Treating Hyperglycemia and Diabetes With Insulin Therapy: Transition From Inpatient to Outpatient Care

Abstract and Introduction

Abstract

Context

Intensive insulin therapy is recommended to control glucose elevations in the critically ill and has been shown to significantly improve outcomes among hospital inpatients with acute hyperglycemia or newly diagnosed diabetes. Once discharged, the hyperglycemic patient may require ongoing outpatient care, most often under the attention of a primary care physician.

Evidence acquisition

The purpose of this review is to provide a background of in-hospital hyperglycemia management and discharge planning in preparation for continued outpatient care. Primary data sources were identified through a PubMed search (1990–2007) using keywords, such as diabetes, hyperglycemia, in-hospital, discharge, and insulin.

Evidence synthesis

Hyperglycemia protocols with strict glycemic goals have been shown to improve morbidity and mortality among critically ill inpatients. Discharge planning should prepare patients for self-care and give them the survival skills necessary to maintain glycemic control. In preparation for discharge, patients are usually transitioned from insulin infusions to subcutaneous insulin administered through an appropriate basal-prandial regimen.

Conclusion

A thorough understanding of hyperglycemia history and treatment will allow the primary care physician to deliver optimal diabetes care and thereby improve both short-term and long-term outcomes for those patients with critical illnesses and hyperglycemia or diabetes.

Introduction

Hyperglycemia, when left untreated, can have a negative impact on the patient's prognosis and outcome during the hospital stay and after discharge.[1–5] The prevalence of hyperglycemia in hospitalized patients is high, and may be associated with multiple factors: First, about 20.8 million Americans have diabetes, 6.2 million of whom (around one third) have not been diagnosed.[1,6] Furthermore, diabetes itself may contribute to hospitalization because it can lead to cardiovascular disease, renal damage, stroke, and/or other complications.[7–9] Finally, hyperglycemia may be induced during periods of acute metabolic stress or traumatic injury, develop as a result of surgery, or arise as an adverse effect of treatment with certain medications.[2,9–11] Several investigators have reported newly occurring hyperglycemia in association with acute hospital admissions, and others have shown the progression from normal glucose homeostasis to hyperglycemia during critical illnesses.[12,13] Cely and colleagues[13] found that among medical inpatients, even those with normal baseline glucose levels had hyperglycemia for at least a portion of time while in the intensive care unit (ICU), with duration of hyperglycemia increasing by 19% for each 1.0% increase in glycated hemoglobin A1C (A1C) (Figure 1).[13] For many individuals, therefore, plasma glucose is an important consideration during in-hospital medical care.

Thus, identifying hyperglycemia in new critically ill inpatients is a medical priority necessary to ensure optimal care and improved outcomes. Table 1 lists parameters for normoglycemia and glycemic targets for most patients.[7,14,15] The American Diabetes Association (ADA) also specifies that in critically ill patients, blood glucose values should be kept as close to 110 mg/dL as possible and generally under 140 mg/dL.[7] Stress hyperglycemia and diabetes have been associated with a significantly higher mortality rate, increased length of hospital stay, poor cardiovascular outcomes, and greater need for ICU admission compared with normoglycemia.[1,3,4] In fact, Umpierrez and colleagues[1] reported an in-hospital mortality rate 3 times higher in ICU patients with new hyperglycemia (no prior history of diabetes and admission or in-hospital fasting glucose level > 126 mg/dL or random blood glucose level > 200 mg/dL on ≥ 2 determinations) than in those with normoglycemia (31% vs 10%, P < .01), and 3 times higher than in patients with known diabetes (31% vs 11%, P < .01). Furthermore, this increase in mortality was also evident in non-ICU patients and in both groups combined (Figure 2).

Even small elevations in glucose can substantially increase morbidity and mortality. Krinsley[5] noted that mortality among 1826 critically ill ICU patients increased in parallel with advancing glucose levels but was elevated even at the relatively modest mean glucose levels of 100–119 mg/dL. As a result, it would appear that treatment to reduce glucose might improve the prognosis for all critically ill hospitalized inpatients.

Several investigators have established that controlling glucose with intensive insulin therapy (IIT) during hospitalization reduces morbidity and mortality among critically ill individuals. Working primarily with patients undergoing cardiac surgery, researchers have found that perioperative intensive insulin infusion (to maintain blood glucose at predefined levels ranging variably between 80 mg/dL and 150 mg/dL in different studies) significantly improves both short-term (in-hospital) morbidity and mortality[16–18] and long-term (postdischarge) patient outcomes.[19] Furthermore, many of these patients with hyperglycemia and diabetes will require ongoing monitoring and continued glucose-lowering therapy after discharge, once they have returned to the care of a primary care physician (PCP). Thus, PCPs increasingly need to understand the metabolic changes associated with stress hyperglycemia, to become familiar with the in-hospital regimens used to reduce elevated blood glucose levels, and to learn how to help with the transition from inpatient insulin therapy to long-term glycemic care. This article provides background establishing the correlation between hyperglycemia and morbidity and mortality and reviews how hyperglycemia is managed in the hospital and how this affects the responsibility of PCPs as long-term care providers.

