To test the hypothesis that discharge disposition for adolescents admitted to medical hospitals after attempting suicide varies as a function of hospital type and geographic region.
Retrospective cohort analysis.
The nationally representative Kids’ Inpatient Database for 2000.
Patients aged 10 to 19 years with a diagnosis of suicide attempt or self-inflicted injury.
Main Outcome Measure
Likelihood of transfer to another facility vs discharge to home.
Care for 32 655 adolescents who attempted suicide was provided in adult hospitals (83% of hospitalizations), children’s units in general hospitals (10%), and children’s hospitals (4%). More than half (66%) of medical hospitalizations ended with discharge to home, 21% with transfer to a psychiatric, rehabilitation, or chronic care (P/R/C) facility, 10% with transfer to a skilled nursing facility, intermediate care facility, or short-term acute care hospital facility, and 2% with death or departure against medical advice. After adjustment for individual patient characteristics, children’s units were 44% more likely than adult hospitals to transfer adolescent patients to a P/R/C facility (odds ratio [OR], 1.44; 95% confidence interval [CI], 1.07-1.94). Patients cared for outside the Northeast were significantly less likely to be transferred to a P/R/C facility (South: OR, 0.79; 95% CI, 0.65-0.97; Midwest: OR, 0.63; 95% CI, 0.49-0.80; West: OR, 0.29; 95% CI, 0.22-0.38).
Most adolescents admitted to a medical hospital after a suicide attempt are discharged to home, and the likelihood of transfer to another facility appears to be influenced by the geographic location of the admitting hospital and whether it caters to children.
When adolescents are hospitalized in a medical facility after a suicide attempt, where do they go after discharge? What factors influence, or should influence, these disposition decisions? According to the Centers for Disease Control and Prevention 2003 Youth Risk Behavior Survey,1 9% of high school students reported that they had attempted suicide in the past year, but only 3% had sought medical attention and a smaller, but unknown proportion had been hospitalized. Because medically serious attempts are predictive of similar behaviors in the future,2- 7 patients admitted to a medical hospital after a suicide attempt may be at great risk for future self-destructive behaviors. Despite the increased risk for subsequent self-inflicted harm, little is known regarding the care that adolescents receive after a suicide attempt. Furthermore, there is no firm consensus regarding optimal follow-up care after a suicide attempt8,9 beyond general recommendations that after undergoing medical evaluation and stabilization patients should receive a comprehensive emotional and psychosocial assessment before discharge from the hospital.10
A major decision in care after a suicide attempt is whether to discharge the patient to home with outpatient psychiatric follow-up or to transfer the patient to another facility for further evaluation and management. Scant data inform this decision. While studies have shown a significant proportion of adolescents do not reliably comply with scheduled outpatient psychiatric follow-up after a suicide attempt,9,11,12 to our knowledge, there are no adequate studies to demonstrate the effectiveness of short-term inpatient care for adolescents who attempt suicide13,14 or even the extent to which this option occurs.
Factors that potentially affect the disposition of suicidal adolescents can be classified as either patient-specific or health care system characteristics. Individual patients in part are triaged for posthospital care depending on the severity of the self-inflicted injury, and some patients require rehabilitation or long-term care. Certain patient-specific risk factors, such as active suicidal ideation, available family support systems, previous suicide attempts, and past failure to adhere to outpatient treatment, have been used by clinicians to guide decisions about whether to admit an adolescent who has attempted suicide to an inpatient psychiatric facility.8,15,16 Attributes of the health care system, such as the type of hospital or availability of psychiatric resources, may also influence disposition planning. For example, adolescents are seen in a variety of inpatient medical settings, including adult hospitals, children’s hospitals, and children’s units in general hospitals. Some differences in the care and disposition planning for adolescents have been observed in pediatric-centered vs adult-centered care for suicide-related17 and non–suicide-related problems.18 Insufficient options for inpatient psychiatric care could increase the likelihood of patients being discharged to home rather than being transferred to a hospital setting. The availability of psychiatric resources varies by geographic region and affects the use of these services.19 Despite the potentially important effect of the place of care and the availability of aftercare services on the disposition of suicidal youth, there are few studies of these issues in the research literature.
In this study we examine a national sample of adolescents admitted to medical hospitals after attempting suicide. The sample is drawn from a federally funded hospital discharge data set, which has previously been used to describe hospitalized pediatric patients and to identify variations in their care.18,20- 22 Our objective is to describe the characteristics of hospital admissions of adolescents who attempted suicide and the patterns of discharge disposition, and specifically to test the hypothesis that disposition varies as a function of hospital type and geographic region.
