Blood Sugar Targets in Surgical Intensive Care

Abstract

Background

30–80% of patients being treated in intensive care units in the perioperative period develop hyperglycemia. This stress hyperglycemia is induced and maintained by inflammatory-endocrine and iatrogenic stimuli and generally requires treatment. There is uncertainty regarding the optimal blood glucose targets for patients with diabetes mellitus.

Methods

This review is based on pertinent publications retrieved by a selective search in PubMed and Google Scholar.

Results

Patients in intensive care with pre-existing diabetes do not benefit from blood sugar reduction to the same extent as metabolically healthy individuals, but they, too, are exposed to a clinically relevant risk of hypoglycemia. A therapeutic range from 4.4 to 6.1 mmol/L (79–110 mg/dL) cannot be justified for patients with diabetes mellitus. The primary therapeutic strategy in the perioperative setting should be to strictly avoid hypoglycemia. Neurotoxic effects and the promotion of wound-healing disturbances are among the adverse consequences of hyperglycemia. Meta-analyses have shown that an upper blood sugar limit of 10 mmol/L (180 mg/dL) is associated with better outcomes for diabetic patients than an upper limit of less than this value. The target range of 7.8–10 mmol/L (140–180 mg/dL) proposed by specialty societies for hospitalized patients with diabetes seems to be the best compromise at present for optimizing clinical outcomes while avoiding hypoglycemia. The method of choice for achieving this goal in intensive care medicine is the continuous intravenous administration of insulin, requirng standardized, high-quality monitoring conditions.

Conclusion

Optimal blood sugar control for diabetic patients in intensive care meets the dual objectives of avoiding hypoglycemia while keeping the blood glucose concentration under 10 mmol/L (180 mg/dL). Nutrition therapy in accordance with the relevant guidelines is an indispensable prerequisite.

cme plus

This article has been certified by the North Rhine Academy for Continuing Medical Education. Participation in the CME certification program is possible only over the internet: cme.aerzteblatt.de. The deadline for submission is 23.09.2022.

Intensive care patients frequently develop a metabolic condition known as stress-induced hyperglycemia, regardless of any pre-existing diabetes. A US American study involving 12 559 305 point-of-care measurements reported a prevalence of hyperglycemia (>10 mmol/L [>180 mg/dL]) in 46% of the intensive care patients (1). This proportion is even higher among surgically treated patients: in cardiac surgery up to 80% (2), in liver and pancreas surgery 30 and 60%, respectively (3). In the non-critical care setting, around 25% of patients are affected (4). A trauma registry-based study involving hospitalized patients with diabetes revealed uncontrolled blood sugar levels (>13.9 mmol/L [>250 mg/dL]) in >30% of cases (5). A Swedish observational study (6) identified chronic dysglycemia (HbA1c ≥ 6% or diabetes) at the time of admission in 33% of 943 intensive care unit (ICU) patients. Of these 312 patients, 27% had prediabetes (HbA1c 6.0–6.4%), 14% had undiagnosed diabetes mellitus with an HbA1c of at least 6.5%, and 59% had a history of diabetes. So, there was a combination of chronic dysglycemia and acute stress-induced hyperglycemia in one third of the intensive care unit patients. It has been shown that patients with diabetes have increased hospital morbidity and length of stay as well as increased perioperative mortality (7– 9). Furthermore, diabetes is a known predictor of reduced long-term survival after sepsis (10). With intensive glucose control (i.v. insulin therapy at 6.1 mmol/L [110 mg/dL], target range 4.4–6.1 mmol/L [80–110 mg/dL], Table 1), the Belgian Leuven-I landmark study (11) reduced ICU mortality from 8 to 4.6% (number needed to treat [NNT] 30) (control group: insulin therapy at blood glucose levels >11.9 mmol/L [>215 mg/dL], target range 10–11.1 mmol/L [180–200 mg/dL]). In addition, further studies (12– 14) showed a U-shaped or J-shaped curve for the association between mortality risk and blood glucose levels in critically ill patients (figure 1). There was a similar, but flatter, curve for patients with diabetes (15– 17, Figure 1). The treatment approach of intensive insulin therapy (IIT) was strictly adopted in the management of patients in need of intensive care (18, 19). The multicenter NICE-SUGAR study (20), Leuven-II (21) and other single-center randomized controlled studies (RCT) were unable to reproduce the mortality-lowering effect of IIT from the Leuven-I study (table 1).

Table 1. Overview of the most important randomized controlled studies evaluating glycemic control of critically ill patients*¹.

