 |
 |
|
|
|
Individuals with borderline personality disorder (BPD) frequently present to hospital emergency services with suicidal ideation or behaviour (1). Although threats of suicide outnumber actual suicide attempts, these patients are problematic for mental health professionals because the risk of suicide is high, with rates approaching those found in other major psychiatric disorders. Davis and colleagues estimated that patients with BPD have about a 10% lifetime risk of suicide (2). Clinicians are often troubled when hospitalizing patients with BPD, not knowing whether this course of action will benefit or worsen the patient’s crisis. Clinical lore has long suggested that patients with BPD regress during hospitalization, and an inpatient stay may be associated with disruptive behaviour, including self-harm behaviours, aggression, and noncompliance with ward routine (3). Little empirical evidence is available to guide the clinician in assessing whether hospitalization will improve or worsen these behaviours. Two recent studies provide some empirical evidence on this issue. Faulkner and colleagues compared hospitalized adolescents with and without BPD regarding the occurrence of restraints, seclusions, incidents of self-abuse and aggression, incidents of signing the intent-to-leave form, nonroutine drug and alcohol screens, and discharge against medical advice (4). The charts of 81 former inpatients between the ages of 13 and 17 years were reviewed; patients with BPD were restrained more often, had more time-outs, demonstrated more aggression, and had more nonroutine drug and alcohol screens ordered than did patients without BPD. However, when the analyses were repeated covarying for the length of stay, the 2 groups were no longer significantly different. The study concluded that behaviours such as need for restraint, aggression, and time-outs may be most influenced by increased length of stay. The authors suggest that a subgroup of patients with BPD might have stormy hospitalizations and therefore create a negative reputation for all BPD patients. They acknowledged that further research was needed to subtype BPD patients regarding their likelihood to regress in hospital. More recently, Soliman and Reza studied risk factors for violence committed by patients during an acute hospitalization (5). The clinical characteristics of 49 violent patients were retrospectively compared with all patients admitted to the same unit (n = 474) and with a random sample of nonviolent patients admitted to the same service (n = 140). Using a stepwise logistic regression analysis, the researchers found the following variables to be significant predictors of violence: medication changes during admission, use of medication as needed, past violent behaviour, personality disorders (specifically, antisocial disorder and BPD), and length of hospital stay. The authors point out the need to identify factors that predict violent behaviour during an inpatient admission. They suggest that medication changes, including regular and as-needed medications, might be predictive of patients at high risk for violent behaviour. These studies indicate the need for research clarifying whether patients with BPD are at increased risk for disruptive behaviour during an inpatient stay and whether certain clinical characteristics define a subgroup of patients at risk. This study examines 2 issues: first, whether the diagnosis of BPD or borderline traits differentiates adult patients who demonstrate disruptive behaviour during an inpatient hospitalization from those patients who do not; and second, whether other clinical variables can be assessed during the emergency visit to differentiate patients who are likely to demonstrate disruptive behaviour in hospital from those patients who are not. MethodsStudy Design and Data Collection The study was designed as a case–control, chart-based, retrospective analysis. We reviewed the charts of all psychiatric patients admitted to an urban general hospital inpatient service in Toronto, Ontario, for a 6-month period (from January 2000 through June 2000). The inpatient psychiatric unit is a 36-bed, acute care facility within a general hospital. More than 90% of patient admissions are from the hospital’s emergency room. Charts were examined for evidence of disruptive behaviour in the comprehensive nursing record, and caseness was defined by the presence of the behaviours listed below during the patient’s stay. We matched the subjects with control inpatients from the same time period who did not exhibit disruptive behaviours. Of 259 inpatients during the study period, 154 met exclusion criteria. Of the remaining 105 in- patients, 44 fulfilled criteria for caseness, leaving a total of 61 control subjects. For patients who had multiple admissions during this time period, only the most recent inpatient stay was considered. Disruptive behaviours were categorized as follows: suicide attempt; self-harm behaviour (that is, superficial wrist cutting, other self-inflicted wounds, and some minimal overdoses