This retrospective cohort study examined discharge disposition patterns for suicidal adolescents admitted to medical hospitals. We used data from the nationally representative Kids’ Inpatient Database (KID) for 2000,23 compiled by the Health Care Utilization Project and sponsored by the Agency for Healthcare Research and Quality. This database includes a 20% sample of uncomplicated births and 80% of all other pediatric and adolescent (aged <20 years) admissions from participating hospitals during 2000. The database contains more than 2.5 million discharges of pediatric patients from 2784 US nonrehabilitation hospitals in 27 states organized into 4 geographic regions: Northeast: Connecticut, Maine, Massachusetts, New Jersey, New York, and Pennsylvania; South: Florida, Georgia, Kentucky, Maryland, North Carolina, South Carolina, Tennessee, Texas, Virginia, and West Virginia; Midwest: Iowa, Kansas, Missouri, Wisconsin; and West: Arizona, California, Colorado, Hawaii, Oregon, Utah, and Washington. Hospitals are also grouped into hospital type, including non–children’s hospital, children’s hospital, or children’s unit in a general hospital, according to a classification system developed by the National Association of Children’s Hospitals and Related Institutions. Weights are provided to calculate national discharge estimates from the data set.
We selected all records for hospitalized patients ages 10 to 19 years admitted after a suicide attempt. This age range was chosen to correspond to the 10- to 14-year-old and 15- to 19-year-old age groups typically presented in census data and in morbidity and mortality data on suicide in adolescents.24,25 These patients were identified by a principal or secondary discharge diagnosis of suicide attempt or self-inflicted injury (International Classification of Diseases, Ninth Revision [ICD-9] codes E950-E959). Available patient demographic information included age, race/ethnicity, median household income for the ZIP code in which the patient resided, primary payer, method of suicide attempt, hospital type and geographic region in which the patient was hospitalized, and any noted psychiatric diagnoses (ICD-9 codes 290-319). Patient disposition was coded in the KID using the same disposition coding as the patient status data element on the Centers for Medicare & Medicaid UB-92 claim form: (1) routine discharge to home, (2) transfer to another short-term acute care general hospital, (3) transfer to a skilled nursing or intermediate care facility, (4) transfer to a psychiatric, rehabilitation, or chronic care facility (P/R/C), (5) discharge to home with home health care, (6) left against medical advice, or (7) died in the hospital.
All statistical analyses were performed on weighted data using Stata Statistical Software, Release 8.226 to account for the weighting of each observation, the complete stratified sample frame, and the clustering of observations within hospitals. Descriptive statistics were produced to provide national estimates for youth suicide hospitalization. To compare disposition patterns based on categorical variables such as hospital type and geographic region, we used the χ2 test of statistical significance. We then used multinomial logistic regression to determine the independent effects of variables on the likelihood of a patient being discharged either to home, to a P/R/C facility, to a skilled nursing or intermediate care facility, or to another short-term acute care hospital. Predictor variables, each entered into the regression equation as sets of dummy variables with a specific variable for missing values, included hospital type, geographic region, patient age, sex, race/ethnicity, method of suicide attempt, psychiatric diagnosis (if any), median income for ZIP code, and primary payer. Psychiatric diagnoses were grouped into the following 10 categories: depressive disorders (ICD-9 codes 296.20-296.36 and 311); bipolar disorder (ICD-9 codes 296.00-296.16, and 296.40-296.89); drug and alcohol disorders (ICD-9 codes 291-292 and 303-305); psychoses, including schizophrenia (ICD-9 codes 293-295 and 297-298); disruptive disorders, including conduct disorder and attention-deficit/hyperactivity disorder (ICD-9 codes 312-314); developmental disorders (ICD-9 codes 299, 315, and 317-319); stress disorders (ICD-9 codes 308-309); neurotic disorders (ICD-9 code 300); personality disorders (ICD-9 code 301); and other (ICD-9 codes 290, 302, 306, 307, and 310). The results of the hypothesis tests were deemed significant at an α level of P = .05.
Our research protocol was determined to be exempt from review by the Human Subjects Committee at The Children’s Hospital of Philadelphia.
