According to a recent 2025 review, insulin-based management is recommended for persistent inpatient hyperglycemia, with active insulin management generally prompted when blood glucose is ≥180 mg/dL and typical inpatient targets of 140–180 mg/dL for critically ill patients and 100–180 mg/dL for noncritically ill patients. For patients receiving total parenteral nutrition (TPN), regular human insulin is recommended to be added directly to the TPN admixture when insulin is incorporated into the PN bag, due to its stability and ability to control TPN-associated hyperglycemia. TPN-related hyperglycemia is common due to factors such as preexisting diabetes, baseline hyperglycemia, dextrose infusion rate, critical illness, and acute stress responses. Regular insulin may be added to the TPN at approximately 1 unit per 10-15 g of dextrose, then adjusted according to glycemic response and changes in dextrose delivery; if glycemic goals are not met, the suggested titration is to increase by 0.5 units per 10 g of dextrose daily. The paper also emphasizes that insulin dosing should be adjusted when the TPN dextrose content changes to avoid dysglycemia. Outside of TPN-specific dosing, the authors suggest to avoid correction-only or sliding-scale insulin as the sole strategy and instead use structured protocols that include basal, nutritional/prandial or dextrose-covering insulin, and correction insulin. For TPN patients this supports use of regular insulin in the PN bag as the preferred formulation when feasible, with ongoing bedside glucose monitoring and correction insulin as needed, while IV insulin infusion remains more aligned with critical care, perioperative settings, diabetic ketoacidosis, or uncontrolled hyperglycemia requiring rapid titration rather than routine maintenance TPN coverage. [1]
However, a slightly older 2013 review does not endorse one universal formulation for all patients. The preferred approach, according to these authors, depends on clinical stability. In critically ill or hemodynamically compromised patients, a separate continuous intravenous insulin infusion is preferred because it permits frequent dose adjustment as glucose exposure, insulin sensitivity, vasopressors, steroids, renal function, and illness severity change. For more stable patients receiving parenteral nutrition, the paper describes regular insulin added directly to the PN bag as a clinically safe and effective strategy, with supplemental short-acting correction insulin every 4 to 6 hours. Suggested starting insulin-to-dextrose ratios vary by diabetes status and prior evidence: in patients with known diabetes, 1 unit of regular insulin per 10 to 11 g dextrose is described as an effective initial approach, with daily titration by approximately 0.5 unit per 10 g dextrose if glycemic targets are not achieved; another cited study achieved glycemic goals with a more intensive 1:4 insulin-to-carbohydrate ratio. In patients without known diabetes who develop PN-associated hyperglycemia, the review describes lower starting insulin needs, such as 1 unit per 20 g dextrose with titration to 1:15 if blood glucose remains above 140 mg/dL; average insulin requirements in one protocol were about 0.3 ± 0.2 units/kg/day. One presented algorithm similarly recommends reducing PN dextrose exposure and mixing regular insulin into PN at about 7 to 10 units per 100 g dextrose for known diabetes and 5 to 7 units per 100 g dextrose for patients without diabetes, plus supplemental short-acting insulin every 4 to 6 hours. Non-insulin mitigation is recommended during PN titration, particularly limiting excessive dextrose delivery; dextrose infusion rates above 4 mg/kg/min were associated with hyperglycemia and insulin use, and limiting dextrose to roughly 150 to 200 g/day is presented as a reasonable strategy to reduce hyperglycemia risk. Overall, during unstable/titration phases, especially in critically ill patients, the authors favor IV insulin infusion for adjustability, whereas during more stable PN delivery, adding regular insulin to the PN admixture with daily titration and correctional insulin is presented as an efficient maintenance strategy. [2]
According to a 2022 review, intravenous rapid-acting insulin infusion is preferred for critically ill, unstable, or early titration-phase patients, while regular insulin added to the PN bag is reasonable for stable patients, especially when separate intravenous insulin administration is not feasible on general wards. The recommended inpatient glucose target is 140-180 mg/dL. During the first days of PN, the paper emphasizes frequent glucose monitoring because insulin needs can change rapidly; monitoring can be reduced once the patient reaches a stable metabolic state. For unstable or critically ill patients, continuous IV insulin is described as ideal because of favorable pharmacokinetics and maximal flexibility, allowing rapid adjustment without relying on once-daily PN-bag changes. The paper provides an initial IV insulin dose scheme starting at 1 IU/hour for blood glucose 108–143 mg/dL and increasing stepwise up to 8 IU/hour for glucose ≥360 mg/dL, with glucose checks every 2 hours; it also cautions against IV insulin boluses because they may cause glycemic fluctuations and electrolyte disturbances, especially hypokalemia. For patients with type 1 diabetes, a minimum basal IV insulin rate of 0.5–1 IU/hour should continue even if PN is paused. [3]
