Hypoglycemia Evaluation and Reporting in Diabetes: Importance for the Development of New Therapies

Short Running title: Hypoglycemia evaluation in clinical trials

David C. Klonoff MD,1 G Alexander Fleming MD,2 Douglas B Muchmore MD,2 Brian M Frier MD3


1  Diabetes Research Institute; Mills-Peninsula Health Services San Mateo, California, USA

2  Kinexum, Harpers Ferry, West Virginia, USA

3  The Queen's Medical Research Institute, University of Edinburgh, Edinburgh, UK

Corresponding Author: 

David C. Klonoff, Diabetes Research Institute; Mills-Peninsula Health Services San Mateo,

California, USA, 100 South San Mateo Drive, Room 5147, San Mateo, California 94401,  Phone: 650-696-4262, This email address is being protected from spambots. You need JavaScript enabled to view it.

Word Count: 3900

Number of Tables/Figures: 2 tables, 1 figure

Keywords:  Diabetes, diabetes therapy, glucose-lowering therapy, Hypoglycemia, glucose monitoring, regulatory efficacy endpoint


Abbreviations: American Diabetes Association (ADA), continuous glucose monitor (CGM), Diabetes Control and Complications Trial (DCCT), European Medicines Agency (EMA, previously known as EMEA), Endocrine Society (ES), Food and Drug Administration (FDA), Hemoglobin A1c (A1C), major adverse cardiovascular events (MACE), selfmonitored blood glucose (SMBG), severe hypoglycemia (SH), type 1 diabetes (T1D), type 2 diabetes (T2D), United Kingdom Prospective Diabetes Study (UKPDS)



Hypoglycemia complicating diabetes therapy is well recognized to be an ever-present threat to patients, their families, providers, payers and regulators.  Despite this being widely acknowledged, the regulatory stance on hypoglycemia as an endpoint in clinical trials to support new product registration has not evolved in any meaningful way since the publication of a position paper by an American Diabetes Association (ADA) Workgroup in 2005.  As the impact of hypoglycemia on persons affected by diabetes is of major importance when assessing new treatments, the historical position of regulatory agencies on hypoglycemia is reviewed with respect to product approvals.  The purpose of this article is to present proposals for facilitating development of therapies that reduce hypoglycemia risk though: 1) development of composite measures of benefit for regulatory endpoints; and 2) facilitation of the fulfillment of an unmet clinical need for reducing hypoglycemia.  In view of greater comprehension of the effects of hypoglycemia, coupled with improved methodology to assess its frequency, the authors recommend: (1) a numerical cut point of < 54 mg/dl (<3.0 mmol/L)  as a clinically relevant level with which to define meaningful hypoglycemia for trials of diabetes therapies; (2) utilization in clinical trials of mature glucose monitoring technologies for purposes of regulatory evaluation and clinical decision-making; and (3) development of primary efficacy endpoint composites that include hypoglycemia rates and glycemic control.  



Effective treatment of diabetes is challenging. The importance of strict glycemic control, as measured by hemoglobin A1c (A1C), to minimize the risk of microvascular complications in both type 1 diabetes (T1D) and type 2 diabetes (T2D) was irrefutably confirmed by the

Diabetes Control and Complications Trial (DCCT) and the United Kingdom Prospective Diabetes Study (UKPDS). [1] Thereafter, glucose lowering became the sine qua non for assessing diabetes therapies, despite the fact that aggressive glycemic control may significantly increase the risk of hypoglycemia.  It is our belief that hypoglycemia risk reduction has received insufficient attention to date from both clinical and regulatory viewpoints. In particular, we believe that hypoglycemia risk should be considered both for clinical and regulatory purposes when assessing the value of any new therapy.  We therefore assert that it is time to revise the approaches for evaluation and reduction of hypoglycemia risk, with particular attention being given to defining methods of ascertaining the risk for hypoglycemia posed by therapies for diabetes.  A balanced approach for the evaluation of new treatments is required, considering both the benefit (A1C reduction) and risk

(hypoglycemia) of glucose-targeted interventions.


Scope of the Problem

Hypoglycemia as a consequence of insulin and other glucose-lowering therapies is the principal limiting factor that prevents attainment of glycemic targets in both T1D and T2D.[2]  Each year, around one third of people with type 1 diabetes experience one or more episodes of severe hypoglycemia, defined as an event that requires external assistance for recovery.[3]  In addition to being potentially life-threatening, severe hypoglycemia is associated with cardiovascular events, cognitive impairment, seizures, fall-related fractures, and driving mishaps.   Hypoglycemia is also associated with reduced quality of life and depression, potentially leading to disruption of daily activities, a negative impact on treatment adherence, and poor dietary control.[4] An important component of the reduction in quality of life associated with hypoglycemia is fear of hypoglycemia.[5]  Fear of hypoglycemia contributes to therapeutic inertia and resistance to intensification of therapy, resulting in poorer glycemic control. [5,6] Beyond impact on quality of life, it limits employment opportunities and affects work productivity.[7]   Based on four series, in type 1 diabetes 410% of deaths are attributed to acute hypoglycemia.[8] These adverse consequences combine to make hypoglycemia the most important obstacle to achieving good overall glycemic control.[9] The substantial economic costs of hypoglycemia accrue to the patient, the healthcare system, and to society as a whole.[10]  Drugs or devices that can limit the incidence of hypoglycemia of all severities are greatly needed. 

Although hypoglycemia is often thought to be a consequence of overly aggressive glycemic targets, recent evidence suggests that the incidence of severe hypoglycemia in type 1 diabetes may bear little or no direct relationship to overall glycemic control as reflected by A1C values.[11] Severe hypoglycemia in adults aged 60 years or more is strongly associated with other risk factors including impaired awareness of hypoglycemia and increased glucose variability.[12]  This lack of association of severe hypoglycemia with mean glycemic control underscores the importance of recognizing hypoglycemia as an independent endpoint in evaluation of glucose-lowering therapies. 

How is the Presence of Hypoglycemia determined?

