Leaving Hospital Against Medical Advice Increases Death Risk

Norra MacReady

August 26, 2013

A large, population-based study suggests that leaving the hospital against medical advice increases a patient's risk for readmission or even death.

In an analysis of more than 1.9 million hospital admissions over the course of 20 years, leaving the hospital against medical advice (AMA) more than doubled the odds of 90-day mortality and 30-day readmission compared with complying with medical advice, lead author Allan Garland, MD, from the Departments of Internal Medicine and Community Health Sciences and the Manitoba Center for Health Policy, University of Manitoba, Winnipeg, Canada, and colleagues explain in an article published online August 26 in the Canadian Medical Association Journal.

The adverse effects of premature hospital discharge became apparent as early as 1 week later and persisted for up to 180 days, suggesting that "they are not solely a result of incomplete treatment of acute illness," the authors write, adding that "reducing the persistently elevated risk will likely require longitudinal interventions extending beyond hospital admission."

Dr. Garland and colleagues examined all unscheduled hospital admissions ending in live discharge among adults in Manitoba from April 1, 1990, to February 28, 2009. They compared rates of hospital readmission within 30 days and death within 90 days among patients who left AMA with those of people who stayed in the hospital as recommended (non-AMA).

Of 1,916,104 index admissions among 610,187 people during the study period, 21,417 (1.1%) resulted in patients leaving AMA. There were 47,613 30-day readmissions, including 5076 among patients who left AMA (23.7%) and 42,537 (2.3%) among non-AMA patients (P < .001). AMA patients had a mean age of 42.7 years (standard deviation [SD], 18.0 years) vs a mean of 54.1 years (SD, 22.2 years) for non-AMA patients. Mental disorder was the main admitting diagnosis for 4645 (21.7%) of the AMA patients compared with 102,472 (5.4%) non-AMA patients. For non-AMA patients, the main admitting diagnosis was pregnancy-related complications, seen in 353,508 cases (18.7%), compared with 2227 (10.4%) of the AMA patients (P < .001 for both comparisons).

AMA patients had a cumulative 30-day readmission rate of 24.0% compared with 12.1% for non-AMA patients (P < .001), for a difference of 11.9% (95% confidence interval [CI], 11.3% - 12.4%). By 180 days, the differences were 43.3% and 26.4%, respectively, for a difference of 16.9% (95% CI, 16.1% - 17.5%). In a logistic regression analysis, AMA status was associated with an odds ratio of 30-day readmission of 3.04 (95% CI, 2.79 - 3.30) compared with non-AMA status. In addition to discharge status, the analysis was adjusted for age, sex, year of index admission, diagnosis on index admission, comorbid conditions, length of index admission, number of any previous admissions within the previous 5 years, and household income.

In a conditional logistic regression analysis, AMA discharge was associated with an odds ratio for 90-day mortality of 2.51 (95% CI, 2.18 - 2.89) compared with non-AMA discharge. This analysis was adjusted for AMA discharge status, year of index admission, comorbid conditions, household income, number of admissions within the previous 5 years and the existence of any AMA discharges within the previous 5 years. Exact matching occurred on age and sex.

Study limitations include imperfect identification of AMA status from hospital abstracts and the use of only a small sample of abstracts; the inability to adjust the analysis for the severity of the acute illness; the inability to determine whether hospital readmissions and deaths were related to the original admission; and lack of information about patient ethnicity, the authors write. In addition, findings derived from a publicly funded Canadian healthcare system may not be generalizable to other populations.

Nevertheless, they conclude, "Leaving the hospital against medical advice was associated with increased risks of readmission to hospital and death that persisted for at least 6 months."

This study was supported by grants from the Manitoba Health Research Council and the Winnipeg Health Sciences Centre Foundation/Department of Research. The authors have disclosed no relevant financial relationships.

CMAJ. Published online August 26, 2013.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: