In Brief This article reviews the use of subcutaneous insulin for hospitalized patients. Topics include the rationale for using insulin;scheduled insulin therapy to cover basal and nutritional needs; correction therapy; dose determination; establishment of timing of insulin action appropriate to the pattern of carbohydrate exposure; education of caregivers;and the design of hospital systems that will promote quality and help staff to manage complexity.
This review will focus on the rationale and methodology for management of hospital hyperglycemia under conditions that do not require intravenous insulin infusion.
For many years, tradition held that hospitalization was not the time in the life of a patient with diabetes to focus on glycemic control. The complications of diabetes were divided into 1) acute metabolic emergencies requiring patient stabilization rather than perfection of glycemic control and 2) chronic tissue complications, which had required months or years to evolve. It was thought that safety considerations dictated the prudence of avoiding hospital hypoglycemia.
Nevertheless, adverse hospital outcomes have been linked to hyperglycemia. These include nosocomial infection, sepsis, cardiac mortality after myocardial infarction or cardiac surgery, atrial fibrillation after cardiac surgery,increased infarct size after stroke, acute graft rejection after transplantation, “unit” neuropathy, transfusion requirement and acute renal failure in critical care unit patients, prolongation of stay, and increased cost.1–21 Some of these outcomes, being multifactorial in etiology and not readily classifiable as a manifestation of microvascular or macro-vascular disease,were not linked intuitively to specific diabetes treatment actions. In fact,it took randomized, controlled trials to establish the relationship.
The revolution in present-day thinking about the importance of inpatient glycemic control has occurred because of the findings of a large and long-running cardiac surgery series3,5,14 and because of the convincing evidence of randomized, controlled trials demonstrating, first, the causal nature of the relationship between hyperglycemia and some of the outcomes listed above, and, second, the ability of intensive insulin management to reduce those adverse outcomes.22–27 (Readers are referred to the article by Goldberg and Inzucchi in this issue of Diabetes Spectrum, .)
The strongest evidence for hospital outcomes advantages was obtained,without significant complications from hypoglycemia, during and after the use of intravenous insulin infusion under conditions of intensive monitoring. For general medical conditions treated in hospital wards other than intensive care units, similar outcomes data do not exist. It will be necessary to show, from randomized trials prospectively conducted outside of the intensive care unit setting, that specific benefits will result from intensive subcutaneous insulin management, without prohibitive risk from hypoglycemia. It is necessary to define blood glucose (BG) targets and to determine whether there is any glycemic threshold, other than the upper limit of normal, above which those benefits are jeopardized. Pending the results of such research, upper limits of glycemic targets intended to provide clinicians with guidelines for promoting improved outcomes have been recommended(Table 1).28, 29
Mechanisms of Hospital Hyperglycemia and Protection by Glycemic Control or Insulin
Conditions of hospitalization, such as trauma, hemorrhage, burns, hypoxia,infections, sepsis, or shock, may induce both insulin resistance and relative insulin secretory defect. Both peripheral and hepatic sites of insulin resistance have been implicated. The consequences include increased muscle protein degradation, decreased peripheral glucose disposal, lipolysis resulting in increased circulating nonesterified fatty acids, hyperlactatemia,and increased hepatic glucose output. Mechanisms of the observed defects of insulin signaling and relative β-cell suppression during acute illness are the subject of active research.30
The targets for protective action of insulin in the hospital setting prominently include the heart, the endothelium (vasodilatory action,protection against vessel wall inflammatory processes), the inflammatory pathway, the coagulation pathway, host defenses against infection, and the nutritional status of the patient.
Protection may be conferred in part by metabolic control (correction of hyperglycemia, reduction of circulating nonesterified fatty acids, reduction of lactate, reversal of proteolysis, and protection against the downstream nonmetabolic consequences of these abnormalities). Protection also may be conferred by other direct or indirect nonmetabolic effects of insulin itself(nitric oxide–mediated improvement of endothelial function; regulation of nuclear factor-κB; inhibition of the production of harmful reactive oxygen species; regulation of the transcription of proinflammatory genes,adhesion molecules, and chemokines; and suppression of early growth response gene-1, plasminogen activator inhibitor 1, and matrix metalloproteinases).30–33
Deciding on a Treatment Plan
One of the first challenges on admission is to recognize the presence of diabetes or risk for hospital hyperglycemia by patient history and examination and by measurements of BG and potentially hemoglobin A1c(A1C).34 Another challenge is the need for the admitting team to identify type 1 diabetes and record the classification.28
The timing of BG monitoring should be planned with consideration of the pattern of exposure to carbohydrate and antihyperglycemic treatment(Figure 1). The purpose of monitoring is to determine a pattern that might require initiation or revision of daily therapy with antihyperglycemic medication (for example, during pregnancy postprandial testing should be included), to detect hypoglycemia,and to provide BG measurements at the times of day when correction doses might best be used.
