Secondary Logo

Journal Logo

Brief Report: Clinical Science

HIV-Positive Injection Drug Users Who Leave the Hospital Against Medical Advice

The Mitigating Role of Methadone and Social Support

Chan, Alex C. H., MD, MPH*†; Palepu, Anita, MD, MPH*†; Guh, Daphne P., MSc*‡; Sun, Huiying, MSc*‡; Schechter, Martin T., OBC, MD, PhD*‡§∥; O'Shaughnessy, Michael V., OBC, PhD*‡§¶; Anis, Aslam H., PhD*‡§∥

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: January 1st, 2004 - Volume 35 - Issue 1 - p 56-59
  • Free


Leaving hospital against medical advice (AMA) signifies a major disruption in the patient-physician relationship. When patients leave AMA, health care providers may experience a broad range of emotions, including concern, guilt, disappointment, frustration, resentment, and anger. 1–3 Leaving AMA may result in increased morbidity and hospital readmission. 4–6 Prior studies of leaving the hospital AMA have focused on admissions to general medical wards, psychiatric facilities, and inpatient alcohol treatment programs, where rates are reported to be <5%, 4% to 57%, and 10% to 38%, respectively. 7–21

Male sex, younger age, nonwhite ethnicity, lack of fixed address, comorbid psychiatric illness, lack of health insurance, and presence of drug- or alcohol-related diagnoses have been frequently identified as positive predictors of leaving the hospital AMA. 6,8,13 Substance abuse has been associated with increased risk of discharge AMA. 22,23 We determined the rate of discharge AMA from a specialized HIV/AIDS ward and identified factors associated with leaving AMA among patients with a history of injection drug use. We examined whether discharge AMA was associated with recent injection drug use and with specific drug use prior to hospital admission. We also examined if in-hospital methadone treatment was associated with reduced risk of leaving AMA.


Patient Selection

Consecutive patients admitted to an 18-bed HIV/AIDS ward between April 1997 and October 2000 were included in this study. St. Paul's Hospital is a university-affiliated, inner city, tertiary care center located in Vancouver, British Columbia, Canada. We restricted our attention to this cohort rather than all admissions to the hospital because of the availability of extensive longitudinal data.

Data Collection

Database information was extracted retrospectively from patient medical records at the time of discharge from the hospital. Patient information, including housing status, social supports, community supports, drug use history, and medical history, was collected on standardized forms. Social support was classified into (1) family, friends, or an HIV-negative partner who did not use drugs; (2) a partner who was HIV-positive or used drugs; and (3) absence of social support. Housing status was classified as stable (living in a house, apartment, or long-term care facility) or unstable (residing at a hotel, temporary residence, rooming house, shelter, or detoxification unit or having no fixed address). Community support included counseling, home care, support groups, and palliative care programs. Recent injection drug use was defined as self-reported use in the past 30 days prior to hospitalization. In-hospital methadone use was abstracted from the medication administration records. Ethnicity was categorized as aboriginal and nonaboriginal. At the time of discharge, patients who signed out AMA and those who left hospital without physician consent and did not return within 6 hours (and, as a result, lost their bed per hospital policy) were considered to have left AMA. The diagnosis most responsible for hospitalization was extracted from the discharge summary form routinely completed by the medical staff. Approval for accessing the patient database was obtained from the Ethics Review Board of Providence Health Care.

Statistical Analysis

Bivariate analysis using contingency table analysis for categoric variables and the Wilcoxon rank sum test for continuous variables was used to examine the relation between the variables of interest and discharge from hospital AMA. Because analysis was based on discrete hospital admissions rather than on admissions by unique subjects, a multivariate logistic regression model utilizing the generalized estimating equation algorithm was employed to account for multiple admissions by individual subjects. A multivariate logistic regression model was fitted to examine the relation between recent injection drug use and hospital use of methadone and leaving hospital AMA. This model adjusted for the frequency of hospital admissions during the study period, admission CD4 cell count, social assistance check day (yes/no), and day of discharge. All analyses were performed using SAS 6.12 software (SAS Institute, Cary, NC).


Study Population

Excluding in-hospital deaths, 744 patients admitted to the HIV/AIDS ward between April 1997 and October 2000 incurred 1466 hospitalizations, of which 284 resulted in leaving AMA. Our study focused on the 480 patients who reported injection drug use to be an HIV risk factor. These patients accounted for 1056 hospital admissions, of which 263 resulted in leaving AMA. Patients with a history of injection drug use accounted for 72% of all admissions to the HIV/AIDS ward and 92.6% of all discharges AMA from the HIV/AIDS ward. Of those patients who left AMA, 117 left once, 24 left twice, 10 left on 3 occasions, and 13 left on 4 or more separate hospital admissions.

