OBJECTIVES To examine the relationship between payer status and potential confounders on DAMA using the National Hospital Discharge Survey. METHODS Retrospective analysis of a random, nationally representative sample of inpatient discharges in the US from 2004 to 2006. Patients represent discharges from short-stay non-institutional hospitals in the US. Newborns and patients with an undefined payer status were excluded. Only patients with a routine or DAMA discharge status were included. A logistic regression analysis was performed modeling DAMA versus routine discharge as the outcome variable, with payer status as the primary predictor, and confounder variables, including gender, race, marital status, hospital type, geographic region, admission type and source, and drug abuse/addiction. Logistic regression analysis was also performed after removing drug/alcohol abuse/addiction patients. RESULTS From 2004 to 2006 DAMA accounted for 1.23% (n = 946,232) of all hospital discharges when drug and alcohol abuse/addiction patients are included and 0.76% (n = 510,556) when excluded. Controlling for the confounder variables, public (AOR = 0.72, 95% CI = 0.61-0.80) and privately (AOR = 0.46, 95% CI = 0.40-0.52) insured patients are significantly less likely to DAMA than self-pay patients. Consistent with the literature, males, African-Americans, private hospitals, emergency admissions, and drug (AOR = 4.57, 95% CI = 4.22-4.96) and alcohol (AOR = 5.00, 95% CI = 4.46-5.61) abuse/addiction are significant risk factors for DAMA. Many of these relationships become stronger when substance abuse cases are excluded. CONCLUSIONS Considerable evidence exists that initiatives may be needed to better manage self-pay patients and ensure they do not DAMA. Further research should be done with longitudinal data to examine the relationship between DAMA and health outcomes. Prior research indicates that DAMA predictors may also be responsible for non-compliance of prescribed drug therapy, especially following discharge. Therefore the potential for early pharmacist involvement has also been identified.