a: Subcutaneous insulin therapy may be selected as continuation of ambulatory therapy, as new therapy, or as a transition plan from intravenous infusion of insulin, with choice of insulin and timing dependent on the pattern of carbohydrate exposure. b: Discrete meals. The use of a consistent carbohydrate diet can be ordered in the hospital, facilitating the implementation of basal-prandial-correction therapy with glargine and rapid-acting analog for nonpregnant patients who are eating discrete meals,with BG monitoring before meals, at bedtime, and possibly midsleep. c: Negligible or continuous carbohydrate exposure. When insulin requirements are nearly continuous, as during negligible oral intake or NPO status, or during continuous enteral feedings, intravenous infusion of 5% dextrose containing fluids at rate > 43 cc/hour, or peritoneal dialysis, and frequent doses of split mixed NPH and regular insulin provide easily reversible and almost flat-line coverage. “Hold” orders can be written for low BG or for interruption of carbohydrate exposure. BG monitoring might be at 6:00 am, noon, 6:00 pm, and midnight. d: Transitional meal plan/grazing. For patients tolerating only transitional meal plans or nutritional supplements as during chemotherapy or postoperative recovery, who may be grazing as tolerated between meals, whose exposure to carbohydrate otherwise is negligible, and who require some insulin despite poor overall oral intake, or for patients who may have gastroparesis, the pattern of insulin requirement may be well matched by use of NPH and regular insulin given in divided dosage, approximately two-thirds in the morning and one-third in the evening, with BG monitoring before meals and at bedtime. e: Daytime grazing/overnight enteral feedings. The evening doses of NPH and regular insulin constitute a larger component of therapy than the morning doses. The initial doses of regular insulin seldom exceed 33% of the TDD of insulin. BG monitoring might be before meals and at midnight. f: Corticosteroids. The design of insulin therapy depends on the timing of the corticosteroids and challenges the creativity of the caregiver.
Management during pregnancy or severe hyperglycemia are not shown;please refer to the text.
a: Subcutaneous insulin therapy may be selected as continuation of ambulatory therapy, as new therapy, or as a transition plan from intravenous infusion of insulin, with choice of insulin and timing dependent on the pattern of carbohydrate exposure. b: Discrete meals. The use of a consistent carbohydrate diet can be ordered in the hospital, facilitating the implementation of basal-prandial-correction therapy with glargine and rapid-acting analog for nonpregnant patients who are eating discrete meals,with BG monitoring before meals, at bedtime, and possibly midsleep. c: Negligible or continuous carbohydrate exposure. When insulin requirements are nearly continuous, as during negligible oral intake or NPO status, or during continuous enteral feedings, intravenous infusion of 5% dextrose containing fluids at rate > 43 cc/hour, or peritoneal dialysis, and frequent doses of split mixed NPH and regular insulin provide easily reversible and almost flat-line coverage. “Hold” orders can be written for low BG or for interruption of carbohydrate exposure. BG monitoring might be at 6:00 am, noon, 6:00 pm, and midnight. d: Transitional meal plan/grazing. For patients tolerating only transitional meal plans or nutritional supplements as during chemotherapy or postoperative recovery, who may be grazing as tolerated between meals, whose exposure to carbohydrate otherwise is negligible, and who require some insulin despite poor overall oral intake, or for patients who may have gastroparesis, the pattern of insulin requirement may be well matched by use of NPH and regular insulin given in divided dosage, approximately two-thirds in the morning and one-third in the evening, with BG monitoring before meals and at bedtime. e: Daytime grazing/overnight enteral feedings. The evening doses of NPH and regular insulin constitute a larger component of therapy than the morning doses. The initial doses of regular insulin seldom exceed 33% of the TDD of insulin. BG monitoring might be before meals and at midnight. f: Corticosteroids. The design of insulin therapy depends on the timing of the corticosteroids and challenges the creativity of the caregiver.
