Abstract 15796: Predictors of 30-Day Readmission among Acute Myocardial Infarction Patients do not Include the Red Cell Distribution Width

Abstract

Background: Mortality and other major adverse events, including myocardial infarction (MI), are predicted by the red cell distribution width (RDW). With new reimbursement penalties for worse-than-expected readmission rates of acute MI patients within 30 days of discharge, adequate prediction is needed for which patients have a high readmission risk. This study tested if RDW predicts 30-day readmission (30dR) among acute MI patients.

Methods: Hospitalized patients presenting with acute MI in 1993-2012 (N=6,028) who were enrolled in the Intermountain Heart Collaborative Study were studied. RDW, other complete blood count components, basic metabolic profile components, age, sex, cardiac risk factors, clinical data, comorbidities, treatment variables, 30dR, and 30-day mortality were queried from electronic medical records. Mortality was supplemented by data from Social Security death records and Utah death certificates. Cox regression evaluated association of RDW with 30dR and 30-day mortality adjusted for covariables.

Results: Age averaged 63.4±12.7 years and 26.9% were female. Overall, 1,701 (28.2%) patients were readmitted and 328 patients died. Continuous RDW values were associated with 30dR in univariable analysis (hazard ratio [HR]=1.036 per +1% of RDW, 95% CI=1.01, 1.07; p=0.013) but not after full adjustment, (HR=0.98 per +1% of RDW, 95% CI=0.95, 1.01; p=0.23). This was also found for RDW quintiles: adjusted HR=0.95 (p=0.61), 1.13 (p=0.21), 1.02 (p=0.83), and 0.98 (p=0.87) for quintiles 2, 3, 4, and 5 vs. 1. Predictors of 30dR were age, diabetes, not smoking, heart failure, renal failure, prior MI, atrial fibrillation, depression, bypass grafting, statins, beta-blockers, ACE inhibitors, hematocrit, MCHC, sodium, bicarbonate, and creatinine. In contrast, RDW was associated with mortality in univariable (HR=1.100 per +1%, CI=1.05, 1.15; p<0.001) and multivariable (HR=1.087 per +1%, CI=1.03, 1.15; p=0.004), with adjusted HR=1.70 (p=0.014) for quintile 5 vs. 1.

Conclusions: RDW was not associated with 30dR among acute MI patients, but strongly predicted 30 day mortality. Further investigations should examine whether 30dR can be predicted by a risk score of patient characteristics using those factors that were associated with 30dR.