perceived to be of low lethality); violence, aggression, or vandalism; negative behaviours such as refusal to eat or take medications; violation of contracts to safety or abstinence; absence without leave or escape of a patient who was bound by contract or involuntary status; general disturbance to copatients or staff; and verbal abuse of copatients or staff. Exclusion criteria included a diagnosis of schizophrenia, psychotic disorder, delirium, or dementia or a diagnosis receiving a psychotic specifier, such as “major depressive disorder, psychotic.” We excluded patients with these diagnoses in an attempt to capture disruptive behaviours that had volitional underpinnings and that were not the manifestation of a psychotic disorder or delusional belief. Diagnoses represented in the remaining sample include Axis I mood, anxiety, eating, adjustment, and substance-abuse disorders, as well as Axis II personality disorders and a full range of Axis III and IV problems. Cases were determined by a single rater who was not blinded to admission diagnosis. On a subsample of charts with a blinded second rater, a test of interrater reliability for assigning caseness yielded a concordance of 87.5% (6 = 0.87, n = 8). Additional information derived from the charts included demographic (that is, age, sex, housing, and employment status), clinical (multiaxial assessment and diagnoses), and historical data. Multiaxial assessment and diagnoses were classified using DSM-IV and were derived from the emergency assessment completed by the emergency psychiatry team on the patient’s admission to hospital. Because this study focuses on predicting disruptive behaviour during a hospitalization based on information gathered during a psychiatric emergency intake interview, we did not consider discharge diagnoses. The Brown–Goodwin assessment for Lifetime History of Aggression (LHA) (6) was used to review patient charts for aggressive events; a total of 0 events was assigned an LHA score of 0, 1 event was assigned a score of 1, 2 to 4 events were assigned a score of 2, 5 to 10 events were assigned a score of 3, 10 or more events were assigned a score of 4, and “too many events to count” were assigned a score of 5. Scoring was modified to facilitate the inclusion of patients with “too many events to count.” This study was approved by the Research Ethics Board of St Michael’s Hospital in Toronto. Data Analysis Data were coded and entered into a relational Microsoft Access database and analyses were conducted using SPSS (7). We compared dependent measures between cases and control subjects using chi-square analyses. Significant variables were entered into a forward stepwise logistic regression analysis, predicting the contribution of specific variables to the presence or absence of disruptive behaviour during psychiatric inpatient hospitalization. Odds ratios (ORs) of significant contributors are reported. ResultsOver the study period, 44 patients were identified as meeting criteria for disruptive behaviour during a psychiatric inpatient hospitalization. Clinical records indicate that these 44 in- patients accounted for 148 disruptive events in the hospital, with an average of 3.05 events per person (SD 2.51). Table 1 presents the proportion of patients demonstrating each type of disruptive behaviour. The most frequent types identified include creating a general disturbance on the ward, leaving hospital against medical advice, and being verbally aggressive.
The characteristics we chose to differentiate inpatients with disruptive behaviour from those without were items that can be assessed in a clinical encounter during an emergency room visit. These include age, sex, aspects of personal history, psychiatric diagnosis, and problems precipitating the emergency room visit. There were more men in the group with disruptive behaviour (n = 29, or 66%) than without disruptive behaviour (n = 31, or 50.8%); however, this difference was not statistically significant (P2 = 2.38, df 1, ns). The inpatients with disruptive behaviour had a mean age of 38.64 years (SD 9.85) vs 41.79 years (SD 13.95) for the comparison group (F1,103 = 1.65, ns). Table 2 displays the differences between the inpatients with disruptive behaviour and those without disruptive behaviour, based on aspects of personal history. Patients with disruptive behaviour during their inpatient admission were significantly differentiated from patients without disruptive behaviour, based on the following characteristics: history of suicidal behaviour (P2 = 10.00, df 1, P < 0.001), history of previous psychiatric hospitalizations (P2 = 5.20, df 1, P < 0.05), and LHA score over 1 (c2 = 5.30, df 1, P < 0.05). There was a nonsignificant trend toward reporting a history of having been abused (P2 = 3.67, df 1, P = 0.06). We compared both groups’ diagnoses as recorded at the time of hospital admission. The groups did not differ based on the presence of an Axis I mood disorder diagnosis (84.1% of the disruptive patients vs 80.0% of the comparison group; P2 = 0.19, df 1, ns) or on