Diabetes self-management observational study investigating how CGM use impacts diabetes distress, glycaemia and functions as a technological substitute for hypoglycaemia awareness: a study protocol

Abstract

Introduction

Continuous glucose monitoring (CGM) provides real-time glucose data for people with diabetes. However, detailed knowledge of its use in daily life remains limited. We aim to investigate the interaction between people with type 1 diabetes (T1D) and their CGM data and the impact of the interaction on glycaemia and diabetes distress.

Methods and analysis

This is a two-centre observational study of adults (n=500) with T1D using FreeStyle Libre 2. Over a period of 14 days, participants will continue their regular CGM use, record insulin doses and timing with smart insulin pens, track activity and sleep with an activity tracker, log all food intake in the LibreLink app and answer questions about quality of life and hypoglycaemia two times per day. Before the study period, the participants will complete a survey of 11 validated questionnaires assessing diabetes distress, hypoglycaemia awareness and other patient-reported outcomes (PROs). After the study period, the participants will complete two additional questionnaires assessing diabetes distress and health literacy.

The collected data will be used in two substudies with the overall aims of:

Substudy 1: to investigate how CGM is used in practice and the impact of the interaction on diabetes distress and glycaemia.

Substudy 2: to investigate whether and how CGM functions as a technological substitute for impaired awareness of hypoglycaemia, focusing on alarm data.

Endpoints will include CGM metrics, alarm data and PROs.

Ethics and dissemination

The Danish Data Protection Agency approved the study (P-2024–15985), and the regional committee on health research ethics has granted an ethical waiver (H-24014662). All participants have signed written informed consent forms before participating. The results will be published in an international peer-reviewed scientific journal by the study investigators and shared via www.clinicaltrials.gov. Participants who agreed to receive information about the study will be sent the results after publication.

Trial registration number

ClinicalTrials.gov (NCT06453434).

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study collects a large, real-world dataset from 500 adults with type 1 diabetes (T1D), enhancing generalisability and enabling robust subgroup analyses.

• It incorporates validated patient-reported outcomes (PROs) to assess psychosocial aspects, including diabetes distress—an often overlooked aspect of clinical care—as well as novel exploratory hypothesis-generating questionnaires.

• It integrates multiple data sources, including continuous glucose monitoring (CGM), smart insulin pens, activity trackers, food intake logs and PROs, providing a comprehensive view of diabetes self-management.

• The observational design will not allow for conclusions about causal relationships between CGM use and glycaemic outcomes and PROs.

• The study duration of 14 days is short, which limits the ability to observe long-term patterns.

Introduction

Continuous glucose monitoring (CGM) is part of standard care for type 1 diabetes (T1D) and provides real-time glucose, including levels and trend arrows.¹ Newer CGMs offer optional alarms for low, high or pending glucose levels, which can be set by users or healthcare providers.² These features support people with T1D to make informed decisions regarding meals and insulin treatment and react immediately to prevent acute glycaemic events.³ CGM is superior to blood glucose monitoring in improving glycaemia and reducing hypoglycaemia in people with T1D.^4–6 A systematic review and meta-analysis of 64 randomised controlled trials found that CGM use lowers HbA1c and increases time in range (TIR), both linked to reduced risk of late diabetes complications.^6–9 CGM use is also associated with fewer events of hypoglycaemia and reduced time below range (TBR)<3.9 mmol/L (<70 mg/dL), time above range (TAR)>10.0 mmol/L (>180 mg/dL) and glycaemic variability.⁴ A recent study found that switching from CGM without alarms to CGM with low glucose alarms reduced the risk of hypoglycaemia.¹⁰ Higher hypoglycaemia alarm threshold settings further reduced the risk of hypoglycaemia and glycaemic variability but also reduced TIR.¹¹ The frequency at which glucose data is checked was associated with more TIR.¹²

Despite these benefits, the data load from CGM may be overwhelming, especially for users uncertain about how to interpret or act on it. Limited literacy or numeracy can make acting on CGM data harder, increasing the risk of corrective actions that are not aligned with actual needs and increasing the risk of hypoglycaemia and hyperglycaemia.^{13 14} Overcorrection with carbohydrates or insulin is common: 29% of all hypoglycaemic events in T1D are preceded by overcorrection of hyperglycaemia, and 33–41% of hyperglycaemia episodes follow hypoglycaemia, possibly due to overcorrection with carbohydrate consumption.^15–17 Insulin timing and adherence also play a role; never/rarely missing a bolus was associated with lower HbA1c compared with often missing a bolus.¹⁸ Difficulties interpreting CGM data may be a barrier to optimal use and contribute to diabetes distress.^{19 20}

