“I'm Going Home”: Discharges Against Medical Advice

Abstract

Discharge against medical advice (AMA), in which a patient chooses to leave the hospital before the treating physician recommends discharge, continues to be a common and vexing problem. This article reviews the prevalence, costs, predictors, and potential interventions for this clinical problem. Between 1% and 2% of all medical admissions result in an AMA discharge. Predictors of AMA discharge, based primarily on retrospective cohort studies, tended to be younger age, Medicaid or no insurance, male sex, and current or a history of substance or alcohol abuse. Interventions to reduce the rate of AMA discharges have not been systematically studied. This article offers suggestions for interventions based on studies in other areas of clinical care as well as the psychiatric AMA discharge literature. Studies for this review were identified by searching the relevant MeSH heading (discharge) and key words (against medical advice, leave, elope, hospital, and self-discharge) in PubMed databases and selecting all English-language articles from 1970 through 2008 that included data on adult medical inpatients.

Discharge against medical advice (AMA), in which a patient chooses to leave the hospital before the treating physician recommends discharge, is a problem for many physicians who treat hospitalized patients.^1,2 Leaving the hospital against the physician's advice may expose the patient to risk of an inadequately treated medical problem and result in the need for readmission.³ The ethical dilemma that this issue raises is conceptually relatively straightforward. Many physicians struggle with the desire to respect the patient's wishes to leave AMA (in general, the patient's right to self-determination or autonomy) while attempting to do what they think is best for the patient (to act with beneficence).⁴ In practice, managing this issue presents more complications than simply identifying and potentially prioritizing the relevant ethical principles. Physician-patient communication, informed consent, and underlying psychiatric issues are all relevant to practical management. This article examines the problem of AMA discharges—their prevalence, risks, and costs—and formulates recommendations for managing and preventing them on the basis of available evidence.

METHODS

Studies were identified by searching PubMed databases for English-language articles from 1970 through 2008 that included data on adult medical inpatients. Primary psychiatric admissions and admissions for detoxification or substance abuse were excluded. The search was performed by using the MeSH heading discharge and then combining it with the following key words: against medical advice, leave, elope, hospital, and self-discharge. Bibliographies of all articles were searched for related studies.

PREVALENCE AND RISK TO QUALITY

Against medical advice discharges continue to be a highly prevalent problem of health care quality, representing as many as 2% of all hospital discharges.^5-7 Furthermore, patients discharged AMA, taken as a whole, are an at-risk group for both morbidity and mortality. Patients with asthma who were discharged AMA had a 4-times higher risk of readmission to the emergency department within 30 days (21.7% vs 5.4%) and almost a 3-times higher risk of readmission to the hospital within 30 days (8.5% vs 3.2%).⁸ In a study of a general medicine service, patients who left AMA were 7 times more likely to be readmitted within 15 days (21% vs 3%), almost always for the same diagnosis.³ In a large retrospective study among almost 100,000 patients admitted with acute myocardial infarction, those who left AMA (N=1079) underwent fewer revascularization procedures and had shorter lengths of hospital stay. After adjustment for these phenomena, patients who left AMA had a 40% higher risk of death or readmission for myocardial infarction or unstable angina up to 2 years after discharge.⁹ Although a moderately sized prospective study found no relationship between AMA discharge and death,³ 2 other studies, smaller and variable in their design, found a high rate of mortality among patients who were discharged AMA. In a review of medical records, Link et al¹⁰ found a 15.7% mortality rate at 1 year among 57 patients discharged AMA from a group of academic hospitals in Virginia. Using a case-control design, Corley and Link¹¹ found a 19% rate of mortality at 6 months among 33 medical patients who left AMA from a Veterans Affairs institution.

Few data are available on the estimated total costs to the health care system of unanticipated AMA discharges. Multiple studies have found that patients who leave AMA are at risk for early readmission,^3,12,13 which can result in higher, unnecessary health care costs. Aliyu,¹² using 30-day readmission data, calculated the readmission cost due to an AMA discharge at 56% higher than expected from the initial hospitalization.

