Research: Complications Asymptomatic hypoglycaemia in Type 1 diabetes: incidence and risk factors

Transcription

1 Research: Complications Asymptomatic hypoglycaemia in Type 1 diabetes: incidence and risk factors M. M. Henriksen 1, H. U. Andersen 2, B. Thorsteinsson 1,3 and U. Pedersen-Bjergaard 1,3 1 Department of Endocrinology and Nephrology, Nordsjællands Hospital, Hillerød, 2 Steno Diabetes Center Copenhagen, Gentofte and 3 Department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark Accepted 25 October 2018 DOI: /dme Abstract Aim The epidemiology of asymptomatic (silent) hypoglycaemia is not well-described. We investigated incidence and risk factors for asymptomatic hypoglycaemia in Type 1 diabetes. Methods A cohort of 153 people with Type 1 diabetes participated in 6 days of blinded continuous glucose monitoring (CGM) and recording of hypoglycaemia symptoms. At entry, hypoglycaemia awareness was classified (by three different methods) and HbA 1c and C-peptide were measured. Hypoglycaemic episodes were defined as interstitial glucose 3.9 mmol/l (IG 3.9 )or 3.0 mmol/l (IG 3.0 ) for 15 min, and were considered asymptomatic if no hypoglycaemic symptoms were reported. Results At thresholds IG 3.9 and IG 3.0, the incidence rates of hypoglycaemic episodes were 5.0 (7.9) [median (IQR)] and 1.3 (3.4) episodes/person-week, respectively. Three-quarters of episodes were asymptomatic. In total, 77% and 52% of participants experienced one or more episode of asymptomatic hypoglycaemia at IG 3.9 and IG 3.0 [3.0 (6.2) and 1.0 (2.3) asymptomatic episodes/person-week]. At multivariate analysis, reduced awareness was positively associated with asymptomatic hypoglycaemia, particularly nocturnal events, and negatively with symptomatic hypoglycaemia. High insulin dose was associated with increased risk of both asymptomatic and symptomatic hypoglycaemia, whereas low HbA 1c and long diabetes duration were risk factors only for symptomatic hypoglycaemia. Conclusions Asymptomatic hypoglycaemia constitutes the majority of hypoglycaemic events in Type 1 diabetes. Reduced hypoglycaemia awareness and high insulin dose are risk factors for asymptomatic hypoglycaemia but other conventional risk factors for severe hypoglycaemia do not correlate with risk of asymptomatic episodes. Diabet. Med. 36:62 69 (2019) Introduction The existence of daytime [1] and nocturnal [2] asymptomatic hypoglycaemia (silent; i.e. episodes of hypoglycaemia with a measured low glucose level but without symptoms of hypoglycaemia [3]) has long been known. An older study using blinded self-monitoring of blood glucose (SMBG) to assess the daytime frequency of asymptomatic hypoglycaemia in Type 1 diabetes found that 84% of all episodes with biochemical hypoglycaemia were not accompanied by symptoms of hypoglycaemia [1]. However, this study assessed frequency of asymptomatic hypoglycaemia only during the daytime. The development of continuous glucose Correspondence to: Marie Moth Henriksen. marie.moth.henriksen@regionh.dk monitoring systems (CGM) with increasing reliability in the hypoglycaemic range has made it possible to document the occurrence of asymptomatic hypoglycaemia 24 h a day. Newer studies using CGM have shown that the majority of hypoglycaemic episodes may be asymptomatic [4 6], supporting the findings of previous SMBG-based studies. Despite this, the epidemiology of asymptomatic hypoglycaemia in adults with Type 1 diabetes is neither described nor fully understood. This contrasts with the epidemiology of severe hypoglycaemia, which is well-described and documented [7 10]. Risk factors for severe hypoglycaemia are welldocumented and include impaired hypoglycaemia awareness and unawareness (i.e. reduced or lost ability to perceive hypoglycaemia), loss of endogenous insulin secretion, genetic factors (high angiotensin-converting enzyme activity), long 62 ª 2018 Diabetes UK

2 Research article What s new? The existence of asymptomatic hypoglycaemia in Type 1 diabetes has been described previously but the epidemiology of these episodes has not been explored, thus rates of asymptomatic hypoglycaemia are unclear. This study reveals that asymptomatic hypoglycaemia constitutes the majority of hypoglycaemic episodes in Type 1 diabetes assessed by blinded CGM. Highlighting that self-reported symptomatic hypoglycaemia comprises only a minor fraction of the total burden of hypoglycaemia in Type 1 diabetes; particularly, in individuals with impaired hypoglycaemia awareness. This study highlights the importance of the use of CGM in detecting asymptomatic episodes; particularly in individuals with impaired hypoglycaemia awareness. duration of diabetes and strict glycaemic control (even though this has been markedly weaker in newer population-based studies) [8,11 14]. However, the impact of these factors on the rate of asymptomatic hypoglycaemia has not been examined. The consequences of asymptomatic hypoglycaemia are largely unknown, but as asymptomatic hypoglycaemic events may comprise the majority of