased thoughts about stealing, decreased stealing behavior, and improvement in social and occupational functioning. If an SSRI is only partially effective, we consider augmentation with naltrexone, buspirone, or a mood stabilizer.

Naltrexone Patients taking naltrexone often report less-intense urges to steal. The urges may not disappear but are often sufficiently reduced so that the patient can resist them more easily. Patients also report that the thrill associated with stealing is reduced or eliminated.

Naltrexone was used in the first medication study of kleptomania and showed a significant decline in the intensity of urges to steal, stealing thoughts, and stealing behavior. Average dosage was 150 mg/d;¹¹ a reduced dosage (e.g., 50 mg/d) may work in adolescents with kleptomania.²⁰

Nausea as a side effect can be reduced by starting patients on 25 mg/d for the first 3 or 4 days and possibly adding ondansetron, 4 to 8 mg/d. Nausea and diarrhea are usually mild and resolve within the first week. Clinically, most patients respond to naltrexone within 2 weeks. After that, the dosage usually needs to be adjusted.

In patients with comorbid depression, augmentation with an SSRI may prevent worsening of untreated depressive symptoms. It is prudent to obtain liver function tests prior to naltrexone administration and again 3 to 4 weeks after starting the drug.²¹ Repeat testing should be performed at 2-to 4-week intervals for the next 2 months, then once a month for the following 3 months. After 6 months, testing three to four times a year is usually sufficient.

Nonsteroidal analgesics should not be used with high dosages of naltrexone (>50 mg/d), as concurrent use may increase the risk of hepatic transaminase elevation.²¹

Mood stabilizers Responses to lithium and valproate have been described in two case reports of patients with kleptomania.^14,15 In the case of valproate, the effective dosage was 2,000 mg/d, whereas lithium reduced stealing urges at a serum level of 0.5 mEq/L.

Although it would be premature to recommend the use of mood stabilizers, their possible benefit may be related to their efficacy in bipolar disorder treatment and the existence of features (e.g., impulsivity) shared by kleptomania and bipolar disorder.

Atypical antipsychotics Although there is no evidence that atypical antipsychotics are useful in kleptomania, augmenting an SSRI with an atypical neuroleptic may be beneficial. Atypical antipsychotics have been explored as augmenting agents in the treatment of nonpsychotic disorders and behaviors, including pathologic gambling and obsessive-compulsive disorder.

The role of psychotherapy

Cognitive-behavioral therapy Based on the evidence of its effectiveness in treating pathologic gambling, CBT may hold promise as monotherapy for mild cases of kleptomania.

Combination therapy Combined pharmacologic and behavioral therapy may be the optimal treatment strategy for kleptomania. In our experience, patients who respond only partially or fail to respond to pharmacotherapy alone are more likely to find relief with a combination of drug and cognitive-behavioral therapies.

Related resource

• Goldman MJ. Kleptomania: the compulsion to steal—what can be done? Far Hills, N New Horizon Press, 1998.
• Shoplifters Alternative http://www.shoplifters.org
• Impulse Control Disorders Clinic, University of Minnesota http://www.med.umn.edu/psychiatry/research/impulse.htm

Drug brand names

• Citalopram • Celexa
• Fluvoxamine • Luvox
• Imipramine • Tofranil
• Naltrexone • Revia
• Nortriptyline • Aventyl, Pamelor
• Paroxetine • Paxil
• Quetiapine • Seroquel
• Valproic acid • Depakote

Disclosure

The authors report no affiliation or financial arrangement with any of the companies whose products are mentioned in this article.

What is kleptomania? An independent illness, a symptom of other psychiatric disorders, or merely criminal behavior? Kleptomania—a disorder defined by an inability to resist the impulse to steal—is one of psychiatry’s most poorly understood diagnoses, even though it has been recognized in the literature for almost 200 years.

Kleptomania causes notable distress and impaired functioning.¹ People with kleptomania often suffer from comorbid mood, anxiety, substance use, and other impulse-control disorders.¹⁴ They experience the humiliation of repeated arrests, which leads to guilt, depression, and even suicide.^1,5 Yet kleptomania usually goes undiagnosed and untreated, despite a lifetime prevalence as high as 0.6%.⁶

Case report: ‘I’m a thief’

“I’m a thief,” began Susan, age 39. “I steal something four or five times every week. I steal from grocery stores and clothing stores. Sometimes I might steal something like vanilla extract; other times an expensive men’s tie. I probably steal $200 worth of items every week.

“You probably won’t believe this, but I don’t want or need the stuff I take. I have plenty of money. I have no idea why I take the things I do. That’s why I’m so depressed. What kind of person does something like this?

The urge to steal “I was probably 14 when I started stealing. I would go to stores with my mother. When I saw certain objects, I would get urges to steal them. The odd thing was that the items I stole were so ridiculous. I remember stealing key chains for several months, maybe three or four times a week. When I got older, things got worse. I was having urges more often, and so I needed to steal more often.

