JavaScript has to be enabled to view this site. Learn how to enable JavaScript.


3. Inpatient Diabetes Management

Inpatient care may be appropriate in the following situations (American Diabetes Association, 2004a):

  • Elderly patients with infection or illness, weight loss, dehydration, polyuria or polydipsia
  • Life-threatening acute metabolic complications of diabetes:
    • Hyperglycemic hyperosmolar state with impaired mental status, elevated plasma osmolaity that includes plasma glucose greater than 600 mg/dL
    • Diabetic ketoacidosis with a plasma glucose greater than 250 mg/dL, arterial pH less than 7.30 and serum bicarbonate level less than 15 mEq/L and the presence of moderate ketonuria and/or ketonemia
    • Hypoglycemia with neuroglycopenia that includes blood glucose less than 50 mg
  • Uncontrolled insulin-requiring diabetes during pregnancy
  • Surgery, infection, steroids – if these conditions cause significant hyperglycemia and rapid initiation of rigorous insulin is needed

Hospitalized patients with diabetes suffer increased morbidity, mortality, length of stay and other related hospital costs compared to non-hyperglycemic inpatients. These negative outcomes are observed more frequently in hospitalized patients with newly discovered hyperglycemia. Hyperglycemia is an independent marker of inpatient mortality in patients with undiagnosed diabetes (Umpierrez, 2002).

Hyperglycemia has been associated with increased infection rates and poorer short-term and long-term outcomes in critically ill patients in the intensive care unit, post-myocardial infarction and post-surgical settings. Earlier studies supported that aggressive glucose management in medical and surgical patients improves outcomes (Van den Berghe, 2001). More recently, intensive management has been linked to increased hypoglycemia and increased mortality in a subset of patients including those with a long history of diabetes and cardiovascular disease (NICE-SUGAR Study Investigators, The, 2009).

The following are suggestions for the inpatient setting (American Diabetes Association, 2014; Clement, 2004):

  • Insulin therapy with intravenous insulin in critically ill patients (Van den Berghe, 2001)
  • Oral glycemic agents may need to be held or the dose adjusted if the patient is hospitalized
  • Use of scheduled insulin, with basal coverage (improves glucose control compared to sliding scale coverage alone)
  • For insulin-deficient patients, despite reductions or the absence of caloric intake, basal insulin must be provided to prevent diabetic ketoacidosis
  • Target preprandial plasma glucose levels to 90-140 mg/dL (American Diabetes Association, 2014; NICE-SUGAR Study Investigators, The, 2009; American Diabetes Association, 2004a; Clement, 2004; Garber, 2004)
  • Target random plasma glucose to less than 180 mg/dL (American Diabetes Association, 2014; Holman, 2009; NICE-SUGAR Study Investigators, The, 2009; American Diabetes Association, 2004a; Clement, 2004; Garber, 2004)
  • A protocol should be utilized for patients with hypoglycemia < 70 mg/dL (American Diabetes Association, 2014; Cryer, 2003)
  • Establishing a multidisciplinary team that sets and implements institutional guidelines, protocols and standardized order sets for the hospital results in reduced hypoglycemic and hyperglycemic events

Other considerations include (Clement, 2004):

  • For patients who are alert and demonstrate accurate insulin self-administration and glucose monitoring, insulin self-management should be allowed as an adjunct to standard nurse-delivered diabetes management.
  • Patients with no prior history of diabetes who are found to have hyperglycemia (random fasting blood glucose greater than 125 mg/dL or random glucose of 200 mg/dL or more) during hospitalization should have follow-up testing for diabetes within one month of hospital discharge (Umpierrez, 2002).

Types of Insulin

Based on outpatient studies, consider insulin Glargine or Detemir as the basal insulin (there are limited inpatient studies to date).  In studies comparing Glargine to NPH, the risk of nocturnal hypoglycemia was reduced (Wang, 2003; Yki-Järvinen, 2000).  Treatment with insulin Detemir resulted in more predictable glycemic control than NPH insulin (Vague, 2003).

Consider using rapid-acting insulin analogs (e.g., lispro, aspart, glulisine instead of regular insulin) unless the patient is to have nothing by mouth or is on continuous feedings.  Initial studies comparing rapid-acting insulin with human regular insulin show rapid-acting insulins to be more effective at reducing the peak postprandial glucose concentration (Reynolds, 2004).  They may also lower the demand for endogenous insulin, provide superior postprandial glycemic control, and cause fewer hypoglycemic episodes requiring medical intervention (Rave, 2006; Pettitt, 2003; Gerich, 2002).

Insulin lispro, glulisine and aspart have similar pharmacokinetics; they have an earlier onset and peak of action than regular insulin.  Peak action usually occurs at one hour with a duration of three to four hours, while regular insulin has a peak action of two to four hours and a duration of six to eight hours.  Lispro, glulisine and aspart may then reduce the occurrence of late postprandial hypoglycemia compared to regular insulin (Guerci, 2005; John, 2004).

