Intended for healthcare professionals

Clinical Review State of the Art Review

# Benefits and harms of intensive glycemic control in patients with type 2 diabetes

BMJ 2019; 367 (Published 05 November 2019) Cite this as: BMJ 2019;367:l5887
1. René Rodriguez-Gutierrez, professor of medicine1 2 3,
2. José Gerardo Gonzalez-Gonzalez, professor of medicine1 3,
3. Jorge A Zuñiga-Hernandez, assistant professor of medicine1 3,
4. Rozalina G McCoy, associate professor of medicine4 5
1. 1Plataforma INVEST Medicina UANL – KER Unit (KER Unit México), Subdireccion de Investigacion, Universidad Autónoma de Nuevo León, Monterrey, 64460, Mexico
2. 2Knowledge and Evaluation Research Unit in Endocrinology, Mayo Clinic, Rochester, MN 55905, USA
3. 3Endocrinology Division, Department of Internal Medicine, University Hospital “Dr José E González,” Universidad Autonoma de Nuevo Leon, Monterrey, 64460, Mexico
4. 4Division of Community Internal Medicine, Department of Medicine, Mayo Clinic, Rochester, MN 55905, USA
5. 5Division of Health Care Policy and Research, Department of Health Sciences Research, Mayo Clinic, Rochester, MN 55905, USA
1. Correspondence to: R Rodríguez-Gutiérrez rodriguezgutierrez.rene{at}mayo.edu

## ABSTRACT

Diabetes is a major and costly health concern worldwide, with high morbidity, disability, mortality, and impaired quality of life. The vast majority of people living with diabetes have type 2 diabetes. Historically, the main strategy to reduce complications of type 2 diabetes has been intensive glycemic control. However, the body of evidence shows no meaningful benefit of intensive (compared with moderate) glycemic control for microvascular and macrovascular outcomes important to patients, with the exception of reduced rates of non-fatal myocardial infarction. Intensive glycemic control does, however, increase the risk of severe hypoglycemia and incurs additional burden by way of polypharmacy, side effects, and cost. Additionally, data from cardiovascular outcomes trials showed that cardiovascular, kidney, and mortality outcomes may be improved with use of specific classes of glucose lowering drugs largely independently of their glycemic effects. Therefore, delivering evidence based, patient centered care to people with type 2 diabetes requires a paradigm shift and departure from the predominantly glucocentric view of diabetes management. Instead of prioritizing intensive glycemic control, the focus needs to be on ensuring access to adequate diabetes care, aligning glycemic targets to patients’ goals and situations, minimizing short term and long term complications, reducing the burden of treatment, and improving quality of life.

## Introduction

The ethos of this review is aligned with contemporary guidelines, which advocate that the person with diabetes is at the forefront of the decision making process.1234567 However, guidelines often reach discordant conclusions about desired glycemic targets and/or fall short of specifying why, how, and with what tools individualization of diabetes care can practically be achieved.235678 This review provides a comprehensive summary of the evidence on the benefits and harms of intensive glycemic control for non-pregnant adults with type 2 diabetes. It provides practical advice on how patients and clinicians can combine the best available evidence on diabetes management with patients’ needs, goals, values, preferences, and clinical and psychosocial contexts.

## Epidemiology

Diabetes is, worldwide, the leading cause of blindness, non-traumatic lower extremity amputations, peripheral neuropathy, and end stage kidney disease and is a major risk factor for atherosclerotic cardiovascular disease (ASCVD) and ASCVD mortality, which accounts for 43% of all diabetes related deaths in people younger than 70 years of age.910111213 Globally, more than 425 million people are living with diabetes, and its prevalence is expected to increase at least 50% by 2045.1415 Diabetes was responsible for one death every eight seconds in 2017 and accounted for a global expenditure of $727bn (£590bn; €662bn) or$11 638 per person.14 The relentless surge in diabetes is multifactorial, driven by the increasing prevalence of obesity, changes in dietary and physical activity patterns, and an aging population.16 Improving health outcomes, ensuring access to care, alleviating the burden of disease, and containing the costs of diabetes and its complications are therefore a top priority for patients, clinicians, health systems, and societies.

The vast majority of people with diabetes have type 2 diabetes, characterized by progressive insulin resistance and subsequent hyperglycemia.17 Early epidemiologic studies in type 2 diabetes suggested that the long term risk of microvascular and macrovascular complications increases progressively as glucose concentrations rise,181920 inspiring the pursuit of near euglycemia as a means of preventing these complications in type 1 and type 2 diabetes. This glucocentric approach was bolstered by early large randomized controlled trials (RCTs) of intensive glucose lowering therapy—the Diabetes Control and Complications Trial (DCCT)21 among patients with type 1 diabetes and the United Kingdom Prospective Diabetes Study (UKPDS)22 among patients with newly or recently diagnosed type 2 diabetes—which showed reductions in the early manifestations of microvascular complications with intensive glycemic control.23 Importantly, these trials defined intensive control by plasma glucose rather than glycohemoglobin A1c (HbA1c) targets: fasting glucose 70-120 mg/dL (3.9-6.7 mmol/L) and postprandial glucose below 180 mg/dL (10 mmol/L) in the DCCT intensive treatment arm21; and fasting glucose below 106 mg/dL (<6 mmol/L) versus below 270 mg/dL (<15 mmol/L) in the UKPDS intensive and standard treatment arms, respectively.22 However, most clinical guidelines then used these results to recommend intensive control as defined by HbA1c below 6.5-7.0% (48-53 mmol/mol),24252627 extrapolating from improved outcomes achieved by patients who had mean HbA1c 7.0% (53 mmol/mol) compared with 9.0% (75 mmol/mol) in DCCT or 7.9% (63 mmol/mol) in UKPDS.2223