In 2000, an estimated 32 655 adolescents aged 10 to 19 years were admitted to a medical hospital in the United States after a suicide attempt (Table 1). The median age of adolescents who attempted suicide was 16 years (interquartile range, 15-19 years). Of the patients, 70.3% were female; 63.8% were white, 9.5% were Hispanic, and 7.3% were black. Of the adolescent patients, 83.3% were cared for in adult hospitals, 10.2% in a children’s unit in a general hospital, and 4.4% in a children’s hospital. The median length of stay was 2 days (interquartile range, 1-6 days). Two thirds (65.8%) of patients who had attempted suicide had routine discharge to home after medical hospital admission, 21.2% of patients were transferred to a P/R/C facility, 7.1% were transferred to another short-term acute care hospital, and 3.4% were transferred to a skilled nursing or an intermediate care facility. Fewer than 1% (0.8%) of patients died in the hospital, and 1.4% left against medical advice; these patients were excluded from further analyses regarding disposition.
The likelihood of transfer varied by geographic region, with adolescent patients cared for in the Northeast most likely to be transferred to a P/R/C facility and least likely to be discharged to home (Table 2). Patients in the West were significantly less likely to be transferred to a P/R/C facility and were most likely to be transferred to a skilled nursing or intermediate care facility. Disposition also varied by hospital type. While adult hospitals were more likely to discharge adolescent patients to home after a suicide attempt than were child-focused inpatient settings, child-focused inpatient settings were significantly more likely to transfer patients to a P/R/C facility (Table 2).
After adjusting for individual characteristics in a multinomial logistic regression model (Table 3), children’s units were 44% more likely to transfer adolescent patients to a P/R/C facility than were adult hospitals (odds ratio [OR], 1.44; 95% confidence interval [CI], 1.07-1.94). No statistically significant differences regarding transfer to a skilled nursing or intermediate care facility or to a short-term acute care hospital were evident across hospital types. Compared with patients cared for in the Northeast, patients elsewhere were significantly less likely to be transferred to a P/R/C facility than discharged to home (South: OR, 0.79; 95% CI, 0.65-0.97; Midwest: OR, 0.63; 95% CI, 0.49-0.80; West: OR, 0.29; 95% CI, 0.22-0.38), whereas patients in the West were significantly more likely to be transferred to a skilled nursing or intermediate care facility (OR, 19.11; 95% CI, 11.67-31.30). Transfer to a P/R/C facility was more likely than discharge to home for adolescents with bipolar disorders (OR, 1.64; 95% CI, 1.26-2.14) or depression (OR, 1.16; 95% CI, 1.02-1.32), while those with personality disorders (OR, 0.73; 95% CI, 0.58-0.92), disruptive disorders (OR, 0.66; 95% CI, 0.56-0.78), and stress disorders (OR, 0.45; 95% CI, 0.37-0.56) were less likely to be transferred to these facilities. Patients who were admitted after self-cutting were 65% less likely to be transferred to a P/R/C facility than were adolescents who attempted suicide by ingestion or self-poisoning (OR, 0.35; 95% CI, 0.27-0.44). Transfer to a P/R/C facility was also less likely for female patients (OR, 0.73; 95% CI, 0.64-0.83), black (OR, 0.76; 95% CI, 0.60-0.96) or Hispanic (OR, 0.63; 95% CI, 0.49-0.80) patients, and patients with private insurance (OR, 0.76; 95% CI, 0.66-0.88). Diagnosis, method of suicide attempt, and payer each appear to exert similar patterns of influence on the likelihood of transfer to a skilled nursing or intermediate care facility as opposed to discharge to home.
This national study of variations in disposition among adolescents admitted to medical hospitals after a suicide attempt reveals that most adolescents who attempted suicide were discharged to home after medical stabilization. Transfer to another facility varied as a function of attributes of the health care system (specifically, hospital type and geographic region as hypothesized and insurance status) and patient-specific factors (psychiatric comorbidities and the method of suicide attempt). We discuss the implications of each of these sets of findings below.