	Number of study participants *2	Intervention group*3	Control group *3	Primary outcome	Outcome	Comments
Single-center studies
van den Berghe et al. 2001 (11) Leuven I	765/783	4.4–6.1 (79–110)	11–11.1 (180–200)	ICU mortality	intervention group superior	after combined assessment of both studies (e5): – for the subgroup of the 407 patients with pre-existing diabetes, in-hospital mortality was not significantly different for the achieved mean blood glucose target range (110–150 mg/dL – 21.2%, >150 mg/dL 21.6%, <110 mg/dl 26.2%) – incidence of hypoglycemia not significantly different from other subgroups
Van den Berghe et al. 2006 (21) Leuven II	595/605	4.4–6.1 (79–110)	11–11.1 (180–200)	ICU mortality	no difference; intervention for group of long-term ICUpatients superior
Arabi et al. 2008 (22)	266/257	4.4–6.1 (79–110)	10–11 (180–198)	ICU mortality	no difference
De La Rosa et al. 2008 (23)	254/250	4.4–6.1 (79–110)	10–11 (180–198)	28-daymortality	no difference
Multicenter studies
Brunkhorst et al. 2008 (24)	247/289	4.4–6.1 (79–110)	10–11 (180–198)	28-daymortality and SOFA score	no difference
Finfer et al. 2009 (20) NICE-SUGAR study	3054/3050 or 3015/3014 patients finally analyzed	4.4–6.1 (79–110)	7.8–10 (141–180)	90-daymortality	control group superior	1211 diabetic patients (predefined as subgroup) no significant differences with regard to outcome and side effects of IIT as compared with total study population
Preiser et al. 2009 (25)	542/536	4.4–6.1 (79–110)	7.8–10 (141–180)	ICU mortality	no difference
Kalfon et al. 2014 (26)	1336/1312	4.4–6.1 (79–110)	<10 (<180)	90-daymortality	no difference

*¹ modified from Mesotten et al. 2015 (27); *² Intervention/control group; *³ blood glucose target range (mmol/L or mg/dL); IIT, intensive insulin therapy; ICU, intensive care unit;

SOFA, Sequential Organ Failure Assessment

Figure 1.

Association between mortality and blood glucose levels for critically ill patients. With no pre-existing diabetes mellitus, the curve shows lowest mortality in the range of normal blood glucose levels. This point is displaced towards higher blood sugar levels in diabetic patients. Exactly in which range is the subject of controversial debate. Figure modified from Gunst et al. 2019 (e36).

The NICE-SUGAR study showed increased 90-day mortality under IIT in comparison with the control group (27.5% versus 24.9%; odds ratio [OR] 1.14; 95% confidence interval: [1.02; 1.28]; p = 0.02), primarily as a result of more frequent deaths from cardiovascular causes in the intervention group (41.6% versus 35.8%; p = 0.02). A direct association with the significantly increased incidence of hypoglycemia in the intervention arm (6.8 % versus 0.5 %; OR 14.7; [9.0; 25.9]) was not confirmed. Operative patients were particularly affected by excess mortality. No other subgroup effects were observed.

This inconsistency in the study results led to an ongoing discussion regarding the optimal blood glucose target range for diabetic patients in need of intensive care. While taking into consideration existing evidence and guidelines, the present article aims to provide practical recommendations on blood glucose monitoring for critically ill patients, focusing on patients with diabetes.

Methods

A selective search of the literature was conducted using PubMed and Google Scholar, with an emphasis on the following (variously combined) terms: “Diabetes”, “Critical Care”, “Blood Glucose”, “Glucose Control”, “Glucose Monitoring”, “Glucose Management”, “Intensive Care Unit”, “ICU”, “Hypoglycemia” and “Outcome”. Primary research and review articles published between 2001 and February 2021 and guidelines from leading professional societies were included (German Diabetes Society [Deutsche Diabetes Gesellschaft, DDG], American Diabetes Association [ADA], German Society for Nutritional Medicine [Deutsche Gesellschaft für Ernährungsmedizin, DGEM], European Society for Clinical Nutrition and Metabolism [ESPEN], and American Society for Parenteral and Enteral Nutrition [ASPEN]).

Basic principles and diagnostic approach

Pathophysiologically, the metabolic state of stress-induced hyperglycemia is due to endocrine effects caused by cortisol, glucagon and somatropin and to an acute phase response of the innate immune system secondary to major surgery or critical illness. In patients with diabetes, the resulting acute insulin resistance (28, 29) encounters an already defective glucose regulation (figure 2). There is no generally accepted definition to date of stress hyperglycemia (4, 10, 30). Long-standing considerations stipulate for its diagnosis blood glucose thresholds of 6.9 mmol/L (124 mg/dL) for fasting and 11.1 mmol/L (200 mg/dL) for random measurements in patients without known diabetes (31, 32). Recent concepts assume that the stress hyperglycemia ratio (SHR, relative hyperglycemia) is a predictor for in-hospital mortality. It is calculated by dividing the admission blood glucose level by the average glucose level (derived from the current HbA1c value) (5, 33). The lowest point of ICU mortality was for an SHR of 0.8–1.0 for patients with an HbA1c >6.5% on admission. A higher or lower SHR increased ICU mortality (34).