Insulin Dosing Schedule

Insulin dosing schedules must be individualized based on a variety of factors, including the severity of diabetes, oral intake, severity of illness and other concurrent diabetic medication.  It is not feasible to design a single algorithm for determining an insulin regimen in every patient.  The following information provides general guidance in determining initial insulin doses.

Healthy, non-diabetic people are estimated to secrete approximately 0.4-1.0 units of insulin/kg body weight per day (Polonsky, 1988a; Davidson, 1986).  Approximately 50% of this insulin is secreted as basal insulin and 50% as postprandial boluses following meals (Polonsky, 1988b).  Typical daily insulin doses for people with diabetes range from 0.5 to 0.7 units/kg per day.  In the United Kingdom Prospective Diabetes Study of people with T2DM, the median daily insulin dose for people in the intensive insulin treatment arm of the study after a diabetes duration of approximately 12 years was 36 units/day (UK Prospective Diabetes Study Group, 1998a).

Fifty percent of subjects were receiving between 23 and 53 units of insulin per day.  The average weight of subjects was 75 kg, so the "average" daily insulin requirement was about 0.5 units/kg (UK Prospective Diabetes Study Group, 1998a).  Therefore, in initiating subcutaneous insulin in a hospitalized patient who is eating meals, a total daily insulin dose of 0.6 units/kg is probably reasonable (Clement, 2004).  Modification can be made based on clinical judgment for factors such as severity of illness, fragility, renal function, body weight, expected nutritional intake and medication effects (e.g., glucocorticoid medications).

Based on the normal physiology of insulin release and experience with outpatient regimens for managing diabetes with subcutaneous insulin, it has been recommended that inpatient subcutaneous insulin regimens comprise three components (Clement, 2004):

  • A basal insulin component
  • A prandial insulin component (for patients eating meals)
  • A correction, sometimes referred to as "supplemental," insulin component used to treat hyperglycemia before or between meals (Clement, 2004)

In a small, randomized trial comparing a basal/prandial insulin regimen to a traditional sliding scale insulin regimen in hospitalized patients with T2DM, the basal/prandial insulin regimen resulted in improved glycemic control during the hospitalization.  Hospital length-of-stay or incidences of hypoglycemia did not differ between the basal/prandial insulin regimen or the sliding scale insulin regimen (Umpierrez, 2007).

Basal insulin

Typical approach is to give 40-50% of the estimated total daily insulin dose as the basal insulin component.  Common basal regimens include one injection per day of Glargine insulin, usually given at bedtime or twice daily; Detemir insulin given once daily in the evening or given twice daily; twice-a-day injections of NPH insulin, given at breakfast and either at supper or bedtime; or once-a-day NPH insulin given at bedtime (Vague, 2003).  Basal insulin would generally be appropriate for any patient being managed with subcutaneous insulin, whether eating meals, nothing by mouth or receiving nutrition as continuous enteral feeding or total parenteral nutrition (TPN) (Clement, 2004).

Prandial insulin

For patients eating meals, several approaches have been suggested to initiate a prandial insulin regimen:

  • Divide 50% of the estimated daily insulin requirement into three equal insulin doses given before the three meals.
  • Estimate the prandial insulin dose before each meal as 10-20% of the estimated daily insulin requirement.
  • Count the carbohydrate content of the meal (one carbohydrate unit = 15 gm of meal carbohydrate), and determine the prandial insulin dose as a set number of units of insulin per meal carbohydrate unit.
  • Insulin doses based on grams of carbohydrates consumed.

Typical insulin requirements using this last approach are one to two units of insulin per carbohydrate unit (Clement, 2004).

It is recommended that prandial insulin be given as a rapid-acting insulin analog within 0-15 minutes of the meal (Clement, 2004).  Prandial insulin replacement has its main effect on peripheral glucose disposal into muscle.  Also referred to as "bolus" or "mealtime" insulin, prandial insulin is usually administered before eating.  There are occasional situations when this insulin may be injected immediately after eating, such as when it is unclear how much food will be eaten.  In such situations, the quantity of carbohydrates taken can be counted and an appropriate amount of rapid-acting analog can be injected (Clement, 2004).

Patients who are not eating meals will not typically require a prandial insulin component, although they may need periodic correction insulin.

Correction (supplemental) insulin

Correction dose insulin is given in addition to the scheduled basal and prandial insulin in order to correct hyperglycemia.  For patients eating meals, it is typically given with meals by simply increasing the rapid-acting insulin dose by an additional amount based on the correction schedule.  For patients not eating meals (e.g., nothing by mouth, on continuous enteral feeding, total parenteral nutrition), it is reasonable to give periodic short-acting insulin, either as regular insulin or a rapid-acting analog, based on the correction schedule at four- to six-hour intervals (Guerci, 2005; John, 2004).  If rapid-acting insulin is used in this situation, an every-four-hour schedule may be optimal.  For regular insulin, a four- to six-hour schedule is reasonable (Clement, 2004).