Evidence in Support of Intensive Insulin Control for Critically Ill Inpatients

On the basis of the hypothesis that elevated glucose leads to poor outcomes in critically ill inpatients, 2 landmark studies carried out by Van den Berghe and colleagues[16,20] in Leuven, Belgium, assessed the benefit of normalizing blood glucose levels between 80 mg/dL and 110 mg/dL (IIT) compared with conventional management targeting between 180 mg/dL and 200 mg/dL. In the first study, a total of 1548 patients admitted to the surgical ICU were randomly assigned to either IIT (n = 765) or conventional insulin treatment (n = 783); 13% of the total patients had a documented history of diabetes. Insulin infusion was administered if the blood glucose level exceeded 110 mg/dL (IIT, n = 755) or 215 mg/dL (conventional treatment, n = 307). IIT reduced the unbiased mortality rate during intensive care by 32% (P < .04) compared with conventional therapy, with particular efficacy noted among patients who required more than a 5-day ICU stay (P = .005), and reduced overall in-hospital mortality by 34% (P < .01) (Figure 3).[16]

Figure 3.

These Kaplan-Meier curves show that patients receiving intensive insulin therapy to maintain blood glucose between 80 mg/dL and 110 mg/dL had significantly improved survival both (A) while in the intensive care unit (ICU) and (B) throughout the hospital stay compared with patients on conventional therapy.

Patients discharged alive from the ICU (Panel A) and from the hospital (Panel B) were considered to have survived. In both cases, the differences between the treatment groups were significant (survival in ICU, nominal P = .005 and adjusted P < .04; in-hospital survival, nominal P = .01). P values were determined with the use of the Mantel-Cox log-rank test.

Reprinted with permission from Van den Berghe et al. N Engl J Med. 2001;345:1359–1367.[16] Copyright © 2001 Massachusetts Medical Society. All rights reserved.

Furthermore, the risk for complications, such as bloodstream infections, acute renal failure, and critical illness polyneuropathy, each fell by 41% to 46% with the IIT intervention, limiting the need for intensive care resources.

In the second study, which attempted to expand these findings to a medical ICU population, the investigators followed the outcomes of IIT intervention among 1200 adult patients expected to need at least 3 days of ICU care. In this setting, IIT significantly reduced a number of morbidity measures, including new kidney injury, period of mechanical ventilation needed, and ICU and hospital duration of stay, compared with conventional treatment (P < .01).[20] Although overall mortality was not affected by treatment approach, the death rate fell from 52.5% to 43% among IIT-treated patients who stayed in the ICU for ≥ 3 days (P = .009).[20]

These findings have been supported by several well-designed trials of specific insulin delivery regimens in hospitalized patients. A group in Oregon reported long-term follow-up on cardiac surgical patients with diabetes receiving a unique automated continuous insulin infusion (CII) method known as the Portland Protocol, which titrates insulin delivery to achieve tight glucose control.[17,18,21] This protocol originally targeted blood glucose values of 150–200 mg/dL[21] but has been updated twice so that it now aims for concentrations of 100–150 mg/dL.[17,18] Data from patients who were treated according to this protocol were compared with historical data from patients who were treated with individualized sliding-scale insulin (SSI)-guided subcutaneous (SC) injections administered every 4 hours. A total of 4864 patients with diabetes who had undergone open heart surgery were evaluated between January 1987 and September 2003 under this protocol. In separate multivariate analyses, CII consistently improved a number of hyperglycemia-related endpoints, including risk for death (reduced by 57%) and rate of postsurgical deep sternal wound infection (reduced by 66%; P < .0001 for both).[18]

Similarly, the well-known Diabetes Mellitus Insulin-Glucose Infusion in Acute Myocardial Infarction (DIGAMI) study showed that an insulin-glucose infusion followed by multidose SC insulin administration for at least 3 months after hospital discharge, both titrated to achieve normoglycemia, produced long-term (1 and 3.4 years) improvement compared with conventional therapy in patients with diabetes treated within 24 hours of acute myocardial infarction (AMI).[22,23] Achieving tight metabolic control in the immediate post-AMI period reduced relative mortality by 29% (P = .027) at 1 year, a benefit that was maintained at the 3.4-year follow-up (P = .011). In a follow-up study (DIGAMI 2), the insulin-glucose infusion followed by aggressive SC therapy did not improve survival or reduce the number of nonfatal myocardial reinfarctions or stroke compared with infusion followed by standard glucose therapy.[24] However, target fasting glucose levels were not reached, leaving the question open of whether more aggressive dosing to achieve tight glycemic control might have resulted in more positive findings.