Evidence emerging over the past decade, however, showed that the aggressive efforts often needed to achieve low HbA1c levels can ultimately lead to worse clinical outcomes, greater risk of severe hypoglycemia, and higher burden of treatment.2628293031 At the same time, for most adults with established type 2 diabetes and high risk of ASCVD, the advantage of pursuing intensive (HbA1c <6-7% (42-53 mmol/mol), depending on the trial) compared with conventional (HbA1c 7-8.5% (53-69 mmol/mol)) glycemic targets on hard outcomes important to patients (outcomes that patients appreciate and value, in contrast to surrogate or intermediate outcomes) may be negligible.30313233 More recently, cardiovascular outcome trials (CVOTs) designed to examine the cardiovascular safety of glucose lowering drugs, showed that microvascular (for example, kidney)3435363738394041 and macrovascular (for example, ASCVD, heart failure)35363742434445 health outcomes can be meaningfully improved without substantially lowering the HbA1c. As the paradigm of diabetes management shifts from centering on glucose to overall risk of complications, an urgent need exists to systematically re-examine and contextualize the existing body of evidence on glycemic control in the care of people with type 2 diabetes.

## Sources and selection criteria

With the assistance of an experienced librarian, we did a comprehensive literature search for studies evaluating intensive versus conventional glycemic control in adults with type 2 diabetes in PubMed/Medline, Google Scholar, and Web of Science. We identified and reviewed English language RCTs, observational follow-up studies of these RCTs, systematic reviews, and meta-analyses from 1 January 1970 to 15 June 2019. Search terms with controlled vocabulary used were “intensive”, “tight”, “conventional”, “standard”, “glycemic control”, and “type 2 diabetes”. We excluded the UGDP trial,46 as it was stopped early owing to harm and the data handling was controversial. Data from the UGDP trial, however, was represented in some of the meta-analyses included in this review. We also included published CVOTs that focused on the cardiovascular safety of glucose lowering drugs. We also searched clinicaltrials.gov for ongoing trials evaluating intensive glycemic control in patients with type 2 diabetes. Using the same timeframe and databases, we reviewed original research articles and commentaries discussing shared decision making and minimally disruptive medicine. We examined the most recent consensus statements and guidelines from the American Diabetes Association (ADA), European Association for the Study of Diabetes (EASD), National Institute for Health and Care Excellence (NICE), American Association of Clinical Endocrinologists (AACE), American College of Endocrinology (ACE), American College of Physicians (ACP), US Department of Veterans Affairs/Department of Defense (VA/DoD), and American College of Cardiology/American Heart Association (ACC/AHA).23567847 We reviewed the final manuscript list for completeness and added additional manuscripts identified by the study team and consulted experts in the field.

We defined intensive glycemic control as a treatment strategy used to target HbA1c 6.5-7.0% (48-53 mmol/mol) or below, irrespective of the number and type of drugs used, and conventional glycemic control as targeting HbA1c above 7.0% (53 mmol/mol) but below 8.0-8.5 (64-69 mmol/mol). Risk of bias assessment was done independently and in duplicate using version 2.0 of the Cochrane risk of bias tool for randomized trials included in the review (RoB 2).48

## Early evidence for intensive glycemic control in type 2 diabetes

In epidemiologic studies, chronic hyperglycemia is associated with heightened risk of microvascular and macrovascular complications of diabetes that are important to patients, including end stage kidney disease,49 proliferative retinopathy/blindness,50 clinical neuropathy,11 amputations,51 stroke,52 myocardial infarction,18 heart failure,53 and ASCVD and all cause mortality.54 However, although more than 10 pathophysiologic mechanisms have been attributed to diabetes,55 its diagnosis and monitoring rely almost exclusively on elevated blood glucose concentrations based on its epidemiologic association with retinopathy,56 the main diabetes defining complication. The rational hypothesis that treating to near euglycemia would minimize the risk of complications was therefore pursued from the early 1960s until the 1990s, when it was bolstered by extrapolation of the positive findings of the DCCT.21 The DCCT, conducted among patients with type 1 diabetes, was the first RCT to specifically examine intensive glycemic control by targeting fasting glucose 70-120 mg/dL (3.9-6.7 mmol/L) and postprandial glucose below 180 mg/dL (10 mmol/L) compared with contemporaneous usual care. Positive findings in this trial are often used to rationalize pursuit of intensive glycemic control in type 2 diabetes, despite clear and substantial differences between the two diseases. The first study to specifically examine intensive glycemic control among adults with type 2 diabetes was the Kumamoto study,57 in which 110 non-insulin treated patients with diabetes were randomized to one or two injections of intermediate acting insulin daily with the goal of not showing symptoms of hypoglycemia or hyperglycemia, or to multiple daily insulin injections with a target HbA1c as close to 7% (53 mmol/mol) as possible. Multiple daily insulin injections resulted in 10-25% absolute risk reductions in neuropathy, nephropathy, and retinopathy at six years. The conclusion of the Kumamoto study was that the optimal glycemic thresholds to prevent incident and progressive microvascular disease were HbA1c below 6.5% (48 mmol/mol), fasting glucose below 110 mg/dL (6.1 mmol/L), and two hour postprandial glucose below 180 mg/dL (10 mmol/L).57 Shortly thereafter, the much larger UKPDS trial conducted in 3867 adults with newly diagnosed type 2 diabetes showed that intensive, compared with conventional, glycemic control resulted in 3.2% absolute risk reduction in all diabetes related complications at 10 years.22 These trials, in addition to the DCCT,21 bolstered the paradigm of lowering glucose to near normal concentrations and led to most contemporaneous clinical practice guidelines endorsing a target HbA1c below 7.0% (53 mml/mol) in all patients with diabetes.24252627

