Residency · Residency · Endocrinology

Inpatient Diabetes Management

Scope and Importance

Epidemiology

Diabetes and hyperglycemia are among the most prevalent conditions encountered in the hospital setting, affecting approximately 25-30% of all hospitalized patients, a prevalence that continues to rise in parallel with the diabetes epidemic. Inpatient hyperglycemia, including stress hyperglycemia in patients without a prior diabetes diagnosis, independently predicts worse clinical outcomes across virtually every metric: increased mortality, higher rates of surgical site and nosocomial infections, prolonged length of stay, and elevated healthcare costs. Hypoglycemia in the hospital setting is similarly associated with increased mortality, though this likely reflects its role as a marker of illness severity and the dramatic swings in insulin sensitivity that accompany acute physiological stress rather than a direct causal relationship. The involvement of endocrinology consult services has been shown to improve glycemic control, reduce hypoglycemia episodes, and shorten hospital length of stay, underscoring the value of specialized expertise in this complex clinical arena.

Glycemic Targets (ADA/Endocrine Society)

The glycemic targets recommended by the ADA and the Endocrine Society for hospitalized patients reflect the balance between the harms of hyperglycemia and the dangers of hypoglycemia. For general medical and surgical ward patients, a target range of 140-180 mg/dL is recommended for most patients. A more stringent range of 110-140 mg/dL may be considered for select patients if achievable without significant hypoglycemia, such as clinically stable post-cardiac surgery patients. Conversely, less stringent targets of 180-250 mg/dL or higher are appropriate for terminally ill patients receiving end-of-life care, those with severe comorbidities, or patients at high risk for hypoglycemia. For the intensive care unit setting, the NICE-SUGAR trial established a target of 140-180 mg/dL as the standard, with avoidance of glucose levels below 110 mg/dL. Pre-meal glucose should ideally remain below 140 mg/dL and random glucose below 180 mg/dL. Hypoglycemia definitions are stratified: below 70 mg/dL constitutes an alert value, below 54 mg/dL represents clinically significant hypoglycemia, and any event requiring third-party assistance defines severe hypoglycemia.

Landmark Inpatient Glycemia Trials

The evidence base for inpatient glycemic management has been shaped by several pivotal trials. The Van den Berghe study in 2001 demonstrated a dramatic 34% mortality reduction with tight glycemic control (80-110 mg/dL) versus conventional management (180-200 mg/dL) in a surgical ICU population. While paradigm-shifting at the time, this was a single-center study that proved difficult to replicate. The NICE-SUGAR trial in 2009, a large multicenter ICU trial, delivered the definitive counterpoint: intensive glucose control (81-108 mg/dL) resulted in higher mortality (27.5% versus 24.9%) compared to conventional targets (144-180 mg/dL), driven largely by increased severe hypoglycemia in the intensive arm. This landmark trial established 140-180 mg/dL as the standard ICU glucose target that persists today. Outside the ICU, the RABBIT 2 trial by Umpierrez in 2007 demonstrated that basal-bolus insulin (glargine plus glulisine) achieved significantly lower mean glucose (166 versus 193 mg/dL) compared to sliding scale insulin alone in non-ICU type 2 diabetes patients, without an increase in hypoglycemia. This trial established basal-bolus insulin as the evidence-based standard of care for non-critical inpatient diabetes management. The RABBIT 2 Surgery trial in 2011 confirmed the superiority of basal-bolus over sliding scale in surgical patients.

Insulin Therapy in the Hospital

Basal-Bolus-Correction Regimen (Standard of Care for Non-ICU)

Calculating Initial Insulin Dose

The initial inpatient insulin dose requires careful calculation tailored to the individual patient's circumstances. For patients who are known insulin users, the starting inpatient dose is typically 50-80% of the home total daily dose (TDD), reduced from the outpatient dose to account for variable oral intake, potential changes in renal function, and the risk of NPO status. For insulin-naive type 2 diabetes patients, a starting dose of 0.3-0.5 units/kg/day is appropriate, with the lower end of this range (0.3 units/kg) used for patients older than 70 years, those with eGFR below 45, BMI under 25, or a history of hypoglycemia, and the higher end (0.5 units/kg) reserved for obese patients, those on high-dose corticosteroids, patients with significant insulin resistance, or those presenting with glucose above 300 mg/dL. For type 1 diabetes patients, the dose should be maintained at or near the home TDD, and basal insulin must never be discontinued, as doing so places the patient at risk of DKA within hours.