Management during pregnancy or severe hyperglycemia are not shown;please refer to the text.
Under conditions of hospitalization, metformin commonly is contraindicated. In general, oral agents do not provide the flexibility necessary to gain control rapidly. Sometimes oral agents are stopped on admission but are reintroduced as a patient approaches discharge.
Patients previously taking insulin temporarily may have a new total daily dose (TDD) requirement. In contrast to the ambulatory pattern of insulin requirement consisting of a fairly continuous basal demand with discrete prandial elevations, the hospital pattern of insulin requirement may reflect continuous delivery or interruption of nutrition. The best course of action for demonstrable hyperglycemia usually is to plan a program of scheduled insulin appropriate to the specific conditions of hospitalization.28
Insulin for Specific Conditions of Hospitalization
Scheduled (programmed) insulin
Specific insulin products are not uniquely identified as replacement for particular physiological insulin functions (basal or nutritional). However,for purposes of calculation of an initial TDD of insulin and establishment of a pattern of delivery, it is useful to think of therapy as having the following components: basal, nutritional, and correction-dose insulin. To match the anticipated and changing patterns of insulin requirement under conditions of hospitalization, providers have a palette of insulins and insulin analogs, each having a characteristic peak and duration of action when given subcutaneously, with which to “paint” the day(Figure 1).
Rapid-acting analog therapy (lispro, aspart) is efficacious for coverage of prandial needs for patients eating discrete meals or for correction of hyperglycemia. For acute severe hyperglycemia, BG monitoring and subcutaneous injection of rapid-acting insulin analog can be ordered every 2–2.5 hours until BG correction occurs. Given every 1–2 hours, rapid-acting analog has been used in the treatment of ketoacidosis.35 If perfusion of subcutaneous sites is impaired, however, after repeated doses of subcutaneous insulin, there is a danger of delayed effect occurring beyond the expected time frame of action, with consequent late hypoglycemia. Rapid-acting analog therapy is not well suited for coverage of basal requirements or insulin requirements resulting from continuous delivery of nutritional support or intravenous dextrose.
Long-acting peakless analog (glargine) administered once daily for patients having type 1 diabetes ensures against inadvertent omission of needed basal insulin coverage. A correctly established dose of glargine providing basal coverage may be safely administered in type 1 diabetes, despite NPO (nothing by mouth) status.36 With realization that requirements for exogenous insulin may have been inflated because of overnutrition and may vanish during prolonged NPO status,a safe minimum dose estimating true basal requirements still may be replaced as glargine insulin during NPO status in type 2 diabetes. Because nutrition or dextrose suddenly may be interrupted, as a safety precaution against hypoglycemia in unstable hospitalized patients, the doses of glargine generally should not be titrated to meet nutritional requirements, but rather,they should meet only the estimated basal insulin requirement. Glargine is not recommended during pregnancy.
NPH and subcutaneous regular insulin are especially useful for patients whose carbohydrate exposure is not delivered as discrete meals. When insulin is required, a regimen of mixed NPH and regular insulin given every 6–8 hours provides excellent coverage for NPO status, intravenous dextrose, or continuous enteral feedings. NPH and subcutaneous regular insulin can supplement intravenous insulin coverage used as an additive with total parenteral nutrition. NPH and subcutaneous regular insulin are well suited to cover transitional meal plans and corticosteroid therapy. Intermediate-acting insulin and regular insulin or rapid-acting analog also can be used to cover normal diet with discrete meals. NPH insulin is preferred to glargine for basal coverage during use of multiple daily injection therapy in pregnancy.
Starting doses of insulin
In type 1 diabetes, true basal requirements for exogenous insulin are absolute. Generally, the basal insulin requirement is ∼ 40–50% of the preadmission TDD of insulin, or (in the absence of renal failure or other conditions that reduce insulin requirement) a conservative low estimate is∼ 0.25 units/kg.
In type 2 diabetes, the apparent requirement for exogenous basal insulin may vanish during prolonged NPO status. However, a safe minimum starting point(in the absence of renal failure), which would be tolerated as replacement for endogenous basal insulin production even if the requirement should trend downward, is ∼ 0.15 units/kg of actual body weight.