Over the past two decades, multiple RCTs have been carried out to better examine the effect of intensive glycemic control on microvascular and macrovascular endpoints and mortality (table 1 and table 2). These have been collated and summarized by 13 systematic reviews and meta-analyses, encompassing 24 RCTs and 37 167 patients, examining the benefits and potential harms of intensive versus conventional glycemic control in patients with type 2 diabetes (fig 1; supplementary table).3031323374757677787980818283 Most range from low risk of bias to some concerns about bias (supplementary figure A).2257585960616263646566676869707172738485 In the sections that follow, we present summary data of the findings as they pertain to microvascular and macrovascular outcomes in type 2 diabetes.

Table 1

Main data from pivotal randomized controlled trials included in systematic reviews with meta-analyses of intensive versus conventional glycemic control in type 2 diabetes

View this table:
Table 2

Main data from pivotal randomized controlled trials included in systematic reviews with meta-analyses of intensive versus conventional glycemic control in type 2 diabetes (continuation)

View this table:
Fig 1

Point estimates and 95% CIs for different outcomes reported in systematic reviews with meta-analyses of intensive v conventional glycemic control in type 2 diabetes. CCG=conventional glycemic control; ICG=intensive glycemic control

### Microvascular outcomes

None of the nine systematic reviews (12 trials comprising 36 179 participants) at moderate to low risk of bias found benefit of intensive glycemic control on microvascular outcomes important to patients, such as end stage kidney disease, clinical neuropathy, or blindness.3074787980818283 Whereas Herrera-Gómez et al showed a 38% relative risk reduction (0.62, 95% confidence interval 0.39 to 0.98) for renal death with lower HbA1c,83 this finding was not consistent with the 1% (0.55 to 1.79) risk reduction for this same outcome previously reported by Coca et al,78 likely because Coca et al included both UKPDS 33 and 34 in their analyses.2258 However, consistent improvement was seen in surrogate microvascular endpoints such as microalbuminuria, macroalbuminuria, and photocoagulation. Specifically, two systematic reviews found a 10-14% relative risk reduction for microalbuminuria,3078 and one (of three) systematic reviews showed a 9-14% relative risk reduction for photocoagulation.308081 Additionally, Coca et al reported a 26% relative reduction in the risk of macroalbuminuria but no significant effect on a more meaningful outcome of doubling of serum creatinine (supplementary table).78

### Macrovascular outcomes

Evidence for macrovascular outcomes comes primarily from nine systematic reviews of 19 moderate to high quality RCTs that included 35 450 participants.303133747576778082 None showed improvement in ASCVD or all cause mortality, stroke, or lower extremity amputations with intensive as opposed to conventional glycemic control. However, a consistent 9-18% relative reduction in the risk of non-fatal myocardial infarction was seen with intensive glycemic control (fig 1; supplementary table).

### Harms

Seven systematic reviews examining 12 trials (33 509 participants) at low to moderate risk of bias consistently reported a twofold to threefold increase in the risk of severe hypoglycemia with intensive glycemic control (fig 1; supplementary table).30317475768082 Other harms, although burdensome to patients and acknowledged by clinicians, were rarely detailed in RCTs or systematic reviews. Montori et al reported a 1-4% absolute increase in weight with intensive compared with conventional glycemic control, and Ray et al reported a weight increase of 2.5 kg (supplementary table). Notably, these trials were conducted in an era when the most commonly used second line glucose lowering drugs (sulfonylureas and insulin) led to weight gain; more recently approved drugs (such as sodium-glucose cotransporter-2 (SGLT-2) inhibitors, glucagon-like peptide-1 receptor (GLP-1R) agonists, and dipeptidyl peptidase-4 (DPP-4) inhibitors) may not cause weight gain but may have other adverse effects. Other adverse sequelae of intensive glycemic control raised by observational studies, including increased burden of treatment, distress due to diabetes, higher costs of care, and higher prevalence of polypharmacy, were not captured by the systematic reviews and meta-analyses or examined in the RCTs.298687

## Cardiovascular outcome trials

ASCVD remains the leading cause of death among people with diabetes88; nearly one third of people with diabetes have established ASCVD and 62% have at least two risk factors for ASCVD.8990 After rosiglitazone was linked to increased risk of cardiovascular events and mortality,91 the US Food and Drug Administration (FDA) mandated, in 2008, that all subsequently approved glucose lowering drugs must demonstrate cardiovascular safety during post-marketing surveillance.92 Since then, 15 trials have been reported for thiazolidinediones, DPP-4 inhibitors, GLP-1R agonists, SGLT-2 inhibitors, and ultra-long acting insulin (table 3 and table 4).100 All RCTs were designed to achieve glycemic equipoise and demonstrate non-inferiority,34353637394243449394979899 and in some cases superiority,459596 with respect to cardiovascular safety compared with either active or placebo controls. Most (22-100%) patients included had a previous cardiovascular event or were at high cardiovascular risk.3435363739424393949596979899 All but one of the CVOTs had an overall low risk of bias (supplementary figure B).