Dose Distribution

The calculated TDD is divided with approximately 50% allocated to basal insulin (glargine or degludec administered once daily, or NPH twice daily when cost is a concern) and 50% to bolus insulin (divided equally across three meals using rapid-acting analogs such as lispro, aspart, or glulisine). A correction scale is then added to the scheduled bolus doses, calculated using the insulin sensitivity factor.

Correction Scale Design

The insulin sensitivity factor (ISF) is estimated using the formula ISF = 1800/TDD (or 1700/TDD as a common approximation), which predicts the glucose drop in mg/dL expected from one unit of rapid-acting insulin. Typical correction scale categories are organized by insulin sensitivity: a low-dose (insulin-sensitive) scale administers 1 unit per 50 mg/dL above target, a moderate-dose scale gives 1 unit per 30-40 mg/dL above target, and a high-dose (insulin-resistant) scale provides 1 unit per 20-25 mg/dL above target. The correction dose is critically added on top of the scheduled mealtime dose, not given in isolation.

Why Sliding Scale Alone Is Inadequate

Sliding scale insulin as monotherapy is a reactive approach that treats hyperglycemia only after it has already occurred rather than preventing it proactively. Without basal insulin, there is no physiological background insulin coverage, which is particularly dangerous in type 1 diabetes where ketogenesis can develop within hours. The resulting glucose profile follows a characteristic sawtooth pattern of wide excursions between hyperglycemia and relative hypoglycemia. Sliding scale alone does not improve outcomes and is inferior to basal-bolus therapy in every randomized controlled trial that has compared the two approaches. Despite the overwhelming evidence against it, sliding scale-only insulin persists as a legacy practice in many institutions. The use of correction-only insulin without scheduled basal-bolus dosing should be limited to specific scenarios such as new-onset hyperglycemia with uncertain insulin requirements, patients with very low dose requirements, or the initial 24-48 hour assessment period while determining insulin needs.

<image>A comparison diagram showing two 24-hour glucose profiles. Top panel: "Sliding Scale Only" showing a sawtooth pattern with glucose swinging from 100 mg/dL to 350 mg/dL repeatedly, with reactive insulin doses shown as arrows after each peak, and a "Zone of hyperglycemia" shaded in red above 180. Bottom panel: "Basal-Bolus Regimen" showing relatively stable glucose between 120-180 mg/dL with a steady basal insulin line (shown as a continuous blue line along the bottom) and three mealtime bolus arrows at breakfast, lunch, and dinner, plus correction doses as needed. The target range (140-180) shaded in green. Include labels showing typical glucose values and insulin timing. Use clean medical chart style.</image>

Insulin Dose Adjustments

Daily Titration Approach

Inpatient insulin dosing requires daily review and adjustment based on the preceding day's glucose data. If fasting glucose is consistently above 180 mg/dL, the basal insulin dose should be increased by 10-20%. If fasting glucose falls below 100 mg/dL or nocturnal hypoglycemia occurs, the basal dose should be decreased by 10-20%. If pre-meal glucose values are consistently above 180 mg/dL, the corresponding mealtime bolus dose should be increased by 10-20%. If pre-meal glucose is below 100 mg/dL, the preceding mealtime dose should be decreased by 10-20%. When patients are eating inconsistently, mealtime insulin should be held if no meal tray has been ordered or the patient is NPO, and correction-only insulin should be administered with meals.

NPO Management

Management of patients who are not receiving oral intake requires particular attention to insulin management. For type 1 diabetes patients, basal insulin must be continued at 80-100% of the usual dose; basal insulin should never be stopped in T1D as this risks catastrophic DKA. For type 2 diabetes patients, basal insulin should be continued at 50-75% of the usual dose, or at the full dose if the patient is using degludec or glargine with stable dosing. All mealtime and bolus insulin should be held. Correction insulin using subcutaneous rapid-acting analogs should be provided every 4-6 hours. Glucose monitoring should occur every 4-6 hours, and a dextrose 5% or 10% infusion should be initiated if glucose falls below 100 mg/dL, particularly in type 1 diabetes.