episodes, concerns have been raised as to their possible role in the development of impaired awareness of hypoglycaemia. Recurrent episodes of asymptomatic hypoglycaemia may blunt the symptoms of hypoglycaemia and thereby promote the syndrome of impaired awareness [15]. This is supported by studies finding an increased rate of asymptomatic hypoglycaemia in individuals with unawareness compared with individuals with normal awareness [16 18]. Therefore, the aim of this study is to describe the epidemiology of asymptomatic hypoglycaemia in people with Type 1 diabetes, including: (i) the frequency of asymptomatic hypoglycaemia assessed by CGM in daily life in Type 1 diabetes; (ii) risk factors for asymptomatic hypoglycaemia; and (iii) frequency and risk factors for nocturnal and daytime asymptomatic hypoglycaemia to assess possible diurnal variation. Participants and methods Participants Some 153 people with Type 1 diabetes were included in the study. They were recruited consecutively at the diabetes outpatient clinic at Nordsjællands Hospital, Denmark. All included individuals were adults ( 18 years of age) and diagnosed with Type 1 diabetes for 1 year. Exclusion criteria were end-stage renal disease, concomitant malignant disease and pregnancy. None of the participants used CGM on a regular basis during participation in the study. Written informed consent was obtained from all participants and the Regional Committee on Biomedical Research Ethics (H ) and the Danish Data Protection Agency approved the study. Participants completed a questionnaire upon at entry into the study. The questionnaire asked about the number of episodes of mild hypoglycaemia during the past week (defined as self-treated episodes), and the number of episodes of severe hypoglycaemia during the preceding 1 year (defined as episodes during which assistance from another person was needed to restore glucose levels to normal) [3]. Self-estimated state of hypoglycaemia awareness was assessed using three validated methods [19] described by Gold et al. [20], Clarke et al. [21] and Pedersen-Bjergaard et al. (the Hillerød method) [22]. The Gold and Hillerød methods classify awareness state by asking about the occurrence of hypoglycaemic symptoms, whereas the Clarke method is based upon an eight-item questionnaire. The questionnaires used by Gold et al. and Clarke et al. classify participants as having normal vs. impaired awareness, whereas the Hillerød method further classifies participants as unaware (i.e. lost the ability to perceive hypoglycaemia). History of diabetes, treatments and diabetic late complications were asked for or were extracted from participants medical records as appropriate. The C- peptide level was measured (Advia Centaur XP, Siemens Medical Solutions Diagnostics Inc., Los Angeles, CA, USA) and when undetectable (< 20 pmol/l), the participant was classified as C-peptide negative. HbA 1c was measured using a Tosoh G8 HPLC Analyzer (Tosoh Bioscience Inc., San Francisco, CA, USA) upon entry into the study. Study setting All people with Type 1 diabetes participated in 6 days of continuous interstitial glucose (IG) monitoring (CGM). During the monitoring period, participants carried out their usual daily activities and handled their usual diabetes treatment. The CGM device used was an ipro â 2 using Enlite TM glucose sensors (Medtronic MiniMed, Northridge, CA, USA). The CGM used in the study had an overall mean absolute relative difference (MARD) of 13.9% and 18.4% in the hypoglycaemic range, a sensitivity of 79.5% and a positive predictive value of 83.8% for hypoglycaemia [23]. The CGM was blinded and calibration was done as recommended by the manufacturer using blood glucose meter measurements (Contour â XT meter using Contour â NEXT test strips; Bayer, Basel, Switzerland). During the monitoring period, participants were asked to record any symptoms of hypoglycaemia in a diary every 30 min and, in the case of an episode of hypoglycaemia, to fill in a questionnaire regarding the episode including glucose measurement at time of hypoglycaemia. However, measurement was encouraged but not mandatory when symptomatic hypoglycaemia was experienced. All participants returned the diary after the monitoring period. ª 2018 Diabetes UK 63

3 Asymptomatic hypoglycaemia in Type 1 diabetes M. M. Henriksen et al. Continuous glucose monitoring Table 1 Characteristics of participants with Type 1 diabetes We defined a hypoglycaemic episode as a continuously low IG concentration lasting for 15 min, and recovery when the IG concentration had been continuously above the threshold for 20 min [9,24]. Low IG concentration was classified according to three threshold glucose values [3,24 26]. Predominantly 3.9 mmol/l (IG 3.9 ) [3] as previously recommended by the American Diabetes Association (ADA) and the Endocrine Society, and 3.0 mmol/l (IG 3.0 ) [26] as recently proposed by International Hypoglycaemia Study Group, and endorsed by ADA and the European Association for the Study of Diabetes (EASD). Furthermore, the