Box

“My entire life has been torment. Each day I worry about having the urges, and then I worry about being caught stealing. I can’t relax. I’ve been married for 17 years, and I haven’t told my husband. My secrecy is tearing our marriage apart. My husband thinks I’m having an affair because I’ve distanced myself emotionally from him.”

Table 1

SCREENING TEST FOR KLEPTOMANIA

Susan described urges to steal almost every day. When the urges were mild, she could resist them. Other days they were severe, and Susan felt unable to control her behavior. At work, her urges distracted her from completing projects, and her performance suffered. The urge to steal would often compel Susan to leave work early so she could get to a store.

Calm, then guilt “Every time I steal something I feel both a thrill and a great sense of calm,” she said. “It feels good. The problem is that almost immediately after each theft, I feel guilty and ashamed. After I steal, I usually donate the items to the Salvation Army, throw them away, or give them away as gifts.”

Drug trials We started treating Susan with the selective serotonin reuptake inhibitor (SSRI), citalopram. She reported notable improvement in her mood after 3 weeks on a dosage of 60 mg/d and she had been attending weekly psychotherapy, although her stealing continued unchanged. The addition of naltrexone, 200 mg/d for 2 weeks, decreased the frequency of Susan’s stealing and reduced her urges to steal, but her symptoms continued to interfere significantly with her overall functioning.

We then added the atypical antipsychotic quetiapine, 100 mg bid, and Susan’s urges to steal and stealing behavior went into remission within 3 weeks. She has refrained from stealing for the last 9 months.

Making the diagnosis

In our clinic, we have treated more than 50 patients with kleptomania. Rather than coming to us through the criminal justice system, they are usually self-referred. Often they contact us after discovering on the Internet that we specialize in treating persons with this disorder.

In our experience, kleptomania typically goes undiagnosed in clinical settings, in part because patients are ashamed and embarrassed to discuss their symptoms with physicians unless specifically asked.¹ If left untreated, however, kleptomania frequently becomes chronic.⁴ If persons with kleptomania are to seek treatment, it is important that family, friends, and mental health professionals understand the myths and facts about this disorder (Box).

To make the diagnosis, we use the simple screening instrument shown in Table 1.⁷ In general, because of high comorbidity with certain disorders, we screen every patient presenting to our clinic with a mood, substance use, anxiety, or eating disorder, or who has a problem with impulse control. Kleptomania is likely if the patient answers “yes” to questions 1 through 4 and to question 5 or 6. Stealing may be a symptom of several other psychiatric disorders, however, and misdiagnosis is fairly common (Table 2).^6-8

Data suggest that the female-to-male ratio in kleptomania is approximately 2:1, with onset in adolescence. Typical individuals with kleptomania steal because they have urges to steal, often triggered by specific stimuli such as the sights and sounds of stores or feelings of loneliness or stress.¹ Most patients with kleptomania are fairly specific about the types of stores from which they steal and the items they steal, and most hoard stolen items.¹

Although many patients with kleptomania function quite well, others are severely debilitated in social and occupational realms. In a series of 22 patients with kleptomania:

• 64% had been apprehended
• 23% had served jail time
• 27% had been hospitalized because of their kleptomania symptoms
• 18% had considered or attempted suicide because of the distress associated with their kleptomania.¹

Treatment recommendations

Patient history With patients who steal, we begin by identifying the motivation behind the stealing. Most patients with kleptomania report urges to steal. Some of these patients may have comorbid depression; for them, stealing makes them feel less depressed. Anger or irritability may point to borderline personality disorder. Stealing for the enjoyment of risk may suggest bipolar disorder.

Table 2

DIFFERENTIAL DIAGNOSIS: DISORDERS THAT MAY INVOLVE STEALING

Many patients with kleptomania have comorbid mood, substance, or anxiety disorders. Treating these other symptoms while ignoring the symptoms of kleptomania may be unsuccessful. Comorbidity also may influence the choice of medication.

Medical assessment Case reports have associated the onset of kleptomania with a variety of medical conditions, including presenile cortical atrophy in a 25-year-old, a parietal tumor that caused blackouts and obliterated any memory of stealing episodes, narcolepsy, and an insulinoma that caused severe hypoglycemia.⁹ The relationship of these conditions with the onset of kleptomania is unclear, but the reports suggest that medical causes—although unlikely—should be ruled out before you consider kleptomania as a psychiatric illness.

Patient education Persons with kleptomania often feel that no one else has the same problem. They do not think of their behavior as being an illness. It is helpful to explain that kleptomania is treatable and to connect patients with educational books, self-help groups, and Web sites providing information and support (see “Related resources”).

Cognitive-behavioral therapy (CBT) Although the evidence is quite limited, covert sensitization, exposure and response prevention, and imaginal desensitization have all been shown effective in case reports.¹⁰

What medications are effective?

Only case reports, a case series of five subjects, and a single open-label treatment study involving 10 subjects with kleptomania have been done.