For patients who are eating normal meals, there is a nutritional insulin requirement approximately equal to the basal requirement. For patients receiving intravenous dextrose or nutritional support, the nutritional requirement can be initially estimated at ∼ 1 unit/10 g carbohydrate, an amount that often requires subsequent upward revision.37
The starting TDD of insulin can be envisioned as the sum of the basal and nutritional requirements, calculated as discussed above.
Correction doses, “hold” orders, and daily revision of scheduled insulin
“Sliding scale insulin” as monotherapy generally is ineffective and may be harmful.38–40 Correction doses (supplements) of rapid-acting analog or regular insulin should be proportionate to the TDD insulin requirement or, if that requirement is unknown, then correction doses should be proportionate to an estimate of TDD requirement based on assumptions about typical requirements for a given body weight.41,42 Over several days, use of a dose-finding process for the scheduled insulin requirement reduces reliance on correction doses.
Except for basal insulin in type 1 diabetes, orders might include a provision for nursing staff to withhold or reduce doses of scheduled insulin in the event of downward trending of BG or interruption of carbohydrate exposure.
The dose of scheduled insulin should be recalculated daily based on BG response, the amount of insulin actually administered on the previous day, and the changing condition of the patient. Review of charted doses is necessary to discover withholding of ordered doses of scheduled insulin or use of correction doses. For both type 1 and type 2 diabetes, it is an important precaution to reduce the TDD of scheduled insulin in case of downward trending of BG or development of factors that might predispose to hypoglycemia, such as renal or hepatic failure or resolution of stress- or drug-related hyperglycemia.42
Consistent carbohydrate diets
Patient self-management of diabetes in the hospital
Unless they prefer to have their doses of insulin and meal plans assigned,competent patients who are experienced in self-management generally should continue to determine their dietary selections and administer their insulin in the hospital.45 A multidisciplinary effort is necessary to establish a safe policy on inpatient diabetes self-management. (Readers are referred to the article by Nettles in this issue of Diabetes Spectrum, ).
Hypoglycemia prevention
In one study, 45% of episodes of hospital hypoglycemia resulted from triggering events, such as sudden change of caloric exposure, and 39% resulted from inappropriate adjustment of insulin dose in the presence of medical conditions predisposing to hypoglycemia or previous episodes of hypoglycemia. When there is poor perfusion of subcutaneous injection sites, as with hypotension, use of pressors, or edema, delayed snowballing of repeated doses of subcutaneous insulin can cause hypoglycemia. Because these situations are recognizable in advance, iatrogenic hospital hypoglycemia appears to be largely predictable and frequently should be preventable by fairly simple measures.46,47 (Readers are referred to the article by Tomky in this issue of Diabetes Spectrum, , for nursing precautions.)
Discharge planning
The question may be raised of whether hospital hyperglycemia in fact signifies the presence of diabetes. An A1C of > 6% drawn at admission suggests diabetes, whereas an A1C < 5.2% suggests nondiabetes.34 In outpatient follow-up, the diagnosis should be established by BG criteria.48
Hospitalization may provide an ideal opportunity to introduce patients with diabetes to strategies that might improve their preadmission control or reduce hypoglycemia, such as the addition of basal insulin to oral agents49,50 or intensification from use of regular and intermediate-acting insulin to multiple daily injections of insulin using a rapid-acting analog and basal insulin.51–55 Outpatient follow-up over several visits with a diabetes educator and dietitian is necessary to establish the skill of matching prandial insulin to variable carbohydrate intake. This strategy (advanced carbohydrate counting),at least in type 1 diabetes, yields results superior to fixed-dose use of prandial insulin.56
Education of caregivers
One barrier to improved diabetes care in the hospital, shown by Bernard et al.,57 is that most trainee physicians do not think that additional training in diabetes care is needed. In that study, resident physicians felt that insufficiency of time was a greater barrier than deficiency of training. The authors hypothesized that the difficulties with residents' diabetes practices could result from a lack of knowledge and experience on the part of the supervising physicians. While this study was conducted in an ambulatory care setting, the same conclusions could be extrapolated to the inpatient experience.
Baldwin et al.58 explored this issue further by designing a systematic approach to educate residents in inpatient insulin management. Residents were educated regarding management of hyperglycemia without using sliding scale insulin, which is ubiquitous in most inpatient settings and oftentimes is the sole means of blood glucose management. An endocrinologist supervised a basal-bolus regimen in all patients treated with insulin. A team consisting of two residents managed BG levels for all of the team's patients and rounded twice daily with the supervising endocrinologist.