Table 3

Main data from cardiovascular outcome trials in type 2 diabetes

View this table:
Table 4

Main data from cardiovascular outcome trials in type 2 diabetes (continuation)

View this table:

The first CVOTs successfully demonstrated non-inferiority with respect to cardiovascular events and mortality,93949596 but a major breakthrough came in 2015 when the EMPA-REG study showed a significant 14% relative risk reduction (hazard ratio 0.86, 95% confidence interval 0.74 to 0.99; P=0.04 for superiority) in major cardiovascular events (all cause mortality, cardiovascular mortality, non-fatal myocardial infarction, and stroke) with use of the SGLT-2 inhibitor empagliflozin despite minimal relative reduction in HbA1c compared with the comparator arm (7.6% v 8%).42 This finding was reinforced by the subsequent LEADER (liraglutide, a GLP-1R agonist; hazard ratio 0.87 (0.78 to 0.97; P=0.01 for superiority) for the three point major cardiovascular event (MACE) outcome of death from cardiovascular causes, non-fatal myocardial infarction, and non-fatal stroke),35 SUSTAIN-6 (semaglutide, a GLP-1R agonist; hazard ratio 0.74 (0.58 to 0.95; P<0.001 for non-inferiority) for the three point MACE),36 and CANVAS (canagliflozin, an SGLT-2 inhibitor; hazard ratio 0.86 (0.75 to 0.97; P=0.02 for superiority) for the three point MACE) trials. The DECLARE TIMI-55 trial (dapagliflozin, an SGLT-2 inhibitor) failed to show superiority in the primary three point MACE but did show lower rates of hospital admission for heart failure (hazard ratio 0.73, 0.61 to 0.88).39 More recently, the REWIND trial (dulaglutide, a GLP-1R agonist) showed a reduction in the three point MACE (hazard ratio 0.88, 0.79 to 0.99; P=0.026 for superiority).45 In response to these data, the FDA expanded indications for liraglutide, empagliflozin, and canagliflozin to include cardiovascular disease,101 and clinical practice guidelines began to acknowledge the non-glycemic benefits of these drug classes and recommend their use independently of the sole need to lower HbA1c.2102103104

The CVOTs also showed renal benefits with use of GLP-1R agonists and SGLT-2 inhibitors that were similarly independent of glycemic reduction. For GLP-1R agonists, in the ELIXA trial, lixisenatide resulted in a 10% greater percentage change in urinary albumin to creatinine ratio compared with the control group.34 LEADER investigators reported a 22% relative risk reduction (hazard ratio 0.78, 0.67 to 0.92) in the risk of a composite renal outcome (new onset macroalbuminuria, doubling of serum creatinine, estimated glomerular filtration rate (eGFR) ≤45 mL/min/m2, need for renal replacement therapy, or renal death) in the liraglutide arm.35 The SUSTAIN-6 trial (semaglutide) reported a hazard ratio of 0.64 (0.46 to 0.88) for a composite of macroalbuminuria, doubling of serum creatinine, eGFR 45 mL/min/1.73 m2 or lower, or need for dialysis.36 Dulaglutide, in the REWIND trial, showed 15% relative reduction (hazard ratio 0.85, 0.77 to 0.93) in the risk of new macroalbuminuria, sustained decline in eGFR of at least 30%, or need for chronic renal replacement therapy.41 Similarly, for SGLT-2 inhibitors, the CANVAS trial of canagliflozin reported reductions in the progression of albuminuria (hazard ratio 0.73, 0.67 to 0.79) and in the composite of at least 40% reduction in eGFR, need for renal replacement therapy, or renal death (0.60, 0.47 to 0.77).37 The DECLARE-TIMI 58 trial of dapagliflozin reported a 24% relative reduction (hazard ratio 0.76, 0.67 to 0.87) in the composite risk of at least 40% reduction in eGFR, eGFR 60 mL/min/1.73 m2 or lower, dialysis, or death due to renal causes.39

Because these trials were specifically designed to test the cardiovascular safety of new glucose lowering drugs, with similar HbA1c levels achieved in the intervention and comparator arms by study design,100 variation in study endpoints could be attributed to the study drug, rather than the presumed benefit of HbA1c reduction.100105 Moreover, mean HbA1c achieved in the intervention arms of CVOTs with demonstrable cardiovascular and/or renal benefits was consistently greater than 7.0% (53 mmol/mol), reinforcing the benefits of moderate glycemic control.