Because adolescents who attempt suicide are brought to both pediatric-focused and adult-focused hospitals, it is important to know whether care varies according to hospital type. Compared with patients in adult hospitals, adolescents hospitalized in children’s units are more likely to be transferred to P/R/C facilities than discharged to home after a suicide attempt. Like other studies that have documented differences in the care of adolescents among pediatricians, family physicians, and internists,26,27 our observation raises questions regarding factors that underlie such variation. Compared with their colleagues who treat adults, pediatricians may see more adolescents who engage in self-directed harmful behaviors and are therefore more aware of the acute needs of this population. In addition, quality or availability of consultative psychiatric services for children and adolescents may differ between adult- and child-oriented medical facilities, thereby affecting the assessment of patients’ immediate psychiatric needs. Variations in case management and discharge planning services may also exist between these settings. For example, if adult hospitals are more effective at coordinating outpatient follow-up, health care providers in those facilities might be more comfortable than providers from child-focused centers in sending a patient home directly after medical stabilization. A study of discharge planning for hospitalized suicidal adolescents in France showed that health care providers in pediatric hospital units contacted primary care physicians and proposed follow-up care plans for these patients more frequently than did care providers in other departments such as internal medicine and surgery.17 While the investigators did not examine how disposition decisions were made or how successful patients were in following through with these plans, their study is consistent with our findings of care variation by hospital type.
Aware of other studies that revealed regional variations in pediatric health,18,28 we hypothesized that discharge disposition after a suicide attempt would display geographic differences based on the regional availability of inpatient psychiatric care. Our findings support this hypothesis. Significantly higher numbers of transfers to P/R/C facilities were seen among adolescents who attempted suicide and were hospitalized in the Northeast, a region in the United States where states have the most inpatient psychiatric beds per 100 000 population; conversely, patients least likely to be transferred to psychiatric facilities were in the West, a region in which states have the lowest numbers of inpatient psychiatric beds.29 Furthermore, the substantially larger proportion of patients in the West being transferred to nonpsychiatric facilities suggests that while approximately one third of patients across the United States are deemed to require some form of ongoing inpatient care, the relative dearth of inpatient psychiatric facilities in the West, and perhaps elsewhere, may be leading to transfers to the next best treatment setting. According to federal and state definitions, skilled nursing facilities do not include any institutions that are primarily for the treatment of mental illness and intermediate care facilities do not provide 24-hour supervised nursing care.30,31 Therefore these facilities seem unable to provide the appropriate psychiatric care or continuous supervision needed by hospitalized suicidal adolescents. An alternate hypothesis for the observed geographic differences, namely, that severity of injury varies by region and is worst in the West, thereby requiring ongoing medical care, cannot be excluded.
Our finding that adolescent patients with private insurance are less likely to be transferred to P/R/C facilities than are those covered by Medicaid is consistent with trends in inpatient psychiatric hospitalization observed in other studies, in which publicly insured patients have higher utilization of inpatient psychiatric care than do their privately insured counterparts.32 The observed variation in psychiatric care patterns raises important questions about equal access to appropriate health care services: Are private insurance plans too restrictive in their inpatient benefits for mental health care? Are patients with Medicaid spending relatively more time in inpatient settings because of inadequate outpatient care opportunities? Answering these questions regarding how disparity in access affects overall quality of care requires better comparative assessments of the outcomes experienced by adolescents who have attempted suicide who receive either inpatient or outpatient psychiatric care.
Numerous studies have found that more than 90% of those who attempt suicide meet criteria for a psychiatric diagnosis.2,3,6,33- 35 That 85% of those who attempted suicide in this study received a psychiatric diagnosis is heartening and provides evidence that, for the most part, physicians recognize the importance of diagnosing the psychiatric comorbidity that almost always accompanies a suicide attempt. The observation that youth with different diagnoses had different likelihood of discharge dispositions, however, warrants further exploration. For example, that patients with diagnoses of depression and bipolar disorder are more likely to be transferred to another inpatient facility has at least 2 possible interpretations. First, the diagnosis may be associated with the severity of the attempt, which may determine discharge disposition. A second hypothesis is that providers associate these diagnoses with the likelihood of a second suicide attempt. Certainly, affective disorders are the most common psychiatric comorbidity among adolescents who attempt suicide,3,6,9,36 and providers may perceive these patients as being at higher risk for future attempts and therefore needing more intensive psychiatric intervention. Completed suicide, however, has also been highly associated with conduct disorder,2,6,33,34,37 yet patients with this diagnosis were less likely to be transferred to P/R/C facilities than sent home. Youth with substance use disorders and psychotic disorders, which are major risk factors for suicide,2,6,33,34,37 were no more likely to be transferred to P/R/C facilities than youth with no psychiatric diagnosis. This disparity may be owing to lack of knowledge of the association between suicide and these diagnoses, the belief that the suicide attempt is not integrally related to the disorder, or because youth assigned these diagnoses made less serious attempts.
