Discharge Medication Complexity Effect on 30-Day Heart Failure Readmission

Objective: To evaluate discharge medication regimen complexity index’s (MRCI) effect on heart failure (HF) patients being rehospitalized 30 days after discharge at a large tertiary teaching facility.

Methods: The institutional review board approved this single center, retrospective, cohort study. The University HealthSystem Consortium (UHC) database was used to identify HF patients from January 2011 to December 2013. Inclusion criteria consisted of men and women aged 18 and older who had a primary discharge diagnosis of HF, a discharge medication list, and discharged to home. Patients were excluded if they had received a heart transplant or left ventricular assist device, and/or died during hospitalization. A 30-day readmission was defined as being readmitted to the same hospital within 30 days of discharge with a principal discharge diagnosis of HF. Only patient’s index admission was included in the study, and patients admitted within the first and last 30 days of the study were excluded to avoid excluding a 30-day readmission. A pilot analysis was conducted involving randomly selected 55 patients to compare manual MRCI collection tool and an automated scoring MRCI system. Multivariable logistic regression was used to examine MRCI effect on 30 day rehospitalization after controlling for other factors.

Results: The pilot analysis revealed that the manual and automated MRCIs were moderately correlated with an R of 0.74 and R2 of 0.55. For the main analysis, a total of 1,455 patients were included in the study with 81 patients (5.6%) readmitted within 30 days of discharge. In the 30-day readmission group, 54 (67%) patients were male with a mean age of 67, and 783 (57%) patients were male with a mean age of 68 in the non-readmit group. Bivariate analysis revealed no statistically significant difference in MRCI in patients readmitted within 30 days of discharge versus patients not readmitted (MRCI: 14.5 versus 13, p=0.13). However, significantly more patients had systolic HF and coronary artery disease (CAD) in the 30-day readmit group [SHF: 57 (70.4%) versus 761 (55.4%), p<0.01; CAD: 27 (33.3%) versus 264 (19.2%), p<0.01]. The variables (p<0.25) included in the multivariate logistic regression analysis consisted of sex, systolic HF, diabetes mellitus (DM), hypertension (HTN), CAD, chronic kidney disease (CKD), length of stay, MRCI, ACEI or ARB, digoxin, and loop diuretics. The multivariate logistic regression analysis revealed that patients prescribed ACEI or ARB were less likely to be readmitted 30 days after discharge (OR: 0.59; CI: 0.36-0.96), and patients with CAD were more likely to be readmitted 30 days after discharge (OR: 1.70; CI: 1.00-2.89). However, there was no effect of MRCI on 30 day rehospitalization after controlling for other factors.

Conclusion: The automated MRCI score was moderately correlated with manual MRCI score. Although ACE/ARB and CAD were significantly associated with 30 day readmission for HF, the automated MRCI was found non-significant. More research is needed to automate MRCI and to evaluate its utility in clinical care.