Figure 2.

Concept of stress-induced hyperglycemia with pre-existing diabetes. Insulin resistance and deficiency cause reduced glucose uptake at insulin-dependent organs, increased proteolysis and dyslipidemia (peripheral insulin resistance). Excessive glucose uptake in the liver results in lipogenesis stimulation with the risk of developing non-alcoholic fatty liver disease (NAFLD). In addition, gluconeogenesis is disinhibited (central insulin resistance). Perioperatively, inflammatory stress and iatrogenic interventions increase insulin resistance and subsequently hyperglycemia (stress-induced hyperglycemia), which in turn can potentially influence clinical endpoints. Depiction modified from (28, e37).

SIRS, systemic inflammatory response syndrome

The ADA defines a threshold of 7.8 mmol/L (140 mg/dL) as hyperglycemic for hospitalized patients with diabetes (30). The DDG sees a need for action for blood glucose levels >10 mmol/L (>180 mg/dL) and suggests a target range of 7.8–10 mmol/L (140–180 mg/dL) (9). At any rate, risk stratification by determining HbA1c on admission to hospital or the ICU is to be recommended (4, 30, 35). Firstly, this would allow a previously unrecognized diabetes to be diagnosed. And secondly, there is the possibility of individually tailoring the intensity of glucose-lowering treatment. Pre-existing diabetes can be assumed, or an as yet unrecognized diabetes can be diagnosed, from an HbA1c of 6.5% or higher (35). According to the DDG’s regularly evaluated recommendations for practice, factors influencing and interfering with HbA1c measurement should be carefully heeded to avoid misdiagnosis (36, 37).

A retrospective observation study involving 3084 critically ill patients demonstrated an association between pre-morbid glycemic control (HbA1c levels <6.5, 6.5–7.9, ≥ 8.0%) and the incidence of severe hypoglycemia (0.9, 2.5, 4.3 %; p <0.001) (38). In turn, these were associated with increased in-hospital mortality. It should now be possible to answer the question of whether an individualized approach can improve the outcome of critically ill patients following publication of the results of an already completed multicenter RCT (CONTROLING, NCT02244073).

No treatment without side effects

The drug of choice for managing stress-induced hyperglycemia is insulin. However, its use – regardless of whether the patient is diabetic – is associated with the highest rate of hypoglycemia (39). At least 10% of documented medication errors resulted in insulin-related hypoglycemia (9, 39).

A meta-analysis examined the connection between glycemic control, mortality, and hypoglycemia using the results of 36 RCTs containing the data of 17 996 ICU patients (40) and defined four glycemic control subgroups:

• “tight” (4.4 to ?6.1 mmol/L [79–110 mg/dL])

• “moderate” (6.1 to ?7.8 mmol/L [110–141 mg/dL])

• “mild” (7.8 to <10 mmol/L [141–180 mg/dl])

• “very mild” (10 to ?12.2 mmol/L [180–220 mg/dL])

Using the “very mild” subgroup as a reference, no treatment modality was superior to the others with regard to total mortality. Irrespective of treatment type (conservative versus surgical intensive care unit), treatment duration and diabetes status, a fivefold increased risk for hypoglycemia remained when patients were treated with “tight” glycemic control as compared with “mild” or “very mild”.

The most significant risk under IIT for patients with diabetes receiving intensive care is treatment-associated hypoglycemia (e1). The rating of levels between 2.3 and 3.9 mmol/L (41–70 mg/dL) as mild or moderate and ≤ 2.2 mmol/L (40 mg/dL) as severe hypoglycemia has gained international acceptance (e1). In addition, the ADA and the Endocrine Society demanded in 2013 that all episodes of abnormally low plasma glucose levels in diabetic patients that expose the individual to potential harm be defined as iatrogenic hypoglycemia (e1). Setting uniform limits for diabetic patients is problematic since even relative hypoglycemia (≥ 30% decrease below mean prehospital admission levels) can be associated with negative effects (e2, e3). The accepted threshold is ≤ 3.9 mmol/L (≤ 70 mg/dL) which, according to the ADA, corresponds to level 1 hypoglycemia (30). Beyond this value, particular attention should be directed towards further, often more pronounced hypoglycemia, irrespective of symptoms (e1).