Hypoglycemia is identified using internal cues (symptoms), external cues (based on experience, knowledge and recognition of situations imposing risk), feedback from observers such as family members, and by measuring blood glucose.  Symptoms of hypoglycemia are protean, and unfortunately are not reliable indicators of low blood glucose.  The two major groups of symptoms are autonomic, including, but not limited to, pounding heart, tremor, hunger and sweating, and neuroglycopenic, including odd behavior, speech difficulty, confusion, drowsiness and incoordination; common non-specific symptoms include nausea and headache.[13]  Symptoms of hypoglycemia are subjective and idiosyncratic and are unreliable in determining the actual presence of hypoglycemia in people with diabetes because their perception is subject to modifying factors such as posture, medications, distraction, and increasing age.[14)] Symptoms vary between hypoglycemic events within the same individual [15], and the autonomic symptoms of hypoglycemia diminish with progressive duration of T1D such that neuroglycopenic symptoms eventually predominate.[16]  Sometimes symptoms are ignored, especially if the individual is distracted, or they may be subtle, absent or even misinterpreted in some situations in people who otherwise have intact hypoglycemia awareness.[14, 17] Repeated exposure to hypoglycemia leads to the development of hypoglycemia-associated syndromes[18], including impaired awareness of hypoglycemia, the presence of which increases the risk of subsequent severe hypoglycemia by about 6 fold.[19]  People with both T1D and T2D who have impaired awareness of hypoglycemia demonstrate evidence of much more biochemical hypoglycemia with routine blood glucose monitoring,[20, 21] although the incidence of low values depends on how frequently testing is undertaken. 

In the classification of hypoglycemia, the absence of symptoms in the presence of an identified low blood glucose is acknowledged as “biochemical hypoglycemia.”[22]   The various manifestations of hypoglycemia have different implications for patients, healthcare professionals, regulatory authorities, clinical researchers and payers.  It is important to explore how hypoglycemia is defined, measured and recorded for each set of stakeholders.  In particular, advances in technologies for blood glucose measurement now afford the opportunity to expand the means of ascertaining hypoglycemia.  While symptoms of hypoglycemia may be informative when they occur, and are perceived and recognized as such by the patient, their absence does not exclude the presence of hypoglycemia and is not therefore a reliable measure.  


Definition of Hypoglycemia

 By describing severe hypoglycemia as any episode requiring external assistance for treatment [23] the DCCT investigators helped to promote the use of this definition internationally, both for clinical use and within trial settings.  However, while this definition of severe hypoglycemia has been widely accepted, definitions of other forms of hypoglycemia have varied considerably between studies and have complicated or prevented meta-analyses.  Since then, several position papers and guidance documents have been published in efforts to unify diagnostic approaches to hypoglycemia.  Taking an early lead in these efforts, the American Diabetes Association (ADA) Workgroup on Hypoglycemia published their recommendations in 2005.[22] Based on the observation in healthy nondiabetic adult volunteers that a rise in plasma concentrations of some counterregulatory hormones could first be detected at a threshold of ~68 mg/dl (3.8 mmol/L) when hypoglycemia was induced in the (unphysiological) setting of a hyperinsulinemic glucose clamp, the ADA adopted <70 mg/dl (<3.9 mmol/L) as the level for defining biochemical hypoglycemia (irrespective of the presence of symptoms).  Despite considerable discussion as to the validity and appropriateness of this definition, most advisory groups and regulatory authorities currently recognize and use this definition.  While it is generally accepted that this very conservative level is an appropriate “alert value” to trigger action by individual patients (9), different glycemic threshold values may be more appropriate for evaluating diabetes therapies in the setting of clinical trials. 


Regulatory aspects of Hypoglycemia

The European Medicines Agency (EMA) and the Food and Drug Administration (FDA) in the USA have similar approaches for evaluating hypoglycemia.[24,25] The FDA Guidance for Industry for Diabetes Mellitus: Developing Drugs and Therapeutic Biologics for Treatment and Prevention, published in February 2008, specifies that demonstration of efficacy should be based on reduction of A1C.The FDA also stated that anti-diabetes drugs should be assessed for their tendency to cause or augment hypoglycemia and should yield acceptable hypoglycemic risk (as yet undefined) with comparable A1C outcomes as specified by the ADA Workgroup.[22] To date, a lower risk of hypoglycemia as a primary efficacy endpoint has not been used in any pivotal trial submitted to FDA as part of a new drug application.  This is because the accepted regulatory criterion for defining hypoglycemia as an endpoint in clinical trials is based on the DCCT definition of severe hypoglycemia (SH) as an event that requires the assistance of another person to treat it.  As SH is a relatively infrequent event, large study populations are required to evaluate this serious degree of hypoglycemia.[26,27]

Current Regulatory Practices for Evaluating Hypoglycemia

The EMA and FDA have been increasingly transparent in revealing their underlying thinking and the documentary reviews that are associated with their evaluation of therapies, including those for diabetes.  While hypoglycemia is always a major focus of regulatory review for glucose-lowering therapies, the FDA-approved product information for glucose-lowering drugs consistently focuses on hypoglycemia as a safety concern and not as a comparative performance measure.  The prescribing information for the most recently FDA-approved new insulin analog, insulin degludec (Tresiba®, Novo Nordisk) illustrates this practice.  As is the case for other insulin products, hypoglycemia features prominently in both the Warning and Precaution sections and in the Adverse Reactions section. [28]   Tables are included that summarize the primary efficacy endpoint (A1C) and change in insulin dose results, but no hypoglycemia rates are reported.  Likewise, the Toujeo® prescribing information, which is the recently approved new formulation of insulin glargine (Lantus, Sanofi), is similarly constructed.  Fewer studies were performed in the Toujeo program than in the Tresiba program since Toujeo does not contain a new active molecule.  Even though the summarized trials involved comparisons of a single sponsor’s two products, comparative hypoglycemia data are not provided.[31] 


The Tresiba Product Information that was approved in 2013 by EMA [32] strikingly contrasts with FDA’s version by including data that support a hypoglycemia benefit. Not only are comparative hypoglycemia rates provided in tables in the EMA version, but a paragraph is included that states a pre-specified meta-analysis demonstrated that “Tresiba was superior in terms of a lower number of treatment emergent confirmed hypoglycaemic episodes (driven by a benefit in type 2 diabetes mellitus, see table 2) and nocturnal confirmed hypoglycaemic episodes compared to insulin glargine.”  

