Post–bariatric surgery hypoglycemia (PBH) is a metabolic complication of Roux-en-Y gastric bypass (RYGB). Since symptoms are a key component of the Whipple’s triad to diagnose nondiabetic hypoglycemia, we evaluated the relationship between self-reported symptoms and postprandial sensor glucose profiles.
Thirty patients with PBH after RYGB (age: 50.1 [41.6–60.6] years, 86.7% female, BMI: 26.5 [23.5–31.2] kg/m2; median [interquartile range]) wore a blinded Dexcom G6 sensor while recording autonomic, neuroglycopenic, and gastrointestinal symptoms over 50 days. Symptoms (overall and each type) were categorized into those occurring in postprandial periods (PPPs) without hypoglycemia, or in the preceding dynamic or hypoglycemic phase of PPPs with hypoglycemia (nadir sensor glucose <3.9 mmol/L). We further explored the relationship between symptoms and the maximum negative rate of sensor glucose change and nadir sensor glucose levels.
In 5,851 PPPs, 775 symptoms were reported, of which 30.6 (0.0–59.9)% were perceived in PPPs without hypoglycemia, 16.7 (0.0–30.1)% in the preceding dynamic phase and 45.0 (13.7–84.7)% in the hypoglycemic phase of PPPs with hypoglycemia. Per symptom type, 53.6 (23.8–100.0)% of the autonomic, 30.0 (5.6–80.0)% of the neuroglycopenic, and 10.4 (0.0–50.0)% of the gastrointestinal symptoms occurred in the hypoglycemic phase of PPPs with hypoglycemia. Both faster glucose dynamics and lower nadir sensor glucose levels were related with symptom perception.
The relationship between symptom perception and PBH is complex, challenging clinical judgement and decision-making in this population.
Introduction
Post–bariatric surgery hypoglycemia (PBH) is a metabolic complication characterized by meal-induced hypoglycemic episodes (1,2). The reported prevalence of this condition ranges widely in the literature (6.6–88.0%) depending on the diagnostic methodology used (3,4) and appears to occur more commonly after Roux-en-Y gastric bypass (RYGB) than after other surgical procedures (4,5).
In line with the workup for other forms of nondiabetic hypoglycemia (6), the presence of the Whipple’s triad (low plasma glucose, clinical symptoms consistent with hypoglycemia and their relief after intake of glucose) should be documented as part of the evaluation of PBH (7–9). Despite the emphasis on the presence of symptoms, symptom perception remains largely unexplored in this population. Continuous glucose monitoring (CGM) in combination with applications for symptom logging (10) offers a way to investigate the sensor glucose profile and relate it to the timing and quality of symptom perception during daily living.
The objective of this work was to gain insights into the relationship between postprandial symptoms and glycemic profiles in patients with PBH by examining blinded CGM recordings and self-reported symptoms over a 50-day free-living period.
Research Design and Methods
Study Design, Population, and Procedures
Data were collected as part of an observational clinical study (NCT05212207) conducted at the University Hospital of Bern (Switzerland). The study included individuals aged ≥18 years who underwent RYGB and had biochemically confirmed PBH (defined as a postprandial plasma or sensor glucose of <3.0 mmol/L, as suggested by the International Hypoglycemia Study Group [11], accompanied by symptoms). Exclusion criteria were other causes of hypoglycemia, pregnancy or lactation, and inability to provide written informed consent or follow study procedures. The study was approved by the Ethics Committee Bern (2021-02086) and was conducted in accordance with the Declaration of Helsinki, the principles of Good Clinical Practice, and the local legal requirements. Participants provided written informed consent prior to study-related procedures.
