The Role of Retrospective Data Review in the Personal Use of Real-Time Continuous Glucose Monitoring: Perceived Impact on Quality of Life and Health Outcomes

Abstract

Background:

To explore whether regularly reviewing one's own retrospective continuous glucose monitoring (CGM) data might be linked with perceived quality of life (QoL) and glycemic benefits.

Methods:

Adults with type 1 diabetes (N = 300) or insulin-using type 2 diabetes (N = 198) using the Dexcom G5 Mobile or G6 Real-Time CGM (RT-CGM) system and receiving the weekly CLARITY summary report of their glucose data completed a survey exploring their use of the report and its perceived value and impact on QoL and glycemic outcomes. Regression analyses examined whether personal use of the report was associated with QoL, perceived glycemic outcomes, and RT-CGM metrics.

Results:

The majority reported that receiving and viewing the report contributed to improved hypoglycemic confidence (75.9%) and overall well-being (50.0%), reduced diabetes distress (59.3%–74.1%), and helped to improve A1C (73.1%) and reduce problems with hypoglycemia (61.8%) and chronic hyperglycemia (73.1%). Regularly reviewing the report with family or friends (positive predictor) and doing nothing with the report's information (negative predictor) were independently associated with QoL and perceived glycemic outcomes. Surprisingly, both predictors were also associated with poorer glycemic control (e.g., greater % time above range >180).

Conclusions:

These findings suggest that receiving a weekly RT-CGM summary report may contribute to QoL and health benefits, especially if the individual chooses to actively review and make use of the report's findings and openly reviews the findings with family or friends. Prospective studies are needed to more precisely determine how retrospective RT-CGM data summaries can best be presented and utilized effectively by adults with diabetes to enhance health outcomes.

Introduction

Perhaps the key feature of current real-time continuous glucose monitoring (RT-CGM) systems is the ability to provide users with immediate feedback about their current glucose values as well as real-time trend information. However, these systems also offer a wealth of retrospective data, summarizing weeks and months of the user's glucose data, which can be of significant value for the user as well as for their health care professional (HCP).^1,2 Regular review of retrospective CGM data can help the user, whether on their own or in collaboration with their HCP, make smart treatment adjustments that can contribute to better glycemic control and fewer glycemia-related problems.^3,4

However, early survey data suggested that individuals with type 1 diabetes (T1D) who use RT-CGM generally make little use of these retrospective data outside of health care visits. Among RT-CGM users in the T1D Exchange Clinic Registry, Wong et al. found that <15% of users downloaded their data at least weekly, and 38% never did so at all.⁵ A 2015 survey of a small sample of Dexcom RT-CGM users showed similar findings, with 40% of respondents indicating that they never downloaded, 17% only rarely did so, and 15% only downloaded before an HCP visit.⁶ In 2018, an investigation of a much larger sample of Dexcom RT-CGM users (N = 50,00) found that the majority (58%) never viewed their online RT-CGM summary reports.⁷

Is there a missed opportunity here? If RT-CGM users reviewed their own retrospective data more regularly and thereby gained a broader perspective on their glucose patterns, could this contribute to better glycemic outcomes? Might doing so even add to the user's sense of control over their own diabetes, thereby enhancing quality of life (QoL) and psychosocial functioning? While the amount of glucose data amassing over a week, month, or longer period of time can be overwhelming (whether it be data derived from CGM or via frequent fingersticks), there is evidence that providing patients with a periodic graphical summary of these retrospective data could improve clinical outcomes, advance QoL, and enhance patient engagement in their own diabetes care.⁸

More specifically, an early RT-CGM qualitative study in adults with T1D found that significant glycemic benefits were more likely to occur in those who made use of their own retrospective data versus those who did not.⁹ Also, recent studies have demonstrated that more frequent viewing of one's own RT-CGM retrospective reports is associated with greater time-in-range (TIR).^10,11 In total, while these findings suggest that personal reviews of one's retrospective RT-CGM data could be beneficial, no study has yet examined this from the user's perspective—in particular, surveying how RT-CGM users respond to such data summaries and their perspectives on the personal value and impact of them. Investigations in this area may be timely, especially now that the process of obtaining retrospective summary data with some systems has become easier (e.g., automated download processes) and more comprehensible (e.g., development and consolidation of the Ambulatory Glucose Profile, or AGP, report).