A meta-analysis of 35 randomized controlled trials concluded that rapid insulin treatment in critically ill patients improves short-term mortality.[25] Pittas and coworkers[25] reported a 15% relative risk reduction for death among insulin- vs non-insulin-treated patients and a 29% reduction with insulin treatment to goal compared with insulin therapy without a target glucose level. The benefit was observed in patients with diabetes (regardless of prior insulin therapy), in both medical and surgical inpatients.

The overall conclusion from these reports appears to be that treating hyperglycemia in post-AMI patients in the medical or surgical ICU reduces mortality risk and improves outcomes. This finding supports the need for aggressive treatment of critically ill hospital inpatients to reduce glucose levels, prevent complications, and reduce overall mortality.

Current Concerns Around Intensive Glycemic Control in Patients With Cardiovascular Disease

Reports from 2 recent studies of over 10,000 nonhospitalized patients examined the effects of intensive blood glucose control on the risk for cardiovascular events, such as heart attack, stroke, or death from cardiovascular disease, in patients with type 2 diabetes.[26,27] The Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial studied the effects of intensive glycemic control (defined as striving to achieve a target A1C < 6.0%) and standard glycemic control (target A1C between 7.0% and 7.9%) in middle-aged or older people with type 2 diabetes at high risk of having a cardiovascular disease event. Over 3.5 years of treatment, more patients died in the intensive treatment arm compared with those in the standard treatment arm, and consequently, the intensive glucose control arm was discontinued.[26]

In contrast with the ACCORD trial, data from a similar study of high-risk patients with type 2 diabetes worldwide – the Action in Diabetes and Vascular Disease: PreterAx and DiamicroN-MR Controlled Evaluation (ADVANCE)[27,28] – show that the treatment strategy of intensively lowering blood glucose did not lead to greater morbidity. In the ADVANCE trial, the intensive blood glucose lowering program aimed to reduce levels of A1C to ≤ 6.5%, which was not as aggressive as the target of 6.0% in ACCORD.

Generally, patients at high risk for cardiovascular events are not treated to glucose levels as low as those targeted in these trials. Both of these studies targeted blood glucose levels below those recommended in current treatment guidelines (< 7.0% in most patients, less stringent in patients with comorbid conditions).[7] Because these trials targeted physiologic blood glucose levels in the intensive groups, episodes of hypoglycemia were expected and strategies for prevention and treatment were put in place a priori.[29] Moreover, a considerable number of patients in both trials used a statin, an antihypertensive drug, and an angiotensin-converting enzyme inhibitor; therefore, the cardiovascular benefit of aggressive glycemic control may be overshadowed by the benefits of these drugs. The results from these trials have not undermined the current guidelines for treating patients with type 2 diabetes mellitus, and, in general practice, it may be wise to target the best glycemic control that each individual patient can reach safely. Furthermore, the generalizability of these data to hospitalized patients, especially those in the ICU, is unknown.

Standards of Care

Given the clear connection between hyperglycemia and morbidity and mortality in critically ill hospitalized patients, several organizations, including the Joint Commission (formerly the Joint Commission on Accreditation of Healthcare Organizations, or JCAHO) and the ADA, have adapted their clinical practice guidelines to include inpatient glucose screening and treatment as a standard of care.[7,30,31] In fact, the Joint Commission has developed a professional and patient education initiative and performance measurement standard on the basis of the ADA guidelines that is designed to ensure that all hospitalized patients with diabetes are managed to achieve the best possible outcome.[30]

The Joint Commission/ADA protocol recommends a range of standard strategies necessary to effectively manage hyperglycemia in hospitalized patients with acute illness. On admission, all patients with diabetes should have their blood glucose monitored; diabetes should be recorded in the patient record; and periodic A1C monitoring should be performed.[7,30] Results should be shared with members of a multidisciplinary healthcare team, if available, to ensure that all interventions assigned are glucose-conscious. Although independently not a diagnostic test, obtaining an A1C upon admission provides a measure of the patient's blood glucose level in the 60–90 days prior to hospitalization. In the event of elevated glucose levels, a treatment plan is devised by the multidisciplinary group: The plan should define basal, prandial, and any necessary supplementary/correctional insulin therapies as well as standards of care for managing hypoglycemia if it occurs. Outcomes should adhere to well-defined glucose goals (Table 1). Furthermore, an A1C measure should be used for discharge planning, and prior to release, the team should provide the patient with a hyperglycemia management session to promote techniques for self-management at home.

Table 1.

Normoglycemia and Recommended Glycemic Targets

Glycemic Control	Healthy[1,6]	ADA[7]	AACE[1]	EDPG[6]
A1C, %	< 6.0	< 7.0*	≤ 6.5*	≤ 6.5*
Preprandial glucose, mg/dL
Plasma	< 110	70–130*	≤ 110*	< 110*
Blood	< 100	(62–116)†	(< 100)	< 100*
Peak postprandial glucose, mg/dL
Plasma	< 140	< 180*	(≤ 157)	(< 151)
Blood	< 120	(< 161)	≤ 140*	< 135*

^*Target reported in guideline (converted values).