The correction dose insulin schedule must be individualized for the patient.  A typical assumption is that one unit of insulin will lower the blood glucose 50 mg/dL (Hirsch, 2002).  An empiric "Rule of 1,700" has been proposed as one way of estimating the insulin correction requirement.  This rule estimates that the decrease in glucose in response to one unit of insulin = (1,700/patient's total daily insulin dose) (Davidson, 2003).  The "Low," "Medium" and "High" correction schedules included on the order set assume that one unit of insulin will lower the blood glucose by approximately 50, 25 and 15 mg/dL, respectively.

There does not appear to be a consensus whether correction insulin should be given at bedtime.  Some experts argue against bedtime correction insulin due to a fear of nocturnal hypoglycemia with short- or rapid-acting insulin given at bedtime (Hirsch, 1995).  If correction insulin is given at bedtime, the recommendation is that the correction dose should be reduced (Clement, 2004).

Hyperglycemia induced by corticosteroid therapy is often characterized by predominant postprandial hyperglycemia with lesser effects on fasting glucose levels.  For patients with corticosteroid-induced hyperglycemia, caution is suggested in prescribing correction dose insulin at bedtime due to the increased risk of nocturnal hypoglycemia (Clement, 2004).


The following is an example of one possible initial subcutaneous insulin regimen for a hospitalized patient weighing 100 kg with hyperglycemia who is eating meals.

Estimated total daily insulin dose = 100 kg x 0.6 units insulin/kg = 60 units of insulin daily.

Basal: 50% of total daily insulin dose = 30 units given as Glargine or Detemir insulin at bedtime.

Prandial:  50% of total daily insulin dose/3 = 30 units/3 = 10 units of insulin at each meal given as Lispro, Glulisine or Aspart insulin.

Correction schedule:  Assuming 1 unit of insulin will drop the blood glucose 50 mg/dL, the "Low" correction schedule on the order set could be used.  Using the Rule of 1,700, one would estimate that one unit of insulin = drop in blood glucose of (1,700/60) = 28 mg/dL.  In this case, the "Medium" correction schedule might be chosen.

Whatever insulin regimen is initially implemented, it will likely need to be modified over the course of a patient's hospitalization.  If a patient is frequently requiring use of the correction schedule, common sense would dictate that either the basal component, prandial component or both need to be modified.

Transition from Intravenous to Subcutaneous Insulin

When transitioning from intravenous to subcutaneous insulin, it is generally recommended that an initial subcutaneous basal insulin dose of long- or intermediate-acting insulin be given prior to discontinuation of the intravenous insulin (Furnary, 2006).  Based on the absorption profiles of longer-acting insulins, administering the first subcutaneous insulin dose two hours prior to stopping the insulin infusion would appear to allow sufficient overlap to avoid excessive rebound hyperglycemia when the insulin infusion is discontinued (Furnary, 2006; Clement, 2004).

Determination of the initial basal insulin dose can be made using the guidelines above (e.g., estimating the basal insulin dose as 40-50% of the estimated total daily insulin dose).  An alternative method that has been suggested is to estimate the initial basal dose based on the intravenous insulin requirements over a six- to eight-hour period leading up to the transition time.  Ideally, this six- to eight-hour period would be a time when the patient was not eating and was not receiving intravenous glucose.  The initial basal insulin dose could be calculated as 80% of the estimated 24-hour insulin requirement to provide a margin of safety (Furnary, 2006).


A patient managed on an intravenous insulin drip is to be transitioned to subcutaneous insulin.  Over a recent six-hour period when the patient was not eating and was not receiving intravenous glucose, the patient received a total of 15 units of insulin via the infusion.  The estimated 24-hour basal insulin requirement would be 15 x 4 = 60 units.  The initial basal insulin dose could be estimated as 80% x 60 units = 48 units.

Often the clinician may want to use a bedtime long-acting insulin (e.g., Glargine insulin) as the subcutaneous basal insulin, but the transition from intravenous to subcutaneous insulin is planned to occur during the day.  In these cases, one option would be to give a one-time dose of NPH insulin by subcutaneous injection to act as a bridge until the regularly scheduled long-acting insulin is given (Clement, 2004).  A typical NPH insulin dose might be 40% of the planned long-acting insulin dose.


Using the example above, the clinician plans to give 48 units of Glargine insulin at bedtime as the basal insulin dose on the transition to subcutaneous insulin.  However, the clinician would like to transition the patient to subcutaneous insulin during the day rather than waiting until later in the evening when fewer staff are present.  A one-time order for NPH insulin 20 units (40% x 48 units = 19.2 units, round to 20 units) could be written to be given two hours before the insulin infusion is stopped.  This intermediate-acting insulin would provide temporary basal insulin coverage until bedtime, when the 48 units of Glargine insulin could be given.