the presence of substance abuse or dependence diagnoses (54.5% of the disruptive patients vs 49.2% of the comparison group; P2 = 0.30, df 1, ns). However, the patients with disruptive behaviour reported abusing a greater number of substances (mean 1.40, SD 1.64, median 1, range 0 5) than did the comparison patients (mean 0.85, SD 1.15; median 0; range 0 to 5; F1,102 = 3.95, P < 0.05). In terms of Axis II disorders based on the admission diagnoses, 20.5% (n = 9) of the patients with disruptive behaviour were diagnosed with BPD, compared with 8.2% (n = 5) of patients without disruptive behaviour; this difference was not statistically significant (P2 = 3.38, df 1, ns). When diagnoses of BPD or borderline traits were considered, 32.6% (n = 14) of patients with disruptive behaviour were diagnosed with BPD or traits, compared with 13.4% (n = 9) of patients without disruptive behaviour (>P2 = 4.45, df 1, P < 0.05). The 2 groups did not differ significantly regarding the occurrence of antisocial personality disorder or antisocial traits (11.6% vs 8.3%, P2 = 0.31, df 1, ns). The participants with disruptive behaviour did not differ significantly in the presence of an Axis III disorder on admission, compared with those without; however, all patients frequently had coexisting general medical conditions on Axis III. About 54% of both patients with disruptive behaviour (n = 24) and patients without disruptive behaviour (n = 33) were recorded as having current general medical conditions on admission to hospital. However, patients with disruptive behaviour were diagnosed as having more infectious diseases on admission, compared with patients without disruptive behaviour (23.3% vs 6.8%, P2 = 5.70, df 1, P < 0.05). Table 3 compares the 2 groups, based on the reported Axis IV assessment, regarding the presence of psychosocial and environmental problems at the time of admission to hospital. Patients without disruptive behaviour reported significantly more problems in their primary intimate relationships than did patients with disruptive behaviours (P2 = 11.06, df 1, P < 0.001). This difference may be explained by the fact that patients with disruptive behaviours were significantly less likely to report the presence of a primary support network (that is, parents, a partner, or siblings) than patients without disruptive behaviours (47.7% vs 28.3%; P2 = 4.11, df 1, P < 0.05). Similarly, patients without disruptive behaviour reported problems with work or employment more frequently than did patients with disruptive behaviour (P2 df 1, P < 0.05). This likely reflects the fact that more of these individuals were employed at the time of their crisis leading to admission. As shown in Table 3, the patients with disruptive behaviour were more likely to report housing problems than patients without disruptive behaviour (P2 = 4.54, df 1, P < 0.05). These findings present a picture of disruptive patients as having poorer social adjustment and greater social dis- advantage than control subjects. We entered the significant variables from the bivariate analyses into a forward stepwise logistic regression analysis, predicting the presence or absence of disruptive behaviour in hospital to determine the unique predictors of these behaviours. Four variables significantly contributed to the final model (R2 = 0.37, P < 0.001) with the following ORs: Axis III infectious diseases (OR 7.63; 95% CI, 1.41 to 41.67), Axis IV housing problems (OR 3.58; 95% CI, 1.21 to 10.64), history of suicidal behaviour (OR 3.46; 95% CI, 1.24 to 9.71), and problems with primary supports (OR 0.12; 95% CI, 0.03 to 0.46), which was related to a reduced risk of demonstrating disruptive behaviour in hospital. DiscussionThis retrospective case–control study demonstrates that patients with disruptive behaviour during an inpatient psychiatric admission were significantly more likely to receive the diagnosis of BPD or borderline traits, compared with patients without disruptive behaviour. The results further suggest that patients with disruptive behaviour could be characterized by a history of impulsive behaviours, including previous suicidal behaviour and a lifetime history of aggression, and by a proximal lack of environmental resources, including fewer primary supports and more housing problems. When these variables were entered into a stepwise logistic regression analysis, BPD and borderline traits no longer significantly predicted disruptive behaviour. A history of suicidal behaviour, of not reporting problems with primary social supports, of having a concurrent infectious disease, and of reporting housing problems significantly differentiated patients with disruptive behaviour from patients without disruptive behaviour These results indicate that certain patient characteristics may be more predictive of disruptive behaviour during an inpatient admission than the emergency psychiatric diagnosis of BPD. A history of suicidal behaviour is likely a marker for patients with characterological impulsive-aggressive behaviours (8), whether in the borderline spectrum