Introduction

Presently, 20–30% of all hospitalized patients have diabetes [1, 2]. The impact of diabetes on hospital outcome depends on both long-term and in-hospital glycemic control [2–4]. Although inpatient glycemic targets remain a controversial topic [5, 6], the prevention of hyperglycemia, especially in critically ill patients, is associated with improved outcomes and is nearly universally recommended [2, 4–9]. In addition, high levels of glycemic variability are associated with increased mortality in critical illness [10]. Judicious use of glucose-lowering medications in concert with an appropriate nutrition strategy allows for optimal inpatient glycemic control.

Nutrition support is an important aspect of medical care that provides protein and energy (calories). This slows catabolism in hospitalized patients and greatly impacts patient outcome [11•]. In hospitalized patients, inadequate nutrition is associated with increased rates of infection, complications, increased lengths of hospital stay, and increased ventilator days in critically ill patients [12, 13]. Nutrition support in the form of enterally administered feeds or parenteral nutrition (PN) is indicated in patients who cannot meet their nutritional requirements due to illness or compromise of the gastrointestinal (GI) tract [14]. These methods of feeding provide macronutrients, including carbohydrates, as a source of calories and can greatly impact glycemic control in hospitalized patients [2, 4, 5]. In this article, we review strategies of in-hospital nutritional support for patients with diabetes to improve both glycemic control and overall clinical outcome.

General Nutrition in Diabetes

For the management of both type 1 diabetes mellitus (T1DM) and type 2 diabetes mellitus (T2DM), the role of nutrition as part of a patient’s daily routine is an important consideration. In T1DM, close observation of carbohydrate intake allows patients to match this with corresponding amounts of injected insulin and achieve glycemic targets [15]. In T2DM, reduced carbohydrate diets are recommended and may help to reduce insulin resistance [16]. Unfortunately, hospital food services are limited by economic and other constraints that often lead to a high proportion of carbohydrates as a calorie source in hospital meals [17].

Patients with T2DM who require insulin may also benefit from matching dietary carbohydrate load to the amount of insulin administered for optimal glycemic control [15]. Foods with a high glycemic index, especially sugar-containing beverages, should be restricted to prevent large glycemic excursions in both T1DM and T2DM [15, 16]. Patients who take insulin should consume consistent amounts of carbohydrates with each meal to achieve target blood glucose levels [15, 16]. Consumption of dietary fiber, which inhibits intestinal α-glucosidase activity, reduces postprandial blood glucose levels and is recommended for the treatment and prevention of T2DM [16]. Other nutritional considerations in diabetic patients include an emphasis on portion control for weight loss in overweight and obese T2DM patients [15], as well as supplementation of micronutrient deficiencies that commonly exist in obese patients [18].

In general, the optimal diet for diabetic patients has a reduced carbohydrate content, restricted amount of high glycemic index foods, and increased dietary fiber. The concept of matching medication therapy with dietary carbohydrate intake is fundamental to the treatment of diabetes and is a major component in the nutritional management of hospitalized diabetic patients. The practice of combined nutritional and hormonal therapies is fundamental to a comprehensive strategy for the metabolic management of hospitalized patients [19].

Nutrition in Hospitalized Patients

All hospitalized patients should undergo an initial nutritional assessment upon admission [14]. Often, disease processes that lead to hospitalization are associated with unintentional weight loss, decreased appetite and food intake, and a compromised GI tract. Patients with severe malnutrition prior to admission may exhibit muscle wasting and low levels of circulating visceral proteins such as albumin or prealbumin. A careful history and physical examination can identify patients at high nutritional risk and aid in the decision for nutritional intervention [14].

Nutritional provision is a common aspect in the care of all hospitalized patients and nutritional support may be considered for those who fail to consume adequate nutrition with an oral diet [14, 20]. For the critically ill, nutritional support should begin within 72 h of hospital admission [19]. Similarly, in surgical patients, feeding should start within 72 h following surgery [19]. For surgical patients who are severely malnourished, nutritional support should be started prior to surgery [19].

In most patients, the preferred method of feeding is enterally—through the GI tract [11•, 14]. If oral food intake is inadequate, enteral nutrition (EN) by tube feedings through a nasogastric, nasojejunal, gastric, or jejunal feeding tube should be initiated. PN is indicated in patients who have a compromised GI tract and may be administered through a peripheral (up to 14 days) or central venous access catheter [11•].

Enteral tube feedings are often stopped for many reasons during the hospital course including radiologic testing, patient repositioning, diarrhea, vomiting, or for high gastric residuals. The average critically ill patient receives approximately 60–70% of prescribed tube feeds [21]. Therefore, PN has been used to supplement EN in patients who do not adequately meet their nutritional needs by EN alone [11•, 21].