PREDICTORS OF AMA DISCHARGE

Understanding why patients choose to leave the hospital AMA has obvious importance because of the potential to identify those at higher risk and therefore intervene earlier to prevent excess morbidity, mortality, and health care costs. Most of the published data are from retrospective analyses and case-control studies at single urban institutions, limiting the ability to define a clear causal relationship (eTable^3,5,7-30 online linked to this article). However, the following correlates of AMA discharge have had reasonably consistent results over time: lower socioeconomic class, male sex, younger age, Medicaid or no insurance, and substance abuse.^{5,8,12,15-17,19,22,23,26,27,30}

Some studies have had additional novel findings. Jeremiah et al²⁷ identified lack of a primary care physician as being associated with a higher likelihood of signing out AMA. A history of leaving AMA is associated with a higher likelihood of doing the same in the future.^8,25 Although African American race has been associated with AMA discharge in retrospective studies,^5,12,15,31 the finding is inconsistent in larger studies and may be confounded by socioeconomic and hospital-related factors.^7,8,29

In smaller studies limited to subpopulations of patients with human immunodeficiency virus, additional correlates of AMA discharge include patient-reported factors such as financial issues and sickness within the family.³² Patient-reported reasons for leaving AMA often include these types of personal or financial obligations. Other studies have found that patients report feeling better^21,27 or receipt of social assistance payments^24,32 as their primary reason for leaving AMA.

In most retrospective studies during the past 3 decades, the presence of a drug or alcohol problem has had a consistent association with the decision to sign out AMA. The reason for this association is still unclear, but limited data suggest that underlying addictive behaviors and the wish to acquire more drugs might be reasons.^33-35 Recommendations for reducing the rate of AMA discharges among patients with a drug or alcohol history have consisted primarily of early identification, discussion, and communication about counseling. Chan et al²⁴ found that AMA discharges were less likely among injection drug users with human immunodeficiency virus if they were receiving methadone in the hospital or had social support from family and friends. Protocols to treat alcohol and opioid withdrawal to reduce AMA discharges have not been studied in patients on a medical service. Given the high prevalence of AMA discharges in these patients, this would be a worthy area of study.

INSIGHTS FROM PSYCHIATRIC LITERATURE

Although beyond the scope of this review, a much larger amount of literature has examined psychiatric inpatients and AMA discharges.³⁶ Clinical overlap exists between medical and psychiatric patients studied because of the high burden of psychiatric morbidity in medical inpatients; however, data are otherwise too heterogeneous to allow wide-ranging comparisons. For example, AMA discharges in psychiatric populations range from 3% to 51% (average, 17%),³⁶ far higher than in medical patients studied. However, given the paucity of data on interventions to reduce AMA discharges for medical patients, drawing some insights on the process from the psychiatric literature may be useful.

Targum et al³⁷ found an approximately 30% decrease in total AMA discharges among psychiatric inpatients in a private hospital that used a nurse as a patient advocate. The advocate's responsibility was to help explore a “patient's preconceptions about hospitalization and to address fears and complaints about it.” Involving a consultation-liaison psychiatrist early in the course of a hospital admission may produce similar gains. In a retrospective study, Holden et al²¹ found that medical patients seen early by a consultation-liaison psychiatrist were less likely to sign out AMA.

Another concept drawn from the psychiatric literature is that AMA behavior may be more a consequence of factors related to admission and early hospitalization than a break-down in the therapeutic alliance between physician and patient. In a prospective study that derived and validated an AMA prediction tool, Steinglass et al³⁸ found that patients who signed out AMA were more likely to believe that their hospitalization would be short-term. Thus, the initial “treatment contract” rather than the nature of the hospitalization appeared to be at odds with patients' expectations.

STRATEGIES FOR PREVENTING AMA DISCHARGES

Addressing Substance Abuse

Proactively addressing substance abuse issues early during hospital admission can help prevent discharge dilemmas. Failing to collect these critical elements of the social history can lead to deficiencies in care and prevent the timely evaluation and intervention needed for patients with substance abuse. However, approaching these issues with patients requires skills that physicians may not have fully developed.³⁹ How the physician handles an evaluation for substance abuse can affect the patient's response. Physicians who start with an accusatory attitude are likely to be met with unease at best and with an AMA discharge and termination of care at worst. Maintaining a similar empathetic, dispassionate attitude toward all patients, regardless of their diagnosis, is more likely to result in a clear and effective evaluation.^40,41

