Background: Hospital readmission within 30 days of discharge (30-d readmission) is an undesirable outcome. Readmission of patients with diabetes is common and costly. Most of the studies that have examined readmission risk factors among diabetes patients did not include potentially important clinical data.

Objectives: To provide a more comprehensive understanding of 30-day readmission risk factors among patients with diabetes based on predischarge and postdischarge data.

Research design: In this retrospective cohort study, 48 variables were evaluated for association with readmission by multivariable logistic regression.

Subjects: In total, 17,284 adult diabetes patients with 44,203 hospital discharges from an urban academic medical center between January 1, 2004 and December 1, 2012.

Measures: The outcome was all-cause 30-day readmission. Model performance was assessed by c-statistic.

Results: The 30-day readmission rate was 20.4%, and the median time to readmission was 11 days. A total of 27 factors were statistically significant and independently associated with 30-day readmission (P<0.05). The c-statistic was 0.82. The strongest risk factors were lack of a postdischarge outpatient visit within 30 days, hospital length-of-stay, prior discharge within 90 days, discharge against medical advice, sociodemographics, comorbidities, and admission laboratory values. A diagnosis of hypertension, preadmission sulfonylurea use, admission to an intensive care unit, sex, and age were not associated with readmission in univariate analysis.

Conclusions: There are numerous risk factors for 30-day readmission among patients with diabetes. Postdischarge factors add to the predictive accuracy achieved by predischarge factors. A better understanding of readmission risk may ultimately lead to lowering that risk.