In one of the named studies involving 126 US American centers, at least one level 1 hypoglycemia episode was detected in 10.1% of ICU patient-days and in 3.5% of non-ICU patient-days (1). The ADA considers level 2 hypoglycemia to be blood glucose levels <3.0 mmol/L (<54 mg/dL), which accounts for 5.3% of ICU patient days according to [1]). Level 3 hypoglycemia is defined as a “clinical event characterized by altered mental and/or physical functioning that requires assistance from another person for recovery” (30). Blood glucose levels <2.2 mmol/L (40 mg/dL) were identified in 1.9% of ICU patient days (1). Diabetic patients with severe hypoglycemia have an increased risk of cardiovascular events (absolute risk increase [ARI] for major macrovascular events: 1.3%) and mortality (ARI for death from any cause: 10.5%) (e4). The above examples show why it is absolutely essential to define a lower blood glucose threshold that can be reached by insulin therapy but should not be fallen short of.

Defining a lower blood glucose threshold

If the ADA definition for level-1 hypoglycemia were taken as the lower treatment threshold, then the risk for hypoglycemia would be disproportionately high. Under IIT, the NICE-SUGAR study showed an increased 90-day mortality for the total study population, with a number needed to harm (NNH) of 38, which was affected decisively by cardiac events (20). In the subgroup comparison, 1211 diabetic patients demonstrated no significant difference with regard to adverse reactions (including hypoglycemia) and outcome when compared with non-diabetics and other subgroups (table 1). Nor did a combined analysis of the 407 diabetic patients from the two Leuven studies find any increased hypoglycemia rates in comparison with the other subgroups ([e5], Table 1). The above meta-analysis conducted by Yamada et al. identified the blood glucose target range of 7.8–10 mmol/L (141–180 mg/dL) as most likely to meet the compromise of reducing ICU mortality on the one hand and avoiding hypoglycemia on the other (40). Another meta-analysis looking at the data from 15 RCTs (13 studies included patients with diabetes, five of them exclusively) examined the effect of perioperative glucose target ranges on surgical site infection. It showed as an incidental finding that the hypoglycemia risk (definition here: <4.4 or <2.2 mmol/L [<79 or <40 mg/dL]) was significantly higher (OR: 5.55; [2.58; 11.96]) in the blood glucose target ranges <6.1 mmol/L (<110 mg/dL) and 6–8 mmol/L (108–144 mg/dL) as compared with conventional glycemic control (<12.2 mmol/L [<220 mg/dL]) (e6). Four of the 15 studies, however, reported no cases of hypoglycemia. In the other studies, the risk of developing severe adverse events associated with hypoglycemia (death [OR: 0.74; [0.45; 1.23]], stroke [OR 1.37; [0.26; 7.20]]) was not significantly different on comparing strict and conventional glycemic control.

A problem arises from deriving a lower target blood glucose threshold of 6.1 mmol/L (110 mg/dL) for perioperative care from the data of these studies as a compromise because this value is close to level 1 hypoglycemia. Factors which promote insulin resistance and make it difficult to achieve therapeutic goals should also be considered (figure 2). ESPEN recommends a blood glucose target range of 6–8 mmol/L (108–144 mg/dl) for all intensive care patients (e7). The DDG favors higher threshold values: For patients not at risk from hypoglycemia, levels of 6.1–7.8 mmol/L (110–140 mg/dL) are regarded as tolerable, with values of 7.8–10 mmol/L (140–180 mg/dL) defined as the target range (9). The ADA takes a similar stance (30). Both the NICE-SUGAR study and meta-analyses confirm the range of 8–10 mmol/L (144–180 mg/dl) as one with a low risk of hypoglycemia (20, 40). Thus, it seems feasible in practice to reduce, or pause, intravenous insulin administration as soon as a blood glucose level of 8 mmol/L (144 mg/dL) is reached. Increased monitoring is required for the range of 6.1–8 mmol/L (110–144 mg/dL), and, if below 6.1 mmol/L (110 mg/dL), administration of glucose should be considered, or starting or restarting nutritional therapy. The region ≤ 3.9 mmol/L (≤ 70 mg/dL) should be avoided in all events. This recommendation is supported by the results of a retrospective observational study involving 747 critically ill patients who had had at least one episode of a blood glucose level <3.9 mmol/L (<70 mg/dL): The risk of in-hospital mortality (OR 1.22, 95% CI: [0.77; 1.93], p = 0.39) was independent of the causes of hypoglycemia (spontaneous [induced by most severe illness] or iatrogenic) and diabetes status (e8).