Any pharmacological agents interfering with glucose regulation were stopped prior to study-related procedures. At baseline, the height, weight, and glycated hemoglobin were measured, and participants’ medical history was retrieved. Participants were asked to wear a blinded Dexcom G6 sensor (Dexcom, Inc., San Diego, CA) for 50 days and were instructed to perform a calibration every other day in a fasted state. The Dexcom G6 communicated to a mobile app on the participants’ smartphones, which allowed for the recording of meals and symptoms, using predefined categories with examples: autonomic, neuroglycopenic, and gastrointestinal. The symptom categorization aligned with the Edinburgh Hypoglycemia Symptom Scale and the Sigstad-Dumping Score (12,13), which were used for investigations in the PBH population previously (14–16). The intensity of symptoms was rated on a Likert Scale from 1 to 10. The symptom logging interface of the mobile app is shown in Supplementary Fig. 1. No dietary instructions were given to the participants so that the recording period is representative of usual eating habits and lifestyle patterns.
CGM Data Preprocessing
CGM raw data files were retrieved from the online platform and stored as comma-separated values files before analysis. The rate of sensor glucose change (ROC) was calculated every 5 min using data from the preceding 15 min (ROCt = [sensor glucoset – sensor glucoset-15min]/15). The sensor glucose levels outside the Dexcom G6 reportable range (2.2–22.2 mmol/L) were set to the respective lowest/highest range value.
Identification of Postprandial Periods
Individual sensor glucose curves, meal, and symptom recordings were visualized using a software program implemented in R (version 4.0.1; The R Foundation for Statistical Computing, Vienna, Austria) to manually mark the meal onset on the sensor glucose curve based on the typical meal-induced sensor glucose pattern (see Supplementary Fig. 2). Data review was performed by two independent clinical researchers, skilled in the interpretation of CGM data of post–bariatric surgery patients and the underlying physiology. Criteria for meal-induced sensor glucose excursions were as follows: sensor glucose increment of >1.0 mmol/L within 60 min and subsequent sensor glucose decrease. The meal markings of the two clinical researchers were compared, and only meal markings that were set by both researchers (within 15 min of each other) were included in the analysis. Because of obvious underreporting of meals by participants, the primary analysis was performed using the meal markings of the clinical researchers. In addition, a sensitivity analysis using the participants’ meal recordings was performed (results are displayed in the Supplementary Material).
Postprandial periods (PPPs) were defined as the 3 h following the marked meal onset. In the event of multiple meals within less than 3 h, each PPP extended up to the next marked meal onset. PPPs with less than six available sensor glucose values were excluded from the analysis. We also excluded PPPs if the nadir sensor glucose preceded the peak sensor glucose and was <3.9 mmol/L. This was done to avoid misclassification of these PPPs due to the preceding hypoglycemic sensor value (see below for classification). An example of such a case is provided in Supplementary Fig. 2.
Assessment of Symptomatic PPPs
PPPs which contained participant-recorded symptoms were defined as symptomatic. Postprandial symptoms were further classified as those occurring in PPPs without hypoglycemia (nadir sensor glucose ≥3.9 mmol/L) and PPPs with hypoglycemia (nadir sensor glucose <3.9 mmol/L). If symptoms occurred in PPPs with hypoglycemia, we assessed whether the documentation occurred at the time of hypoglycemic sensor values (hypoglycemic phase) or during the preceding dynamic phase. The hypoglycemic phase was defined as the time period starting 10 min before the first sensor glucose of <3.9 mmol/L (to account for a possible sensor lag time) until 30 min after sensor glucose returned to levels of >3.9 mmol/L (to account for possible delays in symptom recording by the participant). In cases where postnadir sensor glucose values remained at <3.9 mmol/L, the hypoglycemic phase was extended up to 30 min after the end of the postprandial time window. In cases where less than six postnadir sensor glucose values were available, the hypoglycemic phase was extended up to 30 min after the nadir sensor glucose. The dynamic phase was defined as the time period between the meal marking and the start of the hypoglycemic phase. The definition of the hypoglycemic and dynamic phases is illustrated in Supplementary Fig. 3. The percentage of all symptoms (and each type of symptom) occurring in PPPs without hypoglycemia, or in the dynamic and hypoglycemic phase of PPPs with hypoglycemia, was calculated.
