Background. Hospitalized persons who inject drugs are at a greater risk of adverse hospital outcomes including discharge against medical advice, inpatient illicit drug use, overdose, and death. However, there are limited data on the frequency and outcomes of these events in the United States. Methods. This retrospective analysis included patients with injection-related infections receiving a protocol for injection drug use (IDU) at University of Alabama at Birmingham Hospital from 2016 to 2017. In-hospital IDU was suspected or reported drug usage plus confirmatory drug screen, and documented discharges “against medical advice” were deemed patient-directed discharges (PDD). We analyzed the frequency of and associations between in-hospital IDU, PDD, 30-day readmission, and deaths (between 2016 and 2019) using McNemar’s tests. Logistic regression models evaluated the association between PDD, in-hospital IDU, readmission, and death. Results. Overall, 83 patients met inclusion criteria: 28 (34%) with in-hospital IDU, 12 (14%) PDD, 9 (11%) died, and 12 (14%) 30-day readmission. In-hospital IDU was significantly associated with PDD (P = .003), 30-day readmission (P = .005), and death (P = .0003). Patient-directed discharges and 30-day readmission were not significantly associated with death nor with each other. Conclusions. In a cohort of patients receiving inpatient care for injection-related infections, illicit drug use, PDD, 30-day readmissions, and death were common. Furthermore, patients who use illicit drugs while hospitalized are significantly more likely to leave early, be readmitted, and/or die. We must design models of care that prevent adverse outcomes, including drug use and PDD, to reduce barriers to evidence-based treatment of infections.