Defining an upper blood glucose threshold

Since stricter glucose control does not generally reduce mortality, while hypoglycemia remains an independent risk factor (39), a debate was sparked (e9, e10) that led, among other things, to the initiation of two small observational studies (e11, e12). They explored the safety of the blood glucose target range of 10–14 mmol/L (180–252 mg/dl) for critically ill diabetic patients. This produced no indication of a deterioration of the clinical outcome, so Luethi et al. followed this with a larger prospective cohort study involving 750 critically ill diabetic patients. A conventional phase of glucose control (insulin when blood glucose levels >10 mmol/L [>180 mg/dL], target range: 6–10 mmol/L [108–180 mg/dL]) was compared with a liberal phase (insulin when blood glucose levels >14 mmol/L [252 mg/dL], target range 10–14 mmol/L [180–252 mg/dL]) (e13). The overall evaluation revealed non-significantly reduced hypoglycemic events (8.6 % versus 6.6 %, p = 0.32) with liberal glucose control. When only patients with an HbA1c >7% were analyzed, the effect was more pronounced with liberal blood sugar control (9.6% versus 4.1%, p = 0.053). The other clinical outcomes did not differ between liberal and conventional blood sugar control (30-day mortality: 11.1% versus 14%, p = 0.25; hours of mechanical ventilation: 16 hrs. versus 19 hrs., p = 0.3).

One argument against higher target ranges is the positive effect of IIT on the reduction of perioperative, potentially life-threatening (when septic), wound infections (e14). Against this background, a meta-analysis evaluating data from 15 RCTs compared strict (<6.1 mmol/L [<110 mg/dL]) and moderately strict glucose control (6.1–8.3 mmol/L [110–149 mg/dL]) in a total of 1442 patients with conventional glucose control (<12.2 mmol/L [<220 mg/dL]) in 1394 patients (e6). Twelve of the evaluated studies included patients with and without diabetes, and four included only diabetic patients; four studies investigated deep wound infections, two studies investigated sternal wound infections, all others investigated wound infections of any type. Compared with the conventional strategy, the risk of developing wound infections was reduced under strict (OR: 0.5; [0.35; 0.73]) and moderately strict (OR: 0.27; [0.09; 0.78]) control. In the four studies which only included diabetic patients, an NNT between 8 and 12 was demonstrated when comparing strict and moderately strict with conventional treatment protocol for avoiding wound infection. The NNH was between 2 and 19 for additionally induced hypoglycemia in the strict treatment protocol.

Garg et al. examined the effect of preoperative diabetes management in patients with diabetes with the aim of achieving blood glucose levels <11.1 mmol/L (<200 mg/dL) on the day of elective surgery (e15). Over a period of 24 months, a proactive diabetes management program was conducted in 1835 diabetic patients for an average of 7.5 days before surgery, depending on their metabolic status (criteria included an HbA1c >8%). Their mean age was 64.3 years, their mean HbA1c level 7.1 %. The investigation included 2074 diabetic patients who had been treated two years previously and whose epidemiological characteristics did not differ significantly from those of the observational group. The following parameters were lowered in the group with the diabetes management program:

• fasting blood glucose on the day of surgery (8.1 versus 7.7 mmol/L [146 versus 139 mg/dL], p = 0.0028)

• mean blood glucose during hospital stay (9.2 versus 8.8 mmol/L [166 versus 158 mg/dL], p <0.0001)

• hypoglycemia rate (4.93 versus 2.48%, p = 0.004)

• length of hospital stay (4.78 versus 4.62 days, p = 0.02)

• intravenous administration of antibiotics (23.7 versus 20.2 days, p = 0.001).

Rates of in-hospital mortality, perioperative renal failure, stroke, myocardial infarction, and hospital readmissions did not differ significantly.

Optimizing treatment conditions

The large number and complexity of the study results show that optimized blood glucose monitoring cannot be achieved by initiating and modifying continuous intravenous insulin therapy alone. Table 2 provides an overview of other fundamental aspects of glycemic control in critically ill patients (for a more detailed presentation of the contents see eTable). Table 3 presents aspects of perioperative risk management for patients with a history of diabetes.

Table 2. Key aspects that should be considered in the context of blood glucose monitoring in critically ill diabetics*.