In order to assess the relation of postprandial glucose dynamics and nadir sensor glucose levels with symptom perception, all PPPs (irrespective of the presence of symptoms or hypoglycemia) were classified based on the nadir sensor glucose and the maximum negative ROC levels. For nadir sensor glucose levels, PPPs were categorized as without hypoglycemia, level 1 hypoglycemia (nadir sensor glucose <3.9 mmol/L), and level 2 hypoglycemia (nadir sensor glucose <3.0 mmol/L). According to the maximum negative ROC level, PPPs were split into three equally sized categories, which resulted in the following categories: ROC1 (0.00–0.11 mmol/L/min; slow), ROC2 (0.11–0.19 mmol/L/min; medium), and ROC3 (0.19–0.58 mmol/L/min; fast). The percentage of symptomatic PPPs was calculated for each category (and combination thereof) over all symptoms and for each type of symptom.
Statistical Analysis
All proportions were calculated on a participant level, and the median (interquartile range) across participants is presented unless otherwise specified. In a subgroup analysis, we focused on participants recording ≥5 postprandial symptoms during the study period. We also performed separate analyses for each type of symptom (autonomic, neuroglycopenic and gastrointestinal), considering participants with ≥5 occurrences per type. The relation of nadir sensor glucose and maximum negative ROC to postprandial symptom perception was assessed using a logistic mixed-effects model with a random intercept for participants. The programming software R (version 4.0.1; The R Foundation for Statistical Computing, Vienna, Austria) was used for the analysis, and figures were created using Prism (version 9.2.0; GraphPad Software Inc., San Diego) and Adobe Illustrator (version 24.0.2; Adobe, San Jose) software.
Data and Resource Availability
The data set generated and analyzed during the present work is available from the corresponding author upon reasonable request.
Results
Baseline Characteristics
Data from 30 patients with a diagnosis of PBH after RYGB (age: 50.1 [41.6–60.6] years, 86.7% female, BMI: 26.5 [23.5–31.2] kg/m2, total weight loss: 34.9 [26.0–46.1]%) were included in the analysis. The duration of recordings was 48 (42–50) days. Deviations from the intended 50 days of data collection were attributable to connectivity issues and hesitation to wear a blinded CGM for such a long time period (the case for two participants). Ten participants were on prior pharmacotherapy for PBH, which was stopped because of lack of efficacy and/or intolerance. Two participants were on current treatment with semaglutide and empagliflozin, which was stopped 1 month and 1 week prior to study-related procedures, to meet study eligibility criteria. Six participants had a history of invasive treatment for PBH (surgical or endoscopic intervention), dating back several years for the surgery and more than 6 months for the endoscopy. Findings of a sensitivity analysis excluding these six participants are shown in Supplementary Table 1 and Supplementary Fig. 4. Fourteen participants had a history of severe hypoglycemia, defined as hypoglycemia-related severe cognitive impairment requiring external assistance for recovery (17). An overview of baseline characteristics is provided in Table 1. The baseline characteristics of the subgroups reporting more or less than five symptoms of each type during the study period are shown in Supplementary Table 2.