^†Blood glucose should be kept as close to 110 mg/dL as possible and generally < 140 mg/dL in critically ill patients.

A1C = glycated hemoglobin; AACE = American Association of Clinical Endocrinologists; ADA = American Diabetes Association; EDPG = European Diabetes Policy Group

The Joint Commission/ADA program became effective on July 1, 2007, requiring all certified institutions to uphold defined performance standards. The program is designed to maintain quality and integrity via periodic evaluation of program processes and patient outcomes, minimum staff education requirements, standardized blood glucose monitoring protocols, and patient education programs on home self-care. It is hoped that by promoting these practice pathways, the program will ensure that hyperglycemia becomes an inpatient practice priority and that all patients will receive the best care possible to reduce hyperglycemia-related risks.[30]

Insulin is the preferred therapy for hyperglycemia in the hospital setting.[7] Each of the major classes of oral antidiabetic agents has significant limitations for inpatient use and, therefore, is frequently discontinued upon hospital admission.[7,32] Briefly, metformin has a propensity to result in lactic acidosis; sulfonylureas are associated with hypoglycemia in patients not consuming their normal nutrition; and thiazolidinediones exacerbate fluid retention increasing intravascular volume, a particular concern in patients with congestive heart failure.[7,32] Moreover, these agents provide little flexibility or opportunity for titration in a setting that necessitates acute changes; therefore, patients should be converted to SC or intravenous (IV) insulin during hospitalization.[7,32]

Insulin Management

CII in Acute and/or Perioperative Care

CII is the only method of insulin therapy developed specifically for the hospital population and is indicated for general preoperative, intraoperative, and postoperative care in patients with myocardial infarction or cardiogenic shock, and for those with critical care illness.[9] It also provides a good transition therapy during the dose-finding period leading to SC insulin in patients with type 1 or 2 diabetes. In the acute setting with the immediate need for tight glucose control, where insulin requirements fluctuate rapidly or total parenteral nutrition is the norm, the IV route has an advantage over SC delivery. IV therapy offers a rapid onset of action, quick dose adjustment to gain tight glycemic control, and flexibility in the event of hypoglycemia. IV administration also allows for immediate discontinuation and rapid drop in plasma insulin at the first sign of excessive glucose reduction.

Several validated protocols for CII, including the previously mentioned Portland Protocol and the Yale protocol, are processes used to determine the infusion rate on the basis of change in glucose over time rather than on acute glucose measures. They are specifically designed to be implemented by nursing staff and are used to provide effective and standardized infusion regimens that have been shown to reliably achieve tight glycemic control.[17,18,21,33] The most successful insulin infusion protocols share a number of important features that contribute to their acceptability and success. First, they generally use dynamic criteria that respond to individual insulin responses and change in insulin needs over the course of treatment to ensure a stable plasma glucose level.[7,34] For instance, the Yale protocol balances the rate of glucose change against the current blood glucose level and insulin infusion rate to determine highly individualized dose adjustments when needed.[33] The protocols generally are designed by multidisciplinary groups, can be administered by nurses or other support personnel, use monitoring methods approved for use in most hospitals, and can be applied in all institutional departments – from emergency departments to operating rooms and medical and surgical ICUs.[33–36] Furthermore, the best protocols aim for infusion schedules that are as stable as possible while allowing for immediate response in the event of blood glucose imbalance, encourage regular blood glucose monitoring (eg, hourly) to identify hypoglycemia or hyperglycemia early, and describe validated and effective protocols for managing treatment-induced hypoglycemia when needed. Table 2 summarizes safety and efficacy data from several insulin infusion protocols. The particular protocol employed is not important provided that it meets these criteria.

Table 2.

Safety and Efficacy of Published IV Insulin Infusion Protocols[18,33,35,68] (This table reports key, established protocols and should not be considered comprehensive, given the large number of different protocols in clinical use.)

Protocol	Patient Population Studied	Target BG	Efficacy in BG Management	Incidence of Hypoglycemia
Goldberg[33] (Yale Protocol)	Medical ICU (N = 52)	100–139 mg/dL	Target reached in median 9 hrs; 93% of BG values, 80–199 mg/dL	0.3% of BG values < 60 mg/dL, no adverse events
Krinsley[35]	Medical-surgical ICU (N = 800)	< 140 mg/dL	Mean BG ↓ from 152.3 mg/dL before to 130.7 mg/dL after protocol instituted	0.34% of patients with BG < 40 mg/dL; 1.02% with BG 40–59 mg/dL
Furnary[18] (Portland Protocol)	Cardiac surgical patients (N = 4864)	100–150 mg/dL	94% of patients in target range within 3 hrs	0.5% of BG values < 60 mg/dL; 0.04% symptomatic hypoglycemia
Zimmerman[68]	Cardiac surgical patients (N = 168)	80–150 mg/dL	BG < 150 mg/dL reached in median 2.1 hrs; 61% BG values within target range	16.7% of BG values < 65 mg/dL; nonsignificant change vs before protocol