glucose value 2.2 mmol/l (IG 2.2 ) was used as a third explorative threshold because this is the lower detection limit of the CGM used in this study. A hypoglycaemic episode was classified as asymptomatic (silent) when it was not accompanied by self-reported symptoms of hypoglycaemia in the diary. Daytime was defined as 7:00 AM to 11:00 PM. The total valid duration of CGM recording was calculated by subtracting periods of missed CGM recording. Statistical analyses Standard descriptive statistics were used to characterize the cohort and the rates of hypoglycaemia. When the distributions were skewed mean SD as well as median and interquartile range (IQR) and range are shown. Comparison between participant characteristics and hypoglycaemia rates was carried out using parametric or non-parametric tests (Wilcoxon signed rank test) as appropriate. The influence of risk factors on the rate of asymptomatic hypoglycaemia was assessed with a negative binomial model that took into account the skewed distribution of hypoglycaemic episodes, the observation time and participants who had more than one episode of hypoglycaemia during the monitoring period. We included the following variables in the analyses of risk indicators: age, gender, BMI, insulin dose (change per 0.1 IU/ kg), % bolus/total insulin dose, duration of diabetes (per 10 years), HbA 1c (per 10 mmol/mol; ~ 1%), hypoglycaemia awareness status and C-peptide status. At multivariate analyses, only variables with P < 0.1 in univariate analyses were included. Analyses were performed with SPSS software (v. 22.0); P-values < 0.05 (two-sided) were considered statistically significant. Results Participant characteristics Table 1 shows the baseline characteristics of the 153 participants. Core characteristics including a mean ( SD) age of years and a mean duration of diabetes of years resemble those of most unselected cohorts [7,8]. Characteristic No. of participants 153 Gender, men 90 (59) Age (years)* Duration of diabetes (years)* C-peptide negative 108 (71) BMI (kg/m 2 )* 26 4 Total insulin dose (units/day)* Insulin therapy Analogues 123 (80) Analogue/human 22 (15) Human 8 (5) Insulin treatment regimens 1 3 daily injections 6 (4) 4 daily injections 117 (76) CSII 30 (20) HbA 1c ; mmol/mol (%)* ( ) Awareness (%) Gold: Aware/impaired 80/20 awareness [20] Clarke: Aware/unclassified/ 72/13/15 reduced awareness [21] Hillerød: Aware/impaired/ 47/39/14 unaware [22] Episodes of severe hypoglycaemia *; 0(0 8) (per person-last year) Episodes of mild hypoglycaemia * 1(0 16) (per person-last week) Values are given as n (%) unless indicated otherwise; *mean SD, median (range). Undetectable: < 20 pmol/l. CSII, continuous subcutaneous insulin infusion. Overall frequency of hypoglycaemia The mean duration of CGM was days per person with a total of h of valid recording (99.5% of the total recording time) of which 1338 h (6%) were hypoglycaemic at IG 3.9, 499 h (2%) at IG 3.0 and 168 h (1%) at IG 2.2. Some 87% of participants had at least one hypoglycaemic episode during the recording period (IG 3.9 ). In total, 767 episodes of hypoglycaemia were recorded (IG 3.9 ), 196 (26%) of which were symptomatic and 571 (74%) were asymptomatic. The same was found for the thresholds IG 3.0 and IG 2.2, at which 73% and 75% of all recorded hypoglycaemic episodes were asymptomatic, respectively. None of the episodes of hypoglycaemia were severe. At threshold IG 3.9,IG 3.0 and IG 2.2, the overall incidence of hypoglycaemic episodes/person-week assessed by CGM was (mean SD) , and , respectively [median (IQR): 5.0 (7.9), 1.3 (3.4) and 0.0 (1.2), respectively]. Frequency of asymptomatic hypoglycaemia The frequencies of asymptomatic hypoglycaemia at IG 3.9, IG 3.0 and IG 2.2 are presented in Table 2. Some 118 (77%) of the 153 participants experienced at least one episode of asymptomatic hypoglycaemia (IG 3.9 ). At threshold IG 3.9, the 64 ª 2018 Diabetes UK

4 Research article mean incidence of asymptomatic hypoglycaemia assessed by CGM was (mean SD) [median (IQR): 3.0 (6.2)] episodes/person-week, comprising [2.0 (5.2)] daytime episodes of hypoglycaemia/person-week and [0.0 (2.0)] nocturnal episodes/person-week. This corresponds to [2.1 (5.9)] h of daytime asymptomatic hypoglycaemia/person-week and [0.0 (6.1)] h of nocturnal asymptomatic hypoglycaemia/person-week. At threshold IG 3.0, the frequency of asymptomatic hypoglycaemia was [0.0 (1.2)] daytime episodes/ person-week and [0.0 (1.1)] nocturnal episodes/ person-week. Time spent in the hypoglycaemic range was [0.0 (1.2)] h of daytime hypoglycaemia/personweek and [0.0 (1.6)] h for nocturnal episodes. At threshold IG 2.2, the rate of asymptomatic hypoglycaemia was [0.0 (0.0)] daytime episodes/person-week and [0.0 (0.0)] nocturnal episodes/person-week. Thus, at thresholds IG 3.9,IG 3.0 and IG 2.2, 72%, 60% and 49% of the asymptomatic episodes of hypoglycaemia occurred during daytime, respectively. Risk factors of asymptomatic and symptomatic hypoglycaemia At multivariate analysis (Table 