Baseline characteristics
Parameter . | n = 30 . |
---|---|
Age, years | 50.1 (41.6–60.6) |
Sex: female | 26 (86.7) |
Current weight, kg | 70.5 (62.5–95.4) |
Current BMI, kg/m2 | 26.5 (23.5–31.2) |
Presurgery weight, kg | 118.0 (101.0–131.5) |
Presurgery BMI, kg/m2 | 41.0 (38.4–43.5) |
Absolute total weight loss, kg | 42.9 (32.2–49.7) |
Relative total weight loss, % | 34.9 (26.0–46.1) |
Time since bariatric surgery, years | 6.9 (4.2–9.4) |
Time since initial PBH diagnosis, years | 2 (0–4) |
HbA1c, % (mmol/mol) | 5.3 (5.2–5.7) (34.4 [33.3–38.8]) |
Pharmacotherapy for PBH | 12 (40.0) |
Acarbose | 8 (26.7) |
Calcium channel blockers | 0 (0.0) |
Somatostatin analogs | 1 (3.3) |
GLP-1 analogs | 5 (16.7) |
DPP-4 inhibitors | 0 (0.0) |
SGLT-2 inhibitors | 2 (6.7) |
Previous invasive treatment for PBH | 6 (20.0) |
Endoscopic transoral outlet reduction | 5 (16.7) |
Pouch revision | 1 (3.3) |
History of severe neuroglycopenia | 14 (46.7) |
Syncope | 8 (26.7) |
Seizure | 6 (20.0) |
Hospitalization | 5 (16.7) |
Frequency of symptomatic PBH events | |
Less than one per year | 0 (0.0) |
One per year to one per semester | 1 (3.3) |
One per semester to one per trimester | 0 (0.0) |
One per trimester to one per month | 0 (0.0) |
One per month to one per week | 8 (26.7) |
More than one per week | 15 (50.0) |
Every day | 2 (6.7) |
More than one per day | 4 (13.3) |
Parameter . | n = 30 . |
---|---|
Age, years | 50.1 (41.6–60.6) |
Sex: female | 26 (86.7) |
Current weight, kg | 70.5 (62.5–95.4) |
Current BMI, kg/m2 | 26.5 (23.5–31.2) |
Presurgery weight, kg | 118.0 (101.0–131.5) |
Presurgery BMI, kg/m2 | 41.0 (38.4–43.5) |
Absolute total weight loss, kg | 42.9 (32.2–49.7) |
Relative total weight loss, % | 34.9 (26.0–46.1) |
Time since bariatric surgery, years | 6.9 (4.2–9.4) |
Time since initial PBH diagnosis, years | 2 (0–4) |
HbA1c, % (mmol/mol) | 5.3 (5.2–5.7) (34.4 [33.3–38.8]) |
Pharmacotherapy for PBH | 12 (40.0) |
Acarbose | 8 (26.7) |
Calcium channel blockers | 0 (0.0) |
Somatostatin analogs | 1 (3.3) |
GLP-1 analogs | 5 (16.7) |
DPP-4 inhibitors | 0 (0.0) |
SGLT-2 inhibitors | 2 (6.7) |
Previous invasive treatment for PBH | 6 (20.0) |
Endoscopic transoral outlet reduction | 5 (16.7) |
Pouch revision | 1 (3.3) |
History of severe neuroglycopenia | 14 (46.7) |
Syncope | 8 (26.7) |
Seizure | 6 (20.0) |
Hospitalization | 5 (16.7) |
Frequency of symptomatic PBH events | |
Less than one per year | 0 (0.0) |
One per year to one per semester | 1 (3.3) |
One per semester to one per trimester | 0 (0.0) |
One per trimester to one per month | 0 (0.0) |
One per month to one per week | 8 (26.7) |
More than one per week | 15 (50.0) |
Every day | 2 (6.7) |
More than one per day | 4 (13.3) |
Continuous variables are shown as median (interquartile range); categorical variables are shown as n (percentage). Four participants had a history of two different treatment agents in the past. Five participants had a history of different events of severe neuroglycopenia. HbA1c, glycated hemoglobin; GLP-1, glucagon-like peptide 1; DPP-4, dipeptidyl peptidase 4; SGLT-2, sodium–glucose cotransporter 2.