At the end of the rotation, all house officers felt confident in inpatient diabetes management without using a sliding scale. There was a significant decrease in the length of stay compared to a risk-matched control group, and the use of a sliding scale was effectively abolished in this particular training program. Interestingly, the use of a sliding scale on the surgical services not involved in this study also declined by 43%.
This study exemplifies how intensive education, particularly for fellows,residents, and interns, can go a long way in improving glycemic control and achieving overall improved outcomes for diabetic patients. In the future, it would be helpful to have more data regarding the efficacy of different teaching models for education of house staff and other health care providers involved in diabetes management.
Several methods of providing staff nurses with education about diabetes management have been reported in the literature. Adams and Cook59 compared the diabetes nursing care of two groups of home health agency nurses: those with a diabetes educator and those without. They found that although both groups scored similarly on a diabetes knowledge test, the nurses with the diabetes clinical nurse specialist scored higher in their care of diabetes patients on a nursing care intervention tool. The authors speculated that this resulted from the orientation modules, nursing care plans, and teaching tools provided to the staff nurses by the diabetes clinical nurse specialist.
Self-directed learning about diabetes was the approach to educating staff nurses used by Dunning.60 The author developed a nursing care manual for the most frequent admission diagnoses for diabetes patients at a major teaching hospital. This manual was a unit resource for staff nurses. The manual was designed to improve the knowledge of nurses caring for diabetic patients and to be used as a springboard for developing nursing policies and standards for the nursing care of patients with diabetes.
Parker et al.61 studied the impact of a diabetes education program on improving nurses'diabetes knowledge and changing nurse behavior in long-term care facilities. Nurses from two long-term care facilities participated in a multisession education program on diabetes treatment. Diabetes knowledge was assessed using both a pre- and postprogram test. Comparison was made with a control group of nurses from two other long-term care facilities who did not attend the education program. The nurses attending the education program had a significant increase in their postprogram knowledge score compared to those in the control group. However, a chart review revealed that there was no significant improvement postprogram in diabetes nursing care behaviors, such as giving insulin in the abdomen or giving carbohydrate after low blood glucose. The authors postulated that the lack of behavior change was because there was no opportunity for re-enforcement sessions and no contact with a diabetes nurse educator.
The findings of Parker et al. were supported by the work of Asselin.62 She found that experienced hospital staff nurses primarily use new knowledge from unit-based resources to change their clinical practice. In only 1 of 29 examples given by these nurses was nursing practice changed by knowledge gained from attending a formal continuing education program. Asselin recommended that nursing staff development focus on unit-based resources and unit-based trainers. In addition, a network of multidisciplinary resource personnel and unit-based computer access to disease-specific knowledge bases would provide further information. A regular staff member from each unit thus might receive additional training and become a champion for diabetes to promulgate the spread of knowledge among his or her peers.
A nursing clinical practice group can be defined as a group of nurses in the same nursing specialty at an institution who meet periodically to review their current nursing practice as it relates to standards of nursing care,institutional policies and procedures, and evidence-based practice. Clinical practice groups offer an avenue for communicating new ideas and new requirements and challenging the status quo. They allow for the use of critical thinking skills to evaluate current and future clinical practices and to discuss ways to improve nursing care for patients with specific diseases or conditions. This venue supports both novice and experienced nurses'perspectives on patient care. It enhances the likelihood that both the organization and its patients will benefit from performance improvement measures in order to provide the best diabetes care and education possible. Clinical practice groups are an excellent way to enhance clinical nursing practice.
Organization, Safety, Quality, and Systems Development
Who should be the caregiver?
The specialty training of caregivers may influence the hospital course for patients with diabetes. One study of diabetic ketoacidosis (DKA) by Levetan et al.63 compared outcomes in patients cared for by generalists to those cared for by endocrinologists. The endocrinologist-treated group had a significantly lower readmission rate for DKA as well as a significantly shorter length of stay. Endocrinologists treated patients more costeffectively than generalists,keeping their costs to almost half of their generalist counterparts. This savings resulted mainly from avoidance of unnecessary imaging and laboratory studies by those unfamiliar with DKA. In another study by Levetan et al.,64 the length of stay for patients with diabetes was shorter by 35% for those cared for by a diabetes consult team, which was a statistically significant difference.