At least two factors account for the lack of comparative information of hypoglycemia rates in the FDA label of this and other insulin products.  First, FDA requires that comparative claims be supported by substantial evidence.[33,34] FDA generally interprets substantial evidence to consist of positive results from two well-controlled studies, though a single trial may be sufficient in exceptional circumstances.[35] Typically, a pre-specified meta-analysis, even if it was positive, would not be sufficiently convincing to meet this expectation.  However, the FDA reviewers carefully considered the evidence for a benefit in terms of lower hypoglycemia rates in its review of Tresiba, and asked its advisory committee to interpret the clinical implications and value of this evidence.  FDA reviewers even asked the committee, in effect, if the possible hypoglycemia benefit could offset the safety concern resulting from an imbalance across the clinical program of major adverse cardiovascular events (MACE) favoring placebo.[36] This concern ultimately led to a 2-year delay in the approval of the Tresiba New Drug Application until an interim analysis of a cardiovascular safety trial became available.[37] 


Although FDA reviewers concluded that the evidence for a hypoglycemia benefit was not sufficiently consistent across studies, they expressed concern about the methodology used to establish a benefit on nocturnal hypoglycemia in T1D patients with CGM. FDA reviewers questioned whether the observed benefit of less frequent nocturnal hypoglycemia with Tresiba resulted from timing of CGM measurement that favored this insulin.[36] In summary, several factors led the EMA and the FDA to come to different conclusions and actions about the Tresiba hypoglycemia data, including the lack of standardization for classifying hypoglycemia events and analyzing hypoglycemia data.  


ADA Workgroup on Hypoglycemia and the ADA/Endocrine Society Article on


In 2005 the Workgroup concluded “any significant reduction in severe hypoglycemia (that requiring the assistance of another individual) even by as little as 10-20%, would be advantageous” as well as “a significant reduction in the frequency of documented hypoglycemia (plasma glucose ≤70 mg/dl [≤3.9 mmol/L]), with or without symptoms, of 30% by a new drug, device, or management strategy would represent a clinically important improvement over existing therapies.”[22] (Table 1) In 2013, the ADA and The Endocrine Society (ES) reconfirmed the previous definitions of hypoglycemia in diabetes and reviewed the implications of hypoglycemia on both short- and long-term outcomes[8]   This ADA/ES article analyzed the effects of hypoglycemia on mortality in three large outcome trials[38,39,40] that assessed the effect of glucose-lowering on cardiovascular events in patients with type 2 diabetes and concluded that severe hypoglycemia was clearly associated with an increased risk of mortality.  However, the ADA/ES report of 2013 did not revisit the recommendations from 2005 about the regulatory definitions and margins of improvement that should be considered before granting approval for a new product that was claimed to promote a lower risk of hypoglycemia. In the absence of any newer consensus recommendations on the topic, the FDA has continued to adhere to the 2005 recommendations. Specifically, in most instances, the FDA does not recognize a lower risk of hypoglycemia that has been measured on the basis of the number of blood glucose values below a certain cut point value, but instead demands that a lower rate of hypoglycemia be demonstrated by reducing the frequency of severe episodes requiring external assistance. We contend that this regulatory definition of hypoglycemia has become inadequate and should now be expanded.


Consensus standards should be updated periodically.  In view of the heightened recognition of the risks associated with hypoglycemia and the improvements in analytical methodology, a fresh analysis of the recommended targets for lower hypoglycemia rates should consider modern glucose monitoring technologies for identifying hypoglycemia while also weighing current information about the risks and consequences of hypoglycemia.


The Need for an Updated Consensus Statement about Hypoglycemia Reduction The time is propitious for an expert consensus panel (comprised of academic experts in hypoglycemia and clinical trial design, who are not employed by industry) to establish goals for lower hypoglycemia risk, which could now be informed by five important types of information that were not known to the ADA Workgroup when it met in 2004: 

1)    mechanisms and capabilities of the technologies either available or under development for delivery of ultra-fast or ultra-long acting insulins and the potential for these new therapies to modify hypoglycemia risk; 

2)    enhanced knowledge of the current risks of hypoglycemia; 

3)    progressive improvement in the technologies used in self-monitoring of blood glucose

(SMBG) to detect hypoglycemia [41]; 

4)    experience with the use of continuous glucose monitoring (CGM) to assess hypoglycemia during clinical trials of a diabetes therapy [42]; 

5)    recognition that hypoglycemia risk may be unrelated to A1C, bearing witness to the importance of hypoglycemia as a primary endpoint in appropriate clinical trial settings.  In type 1 diabetes the inverse relationship between HbA1c and frequency of severe hypoglycaemia that was observed in the Diabetes Control and Complications Trial (DCCT) (43) has not been apparent in free-living populations studies, in which events were recorded outside of a trial setting.  In both the Type 1 Diabetes Exchange Clinic Registry of 7012 patients (44, 2) a survey of 422 type 1 patients attending three specialty clinics in Melbourne, Australia (45) severe hypoglycemia occurred at all levels of A1C.  The same lack of a linear relationship or even a monotonic relationship between patients’ A1C levels and hypoglycemia risk has also been observed in type 2 patients in the Diabetes Study of Northern California survey of 9094 type 2 patients (46).  In these three studies the risk of severe hypoglycemia was evident at all levels of A1C.