Symptomatic PPPs
The workflow of the classification of PPPs is shown in Fig. 1. A total of 5,851 PPPs (182 [146–249] per participant, 5 [4–5] per participant per day) were recorded, of which 74.1 (67.8–83.1)% were without hypoglycemia. A total of 775 postprandial symptoms (15 [6–29] per participant, corresponding to 3 [1–5] symptoms per participant per week) were reported. Of these, 39.4 (31.0–52.9)% were autonomic, 33.3 (23.4–42.1)% were neuroglycopenic, and 20.0 (1.2–31.4)% were gastrointestinal. The median of the mean intensity was 4.6 (3.6–5.6), 4.4 (3.6–5.6), and 3.8 (3.0–4.6), respectively. One participant reported exclusively neuroglycopenic symptoms, while two participants reported exclusively autonomic symptoms. Four participants reported only autonomic and neuroglycopenic symptoms, while two participants reported only autonomic and gastrointestinal symptoms. The other participants reported a combination of different types of symptoms. A median percentage of 30.6 (0.0–59.9)% of all postprandial symptoms was perceived in PPPs without hypoglycemia. Further, 16.7 (0.0–30.1)% of all postprandial symptoms were recorded during the preceding dynamic phase of PPPs with hypoglycemia, and 45.0 (13.7–84.7)% were perceived during the hypoglycemic phase. When analyzed within each type of symptom, 53.6 (23.8–100.0)% of the autonomic symptoms, 30.0 (5.6–80.0)% of the neuroglycopenic symptoms, and 10.4 (0.0–50.0)% of the gastrointestinal symptoms were reported in the hypoglycemic phase of PPPs with hypoglycemia. Results are illustrated in Fig. 2 (exact numbers are reported in Supplementary Table 3). The subgroup analysis focusing on participants with ≥5 symptoms during the study period revealed similar patterns, with less neuroglycopenic symptoms in the hypoglycemic phase of PPPs with hypoglycemia (12.9 [6.3–50.5]%).
Percentages of symptoms (overall and per type of symptom) occurring in PPPs with (red) and without (white) hypoglycemia and during the preceding dynamic (red with white filled pattern) or hypoglycemic (red with black filled pattern) phase of PPPs with hypoglycemia. The top of the bars shows the median percentage, and the whiskers correspond to the interquartile range.
Percentages of symptoms (overall and per type of symptom) occurring in PPPs with (red) and without (white) hypoglycemia and during the preceding dynamic (red with white filled pattern) or hypoglycemic (red with black filled pattern) phase of PPPs with hypoglycemia. The top of the bars shows the median percentage, and the whiskers correspond to the interquartile range.
In the analysis using the participants’ meal recordings, the PPPs were numerically less (n = 4,419, 147 [81–212] per participant, 3 [2–4] per day), but results on symptom perception overall and per type of symptom were comparable to the primary analysis. A median percentage of 33.3 (0.0–66.1)% of all postprandial symptoms was perceived in PPPs without hypoglycemia, 15.0 (0.0–29.7)% during the preceding dynamic phase, and 34.8 (13.3–75.0)% during the hypoglycemic phase of PPPs with hypoglycemia. Detailed results are reported in Supplementary Table 4.
Classification of PPPs According to the Nadir Sensor Glucose and Maximum Negative ROC
As shown in Fig. 3, when all PPPs (irrespective of the presence of symptoms) were classified based on the nadir sensor glucose, the highest frequency of symptoms was observed in PPPs with level 2 hypoglycemia (18.9 [8.7–30.3]%). Similarly, the highest frequency of symptoms was observed in PPPs with the fastest maximum negative ROC (10.8 [5.1–13.4]%). The combined category with the lowest nadir sensor glucose and the fastest maximum negative ROC (level 2–ROC3) had the highest frequency of symptomatic PPPs, 39.4 (11.8–81.3)%. Although the lowest frequency of symptoms was observed during PPPs without hypoglycemia, because of the high number of these PPPs relative to PPPs with hypoglycemia, the proportion of all postprandial symptoms in PPPs without hypoglycemia was still 30.6 (0.0–59.9)% (as shown above). In addition, for a 1 mmol/L decrease in nadir sensor glucose, the odds of experiencing symptoms increased by 92.8% (odds ratio: 1.928, 95% CIs [1.692, 2.20]) and for a 0.1 mmol/L/min faster maximum negative ROC, the odds of experiencing symptoms increased by 49.0% (odds ratio on the original scale: 54.058, 95% CIs [14.591, 199.48]).