For stable patients, adding regular/short-acting insulin directly to the PN bag is presented as convenient and physiologically favorable, particularly for cyclic PN or when PN is frequently stopped for procedures because insulin delivery stops with nutrition delivery. However, this method is not appropriate when insulin requirements are unstable, since the dose should not be adjusted more often than every 24 hours and cannot be changed once mixed into the PN bag. The paper notes that insulin requirements vary widely: obese, insulin-resistant patients with type 2 diabetes may need as much as 1 unit per 5 g glucose, whereas thin patients with type 1 diabetes may require closer to 1 unit per 20 g glucose. [3]
Subcutaneous insulin is treated as a post-acute/stable-patient option, not the preferred approach for unstable PN titration. The paper recommends overlapping subcutaneous insulin by 2–3 hours before stopping IV insulin. For continuous 24-hour PN, the authors suggest giving 100% of the insulin as long-acting basal insulin subcutaneously; for continuous PN plus oral or enteral intake, 50% may be given as long-acting basal insulin and 50% as short-acting insulin; for cyclic overnight PN plus oral or enteral intake, 50% long-acting insulin with emphasis on nocturnal coverage and 50% short-acting insulin is suggested. The INSUPAR trial (see Table 1) is highlighted as showing no difference in overall metabolic control between 100% regular insulin added to PN versus 50% regular insulin in PN plus 50% subcutaneous glargine in noncritically ill type 2 diabetes patients, although glycemic control after PN interruption was better with glargine and hypoglycemia was more frequent in the glargine group. Notably, only regular/short-acting insulin is chemically stable for PN admixture; NPH, lispro, aspart, and glargine are described as incompatible with PN admixtures, and insulin should only be added after stability/compatibility is confirmed with pharmacy input. Overall, this paper supports a staged approach, recommending IV insulin during critical illness or active titration, regular insulin in the PN bag once PN and glycemia are stable, and subcutaneous basal with or without bolus strategies only in stable patients receiving continuous PN or mixed oral/enteral nutrition. [3]
Pooled data is presented in a 2023 systematic review/network meta-analysis which included 21 studies and 1,459 adult inpatients receiving TPN. The paper compared six insulin strategies: regular insulin added to PN (RI-in-PN; n=524), continuous IV insulin infusion (CVII; n=392), multiple subcutaneous insulin injections (MSII; n=124), continuous subcutaneous insulin infusion (CSII; n=196), subcutaneous glargine insulin (s.c. GI; n=142), and 50% RI-in-PN + 50% s.c. GI (n=81). Overall, the authors concluded that CVII was the most effective overall strategy for TPN-associated hyperglycemia because it performed best for reducing hyperglycemia, shortening hospital length of stay, and lowering total daily insulin dose, without a statistically significant increase in hypoglycemia. However, CVII was not best for mean blood glucose (MBG): MSII had the poorest MBG control, with significantly higher MBG than s.c. GI, CSII, RI-in-PN, 50% RI-in-PN + 50% s.c. GI, and CVII; reported WMDs versus MSII favored s.c. GI 3.24, CSII 3.13, RI-in-PN 2.73, 50% RI-in-PN + 50% s.c. GI 2.33, and CVII 1.60. CVII also had significantly higher MBG than s.c. GI and CSII, with WMDs of 1.64 and 1.53, respectively. By SUCRA ranking for MBG control, the best-ranked strategies were s.c. GI 85.8%, CSII 79.5%, RI-in-PN 64.5%, 50% RI-in-PN + 50% s.c. GI 42.5%, CVII 27.4%, and MSII 0.2%. [4]
Regarding safety, incidence of hypoglycemia did not differ significantly among the six strategies; all pairwise comparisons had confidence intervals crossing 1, so the authors considered the hypoglycemia risk generally comparable between MSII, CVII, s.c. GI, CSII, RI-in-PN, and 50% RI-in-PN + 50% s.c. GI. For hospital length of stay, CVII was the best-ranked intervention, while MSII was worst. Compared with CVII, length of stay was significantly longer with RI-in-PN by 2.74 days, s.c. GI by 2.83 days, CSII by 3.34 days, and MSII by 9.14 days; by SUCRA, the ranking from shortest to longest stay was CVII 92.5%, 50% RI-in-PN + 50% s.c. GI 65.8%, RI-in-PN 65.8%, s.c. GI 49.3%, CSII 39.8%, and MSII 0.6%. For hyperglycemia events, CVII was clearly favored, with lower risk than RI-in-PN, s.c. GI, and CSII; reported RRs favoring CVII were 0.56 vs RI-in-PN, 0.55 vs s.c. GI, and 0.54 vs CSII, and the SUCRA ranking for hyperglycemia reduction was CVII 99.8%, RI-in-PN 36.5%, s.c. GI 35.5%, and CSII 28.2%. [4]
For surgical site infection (SSI), CSII and CVII performed best. CSII had significantly lower SSI risk than RI-in-PN and s.c. GI, with reported RRs of 0.05 vs RI-in-PN and 0.08 vs s.c. GI, while CVII had significantly lower SSI risk than RI-in-PN, with an RR of 0.34. SUCRA ranking for SSI reduction was CSII 98.2%, CVII 60.4%, s.c. GI 34.4%, and RI-in-PN 7.0%. For mean total daily insulin dose, CVII required significantly less insulin than RI-in-PN and 50% RI-in-PN + 50% s.c. GI, with reported differences of 13.89 and 18.59 units/day, respectively; SUCRA ranking was CVII 100%, RI-in-PN 40.8%, and 50% RI-in-PN + 50% s.c. GI 9.2%. Overall, the data suggest that CVII is the preferred method when rapid titration and close monitoring are feasible, particularly for unstable or critically ill patients, while RI-in-PN, CSII, s.c. GI, and 50% RI-in-PN + 50% s.c. GI appear broadly interchangeable in stable patients based on overall efficacy and safety. MSII was the least favorable option, associated with worse MBG control and longer hospital stay. [4]