Glycemic Management of Hospitalized Patients Receiving Nutrition Support

Hyperglycemia is seen in 22–46% of hospitalized patients (1,2) and is particularly common in those receiving EN or PN, with and without diabetes (3,4). The prevalence of hyperglycemia in patients receiving EN is ∼30% (5). It is even higher in those receiving total parenteral nutrition (TPN), with estimates of 28–44% when a diagnostic blood glucose cut-off value of >200 mg/dL is used and up to 90% when the cut-off value is >150 mg/dL (6–9). Predisposing factors include age >65 years, high C-reactive protein level, high dextrose content in the PN, A1C >5.7%, presence of diabetes, infection, and concomitant use of glucose-elevating drugs such as corticosteroids (6).

Hyperglycemia related to nutrition support is associated with an increase in mortality and morbidity. A retrospective observational study (10) found that patients with EN-associated hyperglycemia (with or without prior diabetes) had higher mortality rates than normoglycemic patients (relative risk 1.34 vs. 1.56 vs. 1, respectively). Patients with newly diagnosed hyperglycemia required more time to come off of tube feeding (TF) than those with prior diabetes. A longitudinal observational study (11) in 115 patients without diabetes who were admitted for stroke noted higher mortality in those who developed hyperglycemia while receiving EN (45.5%) than in both a normoglycemic group (10.7%) and a group with previously identified stress hyperglycemia (15.8%). Development of hyperglycemia after EN initiation was also associated with worsening evolution of stroke, as indicated by inability to take oral intake.

Similarly, multiple studies have demonstrated that PN-associated hyperglycemia is associated with high risks of mortality and morbidity. Blood glucose values >180 mg/dL within 24 hours of PN initiation were associated with increased risks of pneumonia, acute renal failure, and cardiac complications (4,12,13). PN-associated hyperglycemia was associated with increased mortality ranging from 5.6-fold (with a mean blood glucose >180 mg/dL) to 11-fold (with a mean blood glucose >164 mg/dL) (13,14). Interestingly, this increased risk is more pronounced in patients without diabetes than in those with known diabetes (13).

Although many studies have shown adverse outcomes with EN- and PN-associated hyperglycemia, few intervention studies have been published showing that glucose management improves outcomes. Improved glycemic outcomes using continuous insulin infusions compared with conventional glycemic management was associated with a lower surgical site infection rate (4.7 vs. 13.2%, P <0.030) in a retrospective observational study involving 212 patients with diabetes receiving EN after gastrectomy. However, the difference in mean blood glucose between the two groups (97.3 ± 21 vs. 171.2 ± 32.4 mg/dL, P <0.001) was achieved at a significant cost of hypoglycemia (7.5 vs. 0.9%, P = 0.035) (15). Similar studies involving patients receiving PN are not available.

It is vital to recognize that there are no randomized controlled trials (RCTs) showing improved outcomes with better glycemic control in patients receiving EN or PN. Therefore, by extrapolating data gathered in the general hospitalized population, experts recommend a blood glucose target range of 140–180 mg/dL in patients receiving nutrition support (1,16).

The detrimental effect of EN and PN is additive to the effect of stress hyperglycemia, which results from the complex interplay of counterregulatory hormones such as cortisol, growth hormone, catecholamines, and cytokines, leading to increased hepatic glucose production and insulin resistance (17–19). More pronounced hyperglycemia is observed with PN than with EN because of a lack of incretin effect, as PN bypasses glucagon-like peptide-1 and gastric inhibitory polypeptide in the gastrointestinal tract (20–22).

Meta-analysis of trials comparing hypocaloric (20 kcal/kg/day) to full energy goal (25 kcal/kg/day) PN has shown reductions in risks of hyperglycemia and insulin resistance, infection, duration of mechanical ventilation, and length of stay (23,24). Hence, current nutritional guidelines from the Society of Critical Care Medicine (SCCM)/American Society of Parenteral and Enteral Nutrition (ASPEN) recommend the use of hypocaloric PN with adequate protein (>1.2 g/kg/day) over the first week of intensive care unit (ICU) stays in high-risk and severely malnourished patients (25).