Diabetes distress, defined as psychological distress related to the burden of living with diabetes, affects 20–40% of adults with T1D,^21–23 with higher prevalence among females,²⁴ younger people,^{23 24} people with shorter diabetes duration^{23 24} and people with a perceived lack of help from their network.^{23 25} Diabetes distress remains common despite CGM use, particularly in people with impaired awareness of hypoglycaemia (IAH), defined as reduced perception of hypoglycaemic symptoms.^{26 27} Some studies have shown that CGM use may reduce diabetes distress.^{28 29} The inconsistency between studies may imply that some people benefit from CGM, while others currently do not.

For people with IAH, CGM may function as an artificial hypoglycaemia awareness. IAH is common among CGM users and is associated with a sixfold risk of severe hypoglycaemia (SH), defined as hypoglycaemia requiring assistance from others.^{30 31} While CGM provides early warnings when physical symptoms are absent, SH still occurs.^{30 31} Alarm fatigue, individual differences in alarm threshold settings, response behaviours and beliefs about hypoglycaemia may be contributing factors.

Taken together, research gaps remain in understanding how CGM is used in daily life, the behavioural and psychological mechanisms behind its effect, how CGM functions as an artificial hypoglycaemia awareness in IAH and the extent to which behaviours and engagement influence diabetes distress and glycaemia. Increased knowledge in this field may inform educational programmes and clinical guidance and improve both patient-reported and glycaemic outcomes of CGM use.

The overarching aim of this study is twofold:

1. To investigate how CGM is used in practice and the impact of the interaction on diabetes distress and glycaemia.

2. To investigate whether and how CGM functions as a technological substitute for IAH, focusing on alarm data.

Study design

A two-centre observational study is currently being conducted in Denmark, including 500 adults with T1D on multiple daily injection therapy (MDI) who are already using a FreeStyle Libre 2 (FSL2). For 14 days, participants will continue their regular FSL2 use. Smart insulin pens (NovoPen 6 or NovoPen Echo Plus (Novo Nordisk, Denmark)) will record insulin doses and timing, and an activity tracker will track physical activity and sleep. Participants will use the LibreLink app as a food diary and answer questions about quality of life and hypoglycaemia two times per day. On recruitment, participants will complete a survey of 11 validated questionnaires assessing diabetes distress, hypoglycaemia awareness, treatment and glucose monitoring satisfaction, fear of hypoglycaemia, perceived competence in diabetes, sleep quality, quality of life and personality. At the end of the study, participants will complete two questionnaires assessing diabetes distress and health literacy.

The study design is illustrated in figure 1.

Figure 1.

Study design. CGM, continuous glucose monitoring.

Study population

A total of 500 participants will be recruited at Copenhagen University Hospital—North Zealand, Hillerød (NOH) and Steno Diabetes Center Copenhagen (SDCC).

Inclusion criteria

1. Aged between 18 and 85 years.

2. Diagnosed with T1D for more than 12 months.

3. Used FSL2 for more than 3 months.

4. On MDI treatment.

5. Capable of providing written informed consent.

6. Willing and able to complete study procedures, including using smart pens, wearing activity trackers and completing questionnaires at the investigator’s discretion.

Exclusion criteria

1. History of allergic reactions to materials or adhesives used in CGM devices.

2. Presence of severe cognitive or psychiatric conditions that could hinder the use of CGM or smart pens (at the investigator’s discretion).

3. Current use of corticosteroids, unless part of a chronic therapy plan (at the investigator’s discretion).

4. Daily consumption of vitamin C≥500 mg.

Recruitment

The recruitment period will run from 1 July 2024 to 1 February 2026. Participants will be recruited from diabetes outpatient clinics at NOH and SDCC by certified research nurses and physicians following Good Clinical Practice guidelines. Recruitment will occur either during routine outpatient appointments or through phone or email contact with individuals who respond to a recruitment poster or a letter of invitation. At outpatient appointments, potential participants will be identified by their treating physician or research staff using patient lists in the electronic medical record system (EMR).

Investigational sites

NOH, Denmark

The outpatient clinic at the Department of Endocrinology and Nephrology of NOH treats approximately 1000 people with T1D, and 700 use FSL2.

SDCC, Denmark

The outpatient clinic at SDCC treats approximately 6000 adults with T1D, and 3000 use FSL2.