Recognizing Psychological Factors

Although early articles examining the issue of AMA discharge were based on small series of case reports and descriptive studies, they provide important conclusions about the psychological mechanisms behind AMA discharges and suggest ways in which physicians can use these observations to potentially reduce the rate of untimely discharges. Two articles described the association of patients' anxiety and anger, possibly masking feelings of helplessness, with AMA discharges.^14,15 Albert and Kornfeld¹⁴ described a series of case studies on a hospital service that emphasized the need for early recognition and also showed that an AMA threat was a way for patients to demonstrate their feelings, often anger, anxiety, or depression. Furthermore, these investigators cautioned that many patients show overt signs of emotional distress before threatening to leave AMA, signs that can be missed or misinterpreted by physicians untrained to recognize them. More recently, this phenomenon has been studied in outpatient encounters by Levinson⁴² and others,^43,44 who found that patients offer their physicians “empathic opportunities” through their speech. Patients' comments, when physicians are attuned to them, give clues about the patients' underlying psychological state. An emerging body of research in outpatients suggests that physicians' recognition and acknowledgment of patients' underlying emotions can result in improved trust in the physician-patient relationship and more efficient and better quality of care.^45-49 The role in preventing or reducing AMA discharges, although potentially useful, remains to be evaluated.

Another consistency in the literature on AMA discharges is the recommendation for proactive physician-patient communication. Indeed, many problems like these originate in incomplete physician-patient communication.⁵⁰ Persistent somatic complaints, increasing anger, and insomnia may be related to uncommunicated feelings of dependency or loss of control; they may distract, and in the worst-case scenario, alienate the staff. Indeed, nursing and physician staff may become angry when dealing with this type of difficult patient, which may result in premature discharges.^21,51 However, if physicians are mindful of their own naturally occurring emotions, such as anger, during these difficult patient encounters, they will be better able to refocus the discussion back on the patient.⁵² Psychiatric consultants can assist in this process. A dispassionate, empathetic, and nonjudgmental assessment of patients is challenging, takes time, and requires continual practice,⁵³ but it remains the backbone of the therapeutic alliance. Navigating this process is especially important because sick patients are vulnerable to making decisions that may not reflect their best interests. Patients often respond to hospitalization with anxiety, depression, and fear, factors that can impair their ability to reason effectively.^54-56

Motivational Interviewing

Regarding the issue of communication in reducing AMA discharges, the literature contains limited explanations of the specifics for improvement. However, improving physician-patient communication has multiple elements and therefore can be studied and evaluated in different ways. Motivational interviewing, which relies on the principle of patient-centered interviewing to help physicians examine decisions through the particular perspective of the patient, has attempted to improve chronic disease outcomes by focusing on communication about patient behavior.^57-59 Motivational interviewing assumes that patients who make difficult decisions about behavior change are often ambivalent about the pros and cons of that change.⁶⁰ Using nonjudgmental, empathetic questioning, physicians can more fruitfully engage patients in a process designed to uncover the unspoken motivations behind their particular behaviors.⁶¹ Once the physician and patient can agree about this ambivalence and about the positive and negative aspects of either maintaining the status quo or making a change, the physician can more accurately approach the problem and, with the patient's involvement, formulate a plan.

Given the limited literature on the actual motivations for patients to sign out AMA, the applied use of motivational interviewing can potentially help in understanding how patients make these decisions. Information that more clearly represents a patient's motivating behavior will allow a physician to target counseling more appropriately and to potentially negotiate a discharge to a more medically appropriate time. For example, when a physician determines that an increasingly angry and “demanding” patient wants to leave the hospital to care for his homebound mother, not because he has little concern for his elevated blood pressure, the physician can attempt to reduce the patient's burden by focusing on that issue, rather than on the mounting discharge conflict between physician and patient. Particularly because many patients request to leave the hospital for financial and personal reasons, the clearer these motivations are, the better the physician can discuss the needs for hospitalization. In this way, physicians can “broaden the terms of engagement”²² for discussing and negotiating care during a hospitalization to leave room for the patient's prioritized needs. Patients who decide to leave the hospital for “personal” reasons may be prioritizing financial concerns over health concerns, and while this may make physicians uneasy, if it is an informed decision, it should be respected. Negotiating about the hows and whens of this decision is part of a harm-reduction strategy that physicians can undertake to ensure the best care possible while respecting patients' right to choose the manner and timing of their treatment. Discussing with patients the data on risk for morbidity can help with achieving fully informed consent.

