Effect of Impaired Awareness of Hypoglycemia on Glucose Decline During and After Exercise in the T1DEXI Study

Abstract

Context

Adults with type 1 diabetes (T1D) face the necessity of balancing the benefits of exercise with the potential hazards of hypoglycemia.

Objective

This work aimed to assess whether impaired awareness of hypoglycemia (IAH) affects exercise-associated hypoglycemia in adults with T1D.

Methods

We compared continuous glucose monitoring (CGM)-measured glucose during exercise and for 24 hours following exercise from 95 adults with T1D and IAH (Clarke score ≥4 or ≥1 severe hypoglycemic event within the past year) to 95 “aware” adults (Clarke score ≤2 and no severe hypoglycemic event within the past year) matched on sex, age, insulin delivery modality, and glycated hemoglobin A_1c. A total of 4236 exercise sessions, and 1794 exercise days and 839 sedentary days, defined as 24 hours following exercise or a day without exercise, respectively, were available for analysis.

Results

Participants with IAH exhibited a nonsignificant trend toward greater decline in glucose during exercise compared to “aware” (−21 ± 44 vs −19 ± 43 mg/dL [−1.17 ± 2.44 vs −1.05 ± 2.39 mmol/L], adjusted group difference of −4.2 [95% CI, −8.4 to 0.05] mg/dL [−0.23 95% CI, −.47 to 0.003 mmol/L]; P = .051). Individuals with IAH had a higher proportion of days with hypoglycemic events below 70 mg/dL [3.89 mmol/L] (≥15 minutes <70 mg/dL [<3.89 mmol/L]) both on exercise days (51% vs 43%; P = .006) and sedentary days (48% vs 30%; P = .001). The increased odds of experiencing a hypoglycemic event below 70 mg/dL (<3.89 mmol/L) for individuals with IAH compared to “aware” did not differ significantly between exercise and sedentary days (interaction P = .36).

Conclusion

Individuals with IAH have a higher underlying risk of hypoglycemia than “aware” individuals. Exercise does not appear to differentially increase risk for hypoglycemia during the activity, or in the subsequent 24 hours for IAH compared to aware individuals with T1D.

Exercise has multiple well-established health benefits for individuals with type 1 diabetes (T1D) but can also lead to hypoglycemia (1). Fear of experiencing hypoglycemia poses a considerable barrier to physical activity in this population. Therefore, navigating the delicate balance between harnessing the benefits of exercise and mitigating the potential risks of hypoglycemia becomes a significant and recurrent challenge for individuals diagnosed with T1D (2). Nearly 1 in 5 individuals with T1D develops impaired awareness of hypoglycemia (IAH) during their lifetime (3). IAH is characterized by a reduced ability to detect hypoglycemia symptoms, and is associated with an at least 6-fold increased risk of experiencing severe hypoglycemic events (4). Both previous exposure to hypoglycemia and exercise can further impair the counterregulatory response to defend against a subsequent episode of hypoglycemia, further exacerbating IAH and heightening the risk of severe hypoglycemia (5). Despite substantial advancements in diabetes technologies, such as continuous glucose monitoring (CGM) with predictive and threshold alerts and alarms and automated insulin delivery systems (ie, hybrid closed-loop insulin delivery systems) with predictive suspension and resumption of insulin delivery, IAH remains highly prevalent among individuals with T1D (4).

Existing research has primarily investigated the alterations in counterregulatory hormone responses and glycemic patterns during exercise among individuals with T1D in controlled settings. These studies often involved assessment with hypoglycemic clamps, stringent exercise and glycemic management protocols, and close supervision by health care professionals (6, 7). Real-world data encompassing various types of activities, ranging from vigorous exercise to everyday household chores, while accounting for patient behavior variations, remain scarce. The T1DEXI study was specifically designed to evaluate glycemic patterns during and after exercise in individuals with T1D using CGM data (8). Our present study focuses on individuals with IAH and aims to determine whether glucose changes with exercise differ in those with IAH compared to those with intact awareness of hypoglycemia (denoted “aware’). We also sought to determine whether the risk of hypoglycemia in the 24 hours following exercise differed in those with IAH compared to aware, and whether exposure to hypoglycemia during the 24 hours prior to an exercise or sedentary day affected the incidence of hypoglycemia differently in those with IAH compared to aware.

