Association of Glucose Concentrations at Hospital Discharge With Readmissions and Mortality: A Nationwide Cohort Study

Abstract

Context: Low blood glucose concentrations during the discharge day may affect 30-day readmission and posthospital discharge mortality rates.

Objective: To investigate whether patients with diabetes and low glucose values during the last day of hospitalization are at increased risk of readmission or mortality.

Design and outcomes: Minimum point of care glucose values were collected during the last 24 hours of hospitalization. We used adjusted rates of 30-day readmission rate, 30-, 90-, and 180-day mortality rates, and combined 30-day readmission/mortality rate to identify minimum glucose thresholds above which patients can be safely discharged.

Patients and setting: Nationwide cohort study including 843,978 admissions of patients with diabetes at the Veteran Affairs hospitals 14 years.

Results: The rate ratios (RRs) increased progressively for all five outcomes as the minimum glucose concentrations progressively decreased below the 90 to 99 mg/dL category, compared with the 100 to 109 mg/dL category: 30-day readmission RR, 1.01 to 1.45; 30-day readmission/mortality RR, 1.01 to 1.71; 30-day mortality RR, 0.99 to 5.82; 90-day mortality RR, 1.01 to 2.40; 180-day mortality RR, 1.03 to 1.91. Patients with diabetes experienced greater 30-day readmission rates, 30-, 90- and 180-day postdischarge mortality rates, and higher combined 30-day readmission/mortality rates, with glucose levels <92.9 mg/dL, <45.2 mg/dL, 65.8 mg/dL, 67.3 mg/dL, and <87.2 mg/dL, respectively.

Conclusion: Patients with diabetes who had hypoglycemia or near-normal glucose values during the last day of hospitalization had higher rates of 30-day readmission and postdischarge mortality.

Figure 1.

Study flow diagram. ICU, intensive care unit.

Figure 2.

Relation of 17 glucose-concentration categories to mortality, readmission, and readmission or mortality rates. The red circles represent adjusted event rates; the red lines are 95% CIs. The blue lines are smoothed rates obtained by fitting the adjusted rates to a piecewise, continuous nonlinear regression with a single knot, resulting in two straight lines meeting at the knot. The knot can be interpreted as the glucose concentration separating normal glucose concentrations from hypoglycemia. The analyses are adjusted for age, BMI, BMI², admission source, admitting service, diabetes medications, comorbidities (i.e., cardiac arrythmia, CHF, fluid or electrolyte disorder, hypertension, metastatic cancer, solid tumor without metastasis, renal failure, and weight loss).

Figure 3.

Relation of 17 glucose-concentration categories to mortality, readmission, and readmission or mortality rates. The red circles represent adjusted event rates; the red lines are 95% CIs. The blue lines are smoothed rates obtained by fitting the adjusted rates to a piecewise, continuous nonlinear regression with a single knot, resulting in two straight lines meeting at the knot. The knot can be interpreted as the glucose concentration separating normal glucose concentrations from hypoglycemia. The analyses are adjusted for age, BMI, BMI², sex, admission source, admitting service, diabetes medications, comorbidities [hypothyroidism, lymphoma, liver disease, paralysis, pulmonary circulatory disorder, renal failure, peripheral vascular disease, vascular disorder, cardiac arrhythmia, neurologic disorder, fluid or electrolyte disorder, deficiency anemia, alcohol abuse, drug abuse, depression, chronic obstructive pulmonary disease, psychoses, blood-loss anemia, coagulopathy, rheumatologic diseases (i.e., rheumatoid arthritis/collagen vascular disease), HIV/AIDS, peptic ulcer, CHF, metastatic cancer, solid tumor without metastasis, myocardial infarction].