Type 2 diabetes has become a worldwide epidemic1 and is associated with multiple complications that can be prevented by modifying risk factors and optimizing glycemic control.2 The optimization of glycemic control often requires the use of multiple agents, including insulin.
Insulin is an important component of antihyperglycemic therapy, yet there are many perceived barriers.3 Existing guidelines do not specifically address the topic of insulin initiation.4 We review and analyze the evidence from randomized controlled trials on insulin initiation and address adverse effects and barriers. We also discuss the selection of an insulin regimen, titration and delivery of care, as well as when and how to combine insulin therapy with oral antihyperglycemic agents. A summary of our systematic review and meta-analysis is available in Box 1.
Summary of the literature review and meta-analysis
We performed a systematic review of studies examining the effect of the initiation of subcutaneous insulin therapy on glycemic control, weight gain, risk of hypoglycemia, other adverse effects and diabetic complications in outpatients with type 2 diabetes. We excluded studies involving children, adolescents, or patients with type 1 diabetes or gestational diabetes, as well animal studies and trials of inhaled, intravenous, intraperitoneal or continuous subcutaneous insulin treatment. We focused on studies involving insulin-naive patients in the ambulatory care setting and excluded those comparing various insulin regimens in patients already taking insulin. Details regarding our search strategy and meta-analysis are included in Appendix 1.*
We included 56 studies for quantitative review, and 39 studies and 3 systematic reviews for qualitative review (Appendix 2*). Study quality is summarized in Appendix 3.* All studies were randomized controlled trials, and most were sponsored by industry.
Methodologic quality varied among the trials. In general, participants were 50–70 years old, had diabetes for 8–10 years and were taking at least one oral antihyperglycemic agent. In our meta-analysis, although some comparisons had high heterogeneity, we used a random-effects model that incorporated this uncertainty in the resulting estimates of treatment effect and confidence intervals.
↵* The appendices are available at www.cmaj.ca/lookup/suppl/doi:10.1503/cmaj.110779/-/DC1.
When should insulin be started?
Clinical practice guidelines vary as to the recommended criteria for the initiation of insulin therapy in patients with type 2 diabetes. Factors that are considered include the control of blood glucose levels and comorbidities that affect choice of treatment.
Glucose control
The American Diabetes Association and the European Association for the Study of Diabetes developed a consensus algorithm wherein basal insulin is recommended as a second-line agent if the glycated hemoglobin (HbA1c) value is greater than 7.0% after metformin monotherapy.5 Similarly, the International Diabetes Federation recommends that insulin be started if optimized oral antihyperglycemic therapy and lifestyle interventions are unable to maintain blood glucose at target levels.6 In contrast, the Canadian Diabetes Association recommends that insulin be considered as a first-line agent if the HbA1c value is 9.0% or greater in patients with newly diagnosed diabetes or if there is symptomatic hyperglycemia with metabolic decompensation (defined as polyuria, polydipsia and weight loss), and as a second-line agent if the HbA1c is still not at target levels (consensus recommendation).4
Although no randomized controlled trials have looked at the impact on cardiovascular outcomes of insulin initiation early in the course of type 2 diabetes, early intensive control of blood glucose levels was assessed in the United Kingdom Prospective Diabetes Study (UKPDS). The study compared intensive glycemic control (with metformin, secretagogue or insulin therapy) and conventional glycemic control in patients with newly diagnosed diabetes and found that those in the intensive treatment group had reduced microvascular and macrovascular complications in long-term follow-up.7,8 Early insulin initiation has been shown to improve and preserve β-cell function,9 and the ongoing Outcome Reduction With Initial Glargine Intervention (ORIGIN) trial will assess the impact of an early basal insulin strategy on cardiovascular outcomes.10
Comorbidities
Careful monitoring of glycemic control is necessary when treating diabetes in patients with renal or hepatic failure. Many oral antihyperglycemic agents are contraindicated or require dose modification in these patients, necessitating initiation of insulin therapy. Because insulin is cleared renally, its dose must be reduced to prevent hypoglycemic episodes.11 Compensated cirrhosis is associated with insulin resistance and often requires higher doses of insulin, whereas decompensated cirrhosis (associated with complications such as variceal hemorrhage, ascites, spontaneous bacterial peritionitis or hepatic encephalopathy) results in reduced metabolism of insulin in the liver and hence would require lower doses of insulin.12
What adverse effects have been reported with insulin use?