Materials and Methods

Study Design and Population

The present analyses used data from the T1DEXI study, which enrolled participants from November 2019 to June 2021. The design and main results of the T1DEXI study have been previously published (8). In brief, the T1DEXI study was an observational cohort study of adults living with T1D (disease duration ≥2 years; age ≥18 years) who were randomly assigned to complete 1 of 3 types of “at-home” structured, video-guided exercise (aerobic, interval, or resistance), in addition to their usual types of physical activity, over a 4-week period. Each participant completed approximately 6 structured study-assigned exercise sessions in addition to their usual physical activities. To track their activities, patients used a study-developed smartphone application (T1DEXI app), which allowed them to enter information such as the activity time of day, duration, type, self-reported intensity rating, whether the activity was competitive, and timing since their last meal. All nonstudy-assigned activities were categorized to a general activity type (9). For these analyses, each instance of physical activity captured, including both structured study-assigned exercises and other forms of personal physical activity, is termed either “exercise” or “exercise session.” Both structured study-assigned and other forms of personal exercises were combined for analysis.

Sensor glucose data were captured with the participants’ personal Dexcom G6 CGM. If a participant did not use CGM at baseline or had a personal CGM other than Dexcom G6, then a blinded Dexcom G6 Pro CGM was provided for collection of sensor glucose data during the 4 weeks of observation.

For the present analyses, adult participants in the T1DEXI study were defined as having IAH based on a Clarke survey score of 4 or greater (10) or the occurrence of 1 or more severe hypoglycemic events requiring third-party assistance (11) within the year before study enrollment. Participants with IAH were compared to participants with intact awareness of hypoglycemia (defined here as “aware’) defined by a Clarke survey score of 2 or less and no severe hypoglycemic events within the past year (12). Participants in the “aware” group were selected to match those identified as having IAH based on sex, age (within ±10 years), insulin delivery modality, and glycated hemoglobin A_1c (HbA_1c) (within ±0.5%) at the time of enrollment. An institutional review board approved the study, and electronic informed consent was obtained from all participants.

Exercise sessions between 20 and 120 minutes were considered for analysis. Change in glucose during study and nonstudy exercise, defined as glucose level at end of exercise minus glucose level at the start of exercise, percentage of exercise sessions with a glucose value below 70 mg/dL(<3.89 mmol/L), and nadir glucose during the hour after exercise were compared between IAH and “aware” participants using a linear mixed-effects model adjusting for sex, age at enrollment, insulin delivery modality, HbA_1c, exercise type, exercise time of day, glucose at the start of exercise, and insulin on-board at the start of exercise as fixed effects (9), with a random matched-pair effect and a compound symmetry covariance structure to handle the repeated exercise events within a participant. Percentage of exercise sessions with a glucose value of less than 70 mg/dL (<3.89 mmol/L) during exercise were compared between the IAH vs aware groups using a similar logistic regression model.

Glycemic control on exercise days and sedentary days was assessed using consensus-based CGM metrics (13) both for the IAH and “aware” groups. An “exercise day” was defined as a 24-hour period following the completion of a study-assigned or personal exercise session not interrupted by another exercise session, while a “sedentary day” was defined as a 24-hour period without any exercise in the current or previous 24 hours. Participants were required to have at least 1 exercise and 1 sedentary day with 20 hours or more of CGM data on each day to be included in this analysis. A hypoglycemic event below 70 mg/dL (<3.89 mmol/L) was defined as at least 15 consecutive minutes with a CGM glucose below 70 mg/dL (<3.89 mmol/L) and a hypoglycemic event below 54 mg/dL(<3.00 mmol/L) was defined as at least 15 consecutive minutes with a CGM glucose of less than 54 mg/dL (<3.00 mmol/L) (14).

Statistical Analysis

Glycemic metrics on exercise days and separately on sedentary days were compared between the IAH and “aware” groups using a linear mixed-effects model adjusting for sex, age, insulin delivery modality, and HbA_1c with a random matched-pair effect and a compound symmetry covariance structure to handle the repeated days within a participant. For binary glycemic outcomes, a similar mixed-effects logistic regression model was used. If the IAH vs “aware” group difference in the percentage of exercise days with a hypoglycemic event below 70 mg/dL (<3.89 mmol/L) was significant, then an exercise type by hypoglycemic awareness group interaction effect on exercise days was assessed using a logistic regression model adjusting for the same covariates. For skewed continuous outcomes, a mixed-effects robust regression model adjusting for sex, age at enrollment, insulin delivery modality, and HbA_1c with a random participant effect was used (15). To assess if differences in CGM-measured glucose levels existed between IAH vs “aware” on exercise vs sedentary days, an interaction effect between group and day type (ie, exercise or sedentary) was added to each model and tested.

To assess the effect of antecedent hypoglycemia on the incidence of hypoglycemia, the proportion of exercise and sedentary days with a hypoglycemic event below 70 mg/dL (<3.89 mmol/L) following another hypoglycemic event below 70 mg/dL (<3.89 mmol/L) the day before vs 2 days before was tested using a mixed-effects logistic regression model adjusting for sex, age, insulin delivery modality, and HbA_1c as fixed effects with a random matched-pair effect and a compound symmetry covariance structure.

