DISCUSSION
Hospital hyperglycemia which is defined as random blood glucose > 7.8 mmol/L with or without diabetes history occurs frequently in inpatients 14-16. Inpatient hyperglycemia has been associated with an increase in mortality, complications, hospital stay length, and overall hospital cost17,18. Previous reports have indicated that the inpatient hyperglycemia incidence is around 32.2% of ICU patients and 32% of patients admitted to non-ICU wards in the U.S.19. The reported incidence of hyperglycemia was 45.7% in China 20. In our study, 26.1% of total adult inpatients were noted to experience hyperglycemia (> 7.8 mmol/L) with 12.8% having a significant episode of hyperglycemia (> 10 mmol/L). The HH and SHH prevalence of ICU inpatients are as high as 73.6% and 30.6%, respectively. Meanwhile, there are 20.7% of non-ICU inpatients with HH and 10.8% with SHH. The inpatient wards with higher hyperglycemia rates can be divided into two categories: the first category is those who tend to admit more patients with pre-existing diabetes history who are at risk for developing excursions in blood glucose control, such as the Cardiology ward; the second category consists of patients without a history of diabetes but who experience illness, stress, and/or medication related hyperglycemia. Surgical and ICU wards typically represent patients falling into the latter category 21,22. As is known, medications other than glucocorticoids can induce hyperglycemia, such as lipid-modifying agents, antidepressants and estrogen 23-26. In our hospital, patients following craniocerebral operations who are admitted to the Neurosurgery ward regularly receive dexamethasone therapy. However, episodes of hyperglycemia were low which appeared to be related to intensive glucose monitoring whenever glucocorticoid therapy was prescribed. This finding seems to support the belief that intensive monitoring of blood glucose is of benefit for inpatient glycemic control9,27,28. Illness associated stress which occurs in 12 -22% of inpatients has been reported to induce hyperglycemia which a resultant increase in both morbidity and mortality5,22,29-31. Stress-induced hyperglycemia is felt to be related to over activation of the hypothalamic–pituitary–adrenal axis and/or the sympathetic autonomic nervous system thereby causing insulin resistance followed by hyperglycemia 32. Sleep loss which is common in inpatients is considered as an independent risk factor which can disrupt glycemic homeostasis 33,34. Our study also indicates that thoracic surgery is unique in character as the rate of mild hyperglycemia (7.8-10mmol/L) is nearly three times higher than the rate of pre-existing diabetes history. This association may be related to the fact that many thoracic surgery cases were for a malignant tumor, and both lung cancer itself and its treatment have been associated with the development of hyperglycemia35-37. Traditional inpatient glycemic care is typically passive in nature and not very efficient14,20,38. Lack of timely reaction to glucose excursions, delayed treatment, and undiagnosed new onset diabetes or pre-diabetes have all been identified as common factors related to poor inpatient glycemic control 39. Our facility developed an alert system for glucose excursions which sent real time cellphone message to the hospitalized patient’s designated inpatient team physician with the aim of improving the reaction time and intervention for episodes of hypo- or hyperglycemia. Previously studied glucose alert systems have shown improvement of inpatient glycemic control40,41. With the HMA system, the monthly glycemia related consultation rate for all inpatients increased nearly 66%. While the HMA system was not noted to reduce the rate of total inpatient hyperglycemic episodes, the HH rate for the surgical wards was decreased. These findings indicate that the effect of HMA system on utilization of formal consultations for blood glucose excursions varied by the specialty service of the ward. The surgical wards actively utilized the alert by requesting endocrinology team consultation. It should be noted that two kinds of endocrinologist consultation are performed in our hospital: 1) individual endocrinology consults and 2) glycemic care team (GCT) assessments. The former is the traditional way to deal glucose excursions in Chinese hospitals, and is performed during a single visit by an endocrine physician to the inpatient with recommendations being provided to fulfill the consultation requirement. The GCT is composed of a nurse, a diabetes educator, nutritionist and an endocrinologist. GCT assessment is led by the endocrinologist who takes the initiative to follow-up the patient during the period that glucose excursions are occurring. In order to cover the entire complement of hospitalized patients, the GCT would require a large number of endocrinologists which presently is beyond the capacity of our facility. Therefore, at present, GCT assessments are only available to a select number of inpatients in our hospital. The GCT glucose assessments were noted to correlate with a reduced rate of hyperglycemia and a shorter length of hospital stay, while individual endocrinologist consults were noted to decrease the average length of hospital stay and medical care expenses for inpatients experiencing hyperglycemia 42. However, the HMA system was found to reduce the HH rate for the surgical wards even when just individual endocrinologist consultations were utilized. Within internal medicine wards, a quite different response to HMA alerts was noted. Nephrology was the fourth highest ward receiving HMA alerts but the third lowest ward in increased utilization of endocrinologist consults. The glycemia related consultation rate for Cardiology + CCU ward even declined by 20%. There is rare previous report describing the relationship between inpatient ward response to glycemic alert, hyperglycemia episodes and glycemia related consultations. We postulate that the less active response to the glucose excursions alerts may lead to suboptimal glycemic control secondary to a delay in endocrinologist consultation, or possibly that individual endocrinologist consults were not felt to be effective enough for internal medicine wards patients with more complicated glucose control issues. Future investigation is warranted as to determine the exact causes for variations in utilization of consultations for HMA alerts.
In conclusion, inpatient hyperglycemia which is associated with inferior outcomes and longer hospital stays is commonly encountered in hospitalized adults, and real time glycemic alert systems for inpatients hold the potential to improve glycemic control thereby reducing complications, morbidity and mortality related to excursions in glucose from either hyper- or hypoglycemia. This study revealed the prevalence of hospital hyperglycemia at 26.1% overall, 20.7% in non-ICU inpatients, and 73.6% in ICU inpatients. It is noted that real time glycemic alert cellphone messages to the team physician who is directly responsible for the inpatient with the glucose excursions can increase the glycemia related consultation rate overall and also appears to decrease hyperglycemia episodes in patients on the surgical wards.