Modulating factors	Consequences
Nutrition therapy (based on DGEM guidelines [e16])	Synchronization of nutrition and insulin therapy (e17) – consider limiting energy supply if insulin requirement >4 IU/hr, selectively reduce glucose supply in patients on PN – do not start PN early in patients who are not malnourished; consider the negative effects of PN (e.g., risk of hepatic steatosis, e20)
Measuring intervals	With continuous intravenous insulin therapy: monitoring frequency every 3–4 hrs. (e16) – with metabolic instability or high insulin requirement: shorten the measuring interval, even down to every 2 hrs. – mild hypoglycemia: considered a „red flag“ (30), so adapt measurement intervals
Measuring methods	Bear in mind the degree of standardization and validity of the measurement systems – favor point-of-care systems in ICU setting – favor blood gas analyzers with higher measuring accuracy over portable measuring devices (e23) and parallel monitoring of electrolyte status
Protocol-based approach	Adapt the protocol-based course of action to the circumstances of individual ICU – pay particular attention to the level of training and the role of the intensive care nursing staff (including nursing presence, patient observation [symptoms of hypoglycemia], nutrition management)
Overall treatment context (see also Table 3)	Preoperative risk prevention (premedication visit): gather information on diabetes diagnosis, diabetes type, diabetes complication status, HbA1c level (e32) – special type 1 diabetes management: generous indication for diabetologist consult, where vasopressor need is expected: suspend insulin pump therapy – aim for intraoperative blood glucose target of 7.8–10 mmol/L (140–180 mg/dL) (30) – postoperative: anticipate hyperglycemia (Figure 2) and interpret this in the context of the respective intervention (hepatic, pancreatic surgery!) and preoperative diabetes management (for example, glucocorticoids for PONV prevention)

* The eTable provides a more comprehensive version of this table.

ICU, intensive care unit; PN, parenteral nutrition; PONV, postoperative nausea and vomiting

eTable. Supplement to key perioperative risk constellations in patients with diabetes*.

Risk factors	Consequences
Nutrition therapy*	– Synchronization of nutrition and insulin therapy: avoid hypoglycemia (e17) – Insulin requirement >4 IU /hr. (individual metabolic intolerance): a) no immediate dose increase, consider reduction or pause of energy supply (e16) b) Insulin dose may be increased up to 6 IU/hr. (maximum of 144 IU/day!), no specific recommendations for diabetic patients (e7) – Prolonged hyperglycemic phases: a) reduce or restrict glucose supply to 4 g/kg/24hrs. (expert opinion, in contrast with the US-American guidelines) (e16) b) consider possibility of developing non-alcoholic fatty liver disease (NAFLD) in diabetic patients in whom excessive glucose intake can be particularly detrimental (e18, e19) – Use of parenteral nutrition (PN) a) glucose administration should be strictly reduced in diabetic patients on parenteral nutrition suffering from prolonged hyperglycemia b) commence additional PN in non-malnourished patients on enteral nutrition only if 60% of the daily calory requirement is not achieved within 7 days due to patient-related factors (reflux, bowel paralysis) (e16) c) bear in mind adverse effects on liver function since there is already a risk of developing hepatic steatosis by itself after one to up to four weeks due to high calory and glucose administration (with subsequent increase of insulin dose [insulin-dependent weight gain]) (e20) d) recovery of liver function after major liver surgery may be impaired with pre-existing NAFLD; length of hospital stay and complication rate appear to increase in elective cardiac surgery patients (e21, e22)
Measuring intervals	– Phases of stress metabolism and early artificial nutrition: monitoring frequency at least every 6–8 hrs. (e16) – Continuous intravenous insulin therapy: monitoring frequency every 3–4 hours – Metabolic instability and/or high insulin requirement: shorten the measuring interval, even down to every 2 hours – Risk of hypoglycemia due to particularly severe disease (liver failure with loss of gluconeogenesis or adrenal insufficiency): highly individual planning of the measurement intervals – Factors affecting nutritional supply (e.g.: reflux, vomiting, bowel paralysis), interruption of medication affecting nutrition and blood glucose levels (e.g., glucocorticoids): shorten the usual measuring intervals (imminent uncontrolled hyper- or hypoglycemia) (30) – In >80% of cases, mild hypoglycemic episodes often already precede severe hypoglycemia in the course of treatment (red flags) (30) – adapt the measuring intervals
Measuring methods	– Respect the degree of standardization and validity of the measurement systems – Favor point-of-care systems for intensive care (shorter waiting time for results, no need for sample preparation and transport) – Blood gas analyzers vs. portable blood glucose monitoring systems a) BGA devices with high measuring accuracy which is not lost even in the presence of vasopressor need, edema and continuous insulin therapy (e23) b) BGA devices also show the electrolyte constellation as per protocol. c) Measurement deviations (over- and underestimations) are higher for both BGA devices and portable meters in the hypoglycemic range (e23) – arterial blood is preferable to capillary blood, central venous blood is an alternative (e24) – continuous blood glucose monitoring systems: a) fundamental system requirements (e25): high measurement accuracy, the possibility of trend analysis (trend compass), high user practicability as well as time and cost savings b) future goals: anticipate hypoglycemia before it develops, reduce phases of hypoglycemia and glucose variability (e25, e26) c) Flash glucose monitoring systems (subcutaneous sensor): use in the intensive care setting not (yet) recommended apart from studies (e27, e28); note: measurement accuracy requirements must match those for blood glucose monitoring systems (e29); real world data reporting on the use of flash glucose monitoring systems during the COVID-19 pandemic provide positive signals regarding usability even for critically ill patients (e30).
Protocol-based approach	– Protocol-based procedure: should be adapted to the conditions of each individual ICU – Role of the intensive care team: a) nursing presence b) patient monitoring (symptoms of hypoglycemia) c) monitoring enteral /parenteral energy consumption d) Acquisition of measurement results and treating accordingly – standard operating procedures for the perioperative care of diabetic patients are recommended in principle (9, 30) – consider the level of training: The specialized study center in Leuven was able to stabilize 70% of the study patients within the desired blood glucose target range. This figure was significantly below 50% in the multicenter NICE-SUGAR trial (e31). Yamada et al. showed that only 50% of the 72 control groups in 36 RCTs achieved a mean glucose level within the target range (39).
Overall treatment context (see Table 3)	– Preoperative risk prevention (premedication visit) a) gather information on diabetes diagnosis, diabetes type, diabetes complication status, HbA1c level (e32) b) type 1 diabetes management: generous indication for diabetologist consult, expertise and specific protocols should be in place c) abnormal preoperative HbA1c levels: Discuss scheduling elective surgery within the multidisciplinary team since HbA1c levels >8.5 % are associated with poorer surgical outcomes (9) – Stop insulin pump therapy for major operations (anticipated treatment with catecholamines) and switch to intravenous insulin (ensures adequate controllability of insulin pharmacokinetics and dynamics) (9) – Aim for intraoperative blood glucose target of 7.8–10 mmol/L (140–180 mg/dL) (30); lower glucose levels are associated with increased mortality and higher incidence of stroke, for example in cardiac surgery (e33, e34) – Anticipate immediately postoperative phases of hyperglycemia (see Figure 2), interpret and treat within the context of the respective operation (e.g., liver and pancreas surgery, (see Table 3), preoperative diabetes management (HbA1c level), prescribed medication (e.g., glucocorticoids for PONV prevention)