BG = blood glucose; ICU = intensive care unit

Transition to SC Insulin for Extended In-Hospital Care and Preparation for Discharge

As a patient's clinical status begins to improve, it is important that glucose is followed especially closely to see how hyperglycemia is altered under conditions of improving health. Many patients will continue to require insulin to maintain normoglycemia, but with stabilization of the acute crisis – particularly at the point when volume resuscitation and/or pressor support are no longer necessary and the patient begins to eat solid foods – the patient often can be transitioned to SC bolus insulin injections.[34,37] SC dosing, particularly among feeding patients, is based on discrete requirements for basal insulin (exogenous insulin needed to manage gluconeogenesis and ketogenesis not met by endogenous insulin production) and prandial insulin (insulin requirement to cover nourishment, whether the patient is receiving IV dextrose, total parenteral nutrition, enteral feeding, or meals by mouth), and often includes supplemental or correction dosing to respond to unexpected hyperglycemic episodes.[7,9,34,38] SC insulin therapy may comprise regular crystalline insulin or short-, intermediate-, or long-acting analogs, depending on clinical status or physician preference.

Although formal research has not established an optimal strategy for transitioning from IV to SC insulin, there are several standards described by the ADA and others that are considered benchmarks for best care (Table 3). First, adequate glycemic coverage must be sustained while terminating IV infusion and initiating SC delivery, and thus most practitioners will administer the first SC dose of an intermediate- or long-acting insulin at least 2–3 hours prior to discontinuation of the IV formulation.[7,34] A long-acting, peakless analog may be the best means to ensure a safe yet effective insulin level during the overlap period, minimizing the risk for hypoglycemia that may occur with insulins that release in surges.[34] The first SC dose is estimated at about 80% of the projected daily requirement, whereas subsequent evening doses can be adjusted on the basis of patient response to the initial injection. In addition, ongoing monitoring is the key to ensuring that glucose stabilization is achieved and correction doses of insulin (generally, a rapid-acting analog) are administered as needed. Table 4 provides a list of SC insulins by their time-action profile. Prandial dosing also can be adjusted in response to physiologic changes, such as the decrease in stress-induced hyperglycemia and the increase in appetite associated with improved health. Prandial requirements generally rise with increasing food intake, until total daily doses of basal and prandial insulin are about equal.[34]

Table 3.

Sample Recommendations^* for Transition From IV to SC Insulin Therapy

Administer first SC dose (basal insulin) at least 2 hrs prior to terminating IV infusion TDD for first SC administration is 80% of estimated 24-hr insulin requirement Subsequent doses over next few days are adjusted according to patient response and improving health until ○ Basal dose is ∼50% of SC TDD ○ Bolus total dose is ∼50% of SC TDD Correction dose is determined as ○ Actual BG minus target BG divided by the CF (CF is 1700 divided by TDD)

^*Unvalidated, no clinical trials have tested any transition strategies.

BG = blood glucose; CF = correction factor, representing the BG reduction associated with 1 U of rapid-acting insulin; IV = intravenous; SC = subcutaneous; TDD = total daily dose

Adapted with permission from Bode BW, Braithwaite SS, Steed RD, Davidson PC. Endocr Pract. 2004;10(suppl 2):71–80.[34]

Table 4.

Time-Action Profiles of Subcutaneous Insulins

Insulin Type	Onset	Peak (hrs)	Duration of Action (hrs)
Rapid-acting
Lispro, aspart, glulisine	5–15 min	0.5–1.5	2–4
Short-acting
Regular human	30–60 min	2–3	3–6
Intermediate-acting
Human NPH	2–4 hrs	4–10	10–16
Long-acting (basal)
Insulin glargine	1–2 hrs	No pronounced peak	≈24
Insulin detemir	1–2 hrs	Less pronounced peak (≈6)	18–24

Reproduced with permission from Edelman S, Dailey G, Flood T, Kuritzky L, Renda S. Oestopath Med Primary Care. 2007;1:9. doi:10.1186/1750-4732-1-9.[69]

Correction insulin administration is an art in itself. Correction-dose prandial insulin therapy is used in conjunction with scheduled basal/prandial insulin either to determine the optimal dose of insulin or to supplement scheduled insulin when rapid changes are needed. Bode and colleagues[34] recommend that when switching from IV to SC insulin, correction doses with rapid-acting SC insulin analogs should be administered according to food intake: Patients who are able to eat discrete meals should be dosed before mealtimes, at bedtime, and at 3:00 am, whereas those on total parenteral nutrition or nothing by mouth should be dosed every 4–6 hours. However, the ADA offers a caveat, noting that if correction dosing becomes frequent, the scheduled doses for the next day should be increased to address the apparently increasing insulin needs.[7,34]

