The association between glycaemic control during hospitalization and risk of adverse events: A retrospective cohort study

Abstract Introduction Hyperglycaemia is common during hospitalization; glycaemic targets in non‐critical care settings have not been well studied. We assessed associations between inpatient glycaemic control and adverse events. Methods We conducted a retrospective cohort study on non‐critically ill medical patients hospitalized in a tertiary care hospital between 2015 and 2018. Mean glycaemia during the first four days of hospitalization was categorized as 4.0–7.0 mmol/L, 7.1–10.0 mmol/L and >10.0 mmol/L. The primary outcome was a composite of adverse events including mortality, infections, acute kidney injury, thromboembolic and cardiovascular events. The secondary outcome was hypoglycaemia, defined as any glycaemia <4.0 mmol/L. Logistic regression was used to assess adverse events, and a Cox proportional hazards model was used to estimate hypoglycaemia risk. Results Our cohort included 1,368 patients, of whom 407 (29.8%) experienced an adverse event. We did not find associations between glycaemia of 4.0–7.0 mmol/L (adjusted odds ratio [OR]: 0.88, 95% confidence interval [CI]: 0.63–1.23) or glycaemia of >10.0 mmol/L (adjusted OR: 0.98, 95% CI: 0.75–1.28) and the occurrence of adverse events, compared to a glycaemia of 7.1–10.0 mmol/L. Glycaemia of >10.0 mmol/L was associated with an increased risk of hypoglycaemia (adjusted hazard ratio [HR]: 1.72, 95% CI: 1.21–2.45). Hypoglycaemia was associated with adverse events (adjusted OR 1.85, 95% CI 1.31–2.60). Conclusions Neither glycaemia of 4.0–7.0 mmol/L nor glycaemia of >10.0mmol/L during non‐critical care hospitalization was associated with increased adverse events. Glycaemia of >10.0 mmol/L was associated with increased hypoglycaemia, likely due to aggressive glucose lowering. These findings highlight the need for further studies to discern optimal inpatient glycaemic targets.


| INTRODUC TI ON
Hyperglycaemia is common among hospitalized patients with a prevalence of up to 38%. 1 Common causes for hyperglycaemia among hospitalized patients include increased secretion of stress hormones, use of glucocorticoids and failure to re-initiate anti-diabetic medications. 2,3 Studies have shown that hyperglycaemia in various clinical settings is associated with adverse patient outcomes, including infections, cerebrovascular and cardiovascular events, prolonged hospital stay and death. 1,[4][5][6][7][8][9][10][11] Potential mechanisms by which hyperglycaemia may lead to these adverse events include impairment of neutrophil and macrophage function, decreasing lymphocytes, enhancing platelet activation, decreasing tissue plasminogen activator and plasma fibrinolytic activity, and impairment of myocardial glucose utilization. [12][13][14][15][16][17] Furthermore, hyperglycaemia has been found to cause endothelial dysfunction and increase oxidative stress. 18,19 Some of these processes have been shown to improve with lowering of glucose levels to normal range. [20][21][22] Nevertheless, in-hospital glycaemic control is often neglected as care is focussed on the underlying presentation of illness. 23 The American Diabetes Association recommends a target random blood glucose of 7.8 mmol/L to 10.0 mmol/L for the majority of hospitalized patients. More stringent goals between 6.1 and 7.8 mmol/L may be appropriate for selected patients if they can be achieved without significant hypoglycaemia. These targets are extrapolated from randomized controlled trials conducted mainly in the critically ill patient population. 24 For non-critically ill patients, the association between glycaemic control and adverse outcomes has not been extensively studied, and the limited number of studies conducted to date has inconsistent results. 25 A meta-analysis of 19 studies by Murad et al. 25 reported no association between intensive glycaemic control and the risk of mortality, myocardial infarction or stroke among non-critically ill hospitalized patients with diabetes; however, the studies included were heterogeneous and the evidence was mainly derived from surgical patients. To further understand the effects of glycaemic control during non-critical care hospitalization in medical patients, we conducted a retrospective study to determine the association between glycaemic control and adverse events among medical patients admitted to non-critical care units.