The likelihood of being transferred varied by method of suicide attempt. Compared with patients who had ingested poisons, adolescents who engaged in cutting behaviors were significantly less likely to be transferred to P/R/C facilities than discharged to home whereas those who combined the ingestion of a poison with another method were more likely to be transferred to another facility. The perceived lethality or resulting morbidity of the attempt method may explain this finding, given that cutting is common in uncompleted attempts, whereas combined methods, firearms, hanging, jumping, and drowning are more commonly associated with suicide completion.2,35- 37 Indeed, these associations of method and likelihood of death are incorporated into tools used to assess the degree of suicidal intent in patients,38 and the American Academy of Child and Adolescent Psychiatry recommends that adolescents who attempt suicide using more lethal methods should be treated as if the risk for future completion is high.2
The demographics of our study sample are consistent with those in other studies of adolescents who attempted suicide,1,2,6,33,34 which suggests that this is a representative sample of adolescents who attempt suicide. Our ratio of 2.4 female patients for every male patient admitted after an attempt reflects general population sex differences in suicide attempts; female adolescents are at least twice as likely as male adolescents to attempt suicide.1Our study sample also has an age distribution,34 used methods,2 and had comorbid psychiatric disorders6 similar to those found in adolescent suicide attempters in general. The racial/ethnic distribution of adolescents in our sample proportionally reflects the 10- to 19-year-old US population. Given that Hispanic youth attempt suicide at higher rates than other ethnic groups,1 it might have been expected that the proportion of Hispanic youth in our sample would have been higher. The overrepresentation of white patients in our hospitalized sample may reflect cultural and economic barriers that influence racial and ethnic disparities in access to health care, especially mental health care.39
Several study limitations should be considered. First, because the KID is an administrative database that extracts information from discharge abstracts, the available clinical and historical details about the individual patients are limited. Therefore no conclusions can be drawn from this study about the influence on disposition decisions of important risk factors such as previous suicide attempts, severity of mental illness, family history of suicidal behavior or psychiatric disorders, abuse history, or involvement in ongoing therapy. In addition, our finding of 32 655 adolescent admissions reflects the number of discrete admissions, not unique patients. Because of limitations of the data set, we cannot determine whether any of these patients had multiple admissions because of suicide attempts during 2000 or any previous year. Second, we were unable to make a direct comparison between routine discharges to home and transfers specifically to inpatient psychiatric hospitals because these hospitals were grouped into a broader disposition category, “transfer to another facility,” which encompasses P/R/C facilities. Third, a number of adolescents may not have been included in our sample if their physicians did not or were unwilling to code the injuries as self-inflicted or a suicide attempt, a problem that has been described in the reporting of completed suicides.40- 42 Similarly, we may have included too many patients because it is difficult to know whether the adolescents included in our sample had true suicidal intent or inflicted self-harm without intent to die. Although self-injurious behavior exists on a continuum of potential lethality, recommendations have been made to take such actions seriously, especially given the developmental immaturity of many teenagers in understanding the true long-term consequences of their actions.2,42,43 Therefore we erred on the side of including all self-inflicted injuries as potential attempts at suicide.
In conclusion, most adolescents admitted to medical hospitals after a suicide attempt are discharged to home. While hospitals are identifying comorbid psychiatric diagnoses in most patients, more rigorous evaluations of outcomes comparing outpatient vs inpatient psychiatric follow-up are needed to determine which strategy is most beneficial for these patients. Furthermore, variations in disposition associated with geographic region, insurance type, and to a lesser degree hospital type suggest that factors other than the needs of the patient are driving care, including the availability of resources inside and outside the hospital, and warrant further study.
Correspondence: Leonard J. Levine, MD, Division of Adolescent Medicine, St Christopher’s Hospital for Children, Erie Avenue at Front Street, Philadelphia, PA 19134 (firstname.lastname@example.org).
Disclaimer: Drs Levine and Feudtner had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Accepted for Publication: March 24, 2005.
Funding/Support: Dr Feudtner was supported in part by grant K08 HS00002 from the Agency for Healthcare Research and Quality, Rockville, Md.
Acknowledgment: We thank Chrissie Forke, CRNP, for assistance and advice in the conduct of this study.
Levine LJ, Schwarz DF, Argon J, Mandell DS, Feudtner C. Discharge Disposition of Adolescents Admitted to Medical Hospitals After Attempting Suicide. Arch Pediatr Adolesc Med. 2005;159(9):860-866. doi:10.1001/archpedi.159.9.860