Data are expressed as median (interquartile range [IQR]) or mean ± SD unless otherwise noted. Multiple comparisons were corrected using the Benjamini-Hochberg adaptive false discovery rate correction procedure (16). Analyses were performed with SAS software, version 9.4 (SAS Institute), and R software, version 4.2.2 (17).

Results

We identified 95 adults with IAH and matched them with 95 “aware” adults in the T1DEXI cohort, for a total of 190 participants included in these analyses. Each group had 74% identifying as female, with a mean age of 43 ± 14 years, HbA_1c of 6.5 ± 0.7% (48 ± 7.7 mmol/mol), and with 36% using hybrid-closed loop insulin delivery systems also in each group (Table 1). The high proportion of female participants was similar to the 73% of participants identifying as female in the larger T1DEXI cohort (8). The number of women aged 45 years or older was similar in both groups (34 IAH and 31 aware). T1D duration was 16 (IQR: 7-34) years in the IAH group and 17 (IQR: 9-30) years in the “aware” group; body mass index was 25.4 ± 4.2 in the IAH group and 25.4 ± 4.0 in the “aware” group. The percentage of White non-Hispanic participants was 92% in the IAH group and 89% in the “aware” group.

Table 1.

Baseline characteristics of participants with and without impaired awareness of hypoglycemia

Characteristic	Hypoglycemia awareness status
	IAH (n = 95) (Clarke score ≥4 or SH in past y)	Aware (n = 95) (Clarke score ≤2 with no SH in past y)
Female, n (%)	70 (74)	70 (74)
Age, y	43 ± 14	43 ± 14
Type 1 diabetes duration, years	16 (7-34)	17 (9-30)
HbA1c, %	6.5 ± 0.7	6.5 ± 0.7
Body mass index	25.4 ± 4.2	25.4 ± 4.0
Race/Ethnicity
White non-Hispanic, n (%)	87 (92)	85 (89)
Non-White or Hispanic/Latino, n (%)	8 (8)	10 (11)
Highest education level
<Bachelor's, Bachelor's, >Bachelor's, n (%)	20 (21), 40 (42), 35 (37)	7 (7), 48 (51), 40 (42)
Insulin Delivery Modality
Hybrid closed-loop pump, multiple daily injections, standard insulin pump, n (%)	46 (48), 15 (16), 34 (36)	46 (48), 15 (16), 34 (36)
Current CGM user, n (%)a	94 (99)	93 (98)
IPAQ	n = 88	n = 90
IPAQ scoreb (MET-min/wk)	2437 (1246-4060)	2114 (1293-3459)
IPAQ category
Inactivec, minimally actived, HEPA activee, n (%)	9 (10), 39 (44), 40 (45)	7 (8), 49 (54), 34 (38)
Time spent sitting per wkf, h	42 (28-63)	49 (35-63)
Clarke IAH score	4.4 ± 1.2	1.4 ± 0.5
Impaired, Aware, n (%)	87 (92), 8 (8)	0 (0), 95 (100)
Most recent SH event in past yg
<1 y ago, ≥ 1 y ago, never, n (%)	14 (15), 40 (42), 41 (43)	0 (0), 36 (38), 59 (62)

Data presented as mean ± SD or median (interquartile range) unless otherwise indicated.

Abbreviations: IAH, impaired awareness of hypoglycemia; HbA_1c, glycated hemoglobin A_1c; CGM, continuous glucose monitoring; HEPA, health-enhancing physical activity; IPAQ, international physical activity questionnaire; MET, metabolic equivalent of task; SH, severe hypoglycemia.

^a CGM device type for participants at baseline, IAH: Abbot 3 (3%), Dexcom 83 (88%), Medtronic 8 (9%); Aware: Abbot 2 (2%), Dexcom 86 (92%), Medtronic 5 (5%).

^b Physical activity durations are calculated as the number of minutes/week each participant spent walking, in moderate physical activity, or in vigorous physical activity. The physical activity durations are multiplied by a MET score to calculate the MET-minutes/week: 3.3 for walking, 4.0 for moderate physical activity, and 8.0 for vigorous physical activity. The MET-minutes/week for all 3 physical activity types are then pooled for an overall IPAQ score.

^c Participant does not qualify as either minimally active or HEPA active.

^d Participant does not qualify as HEPA active but satisfies one or more of the following: a) 3 days or more of vigorous activity of 20 minutes per day or more, b) 5 days or more of moderate-intensity activity OR walking of 30 minutes per day or more, or c) 5 days or more of any combination of walking, moderate-intensity, or vigorous intensity activities totaling 600 MET-minutes/week or more.