We hypothesize that regularly reviewing one's own retrospective CGM data would be associated with perceived QoL and glycemic benefits. Similarly, we hypothesize that the manner in which these data summaries are reviewed (e.g., how the user typically made use of these data presentations to adjust his/her self-management) would contribute to the perceived QoL and glycemic benefits and would be associated with current CGM-derived glycemic control. To study these issues, we surveyed adults with T1D or insulin-using T2D who were current users of the Dexcom G5 or G6 RT-CGM systems, had downloaded the Dexcom CLARITY app, and had chosen to receive the weekly CLARITY email summary of their glucose data. We explored how frequently these retrospective data reports were reviewed and how they were valued, reactions and behaviors of users in response to the reports, and their perceived impact on key glycemic and diabetes-related QoL outcomes.

Research Design and Methods

Adults with T1D or insulin-using T2D were recruited from the Dexcom database via an email invitation. Inclusion criteria were as follows: T1D or T2D for at least 12 months, ages 21–75 years, currently using the Dexcom RT-CGM (either the Dexcom G5 Mobile or Dexcom G6 Systems), have downloaded the Dexcom CLARITY app, and have been receiving the weekly CLARITY email summary ≥3 months. Of note, CLARITY is a diabetes management application that provides a suite of analytic tools and retrospective data reports for Dexcom RT-CGM users, including the weekly email summary and a selection of four push notifications that highlight more specific summaries of recent RT-CGM outcomes and achievements (“Best day this week,” “Weekly time in range,” “Time in range goal,” and “Weekly patterns”). The user can choose to opt-in or opt-out to any of these individual features.

The email invitation explained that the study was being conducted in collaboration between Dexcom and the Behavioral Diabetes Institute, survey responses were anonymous, and participation was voluntary. Potential participants accessed an online portal and completed eight screening items, and if eligible, an informed consent document and an ∼85 item questionnaire. Completers received a $25 electronic gift card for participation. The collected data were entered into a central database using a HIPAA-protected server, with no linkages to personal health information or personal identifiers. The research protocol was approved by Ethical and Independent Review Services, a community-based, Institutional Review Board.

Measures

Four HCPs with large diverse populations of adult patients with diabetes in the San Diego area were asked to refer one or more patients who were using the Dexcom G5 or G6 RT-CGM system, were receiving the weekly CLARITY email summary, and would be willing to be interviewed. In total, semistructured interviews were completed with seven adults: four with T1D and three with insulin-using T2D. The interviews focused on their thoughts and feelings toward the receipt of retrospective data, how they were (or were not) making use of the weekly summary, and how these reports had impacted on their glycemic control and QoL. Based on their responses, a four-section, self-report questionnaire battery was developed in collaboration with our seven interviewees.

Section 1 (“Sample characteristics”) detailed participants' demographic information (age, gender, ethnicity, education attainment, number of years since T1D, or T2D diagnosis), current insulin pump use (yes/no), time since initiating RT-CGM and since first receiving the weekly CLARITY summary, and which if any of the four CLARITY push notifications they were currently receiving.

Section 2 (“Use of the weekly CLARITY report”) examined how frequently the weekly report was opened and reviewed (“never,” “less than half the time,” “about half the time,” and “more than half the time”) and how it was used (“helps me to plan for making insulin adjustments,” “helps me to plan for making changes about what I eat,” “helps me to plan for making changes to how I exercise,” “I review the report together with a friend or family member,” and “I don't really do anything with this information”). Section 3 (“Perceived value of the weekly CLARITY report”) assessed how participants perceived the personal utility of the weekly report. This included items to identify positive as well negative evaluations (all items are listed in Table 3). Note that for Sections 2 and 3, items were scored on a five-point Likert scale (“strongly disagree,” “somewhat disagree,” “neutral,” “somewhat agree,” and “strongly agree”).