It is also relevant to regard the large amount of SMBG data that are measured in diabetes trials as a resource that is narrowly utilized in regulatory review at present.  In FDA reviews, SMBG data are not typically presented, analyzed, and interpreted as a standalone metric, but are only used to qualify a hypoglycemia event by one of several definitions.  For example, in its insulin degludec New Drug Application, the definition of ADA Documented Symptomatic Hypoglycemia was used to define an event associated with both symptoms of hypoglycemia and a documented blood glucose < 70 mg/dl (<3.9 mmol/L).  Novo Nordisk also provided data by another definition, referred to as Novo Nordisk Confirmed Hypoglycemia, and defined as the sum of ADA-severe events and episodes where blood glucose < 56 mg/dl (<3.1 mmol/L) was recorded (i.e., with or without symptoms).[36] 


The SMBG data utilized above presumably represent a small proportion of the total SMBG data collected in the degludec and other programs.  SMBG data also represent the great majority of total glucose data collected trials of glucose-lowering therapies. In a recent systematic literature review of the use of blood glucose monitoring in phase 3 clinical studies of insulin analogs, it was found that of the total number of measured glucose values across all studies, 84% were with SMBG alone, 9% were with both SMBG and laboratory measurement, and 7% were with laboratory measurement alone.[47] While laboratory measurement of glucose is inherently more accurate than SMBG, laboratory measurement is limited by its impracticality to assess other than fasting glucose values.  The wealth of SMBG data collected in clinical trials of diabetes products, compared to laboratory measurements alone, provides a fuller picture of fluctuating glucose levels during the entire 24-hour day. A disadvantage of SMBG compared to lab measured values in accuracy is well offset by the quantity and varied timing of SMBG measurements.  The density of SMBG data can support the use of a clinically meaningful and physiologically based cutoff for defining hypoglycemia irrespective of whether or not symptoms are associated.


These developments have created an environment whereby many patients and healthcare professionals are seeking therapies for diabetes that will reduce the risk of hypoglycemia by any margin whatsoever.  An expert panel could assess the current evidence and provide meaningful updates to existing consensus statements.


Proposals for Facilitating Development of Therapies that Reduce Hypoglycemia Risk

(1)Develop composite measures of benefit for regulatory endpoints

 Understandably, the FDA will not recognize the value of a lower rate of observed hypoglycemia in a clinical trial if it is associated with an unfavorable rise in A1C. [48] A preferred approach would be to offer composite or categorical primary efficacy endpoints that reflect wider benefit than overall glycemic control alone.  In general, this approach would not trade inferior glycemic control for hypoglycemia benefit.  Under special circumstances, however, we foresee that a diabetes product could be approved in which A1C as a secondary endpoint could fail a non-inferiority test in the presence of the test article having demonstrated a clear beneficial effect on a primary hypoglycemia endpoint.

We call for stakeholders to consider these alternatives and to work with regulatory authorities to update and broaden approaches to the efficacy evaluation of diabetes therapies.  This could include recommending a composite endpoint that combines A1C and hypoglycemia risk, such as the percentage of participants achieving an A1C < 7% without developing unacceptable hypoglycemia. [49, 50] 

2) Incorporate established diabetes technology in the regulatory evaluation of hypoglycemia.  We recommend eliminating the discontinuity between the historic definition of severe hypoglycemia and hypoglycemia that is less severe but still very consequential. A hierarchy of responses to induced hypoglycemia occurs in healthy individuals.  Sequential thresholds include: 1) detection of a counterregulatory hormone response at approximately 68 mg/dl (3.8 mmol/L); 2) generation of autonomic symptoms at approximately 58 mg/dl (3.2 mmol/L); and 3) onset of cognitive dysfunction at approximately 51 mg/dl (2.8 mmol/L).[51] Difficulty in perceiving the onset of hypoglycemia, known as impaired hypoglycemia awareness,[52] is both a disability and an explanation for the short temporal gap between a patient noticing or questioning whether hypoglycemia is present and the sudden onset of disabling neuroglycopenia that requires external assistance for its treatment. 


The first response threshold to hypoglycemia (counterregulatory hormone secretion) may be a homeostatic mechanism, activated to maintain glucose within the “normal” range, and not a consequence of hypoglycemia.  Based on this hierarchy, the clinical importance of the onset of cognitive impairment resulting from hypoglycemia (which may manifest as neuroglycopenic symptoms), the unreliability of autonomic symptoms per se and the suitability of available technology to detect a low glucose concentration we propose utilizing a numerical cut point of <54 mg/dl (<3.0 mmol/L) as a clinically relevant level with which to define meaningful hypoglycemia for trials of diabetes therapies.[26] A glucose concentration < 3.0 mmol/L (< 54 mg/dL) causes defective glucose counterregulation and impaired awareness of hypoglycemia, the core components of hypoglycemia-associated autonomic failure in diabetes. Avoidance will reverse impaired awareness of hypoglycemia and some aspects of defective glucose counterregulation in most affected patients.  In type 1 diabetes, failure to recognize one’s own hypoglycemia at a glucose concentration < 3.0 mmol/L (54 mg/dL) increases the risk of severe hypoglycemia (defined as needing the help of another person for recovery) four-fold.  In type 2 diabetes, this glucose concentration is associated with cardiac arrhythmias.  A glucose concentration <3.0 mmol/L (<54 mg/dL) was associated with mortality in the NICE-SUGAR trial. In view of the increasing risk as blood glucose falls below < 3.0 mmol/L (< 54 mg/dL), this level can therefore be taken to indicate the onset of significant neuroglycopenia in response to hypoglycemia.  This definition is consistent with the consensus view of other expert groups such as the International Hypoglycaemia Study group, and we believe this proposal has recently been agreed by the Professional Practice Committee of the American Diabetes Association. Revision of the biochemical definitions of hypoglycemia is now under consideration by the FDA following a recent workshop in August 2016. [53]