^e Participant satisfies one or more of the following: a) 3 days or more of vigorous activity of 1500 MET-minutes per week or more, or b) 7 days or more of any combination of walking, moderate-intensity, or vigorous intensity activities totaling 3000 MET-minutes/week or more.

^f n = 164 (84 aware, 80 impaired) participants for IPAQ time spent sitting per week.

^g Severe hypoglycemia is defined as severe if the event required assistance of another person to actively administer carbohydrate, glucagon, or other resuscitative actions due to altered consciousness (11).

A total of 2081 exercise sessions from 94 IAH participants and 2155 exercise sessions from 94 “aware” participants were included in the analysis to assess the glucose responses to exercise and early recovery (ie, 1 hour post activity). Glucose at the start of exercise was 141 ± 48 mg/dL (7.83 ± 2.66 mmol/L) in the IAH group and 145 ± 47 mg/dL (8.05 ± 2.61 mmol/L) in the “aware” group. During exercise there was a nonsignificant trend toward greater decline in glucose in the IAH group compared to the “aware” group (−21 ± 44 vs −19 ± 43 mg/dL [−1.17 ± 2.44 vs −1.05 ± 2.39 mmol/L], adjusted group difference = −4.2 [95% CI, −8.4 to 0.02] mg/dL [−0.23; 95% CI, −0.47 to 0.003 mmol/L]; P = .051) with 12% and 8% of exercise sessions in the IAH and “aware” group having a glucose value of less than 70 mg/dL (<3.89 mmol/L) during exercise, respectively (P = .005). In the hour immediately after exercise, nadir glucose was comparable in the IAH and “aware” groups (105 ± 41 vs 110 ± 42 mg/dL [5.83 ± 2.28 vs 6.11 ± 2.33 mmol/L]; P = .08). The change in glucose during exercise and nadir glucose level 1 hour after exercise overall and by activity type are displayed in Fig. 1.

Figure 1.

Box plots of A, change in glucose during exercise by awareness group; B, change in glucose during exercise by awareness group and exercise type; C, nadir 1 hour after exercise by awareness group; and D, nadir 1 hour after exercise by awareness group and exercise type (n = 4236 exercise sessions from 188 participants). Black dots in the middle of boxes represent means, lines in the middle of boxes represent medians, and bottom and top of boxes represent the 25th and 75th percentile, respectively. Numbers on top of boxes represent the number of exercise sessions in each subdivision.

There were 81 IAH and 76 “aware” participants with sufficient data to compare CGM-measured glucose levels on exercise and sedentary days. The percentage of exercise days with a CGM-measured hypoglycemic event below 70 mg/dL (<3.89 mmol/L) (51% vs 43%; P = .006) or a hypoglycemic event below 54 mg/dL (<3.00 mmol/L) (15% vs 9%; P = .001) was higher in the IAH group when compared to the “aware” group (Table 2). Similar to exercise days, the percentage of sedentary days with a CGM-measured hypoglycemic event below 70 mg/dL (<3.89 mmol/L) (48% IAH vs 30% aware; P = .001) or a hypoglycemic event below 54 mg/dL (<3.00 mmol/L) (13% vs 8%; P = .12) was also higher in IAH when compared to “aware” participants. The percentage of overnight periods with a CGM-measured hypoglycemic event of less than 54 mg/dL (<3.00 mmol/L) was comparable in both groups. When exercise days were further stratified by activity type, there was no interaction found between activity type and hypoglycemia awareness status (P = .46; data not shown), thereby suggesting that the IAH group experiences higher rates of hypoglycemic events below 70 mg/dL (<3.89 mmol/L) than the “aware” group on exercise days regardless of the type of activity.

Table 2.