Data collection

All study-related data will be recorded in an electronic Case Report Form (REDCap), which is a secure and General Data Protection Regulation (GDPR)-compliant web application designed for building and managing online surveys and databases.³²

Baseline data

At visit 1, baseline data will be collected, including demographic information, such as age, sex, educational level, occupation, income, household composition, marital status, ethnicity and first language; clinical information, such as duration of T1D, age at diagnosis, current insulin treatment, microvascular and macrovascular complications, other medical history and comorbidities, current medications and most recent HbA1c and lifestyle habits, such as smoking, exercise and alcohol consumption. Information on the use of the FSL2, including duration of use, will also be assessed. The information will be obtained either from the EMR or by asking the participant.

Questionnaires

At visit 1, participants will complete a survey consisting of the following 11 validated questionnaires: Type 1-Diabetes Distress Scale-28 (T1-DDS-28),³³ Diabetes Treatment Satisfaction Questionnaire,³⁴ Glucose Monitoring Satisfaction Survey,³⁵ Perceived Competence in Diabetes,³⁶ Pittsburgh Sleep Quality Index,³⁷ the WHO Well-Being Index,³⁸ Hypoglycaemia Fear Survey-Short Form,³⁹ Type D Scale-14,⁴⁰ Clarke method,⁴¹ Gold Score,⁴² Pedersen-Bjergaard method⁴³ and supplementary questions developed specifically for this study assessing alarm fatigue and experiences with CGM. At visit 2, the Health Literacy Questionnaire⁴⁴ and T1-DDS-28 will be (re)assessed.

Daily surveys

REDCap sends automated text messages to the participants each morning and evening through CPSMS.DK, a GDPR-compliant third-party web service owned by Compaya A/S.⁴⁵ Each message contains a link to a REDCap survey that participants are instructed to complete. The survey includes the Visual Analogue Scale for Anxiety,⁴⁶ the European Quality Visual Analogue Scale (EQ-VAS)⁴⁷ and a sleep quality assessment using a Likert scale ranging from 1 (‘Very good’) to 5 (‘Very bad’). Additionally, participants are asked whether they have experienced hypoglycaemia events, defined, in this case, as either having glucose values below 3.9 mmol/L or receiving an alarm for low glucose if the setting is higher than 3.9 mmol/L. For each hypoglycaemia event, participants are asked two questions. The first question asks how they became aware of their low glucose. Possible responses include (1) noticing symptoms themselves, (2) receiving an alarm from FSL2 and feeling symptoms, (3) receiving an alarm from FSL2 without feeling symptoms or (4) discovering low glucose by coincidence. The second question asks how they responded to the event. Possible answers include (1) taking no action, (2) advancing their next main meal or (3) consuming fast-acting carbohydrates.

CGM data

CGM data are collected from the FSL2 device (Abbott Laboratories, Illinois, USA), a Class IIb Conformité Européenne (CE)-marked and US Food and Drug Administration-approved CGM system. At visit 1, CGM data from the previous month will be downloaded from LibreView, a cloud-based platform for sharing data from FSL2 with healthcare professionals. The data will be uploaded to REDCap. At visit 2, CGM data from the observation period will be downloaded. Additionally, insulin and food intake data will be recorded in the FreeStyle LibreLink app and downloaded from LibreView.

CGM data includes continuous glucose values recorded every 15 min and scanned glucose values, which reflect glucose levels when participants check their CGM data. The number of scanned values indicates the frequency at which glucose data is checked on the CGM app. Abbott will provide alarm data for each participant, including timestamps for low and high glucose alarms and the threshold settings at each timestamp. Therefore, we will not provide specific recommendations for alarm threshold settings during the study period. Participants were free to choose and adjust their own alarm thresholds based on personal preference. Importantly, any changes made by participants to these settings during the study will be captured in the dataset. This will allow us to account for individual threshold settings and any changes over time when analysing the frequency and distribution of glucose alarms.

Activity and sleep data

Activity and sleep data are collected from the AX3 device (Axivity, Newcastle, UK), a CE-marked 3-axis accelerometer. Activity data will be downloaded using the OMGUI platform.

Insulin data

Data on insulin dosing and timing for each injection of insulin degludec and insulin aspart will be recorded by the NovoPen 6 and NovoPen Echo Plus smart pens. NovoPens are only compatible with Novo Nordisk insulins. Therefore, participants using other long-lasting insulins than insulin degludec will manually document insulin doses in the LibreLink app. Participants on short-acting insulin other than insulin aspart will be transitioned to equivalent doses of insulin aspart. This information will be downloaded with CGM data from LibreView at visit 2.

Food intake data

Participants are instructed to log all food intake in the LibreLink app, either by entering the amount of carbohydrate (if calculated) or simply noting the time of the meal. This information will be downloaded with CGM data from LibreView at visit 2.