The minimum amount of recommended dextrose content in PN is 2 g/kg/day, as outlined in guidelines from the European Society of Clinical Nutrition and Metabolism (ESPEN), SCCM, and ASPEN (25,26). Studies have demonstrated that patients receiving PN dextrose content of 2.6 ± 1.4 g/kg/day have increased risks of hyperglycemia and related complications, including mortality (27–29). One approach to reduce the incidence and severity of hyperglycemia and lower daily insulin requirements is to decrease the dextrose content in the PN to 150–200 g/day (30). More prospective RCTs are needed to determine the effects of the dextrose infusion rate on hyperglycemia and clinical outcomes.

Below, we present four approaches to the treatment of PN-associated hyperglycemia. The evidence supporting each of these approaches is limited, and there is no gold-standard regimen. A summary of selected studies providing safety and efficacy data are presented in Table 1. Here, we discuss the rationale for each approach and provide clinical context for the published literature.

Intravenous (IV) regular human insulin (RHI) infusion, titrated based on hourly monitoring of blood glucose levels, is the recommended strategy for critically ill patients receiving PN because of their risks of hypoglycemia and hyperglycemia (31,32). This strategy also can be used as an initial dose-finding method for stable patients. Although it is an effective strategy in the ICU, it is labor-intensive and may not be available in noncritical care settings in many institutions. In addition, hourly point-of-care (POC) glucose checks needed for adequate titration may result in patient and staff dissatisfaction with treatment.

Basal insulin can be provided as intermediate-acting (NPH insulin given twice daily) or long-acting (glargine/detemir given once daily) insulin. Prandial and correction insulin can be given as RHI every 6 hours or as a rapid-acting analog (RAA) insulin every 4 hours.

Studies evaluating scheduled subcutaneous (SQ) insulin regimens typically used a basal-plus approach involving a weight-based dose of long-acting insulin used in combination with correctional insulin given every 4–6 hours (Table 1). These observational studies highlight the challenges of achieving and maintaining treatment goals, as only 50% of the subjects were able to achieve a target blood glucose <180 mg/dL in one study (33), whereas, in the other, blood glucose was in the target range for the first 4 days but remained above goal during the last 3 days of the 7-day study (34).

Adding RHI to the PN bag delivers insulin continuously along with the dextrose and provides an opportunity to achieve smoother blood glucose management outcomes compared with an SQ insulin regimen, given the frequent discordance between insulin dosing and PN administration with SQ insulin delivery (35). In theory, the risk of hypoglycemia should be lower when RHI is added to the PN bag because cessation of PN will lead to discontinuation of insulin as well (23,36). Although metabolic outcomes are similar with both an SQ regimen and RHI in the PN bag, target blood glucose levels are reached faster with fewer hyperglycemic episodes when adding RHI to the PN bag (36). Hypoglycemia did not develop after PN interruption in the group receiving RHI in the PN bag because insulin infusion terminated at PN cessation (36).

A strategy involving the combination of SQ insulin and RHI in the PN bag follows the physiologic principle of providing basal, nutritional, and correctional insulin. Basal coverage is given either as intermediate (NPH insulin twice to three times daily) or long-acting (glargine/detemir given once daily) insulin. Prandial coverage can be given as RHI in the PN bag and calculated based on the insulin/dextrose (I/D) ratio or RHI/RAA insulin requirements. The advantage of combination therapy is better metabolic outcomes and a possibly smoother transition once PN is stopped (37). A prospective cohort study evaluating protocol-directed insulin dosing linked to dextrose showed improved glycemic outcomes without a significantly increased risk of hypoglycemia (38). Hypoglycemia risk after the interruption of PN is lower in combination therapy because all of the insulin is not given as once- or twice-daily basal insulin.