A cut point at 54 mg/dL (3.0 mmol/L) would also be diagnostically relevant if hypoglycemia is to be evaluated with both SMBG and CGM systems. A CGM system measures interstitial tissue glucose, and therefore reports concentrations that lag behind those in capillary or venous glucose. Nonetheless, CGM systems now provide reasonably good accuracy and detect most, if not all, defined hypoglycemia events.[54,55,56] A major limitation of SMBG compared to CGM is that often patients will fail to test themselves with SMBG during mild nocturnal hypoglycemia or in the presence of impaired awareness of hypoglycemia, unless the event is severe.  Blinded (professional) CGM is preferable to real time (personal) CGM for clinical trials that measure the incidence of hypoglycemia. Personal CGM, unlike professional CGM, would decrease the incidence of the hypoglycemia endpoint that is being measured by changing a subject’s behavior pattern. [57] This is because the alarms and alerts that are present in real-time monitoring can trigger a change in behavior. Even if they are turned off, the monitor’s displayed glucose values and trend arrows might also trigger a behavior change, which could include checking a fingerstick blood glucose level. CGM measurement performance is more than adequate for the purpose of providing endpoints in clinical trials.  In contrast to the importance of a single hypoglycemia warning for an individual, clinical trials involve comparisons of aggregate results and statistical methods for sizing trials based on variance of the measure. A glucose level <54 mg/dl (<3.0 mmol/L) as measured by CGM reflects a serious degree of hypoglycemia.  We propose that the duration of time and/or number of events (i.e. the frequency) recorded by CGM at or below this level would serve as highly clinically meaningful endpoints for trials.  This proposed level is considerably lower than the current ADA definition of <70 mg/dl (<3.9 mmol/L) for biochemical hypoglycemia,[26] which is recognized to be an appropriate glucose alert level for the risk of developing hypoglycemia.[9]  Setting the definition of hypoglycemia at levels lower than 54 mg/dl (3.0 mmol/L) would miss many clinically relevant symptomatic episodes. Setting the definition of hypoglycemia at levels higher than 54 mg/dl (3.0 mmol/L) or higher would substantially increase the event rate, but would capture biochemical hypoglycemia events of debatable clinical relevance. [58] It should be noted that when FDA approved the Medtronic 530G (an insulin pump that is integrated with a glucose sensor), the definition of severe hypoglycemia as described in the DCCT was not used as a primary endpoint, and CGM rather than SMBG or symptoms was permitted to document the outcome. [59,60]


Severe hypoglycemic endpoints occur less frequently than less severe hypoglycemic endpoints.  In an analysis of data from two trials of insulin glargine dose titration in 12,837 participants with type 2 diabetes who were commencing insulin therapy the glucose cut-off point defining hypoglycemia greatly influenced the reported frequency of hypoglycemia.  For example, decreasing the hypoglycemia cutoff point from 70 mg/dl to 56 mg/dl (3.9 mmol/L to 3.1 mmol/L) decreased the percentage of affected patients from 43.3 to 17.7% among participants whose A1C level was 7.0–7.2%. (See Figure 1) [58] Similarly, a review of 22 clinical trials of glucose-lowering therapy for diabetes using sulfonylureas demonstrated that with lower cutoffs for diagnosing hypoglycemia, the incidence of hypoglycemia decreased.[61] The number of severe hypoglycemic events necessary to demonstrate statistical significance is so high in many therapeutic trials that it is too difficult to power a study to demonstrate a benefit of reduced hypoglycemia using the current definition.  The potential benefit of some treatments that can be shown to significantly diminish hypoglycemia risk at a threshold level of below 54 mg/dl (3.0 mmol/L) might go unnoticed at lower hypoglycemic thresholds. This washout of statistically significant benefit could occur with underpowered studies specifying hypoglycemic thresholds below 54 mg/dl (3.0 mmol/L) if hypoglycemia were only defined by disabling neuroglycopenia requiring assistance.  


We contend that CGM is sufficiently accurate for estimating plasma glucose values for this purpose.  Recognition of hypoglycemia can be delayed by 5-10 minutes with a CGM (which measures interstitial fluid glucose) compared to a fingerstick test glucose monitor (which measures capillary blood glucose). [62] Inaccuracy due to such delays is “baked into” the accuracy statistics of CGM sensors, which we believe are adequate for this purpose. 

Although accuracy of measurement of both CGM and SMBG measurements declines with increasing profundity of hypoglycemia, this factor will work against demonstrating a hypoglycemia benefit of one treatment over another in a well-controlled trial.   In the context of a non-inferiority comparison, variance works against detecting a difference, but this is also overcome with appropriately determined sample size.  CGM typically does not carry a sampling bias as is the case for unscheduled SMBG measures in an unblinded trial, but both modalities play complementary roles in clinical trials.  Thus, both CGM and SMBG measurements with the proposed cut-point can be used with confidence to establish a statistically significant and clinically meaningful difference between treatments for informing clinical and regulatory decision making.


Facilitating the fulfillment of unmet clinical need for reducing hypoglycemia The development of better therapies for diabetes management depends on establishing regulatory endpoints that are both scientifically based and achievable. Hypoglycemia should not be defined in a way that requires massive clinical trial expenditure and/or reduces the personal and societal impacts that solid evidence should encourage. The means to obtain robust evidence of hypoglycemia have expanded dramatically since the era in which the DCCT was conducted.  Likewise, awareness of the importance of hypoglycemia is burgeoning, with a recent proposal that hypoglycemia rates be considered as an indicator of good diabetes care[63] and evidence-based clinical practice guidelines for treatment of type 1 diabetes complicated by problematic hypoglycemia have recently been published.[64] (Table 2)   New technologies are being developed to reduce hypoglycemia risks (e.g. threshold and predictive suspend sensor-augmented CSII, new faster and longer insulins, non-insulin therapeutics), and unification of methods for evaluation of these technologies will aid in their development and application.[65]  All stakeholders should apply recent advances in awareness, diagnosis and treatment of hypoglycemia toward developing better therapies for people with diabetes.


Author Contributions: All four authors researched data and wrote the manuscript.  Funding: No financial support was received by the authors


Acknowledgements: We thank Annamarie Sucher, Diabetes Technology Society, for her support in the preparation of this article.


Disclosures: The authors report no potential conflicts of interest relevant to this article.




1)      American Diabetes Association. Implications of the United Kingdom Prospective Diabetes Study. Diabetes Care 2003: 25; suppl 1 s28-s32; doi:10.2337/diacare.25.2007.S28.


2)      McCall AL.  Insulin therapy and hypoglycemia. Endocrinol Metab Clin North Am



3)      Frier BM. Hypoglycaemia in diabetes mellitus: epidemiology and clinical implications.

Nat. Rev. Endocrinol. 2014;10:711–722.