Outcomes

Substudy 1

Primary outcome

The primary outcome is the association between diabetes distress (assessed by T1-DDS-28) and:

1. The frequency at which glucose data is checked on the CGM app.

2. The frequency of glucose alarms.

3. The frequency of daily between-meal insulin corrections.

4. The frequency of hypoglycaemic/hyperglycaemic events preceded by corrective actions.

5. Sociodemographic and psychosocial characteristics.

Secondary outcomes

The secondary outcomes are the associations between glycaemic metrics and:

1. The frequency at which glucose data is checked on the CGM app.

2. The frequency of glucose alarms.

3. The frequency of daily between-meal insulin corrections.

4. The frequency of hypoglycaemic/hyperglycaemic events preceded by corrective actions.

5. Sociodemographic and psychosocial characteristics.

Tertiary outcomes

The tertiary outcomes are the association between:

1. The frequency at which glucose data is checked on the CGM app and the frequency of daily between-meal insulin corrections.

2. The frequency at which glucose data is checked on the CGM app and the frequency of hypoglycaemic/hyperglycaemic events preceded by corrective actions.

3. Sociodemographic and psychosocial characteristics and the frequency at which glucose data is checked on the CGM app.

4. Sociodemographic and psychosocial characteristics and the frequency of daily between-meal insulin corrections.

5. Sociodemographic and psychosocial characteristics of CGM users and the frequency of hypoglycaemic/hyperglycaemic events preceded by corrective actions.

Substudy 2

Primary outcomes

The association between hypoglycaemia awareness status (assessed by Gold Score, Clarke method and Pedersen-Bjergaard method) and:

1. The frequency of CGM-detected hypoglycaemia alarms.

2. The alarm threshold settings.

3. The frequency of CGM-detected hypoglycaemic events.

Secondary outcomes

The associations between the frequency of hypoglycaemic events among people with normal and impaired awareness and:

1. The frequency of CGM-detected hypoglycaemia alarms.

2. The frequency of person-reported hypoglycaemia alarms.

3. The frequency of symptomatic hypoglycaemia events.

4. The frequency of corrective actions in response to hypoglycaemia alarms.

5. The frequency of corrective actions in response to symptomatic hypoglycaemia.

Tertiary outcomes

The tertiary outcomes are the associations between glycaemic metrics among people with normal and impaired awareness and:

1. The frequency of CGM-detected hypoglycaemia alarms.

2. The frequency of person-reported hypoglycaemia alarms.

3. The frequency of corrective actions in response to hypoglycaemic events.

Glycaemic metrics are reported as HbA1c and by CGM metrics, as per international consensus,⁴⁸ including:

TIR per day 3.9–10.0 mmol/L (70–180 mg/dL).
Time in tight range 3.9–7.8 mmol/L (70-–140 mg/dL).
TAR >10 mmol/L (>180 mg/dL).
TAR 10.1–13.9 mmol/L (181–250 mg/dL), level 1.
TAR >13.9 mmol/L (>250 mg/dL), level 2.
TBR <3.9 mmol/L (< 70 mg/dL).
TBR 3.0–3.9 mmol/L (54–70 mg/dL), level 1.
TBR <3.0 mmol/L (< 54 mg/dL), level 2.
Mean sensor glucose.
Standard deviation (SD) of mean sensor glucose.
Coefficient of variation (CV); SD divided by mean glucose.
Glucose management indicatorManagement Indicator.

Further non-consensus metrics include:

Extended hypoglycaemic event rate <3.9 mmol/L (<70 mg/dL) lasting ≥ 120 min; ends when glucose returns to ≥3.9 mmol/L (≥70 mg/dL) for ≥ 15 min.
Extended hyperglycaemic event rate >13.9 mmol/L (>250 mg/dL) lasting ≥ 120 min; ends when glucose returns to ≤13.9mmol9 mmol/L (≤250 mg/dL) for ≥ 15 min.
Event of hyperglycaemia (level 1) 10.1–13.9 mmol/L (181–250 mg/dL).
Event of hyperglycaemia (level 2) >13.9 mmol/L (181–250 mg/dL).
Event of hypoglycaemia (level 1) 3.0–3.9 mmol/L (54–69 mg/dL) lasting ≥ 15 min.
Event of hypoglycaemia (level 2) <3.0 mmol/L (<54 mg/dL) lasting ≥ 15 min.
Recurrent events of hypoglycaemia; multiple, separate hypoglycaemic episodes <3.9 mmol/L (<70 mg/dL) each lasting ≥ 15 min and separated by periods of normoglycaemia ≥3.9 mmol/L (≥70 mg/dL) for ≥ 15 min.
TAR 10.1–13.9 from 00:00h to 05:59h, level 1 night.
TAR 10.1–13.9 from 06:00h to 23:59h, level 1 day.
TAR >13.9 from 00:00h to 05:59h, level 2 night.
TAR >13.9 from 06:00h to 23:59h, level 2 day.
TBR 3.0–3.9 from 00:00h to 05:59h, level 1 night.
TBR 3.0–3.9 from 06:00h to 23:59h, level 1 day.
TBR <3.0 from 00:00h to 05:59h, level 2 night.
TBR <3.0 from 06:00h to 23:59h, level 2 day.
Low Blood Glucose Index; a risk metric quantifying the frequency and severity of hypoglycaemia.49
High Blood Glucose Index; a risk metric quantifying the frequency and severity of hyperglycaemia.50
Mean amplitude of glucose excursionAmplitude of Glucose Excursions; mean amplitude of glucose excursions exceeding the SD.51
Postprandial 2 -hour glucose.

Statistical considerations

Sample size

The study is an exploratory study aimed at identifying patterns in how participants interact with their CGM, examining their sociodemographic and psychosocial characteristics and assessing the impact on glycaemic metrics and diabetes distress. A sample size of 498 was calculated to have 80% power and a type-I error rate of 5% to detect a 0.19 difference in T1-DDS-28 score,²² with an SD of 0.72,³³ while accounting for potential participant dropout or data loss. The study was originally designed to assess changes in diabetes distress, and therefore, the sample size was based on the expected effect size and variability for that outcome. The sample size was also adequate for analysing alarm data in people with IAH, which has an estimated prevalence of 15%, resulting in an acceptable margin of error of 3.13%. Since no previous studies have examined this type of alarm data in people with IAH, a specific sample size calculation for that outcome was not performed.

Data analysis

Data will be analysed using descriptive, non-parametric and parametric methods. CGM metrics will be analysed using statistical R-packages designed for CGM data analysis.⁵² Descriptive statistics will be used to elucidate participant characteristics and the frequency and distribution of activities, such as the frequency at which glucose data is checked on the CGM app, the frequency of daily between-meal insulin corrections and the incidences of hypoglycaemic and hyperglycaemic events following corrections. These descriptive statistics will encompass numbers (n), proportions (%), means, SD, medians and IQR. Parametric analyses will be conducted using ordinal regression to explore the association between the T1-DDS-28 scores and the independent variables due to the ordinal nature of the T1-DDS-28 score. Linear regression will be used to investigate the relationship between glycaemic metrics and the continuous variables. Non-parametric analyses, such as the χ² test and the Kruskal-Wallis test, will be used if the normality assumption is unmet.

Considerations for multivariate analysis will be made, incorporating techniques like multiple regression to account for confounding variables (eg, age, sex, education, occupation, duration of diabetes, etc). Data transformation may be applied if the assumption of normality is unmet, and outlier analysis will be performed to identify and address influential data points. Missing data will be assessed to identify patterns and mechanisms, and appropriate imputation techniques will be applied based on the nature and extent of the missingness. The patterns and handling of missing data will be transparently reported in the study results. Any modifications to the initial statistical analysis plan will be described in future publications. A two-sided p value≤0.05 is considered statistically significant for all analyses.

Patient and public involvement statement

Patients were involved in developing and testing the survey by participating in a pilot test. They provided feedback on its length and clarity. The study results will be shared with participants and relevant patient organisations on completion.

Ethics and dissemination

The study was approved by the Danish Data Protection Agency in the Capital Region of Denmark (P-2024–15985). The study was reviewed by the regional committee on health research ethics (Scientific Ethics Committees for the Capital Region of Denmark) and received an ethical waiver (H-24014662), as it is purely observational and does not involve the collection of biological samples. All participants provided informed written consent before participating in the study. Data collected in this study will be available on request. Anonymised participant data will be stored securely by NOH and the Capital Region of Denmark for 10 years and can be shared with qualified researchers on approval by the study investigators and in accordance with the General Data Protection Regulation (EU) 2016/679 and the Danish Data Protection Act (Databeskyttelsesloven). Positive, negative and inconclusive study results will be published in an international peer-reviewed scientific journal by the investigators of the study group and via www.clinicaltrials.gov. Participants who agreed to receive information about the study will be sent the results after publication. MJN will draft the first manuscript.

Ethics statements

Patient consent for publication

Not applicable.