Prandial and correction insulin orders should also be written as appropriate for the patient's situation (eating, on tube feeding, etc.) on transition to subcutaneous insulin.  This insulin would then be given in addition to the basal insulin in accordance with the order set.

Medication adherence

Non-adherence with medications can limit the success of therapy and help to explain why a patient is not achieving treatment goals. To screen for non-adherence, clinicians can ask patients open-ended, non-threatening questions at each office visit. The assessment should include probes for factors that can contribute to non-adherence (fear of adverse reactions, misunderstanding of chronic disease treatment, depression, cognitive impairment, complex dosing regimens or financial constraints).

  • Assess the patient's knowledge of his/her condition and his/her expectations for treatment
  • Assess the patient's medication administration process
  • Assess the patient's barriers to adherence

Interventions to enhance medication adherence should be directed at risk factors or causes of non-adherence. Interventions may include simplifying the medication regimen, using reminder systems, involving family or caregivers in care, involving multiple disciplines in team care, providing written and verbal medication instructions, setting collaborative goals with patients, and providing education about medications (including potential adverse effects) and about diabetes in general (Nichols-English, 2000).

Depression evaluation

There is a substantial increase in the prevalence of depression among people with diabetes as compared to the general adult population (Anderson, 2001). Depression impacts the ability of a person with diabetes to achieve blood glucose control, which in turn impacts the rate of development of diabetes complications (de Groot, 2001; Lustman, 2001).

Identification and management of depression are important aspects of diabetes care. Self-administered or professionally administered instruments, such as PHQ-9, are useful adjuncts to the clinical interview in the identification of depression. The ICSI Major Depression in Adults in Primary Care guideline provides more suggestions for the identification and management of depression. Intervention studies have demonstrated that when depression is treated, both quality of life and glycemic control improve. Counseling may be effective, especially among those who are having difficulty adjusting to the diagnosis of diabetes or are having difficulty living with diabetes. Pharmacotherapy for depression is also effective.

Obstructive sleep apnea

Sleep apnea is a prevalent condition in obese patients with T2DM and is associated with significant comorbidities including hypertension, cardiovascular disease and insulin resistance. Consider referral of symptomatic patients for sleep evaluation. Clinicians should be cognizant of potential obstructive sleep apnea, especially among obese patients (Foster, 2009a; Foster, 2009b).

Referral to an Extended Care Team

Diabetes educator

Consultation with a diabetes educator is suggested if the patient is having difficulty adhering to a nutrition, exercise and medication regimen, and the patient is having difficulty adhering to or accurately completing blood glucose monitoring or may need answers to his/her questions.

Primary care clinician should develop a relationship with a diabetes education program to provide other options for management. The American Diabetes Association publishes a list of recognized educational programs in each state. These programs may be staffed with endocrinologists or primary care clinicians plus diabetes educators including dietitians, nurses and other health care clinicians who are Certified Diabetes Educators or have didactic and experiential expertise in diabetes care and education.


Most T2DM management can be managed by a primary care clinician with periodic consultation as needed by an endocrinologist.  Consultation with a specialist is suggested if persistent proteinuria, worsening microalbuminuria and elevation in serum creatinine or blood urea nitrogen, or hypertension unresponsive to treatment is seen.


Consultation with a specialist is suggested if neuropathy progresses and becomes disabling.

Endocrinologist/cardiologist/hypertension specialist

Consultation with a specialist is suggested if blood pressure is refractory to treatment, or the patient has marked associated postural hypotension or symptoms of coronary artery disease.

Foot care specialist

A consultation with a specialist is suggested if the patient is unable to care properly for his/her own feet, needs prescriptive footwear and/or more serious problems such as foot deformities (e.g., Charcot deformity), infected lesions, and ulcers, deformed nails or thick calluses are present.


Retinopathy is estimated to take at least five years to develop after the onset of hyperglycemia begins.  Patients with T2DM who generally have had years of undiagnosed diabetes and who have a significant risk of prevalent diabetic retinopathy at time of diabetes diagnosis should have an initial dilated and comprehensive eye examination soon after diagnosis. Examinations should be performed by an ophthalmologist or optometrist who is knowledgeable and experienced in diagnosing the presence of diabetic retinopathy and is aware of its management. Subsequent examinations are generally repeated annually. Less frequent exams (every two to three years) may be cost effective after one or more normal eye exams, while examinations will be required more frequently if retinopathy is progressing (American Diabetes Association, 2014).

Vascular specialist/surgeon

Consider referral if patient has symptoms of peripheral vascular disease such as loss of pulses and/or claudication.