4)      Rubin RR, Wadden TA, Bahnson JL, Blackburn GL, Brancati FL, Bray GA, Coday M,

Crow SJ, Curtis JM, Dutton G, Egan C, Evans M, Ewing L, Faulconbridge L, Foreyt J,

Gaussoin SA, Gregg EW, Hazuda HP, Hill JO, Horton ES, Hubbard VS, Jakicic JM, Jeffery

RW, Johnson KC, Kahn SE, Knowler WC, Lang W, Lewis CE, Montez MG, Murillo A,

Nathan DM, Patricio J, Peters A, Pi-Sunyer X, Pownall H, Rejeski WJ, Rosenthal RH, Ruelas

V, Toledo K, Van Dorsten B, Vitolins M, Williamson D, Wing RR, Yanovski SZ, Zhang P; Look AHEAD Research Group. Impact of intensive lifestyle intervention on depression and health-related quality of life in type 2 diabetes: the Look AHEAD Trial. Diabetes Care

2014;37:1544-53. doi: 10.2337/dc13-1928.


5)      Wild D, von Maltzahn R, Brohan E, Christensen T, Clauson P, Gonder-Frederick L. A critical review of the literature on fear of hypoglycemia in diabetes: Implications for diabetes management and patient education. Patient Education and Counseling 2007;68:10–15.


6)      Ross SA. Breaking down patient and physician barriers to optimize glycemic control in type 2 diabetes. Am J Med. 2013 Sep;126(Suppl 1):S38-48.


7)      Trief PM, Sandberg JG, Dimmock JA, Forken PJ, Weinstock RS. Personal and relationship challenges of adults with type 1 diabetes: a qualitative focus group study.

Diabetes Care 2013;36:2483-2488. 


8)      Seaquist ER,Anderson J,Childs B, et al. Hypoglycemia and diabetes: a report of a workgroup of the American Diabetes Association and the Endocrine Society.

Diabetes Care 2013;36:1384-95.


9)      International Hypoglycaemia Study Group. Minimizing hypoglycemia in diabetes.

Diabetes Care 2015;38:1583–1591.


10)  Ahrén, B. Avoiding hypoglycemia: a key to success for glucose-lowering therapy in type 2 diabetes. Vasc Health Risk Manag 2013;9:155-63.


11)  Weinstock R, Xing D, Maahs, Michels A, Rickels M et al. Severe hypoglycemia and diabetic ketoacidosis with type 1 diabetes: results from the T1D exchange clinic registry. J

Clin Endocrinol Metab 2013;98:3411-19.


12)  Weinstock R, Dubose S, Bergenstal R, Chaytor N, Peterson C et al. Risk factors associated with severe hypoglycemia in older adults with type 1 diabetes. Diabetes Care



13)  Deary IJ, Hepburn DA, MacLeod KM, Frier BM. Partitioning the symptoms of hypoglycaemia using multi-sample confirmatory factor analysis. Diabetologia 1993;36:771777.


14)  McAulay V, Deary IJ, Frier BM. Symptoms of hypoglycaemia in people with diabetes.

Diabet Med 2001;18:690-705.


15)  Zammitt, N.N., Streftaris, G., Gibson, G.J. et al. Modeling the consistency of hypoglycemic symptoms: high variability in diabetes. Diabetes Technol Ther 2011;13:571– 578.


16)  Olsen SE, Asvold BO, Frier BM, Aune SE, Hansen LI, Bjørgaas MR. Hypoglycaemia symptoms and impaired awareness of hypoglycaemia in adults with Type 1 diabetes: the association with diabetes duration.Diabet Med 2014;31:1210–1217.


17)  Cox DJ, Gonder-Frederick L, Antoun B, Cryer PE, Clarke WL. Perceived symptoms in the recognition of hypoglycemia. Diabetes Care 1993;16(2):519-527.


18)  Cryer PE. Mechanisms of hypoglycemia-associated autonomic failure in diabetes.N Engl J Med 2013;369:362-72.  


19)  Gold AE, MacLeod KM, Frier BM. Frequency of severe hypoglycemia in patients with type I diabetes with impaired awareness of hypoglycemia.Diabetes Care 1994;17:697-703. 20) Schopman JE, Geddes J, Frier BM. Prevalence of impaired awareness of hypoglycaemia and frequency of hypoglycaemia in insulin-treated type 2 diabetes. Diabetes Res Clin Pract.

2010 Jan;87(1):64-8. doi: 10.1016/j.diabres.2009.10.013. Epub 2009 Nov 24.


21)  Schopman JE, Geddes J, Frier BM. Frequency of symptomatic and asymptomatic hypoglycaemia in Type 1 diabetes: effect of impaired awareness of hypoglycaemia. Diabet Med. 2011 Mar;28(3):352-5. doi: 10.1111/j.1464-5491.2010.03203.x.


22)  The American Diabetes Association Workgroup on Hypoglycemia. Defining and

Reporting Hypoglycemia in Diabetes. Diabetes Care 2005;28:1245-1249.


23)  Diabetes Control and Complications Trial (DCCT): results of feasibility study. The

DCCT Research Group. Diabetes Care 1987;10:1-19.


24)  European Medicines Agency. Guideline on clinical investigation of medicinal products in the treatment or prevention of diabetes mellitus. 2012. Available from 00129256.pdf. Accessed 21 October 2016.


25)  U.S. Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research (CDER). Guidance for Industry Diabetes Mellitus: Developing Drugs and Therapeutic Biologics for Treatment and Prevention. 2008. Available from Accessed 21 October 2016.


26)  Muchmore DB, Heinemann L, Tamborlane W, Wu XW, Fleming A. Assessing rates of hypoglycemia as an end point in clinical trials. Diabetes Care 2015;10:e160-1.


27)  The Diabetes Control and Complications Trial Research group. Epidemiology of severe hypoglycemia in the Diabetes Control and Complications Trial. Amer J Med 1991; 90:450459.


28)  Prescribing information for degudec (Tresiba, Novo). 2015. Available from Accessed 21 October 2016. 


29)  European Agency for the Evaluation of Medicinal Products. Committee of Proprietary Medicinal Products. Note for guidance on the clinical investigation of medicinal products in the treatment of diabetes mellitus. Published 2002. Available from 00003262.pdf. Accessed 21 October 2016. 