Common concerns about the use of insulin include the risk of hypoglycemia and weight gain. Other adverse events, such as congestive heart failure and lipodystrophy, are much less common.
Hypoglycemia
Fear of hypoglycemia remains a barrier to insulin initiation for patients and physicians, although physicians’ perception of this risk is amplified relative to patients (Table 1).13 The prevalence of hypoglycemia associated with the use of insulin varies widely: from 9% of patients when defined as a glucose level below 3.1 mmol/L,14 to 64% of patients when defined as a glucose level below 3.0 mmol/L.15 In our meta-analysis comparing insulin use with oral agents used alone, which included data from six studies, the risk of hypoglycemia was higher with insulin (odds ratio [OR] 2.23); however, the wide 95% confidence interval (95% CI 0.59 to 8.42) and the high level of heterogeneity (I2 = 89.5%) make this risk difficult to interpret (Table 2).
Risk factors for hypoglycemia include inappropriate dose, timing or type of insulin, decreased glucose delivery (e.g., a missed meal), decreased endogenous glucose production (e.g., alcohol intake), increased glucose utilization (e.g., through exercise), increased insulin sensitivity (e.g., weight loss, treatment with insulin sensitizer) and decreased insulin clearance (e.g., renal failure).57 The risk of hypoglycemia can be reduced by addressing these factors. Strategies for preventing hypoglycemia in patients using insulin include asking about hypoglycemia at each visit, education, frequent self-monitoring, individualized glycemic goals and continued professional guidance.57
Weight gain
Weight gain that occurs with the initiation of insulin therapy has been reported to vary from 0.3 kg43 to 6.4 kg30 and may contribute to a patient’s reluctance to start using insulin.58 Weight gain occurs early — in the first weeks to months after insulin initiation59 — then levels off, correlating with the intensity of insulin titration. In one study in which insulin was rapidly titrated over the first 12 weeks, almost 80% of weight gain occurred during this period.31 In contrast, when titration occurred much more slowly, weight gain occurred gradually over the two-year study period.16 The amount of weight gained should be tempered by consideration that weight lost previously because of poor glycemic control is regained after the initiation of insulin60 and that an increase in lean body mass accounts for 30% of this weight gain.16,61
Risk factors for weight gain include fear of hypoglycemia, depression, use of antidepressants and choice of insulin regimen (Table 1).13 Although cardiovascular implications of this weight gain are undefined, weight gain that follows insulin initiation has been correlated with a deterioration in cardiometabolic risk factors such as blood pressure and serum levels of triglycerides, low-density lipoprotein (LDL) cholesterol and high-density lipoprotein (HDL) cholesterol.16,62
Strategies to minimize weight gain following insulin initiation include identification of and addressing the fear of hypoglycemia;58 treatment of depression;58 selection of weight-neutral antidepressant agents;58 promotion of regular exercise;58,62 dietary modifications, in conjunction with a dietitian, to restrict energy intake and reduce fat intake,58,62 particularly for patients with pre-existing obesity; and selection of an appropriate insulin regimen (discussed later in the article). The risk of weight gain can be addressed through the use of an interprofessional approach.
Other reported adverse effects
Observational studies have reported a number of rare adverse events associated with insulin use, including congestive heart failure, edema, lipodystrophy, allergic reactions, reversible transaminitis, reversible nephrotic syndrome and β-cell destruction. The prevention and management of these adverse events are summarized in Appendix 4 (available at www.cmaj.ca/lookup/suppl/doi:10.1503/cmaj.110779/-/DC1).
What barriers affect insulin initiation?
Barriers to insulin initiation exist at the patient, physician and system levels. As mentioned previously, barriers perceived by physicians were amplified compared with patients’ perceptions (Table 1).13 For example, fear of hypoglycemia, weight gain, and injection-related pain and anxiety were ranked higher as barriers by physicians than by patients, whereas patients’ concerns included fear of “addiction” to insulin and a lack of understanding of the seriousness of their disease.
Although quality of life is commonly assumed to be reduced with insulin initiation, quality of life and treatment satisfaction have been found to be unchanged63 or improved64,65 in randomized controlled trials.