## Interpretation of the evidence

The body of evidence stemming from systematic reviews and meta-analyses of RCTs (all at moderate to low risk of bias) suggests no demonstrable benefit of intensive (HbA1c <7.0%) as opposed to moderate (HbA1c 7-8.5%) glycemic control on microvascular (for example, end stage kidney disease, renal death, blindness, clinical neuropathy) and macrovascular (for example, all cause mortality, cardiovascular mortality, non-fatal stroke, amputations) outcomes important to patients, with the exception of a 9-18% relative reduction in the risk of non-fatal myocardial infarction.303133747576778082 Although reductions in surrogate microvascular endpoints were more consistent (10-14%, 26%, and 10-20% relative risk reductions for microalbuminuria, macroalbuminuria, and photocoagulation, respectively),307881 these outcomes are not as meaningful to patients and do not necessarily translate to downstream improvements in their corresponding hard endpoints.106 At the same time, intensive glycemic control—or, more specifically, the aggressive glucose lowering treatment that is often needed to achieve low HbA1c targets—results in a twofold to threefold (200-300% relative risk) increase in the risk of severe hypoglycemia. Intensive glucose lowering treatment has additional harms, including increased risk of adverse drug reactions, polypharmacy, treatment burden, and high costs of care,298687 but these adverse effects cannot be adequately captured by RCTs and have not been thoroughly examined in observational real world studies.

Most,467103104 but not all,8 guidelines project a high degree of confidence related to the benefits of intensive glycemic control on microvascular outcomes and low confidence related to macrovascular outcomes, when the evidence shows the opposite. Furthermore, results of the CVOTs cannot be used to justify intensive glycemic control because they did not examine it; instead, they showed that it is the means of controlling hyperglycemia to achieve moderate HbA1c targets—not intensive control to achieve low HbA1c targets—that may ultimately improve hard microvascular (renal) and macrovascular (ASCVD, heart failure) outcomes.

The use of composite outcomes may have contributed to blurred interpretations of the treat-to-target trials and the presumption of meaningful benefit from intensive glycemic control.107108 For instance, the primary endpoint in the UKPDS trial—a 14 component indicator of “any diabetes related complications”—was reduced by 3.2% (absolute rates of 35.2% in the intensive group versus 38.4% in the conventional group) after 10 years.22 However, almost the entirety of this risk reduction was driven by lower rates of retinal photocoagulation and cataract extraction (approximately 2.7% absolute risk reduction), with negligible effects on endpoints of far greater importance such as mortality, stroke, blindness, amputation, or end stage kidney disease.2232 Similarly, the ADVANCE trial reported a 10% relative risk reduction (1.9% absolute risk reduction) in the primary outcome of combined microvascular and macrovascular events with intensive glycemic control; however, no difference was seen in the individual outcomes of non-fatal stroke, non-fatal myocardial infarction, death from cardiovascular causes, or new or worsening retinopathy.65 In ADVANCE, the 30% relative reduction in the risk of macroalbuminuria (hazard ratio 0.70, 0.57 to 0.85) was the only endpoint driving positive results for the composite microvascular disease endpoint (no effect on visual deterioration or new/worsening neuropathy) and the new or worsening nephropathy endpoint (no effect on doubling of the serum creatinine, renal replacement therapy, or renal death).65 The ADVANCE trial showed a 65% reduction in the risk of end stage kidney disease, but this was based on a small absolute number of events in both groups (20 v 7 events; 0.3% absolute risk reduction; number needed to treat 333 for 11 years), making this finding susceptible to statistical inference fragility.109 ACCORD researchers reported benefits in the microvascular disease composite endpoint of microalbuminuria and macroalbuminuria (19% and 32% relative risk reductions, respectively), three line worsened visual acuity (9% relative risk reduction), cataract extraction (11% relative risk reduction), score of more than 2.0 on the Michigan Neuropathy Screening Instrument (8% relative risk reduction), and lost ankle jerk and light touch (10% and 15% relative risk reduction, respectively). However, no reduction was seen in end stage kidney disease or severe vision loss.110 The VADT also reported a significant reduction in the risk of progression to albuminuria (4.7% absolute risk reduction; number needed to treat 21 for 5.6 years) but no effect on cataract surgery, photocoagulation, vitrectomy, retinopathy, creatinine, end stage kidney disease, mononeuropathy, or peripheral or autonomic neuropathy.67 Thus, a positive result on a composite endpoint that stems from benefits in surrogate disease endpoints is misleading if misperceived to suggest that all components of the composite, most notably hard endpoints important to patients, had shared in the benefit.108

Post-RCT observational extensions of the large treat-to-target trials have also shown mixed effects on hard microvascular and macrovascular outcomes. The UKPDS follow-up study reported, after a mean follow-up of 16.8 years, 17%, 13%, and 15% relative reductions in the risks of diabetes related death, all cause mortality, and myocardial infarction, respectively.66 However, these data are to be interpreted with caution, as results are subject to confounding by events that occurred during observation, particularly as HbA1c levels in the two groups converged throughout the follow-up period. These findings were also not reproduced by the other extension studies.71727385 Although the 10 year VADT follow-up study suggested a small benefit on the primary composite outcome of myocardial infarction, stroke, new or worsening heart failure, amputation for ischemic gangrene, or cardiovascular related death (hazard ratio 0.83, 0.70 to 0.99), no reduction was seen in cardiovascular or all cause mortality.72 Moreover, this benefit was not apparent in the recently reported 15 year follow-up analysis (hazard ratio 0.91, 0.78 to 1.06).73

In contrast to inconsistent and inconclusive data on the benefits of intensive (compared with conventional) glycemic control, both RCTs and observational studies consistently show a significant increase in severe hypoglycemia. Severe hypoglycemia is strongly associated with increased risk of death, cardiovascular events, cognitive decline, falls and fall related fractures, fear of hypoglycemia, and reduced quality of life.111112113114115116117118119120 The incidence of emergency department visits and hospital admissions for severe hypoglycemia is estimated at 0.2-2.0 per 100 person years,121 although these events are a small fraction of all severe hypoglycemic events experienced by people with diabetes.122 The true prevalence of clinically meaningful hypoglycemia remains unknown, as no standardized or comprehensive mechanism is available to ascertain it.121 Intensive glucose lowering treatment has other important harms, which need to be acknowledged and better understood.