3), hypoglycaemia unawareness (assessed by the Hillerød method [22]) was positively associated with asymptomatic hypoglycaemia and negatively associated with symptomatic hypoglycaemia at IG 3.9. The same was found when state of awareness assessed by the Clarke method was used in the analyses; however, state of awareness assessed by the Gold method was not a significant risk factor for asymptomatic hypoglycaemia in this study. High insulin dose was associated with an increased risk of both asymptomatic and symptomatic hypoglycaemia. Low HbA 1c and long duration of diabetes were significant risk factors for symptomatic hypoglycaemia, but not asymptomatic hypoglycaemia. C-Peptide negativity was not a significant risk factor for symptomatic or asymptomatic hypoglycaemia. At threshold IG 3.0, the same association between different risk factors and risk of hypoglycaemia except hypoglycaemia unawareness was only associated with increased risk of asymptomatic hypoglycaemia, not symptomatic episodes (Table 3). Risk factors and daytime and nocturnal asymptomatic hypoglycaemia Table 4 shows the influence of different known risk factors for severe hypoglycaemia on the rate of asymptomatic hypoglycaemia assessed by univariate and multivariate analyses, divided into daytime and nocturnal episodes at IG 3.9.At univariate analyses, high insulin dose was associated with increased risk of both daytime and nocturnal asymptomatic hypoglycaemia. Furthermore, C-peptide negativity and long duration of diabetes were significant risk factors for daytime asymptomatic hypoglycaemia, and high % bolus/total insulin dose, reduced awareness and low HbA 1c of nocturnal asymptomatic hypoglycaemia. At multivariate analyses, high insulin dose was still positively associated with both daytime and nocturnal asymptomatic hypoglycaemia. Furthermore, hypoglycaemia unawareness was associated with increased risk of nocturnal asymptomatic hypoglycaemia, but the remaining factors were not significantly associated with rates of daytime and night-time asymptomatic hypoglycaemia at multivariate analysis. When we used glucose values IG 3.0 (Table 4) and IG 2.2 as thresholds for hypoglycaemia, an increased risk of nocturnal asymptomatic hypoglycaemia for individuals with unawareness of hypoglycaemia remained, but not an increased risk of daytime asymptomatic hypoglycaemia. The incidence rate ratio (IRR) of nocturnal asymptomatic hypoglycaemia was 2.90 (95% CI ; P = 0.008) Table 2 Daytime and nocturnal episodes of asymptomatic hypoglycaemia (interstitial glucose 3.9, 3.0 and 2.2 mmol/l) Asymptomatic hypoglycaemia Total Daytime Nocturnal Interstitial glucose 3.9 mmol/l One or more episodes (%) Time with IG 3.9 mmol/l (h/person-week) *; 3.8 (11.6) *; 2.1 (5.9) *; 0.0 (6.1) Rate (episodes/person-week) *; 3.0 (6.2) *; 2.0 (5.2) *; 0.0 (2.0) Interstitial glucose 3.0 mmol/l One or more episodes (%) Time with IG 3.0 mmol/l (h/person-week) *; 0.4 (3.5) *; 0.0 (1.2) *; 0.0 (1.6) Rate (episodes/person-week) *; 1.0 (2.3) *; 0.0 (1.2) * 0.0 (1.1) Interstitial glucose 2.2 mmol/l One or more episodes (%) Time with IG 2.2 mmol/l (h/person-week) *; 0.0 (0.0) *; 0.0 (0.0) *; 0.0 (0.0) Rate (episodes/person-week) *; 0.0 (0.0) *; 0.0 (0.0) *; 0.0 (0.0) Values are given as percentage except *mean SD or median (IQR). Daytime was defined as 7:00 AM to 11:00 PM. IG, interstitial glucose. ª 2018 Diabetes UK 65

5 Asymptomatic hypoglycaemia in Type 1 diabetes M. M. Henriksen et al. Table 3 Association of risk factors with incidence of asymptomatic and symptomatic hypoglycaemia in 153 people with Type 1 diabetes assessed by univariate and multivariate analyses IG 3.9 mmol/l IG 3.0 mmol/l Asymptomatic Symptomatic Asymptomatic Symptomatic Risk factors IRR (95% CI) IRR (95% CI) IRR (95% CI) IRR (95% CI) Univariate analyses Gender (women) 1.19 ( ) 0.98 ( ) 0.88 ( ) 0.95 ( ) BMI (kg/m 2 ) 0.99 ( ) 1.02 ( ) 1.01 ( ) 1.02 ( ) Insulin dose (IU/kg) 1.12 ( )* 1.09 ( ) 1.20 ( )* 1.12 ( )* % bolus/total insulin 1.01 ( ) 1.00 ( ) 1.02 ( )* 1.01 ( ) Hypoglycaemia unawareness 1.73 ( )* 0.43 ( )* 2.31 ( )* 0.67 ( ) C-peptide (negative) 1.67 ( )* 1.53 ( ) 1.76 ( ) 1.66 ( ) Duration of diabetes 1.13 ( )* 1.19 ( )* 1.11 ( ) 1.17 ( ) HbA 1c (mmol/mol) 0.85 ( ) 0.77 ( )* 0.98 ( ) 0.97 ( )* Multivariate analyses Insulin dose (IU/kg) 1.13 ( )* 1.18 ( )* 1.19 ( )* 1.12 ( )* % bolus/total insulin 1.02 ( )* Hypoglycaemia unawareness 1.66 ( )* 0.45 ( )* 2.42 ( )* C-peptide (negative) 1.61 ( ) 1.13 ( ) 1.23 ( ) Duration of diabetes 1.04 ( ) 1.16 ( )* 1.01 ( ) HbA 1c (mmol/mol) 0.85 ( ) 0.75 ( )* 0.97 ( )* Values in bold indicate factors included in the multivariate analyses with P < *P < Gender: women compared with men; BMI: per 1 kg/m 2 increment; insulin dose: per 0.1 IU/kg increment; % bolus/total insulin: per 1% increment; hypoglycaemia unawareness compared with normal awareness (Hillerød method [22]); C-peptide negative (undetectable: < 20 pmol/l) compared with C-peptide positive; duration of diabetes: per 10-year increment; HbA 1c : per 10 mmol/mol