MANAGING AN AMA DISCHARGE

Informed Consent

Informed consent in deciding to leave AMA is one of the most important elements of care for patients who make this decision. An informed decision means that the patient has arrived at the decision in consultation with his or her physician without being subjected to coercion and with a full understanding and appreciation of the risks, benefits, and alternatives of the decision.⁵⁹ In practice, this often involves an evaluation of decision-making capacity, which in most cases can be handled by the primary physician without specialist input. Indeed, the primary physician is in the best position to evaluate capacity because he or she usually has had the most time and consistent communication with the patient. Psychiatrists are particularly useful in determining the extent to which mental illness impairs capacity and whether it can be ameliorated. However, in regard to AMA decisions, the limited literature suggests that most patients do not receive a capacity evaluation before AMA discharge^30,62 and therefore may not be making a fully informed decision.

Informed consent in discharge planning can be a fairly straightforward process if the steps are clear. A systematic approach is useful in ethical situations that involve uncertainty and complexity^63-66 and can be useful in performing a capacity evaluation.^67,68 The following questions should be included in the evaluation of a patient who wishes to leave the hospital AMA: Does the patient understand and appreciate the admission diagnosis, its prognosis, and the likelihood of risks and benefits of leaving the hospital? Is the patient aware of the alternatives to treatment in the hospital and the risks and benefits associated with them? Can the patient make and communicate a choice? Can the patient articulate a reason for the refusal that is consistent with his or her values?⁶⁹

A clear conversation with the patient about these questions, followed by clear documentation of the answers in the medical record, ensures the best care possible for the patient and may reduce liability.^70,71 What should the physician do if answers to these questions are ambiguous or are subject to debate about the patient's intent? Multiple investigators have advocated the use of a “sliding scale” of capacity assessment; that is, the greater the risk due to the patient's refusal, the more certain the physician should be that the patient has decisional capacity.^68,72,73 For example, patients may have capacity to refuse a blood draw but not to refuse a lifesaving blood transfusion. In regard to hospital discharges, as the risk due to leaving AMA rises, so should the standard for decisional capacity.

Some patients may be at risk to themselves or others and insist on signing out AMA. If such a patient is deemed to be without decision-making capacity and has no surrogate, the physician may be able to keep the patient in the hospital against his or her will. Consultation with an ethicist or psychiatrist can be helpful in understanding the legal requirements for this type of involuntary hospitalization, which vary by state.

Finally, physicians should pay attention to the relative health literacy of patients being discharged AMA.⁷⁴ Health literacy, defined as “the degree to which individuals have the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions,” has been shown in numerous studies to have an effect on patients who obtain health care.⁷⁵ Reasonable efforts should be made to ensure that patients clearly understand the terminology and language used in the discharge discussion. Even in ideal settings, patients can be confused about discharge medications and plans.^76-78 Therefore, providing clear instructions during an untimely discharge is all the more important.

Follow-up

After a patient has fully discussed the discharge with his or her physician and made an informed decision to leave AMA, the physician's responsibility is to ensure that the discharge is as safe and appropriate as possible under the circumstances.⁷⁹ This responsibility includes helping the patient to follow up after discharge. Maintaining the therapeutic alliance does not end with an AMA discharge; it only transfers the alliance to another setting, usually an outpatient setting.⁷⁴

CONCLUSION

Against medical advice discharge continues to be a prevalent and frustrating problem for patients and their physicians. The literature is limited primarily to medical record reviews and retrospective analyses of associations with AMA discharges. Data for physicians on how to effectively manage and intervene in these complicated patient encounters are scant. Prospective studies of medical patients, focusing on patient, physician, and hospital variables, are most likely to reveal reliable and valid data about how best to address, prevent, and treat AMA behavior. Prospective controlled trials randomizing high-risk patients to intervention from either a patient advocate or a psychiatric consultation-liaison physician could examine the effectiveness of one of these approaches. Focusing on providing informed consent, with attention to the vulnerabilities and health literacy levels of hospitalized patients, can ensure the best care possible for patients while respecting their autonomy.