## Guidelines

Multiple clinical practice guidelines make recommendations on glycemic control in adult patients with type 2 diabetes (table 5).23567847 All (with the exception of the ACP position statement), advocate targeting HbA1c below 7.0% (53 mmol/mol) for most adults with diabetes. The AACE goes further and advocates HbA1c less than 6.5% (48 mmol/mol) as the optimal goal for most patients. Although treatment goals remain largely informed by HbA1c, guidelines also acknowledge the implications of CVOTs for both choice of drug treatment and non-glycemic considerations for pursuing treatment.

Table 5

Position of different professional organizations regarding glycemic control

View this table:

The ADA guidelines and ADA/EASD consensus for the management of hyperglycemia in patients with type 2 diabetes also specify that the main goals of diabetes treatment are not simply control of glucose, blood pressure, and cholesterol, but rather the prevention or delay of complications and maintenance of patients’ health related quality of life. They further recommend using shared decision making (SDM) as a means of achieving patient centered care and reinforce the centrality of the patient in the decision making process. However, the guidelines fall short of specifying how and with what tools SDM is achieved and how it is to be reconciled with the pursuit of clearly defined surrogate endpoints of HbA1c, blood pressure, and lipid management.

Moreover, guidelines increasingly acknowledge the importance of the patient’s overall health, functional status, and life expectancy, as well as social determinants of health, access to healthcare resources, and social support for diabetes management.236 However, these factors are introduced primarily as potential justifications for tolerating suboptimal care with relaxed HbA1c targets and preferential use of inexpensive, but presumably clinically suboptimal, glucose lowering drugs. Data suggest that relaxed glycemic targets and simplified treatment regimens may be the safest and clinically optimal for patients with limited life expectancy, multiple comorbidities, and high risk of hypoglycemia. Moreover, although awareness of the patient’s clinical and psychosocial context is a necessary first step toward patient centered diabetes care, clinicians and patients would truly benefit from specific guidance on how to reduce, not merely acknowledge, these barriers to care and health.

## Person centered care

Most people with diabetes have multiple acute and chronic health conditions, personal and professional obligations, and financial and social constraints.123124125 Holistic patient centered, as opposed to disease centered, care needs to consider each person’s unique situation, both clinical and psychosocial, and carefully weigh the balance of benefits and harms of glucose lowering treatment in the broader context of their lives.125126 For example, Mrs J, our 67 year old patient who has been living with type 2 diabetes for 15 years, has at first glance poorly controlled disease. Her HbA1c is 8.7% (72 mmol/mol) and she smokes three cigarettes a day, has a body mass index of 34.2, blood pressure of 145/90 mm Hg, and low density lipoprotein cholesterol of 130 mg/dL (3.36 mmol/L). However, she works hard to manage her diabetes and achieve these goals. She tries to walk five times a week, cook healthy food, and take her 15 daily drugs, including metformin, SGLT-2 inhibitor, DPP-4 inhibitor, long acting insulin, statin, aspirin, thiazide diuretic, angiotensin converting enzyme inhibitor, multiple non-steroidal anti-inflammatory drugs, inhaled long acting β2 adrenergic agonist/corticosteroid, selective serotonin reuptake inhibitor, benzodiazepine, melatonin, and vitamin D3. What hides beneath the surface is that Mrs J also has osteoarthritis with chronic low back and knee pain, bothersome lower extremity neuropathy, chronic obstructive pulmonary disease, mild visual impairment, insomnia, and depression. She is a caregiver to her husband, who recently had an ischemic stroke, and continues to work full time to support herself, her husband, their two children, and their grandchildren. Despite her best efforts, none of her comorbidities is optimized, she has frequent non-severe hypoglycemia, and she feels overwhelmed by her illnesses, her drug treatments, and life’s demands.

One of the major challenges clinicians face today is how to effectively and efficiently implement the best available evidence for the individual patient, given the time and resource constraints of clinical practice.127 This is particularly pertinent to diabetes, for which the benefits of intensive glycemic control may be outweighed by the potential harms, and which occurs in the context of other clinical, psychosocial, and systemic barriers.126 Although clinical practice guidelines are increasingly acknowledging multimorbidity and the importance of tailoring treatment plans for the individual patient, they often fall short of providing specific and actionable guidance on how this is to be achieved.128 A person centered treatment plan needs to be both evidence based and context specific, taking into consideration patients’ clinical and psychosocial complexities, as well as their values, preferences, goals, and life expectations that stem from their unique family, social, and economic situation.129130131

## Models of care

We propose three constructs that can facilitate and support goal and context concordant care: SDM, minimally disruptive medicine (MDM), and the chronic care model (CCM).