Perioperative Insulin Management

Perioperative diabetes management requires coordination between the endocrine, surgical, and anesthesia teams. Oral diabetes medications should be held the morning of surgery, with particular attention to SGLT2 inhibitors (held 3-4 days preoperatively due to the risk of euglycemic DKA) and sulfonylureas (held the morning of surgery due to hypoglycemia risk). On the day of surgery for type 2 diabetes patients, 50-80% of the basal insulin dose should be given, mealtime insulin held, and a dextrose 5% infusion started if the patient is NPO. For type 1 diabetes patients, 80-100% of the basal insulin dose should be maintained, mealtime insulin held, and an intravenous insulin drip initiated for major surgical procedures. Intraoperative management includes hourly glucose monitoring and IV insulin infusion for major surgery or critical care settings.

IV Insulin Infusion (ICU Setting)

Indications

Intravenous insulin infusion is indicated in several clinical scenarios: DKA and hyperosmolar hyperglycemic state, critically ill patients with glucose persistently above 180 mg/dL, perioperative management of major surgery, type 1 diabetes patients who are NPO with unstable glucose, organ transplant perioperative periods, and high-dose glucocorticoid-induced hyperglycemia not adequately controlled with subcutaneous insulin.

Protocol Design

Numerous validated IV insulin protocols exist, including the Yale, Portland, Emory, and various institutional protocols. A typical starting rate is calculated as (glucose minus 60) multiplied by a coefficient (0.01-0.03, depending on estimated insulin sensitivity). Glucose monitoring should occur every 1-2 hours initially and may be extended to every 2-4 hours once glucose values are stable. Nurse-driven protocols with clear titration instructions should specify rate decreases when glucose is falling rapidly, increases when glucose is rising, and insulin holds when glucose drops below 70 mg/dL.

Transition from IV to SC Insulin

The transition from intravenous to subcutaneous insulin is a critical juncture that demands meticulous timing. The first dose of subcutaneous basal insulin (glargine or degludec) should be administered 2-4 hours before discontinuing the drip; if NPH is used as the basal agent, a 1-2 hour overlap is sufficient. The total daily dose of subcutaneous insulin is calculated from the last 6-24 hours of stable IV infusion: TDD (subcutaneous) equals the IV rate in units per hour multiplied by 24 and then by 0.8 (the 80% adjustment accounts for the difference in bioavailability between IV and subcutaneous routes). This calculated dose is split 50/50 between basal and bolus insulin if the patient is eating, or given primarily as basal insulin if the patient remains NPO. The overlap period is critical because the half-life of IV insulin is only 5-7 minutes, and any gap in coverage will result in rapid hyperglycemia.

Special Inpatient Situations

Glucocorticoid-Induced Hyperglycemia

Glucocorticoid-induced hyperglycemia is exceedingly common and can occur within hours of the first steroid dose. The glycemic pattern depends on the pharmacokinetic profile of the specific corticosteroid used. Prednisone or prednisolone administered as a single morning dose produces predominantly afternoon and evening hyperglycemia due to the peak pharmacodynamic effect occurring several hours after administration, while fasting glucose may remain near-normal. NPH insulin given concurrently with the morning steroid dose provides an excellent pharmacokinetic match, with a starting dose of 0.1-0.3 units/kg titrated upward as needed; alternatively, mealtime insulin for lunch and dinner may be increased. Dexamethasone, with its longer duration of action, produces more sustained hyperglycemia throughout the day, typically requiring basal insulin (glargine) in addition to prandial insulin. High-dose pulse corticosteroids such as methylprednisolone 1 gram can cause severe hyperglycemia that may necessitate an intravenous insulin drip. A critical principle is that insulin requirements correlate directly with steroid dosing: as steroids are tapered, insulin must be reduced proportionally to prevent hypoglycemia. An important screening recommendation is to check glucose 4-6 hours after the first steroid dose in all patients, as fasting glucose alone will miss steroid-induced hyperglycemia.