Nutritional goals of both protein and calories should be assessed for individual patients based on their degree of malnutrition and other clinical factors such as renal dysfunction, obesity, and diabetes. In otherwise well-nourished individuals, daily caloric goals should range from 25 to 35 kcal/kg and protein intake should be 1–1.5 g/kg [11•]. For malnourished patients, a lower daily calorie target is used (15–25 kcal/kg) to avoid overfeeding and minimize the risk of refeeding syndrome [11•, 19]. Daily protein intake may be restricted in patients with impaired renal function (0.6–1.0 g/kg) [22], and should be enhanced (1.5–2 g/kg) for those on renal replacement therapy [11•, 19].

For obese patients, low calorie and higher protein nutritional support is recommended [23]. One strategy is to permissively underfeed obese patients. Caloric goals may be determined based on an adjusted body weight (ABW = ideal body weight + ¼ of the difference between actual and ideal body weight), rather than using their actual weight [19, 23]. When using ABW, protein intake should be increased to 1.5 g/kg to prevent the loss of lean body mass, which otherwise may be induced with permissive underfeeding [11•, 19, 23].

Enteral Nutrition in Diabetic Patients

In patients who require EN to meet their dietary protein and calorie needs, one must consider the type of tube feed formula and schedule. By definition, tube feeding formulas are highly processed foods made from water, corn syrup, starches, oils, soy protein, caseins, and other components. Semi-elemental feeds are enzymatically processed to hydrolyze proteins and form peptides of varying lengths for easier digestion and absorption. They also typically contain higher amounts of medium-chain triglycerides as their fat source compared with standard intact protein formulas.

Standard tube feeding formulas contain 1–2 kcal/mL. Approximately 50–55% of the calories are derived from carbohydrates, 30–35% from lipids, and 15–20% from protein. A wide array of specialized EN formulas are available for clinical use and are marketed for the use in specific disease states [14].

For patients with diabetes, EN formulas have been developed that contain relatively low total carbohydrates with higher amounts of fiber and monounsaturated fatty acids (Table 1). In these specialized formulas, carbohydrates account for only 35–40% of total calories and are comprised of corn maltodextrin and other polysaccharides as well as high amounts of fructose [24]. Fructose accounts for approximately 30% of the carbohydrate calories. The fiber content is approximately 10–15 g per 1000 kcal of formula.

Table 1 Characteristics of a diabetes-specific formula
Full size table

The design of a diabetes-specific formula (DSF) is based on dietary recommendations of the American Diabetes Association, which suggest patients with diabetes limit carbohydrates, consume mainly complex carbohydrates, and increase dietary fiber intake [15]. The use of fructose as a source of calories in a DSF is rationalized by the substantially smaller effect fructose has on raising blood glucose and blood insulin levels compared with other mono- and disaccharides [25].

Several randomized trials have demonstrated diminished postprandial blood glucose and insulin levels in T2DM patients following a single bolus of a DSF [26–29]. In these studies, peak postprandial glucose levels were 10–30 mg/dL lower in patients who received a DSF compared with standard formula.

Only a handful of longitudinal randomized trials using a DSF in T2DM patients have been performed, with durations that range from 2 weeks to 3 months [30–33]. Glycated hemoglobin levels (hemoglobin A1c) decreased between 0.4% to 0.8% over this brief period from baseline levels that ranged between 6.9% and 7.7% [30, 33]. Postprandial glucose decreased by 12–47 mg/dL [31, 32]. Fasting glucose and insulin requirements also decreased with a DSF [30, 32, 33].

DSF tube feeds have been studied in T1DM patients and show similar reductions in postprandial glucose and insulin requirements in single bolus trials [34–36]. One longitudinal study demonstrates decreased postprandial glucose and lower insulin requirements over 2 weeks of observation [32].

Because there are multiple differences between a DSF and a standard formula in the above trials, it is difficult to isolate which specific DSF components contribute to their favorable glycemic properties. Two of the studies compared formulas with similar carbohydrate content, but differing amounts of fiber in patients with diabetes. del Carmen et al. [34] showed no difference in blood glucose levels after a single bolus of standard or high-fiber formula in T1DM patients. Similarly, Thomas et al. [28] demonstrated no difference in area under the curve for postprandial blood glucose after a single bolus of standard or high-fiber formula in T2DM patients. The results of these two trials suggest that the glucose-lowering effect of dietary fiber is minimal in patients with diabetes receiving EN.