increment. Factors not included in multivariate analyses due to P 0.10 in univariate analyses. IG, interstitial glucose; IRR, incidence rate ratio; CV, coefficient of variation. Table 4 Association of risk factors with incidence of asymptomatic hypoglycaemia in 153 people with Type 1 diabetes assessed by univariate and multivariate analyses - divided by day and night IG 3.9 mmol/l IG 3.0 mmol/l Daytime Nocturnal Daytime Nocturnal Risk factors IRR (95% CI) IRR (95% CI) IRR (95% CI) IRR (95% CI) Univariate analyses Gender (women) 1.27 ( ) 0.99 ( ) 0.98 ( ) 0.76 ( ) BMI (kg/m 2 ) 0.98 ( ) 1.02 ( ) 1.00 ( ) 1.02 ( ) Insulin dose (IU/kg) 1.12 ( )* 1.12 ( )* 1.19 ( )* 1.19 ( )* % bolus/total insulin 1.01 ( ) 1.02 ( )* 1.03 ( )* 1.01 ( ) Hypoglycaemia unawareness 1.55 ( ) 2.31 ( )* 2.20 ( ) 2.47 ( )* C-peptide (negative) 2.04 ( )* 1.08 ( ) 2.38 ( )* 1.24 ( ) Duration of diabetes 1.18 ( )* 1.02 ( ) 1.14 ( ) 1.07 ( ) HbA 1c (mmol/mol) 0.87 ( ) 0.77 ( )* 0.98 ( ) 0.99 ( ) Multivariate analyses Insulin dose (IU/kg) 1.12 ( )* 1.18 ( )* 1.15 ( )* 1.20 ( )* % bolus/total insulin 1.01 ( ) 1.02 ( )* Hypoglycaemia unawareness 2.02 ( )* 2.05 ( ) 2.90 ( )* C-peptide (negative) 1.61 ( ) 1.53 ( ) Duration of diabetes 1.01 ( ) HbA 1c (mmol/mol) 0.85 ( ) Values in bold indicate factors included in the multivariate analyses with P < *P < Gender: women compared with men; BMI: per 1 kg/m 2 increment; insulin dose: per 0.1 IU/kg increment; % bolus/total insulin: per 1% increment; hypoglycaemia unawareness compared with normal awareness (Hillerød method [22]); C-peptide negative (undetectable: < 20 pmol/l) compared with C-peptide positive; duration of diabetes: per 10-year increment; HbA 1c : per 10 mmol/mol increment. Factors not included in multivariate analyses due to P 0.10 in univariate analyses. at IG 3.0 and increased to IRR 7.73 (95% CI ; P = 0.001) with IG 2.2 for individuals with unawareness compared with individuals with normal state of awareness. Discussion This prospective study in a cohort of unselected individuals with Type 1 diabetes confirms that asymptomatic 66 ª 2018 Diabetes UK

6 Research article hypoglycaemia is frequent, constituting three-quarters of all episodes of hypoglycaemia detected by CGM. This is in line with earlier studies exploring the proportion of asymptomatic hypoglycaemia in Type 1 diabetes, which also demonstrated that SMBG-detected asymptomatic hypoglycaemia comprises a large proportion of all episodes of hypoglycaemia in Type 1 diabetes [1,18]. In the study by Pramming et al. [1], 84% of daytime episodes of hypoglycaemia < 3 mmol/l were asymptomatic, corresponding to an incidence rate of asymptomatic hypoglycaemia of 1.75 episodes/person-week. In another study using SMBG to detect hypoglycaemia in Type 1 diabetes, the prevalence of daytime asymptomatic hypoglycaemia (IG 3.0 ) was 14% and 47% corresponding to an incidence rate of 0.13 and 0.93 episodes/person-week in individuals with normal and impaired awareness, respectively [18]. These studies assessed hypoglycaemia by blinded [1] or unblinded [18] SMBG only during daytime with a limited number of blood glucose recordings and therefore the prevalence of asymptomatic hypoglycaemia is presumed to be underestimated. The incidence of asymptomatic hypoglycaemia in our study was ~ 4 episodes/person-week at IG 3.9 and 2.4 episodes/personweek at IG 3.0. Assessing the rate of asymptomatic hypoglycaemia by SMBG is difficult to do with certainty due to the lack of symptoms of hypoglycaemia. Therefore, the frequency depends heavily upon how frequently plasma glucose is measured (which is normally only feasible during the daytime, resulting in an underrated frequency of nocturnal asymptomatic hypoglycaemia) and on the reliability of recording hypoglycaemia symptoms. This may explain the lower incidence in these studies compared with the findings in our study and emphasizes the importance of CGM in detecting the frequency of asymptomatic hypoglycaemia. Development of CGM has made it possible to assess glucose levels 24 h a day. However, newer studies using CGM to assess hypoglycaemia report a highly variable proportion of asymptomatic hypoglycaemia. The majority of studies report only the night-time prevalence of asymptomatic hypoglycaemia and find a prevalence of nocturnal asymptomatic hypoglycaemia of between 50% and 93% [5,6,27,28]. Studies exploring the rate during 24 h report a prevalence ranging between 59% and 70% [4,17,29,30]. This wide variation may be due to differences in study populations, selection criteria and study setting. Furthermore, definitions of hypoglycaemia, including its threshold, and detection methods differ between studies. Also, the classification of asymptomatic hypoglycaemia is not always clearly defined, resulting in difficulty when comparing frequencies of asymptomatic hypoglycaemia between studies. The majority of studies exploring the frequency of asymptomatic hypoglycaemia do not include an unselected cohort of people with Type 1 diabetes because participants are selected due to