Enteral and Parenteral Nutrition

For patients receiving continuous enteral tube feeds, basal insulin covers the continuous carbohydrate delivery, supplemented by rapid-acting correction insulin every 4-6 hours. A reasonable starting dose is 1 unit of insulin per 10-15 grams of carbohydrate in the formula. A critical safety consideration is that if tube feeds are interrupted or held, the insulin regimen must be immediately reduced or held to prevent hypoglycemia. For bolus enteral feeds, each bolus is treated as a meal with scheduled rapid-acting insulin. Total parenteral nutrition (TPN) management involves adding regular insulin directly to the TPN bag, starting at 0.1-0.2 units per gram of dextrose in the TPN, supplemented by subcutaneous correction insulin. Because insulin in the TPN bag takes 24 hours to adjust (the bag cannot be easily modified once mixed), an external insulin drip may be needed initially for titration, with transition to in-bag insulin once the dose is stabilized.

Insulin Pump Management in Hospital

The optimal approach to insulin pump management during hospitalization depends on patient status and institutional capability. When the patient is competent, alert, and willing to self-manage, patient-directed pump management with endocrine team oversight is ideal. Some institutions have protocols for hospital staff to manage pumps, though this requires specialized training and is not universally available. Pump therapy should be transitioned to IV or subcutaneous basal-bolus insulin (using home TDD for dose calculation) in situations including altered mental status, ICU admission, surgery, MRI, or other imaging. All pump settings including basal rates, ICR, ISF, and active insulin time should be documented upon admission. CGM may be continued during hospitalization if institutional policy allows, though supplementation with point-of-care glucose testing per protocol is typically required.

Discharge Planning

Effective discharge planning is essential to prevent post-discharge glycemic deterioration and readmission. Home diabetes medications should be reconciled with any inpatient changes. Patients newly started on insulin require comprehensive education covering injection technique, self-monitoring of blood glucose, hypoglycemia recognition and management, and sick day rules. An HbA1c above 8-9% at admission suggests that therapy intensification at discharge is warranted. Referral to outpatient diabetes education and endocrinology follow-up within 1-2 weeks should be arranged. Prescription review should ensure the patient has all necessary supplies including insulin, syringes or pens, glucose meter, test strips, and glucagon. Barriers to effective diabetes management including insurance coverage, medication cost, health literacy, and food security should be identified and addressed before discharge.

<image>A practical decision tree for inpatient diabetes management. Start with: "Patient admitted with diabetes or hyperglycemia." First decision: Is patient eating? If YES: start basal-bolus-correction regimen (calculate TDD, 50% basal, 50% bolus divided across meals, add correction scale). If NOT eating (NPO): continue basal insulin at 50-80% (T2D) or 80-100% (T1D), hold mealtime insulin, give correction q4-6h, consider D5 drip. Second decision branch: Is patient in ICU with glucose >180? If YES: start IV insulin protocol targeting 140-180 mg/dL. Additional branches for special situations: on steroids (add NPH with morning prednisone), on tube feeds (basal + correction q4-6h for continuous; basal + bolus for bolus feeds), on TPN (add insulin to bag). Include a transition box showing IV-to-SC protocol with 2-4 hour overlap. Use clean clinical algorithm format.</image>

Hypoglycemia Prevention and Management

Hypoglycemia prevention in the hospital setting requires systematic identification of high-risk patients, including the elderly, those with renal or hepatic failure, patients with sepsis, those with reduced oral intake, and patients with adrenal insufficiency. Insulin dose reduction should be triggered by NPO status, reduced meal consumption, improving insulin sensitivity (as infection or stress resolves), decreasing steroid doses, and new renal impairment. A "hypoglycemia bundle" approach incorporating nurse education, timely meal tray delivery, bedside glucose assessment before insulin dosing, and insulin dose holding when meals are not consumed has been shown to reduce inpatient hypoglycemia rates. When hypoglycemia occurs, treatment consists of 15-20 grams of fast-acting glucose (juice, glucose tablets, or D50 intravenously) with glucose rechecked at 15 minutes. Root cause analysis should be performed for every hypoglycemic event, and insulin orders should be adjusted the same day.