Aside from improvements in measures of glycemic control and insulin requirements, these few trials do not address clinical outcomes and do not demonstrate reductions in diabetes-related complications in patients who use DSF tube feeds over the long term [24]. Improvements in hemoglobin A1c, fasting and postprandial blood glucose levels suggest that a DSF may be beneficial in diabetes, but larger clinical studies are required. Recent guidelines published by the European Society for Parenteral and Enteral Nutrition state that a DSF is not necessary to achieve glycemic control in an intensive care unit (ICU) [20]. The American Society for Parenteral and Enteral Nutrition does not specifically address the use of a DSF, but suggests that specialized tube feeds may be applied where clinically appropriate [14].

One potential source for adverse effects of a DSF may be the relatively high content of fructose in some brands. Fructose has a 7.5-fold higher protein glycation rate than glucose and its use in a DSF may lead to increased production of advanced glycosylated end products [25]. High circulating levels of fructose are associated with heart disease, the metabolic syndrome, and nonalcoholic steatosis [37, 38]. At this point, it is only speculative to mention this potential. Many standard and specialized tube feed formulations contain fructose added either as pure fructose, high fructose corn syrup, or as part of the disaccharide sucrose. The potential metabolic detriment of high levels of fructose exposure continues to be investigated.

As an alternative to EN using a DSF for patients with diabetes, there are several tube feed formulas that, although they are not designed specifically for diabetes, contain low levels of carbohydrates. Some contain relatively high fiber as well. For example, Peptamen AF (Nestle Nutrition, Vevey, Switzerland) derives 36% of its calories from carbohydrates in the form of corn maltodextrin and has 5.2 g of dietary fiber per 1000 mL. Oxepa, Vital AF, and Pulmocare (Abbott Nutrition, Columbus, OH) also have comparably low carbohydrate content (Table 2). With the exception of Pulmocare, which showed similar postprandial blood glucose reduction to a DSF in a single study [35], none of the other above-mentioned formulas have been studied for specific use in diabetes. However, the similarities in carbohydrate content of these formulas with a DSF are striking and suggest that patients with diabetes may benefit from their use as well.

Table 2 Enteral formula components
Full size table

Diabetes Medication Administration with Enteral Nutrition

Regardless of the tube feed formula administered, medications are often required to maintain targeted glycemic levels in hospitalized patients [2]. Several of the oral hypoglycemic agents are suitable to achieve this goal in mildly hyperglycemic patients. Metformin is generally not used for glycemic control with tube feeds, especially when there is the potential for reduced renal function [2]. Second-generation sulfonylureas provide effective glycemic control, although care should be taken to avoid hypoglycemia, especially with tube feed interruptions. Glucosidase inhibitors may also be useful for glycemic control in hospitalized patients but their effect with tube feeds is minimal. Meglitinides may be useful in patients who receive bolused tube feeds, although in our experience, they are less efficacious as rapid-acting insulin.

Other oral agents, such as dipeptidyl peptidase-4 inhibitors, and glucagon-like peptide-1 analogues may be effective with EN. To our knowledge, the use of either of these agents has not been studied in this setting. The latency period that many patients experience on initiation of thiazolidinedione therapy may preclude their use in hospitalized patients receiving EN [2].

In patients with moderate to severe hyperglycemia or for those who cannot be controlled with oral agents, treatment with subcutaneous insulin can normalize glucose levels. Patients who are glucotoxic from prolonged hyperglycemia should also be treated with insulin, but achieving caloric targets may need to be delayed until glycemic control achieved.

The type of insulin and administration schedule depends mainly on the tube feeding schedule, which can be either continuous over 24 h, cycled for usually 12–18 h each day, or as bolused feed usually three or four times each day. The tube feed schedule often depends on the clinical situation for each patient.

Glycemic control can be achieved in patients who receive continuous tube feeds without expected interruption with once- or twice-daily insulin glargine or detemir. Often, interruptions in tube feeds cannot be predicted or anticipated. In patients who are prescribed glargine or detemir, 5% or 10% dextrose solutions should be kept at the bedside and infused if tube feeding is stopped.

Another strategy to prevent hypoglycemia that can result from sudden tube feed discontinuation is to use neutral protamine Hagedorn (NPH) insulin. NPH insulin can be administered safely, without insulin stacking every 6–8 h in place of longer-acting insulin. Because of the shorter duration of activity, doses of NPH insulin may be adjusted more frequently, and if tube feed administration is interrupted, subsequent NPH insulin are held until tube feeds are resumed. For patients on bolused tube feeds, basal/bolus insulin dosing is used. A rapid-acting insulin bolus should be administered with each tube feed bolus, titrated to the 1- to 2-h post-bolus glucose level, and insulin glargine or detemir titrated to the AM (fasting) glucose level. Patients on cycled tube feeds should be treated with NPH insulin every 6–8 h during the period in which the tube feeds are infusing.