### Shared decision making

SDM is a process for clinicians, patients, and caregivers to partner in the development of the most appropriate and feasible therapeutic plan at a given time based on expert input from both the clinician (grounded in the best available scientific evidence) and the patient/caregiver (providing expertise in the patient’s goals, preferences, and context of care).132 SDM is facilitated by decision aids for diabetes management,133 conversational elicitation of the patient’s goals/preferences,134 and/or integrated, holistic clinical decision support tools.135136137 Use of decision aids, in particular, allows therapeutic goals and treatment options to be discussed and considered in the context of the patient’s preferences. For example, the Diabetes Medication Choice Decision Aid (available at http://diabetesdecisionaid.mayoclinic.org) first elicits from patients the aspects of diabetes care that are most important to them (HbA1c reduction, daily routine complexity, hypoglycemia, weight gain, cardiovascular benefits, or cost) and then guides patients and clinicians to choose from among the available drug treatments those that best serve their identified needs (fig 2).138 This decision aid has been used in a variety of clinical contexts, and trials have shown that decision aids can effectively and efficiently facilitate SDM without increasing the burden on clinicians or systems.139 Decision support tools can also be designed to be holistic (that is, not focused on glycemic reduction alone) and adapted to facilitate and inform evidence based SDM,135140 particularly in primary care where most people with type 2 diabetes receive their care.

Fig 2

Diabetes Medication Choice Decision Aid

Implementation of health electronic records based clinical decision support has been shown to improve diabetes management in a variety of clinical settings.135

### Minimally disruptive medicine and the chronic care model

Diabetes is a complex and multifaceted disease, and multidisciplinary team based care founded on the CCM can help patients to receive the whole spectrum of care they need in a way that is accessible and meaningful to them.141142143 The goal of CCM is to achieve MDM, in which the patient’s goals for life and health are prioritized while minimizing the burden of treatment and disease and maximizing the patient’s capacity to enact the work of being a patient.144145 Decision support tools can facilitate implementation of CCM. For example, the Instrument for Patient Capacity Assessment (ICAN) tool (available at https://minimallydisruptivemedicine.org/ican/) helps patients and clinicians to appraise how the healthcare program fits within the constructs of the patient’s life with the ultimate goal of creating a therapeutic plan that minimizes treatment burden and maximizes medical and psychosocial capacity. ICAN is intended to shift the emphasis from the illness to the broader situation by eliciting patients’ personal values and preferences, exploring how the healthcare system assists or hinders them, and cultivating opportunities to advance each unique patient’s context and situation (fig 3).146

Fig 3

ICAN (Instrument for Patient Capacity Assessment) discussion aid

Importantly, these conversations need to take place in the context of scientific evidence pertaining to diabetes management and health outcomes, with clinicians facilitating therapeutic decisions that best achieve the goals identified by patients in ways that are least disruptive, and hence most achievable, to the patient’s life. In the case of glycemic control in type 2 diabetes, this means identifying the therapeutic goals and outcomes that are important to patients; eliciting their capacity for diabetes management and self care; and weighing the best available evidence about optimal glycemic targets, intensity of treatment (intensive versus moderate versus relaxed glycemic control), and specific treatment regiments in the context of this information (fig 4).

Fig 4

Algorithm for therapeutic decision making. Not all clinical encounters should include everything, and nor should they start always from step 1. Much value should be placed on the continuity of the enduring patient-clinician relationship over time

In the example of Mrs J, a long and meaningful dialog revealed the presence of distress due to diabetes and major depressive disorder. She began to work with a psychiatrist and six months later successfully discontinued seven drugs, including insulin; her HbA1c improved to 62 mmol/mol (7.8%), she achieved better blood pressure and cholesterol control, and hypoglycemic episodes resolved. Importantly, this approach does not diminish the importance of diabetes management and control of hyperglycemia. Instead, the case of Mrs J illustrates that pursuit of moderate glycemic control, which, in contrast to intensive glycemic control, has been shown to improve health outcomes important to patients with type 2 diabetes, and prioritization of non-glycemic aspects of her life that were identified as most important to her, may ultimately yield the best achievable overall health outcomes.131

## Knowledge gaps and future research

Although controlling hyperglycemia is an essential component of diabetes management, emerging evidence has challenged the paradigm of pursuing near normal glucose concentrations and relying on HbA1c as a central indicator of quality of diabetes care.106 Other aspects of diabetes management and control may be more important, including real time glucose concentrations (that is, avoidance of symptomatic hypoglycemia and hyperglycemia), specific glucose lowering drugs being used, burden of treatment, and quality of life. Moreover, therapeutic goals need to be individualized and contextualized to fit each patient’s clinical and personal situation, balancing the anticipated benefits of achieving the specific level of glycemic control with the burden and potential harms of pursuing it. Evidence from treat-to-target RCTs, CVOTs, and epidemiologic studies of type 2 diabetes suggests that HbA1c levels between 7% and 8% (53-63 mmol/mol) are appropriate for most non-pregnant adults with type 2 diabetes, although this target needs to be weighed against the actual glucose values it represents (that is, absence of symptomatic hypoglycemia and hyperglycemia) and the treatment modality used to achieve it. Younger, healthier people with diabetes may benefit from more intensive glycemic control, but additional trials using contemporaneous treatment modalities are needed to confirm this. Conversely, patients with limited life expectancy, frailty, or high clinical complexity may benefit more from a conservative treatment strategy that prioritizes avoidance of symptomatic dysglycemia, although operationalizing this is challenging in the setting of multimorbidity and potentially varied manifestations of dysglycemia in this population.

Moreover, long term complications of type 2 diabetes are incurred not only through hyperglycemia but also through hypertension, dyslipidemia, obesity, smoking, and other components of metabolic dysfunction. As new treatment modalities emerge for all these conditions, additional studies are needed to determine optimal therapeutic regimens and targets for these risk factors as well.