Additional barriers are listed in Table 1. Insulin initiation is often linked to patients’ feelings of blame and failure.3 Patients may also be deterred by inconvenience and lack of portability of insulin.66 The identification and addressing of patient barriers through counselling (including education regarding modern devices for insulin delivery) may improve acceptance by patients.13,66
The care of a patient with type 2 diabetes is a time- and resource-intensive process.13,67 Practice- and system-level factors contribute substantially to the successful initiation of insulin therapy. An integrated health care team is required that includes the primary care physician, a diabetes educator and consultants.68 The family physician serves as the principal medical contact for the patient and provides continuity of care. The diabetes educator facilitates education programs that support self-management.69,70 Shared care with a diabetes specialist has been shown to result in improved glycemic control.71 Diabetes case managers can help improve the delivery of care and clinical outcomes by coordinating care and facilitating timely medication changes.72
Organizational and technologic interventions such as electronic databases and automated reminders about appointments, investigations and interventions can improve the efficiency and effectiveness of the diabetes health care team.72 At a systems level, adequate funding to support comprehensive diabetes health care teams is important. Funding for these teams may be threatened by limited health care resources in the setting of an aging population.73
How is the starting regimen chosen?
Three types of insulin regimen are commonly used: basal, premixed and basal–bolus. Premixed and basal–bolus regimens result in greater reductions in HbA1c compared with basal regimens. However, they are associated with more weight gain and, in the case of premixed regimens, an increased risk of hypoglycemia, and are more complex, which may affect adherence (Table 2; see also Appendix 5, available at www.cmaj.ca/lookup/suppl/doi:10.1503/cmaj.110779/-/DC1). We suggest that a basal regimen is the ideal one to start with, given its simplicity and favourable safety profile, recognizing that modifications to this regimen will be required over time to maintain glycemic control. In addition, the acceptability, feasibility and effectiveness of this approach in a primary care setting have been shown and will be detailed later in this paper.
The types of insulin, their pharmacokinetics and regimens are outlined in Appendices 6 and 7 (available at www.cmaj.ca/lookup/suppl/doi:10.1503/cmaj.110779/-/DC1). The effects of various regimens on HbA1c, weight gain and risk of hypoglycemia are summarized in Table 2. Data regarding the effect of different regimens on vascular outcomes, lipid profile and quality of life are limited (see Appendix 8, available at www.cmaj.ca/lookup/suppl/doi:10.1503/cmaj.110779/-/DC1).
Choice of regimen
Our review identified 15 randomized trials that compared basal regimens with premixed regimens (Table 2).14,18,30–42 Patients given a premixed regimen were found to have a greater reduction in HbA1c, but they also had greater weight gain and a higher risk of hypoglycemia.
Similarly, our analysis of eight randomized trials comparing basal regimens with basal–bolus regimens18,19,30,32,43–46 found that the latter were associated with a greater reduction in HbA1c, greater weight gain and a trend, although not significant, toward more frequent hypoglycemia (Table 2).
Our analysis of three trials comparing premixed regimens with basal–bolus regimens found no significant difference in effect on HbA1c.18,30,47 Data were insufficient for a pooled estimate of the effect on weight gain and hypoglycemia, but the individual studies showed no difference between these two regimens on weight gain or risk of hypoglycemia.18,30 Thus, although premixed and basal–bolus regimens may be better than basal regimens at reducing HbA1c, this strength should be balanced against the increased weight gain associated with both regimens and the increased risk of hypoglycemia associated with premixed regimens.
Choice of insulin
Basal insulin
A basal insulin in the form of a long-acting analogue is preferable because it offers glycemic control comparable to that of an intermediate-acting insulin but is associated with a lower risk of hypoglycemia. Our analysis of seven randomized trials comparing intermediate- and long-acting insulin analogues15,48–53 showed that they performed equally in terms of HbA1c reduction (Table 2). Patients taking intermediate-acting analogues had slightly lower weight gain; however, they had significantly more episodes of hypoglycemia (Table 2). A recent Cochrane review comparing intermediate- and long-acting insulins (which did not focus specifically on studies of insulin initiation) reported similar findings: no difference in HbA1c reduction, but a significantly greater reduction in hypoglycemic events with the long-acting insulin analogues glargine (OR 0.84, 95% CI 0.75 to 0.95) and detemir (OR 0.56, 95% CI 0.42 to 0.74).74
We identified two studies that compared glargine and detemir for insulin initiation.75,76 A meta-analysis was not done given the number of studies. These studies showed inconsistent findings: one reported a greater reduction in HbA1c of 0.3% with glargine than with detemir (p < 0.004),75 but the other found no difference between the two products.76 Both studies showed a greater weight gain (0.77–1.37 kg) with glargine than with detemir, but a similar risk of hypoglycemia.