or not. Their disruptive behaviours in hospital may best be considered a continuation of their long-standing difficulties with impulsive aggressiveness rather than a behavioural feature unique to their inpatient stay. Disruptive patients were more socially disadvantaged than were nondisruptive patients, as indicated by their problems with housing and their lack of a primary support network. Discharge planning may be particularly stressful for these individuals, given their lack of community resources and housing. The significant disturbance associated with the transition from being an inpatient to returning to the community has been demonstrated: the UK’s national confidential inquiry into suicide and homicide by people with mental illness found that 41% of inpatient suicides occurred when discharge was being planned (9).The finding of more disruptive patients being diagnosed with infectious diseases, vs the comparison patients, is difficult to explain. Perhaps these individuals with disruptive behaviours demonstrate, in general, poor self-care in adulthood. A similar explanation was given by Laub and Vaillant, who found in their 50-year follow-up study of delinquent vs nondelinquent boys that poor self-care and alcohol abuse were related to increased mortality rates (10). We acknowledge the limitations of this study. The predictive validity of the currently identified clinical variables bears replication using a prospective design. The independent variables need to be assessed blindly, and the diagnoses need to be established with reliable and valid structured interviews. Because of our decision to exclude patients with psychotic disorders, the findings of this study may not generalize to all psychiatric inpatients. We recommend multisite studies with larger samples to determine whether these predictors are replicable and generalize to nonurban settings. Finally, although predicting disruptive behaviours during an inpatient stay is important for successful management of the patient and inpatient setting, the behaviours do not necessarily indicate that the particular admission was unwarranted or ineffective. However, this preliminary study identifies clinical characteristics that can be assessed during an emergency visit to determine the likelihood that a patient may demonstrate disruptive behaviour during an inpatient hospitalization. These findings bear replication using a prospectively designed cohort study. Funding and SupportDr Boggild received funding from St Michael’s Hospital, Summer Student Scholarship, to conduct this research as an undergraduate medical student. Dr Heisel held the Stephen Godfrey Fellowship in suicide studies at the time this study was completed. References1. Hull JW, Yeomans F, Clarkin J, Li C, Goodman G. Factors associated with multiple hospitalizations of patients with borderline personality disorder. Psychiatr Serv 1996;47:638–41. 2. Davis T, Gunderson JG, Myers M. Borderline personality disorder. In: Jacobs DG, editor. The Harvard Medical School guide to suicide assessment and intervention. San Francisco: Jossey-Bass Publishers; 1999. p 311–31. 3. Rosenbluth M. The inpatient treatment of the borderline personality disorder: a critical review and discussion of aftercare implications. Can J Psychiatry 1987;32:228–37 4. Faulkner CJ, Grapentine WL, Francis G. A behavioral comparison of female adolescent inpatients with and without borderline personality disorder. Compr Psychiatry 1999;40:429–33. 5. Soliman AE, Reza H. Risk factors and correlates of violence among acutely ill adult psychiatric inpatients. Psychiatr Serv 2001;52:75–80. 6. Brown GL, Goodwin FK, Ballenger JC, Goyer PF, Major LF. Aggression in humans: correlates with cerebrospinal fluid amine metabolites. Psychiatry Res 1979;1:131–9 7. SPSS for Windows. Version 10.0. Chicago (IL): SPSS Inc; 2000 8. Mann JJ, Waternaux C, Haas GL, Malone KM. Toward a clinical model of suicidal behavior in psychiatric patients. Am J Psychiatry 1999; 156:181–9. 9. Robinson J, Hunt IM, Meehan J, Bickley H, McCann K, Parsons R, and others. Suicide within 12 months of contact with mental health services. Presented at the 9th European Symposium on Suicide and Suicidal Behaviour; 2002 September 14–17; Warwick (UK). 10. Laub JH, Vaillant GE. Delinquency and mortality: a 50-year follow-up study of 1000 delinquent and non-delinquent boys. Am J Psychiatry 2000;157:96–102 Author(s)Manuscript received February 2003, revised, and accepted June 2003. 1. Resident Physician, University of Toronto, Toronto, Ontario 2. Senior Instructor of Psychiatry (Psychology), Department of Psychiatry, University of Rochester School of Medicine and Dentistry, Rochester, New York. 3. Arthur Sommer Rotenberg Chair in Suicide Studies, Professor of Psychiatry, St Michael’s Hospital, Department of Psychiatry, University of Toronto, Toronto, Ontario Address for correspondence: Dr PS Links, St Michael’s Hospital, 30 Bond Street, Suite 2-010, Shuter Wing, Toronto, ON M5B 1W8 e-mail: paul.links@utoronto.ca
1 | 2
|
|||||||||||||||||||||||||||||||||||||||