Our study looked at the association between changes in serum albumin (SA) levels during hospitalization and incidence of hypoglycaemia among non-critically ill patients.

Included were patients discharged from internal medicine units with hospital stays ≤ 14 days. Patients were allocated to three groups: (1) admission SA > 3.5 g/dL with no decrease during hospitalization; (2) admission SA < 3.5 g/dL with no decrease during hospitalization; and (3) decrease in SA regardless of admission SA level. Incident hypoglycaemia (glucose ≤ 70 mg/dL) was predicted by applying regression analysis, using hypoglycaemia as a dependent variable. Mortality studies were performed using Cox regression.

Included were 7718 patients (mean age 71.8 ± 17.4 years, 49.9% males, 27.1% with diabetes). Of these patients, 12.7% had at least one documented hypoglycaemia episode during hospitalization. Patients with decreases in SA levels during hospitalization (group 3) had higher rates of incident hypoglycaemia compared with patients in groups 1 and 2 (21.0% vs. 6.0% and 16.3%, respectively; P < 0.001 for both). Results remained significant after controlling for admission SA. Strong negative correlations were observed between SA and serum osmolarity (r = −0.204, P < 0.0001) and, separately, between changes in SA with changes in serum osmolarity (r = −0.157, P < 0.001), indicating that SA changes were not due to haemodilution. Overall 1-year mortality was 16.7%, and Cox regression analysis showed an increased 1-year mortality in patients in group 3 (27.9%) compared with those in groups 1 and 2 (15.2% and 13.8%, respectively).

Changes in SA during hospitalization are associated with an increased risk of hypoglycaemia during hospitalization of non-critically ill patients.