So far, uses of tricyclic antidepressants (imipramine, nortriptyline), SSRIs (fluoxetine, fluvoxamine, paroxetine), the opioid antagonist naltrexone, and mood stabilizers (lithium, valproate) have met with varying degrees of success. Strategies targeting urge and behavior reduction and mechanisms for coping with urges and behavior (e.g., cognitive-behavioral therapies) may represent important adjunctive components.^2,11-17

No medications are FDA-approved for treating kleptomania. Therefore, it is important to inform patients of any off-label use of medications for this disorder, as well as the empirical basis for considering pharmacologic treatment.

SSRIs Only case reports exist on the use of SSRIs in treating kleptomania. The disorder may share a common pathology with pathologic gambling, and in our clinical experience appears to respond to similar treatments.¹⁸ We draw on research of pathologic gambling as well as our clinical experience in choosing SSRIs as first-line treatment, especially for patients with significant mood symptoms.¹⁹

We suggest titrating SSRIs to the maximum recommended dosage. As in the treatment of pathologic gambling, dosages of SSRIs required to treat kleptomania symptoms appear to be higher than average dosages required to treat depressive disorders. An SSRI should not be considered ineffective unless it has been tried for at least 10 to 12 weeks and the highest dosage tolerated or recommended by the manufacturer has been reached.

Response to SSRIs usually is characterized by decreased thoughts about stealing, decreased stealing behavior, and improvement in social and occupational functioning. If an SSRI is only partially effective, we consider augmentation with naltrexone, buspirone, or a mood stabilizer.

Naltrexone Patients taking naltrexone often report less-intense urges to steal. The urges may not disappear but are often sufficiently reduced so that the patient can resist them more easily. Patients also report that the thrill associated with stealing is reduced or eliminated.

Naltrexone was used in the first medication study of kleptomania and showed a significant decline in the intensity of urges to steal, stealing thoughts, and stealing behavior. Average dosage was 150 mg/d;¹¹ a reduced dosage (e.g., 50 mg/d) may work in adolescents with kleptomania.²⁰

Nausea as a side effect can be reduced by starting patients on 25 mg/d for the first 3 or 4 days and possibly adding ondansetron, 4 to 8 mg/d. Nausea and diarrhea are usually mild and resolve within the first week. Clinically, most patients respond to naltrexone within 2 weeks. After that, the dosage usually needs to be adjusted.

In patients with comorbid depression, augmentation with an SSRI may prevent worsening of untreated depressive symptoms. It is prudent to obtain liver function tests prior to naltrexone administration and again 3 to 4 weeks after starting the drug.²¹ Repeat testing should be performed at 2-to 4-week intervals for the next 2 months, then once a month for the following 3 months. After 6 months, testing three to four times a year is usually sufficient.

Nonsteroidal analgesics should not be used with high dosages of naltrexone (>50 mg/d), as concurrent use may increase the risk of hepatic transaminase elevation.²¹

Mood stabilizers Responses to lithium and valproate have been described in two case reports of patients with kleptomania.^14,15 In the case of valproate, the effective dosage was 2,000 mg/d, whereas lithium reduced stealing urges at a serum level of 0.5 mEq/L.

Although it would be premature to recommend the use of mood stabilizers, their possible benefit may be related to their efficacy in bipolar disorder treatment and the existence of features (e.g., impulsivity) shared by kleptomania and bipolar disorder.

Atypical antipsychotics Although there is no evidence that atypical antipsychotics are useful in kleptomania, augmenting an SSRI with an atypical neuroleptic may be beneficial. Atypical antipsychotics have been explored as augmenting agents in the treatment of nonpsychotic disorders and behaviors, including pathologic gambling and obsessive-compulsive disorder.

The role of psychotherapy

Cognitive-behavioral therapy Based on the evidence of its effectiveness in treating pathologic gambling, CBT may hold promise as monotherapy for mild cases of kleptomania.

Combination therapy Combined pharmacologic and behavioral therapy may be the optimal treatment strategy for kleptomania. In our experience, patients who respond only partially or fail to respond to pharmacotherapy alone are more likely to find relief with a combination of drug and cognitive-behavioral therapies.

Related resource

• Goldman MJ. Kleptomania: the compulsion to steal—what can be done? Far Hills, N New Horizon Press, 1998.
• Shoplifters Alternative http://www.shoplifters.org
• Impulse Control Disorders Clinic, University of Minnesota http://www.med.umn.edu/psychiatry/research/impulse.htm

Drug brand names

• Citalopram • Celexa
• Fluvoxamine • Luvox
• Imipramine • Tofranil
• Naltrexone • Revia
• Nortriptyline • Aventyl, Pamelor
• Paroxetine • Paxil
• Quetiapine • Seroquel
• Valproic acid • Depakote

Disclosure

The authors report no affiliation or financial arrangement with any of the companies whose products are mentioned in this article.