The SSI regimen is no longer endorsed by the ADA because it is considered a static method that assigns insulin dose on the basis of hospital admission glucose concentration (and is rarely modified or adapted), and therefore provides poor long-term glycemic control. Furthermore, the episodic, reactive nature of SSI and the lack of intermediate- or long-acting basal insulin in these protocols, combined with their failure to consider the patient's response to previous regimens, the timing of food intake, the patient's changing insulin needs, or the patient's sensitivity to insulin, result in a high risk for rapid and frequent changes in blood glucose thus predisposing patients to hyperglycemic or hypoglycemic exacerbations.[7,9,39,40] In fact, there has been no association found between the use of single-insulin, sliding-scale regimens in hospitalized patients and improved clinical outcome.[40,41] Moreover, a recent study[42] confirmed that SSI is inferior to a basal-prandial regimen in achieving glycemic control. Patients on a basal-prandial insulin regimen were more likely to reach target blood glucose levels than those on SSI (66% vs 38%), with an overall difference in mean blood glucose of 27 mg/dL between groups (P < .01). Although the mean daily dose of insulin was significantly higher in the basal-prandial regimen, the incidence of hypoglycemia remained low (only 0.4% of glucose readings were < 60 mg/dL).

Discharge Planning

Survival Skills for Patient Self-care

A rational and well-instituted discharge plan is essential to maintaining “continuity of effect,” ie, retaining the benefit of in-hospital insulin therapy while establishing a regimen that can be adapted to the patient's self-care. Generally, the plan is devised as a multidisciplinary collaboration, relying on the efforts of physicians, nurses, certified diabetes educators (CDEs), and other healthcare professionals involved in their care (eg, dietitian, pharmacist, psychologist, ophthalmologist, podiatrist, cardiologist, or other medical specialist).[7,12] If a multidisciplinary team is not available, diabetes education can be provided by a physician or nurse. The multidisciplinary approach is particularly important because the in-hospital specialist will likely transfer the care of the patient to the practitioner in the primary care setting, necessitating a readily implemented and practical plan with a high likelihood of success. Furthermore, chronic diseases, such as diabetes, require that patients gain the skills and understanding necessary to address the medical needs of their condition and to be involved in a successful partnership with their healthcare providers to ensure thoughtful and continuous disease management.[43] A chronic care model developed by Rothman and Wagner[43,44] promotes active participation by the patient and family but acknowledges that productive cooperation with medical providers is key to ensuring optimization of therapy, proper support for self-care challenges, and the medical follow-up needed for successful glucose control.

The ultimate goal of discharge planning is to provide the patient with the survival skills needed to manage and take responsibility for his or her own health. Diabetes self-management education is a key component of the transition plan. Optimal diabetes care requires that the patient have some understanding of the pathophysiology of the disease, is comfortable with glucose self-monitoring, and has received instruction on proper insulin administration techniques.[12] It is also important that the patient understands the risk for hypoglycemia and learns how to recognize it, how to prevent it, how to treat it, and when it is imperative to call a clinician. In addition, patients will be responsible for proper meal planning outside the hospital and should be instructed on acceptable food choices until they meet with a CDE-trained dietitian. Without proper education and good follow-up planning, as many as 30% of patients with diabetes are at risk for rehospitalization within 30 days of discharge.[45]

Barriers to Insulin Use Must Be Addressed

One of the challenges to achieving tight glycemic control with insulin therapy is related to patient concerns with regard to injectable insulin.[46–49] A number of issues are consistently cited as reasons that patients are reluctant to commit to an insulin regimen, and without addressing these concerns, it is unlikely that the postdischarge patient will adapt to a self-management regimen. An informal survey among diabetes educators revealed the following common barriers to insulin use:

• Patient fear;

• Patients identify needles and injections with pain;

• Fear of diabetic complications;

• Weight gain;

• Inconvenience;

• Complex and time-consuming regimen; and

• Cost.

Misperceptions have even been promoted in the medical literature. For instance, Peyrot and colleagues[48] and Ratzmann[50] suggested that patients consider the switch to insulin a sign of a medical “crisis” in their disease and evidence that they personally “failed” previous therapies. Other misperceptions creating barriers to insulin use relate to fear of injections and concern about the complex regimen and some of its purported outcomes, such as weight gain and hypoglycemia.[46,48] Some patients are overwhelmed at the thought of determining their own insulin doses or handling syringes, and many patients resist the idea of a daily injection. However, the availability of inhaled insulin powder has not been embraced, to the extent that one of the most common inhaled insulins (Exubera) was taken off the market as of October 18, 2007. Reluctance to accept insulin therapy until all other agents and interventions have failed can mean regular bouts of hyperglycemia and the associated increased risk for morbidity and mortality. Moreover, patients must contend with these concerns in addition to managing the primary condition that resulted in their hospitalization.