state of awareness, age etc. [17,30]. This makes their results difficult to compare with our frequency of asymptomatic hypoglycaemia in an unselected cohort of persons with Type 1 diabetes. Three studies using CGM to assess hypoglycaemia in Type 1 diabetes found an incidence of 0.9 episodes of daytime hypoglycaemia/person-day [4] and between one and four episodes of hypoglycaemia/person-day [16,29], respectively. However, the studies have varying definitions of hypoglycaemia ( mmol/l) and furthermore, do not specifically report the incidence rate of asymptomatic hypoglycaemia which makes it difficult to compare incidence rates of biochemical episodes of hypoglycaemia with the rates of asymptomatic episodes in our study. Additionally, the cohorts in these studies consist of a mixture of children and adults, making it difficult to extrapolate to an adult population. In this study, multivariate analyses were performed to assess the influence of different factors on the risk of asymptomatic and symptomatic hypoglycaemia. Our analyses included factors known to be risk factors for severe hypoglycaemia in Type 1 diabetes. Hypoglycaemia unawareness was a risk factor for both asymptomatic (positive association) and symptomatic (negative association) hypoglycaemia, which is in line with previous studies [8,14]. In our study, lower HbA 1c was associated with an increased risk of symptomatic hypoglycaemia, but not of asymptomatic hypoglycaemia. The association between HbA 1c and increased risk of severe hypoglycaemia has been found in older studies in selected groups [14]; however, newer studies [11 13] challenge this as they do not find an association with severe hypoglycaemia [11,12] or with overall hypoglycaemia [13] in less selected groups. To further assess the effect of diurnal variation in risk factors on asymptomatic hypoglycaemia, we separated the analyses into daytime and nocturnal episodes. High insulin dose and hypoglycaemia unawareness were risk factors for asymptomatic hypoglycaemia but divided into daytime and nocturnal asymptomatic hypoglycaemia, unawareness was only a risk factor for nocturnal asymptomatic hypoglycaemia with increasing power at lower glucose levels. To explore this further, we analysed the association between unawareness and asymptomatic hypoglycaemia at threshold IG 3.0 and IG 2.2. Again, we found that unawareness only was associated with higher risk of nocturnal asymptomatic hypoglycaemia. Thus, the risk of nocturnal asymptomatic hypoglycaemia increased from a factor 2 at IG 3.9 to a factor 3 and a factor 8 at IG 3.0 and IG 2.2, respectively, for individuals with unawareness of hypoglycaemia, compared with individuals with normal awareness. This may be explained by an association between nocturnal asymptomatic hypoglycaemia and the development of reduced awareness of hypoglycaemia; the more severe the episodes, the stronger the association seems to be. This is supported mechanistically by data showing that nocturnal asymptomatic hypoglycaemia reduces the counter-regulatory response to subsequent hypoglycaemia [31,32]. However, it is still not known what ª 2018 Diabetes UK 67

7 Asymptomatic hypoglycaemia in Type 1 diabetes M. M. Henriksen et al. initiates this vicious cycle, whether asymptomatic hypoglycaemia results from reduced awareness or reduced awareness results from recurrent asymptomatic hypoglycaemia. Nevertheless, our findings show that asymptomatic hypoglycaemia is frequent in Type 1 diabetes and that there is an association between asymptomatic hypoglycaemia, especially profound nocturnal asymptomatic hypoglycaemia, and the syndrome of impaired awareness of hypoglycaemia. This emphasizes the importance of use of CGM in detecting these episodes [33]. The rate of severe hypoglycaemia in this study was lower than expected and described in previous studies, which usually report a rate of episodes of severe hypoglycaemia per one-person-year [7,8,10,34]. The low rate in our study could be due to an improvement in diabetes management with the increased use of insulin pumps and insulin analogues [35], and studies using national registers and cohorts support this reduction in rate of severe hypoglycaemia in Type 1 diabetes [12,36]. The duration of diabetes, level of HbA 1c, frequency of unawareness and C-peptidenegative participants are as expected in an unselected cohort of people with Type 1 diabetes [37]; therefore, the lower rate is not presumed to be explained by a selection bias. The lower rate of severe hypoglycaemia could be influenced by the harmonized driver s licence legislation in the European Union, which may have affected the willingness of participants to report episodes of severe hypoglycaemia [38]. The strength of this study is that it included a large number of randomly selected people with Type 1 diabetes. This makes it possible to extrapolate our findings to adult populations in comparable settings. The use of CGM enabled a more complete detection of hypoglycaemia episodes, and importantly, the thorough use