| Study design and population
We conducted a retrospective cohort study using electronic health records of patients treated at the Jewish General Hospital, a tertiary care teaching hospital for adult patients located in Montréal, Quebec, Canada. For each patient admitted to internal medicine units between 1 January 2015 and 31 December 2018, we obtained the discharge abstract from the medical records department and laboratory data from the biochemistry database.
The discharge abstracts included the primary and secondary diagnoses of each patient, as well as new diagnoses and complications that arose during hospitalization documented by the treating physician. Patients carried a variety of common internal medicine admission diagnoses, including haematology-oncology patients who required hospitalization. All medical diagnoses were recorded using International Classification of Diseases (ICD)-10 codes. The study protocol was approved by the Research Ethics Committee at the Jewish General Hospital, Montréal, Canada.
We included patients aged 18 years or older with at least two capillary glucose measurements performed daily during the first four days of hospitalization. Capillary glucose measurements are performed routinely in patients with a history of diabetes during hospitalization, before meals and at bedtime, and more frequently should hypoglycaemia occur. As such, patients with various types of diabetes were included. Patients with diagnoses of pregnancy, diabetic ketoacidosis and non-ketotic hyperglycaemic-hyperosmolar state at the time of admission were excluded.

| Exposure
The mean glycaemia during the first four days of hospitalization were calculated and classified into three categories for the purpose of this study: 4.0-7.0 mmol/L, 7.1-10.0 mmol/L (reference group) and >10.0 mmol/L. In an attempt to minimize protopathic bias, the first four days of glycaemic data were arbitrarily collected, based on previous studies that demonstrated length of medical hospitalization to be around 6 to 8.5 days. 8,26 Hypoglycaemia was defined as having any glycaemia less than 4.0 mmol/L during hospitalization. 24,27 95% CI: 1.21-2.45). Hypoglycaemia was associated with adverse events (adjusted OR 1.85, 95% CI 1.31-2.60).
Conclusions: Neither glycaemia of 4.0-7.0 mmol/L nor glycaemia of >10.0mmol/L during non-critical care hospitalization was associated with increased adverse events.
Glycaemia of >10.0 mmol/L was associated with increased hypoglycaemia, likely due to aggressive glucose lowering. These findings highlight the need for further studies to discern optimal inpatient glycaemic targets.

K E Y W O R D S
adverse events, cohort study, glycaemic target During the study period, clinical practice in the management of hyperglycaemia was relatively unchanged. The only novel antidiabetic agent introduced was the sodium-glucose cotransporter (SGLT)-2 inhibitors which were available from 2 February 2015 in the Quebec public formulary. 28,29 At the Jewish General Hospital, there are standardized insulin sliding scale protocols that physicians generally prescribe for patients with diabetes. The insulin sliding scale protocol can be adjusted by the treating physician if necessary, to help prevent hypo-or hyperglycaemia during hospitalization.

| Outcomes
The primary outcome was a composite of infections (urinary tract in-

| Sensitivity analyses
The primary analysis was performed to assess the association between having a mean glycaemia of 4.0-7.0 mmol/L or a mean glycaemia of >10.0 mmol/L versus a mean glycaemia of 7.1-10.0 mmol/L during the first four days of hospitalization and the risk of all-cause mortality. The primary and secondary analyses were repeated using average glycaemia during the entire hospitalization rather than only the first 4 days of hospitalization, to assess the association with the risk of primary composite outcome and hypoglycaemia.

| RE SULTS
A total of 1368 patients were included in the study. The distributions for age, sex and serum creatinine were comparable among the groups (Table 1) The majority of adverse events recorded was all-cause mortality in all three exposure groups (Table S1).
Compared with a mean glycaemia of 7.1-10.0 mmol/L, a mean glycaemia of >10.0 mmol/L during the first four days of hospitalization was associated with an increased risk of hypoglycaemia (adjusted HR 1.72, 95% CI 1.21-2.45) ( Table 3). In contrast, a mean glycaemia   data from the entire hospitalization was comparable to using glycaemic data from the first four days of hospitalization (Table S3). Similar to the primary analysis findings, there was no association between average glycaemia during the entire hospitalization and the risk of primary composite outcome (Table S4). There was no association between average glycaemia during the entire hospitalization and the risk of hypoglycaemia (Table S5).