Given the myriad of potential barriers and the risks associated with not properly maintaining glycemic control, discharge planning (and postdischarge care) must dispel these myths and offer a positive message with regard to insulin therapy in order to demystify its use. If patients understand the progressive nature of diabetes and the role of insulin as a sensible method for replacing a hormone that the body is no longer making efficiently on its own, then they may feel more comfortable with the physiology of the disease and their role in managing it.[46] Furthermore, offering reassurance to the patient that insulin injection with today's ultrafine needles (31-gauge) can be much less painful and remarkably simple, in addition to providing alternatives (eg, insulin pump), can reduce anxiety and remove some of the psychological resistance.

CDEs play an important role in conveying these messages and promoting active communication. Siminerio and colleagues[51] found that diabetes outcomes in a rural setting improved when a chronic care model supported by a CDE was initiated. Decision support was implemented on the basis of the ADA Standards of Care; patients received diabetes self-management education taught by an on-site CDE; and the information delivery system was redesigned so that a CDE could meet and speak directly with patients with diabetes. With these changes, healthcare provider adherence to practice guidelines improved; provider knowledge and sense of empowerment increased; and A1C results in the patient population improved.[51] By assisting patients in understanding their medications, learning glucose self-monitoring techniques, and overseeing individualized self-management plans, support personnel were able to significantly improve disease management and enhance medical and behavioral outcomes. In this study, the CDE worked 75 miles away but traveled to the rural practice twice a month. A similar arrangement can be adapted to most rural settings by referring to a national registry of CDEs, such as the one provided by the American Association of Diabetes Educators, which can be accessed at www.diabeteseducator.org.[52]

Maintaining Glycemic Control After Discharge

Rapid Transition to Primary Care

The need for attention to glycemic control does not stop at the hospital door, and attention to glycemic care is an ongoing process, with care passed on to an outside provider after discharge. Norhammar and coworkers[53] found that less than 35% of 181 consecutive patients hospitalized for AMI had normal glucose homeostasis 3 months after hospital discharge, highlighting the high risk for sustained hyperglycemia among patients with hyperglycemia first noted during an acute event. These patients must be aggressively treated and followed after discharge to ensure that glycemic control is maintained.

Postdischarge patients with diabetes should see their PCPs within 7–30 days of leaving the hospital or at 1 month post discharge if he or she had no diagnosis of diabetes prior to hospitalization and illness/stress was the suspected source of hyperglycemia.[9,12,37] The PCP should immediately document the current baseline outpatient blood glucose measure, review the patient's treatment history, and assess the state of glycemic control with the discharge medications. Glycemic control is central to diabetes management, and every clinician involved in the care of patients with diabetes should be familiar with the A1C and blood glucose goals established by the ADA, the American Association of Clinical Endocrinologists, and the European Diabetes Policy Group (Table 1).[7,15,54] If these goals are not maintained, the PCP must adjust the treatment schedules and consider appropriate referrals (eg, endocrinologist).

Oral Antidiabetic Agents

Not all patients will require insulin therapy after discharge. Patients who had been taking oral antidiabetic drugs before hospitalization may warrant a transition back to the prehospitalization regimen, with appropriate monitoring and follow-up. Similarly, patients who do not have documented prehospitalization diabetes but who are dependent on low insulin doses to maintain glycemic control may be switched to a test period of oral therapy.[12,55] These oral agents can be restarted in the postoperative period when oral intake of medications is possible and hepatic and renal function are stable.[55] However, patients on oral agents must be followed closely after discharge. If their A1C exceeds 7.0% after the switch, supplementation with a basal insulin may be warranted.[56]

Progression of Treatment

Many patients will still need to be controlled with SC insulin therapy upon release from the hospital, either in addition to oral agents or alone. Riddle and colleagues[57] tested the benefit of adding basal insulin to an oral regimen to gain better glucose control. The Treat-to-Target Trial was intended to establish a rationale for adding basal insulin to the regimen of patients inadequately controlled on 1 or 2 oral agents as a safe and effective means to reach the 7.0% A1C target.[57] The investigators reported that systematic titration of basal insulin helped 60% of patients achieve A1C ≤ 7.0% and lowered fasting glucose levels, highlighting the value of add-on insulin therapy in targeting tight glycemic control among patients no longer optimally responding to oral agents.[57]

Insulin Options

Ambulatory, feeding patients will generally rely on a multiple-dose regimen, comprising a once-daily basal insulin injection plus mealtime treatment with a rapid-acting insulin analog for prandial control.[9] The historical reliance on intermediate-acting insulins for basal therapy is being replaced by long-acting analogs, such as once-daily insulin glargine or once- or twice-daily insulin detemir, both of which have been shown to reduce the risk for hypoglycemia and to better control fasting plasma glucose and A1C compared with NPH insulin.[9,58–60] The rapid-acting insulin analogs – insulin aspart, insulin glulisine, and insulin lispro – are effective prandial therapies and provide more discrete prandial control than regular human insulin due to the latter's long duration of action.[9] The analogs were formulated to rapidly dissociate into monomers in tissue and therefore reach twice the maximum concentration in half the time of regular human insulin.[61,62] As a result, they provide a relatively rapid peak insulin action and early elimination, reducing the risk for postabsorptive hypoglycemia. These agents are usually administered immediately prior to a meal (5–15 minutes prior); however, insulin glulisine can be given up to 20 minutes after starting a meal,[63] which may provide a benefit because the dose may be modified on the basis of actual meal consumption.