of diaries, which were scrutinized together with the participants, enabled precise classification of events according to symptomology. This study shares the limitations of all CGM-based studies, namely the potential time lag in glucose dynamics and reduced accuracy in the hypoglycaemic range compared with conventional blood glucose measurement. This limitation of CGM may result in low glucose measured artefactually, particularly suspected in the cases of asymptomatic events. However, asymptomatic hypoglycaemia is not specific to CGM because such episodes have also been with SMBG even at very low glucose levels. The requirement for at least 15 min duration of CGM recorded hypoglycaemic events minimizes the effect of CGM fluctuations around the cut-off value and provides robustness to the measured hypoglycaemic events. In conclusion, asymptomatic hypoglycaemia assessed by blinded CGM is frequently detected in the majority of people with Type 1 diabetes. Unawareness of hypoglycaemia is associated with increased risk of asymptomatic hypoglycaemia, primarily due to an increased risk of profound nocturnal episodes. High insulin dose is positively associated with both asymptomatic and symptomatic hypoglycaemia but other conventional risk factors for severe hypoglycaemia do not correlate with risk of asymptomatic episodes. Funding sources This work was supported in part by unrestricted research grants from Nordsjællands Hospital Denmark, The Danish Medical Research Grant, and from the Toyota-Fonden, Denmark. The funders were not involved in study design, data collection, data analyses, and preparation of manuscript or publication decisions. Competing interests H.U.A. owns stocks in Novo Nordisk. H.U.A. is on advisory boards for Novo Nordisk and Astra Zeneca. H.U.A. has served on advisory boards for Abbott. B.T. has served on advisory boards for Novo Nordisk and Eli Lilly. U.P.-B. has served on advisory boards for AstraZeneca, Bristol-Myers Squibb, Sanofi-Aventis and Novo Nordisk and has received lecture fees from AstraZeneca, Bristol-Myers Squibb, Sanofi- Aventis and Novo Nordisk. No other competing interests are declared. Acknowledgements The authors thank the participants and the staff at the research unit at Nordsjællands Hospital for their important contributions. References 1 Pramming S, Thorsteinsson B, Bendtson I, Binder C. The relationship between symptomatic and biochemical hypoglycaemia in insulin-dependent diabetic patients. J Intern Med 1990; 228: Gale EA, Tattersall RB. Unrecognised nocturnal hypoglycaemia in insulin-treated diabetics. Lancet 1979; 1: Seaquist ER, Anderson J, Childs B, Cryer P, Dagogo-Jack S, Fish L et al. Hypoglycemia and diabetes: a report of a workgroup of the American Diabetes Association and the Endocrine Society. Diabetes Care 2013; 36: Boland E, Monsod T, Delucia M, Brandt CA, Fernando S, Tamborlane WV. Limitations of conventional methods of selfmonitoring of blood glucose: lessons learned from 3 days of continuous glucose sensing in pediatric patients with type 1 diabetes. Diabetes Care 2001; 24: Woodward A, Weston P, Casson IF, Gill GV. Nocturnal hypoglycaemia in type 1 diabetes frequency and predictive factors. QJM 2009; 102: Ahmet A, Dagenais S, Barrowman NJ, Collins CJ, Lawson ML. Prevalence of nocturnal hypoglycemia in pediatric type 1 diabetes: a pilot study using continuous glucose monitoring. J Pediatr 2011; 159( ): e1. 7 ter Braak EW, Appelman AM, van de Laak M, Stolk RP, van Haeften TW, Erkelens DW. Clinical characteristics of type 1 diabetic patients with and without severe hypoglycemia. Diabetes Care 2000; 23: ª 2018 Diabetes UK

8 Research article 8 Pedersen-Bjergaard U, Pramming S, Heller SR, Wallace TM, Rasmussen AK, Jørgensen HV et al. Severe hypoglycaemia in 1076 adult patients with type 1 diabetes: influence of risk markers and selection. Diabetes Metab Res Rev 2004; 20: UK Hypoglycaemia Study Group. Risk of hypoglycaemia in types 1 and 2 diabetes: effects of treatment modalities and their duration. Diabetologia 2007; 50: Kristensen PL, Hansen LS, Jespersen MJ, Pedersen-Bjergaard U, Beck-Nielsen H, Christiansen JS et al. Insulin analogues and severe hypoglycaemia in type 1 diabetes. Diabetes Res Clin Pract 2012; 96: Weinstock RS, Xing D, Maahs DM, Michels A, Rickels MR, Peters AL et al. Severe hypoglycemia and diabetic ketoacidosis in adults with type 1 diabetes: results from the T1D Exchange clinic registry. J Clin Endocrinol Metab 2013; 98: Karges B, Rosenbauer J, Kapellen T, Wagner VM, Schober E, Karges W et al. Hemoglobin A1c levels and risk of severe hypoglycemia in children and young adults with type 1 diabetes from Germany and Austria: a trend analysis in a cohort of 37,539 patients between 1995 and PLoS Med 2014; 11: e Khunti K, Alsifri S, Aronson R, Cigrovski Berkovic M, Enters- Weijnen C, Forsen T et al. Rates and predictors of hypoglycaemia in people from 24 countries with insulin-treated type 1 and type 2 diabetes: the global HAT study. Diabetes Obes Metab 2016; 18: The Diabetes Control and