Glycemia on postexercise days vs sedentary days by hypoglycemia awareness status

	Exercise days			Sedentary days			Day type by awareness interaction Pa
	IAH	Aware	Pa	IAH	Aware	Pa
No. of participants	81	76		81	76
No. of d	909	885		405	434
H of glucose readings	24 (22-24)	24 (22-24)		24 (22-24)	24 (23-24)
Mean glucose, mg/dL (mmol/L)	137 ± 28 (7.60 ± 1.55)	140 ± 25 (7.77 ± 1.39)	.23	144 ± 29 (7.99 ± 1.61)	149 ± 29 (8.27 ± 1.61)	.06	.76
Coefficient of variation, %	28 ± 9	27 ± 8	.12	29 ± 8	27 ± 8	.01	.36
Time in range 70-180 mg/dL (3.89-9.99 mmol/L), %	79 ± 18	80 ± 17	.54	74 ± 19	75 ± 21	.90	.99
Time below 70 mg/dL (3.89 mmol/L), %	1.7 (0.0-4.9)	0.7 (0.0-3.5)	.051	1.4 (0.0- 4.9)	0.0 (0.0-2.1)	.004	.36
D with hypoglycemic event <70 mg/dL (<3.89mmol/L)b, %	51	43	.006	48	30	.001	.36
Time < 54 mg/dL (3.00 mmol/L), n (%)			.02			.04	.76
0%	660 (73)	715 (81)		314 (78)	376 (87)
>0%-<2%	147 (16)	110 (12)		53 (13)	34 (8)
≥2%	102 (11)	60 (7)		38 (9)	24 (6)
D with hypoglycemic event <54 mg/dL(3.00mmol/L)b, %	15	9	.001	13	8	.12	.68
Time > 180 mg/dL (9.99 mmol/L), %	11 (3-25)	12 (3-27)	.92	17 (6-36)	19 (6-36)	.35	.76
Time > 250 mg/dL (13.88 mmol/L), n (%)			.34			.38	.62
0%	618 (68)	613 (69)		222 (55)	253 (58)
>0-<2%	56 (6)	45 (5)		26 (6)	27 (6)
2-<5%	50 (6)	68 (8)		39 (10)	36 (8)
5-<10%	73 (8)	67 (8)		49 (12)	40 (9)
≥10%	112 (12)	92 (10)		69 (17)	78 (18)
Overnight (midnight to <6 Am) periods with hypoglycemic event <54 mg/dL (<3.00 mmol/L)b, %	4	2	.10	4	2	.18	.99

Data presented as mean ± SD or median (interquartile range).

Abbreviations: CGM, continuous glucose monitoring; HbA_1c, glycated hemoglobin A_1c; IAH, impaired awareness of hypoglycemia.

^a For continuous outcomes, P values for exercise day and sedentary day fixed effects are from a linear mixed-effects model adjusting for baseline HbA_1c, age, sex, and insulin delivery modality with a compound symmetry covariance structure to handle the repeated exercise events within subject and a random matched-pair effect. For binary outcomes, P values are from a logistic regression model adjusting for baseline HbA_1c, age, sex, and insulin delivery modality as fixed effects with a compound symmetry covariance structure to handle the repeated exercise events within subject and a random matched-pair effect (with the exception of the overnight hypoglycemia event outcome, which did not have a random matched-pair effect). Interaction effect P values are from testing an additional day type (exercise, sedentary) by hypoglycemia awareness status group interaction effect. Due to a skewed distribution, percentage time less than 70 mg/dL (<3.89 mmol/L), percentage time less than 54 mg/dL (<3.00 mmol/L), and percentage time greater than 250 mg/dL (>13.88 mmol/L) were tested using a mixed-effect robust regression model adjusting for baseline HbA_1c, age, sex, and insulin delivery modality with a random subject effect. Multiple comparisons were adjusted using the Benjamini-Hochberg adaptive false discovery rate correction procedure.

^b A CGM-defined hypoglycemic event less than 70 mg/dL (<3.89 mmol/L) is defined as at least 15 consecutive minutes with sensor glucose less than 70 mg/dL (<3.89 mmol/L). A CGM-defined hypoglycemic event less than 54 mg/dL (<3.00 mmol/L) is defined as at least 15 consecutive minutes with sensor glucose less than 54 mg/dL (<3.00 mmol/L).

Importantly, the increased amount of hypoglycemia exposure on exercise days when compared to sedentary days was not different between IAH and “aware” groups (ie, there were no significant group by day type interactions). Specifically, the IAH minus “aware” mean adjusted group difference in percentage time below 70 mg/dL (<3.89 mmol/L) was 0.6% (95% CI, −0.0% to 1.3%) on exercise days and 1.0% (95% CI, 0.3%-1.6%) on sedentary days (Fig. 2; group by day type interaction P = .36). Likewise, the adjusted odds of a CGM-measured hypoglycemic event below 70 mg/dL (<3.89 mmol/L) was 1.4 times greater in the IAH compared to “aware” group on exercise days and 1.9 on sedentary days with no difference between exercise vs sedentary days (group by day type interaction P = .36). Thus, the increased hypoglycemia observed in the IAH group, as compared with the “aware” group, was not significantly higher on days following exercise when compared to sedentary days.

Figure 2.

Box plots of percentage time less than 70 mg/dL (<3.89 mmol/L) on exercise vs sedentary days by hypoglycemia awareness status (n = 2633 days from 157 participants). Black dots in the middle of boxes represent means, lines in the middle of boxes represent medians, and bottom and top of boxes represent the 25th and 75th percentile, respectively. Numbers on top of boxes represent the number of days in each subdivision.