* follow the DGEM guidelines with regard to the use of artificial food (including concepts for increasing and modifying nutrition according to calory requirement as well as the limits set by critical illness or surgery that has taken place) (e16).

DGEM, German Society for Nutritional Medicine; ICU, intensive care unit; PONV, postoperative nausea and vomiting; RCT, randomized controlled study

Table 3. Selected perioperative risk constellations*¹ in patients with diabetes.

Risk factors	Consequences
Undiagnosed diabetes	imminent absolute/relative insulin deficiency, uncontrolled ketoacidosis/hyperosmolar state → check preoperative HbA1c and blood glucose, no major elective surgery if HbA1c >8.5%
Pre-existing antidiabetic therapy	hypoglycemia (e.g., caused by insulin, sulfonylureas), ketoacidosis (caused by SGLT2 inhibitors), lactate acidosis (caused by metformin) → pause in good time, monitor daily blood glucose profiles, consider replacing with insulin while bearing in mind fasting before surgery (˜ 6 hrs.)
Distinguish type 1, type 2, and pancreatogenic diabetes	imminent absolute insulin deficiency, uncontrolled ketoacidosis/hyperosmolar state → patient self-management (insulin pump therapy) vs. continuous intravenous insulin therapy (major surgery, catecholamines, insulin pump needle located at surgical site)
Uncontrolled blood glucose during surgery	hypo-/hyperglycemia → monitor blood glucose, intravenous insulin therapy in justified exceptions
Continuous intravenous insulintherapy	hypoglycemia → monitor blood glucose, maintain target range
Start (medical) nutrition therapy	hyperglycemia → intravenous insulin therapy, nutritional management
Non-alcoholic fatty liver disease (NAFLD to NASH with cirrhosis)*2	Postoperative liver dysfunction, even acute liver failure, with impairment of gluconeogenesis and lactate clearancecapacity, hypoglycemia → blood glucose monitoring, nutritional management, glucose administration
Hyperglycemic reaction toSIRS/sepsis after surgery associated with immunosuppression	Wound and surgical site infections, impaired wound healing, increased risk of infection, septicemia → monitor blood glucose → intravenous insulin therapy, nutritional management
Type 1 and pancreatogenic diabetes (pancreas surgery)	→ monitor blood glucose, intravenous insulin therapy, maintain target range, endocrinology consult

*¹ risk constellations: non-alcoholic fatty liver disease (NAFLD); non-alcoholic steatohepatitis (NASH);

*² primarily in association with major (liver) surgery/liver shock/drug-induced liver damage/bloodstream infections (e.g., Staphylococcus aureus), etc.