## Emerging therapies

Despite lack of conclusive evidence on optimal glycemic targets, no treat-to-target RCTs for type 2 (or type 1) diabetes are ongoing or planned. Instead, we anticipate continued post-marketing examination of recently approved therapeutic classes with semaglutide (SOUL) and ertugliflozin (VERTIS) trials yet to be completed.147148 With multiple available classes of glucose lowering drugs, interventional and quasi-experimental comparative effectiveness and safety trials are needed to determine therapeutic approaches that are optimally aligned with patients’ risk profiles and goals of treatment. This includes the ongoing pragmatic unmasked GRADE trial, which may not be as clinically informative as is necessary given the absence of SGLT-2 inhibitors from the treatment arms and the primary focus on HbA1c, not on microvascular or macrovascular outcomes important to patients.149

Diabetes technologies are also used increasingly by patients with type 2 diabetes, albeit at much lower rates than those with type 1 diabetes. This includes flash and real time continuous glucose monitors, insulin pumps, smart insulin pens, and mobile health platforms. Emerging, albeit inconsistent and weak, evidence suggests lower rates of hypoglycemia, improved glycemic control, and better quality of life with some of these devices.150 Finally, increasing focus on population health management and team based care, as well as rapid improvements in health information technology and data science, have enabled new approaches to diabetes management and care. How these approaches improve access to care, quality of delivered care, and ultimately patients’ health outcomes remains to be determined.151152153

## Conclusions

Evidence suggests that moderate glycemic targets, with HbA1c levels between 7% and 8% (53-63 mmol/mol), are adequate for most people with type 2 diabetes if this is achieved without symptomatic hypoglycemia or hyperglycemia and unless lower targets are easily achievable without treatment burden or adverse effects. Glucose lowering regimens and targets need to be individualized to reflect each patient’s balance of anticipated benefit versus potential harm and to align with their goals, preferences, and situation. Individualization of care can be achieved through the process of SDM, whereby clinicians and patients partner to identify an evidence based and minimally disruptive therapeutic approach that is right for the patient at the particular time. Ultimately, the body of evidence stemming from treat-to-target RCTs, CVOTs, and epidemiologic studies of diabetes management calls for a paradigm shift in diabetes management to recalibrate the notion of presumed benefits from intensive glycemic control and refocus on outcomes that matter to patients, ensuring that the care we deliver is evidence based, safe, effective, equitable, and accessible to all.

### Glossary of abbreviations

AACE—American Association of Clinical Endocrinologists

ACC—American College of Cardiology

ACE—American College of Endocrinology

ACP—American College of Physicians

AHA—American Heart Association

ASCVD—atherosclerotic cardiovascular disease

CCM—chronic care model

CVOT—cardiovascular outcome trial

DCCT—Diabetes Control and Complications Trial

DPP-4—dipeptidyl peptidase-4

EASD—European Association for the Study of Diabetes

eGFR—estimated glomerular filtration rate

FDA—US Food and Drug Administration

GLP-1R—glucagon-like peptide-1 receptor

ICAN—Instrument for Patient Capacity Assessment

MACE—major cardiovascular event

MDM—minimally disruptive medicine

NICE—National Institute for Health and Care Excellence

RCT—randomized controlled trial

SDM—shared decision making

SGLT-2—sodium-glucose cotransporter-2

UKPDS—United Kingdom Prospective Diabetes Study

VA/DoD—US Department of Veterans Affairs/Department of Defense

### Research questions

• What aspects of diabetes risk factor control (HbA1c, real time glucose, blood pressure, cholesterol, smoking status) and/or therapeutic management (use of particular glucose lowering, lipid lowering, and/or blood pressure lowering drugs) translate to meaningful improvements in microvascular and macrovascular outcomes?

• How can person centered diabetes care be meaningfully implemented in the context of systemic constraints on clinicians and health systems?

• How can patients and caregivers be empowered and supported as they manage diabetes, its complications, and related comorbidities?

• How can the social determinants of health that predispose to the incidence and progression of diabetes, and hinder its optimal management, be identified and corrected?

## Acknowledgments

We thank Victor M Montori and Edgar Gerardo Dorsey-Treviño for their perspectives and insights to strengthen this manuscript.

## Footnotes

• Series explanation: State of the Art Reviews are commissioned on the basis of their relevance to academics and specialists in the US and internationally. For this reason they are written predominantly by US authors

• Contributors: RRG, JGGG, RGM, and JAZH reviewed and provided valuable information, insight, and edition to the manuscript. All authors have agreed on the final version of the manuscript. RRG is the guarantor.

• Funding: RGM is supported by the Mayo Clinic Robert D and Patricia E Kern Center for the Science of Health Care Delivery and by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health under award number K23DK114497. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

• Competing interests: We have read and understood the BMJ policy on declaration of interests and declare the following interests: none.

• Provenance and peer review: Commissioned; externally peer reviewed.

• Patient involvement: Mrs J is a patient in the Diabetes Clinic of the Endocrinology Division at the University Hospital “Dr José E Gonzalez” of the Universidad Autonoma de Nuevo Leon. On being invited to participate in this review, Mrs J provided informed consent to share her personal experience as a person living with type 2 diabetes, which framed our discussion of person centered care.

View Abstract