In the absence of large RCTs, current guidelines for the management of PN-associated hyperglycemia are predominantly based on expert opinion (1,16,39). Both the Society of Hospital Medicine (SHM) and the Endocrine Society recommend the initial use of IV insulin infusion for estimating patients’ total daily dose (TDD) of insulin. The American Diabetes Association (ADA), the SHM, and the Endocrine Society recommend adding insulin to the PN bag with a dose calculation based on either IV insulin infusion or the I/D ratio (Table 2).

In the absence of data showing the superiority of one protocol over the others, we recommend that the approach to treatment should take into account the patient’s degree of hyperglycemia, clinical stability, and diabetes status, as well as institutional policies. We further recommend:

1. In clinically unstable patients (e.g., those requiring vasopressors, taking steroids, or with changing renal function), a separate IV insulin infusion should be used until clinical stability is achieved.

2. In clinically stable patients with no history of diabetes, POC blood glucose monitoring should be performed for the initial 48 hours, with correctional insulin provided as RHI every 6 hours or RAA insulin every 4 hours. If the blood glucose is <140 mg/dL after reaching the desired caloric intake, then no further blood glucose testing or treatment is needed. For patients with persistent hyperglycemia (defined as more than two POC blood glucose values >180 mg/dL while on correctional insulin alone), insulin therapy should be scheduled based on the correctional insulin requirements (40).

3. In clinically stable patients with preexisting diabetes, both basal and nutritional coverage should be provided, along with correctional insulin. This plan can be achieved in one of three ways:

   • Approach 1. Basal insulin can be started as a weight-based dose (0.1–0.25 units/kg) or the home basal insulin dose can be resumed (1,41). Correctional insulin can be given as RHI every 6 hours or RAA insulin every 4 hours. The nutritional component to cover dextrose in PN can be given as RHI added to the PN bag.

   • Approach 2. Provide all insulin needs (both nutritional and basal) added to the PN bag. The dose needed can be calculated in one of the following three ways:

     • Eighty percent of the TDD calculated from the previously administered IV insulin infusion can be added as RHI to the PN bag.

     • The dose of RHI in the PN bag can be extrapolated from the initial requirement, with correctional insulin given every 4 or 6 hours.

     • The dose of RHI in the PN bag can be based on an I/D ratio of 1:20 for patients without diabetes and 1:10–15 for patients with diabetes. Basal insulin can be estimated based on the previous home dose or weight-based dosing as outlined above. Subsequent uptitration can be achieved using supplemental RHI or RAA correctional insulin given every 4–6 hours.

   • Approach 3. The nutrition component can be given subcutaneously with either NPH insulin (given two or three times daily), RHI or RAA insulin with the dose calculated based on the I/D ratio described above and given every 4–6 hours along with correctional insulin given as RHI every 6 hours or RAA insulin every 4 hours. Either weight-based dosing or the home basal insulin dose should be started to cover basal needs. This is an option for institutions that do not allow insulin in the PN bag.

Many institutions do not allow adding RHI in the PN bag because of safety concerns. In the authors’ experience, adding RHI to the PN bag is safe when used in carefully selected patients and with proper institutional safeguards such as the guidance of endocrinology and pharmacy teams.

In our institution, we follow the specific patient criteria listed below for safe and effective use of insulin in PN in the inpatient setting.