Critically ill patients receiving EN often require intensive insulin therapy (IIT) to achieve glycemic control [2, 19]. Although there is continued debate over blood glucose targets in critically ill patients, IIT remains a powerful tool that addresses rapid changes in nutrient administration and insulin sensitivity commonly seen in critically ill patients [39]. In general, glycemic targets of 140–180 mg/dL should be used in the ICU, although tighter targets of 100–140 mg/dL or even 80–110 mg/dL in ICU settings have demonstrated safety and efficacy in some trials, especially in cardiothoracic surgery patients [2, 9, 39, 40].

For all patients receiving EN, the diabetic medication regimen should be titrated, based on tube feed rate and formula adjustments, to prevent fluctuations in blood glucose levels.

Parenteral Nutrition in Diabetes

Generally speaking, intravenously administered nutrition is indicated in patients who are malnourished and have compromised GI tract function, for those in which successful access to the GI tract cannot be achieved, or for supplementation if adequate EN cannot be delivered [11•, 41•]. Hyperglycemia is commonly seen in patients receiving PN, especially in association with diabetes, and results in increased mortality and infection rates [42, 43]. Hyperglycemia in patients receiving PN is a consequence of both insulin resistance with illness and intravenous dextrose, which circumvents the incretin response [19, 41•, 43]. Muscle wasting that occurs in malnourished critically ill patients also contributes to insulin resistance due to decreased insulin responsive tissue [11•, 19, 41•, 43].

The PN formula can be modified to prevent or treat hyperglycemia (Table 3). Dextrose (D-glucose) is the carbohydrate source for PN. Reducing the carbohydrate content and adding insulin to the PN formula can lower blood glucose levels. The dextrose content of PN is based on actual body weight for malnourished individuals and ABW in obese individuals [11•]. Limiting the initial dextrose infusion rate to approximately 1.5–2 mg/kg of body weight per minute is recommended to avoid hyperglycemia and to prevent the development of hepatic steatosis [11•, 19, 41•]. Once potential detriments of PN are controlled, then formulas can be advanced to targets. This paradigm differs from a less favored approach in which a priori targets are achieved rapidly and then detriments are managed, after they have already occurred [44].

Table 3 Parenteral nutrition formula adjustments that may improve glycemic control
Full size table

A second PN formula modification that may help normalize blood glucose levels is to reduce the lipid content. Theoretically and according to the Randle hypothesis, fatty acid oxidation inhibits glucose oxidation, or high plasma fatty acid concentration and high muscular fat content increase insulin resistance [45]. Consequently, the intravenous infusion of lipid can increase insulin resistance and promote hyperglycemia [46, 47]. By cutting down on the lipid content in the PN formula, insulin resistance may be reduced and blood glucose levels may improve.

The addition of regular insulin directly into the PN formula should be considered in all diabetic patients, prediabetic patients, and nondiabetic patients already on insulin for stress hyperglycemia. This method is preferred over the exclusive use of subcutaneous insulin to manage hyperglycemia with PN.

The dose of regular insulin added to the PN formula can be based on the total subcutaneous or intravenous insulin that the patient was receiving prior to the initiation of PN or it can be based on the amount of dextrose given in the PN formula. A reasonable starting point is 0.1 units of regular insulin for every gram of dextrose with titration as necessary [11•, 41•]. If hyperglycemia persists, especially in patients with loss of lean body mass, combining down-adjustments in dextrose content and up-adjustments in insulin content should be considered [11•]. For T1DM patients receiving PN, a minimal dose of insulin glargine or detemir (we suggest 0.05–0.1 units/kg) should also be administered to prevent diabetic ketosis and ketoacidosis in case the PN infusion is stopped.

Micronutrient additives to the PN formula may also improve glycemic control. In deficient patients, replacement of the trace element chromium (usually in doses of 20–40 μg/day) can improve blood glucose levels and insulin resistance [48, 49]. Doses of chromium as high as 200 μg/day have been given in PN without clinical consequence [48]. Presently, accurate tests for total body chromium stores are not available. Thus, empiric therapy with chromium should be considered in patients at high nutritional risk with evidence of severe insulin resistance requiring very high amounts of insulin or continued carbohydrate underfeeding [49].

Conclusions

Over the past decade, closely monitored blood glucose levels have been recognized as an important aspect of medical care in hospitalized patients, particularly those who are critically ill. The achievement of glycemic control is associated with improved clinical outcome [2, 7–9]. The administration of nutrition support to hospitalized patients to treat or prevent malnutrition is independently associated with better hospital outcomes. Special care must be taken to treat and prevent hyperglycemia in patients with diabetes receiving either EN or PN. The optimal approach to glycemic control is to synchronize nutrition support with insulin or other diabetes medications.