The concept of combined basal and prandial insulin is based on an effort to provide supplementation that closely mimics physiologic insulin release from the pancreas. Prandial insulin parallels the endogenous insulin surge in response to food intake, with basal insulin providing a small but constant plasma concentration of insulin to yield baseline coverage and simulate the late-phase prandial response that corrects for overnight and between-meal hepatic glucose output (Figure 4).[61,64]

Figure 4.

Treatment with a basal/prandial analog insulin regimen closely replicates endogenous insulin secretion patterns. Adapted with permission from Leahy JL, Cefalu WT, eds. Insulin Therapy. New York, NY: Marcel Dekker, Inc.; 2002:87–112.[64]

Many patients also will require correction-dose insulin to neutralize between-meal hyperglycemic episodes if they occur. According to Hirsch,[61] regular and NPH insulin, because of their longer duration of action, yield both prandial and basal coverage compared with the analogs; however, rapid-acting and long-acting insulin analogs are designed to target each component separately, allowing for more targeted glucose control.

Although a basal-bolus regimen may be viewed as complicated and challenging to some patients, the alternative of premixed insulins (those with fixed doses of basal and prandial insulin in 1 injection) has several limitations.[65,66] The fixed insulin components do not closely adhere with physiologic plasma insulin profiles, and the formulations provide limited flexibility to address shortcomings in the plasma levels achieved. In patients with severely diminished endogenous insulin, the insulin delivery patterns of premixed formulations may induce hypoglycemia at certain times of the day and/or may deliver plasma concentrations inadequate to meet the needs of the midday meal.[61] Furthermore, it is not possible to adjust the individual components in response to results of self-monitored glucose assessments. In contrast, a basal-bolus regimen affords flexibility, allowing for dose adjustments in response to changes in eating patterns or variations in blood glucose levels.

Conclusions

Diabetes and hyperglycemia are significant risks to the health and prognosis of hospitalized critically ill patients. The last decade has seen an increased emphasis on diagnosis and therapy for maintaining glucose control among critically ill patients, with multiple national and international organizations promoting hyperglycemia management with standards of care and standardized treatment pathways. Protocols for CII have been uniquely created to address the need for tight glucose control in the hospital and have proven to be successful at maintaining desirable glucose levels and reducing risks associated with hyperglycemia.

The transition to an SC insulin regimen heralds the improving health and likely discharge of a patient from the hospital. As they are readied for hospital release, discharge planning should prepare patients for self-monitoring and self-care at home and give them the survival skills necessary to maintain glycemic control. A treatment plan devised by a multidisciplinary team is the best means to ensure that patients receive a practical and successful treatment regimen that can be readily overseen by themselves, their families, and their postdischarge medical team.

The role of the PCP in postdischarge care demands that these professionals understand the treatment history of the patients and the options for care moving forward. Maintaining glycemic control in postdischarge patients may improve patient outcomes.[24,54,67] Many patients may require continued treatment with an appropriate basal-prandial regimen, although oral agents may be a rational next move for some patients. This may necessitate greater resource demands from the PCP; however, the use of efficient record keeping, such as through the use of electronic medical records, may help ameliorate any additional burden, and ultimately the patient should benefit. The properly prepared PCP can deliver optimal diabetes care and thereby improve both short-term and long-term outcomes for those patients with critical illnesses and hyperglycemia or diabetes.

Figure 1.

Most patients in the MICU spent time with hyperglycemia, even those with good glucose control at baseline; percentage of time (median and quartile bounds) above different glucose thresholds (within the first 120 hours of admission) are shown here. Reprinted with permission from Cely et al. Chest. 2004;126(3):879–887.[13]

A1C = glycated hemoglobin A1C; MICU = medical intensive care unit.

Figure 2.

Patients with newly diagnosed hyperglycemia had significantly higher rates of in-hospital mortality than patients with normoglycemia or known diabetes, a pattern that occurred in both intensive care unit (ICU) and non-ICU patients. Reprinted with permission from Umpierrez et al. J Clin Endocrinol Metab. 2002;87(3):978–982.[1] Copyright 2002, The Endocrine Society.

*P < .01 vs both normoglycemia and known diabetes groups.