Non-Insulin Agents in Hospital

Oral and Injectable Agent Use

Most oral diabetes medications should generally be held during acute hospitalization. Metformin should be held during acute illness, in the setting of renal impairment, for contrast procedures, and in hypoperfusion states. SGLT2 inhibitors should be held during acute illness due to DKA risk, although there is growing evidence supporting their continuation for the heart failure indication in stable patients who are eating, with plans to restart at discharge. Sulfonylureas should be held due to unpredictable meal intake and hypoglycemia risk. DPP-4 inhibitors may be continued and have been shown to be safe as monotherapy or as an adjunct to basal insulin for mild-to-moderate hyperglycemia. The SITA-HOSPITAL trial demonstrated that sitagliptin combined with basal insulin achieved similar glycemic control to basal-bolus insulin in patients with mild type 2 diabetes (glucose below 250 mg/dL) with less hypoglycemia, suggesting a promising simplified approach for less complex patients. GLP-1 receptor agonists have limited inpatient data and are generally held during acute illness, with particular concern about delayed gastric emptying and aspiration risk; they may be continued in clinically stable patients. Home diabetes medications should be restarted at discharge when the acute illness has resolved and no contraindications remain.

Steroid Management at Discharge

Patients discharged on corticosteroids require clear instructions for insulin dose titration during steroid taper. A useful rule of thumb is to reduce the insulin dose by the same percentage as the steroid dose reduction. Glucose monitoring supplies should be provided along with follow-up within one week. If steroid use has induced new-onset diabetes, the hyperglycemia may resolve completely when steroids are discontinued, but close monitoring is essential.

Quality Metrics and Institutional Programs

Glycemic Management Programs

Institutional glycemic management programs have been shown to systematically improve inpatient diabetes care. Key components include standardized insulin order sets that mandate basal-bolus with correction dosing and eliminate sliding-scale-only orders, glucose monitoring protocols (point-of-care glucose before meals and at bedtime for eating patients, every 4-6 hours for NPO or tube-fed patients), hypoglycemia reduction initiatives with root cause analysis, and dashboard metrics tracking the percentage of patient-days within the target range, hypoglycemia rates (at both below 70 and below 54 mg/dL thresholds), and insulin appropriateness audits. Consultation triggers for the endocrine or glycemic management team should be established, including glucose above 250 mg/dL on two or more occasions, insulin drip use exceeding 48 hours, DKA or HHS, and type 1 diabetes admission.

Key Clinical Pearls

• NEVER discontinue basal insulin in a T1D patient, even when NPO; this is the single most common cause of preventable DKA in hospitalized T1D patients; reduce dose to 80% if NPO but do not stop
• Sliding scale insulin alone is a substandard practice; basal-bolus-correction is the evidence-based standard of care per RABBIT 2 and subsequent trials
• After NICE-SUGAR, the ICU glucose target of 140-180 mg/dL became standard; intensive glucose control (<110 mg/dL) increases mortality from hypoglycemia
• Glucocorticoid-induced hyperglycemia peaks in the afternoon/evening with morning prednisone; fasting glucose alone will miss it; NPH given with the steroid dose is a pharmacokinetically rational approach
• When transitioning from IV to SC insulin, give the first SC dose 2-4 hours BEFORE stopping the drip; IV insulin has a 5-7 minute half-life, and gaps in coverage lead to rapid hyperglycemia
• DPP-4 inhibitors (sitagliptin) represent a promising simplified approach for mild-moderate inpatient hyperglycemia in T2D (glucose <250 mg/dL) with less hypoglycemia than basal-bolus; may reduce insulin burden for less complex patients

References

1. Umpierrez GE, et al. "Randomized Study of Basal-Bolus Insulin Therapy in the Inpatient Management of Patients with Type 2 Diabetes (RABBIT 2 Trial)." Diabetes Care. 2007;30(9):2181-2186.
2. NICE-SUGAR Study Investigators. "Intensive versus Conventional Glucose Control in Critically Ill Patients." N Engl J Med. 2009;360(13):1283-1297.
3. Umpierrez GE, et al. "Management of Hyperglycemia in Hospitalized Patients in Non-Critical Care Setting: An Endocrine Society Clinical Practice Guideline." J Clin Endocrinol Metab. 2012;97(1):16-38.
4. Pasquel FJ, et al. "Efficacy of Sitagliptin for the Hospital Management of General Medicine and Surgery Patients with Type 2 Diabetes (SITA-HOSPITAL)." Diabetes Obes Metab. 2017;19(11):1524-1529.
5. ElSayed NA, et al. "Diabetes Care in the Hospital: Standards of Care in Diabetes - 2024." Diabetes Care. 2024;47(Supplement_1):S295-S306.

Read this lecture as Markdown