In our analysis of four studies that compared morning with bedtime administration of basal insulin,18,49,54,55 we found no significant difference in HbA1c reduction or weight gain. Data were insufficient for a pooled estimate of the effect on hypoglycemia.
Bolus insulin
Current evidence shows that both rapid-acting and short-acting insulins are reasonable choices for bolus insulin. Rapid-acting analogues more closely mimic physiologic insulin secretion than short-acting ones do. However, the literature has not consistently shown improved glycemic control when comparing insulin initiation with rapid-acting and short-acting insulins. One randomized controlled trial showed reductions in HbA1c but an increased risk of hypoglycemia among patients given rapid-acting insulin.77 Another study comparing rapid-acting and short-acting premixed insulin revealed no difference in effect on HbA1c.42
Who should initiate and titrate insulin?
Insulin therapy can be initiated by primary care physicians and specialists, and the dose titrated by patients themselves with support from their health care provider. Before initiating insulin therapy, the clinician should ensure that the patient is able to monitor his or her blood glucose level, is aware of the symptoms of hypoglycemia and has adequate knowledge about dealing with these episodes. Patient education can be facilitated by an interprofessional diabetes care team and implementation of organization-level processes of care that enable regular diabetes monitoring and recall (e.g., telephone reminders for upcoming appointments).72
In the Implementing New Strategies with Insulin Glargine for Hyperglycaemia Treatment (INSIGHT) trial, patients of either family physicians or diabetes specialists were randomly assigned to receive oral antihyperglycemic agents alone or in combination with insulin glargine.21 In a post-hoc analysis of whether patients’ outcomes differed depending on the type of physician managing their care, the reductions in HbA1c values and rates of hypoglycemia were comparable among patients who had insulin glargine initiated by either family physicians or diabetes specialists. Similarly, post-hoc analysis of the AT.LANTUS trial (A Trial Comparing Lantus Algorithms to Achieve Normal Blood Glucose Targets in Subjects With Uncontrolled Blood glucose), which randomly assigned patients to self-titration or clinic-driven titration of glargine, found that rates of glycemic control achieved in primary care settings were similar to rates achieved in specialist care settings.40
Several randomized trials have shown that protocols for self-titration of insulin by patients are effective in safely lowering HbA1c.78–83 Patients using these self-titration algorithms coupled with daily blood glucose monitoring were able to achieve HbA1c levels similar to those achieved in clinics, reducing their HbA1c by 1.0% to 2.5%. A variety of insulin types were used in these studies, including 30/70 twice-daily insulin,79 detemir once daily,80 glargine once daily78,81 and rapid-acting insulin before meals.83 Self-titration by patients resulted in similar rates of hypoglycemia compared with physician-managed titration, even though the patients used higher doses of insulin (0.59 units/kg v. 0.40 units/kg).80 In one randomized controlled trial, self-titration was taught to individuals and to groups, with the two approaches achieving similar reductions in HbA1c (1.8% and 2.0% respectively); the group sessions required half the time.81 Regarding titration of bolus insulin, simple titration based on postprandial values was as effective as carbohydrate counting, lowering HbA1c by 1.5%.83
The principles of insulin titration are described in Box 2, and sample protocols for insulin titration are described in Table 3. Insulin can be initiated in either primary care or specialist settings, and patients can titrate insulin doses themselves with no increased risk of hypoglycemia. Group sessions offer an efficient and effective alternative to individual counselling.
