Introduction: Linkage to care following a human immunodeficiency virus (HIV) diagnosis is critical. In the U.S. only 69% of patients are successfully linked to care, which results in delayed receipt of antiretroviral therapy leading to immune system dysfunction and risk of transmission to others. Methods: We evaluated predictors of failure to link to care at a large urban healthcare center in Philadelphia in order to identify potential intervention targets. We conducted a cohort study between May 2007 and November 2011 at hospital-affiliated outpatient clinics, emergency departments (EDs), and inpatient units.

Results: Of 87 patients with a new HIV diagnosis, 63 (72%) were linked to care: 23 (96%) from the outpatient setting and 40 (63%) from the hospital setting (ED or inpatient) (p<0.01). Those who were tested in the hospital-based settings were more likely to be black (p=0.01), homeless (p=0.03), and use alcohol or drugs (p=0.03) than those tested in the outpatient clinics. Patients tested in the ED or inpatient units had a 10.9 fold (p=0.03) higher odds of failure to link compared to those diagnosed in an outpatient clinic. When testing site was controlled, unemployment (OR 12.2;p<0.01) and substance use (OR 6.4;p<0.01) were associated with failure to link.

Conclusion: Our findings demonstrate the comparative success of linkage to care in outpatient medical clinics versus hospital-based settings. This study both reinforces the importance of routine opt-out HIV testing in outpatient practices, and demonstrates the need to better understand barriers to linkage.