SIRS, systemic inflammatory response syndrome

(Supplement to key perioperative risk constellations: see eTable)

Conclusion

There is no level I evidence currently available which would justify glucose control below a target range of 7.8–10 mmol/L (141–180 mg/dL) for critically ill patients with diabetes. Recent studies suggest that high priority should also be given to individualizing treatment target ranges and avoiding relative and absolute hypoglycemia. Apart from insulin administration, nutrition therapy is an equally important pillar of glucose control. Australian observational studies provide evidence that liberal glucose control in ICU patients with diabetes has no negative impact on short-term mortality and morbidity. Further studies (LUCID trial [e35]) will have to show whether in the future standard intensive care will also include values between 10 and 14 mmol/L (180–252 mg/dL) for critically ill individuals with diabetes. Until then, and in accordance with the guidelines of the leading professional societies, the following should apply:

• Insulin therapy should not be initiated until a level of 10 mmol/L (180 mg/dL).

• A target range between 7.8 and 10 mmol/L (140–180 mg/dL) should be aimed for.

• A blood glucose value of 15 mmol/L (270 mg/dL) is just as undesirable as a value of 3.8 mmol/L (68 mg/dL).

Questions on the article in issue 38/2021: Blood Sugar Targets in Surgical Intensive Care—Management and Special Considerations in Patients With Diabetes.

The submission deadline is 23 September 2022. Only one answer is possible per question. Please select the answer that is most appropriate.

Question 1

Which threshold range is recommended by the professional societies for hospitalized patients with diabetes?

1. 4.4–6.1 mmol/L

2. 3.3–5.8 mmol/L

3. 6.1–7.2 mmol/L

4. 6.8–7.9 mmol/L

5. 7.8–10 mmol/L

Question 2

What monitoring frequency should normally be selected for continuous intravenous insulin therapy?

1. every half hour

2. every 1–2 hours

3. every 3–4 hours

4. every 5–6 hours

5. every 8–10 hours

Question 3

For which surgical specialty is stress-induced hyperglycemia reported in up to 80% of patients?

1. Cardiac surgery

2. Pancreatic surgery

3. Hepatic surgery

4. Oral maxillofacial surgery

5. Traumatology

Question 4

At what threshold does the German Diabetes Association see a need for action in hospitalized patients with diabetes?

1. >7 mmol/L

2. >8 mmol/L

3. >9 mmol/L

4. >10 mmol/L

5. >11 mmol/L

Question 5

Measurement of HbA1c is generally recommended for admission to hospital or the ICU. At what value can pre-existing diabetes be assumed or a previously undiagnosed diabetes be diagnosed?

1. ≥ 5.5%

2. ≥ 6.5%

3. ≥ 7.8%

4. ≥ 8.5%

5. ≥ 9.7%

Question 6

In which range of blood glucose levels do we speak of mild/moderate hypoglycemia or severe hypoglycemia?

1. 1.9–3.9 mmol/L or ≤ 1.8 mmol/L

2. 2.9–3.9 mmol/L or ≤ 2.8 mmol/L

3. 2.3–3.9 mmol/L or ≤ 2.2 mmol/L

4. 3.9–6.1 mmol/L or ≤ 3.8 mmol/L

5. 3.2–4.3 mmol/L or ≤ 3.1 mmol/L

Question 7

What does the abbreviation SIRS stand for?

1. sudden inflammatory repression syndrome

2. systemic insulin-dependent response syndrome

3. systemic insulin-dependent regulation syndrome

4. sudden insulin-dependent response syndrome

5. systemic inflammatory response syndrome

Question 8

What measure should be considered with an insulin requirement >4 IU/hr.?

1. increase energy supply

2. phase out insulin administration

3. restrict energy supply

4. switch to another type of insulin

5. switch from insulin to gliptins

Question 9

What happens metabolically in stress-induced hyperglycemia in patients with pre-existing diabetes?

1. decreased glucose uptake in the liver

2. decreased lipolysis in insulin-dependent organs

3. decreased gluconeogenesis in the liver

4. decreased glucose uptake in insulin-dependent organs

5. decreased protein catabolism in insulin-dependent organs

Question 10

At what HbA1c level should major elective surgery be postponed?

1. >6.5%

2. >7.2%

3. >7.8%

4. >8.1%

5. >8.5%