We next examined if exposure to a hypoglycemic event influenced the risk of developing a subsequent hypoglycemic event. To assess this, we compared the proportion of days with a hypoglycemic event if another hypoglycemic event occurred on the day prior vs 2 days prior. An exercise or sedentary day with a hypoglycemic event below 70 mg/dL (<3.89 mmol/L) occurred 49% of the time following exposure to a hypoglycemic event below 70 mg/dL (<3.89 mmol/L) the day prior compared with 42% of the time when there was a hypoglycemic event at less than 70 mg/dL (<3.89 mmol/L) 2 days prior (P = .18). Therefore, exposure to a hypoglycemic event below 70 mg/dL (<3.89 mmol/L) the day prior did not significantly increase the risk for experiencing a subsequent hypoglycemic event below 70 mg/dL (<3.89 mmol/L). An assessment of whether exposure to a hypoglycemic event below 54 mg/dL (<3.00 mmol/L) the day prior significantly increased the risk for experiencing a subsequent event below 54 mg/dL (<3.00 mmol/L) showed similar results. Within the IAH group, there was a higher proportion of days with a hypoglycemic event at less than 70 mg/dL (<3.89 mmol/L) following a hypoglycemic event below 70 mg/dL (<3.89 mmol/L) 1 day prior when compared to 2 days prior, although this nominally increased proportion was similar both for exercise and sedentary days. The “aware” group had a similar proportion of days with a hypoglycemic event below 70 mg/dL (<3.89 mmol/L) following a hypoglycemia event at less than 70 mg/dL (<3.89 mmol/L) 1 day prior when compared to 2 days prior (see Table 3). This suggests that in the IAH group, exercise itself does not increase the risk of experiencing a subsequent hypoglycemic event below 70 mg/dL (<3.89 mmol/L) after exposure to a hypoglycemic event at less than 70 mg/dL (<3.89 mmol/L) on prior days. In addition, the aware group did not seem to have an increased risk for experiencing another hypoglycemic event at less than 70 mg/dL (<3.89 mmol/L) after a day with a hypoglycemic event below 70 mg/dL (<3.89 mmol/L).

Table 3.

Proportion of exercise days and sedentary days with a hypoglycemic event less than 70 mg/dL (<3.89 mmol/L) based on preceding day with a hypoglycemic event and hypoglycemia awareness status

Day type	Hypoglycemia awareness group	% of d with a hypoglycemic event <70 mg/dL (<3.89 mmol/L)a following a hypoglycemic event
		1 d priorb	2 d priorb
Exercise days	IAH	58 (77/133)	48 (63/131)
	Aware	43 (55/128)	44 (52/118)
Sedentary Days	IAH	55 (33/60)	42 (27/65)
	Aware	35 (22/63)	31 (23/75)

Summary statistics are represented as percentage of days (number of days with hypoglycemic event/total number of days assessed).

Abbreviations: CGM, continuous glucose monitoring; IAH, impaired awareness of hypoglycemia.

^a A CGM-defined hypoglycemic event less than 70 mg/dL (<3.89 mmol/L) is defined as at least 15 consecutive minutes with sensor glucose less than 70 mg/dL (<3.89 mmol/L).

^b Hypoglycemic event less than 70 mg/dL (<3.89 mmol/L) following hypoglycemic event 1 day prior indicates an exercise or sedentary day with a hypoglycemic event when there was no hypoglycemic event 2 days before and a hypoglycemic event the prior day. Hypoglycemic event less than 70 mg/dL (<3.89 mmol/L) following a hypoglycemic event 2 days prior indicates an exercise or sedentary day with a hypoglycemic event when there was a hypoglycemic event 2 days before and no hypoglycemic event the prior day.

Differences in glycemic control and hyperglycemic metrics between IAH vs “aware” participants were not significantly different on exercise days vs sedentary days (all interaction P values >.05; see Table 2). Thus, there was no evidence that exercise itself had a meaningful effect on the IAH vs “aware” group differences for any glycemic metric.

Additionally, the comparison of IAH vs “aware” change in glucose during exercise, nadir 1-hour post exercise, and glycemic control on exercise vs sedentary days showed similar trends within each insulin delivery modality (ie, hybrid-closed loop pump, multiple daily injections, or standard insulin pump; Supplementary Tables S1 and S2) (18).

Three severe hypoglycemic events involving third-party assistance or loss of consciousness were reported in the IAH group, none of which were attributed to exercise.