Analyses were based on the 1056 discrete hospital admissions and are presented in Table 1. In-hospital methadone treatment occurred in 365 (35%) admissions. Methadone was prescribed in the hospital for 139 (66%) of 210 patients who reported daily heroin use and 35% of patients who reported intermittent heroin use. The most common diagnoses responsible for admission were pneumonia other than tuberculosis or Pneumocystis carinii pneumonia (29.4%), cellulitis/skin abscess (8.6%), and septicemia/bacteremia (4.8%), and they were more common among patients who left AMA than those who did not leave AMA (P < 0.05).

Factors Associated With Leaving AMA: Bivariate Analysis

Factors Associated With Leaving the Hospital Against Medical Advice

Factors associated with leaving the hospital AMA in the bivariate analysis are shown in Table 1. The multivariate analysis of factors associated with leaving AMA is summarized in Figure 1. Recent injection drug use (adjusted odds ratio [AOR] = 2.07, 95% confidence interval [CI]: 1.40–3.06) and aboriginal ethnicity (AOR = 1.59, 95% CI: 1.08–2.34) were positively associated with leaving AMA. Leaving AMA was also more likely to occur on weekends than on weekdays (AOR = 2.27, 95% CI: 1.49–3.48) and on days when social assistance (“Welfare”) checks were issued (AOR = 2.95, 95% CI: 1.70–5.10). In-hospital methadone use (AOR = 0.49, 95% CI: 0.32–0.77); social support from family, friends, or a partner who was HIV-negative and did not use drugs (AOR = 0.34, 95% CI: 0.21–0.53); and older age (per 10-year increment, AOR = 0.57, 95% CI: 0.44–0.74) significantly reduced the odds of discharge AMA.

Factors associated with leaving against medical advice (AMA); multivariate analysis. SS-FFP indicates social support from family, friends, or an HIV-negative partner who does not use drugs; IDU-active, report of injection drug use within 30 days prior to hospitalization; Weekend, discharge occurred on a Saturday or Sunday; Welfare day, discharge occurred on a day when social assistance payments were released (every second Wednesday).


The rate of hospital discharge AMA in our study (24.9%) is consistent with rates in studies of psychiatric wards and alcohol or drug detoxification units. 15–21 Although alcohol and drug use in general have been associated with discharge AMA, 12,13 to our knowledge, recent injection drug use and the specific drugs of abuse (including heroin, cocaine, and benzodiazepines) have not been previously examined in detail. Pérez de los Cobos et al 21 found that a positive urine drug screen for cocaine at the time of admission to a heroin detoxification unit was a risk factor for discharge AMA, however. Our findings are consistent with those of Anis et al, 23 who identified a history of injection drug use to be a predictor of leaving AMA. The association between in-hospital methadone use and reduced odds of discharge AMA may reflect a more stable subgroup of patients who were already taking methadone prior to their hospital admission or may represent a treatment effect of methadone initiation during the hospital stay (especially for the management of opioid withdrawal). It is likely that both types of patients were represented in our sample, but further information is not available from the collected data. Of note, among subjects who reported daily heroin use prior to hospital admission, those who received methadone had a lower rate of AMA than those who did not, suggesting a treatment effect while in the hospital. The beneficial effect of methadone persisted in the multivariate analysis that controlled for daily heroin use. Endicott and Watson 24 reported a reduced rate of discharge AMA (from 50% to 29%) from a chemical dependency unit after implementing an opiate withdrawal scale, but our literature review did not reveal other studies that examined the effect of substance abuse treatment on leaving AMA.

Although aboriginal ethnicity per se has not been previously described as an independent risk factor of discharge AMA, our findings are consistent with the literature regarding other ethnic minorities; in particular, African Americans have been identified to be at greater risk of leaving AMA. 11,14 Our results are consistent with those of studies that found younger age to be a positive predictor of discharge AMA. 6,7,10,19,20 The association between leaving hospital AMA and the issuing of social assistance checks has only recently been reported among HIV-infected patients admitted to a tertiary care hospital. 23 The increased rate of leaving AMA on weekends and on days when social assistance payments are issued suggests that triggers for illicit drug use may precipitate premature discharge from the hospital. The finding that lack of social supports (as defined in our study) positively predicts leaving the hospital AMA has also not been previously reported.

Our study has several limitations. Although we adjusted for HIV disease severity by controlling for CD4 cell count, we were unable to adjust for the severity of other medical and psychiatric illnesses. Self-reported drug use status may underestimate actual drug use. We were unable to ascertain if methadone was started prior to or during the index hospitalization. An independent association between discharge AMA and daily alcohol, heroin, or cocaine use may not have been observed due to insufficient sample size or colinearity with recent injection drug use. We were unable to examine the effects of specific types of community support, including residential drug treatment programs or peer-based support groups such as Narcotics Anonymous. Finally, the results of this study may not be applicable to other HIV/AIDS wards in which the drug use patterns and availability of substance misuse treatment differ. At our hospital, patients who are identified with substance use disorders are commonly referred to an inpatient Addiction Medicine consultation service, and opioid withdrawal is routinely treated with opioid substitution therapy (including methadone) or medically supervised detoxification.