30)  European Agency for the Evaluation of Medicinal Products. Committee of Proprietary Medicinal Products.  Note for guidance on the clinical investigation of medicinal products in the treatment of diabetes mellitus. Updated guidance published 2012. Accessible from 00129256.pdfAccessed 21 October 2016. 


31)  Prescribing information for Toujeo. 2015. Available from Accessed 21 October 2016.  32) Tresiba (insulin degludec) EMEA Authorisation. 2015. Available from man_med_001609.jsp&mid=WC0b01ac058001d124. Accessed 21 October 2016.


33)  Clinical Studies Section of Labeling for Human Prescription Drug and Biological Products - Content and Format. 2006. Available from Accessed 21

October 2016. 


34)  Food Drug and Cosmetic Act 21 USC § 355(d). 2010. Available from Accessed 21 October 2016. 


35)  Coutant D, Riggs D, Van Sant Hoffman E. Substantial Evidence: When is a single trial sufficient for approval and promotion? Drug Information Journal 2011;45:253-263.


36)  FDA Briefing Document: NDA 203313 and NDA 203314 Insulin Degludec and Insulin

Degludec/Aspart Novo Nordisk, Inc. Endocrinologic and Metabolic Drugs Advisory Committee Meeting. 2012. Available from docrinologicandMetabolicDrugsAdvisoryCommittee/UCM327015.pdf. Accessed 21 October 2016.


37)  U.S. Food and Drug Administration. FDA approves two new drug treatments for diabetes mellitus. Published September 25, 2015. Available from 21 October 2016.


38)  Gerstein HC, Miller ME, Byington RP, et al. Action to Control Cardiovascular Risk in

Diabetes Study Group. Effects of intensive glucose lowering in type 2 diabetes. N Engl J

Med 2008;358:2545–2559.


39)  Patel A, MacMahon S, Chalmers J, et al. ADVANCE Collaborative Group. Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med



40)  Duckworth W, Abraira C, Moritz T, et al. VADT Investigators. Glucose control and vascular complications in veterans with type 2 diabetes. N Engl J Med 2009;360:129–139.


41)  Klonoff DC, Prahalad P. Performance of cleared blood glucose monitors. J Diabetes Sci Technol 2015;9:895-910. 


42)  Danne T, Philotheou A, Goldman D, Guo X, Ping L, Cali A, Johnston P. A randomized trial comparing the rate of hypoglycemia--assessed using continuous glucose monitoring--in 125 preschool children with type 1 diabetes treated with insulin glargine or NPH insulin (the

PRESCHOOL study). Pediatr Diabetes 2013;14:593-601.


43)  The Diabetes Control Complications Trial Research Group, The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus, N. Engl. J. Med. 1993; 329: 977–986.

44)  Weinstock RS, Xing D, Maahs DM et al for the T1D Exchange Clinic Registry, Severe hypoglycemia and diabetic ketoacidosis in adults with type 1 diabetes: results from the T1D Exchange Clinic Registry.  JCEM 2013; 98:3411-3419.


45)  Hendrieckx C, Halliday JA, Bowden JP, Colman PG, Cohen N, Jenkins A, Speight J. Severe hypoglycaemia and its association with psychological well-being in Australian adults with type 1 diabetes attending specialist tertiary clinics. Diabetes Res Clin Pract. 2014



46)  Lipska KJ, Warton EM, Huang ES, Moffet HH, Inzucchi SE, Krumholz HM, Karter AJ. 

HbA1c and risk of severe hypoglycemia in type 2 diabetes: the Diabetes and Aging Study.

Diabetes Care. 2013 Nov;36(11):3535-42.


47)  Miikkulainen K, A Caruso, O Mast, R Zhang and O Borisenko. Systematic literature review of use of blood glucose monitoring in phase III clinical studies of insulin analogs. BMC Endocrine Disorders 2016;16:21.


48)  Abrahamian H, Francesconi M, Loiskandl A, et al. Evaluation of a new insulinotropic agent by using an innovative technology: efficacy and safety of nateglinide determined by continuous glucose monitoring. Diabetes Technol Ther. 2004;6:31-37.


49)  Muchmore, D. The end point is just the beginning. J Diabetes Sci Technol 2011;5:12871289.


50)  Fleming GA. The endpoint: less is more. J Diabetes Sci Technol 2011;5:1290–1293.

51)  Mitrakou A, Ryan C, Veneman T, Mokan M, Jenssen T, Kiss I, Durrant J, Cryer P, Gerich J. Hierarchy of glycemic thresholds for counterregulatory hormone secretion, symptoms, and cerebral dysfunction. Am J Physiol 1991;260(1 Pt 1):E67-74.


52)  Graveling AJ, Frier BM. Impaired awareness of hypoglycaemia: a review. Diabetes and

Metabolism 2010;36:S64–S74.


53)  U.S. Food and Drug Administration. Public Workshop: Diabetes Outcome Measures Beyond Hemoglobin A1c (HbA1c). Updated August 31, 2016. Available at: Accessed 21 October 2016.  


54)  Keenan DB, Mastrototaro JJ, Zisser H, Cooper KA, Raghavendhar G, Lee SW, Yusi J, Bailey TS, Brazg RL, Shah RV. Accuracy of the Enlite 6-day glucose sensor with guardian and Veo calibration algorithms. Diabetes Technol Ther 2012;14:225-231.


55)  Nakamura K, Balo A. The accuracy and efficacy of the Dexcom G4 Platinum Continuous Glucose Monitoring System. J Diabetes Sci Technol 2015;9:1021-1026.


56)  Bailey T, Bode BW, Christiansen MP, Klaff LJ, Alva S.  The Performance and Usability of a Factory-Calibrated Flash Glucose Monitoring System.  Diabetes Technol Ther. 2015



57)  Hermanns N1, Schumann B2, Kulzer B3, Haak T3. The impact of continuous glucose monitoring on low interstitial glucose values and low blood glucose values assessed by pointof-care blood glucose meters: results of a crossover trial. J Diabetes Sci Technol



58)  Swinnen SG, Mullins P, Miller M, Hoekstra J B, Holleman F. Changing the glucose cutoff values that define hypoglycaemia has a major effect on reported frequencies of hypoglycaemia. Diabetologia 2009;52:38–41.