Koproski et al.65 randomly assigned 179 patients to either usual care or care by a diabetes team consisting of a nurse educator and an endocrinologist. The diabetes team intervention reduced the length of hospital stay, improved glycemic control,and reduced the rate of recurrent hospitalizations when compared to usual care. Clearly, the multidisciplinary diabetes care team approach is more effective than usual care, and there is some evidence that specialized nursing care contributes to the difference. Davies et al.66 showed that diabetic patients randomized to a diabetes specialist nursing service had shorter lengths of stay and lower costs of care. Patient satisfaction and knowledge of disease was superior in the intervention group.
Safety and quality
Safety and quality can be best assured by a multidisciplinary effort supported by hospital administration and diabetes experts and involving quality assurance personnel, nurses, dietitians, pharmacists, the point-of-care testing program, clinical informatics, physicians, and members of other departments.1,67–71
Managing complexity with a single signature: paper protocols and electronic health records
If hospital systems present impediments to comprehensive care plans, even the most well-intentioned and knowledgeable caregivers, defeated by the pressure of time and taking the path of least resistance, may order an oversimplified and ineffective plan. In contrast, if the hospital offers standardized order sets, then practitioners may easily implement a complex and multifaceted individualized care plan by entering a few check marks and numbers and a single signature.41,42,72 Whether the practitioner 1) uses knowledge-embedded paper order sets or 2) has access to a robust integrated electronic health record that incorporates a clinical data repository, clinical decision support and rules,and computerized provider order entry,73 the resulting plan of care must reflect the necessary complexity of detailed diabetes care.
Computerized provider order entry for diabetes is in its infancy. Nevertheless, computer prompting has been shown to reduce reliance on sliding scale monotherapy and increase the use of scheduled insulin.74 Computerized systems must make it convenient for users to select and implement the components of a comprehensive diabetes care plan under protocols or algorithms (Table 2). The design of any paper-based or computerized clinical order set based on accepted algorithms or protocols must: 1) query the electronic database or provider about weight, age, pregnancy, and risk for hypoglycemia; 2)encompass the desired input and output; 3) provide options for user override of defaults and for free text entries; 4) have flexibility to encompass future change; 5) accept input from diabetes experts and users and 6) ensure transparency of algorithms for review by the users.
Conclusion
Initiatives to intensify the treatment of hospital hyperglycemia result in improved patient outcomes and cost reductions.
How can hospitals move in the direction of intensifying treatment? One part of the answer is the participation of diabetes experts and champions, not only in the direct care of patients, but also in the education of physicians,nursing staff, dietitians, pharmacists, and others.
Another part of the solution, under enlightened hospital leadership, is the multidisciplinary design of systems that support improvements in hospital management of hyperglycemia. Such efforts potentially would affect screening for hyperglycemia; knowledge of appropriate glycemic targets; recognition of hyperglycemia as an indication for monitoring and treatment; spread of the use of intravenous insulin infusion; recognition and interception of unsafe situations leading potentially to hypoglycemia or ketosis; preservation of basal insulin regimens for type 1 diabetes; standardization of approaches to prevent hypoglycemia; facilitation of inpatient diabetes self-management for experienced and competent patients; elimination of sliding scale monotherapy;generation of effective orders for scheduled subcutaneous insulin therapy;creation of order sets and implementation of computerized clinical systems that facilitate regimen selection, intensification, and daily revision of insulin therapy; and appropriate patient education and discharge planning.
Cara L. Thompson MSN, APRN, BC-ADM, RD, LDN, CDE, is a diabetes clinical nurse education specialist in the Department of Nursing Practice Education and Research at the University of North Carolina Hospitals in Chapel Hill, N.C. Kelli C. Dunn, MD, and Meera C. Menon, MD, are second-year fellows in the Department of Endocrinology/Medicine at the same institution. Lauren E. Kearns, MSN, RN, BC, is a liver transplant coordinator at the same institution and is board certified in nursing informatics. Susan S. Braithwaite, MD, FACE,FACP, is a clinical professor of medicine at the University of North Carolina Diabetes Care Center in Durham, N.C.
Note of disclosure: Dr. Braithwaite has received honoraria for speaking engagements from Aventis and research funding from Bristol-Myers Squibb. Both companies make insulin products for the treatment of hyperglycemia.