Discussion

The findings of our study confirm that individuals with IAH have higher hypoglycemia exposure overall as compared to those who are “aware,” but also suggest that exercise itself does not increase the risk of hypoglycemia during activity or within the subsequent 24 hours for individuals with IAH compared with those who are “aware.” Participants with IAH exhibited a slightly lower baseline glycemia before exercise and tended toward a greater decline in glucose during exercise with slightly lower nadir glucose in the hour following exercise compared to aware participants. However, the estimated small differences highlight that the glucose responses during and immediately following exercise were similar between groups in the context of real-world physical activities.

While individuals with T1D and IAH have a higher underlying risk of hypoglycemia, with a higher proportion of hypoglycemic events (glucose <70 mg/dL [<3.89 mmol/L]) overall both during physically active days and sedentary days, our study demonstrates that exercise itself does not appear to increase this risk. This underscores the importance of emphasizing that exercise remains a safe and valuable component of an overall health regimen, and that increased glucose monitoring with real-time CGM is essential for safety and hypoglycemia mitigation both on active and inactive days. This finding holds true when evaluating clinically significant, serious hypoglycemia (glucose <54 mg/dL [<3.00 mmol/L]) both during the day and overnight between midnight and 6 Am, both of which are known to occur more frequently in individuals with T1D following exercise (19).

Prior research indicates that antecedent hypoglycemia can impair the counterregulatory defense against hypoglycemia during subsequent exercise through decreased epinephrine secretion as well as a decrease in endogenous glucose production in response to hypoglycemia (20). Other work has also shown a correlation between the severity of antecedent hypoglycemia and the magnitude of the blunting of the counterregulatory response, an effect that is seen even with exposure to mild hypoglycemia starting at less than 70 mg/dL (<3.89 mmol/L) (7). Our findings differ from the literature in which we found insufficient evidence to suggest that prior recent hypoglycemia significantly increased the risk for experiencing subsequent hypoglycemia in the overall cohort. However, a further subgroup analysis found that IAH participants may have a slightly elevated risk of hypoglycemia following antecedent exposure to hypoglycemia both on exercise days and sedentary days, while “aware” participants had similar risk of hypoglycemia irrespective of recent hypoglycemia exposure. Therefore, it is possible that a heightened risk of hypoglycemia as a consequence of antecedent exposure to hypoglycemia may be differentially influenced by IAH or the overall greater exposure to hypoglycemia in the IAH group but not by exercise itself. Thus, while it is imperative that patients with IAH in particular practice caution when engaging in physical activity if they have experienced a hypoglycemic episode within the preceding 24 hours, it is also crucial to emphasize that the elevated risk of experiencing a hypoglycemic event remains present irrespective of exercise, and therefore should not deter individuals from participating in physical activities. This knowledge should inform and empower individuals with diabetes to make informed decisions regarding their exercise routines, ultimately contributing to safer and more effective diabetes management strategies.

While CGM devices have reduced the burden and severity of hypoglycemic events in patients with T1D and IAH, they have not shown significant improvements in the counterregulatory deficits observed in patients with IAH (21). Equally important, CGMs are subject to a physiological lag time between the equilibration of interstitial and capillary blood glucose levels resulting in reduced accuracy during periods of dynamic glucose changes, such as those encountered with exercise (22, 23). Nonetheless, automated insulin delivery using hybrid-closed loop systems engineered to emulate physiologic insulin delivery have rapidly emerged as the standard of care for patients with T1D, with some evidence that they may be protective against exercise-associated dysglycemia (24). Multiple studies have demonstrated their efficacy in improving glycemic control and mitigating hypoglycemic events overall (25). However, hypoglycemia prevention during exercise remains a considerable challenge as automated insulin delivery systems cannot respond rapidly enough to the increased glucose utilization during exercise to prevent hypoglycemia (26) primarily due to the delayed appearance of subcutaneous insulin in the circulation and the relatively slow on and off kinetics of currently available insulin delivered in the subcutaneous space. In our study, glycemic differences between IAH vs “aware” during exercise or the subsequent 24 hours were similar among the different insulin delivery modalities, which encompassed hybrid closed-loop pumps, multiple daily injections, or standard insulin pumps. This suggests that the modality of insulin delivery did not have an interaction with the IAH vs “aware” glycemic differences during exercise in our cohort.