Percentage of symptomatic PPPs according to the nadir sensor glucose level (A) and maximum negative ROC (B). No hypo, level 1, and level 2 correspond to PPPs with nadir sensor glucose of ≥3.9 mmol/L, <3.9 mmol/L, and <3.0 mmol/L, respectively. ROC1, ROC2, and ROC3 correspond to PPPs with the slowest, medium, and fastest maximum negative ROC, respectively. The top of the bars shows the median percentage, and the whiskers correspond to the interquartile range.
Percentage of symptomatic PPPs according to the nadir sensor glucose level (A) and maximum negative ROC (B). No hypo, level 1, and level 2 correspond to PPPs with nadir sensor glucose of ≥3.9 mmol/L, <3.9 mmol/L, and <3.0 mmol/L, respectively. ROC1, ROC2, and ROC3 correspond to PPPs with the slowest, medium, and fastest maximum negative ROC, respectively. The top of the bars shows the median percentage, and the whiskers correspond to the interquartile range.
Similarly, when the participants’ meal recordings were analyzed, the highest frequency of symptoms was observed in PPPs with level 2 hypoglycemia (23.6 [13.6–48.6]%) and the fastest maximum negative ROC (11.0 [5.3–21.7]%). Detailed results are reported in Supplementary Fig. 5. The results of the analyses performed separately for each type of symptom are shown in Supplementary Fig. 6.
Conclusions
In this work, we explored the relationship between postprandial symptoms and glycemic profiles in patients with PBH after RYGB. Although both postprandial rapid glucose dynamics and nadir glucose levels were associated with symptom perception, more than one-third of all postprandial symptoms were perceived in PPPs without hypoglycemia. In addition, even in PPPs with hypoglycemia, a considerable proportion of symptoms were perceived in the preceding dynamic phase when glucose values were above the hypoglycemic threshold. When considering neuroglycopenic symptoms, which are considered most suggestive of hypoglycemia, more than two-thirds of them were perceived either in PPPs without hypoglycemia or in the preceding dynamic phase.
Our analysis questions the specificity of hypoglycemic symptoms, since less than half of all symptoms occurred during the hypoglycemic phase of PPPs with hypoglycemia. Furthermore, our results suggest that glucose dynamics, independent of low sensor glucose levels, must be involved in the initiation of symptoms. This is in line with a previous study using flash glucose monitoring and a symptom diary for 14 days in nine patients with symptomatic PBH after RYGB, which found that only one-third of the symptoms was perceived at low glucose levels (18). In part, these findings could be explained by the early dumping syndrome, which is characterized by volume contraction and sympathetic activation in the early PPP, leading to autonomic and gastrointestinal symptoms (7). Previous work observed that patients with PBH after RYGB also tested positive for signs of early dumping syndrome (early postprandial rise in heart rate and hemoconcentration) and suggested a mechanistic link between the two conditions (19). The common denominator could be the rapid postprandial nutrient fluxes in RYGB-operated subjects (20), which may cause discomfort on a multisystem level and even trigger neuroglycopenic-like symptoms, despite euglycemic glucose levels.
A previous study using a provocative test in 32 patients with PBH showed that specificity for hypoglycemia was higher for neuroglycopenic than autonomic symptoms (21). This is in contrast with our work, which suggests that autonomic symptoms were more prevalent than neuroglycopenic symptoms during periods with low glucose levels. Although the reasons for this discrepancy remain speculative, it may be partially explained by differences in the research methodology (e.g., instruction of the participants, outpatient versus in-clinic assessments) and study populations (e.g., comedication, comorbidities, diagnostic criteria, hypoglycemia burden immediately preceding the evaluation).