Complications Trial Research Group. Hypoglycemia in the Diabetes Control and Complications Trial. Diabetes 1997; 46: Cryer PE. Mechanisms of hypoglycemia-associated autonomic failure in diabetes. N Engl J Med 2013; 369: Kubiak T, Hermanns N, Schreckling HJ, Kulzer B, Haak T. Assessment of hypoglycaemia awareness using continuous glucose monitoring. Diabet Med 2004; 21: Choudhary P, Geddes J, Freeman JV, Emery CJ, Heller SR, Frier BM. Frequency of biochemical hypoglycaemia in adults with Type 1 diabetes with and without impaired awareness of hypoglycaemia: no identifiable differences using continuous glucose monitoring. Diabet Med 2010; 27: 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; 28: Høi-Hansen T, Pedersen-Bjergaard U, Thorsteinsson B. Classification of hypoglycemia awareness in people with type 1 diabetes in clinical practice. J Diabetes Complicat 2010; 24: 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: Clarke WL, Cox DJ, Gonder-Frederick LA, Julian D, Schlundt D, Polonsky W. Reduced awareness of hypoglycemia in adults with IDDM. A prospective study of hypoglycemic frequency and associated symptoms. Diabetes Care 1995; 18: Pedersen-Bjergaard U, Pramming S, Thorsteinsson B. Recall of severe hypoglycaemia and self-estimated state of awareness in type 1 diabetes. Diabetes Metab Res Rev 2003; 19: Keenan DB, Mastrototaro JJ, Zisser H, Cooper KA, Raghavendhar G, Lee SW et al. Accuracy of the Enlite 6-day glucose sensor with guardian and Veo calibration algorithms. Diabetes Technol Ther 2012; 14: Bay C, Kristensen PL, Pedersen-Bjergaard U, Tarnow L, Thorsteinsson B. Nocturnal continuous glucose monitoring: accuracy and reliability of hypoglycemia detection in patients with type 1 diabetes at high risk of severe hypoglycemia. Diabetes Technol Ther 2013; 15: Fonseca VA, Grunberger G, Anhalt H, Bailey TS, Blevins T, Garg SK et al.; Consensus Conference Writing Committee. Continuous Glucose Monitoring: a Consensus Conference of the American Association of Clinical Endocrinologists and American College of Endocrinology. Endocr Pract 2016; 22: International Hypoglycaemia Study Group. Glucose concentrations of less than 3.0 mmol/l (54 mg/dl) should be reported in clinical trials: a joint position statement of the American Diabetes Association and the Europian Association for the Study of Diabetes. Diabetologia 2017; 60: Wentholt IM, Maran A, Masurel N, Heine RJ, Hoekstra JB, DeVries JH. Nocturnal hypoglycaemia in Type 1 diabetic patients, assessed with continuous glucose monitoring: frequency, duration and associations. Diabet Med 2007; 24: Høi-Hansen T, Pedersen-Bjergaard U, Thorsteinsson B. The Somogyi phenomenon revisited using continuous glucose monitoring in daily life. Diabetologia 2005; 48: Wiltshire EJ, Newton K, McTavish L. Unrecognised hypoglycaemia in children and adolescents with type 1 diabetes using the continuous glucose monitoring system: prevalence and contributors. J Paediatr Child Health 2006; 42: Novodvorsky P, Bernjak A, Chow E, Iqbal A, Sellors L, Williams S et al. Diurnal differences in risk of cardiac arrhythmias during spontaneous hypoglycemia in young people with type 1 diabetes. Diabetes Care 2017; 40: Veneman T, Mitrakou A, Mokan M, Cryer P, Gerich J. Induction of hypoglycemia unawareness by asymptomatic nocturnal hypoglycemia. Diabetes 1993; 42: Dagogo-Jack SE, Craft S, Cryer PE. Hypoglycemia-associated autonomic failure in insulin-dependent diabetes mellitus. Recent antecedent hypoglycemia reduces autonomic responses to, symptoms of, and defense against subsequent hypoglycemia. J Clin Invest 1993; 91: Seaquist ER, Chow LS. Hypoglycemia in diabetes: does insulin type matter? JAMA 2017; 318: Donnelly LA, Morris AD, Frier BM, Ellis JD, Donnan PT, Durant R et al. Frequency and predictors of hypoglycaemia in Type 1 and insulin-treated Type 2 diabetes: a population-based study. Diabet Med 2005; 22: Pedersen-Bjergaard U, Kristensen PL, Beck-Nielsen H, Nørgaard K, Perrild H, Christiansen JS et al. Effect of insulin analogues on risk of severe hypoglycaemia in patients with type 1 diabetes prone to recurrent severe hypoglycaemia (HypoAna trial): a prospective, randomised, open-label, blinded-endpoint crossover trial. Lancet Diabetes Endocrinol 2014; 2: Fredheim S, Johansen A, Thorsen SU, Kremke B, Nielsen LB, Olsen BS et al. Nationwide reduction in the frequency of severe hypoglycemia by half. Acta Diabetol 2015; 52: Kristensen PL, Pedersen-Bjergaard U, Beck-Nielsen H, Nørgaard K, Perrild H, Christiansen JS et al. A prospective randomised crossover study of the effect of insulin analogues and human insulin on the frequency of severe hypoglycaemia in patients with type 1 diabetes and recurrent hypoglycaemia (the HypoAna trial): study rationale and design. BMC Endocr Disord 2012; 12: Pedersen-Bjergaard U, Faerch L, Allingbjerg ML, Agesen R, Thorsteinsson B. The influence of new European Union driver s license legislation on reporting of severe hypoglycemia by patients with type 1 diabetes. Diabetes Care 2015; 38: ª 2018 Diabetes UK 69