ABSTRACT
AIMS: We compared characteristics of patients with hyperglycemic hyperosmolar state (HHS) and patients with severe hyperglycemia without the signs of hyperosmolarity and ketoacidosis; analyzed long-term all-cause mortality and potential prognostic factors. METHODS: The studied population included 261 749 adults. HHS was diagnosed in patients with plasma glucose >33.0 mmol/L, ketonuria <1+, and serum osmolarity >320 mmol/L. Patients with plasma glucose >33.0 mmol/L, ketonuria <1+ and serum osmolarity <320 mmol/L were considered as controls (nHHS). RESULTS: During the 5-year period, we observed 68 episodes of HHS in 66 patients and 51 patients with nHHS. Patients with HHS were significantly older, had lower BMI, higher serum C-reactive protein and used diuretics and benzodiazepines more frequently. Mortality rates one, three and 12 months after admission were 19.0, 32.1 and 35.7% in the HHS group, and 4.8, 6.3 and 9.4% in the nHHS group (P<0.001). However, after adjustment for patient age, these differences were not statistically significant. In multivariate Cox regression in HHS group, mortality was positively associated with age, male gender, leukocyte count, amylase, presence of dyspnea and altered mental status, and the use of benzodiazepines, ACE inhibitors and sulphonylureas, while it was inversely associated with plasma glucose, bicarbonate, and the use of thiazides and statins. A nomogram derived from these variables had an accuracy of 89% in predicting lethal outcome. CONCLUSIONS: Infection, use of furosemide and benzodiazepines may be important precipitating factors of HHS. Prospective clinical trials are mandatory to analyze the safety of ACE-inhibitors and benzodiazepines in elderly patients with diabetes.