Our study underscores that the relationship between symptom perception and postprandial glucose patterns in patients with PBH after RYGB is complex. In addition to issues related to specificity, symptom perception was also highly variable in our study population, as reflected by the wide interquartile ranges. Some of this variability may be explained by the heterogeneity of the study population. Given our observation that symptoms may be generated by physiological phenomena other than low glucose levels, one could consider classifying postprandial symptoms without hypoglycemia as postprandial syndrome as compared with symptoms that are accompanied by hypoglycemia or hypoglycemia without symptoms. Considering that patients with confirmed PBH also perceive symptoms at times without hypoglycemia challenges the use of the Whipple’s triad in this population. On the other hand, our findings may underscore the significance of the perception of hypoglycemic symptoms during low glucose levels in diagnosing this condition. The use of the Whipple’s triad may therefore help identify patients with high disease burden who would benefit the most from PBH treatment. That is not to say that experiencing hypoglycemia without symptoms is of no concern, as it is well known that hypoglycemia unawareness may predispose to severe hypoglycemia (22). Thus, whether low sensor glucose levels in the absence of symptoms are truly pathological or of safety concern requires further research. Either way, a more thorough phenotyping of symptoms is challenging to implement in clinical practice, even when providing digital tools. This is mainly related to the high level of subjectivity in symptom perception, awareness and interpretation, frequent underreporting, and inaccurate temporal assignment. Wearable technologies such as smartwatches and mobile electroencephalograms may provide valuable and more objective insights in the future (23).
A strength of our analysis is the longitudinal collection of blinded CGM together with symptom recordings in well-identified patients with PBH under real-life conditions, which provides new insights into the relationship between glycemic profiles and symptom perception in this population. However, we acknowledge potential underreporting of symptoms, resulting in an underestimation of symptomatic PPPs. To mitigate the impact of potential underreporting, our analysis primarily focused on the timing of the reported symptoms rather than the categorization of symptomatic and asymptomatic episodes. In addition, although CGM allowed the observation of longitudinal glycemic patterns, it should be considered that CGM accuracy remains unexplored in this population and may be affected by the rapid glucose dynamics characterizing PBH (24). Furthermore, in line with the eating patterns in patients with PBH, in our population, meals were frequently consumed in close succession to one another, and it was difficult to define whether the symptoms occurring in the early PPP were triggered by the preceding hypoglycemia or the rapid postprandial glucose dynamics. To mitigate this, only symptoms that occurred at times clearly not related to the preceding hypoglycemia were counted as part of the dynamic phase (taking also into account a possible sensor lag time and a possible delay in symptom recording). As a result, the actual proportion of symptoms during the dynamic phase may have been underestimated. Finally, although the retrospective manual marking of meals may have introduced bias, a sensitivity analysis using the participants’ meal recordings supported the validity of our findings.
In conclusion, our results emphasize the intricate relationship between symptom perception and glucose dynamics in patients with PBH, which may challenge clinical judgement and decision-making in this population.
D.H. and L.B. share joint last authorship.
This article contains supplementary material online at https://doi.org/10.2337/figshare.23681739.
Article Information
Acknowledgments. The authors are grateful to study participants for their time and effort. The authors thank Vera Zenklusen and Valérie Brägger (study nurses at the Inselspital, Bern University Hospital) and Céline Stebler (master’s student in pharmacy, University of Bern) for their assistance in study logistics and interactions with study participants.
Funding. This work was supported by the Swiss National Science Foundation (grant no. PCEGP3_186978 to L.B.). Dexcom Inc. provided product support.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. A.T., D.H., and L.B. conceptualized this work. A.Fe., K.A.S., and N.H.N. recruited the study participants. L.C., G.C., and A.Fa. developed the integrated mobile platform of the study, where the CGM and symptom data were collected. A.T. and A.Fe. contributed to the data processing. A.T. performed the data analysis under the guidance of D.H. L.E.I. and F.P. contributed to the data analysis. A.T., D.H., and L.B. interpreted the data. The manuscript was written by A.T., D.H., and L.B. and critically reviewed by A.Fe., K.A.S., N.H.N., L.E.I., F.P., L.C., G.C., and A.Fa. All authors approved the submitted and final versions. D.H. and L.B. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Prior Presentation. This work was presented at the 16th International Conference on Advances Technologies and Treatments for Diabetes (22–25 February 2023, abstract no. 859).