1. Patients who previously used home PN containing RHI who are clinically stable

2. Patients who will be discharged home on PN with a stable insulin requirement who are meeting at least 75% of their dextrose goal, in whom PN is anticipated to be continued long-term (∼1 month), and who are clinically stable as defined by meeting all the following criteria:

   • Stable glomerular filtration rate (>45 mL/min/1.73 m²)

   • Stable dose of steroids, if required

   • Absence of severe infection or sepsis

   • Stable dextrose concentration in the PN bag

The choice of EN has important metabolic implications. EN can be provided as a standard formula (SF), elemental formula, or diabetes-specific formula (DSF). SF is high in carbohydrates, has low to moderate amounts of lipids, and does not contain dietary fiber. In contrast, DSF is high in monounsaturated fatty acids and contains low-glycemic-index carbohydrates along with dietary fiber to prolong the time required for glucose digestion and absorption (Table 3) (42–45). DSF use has been shown to lower mean blood glucose, A1C, postprandial blood glucose levels, glycemic variability, insulin requirements, and insulin resistance (44,46–54). Furthermore, decreases in mortality, lengths of stay, health care costs, and risk of acquired infection in the ICU have been noted (52,55,56).

ASPEN guidelines in 2013 concluded that no recommendations can be made regarding the use of DSF for EN-related hyperglycemia (57). Use of DSF is supported by ESPEN and the Endocrine Society in their latest guidelines (40,58).

Insulin therapy is recommended for all the patients experiencing hyperglycemia while receiving EN (40). The pharmacokinetics and pharmacodynamics of the insulin used should match the mode of EN delivery (i.e., continuous, bolus, or nocturnal). Table 4 summarizes five studies that evaluated the efficacy and safety of different subcutaneous insulin regimens.

The take-home points from these studies are that an SQ sliding-scale insulin regimen alone is inadequate, and the addition of basal insulin is required for patients to achieve favorable glycemic outcomes (59,60). Also, NPH insulin given every 6 hours with correctional doses provides similar outcomes but less hypoglycemia than NPH insulin given every 4 hours (61). Similar glycemic outcomes were achieved in patients receiving either basal-bolus therapy or NPH insulin every 8 hours while on the continuous EN (60). The 70/30 biphasic insulin mixtures delivered three times daily can be used for glycemic management in these patients (62). For patients receiving bolus TF, a basal-bolus regimen can be used (63). In summary, in all of these studies, there were little to no significant differences in mean blood glucose among basal-plus-correctional, basal-bolus, or NPH-based insulin regimens in patients on continuous EN.

Recommendations for the management of EN-associated hyperglycemia, as outlined by ADA (16), the Endocrine Society (1,40), the American Association of Clinical Endocrinologists (AACE) (64), and the Society of Hospital Medicine (39), are presented in Table 5. Although these recommendations at first glance seem different, they follow the same principle of scheduled basal-bolus and correctional insulin. We further recommend:

1. As outlined previously, in clinically unstable patients (e.g., those requiring vasopressors, taking steroids, or with infection), a separate IV insulin infusion should be used until clinical stability is achieved.

2. In clinically stable patients without diabetes, POC blood glucose should be checked for 48 hours. If blood glucose is <140 mg/dL, POC testing should be discontinued. For patients with persistent hyperglycemia (more than two POC blood glucose levels >180 mg/dL while on correctional insulin alone), SQ insulin should be started based on the required correctional insulin.

3. Clinically stable patients with a history of diabetes who were previously on basal insulin should have their basal insulin continued, especially those with type 1 diabetes to prevent DKA. The basal insulin dose can be estimated from the patient’s home dose or calculated based on the patient’s weight (0.1–0.25 units/kg depending on patient’s blood glucose level and BMI) (1,41). Nutritional and correctional insulin can be provided depending on the type of EN the patient is receiving.

Nutritional needs can be covered with intermediate-acting NPH insulin given every 8 or 12 hours, RHI given every 6 hours, or RAA insulin given every 4 hours. Typically, the total daily nutritional dose of insulin may be calculated as 1 unit of insulin for every 10–15 g carbohydrate in the formula. Alternatively, IV insulin infusion can be used for dose-finding for nutritional needs. Patients with preexisting diabetes will need their basal insulin resumed. Correctional doses of insulin should be provided, and daily adjustment of insulin therapy will be needed to reach glycemic goals. Recently updated Endocrine Society guidelines recommend the use of either NPH insulin or a basal-bolus insulin regimen for management of EN-associated hyperglycemia (40).