| DISCUSS ION
In this study, there was no association between a mean glycaemia Hypoglycaemia during hospitalization was associated with a nearly twofold higher risk of adverse events.
Although there is growing clinical evidence indicating the need for treating hyperglycaemia among hospitalized patients with diabetes, the management of hyperglycaemia is challenging, and the optimal glycaemia target has not been well studied. 25 Currently suggested targets are difficult to achieve given that it requires more effort and the risk of hypoglycaemia may increase when the glycaemia is targeted to a lower range within normal. Clinical studies have shown that tight glycaemic control may improve outcomes among patients with acute coronary syndrome via reducing oxidative stress and inflammation.. [30][31][32] However, two large randomized clinical trials, DIGAMI and NICE-SUGAR, which involved cardiac and intensive care patients, respectively, have provided conflicting results on the risk of mortality associated with intensive glycaemic control during hospitalization. 33,34 Thus, the practice of using intensive insulin therapy to achieve tight glycaemic control among critically ill patients has not been justified by these studies given such practice may not improve mortality and can increase the risk of hypoglycaemia. 34 Studies performed so far on non-critically ill patients are fewer, and most studies have shown an increased risk of adverse outcomes among hospitalized patients with hyperglycaemia. 1,[4][5][6][7][8][9][10][11]26 In our study, having a glycaemia of >10.0 mmol/L was not found to be associated with an increased risk of adverse outcomes.  The findings from our secondary outcome suggest that having a glycaemia of >10.0 mmol/L during the 4 days of hospitalization was associated with a higher risk of hypoglycaemia, whereas having a glycaemia of 4.0-7.0 mmol/L was not associated with an increased risk of hypoglycaemia. In the literature, having lower glycaemia within the recommended target range is generally associated with a higher risk of hypoglycaemia, with most evidence derived from critically ill or post-myocardial infarction patients. [35][36][37] In our study, the percentage of patients with mean glycaemia 4.0-

| CON CLUS IONS
For non-critically ill patients hospitalized on internal medicine units, neither having a mean glycaemia of 4.0-7.0 mmol/L nor a mean glycaemia of >10 mmol/L was associated with increased risks of adverse events. The result for having a mean glycaemia of >10 mmol/L was unexpected and may be due to increased medical attention given to these patients and timely intervention given to lower glycaemia.
Mean glycaemia of >10 mmol/L was associated with a higher risk of hypoglycaemia, likely attributable to aggressive glucose lowering measures, arguing for more attention on hyperglycaemia management in hospital. Hypoglycaemia during hospitalization is associated with a nearly twofold increase in the risk of adverse events, which may be associated with increased disease severity and emphasizes the need to avoid hypoglycaemia during hospitalization. This study highlights the need for further studies on optimal glycaemic target in the non-critically ill patient population.

ACK N OWLED G EM ENTS
We would like to thank Sophie Dell'Aniello for her helpful advice regarding the statistical analysis of our study and Sylvie Mayer for her help in data extraction for our study.

CO N FLI C T O F I NTE R E S T
All authors declare no conflict of interests relevant to this study.

AUTH O R CO NTR I B UTI O N S
Lan Deng wrote the manuscript. Wusiman Aibibula performed data analysis. Zahra Talat and Shaun Eintracht performed the data collection. All authors contributed to study design, interpretation of data, and reviewed and approved the final manuscript. Oriana Hoi Yun Yu is the guarantor of this work, had full access to the data and takes responsibility for the integrity of the data and data analyses.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used that supports the findings of this study are available from the corresponding author upon reasonable request and after approval is obtained to release data from the Research Ethics Committee at the Jewish General Hospital, Montréal, Canada.