Principles of insulin titration by regimen
Basal (intermediate- or long-acting insulin)
Adjust the dose based on the fasting glucose level
Premixed insulin at breakfast and dinner
Adjust the breakfast dose based on the previous dinner reading (as long as a dose increase does not cause hypoglycemia at lunchtime)
Adjust the dinner dose based on the fasting glucose level (as long as a dose increase does not cause hypoglycemia at bedtime)
Basal–bolus*
Adjust the dose at mealtime based on the previous day’s glucose level measured either two hours after the corresponding mealtime or before the next mealtime (e.g,. adjust the breakfast dose based on the previous day’s two-hour post-breakfast value or the pre-lunch value)
↵* Rapid- or short-acting insulin is used for bolus dose.
Should oral antihyperglycemic agents be continued when initiating insulin therapy?
Continuation of oral antihyperglycemic agents should be considered when initiating insulin therapy. The combination of therapies can result in greater HbA1c reduction and lower daily insulin requirements.
Compared with insulin initiated alone, insulin combined with oral antihyperglycemic agents (except for thiazolidinediones) has comparable effects on glycemic control and better effects on weight gain, insulin dose and hypoglycemia.
In our analysis of studies comparing insulin alone or in combination with various oral antihyperglycemic agents (discussed in more detail later), we found no increased risk of adverse effects when insulin was combined with oral antihyperglycemic agents, with two exceptions. Combination therapy with insulin and thiazolidinediones (e.g., pioglitazone) was associated with greater weight gain, edema and possibly hypoglycemia; combination therapy with insulin and acarbose was associated with an increased risk of gastrointestinal upset. Appropriately powered data on morbidity and mortality, such as cardiovascular events, were not available.
Metformin and secretagogues
We identified 25 studies comparing insulin initiation alone or in combination with metformin or a secretagogue or both.18,22,31,33,37–39,45,54,55,77,84–97 In our meta-analysis, we found no significant reduction in HbA1c with a combination of an oral antihyperglycemic agent and insulin (−0.14%, 95% CI −0.36 to 0.08; n = 2566);18,22,31,33,37–39,45,54,77,84–97 this effect was not moderated by class of oral antihyperglycemic agent or by insulin regimen. However, combination therapy was associated with significantly less weight gain (−0.83 kg, 95% CI −1.32 to −0.33; n = 2060)18,22,31,33,37–39,45,54,55,87–89,91,92,94–96 and significantly lower insulin doses (−17.7 units, 95% CI −26.1 to −9.2; n = 1308).18,31,33,37,39,54,87–89,91,92,94,95,97 It was also possibly associated with a reduced risk of hypoglycemia (OR 0.75, 95% CI 0.56 to 1.00; n = 838).22,33,39,84
Thiazolidinediones
A recent meta-analysis comparing insulin initiation alone or in combination with pioglitazone found that the combination therapy was associated with a significantly greater reduction in HbA1c (mean difference −0.58%, 95% CI −0.70% to −0.46%), a trend toward lower daily insulin requirements (−12 units daily) and higher HDL cholesterol levels (by 0.10 to 0.18 mmol/L).98 However, it resulted in greater weight gain (mean difference 2.91 kg, range 3.85 kg to −3.50 kg) and more peripheral edema; the effect on hypoglycemia was equivocal (relative risk 1.27, 95% CI 0.99 to 1.63).
Acarbose
One trial examined insulin initiation combined with either acarbose or placebo.99 Compared with the patients in the placebo group, those given combination therapy with acarbose had a greater reduction in HbA1c (difference −0.50%, 95% CI −0.93% to −0.07%) and less weight gain (mean difference −10.01 kg, 95% CI −15.5 kg to −4.5 kg), but a higher incidence of digestive disorders (40% v. 17%).99
Oral agents available for combined therapy with insulin
Currently, acarbose is approved for combined therapy with insulin in both Canada and the United States. Pioglitazone, metformin and glimepiride are approved for combined use only in the United States. Other oral antihyperglycemic agents, such as dipeptidyl peptidase-IV (DPP-IV) inhibitors and glucagon-like peptide 1 (GLP1) agonists, are not approved for combination therapy with insulin in Canada or the United States.
Unanswered questions
Although glycemic control can improve morbidity and mortality among patients with type 2 diabetes, the impact of insulin, as well as the comparative effectiveness of different insulin strategies, on cardiovascular outcomes is unknown. Improved glycemic control is associated with an increased risk of hypoglycemia, and the optimal balance between the two remains unclear. Regimens that improve glycemic control often increase the complexity of treatment; limited data exist regarding the comparative effect of different strategies on quality of life and treatment satisfaction. Finally, effective and sustainable delivery of care has yet to be elucidated.