Both the AACE and the SHM caution against overuse of basal insulin, and we agree with this principle to avoid hypoglycemia with a basal-only regimen. In the authors’ experience, insulin-resistant patients with large insulin requirements receiving both basal and nutritional coverage as NPH insulin every 8 hours frequently have hypoglycemia even when dose of insulin is meticulously adjusted.

Nocturnal or overnight nutrition can be covered with NPH insulin only (16). In our experience, addition of a fixed dose of RAA insulin at the beginning of the feeding along with NPH or the use of a 70/30 biphasic insulin mixture works well for nocturnal TF with high insulin needs.

Bolus insulin can be given as an RAA or RHI with each bolus along with correctional insulin every 4–6 hours. In patients with preexisting diabetes, the home dose of basal insulin or weight-based dosing should be started following the principles outlined above.

When prescribing an insulin regimen, consideration should be given to both in-hospital and home use in balancing regimen complexity with safety and feasibility concerns.

Patients with EN are at high risk of hypoglycemia from any mismatch of insulin administration and nutrition or excessive insulin dosing. Common contributing factors are reductions in steroid or vasopressor doses, progressive organ failure, lack of communication between the nutrition team and the clinician ordering insulin, dislodgment of the feeding tube, and change or cycling of TF. Dextrose 10% infusion should be started when artificial nutrition is interrupted, and the next insulin dose should be adjusted.

Because artificial nutrition results in a continuous postprandial state, any attempt to bring the blood glucose level to <140 mg/dL increases the risk of hypoglycemia in these patients (16). If a patient develops hypoglycemia while on artificial nutrition, an institution-specific hypoglycemia protocol should be initiated. There is ongoing interest in using continuous glucose monitoring (CGM) to detect and prevent hypoglycemia in patients receiving EN or PN, but more studies are needed to guide the regulatory approval and effective implementation of this use of CGM technology.

Several studies have shown the efficacy of dipeptidyl peptidase 4 inhibitors in the general medicine and surgical wards for the management of mild to moderate hyperglycemia (65,66). However, there are insufficient data regarding the use of these or other noninsulin therapies in patients receiving EN or PN.

Recent studies have reported excellent outcomes using fully integrated closed-loop automated insulin delivery systems for the treatment of EN- or PN-related hyperglycemia. In patients with diabetes, time in range improved without an increase in hypoglycemia in both ICU and non-ICU settings (67).

During the ongoing coronavirus disease 2019 (COVID-19) pandemic, an increasing number of patients have been on both steroids and artificial nutrition. Use of IV insulin has been shown to reduce inflammatory markers related to COVID-19 (68). Depending on individual patient factors and setting, either IV or SQ insulin can be used on general wards, and IV insulin can be considered in the ICU (69). Clinicians should be cautious regarding rapidly changing steroid doses, and insulin adjustments should be made accordingly despite the use of constant dextrose content.

Hyperglycemia is common in hospitalized patients receiving PN or EN and is associated with a higher risk of complications and mortality. Effective management of hyperglycemia in these patients, while avoiding hypoglycemia, is crucial. Although it seems intuitive to consider using targeted insulin therapy that matches the glycemic profiles of the modes of EN delivery, to date, no single insulin regimen has been shown to offer superior efficacy and safety. Close monitoring of patients and effective hypoglycemia recognition and treatment is of the utmost importance.

No potential conflicts of interest relevant to this article were reported.

P.P. reviewed the literature and wrote the section of the manuscript on EN-associated hyperglycemia, created the tables, and edited the manuscript. S.P. reviewed the literature and wrote the section on PN-associated hyperglycemia. A.D. reviewed the literature, supervised the writing, and reviewed and edited the manuscript. P.P. is the guarantor of this work and, as such, takes full responsibility for the accuracy of the review.