Our study exhibits some limitations that warrant consideration. The participants in our study exhibited good glycemic control at baseline and maintained high levels of physical activity, which may not reflect the overall characteristics of individuals with T1D (27) and differs from previously described studies in patients with IAH (28, 29). Additionally, there was an overrepresentation of women and an underrepresentation of minority groups in our sample, which limits the generalizability of our findings to more diverse populations. Despite our efforts to identify patients with IAH using the Clarke survey and including individuals who experienced severe hypoglycemic events in the year prior to enrollment, it is crucial to recognize the limitations in relying on patient self-reporting and the potential for recall bias. Furthermore, the overall incidence of CGM-measured hypoglycemic events in the T1DEXI study cohort was low, potentially limiting the generalizability of our findings to populations experiencing more frequent hypoglycemic events. Most participants in the T1DEXI study used real-time CGM, which may help reduce the risk for hypoglycemia in those with IAH (21), and CGM may also be useful for the prevention of exercise-associated hypoglycemia if individuals are educated and motivated (30). Lastly, our study did not specifically assess differences in patient behaviors or insulin management prior to exercise, and there remains the possibility that IAH participants approach exercise differently from “aware” participants. Nevertheless, we did not observe significant interactions in the difference in overall glycemic control or glycemic variability between the IAH and “aware” groups on exercise and sedentary days, suggesting that behaviors intended to mitigate the risk of hypoglycemia during exercise did not lead to differences in hyperglycemia or glycemic variability observed between the groups.

In conclusion, this study represents the largest investigation to date on the glycemic patterns of patients with T1D affected by IAH during and after exercise using real-world data. Encouragingly, our findings indicate that, in patients with T1D and IAH, exercise does not differentially increase the risk of hypoglycemia during the activity or in the 24 hours following an exercise session. These results provide a solid foundation for health care providers to actively promote exercise rather than restrict physical activity in this population. Moreover, our findings can inform the development of practice guidelines aimed at improving the care of individuals with T1D affected by hypoglycemia unawareness.

Abbreviations

CGM

continuous glucose monitoring

HbA_1c

glycated hemoglobin A_1c

IAH

impaired awareness of hypoglycemia

IQR

interquartile range

T1D

type 1 diabetes

Disclosures

J.L.J.K. reports no disclosures. Z.L. reports no disclosures. R.L.G. reports no disclosures. J.R.C. reports receiving grants from the Juvenile Diabetes Research Foundation (JDRF), the National Institutes of Health (NIH), Dexcom, and Medtronic and receiving consultancy fees from NovoNordisk and Zealand, outside the submitted work. F.J.D. reports no disclosures. P.G.J. reports receiving grants from the National Institutes of Health, the Leona M. and Harry B. Charitable Trust, the JDRF, Dexcom, and the Oregon Health & Science University Foundation; consultancy fees from CDISC; US patents 62/352,939, 63/269,094, 62/944,287, 8810388, 9,480,418, 8,317,700, 61/570382, 8,810,388, 7,976,466, and 6,558,321; and reports stock options from Pacific Diabetes Technologies, outside the submitted work. C.K.M. reports research funding from Evidation Health, The Leona M. and Harry B. Helmsley Charitable Trust, State of Louisiana- Federal American Rescue Plan, USDA, The Henry M. Jackson Foundation for the Advancement of Military Medicine Inc, University of Rochester, Foundation for Food and Agriculture Research–Kroger Co. Zero Hunger/Zero Waste Foundation, National Institute for Health Research, WW, American Diabetes Association, NSF, Lilly, and the NIH. C.K.M. receives honoraria for giving scientific talks to academic institutions and nonprofit organizations, including the Obesity Action Coalition, Indiana University Bloomington, University of Alabama Birmingham, and the University of Kansas Medical Center. C.K.M. also served on a data and safety monitoring board for an NIH-funded trial at Duke University, is a paid external advisory board member for the University of Alabama Birmingham's NIH-funded Nutrition Obesity Research Center, and he was a paid member of the planning committee for the Bray Obesity Science Summit. He is also a paid facilitator for continuing education events sponsored by the Commission on Dietetic Registration, he is a paid mentor on an NIH-training program at the University of Nebraska Lincoln, he served on an advisory board for EHE Health, and he serves on an advisory board for Wondr Health. Finally, Dr Martin receives royalties and is on patent applications for a weight management platform and his institution owns the intellectual property for, and he is an inventor of, smartphone apps to quantify food intake–related behaviors. R.W.B. reports receiving consulting fees, paid to his institution, from Insulet, Bigfoot Biomedical, vTv Therapeutics, and Eli Lilly; grant support and supplies, provided to his institution, from Tandem and Dexcom; and supplies from Ascenia and Roche. P.C. reports no disclosures. M.C.R. reports receiving consulting fees from the Jaeb Center for Health Research, Eli Lilly, Zealand Pharma, and Zucara Therapuetics’; speaker fees from Sanofi Diabetes, Eli Lilly, Dexcom Canada, and Novo Nordisk; and stock options from Supersapiens and Zucara Therapeutics. M.R.R. reports consultancy fees from Zealand Pharma.

Data Availability

The study data are available on the Vivli platform (https://doi.org/10.25934/PR00008429).