59)  Gutierrez A, DEPARTMENT OF HEALTH & HUMAN SERVICES. P120010 - Lusin

Markaryan. Available at: Accessed 21 October 2016. 


60)  Bergenstal RM, Klonoff DC, Garg SK, Bode BW, Meredith M, Slover RH, Ahmann AJ, Welsh JB, Lee SW, Kaufman FR; ASPIRE In-Home Study Group.  Threshold-based insulinpump interruption for reduction of hypoglycemia. N Engl J Med. 2013 Jul 18;369(3):224-32


61)  Schopman JE, Simon AC, Hoefnagel SJ, Hoekstra JB, Scholten RJ, Holleman F. The incidence of mild and severe hypoglycaemia in patients with type 2 diabetes mellitus treated with sulfonylureas: a systematic review and meta-analysis. Diabetes Metab Res Rev



62)  Basu A, Dube S, Veettil S, Slama M, Kudva YC, Peyser T, Carter RE, Cobelli C, Basu R. 

Time lag of glucose from intravascular to interstitial compartment in type 1 diabetes.  J Diabetes Sci Technol. 2015 Jan;9(1):63-8.


63)  Rodriguez-Gutierrez R, Lipska KJ, McCoy RG, Ospina NS, Ting HH, Montori VM, et al. Hypoglycemia as an indicator of good diabetes care. BMJ2016;352:i1084.


64)  Yeoh E, Choudhary P, Nwokolo M, Ayis S, Amiel SA. Interventions that restore awareness of hypoglycemia in adults with type 1 diabetes: a systematic review and metaanalysis. Diabetes Care 2015;38:1592-609. 


65)  Vigersky RA. Escaping the hemoglobin A1c-centric world in evaluating diabetes mellitus interventions. J Diabetes Sci Technol 2015; 9:1148-51. 

















Table 1:  Key Clinical and Regulatory Milestones in Hypoglycemia






Diabetes Control and

Complications Trial (27)

Established a durable definition of severe hypoglycemia


European Agency for the

Evaluation of Medicinal

Products Committee on

Proprietary Medicinal

Products (29)

In order to “establish a high level of specificity” to make product claims, this guidance indicated that “only blood glucose values less than 54 mg/dl (3.0 mmol/L) would be considered for evaluation of hypoglycemia.  Defined

“minor” episodes (with or without symptoms) as values <3.0 mmol/L, and

“major” episodes as <3.0 mmol/L and requiring external assistance 


American Diabetes

Association White Paper on

Diabetes (22)

Proposed five categories of hypoglycemia (severe, documented symptomatic, asymptomatic, probable symptomatic and relative) and introduced 70 mg/dl (3.9 mmol/L) as the threshold value for documented hypoglycemia (with or without symptoms) based on published data from studies of counterregulatory hormone responses to induced hypoglycemia.  Suggested that




reduction of severe hypoglycemia by as little as 10-20% or documented hypoglycemia by 30% would be clinically important


US FDA Draft Guidance for

Development of Diabetes 

Therapeutics (25)

Regulatory recognition of ADA definitions of hypoglycemia while emphasizing A1C as the primary endpoint for diabetes treatment trials.  Indicated that new treatments should not cause an “undue” increase in hypoglycemia


European Agency for the

Evaluation of Medicinal

Products Committee on

Proprietary Medicinal

Products (30)

“Minor” and “major” terms abandoned with adoption of ADA categories 


Hypoglycemia and Diabetes:  A report of a workgroup of the American Diabetes Association and the

Endocrine Society (8)

Re-named “relative” hypoglycemia to be

“pseudo-hypoglycemia.”  Re-affirmed ADA definitions and provided clinical management strategies but did not call for re-evaluation of regulatory guidance regarding hypoglycemia endpoints in

clinical trials

Table 2:  Endpoints/Methods of identifying and measuring hypoglycemia




Subjective symptoms

Relatively easy to collect

Unreliable as markers for hypoglycemia in persons with diabetes

Severe hypoglycemia events

Widely accepted, robust definition of event

Rare, especially in clinical trial populations, requiring very large sample sizes to identify modest improvements

Glucose cut point threshold

≤70 mg/dl (3.9 mmol/L)

Useful “action” threshold for individual subject intervention.  Widely accepted by advisory and regulatory groups

Too non-specific for clinical trial endpoints since many observed occurrences of hypoglycemia are simply due to measurement


Glucose cut point threshold

<56 mg/dl (3.1 mmol/L)

Large legacy database from previous clinical trials that have used this cut point, which was originally stipulated by the EMA.  Minimizes false positive rates, increasing specificity.  Frequent enough occurrence to allow moderate sized

Available measurement methods in the ambulatory care setting are less accurate than reference methods



clinical trials 


Self-Monitoring of Blood

Glucose (SMBG)

Widely available technology with everimproving precision and accuracy.  Little systematic bias, so increased variability can be overcome by increasing sampling

Less accurate at hypoglycemic levels than reference standard methods.  Potentially biased data ascertainment due to lack of adherence to scheduled data collection times

Continuous glucose monitoring (CGM)

May be used masked or unmasked (real time feedback).  Very large datasets can be generated.  Little systematic bias, so increased variability can be overcome by increasing sampling.  Less biased than SMBG since data collection occurs without study subject intervention.  Can identify directional trends

(e.g. impending hypoglycemia).  Can assess many parameters (time in range, glucose variability, etc.)

Requires more training than SMBG and more costly to implement.  Less accurate than SMBG in hypoglycemic range, but technology steadily improving  




Fig. 1 Proportion of patients experiencing at least one non- severe hypoglycaemic episode during the 12-week analysis period for a range of predefined glucose cut-off points for the definition of hypoglycaemia, plotted against endpoint HbA1c categories. [58] (Reproduced from an open access article distributed under the terms of the Creative Commons Attribution Noncommercial License)