Conclusion
The choice of an insulin regimen and the timing of starting insulin therapy remain a discussion between the physician and patient. Factors such as accessibility, cost, patient preference and adverse effects (especially hypoglycemia and weight gain) must be considered. Box 3 provides an example of when and how to initiate insulin using the results of our literature review. Appendix 9 (available at www.cmaj.ca/lookup/suppl/doi:10.1503/cmaj.110779/-/DC1) provides information on initiating insulin in older adults.
Applying the results of this review in clinical practice (fictional case)
A 54-year-old man was diagnosed with type 2 diabetes 12 months ago with a presenting HbA1c value of 9.8%. At the time of diagnosis, he attended a diabetes education class, began taking metformin and gliclazide and started a walking program. To date, he has no identified macrovascular or microvascular complications of diabetes. Despite the use of the two oral antihyperglycemic agents, his HbA1c value is 8.2%, and his fasting blood glucose level is 9.8 mmol/L. The findings on physical examination are normal except for obesity. Should insulin therapy be started? If so, what type and dosing regimen should be chosen?
Insulin therapy is indicated because the addition of a third oral antihyperglycemic agent will likely not achieve the target HbA1c value. Also, early intensive glycemic control reduces the risk of macrovascular and microvascular complications. The role of insulin is discussed with the patient. Although he is not worried about the increased self-monitoring of blood glucose levels required, he expresses a fear of becoming “addicted” to insulin. This fear is assuaged by further discussion regarding the natural history of diabetes.
Given the patient’s HbA1c value of 8.2%, a once-daily long-acting insulin is chosen because of its simplicity and its reduced risk of hypoglycemia. It is prescribed for use at bedtime for convenience. Although the patient is counselled about the off-label use of his oral antihyperglycemic agents in combination with insulin, he elects to continue using them, because combination therapy reduces weight gain, the risk of hypoglycemia and the insulin dose. He is told to start with 10 units of insulin and to increase the dose by 1 unit each day until his fasting blood glucose level is less than 7 mmol/L.
At a brief follow-up visit one week later, the patient is found to be adapting well to the new regimen. His fasting blood glucose level is 8.7 mmol/L, he has titrated his insulin dose to 17 units at bedtime, and he has had no hypoglycemic episodes. Three months later, he is injecting 26 units of insulin at bedtime, his HbA1c value is on target at 6.9%, and he reports improved energy and well-being.
Future research on the effect of different insulin regimens on morbidity and mortality may further help us meet the individual needs of patients.
Key pointsInsulin initiation should be considered early in the course of type 2 diabetes.
Insulin is an effective and safe agent in reducing glycated hemoglobin A1c; although its use is associated with weight gain, it is not associated with an increased risk of hypoglycemia.
A basal regimen is an ideal start given its simplicity and favourable impact on weight and risk of hypoglycemia; intensification of this regimen will be required over time to maintain glycemic control.
Combination therapy that includes insulin and oral antihyperglycemic agents reduces weight gain, insulin dose and risk of hypoglycemia.
Acknowledgements
The authors thank Laure Perrier for performing the literature search and Jisun Oh for participating in the review of abstracts.
Footnotes
Competing interests: Henry Halapy is a member of an advisory board for Eli Lilly, has conducted research funded by a grant from NovoNordisk and has received speaker fees from Shoppers Drug Mart. Catherine Yu has received speaker fees from Shoppers Drug Mart. No competing interests declared by Adrian Lau, Terence Tang or Kevin Thorpe.
This article has been peer reviewed.
Contributors: Adrian Lau participated in the literature searches, reviewed abstracts, analyzed and synthesized the results and drafted portions of the manuscript. Terence Tang abstracted data, assisted with statistical interpretation and drafted portions of the manuscript. Henry Halapy abstracted data and drafted portions of the manuscript. Kevin Thorpe performed statistical analyses and drafted portions of the manuscript. Catherine Yu participated in the literature searches, reviewed abstracts, abstracted data, analyzed and synthesized the results and drafted the manuscript. All of the authors critically reviewed the manuscript and approved the final version submitted for publication.