Relationship of HIV care engagement to emergency department utilization.

To determine if HIV primary care engagement (PCE) is associated with Emergency Department (ED) utilization, a retrospective cohort study using the South Carolina HIV surveillance data from January 1986 to December 2006 linked to the hospital discharge data-set was used to assess utilization at statewide EDs during January 2007-December 2010. Suboptimal PCE was defined as <2 reports of a CD4 + T-cell count or viral load value to surveillance in each calendar year from January 2007 to December 2010. Multivariable logistic regression explored associations of HIV PCE with ED utilization after accounting for sociodemographic characteristics and disease stage. Poisson and negative binominal regression examined PCE, sociodemographic characteristics, and disease stage on the frequency of ED utilization. Suboptimal PCE was associated with increased odds of ED utilization for NIR/NRR (no identified risk/no risk reported; aOR [adjusted odds ratio] = 2.25; CI = 1.69-2.99), self-payers (aOR = 1.81; CI = 1.38-2.39), and those diagnosed with an AIDS-defining illness (ADI; aOR = 1.51; CI = 1.14-2.00), who also had the most median ED visits (six). More ED visits were associated with young age, female (incidence rate ratio [IRR] = 1.16; CI = 1.06-1.27), ADI (IRR = 2.17; CI = 1.93-2.45), Medicaid recipients (IRR = 1.34; CI = 1.21-1.49), indigent/charity recipients (IRR = 1.86; CI = 1.57-2.21), or AIDS > 1 year (IRR = 1.23; CI = 1.13-1.35). Fewer visits to the ED were associated with MSM (males having sex with males IRR = 0.81; CI = 0.72-0.90), NIR/NRR (IRR = 0.86; CI = 0.78-0.95), self-payers (IRR = 0.56; CI = 0.50-0.62), or Medicare recipients (IRR = 0.85; CI = 0.77-0.95). Disease stage and insurance type were differentially associated with primary care and ED utilization. There is a need to evaluate HIV primary care systems to increase access and develop interventions to reduce preventable ED visits.