In conclusion, HIV-infected patients who reported recent injection drug use prior to hospitalization and those of aboriginal ethnicity were more likely to leave the hospital AMA. Factors that reduced the odds of discharge AMA were receipt of in-hospital methadone, older age, and social supports. Therefore, interventions such as in-hospital drug addiction treatment programs (including methadone maintenance for opioid dependence) and protocols to deal with alcohol or drug withdrawal symptoms adequately might reduce the risk of discharge AMA. Health care provider attitudes toward patients with substance use disorders and toward treatment of substance use disorders should be explored. Furthermore, strategies to enhance social supports in marginalized populations at risk for hospital discharge AMA should also be investigated. Given that discharge AMA has been associated with increased morbidity and readmission, 4–6 interventions that reduce hospital discharge AMA may lead to improved patient outcomes.


The authors thank everyone who contributed to the development and implementation of the patient database used in this study.


1. Albert HD, Kornfeld DS. The threat to sign out against medical advice. Ann Intern Med. 1973; 79:888–891.
2. Jeremiah J, O'Sullivan P, Stein MD. Who leaves against medical advice? J Gen Intern Med. 1995; 10:403–405.
3. Smith DB, Telles JL. Discharges against medical advice at regional acute care hospitals. Am J Public Health. 1991; 81:212–215.
4. Corley MC, Link K. Men patients who leave a general hospital against medical advice: mortality rate within six months. J Stud Alcohol. 1981; 42:1058–1061.
5. Glick ID, Braff DL, Johnson G, et al. Outcome of irregularly discharged psychiatric patients. Am J Psychiatry. 1981; 138:1472–1476.
6. Weingart SN, Davis RB, Phillips RS. Patients discharged against medical advice from a general medicine service. J Gen Intern Med. 1998; 13:568–571.
7. Baile WF, Brinker JA, Wachpress JD, et al. Signouts against medical advice. J Behav Med. 1979; 2:85–92.
8. Jankowski CB, Drum DE. Diagnostic correlates of discharge against medical advice. Arch Gen Psychiatry. 1977; 34:153–155.
9. Jones AA, Himmelstein DU. Leaving a county hospital against medical advice. JAMA. 1979; 242:2758.
10. Long JP, Marin A. Profile of patients signing against medical advice. J Fam Pract. 1982; 15:551–556.
11. Moy E, Bartman BA. Race and hospital discharge against medical advice. J Natl Med Assoc. 1996; 88:658–660.
12. Pennycook AG, McNaughton G, Hogg F. Irregular discharge against medical advice from the accident and emergency department—a cause for concern. Arch Emerg Med. 1992; 9:230–238.
13. Saitz R, Ghali WA, Moskowitz MA. Characteristics of patients with pneumonia who are discharged from hospital against medical advice. Am J Med. 1999; 107:507–509.
14. Schlauch RW, Reich P, Kelly MJ. Leaving against medical advice. N Engl J Med. 1979; 300:22.
15. Crowe DB, Rosse RB, Sheridan MJ, et al. Substance use diagnoses and discharge patterns among psychiatric inpatients. Hosp Community Psychiatry. 1991; 42:403–405.
16. Dalrymple AJ, Fata M. Cross-validating factors associated with discharges against medical advice. Can J Psychiatry. 1993; 38:285–289.
17. Phillips MS, Ali H. Psychiatric patients who discharge themselves against medical advice. Can J Psychiatry. 1983; 28:202–205.
18. Vander Stoep A, Bohn P, Melville E. A model for predicting discharge against medical advice from adolescent residential treatment. Hosp Community Psychiatry. 1991; 42:725–728.
19. Booth BM, Cook CAL, Blow FC. Comorbid mental disorders in patients with AMA discharges from alcoholism treatment. Hosp Community Psychiatry. 1992; 31:415–416.
20. Greenberg WM, Otero J, Villaneuva L. Irregular discharges from a dual diagnosis unit. Am J Drug Alcohol Abuse. 1994; 20:355–371.
21. Pérez de los Cobos J, Trujols J, Ribalta E, et al. Cocaine use immediately prior to entry in an inpatient heroin detoxification unit as a predictor of discharges against medical advice. Am J Drug Alcohol Abuse. 1997; 23:267–279.
22. Ochitill HN, Havassy B, Byrd RC, et al. Leaving a cardiology service against medical advice. J Chron Dis. 1985; 38:79–84.
23. Anis AH, Sun H, Guh DP, et al. Leaving hospital against medical advice among HIV-positive patients. CMAJ. 2002; 167:633–637.
24. Endicott P, Watson B. Interventions to improve the AMA-discharge rate for opiate-addicted patients. J Psychosocial Nurs Ment Health Serv. 1994; 32:36–40.

HIV-positive; injection drug users; against medical advice; methadone; social support

© 2004 Lippincott Williams & Wilkins, Inc.