Section 4 (“Perceived impact of the weekly CLARITY report and push notifications”) examined how participants viewed the overall impact of all the various CLARITY reports/notifications they may have received on key aspects of their QoL and glycemic outcomes. We found that participants in the initial interview phase could not easily distinguish how one CLARITY feature (i.e., the weekly report) versus another (i.e., one or more of the push notifications) might have broadly influenced long-term QoL or glycemic outcomes, and so it was decided that this group of survey questions would simply focus on this larger question (“Overall, how have the various CLARITY notifications you have been receiving, the weekly email report plus any of the 4 push notifications you may be receiving, affected you…”).

There are no validated QoL instruments assessing the individual's perception of change in this realm retrospectively; therefore, following the format used in previous RT-CGM studies,¹² we adapted three well-validated measures for use in the current study.

To evaluate hypoglycemic concerns, the nine-item Hypoglycemic Confidence Scale was used (HCS; sample item: “feeling confident that I can stay safe from serious problems with hypoglycemia when exercising”).¹³ In the modified version, participants indicated how their thoughts and feelings about addressing and avoiding hypoglycemia had changed, comparing how they feel now with how they felt before they began receiving the various CLARITY notifications (on a five-point, Likert scale)—“much less confident than before,” “somewhat less confident than before,” “neutral,” “somewhat more confident than before,” and “much more confident than before.”

To assess worries and concerns related to diabetes and its management, we included two subscales from the Diabetes Distress Scale for Adults with T1D (T1-DDS)—powerlessness (e.g., “Feeling worried that I will develop serious long-term complications, no matter how hard I try”) and management (e.g., “Feeling that I don't give my diabetes as much attention as I probably should”).¹⁴ In this modified version, paralleling the modified HCS, participants again indicated how their thoughts and feelings had changed by comparing how they feel now with how they felt before receiving CLARITY notifications. A similar five-point scale was used—“much more of a problem than before,” “somewhat more of a problem than before,” “no change,” “somewhat less of a problem than before,” and “much less of a problem than before.”

Finally, to evaluate overall well-being, the World Health Organization-5 scale was used (WHO-5 sample item: feeling “cheerful and in good spirits”), with participants in this similarly modified version comparing how they feel now with how they felt prior to receiving CLARITY notifications. As before, a five point scale was used—“much less of the time,” “somewhat less of the time,” “no change,” “somewhat more of the time,” or “much more of the time.”¹⁵

Regarding perceived glycemic outcomes, participants rated the degree to which the CLARITY notifications had helped them to “avoid problems with hypoglycemia,” “avoid problems with chronic hyperglycemia” and “improve my A1C.” Participants responded to each of the three items on a four-point scale: “has not helped,” “has helped a little,” “has been moderately helpful,” and “has helped a great deal.” Of note, throughout the questionnaire battery, participants were frequently reminded that they were not being asked to evaluate the overall impact of RT-CGM overall, but only the specific impact of CLARITY notifications.

In addition to the battery, participants provided access to the last 90 days of CGM data, allowing us to determine these five critical glycemic metrics: percentage TIR (%TIR), calculated as the percentage of daily values in the standard target range, 70–180 mg/dL, percentage time below range (%TBR) <70 mg/dL, and percentage time above range (%TAR) >180 mg/dL, mean glucose level and glucose coefficient of variation (CoV).

Data analysis

Descriptive statistics (N, %, mean, standard deviation) were used to describe participants' demographic characteristics and summary RT-CGM metrics. Frequencies, reported as N (%), were obtained to summarize the participants' personal use of the weekly CLARITY report, the perceived value of the weekly CLARITY report, and the perceived impact of CLARITY notifications (i.e., weekly report and all push notifications) on QoL (HCS, WHO-5 and the two T1-DDS subscales, powerlessness, and management distress). In addition to computing descriptive statistics and frequencies for the entire sample, significant group differences between participants with T1D versus T2D were examined with independent samples t-tests (for continuous variables) or chi-square tests (for categorical variables). Two-tailed P-values (alpha = 0.05) are reported.

All outcomes were item-level, with the exception of the four modified QoL measures, for which a mean composite score was computed for each. Mean scores for all QoL outcomes ranged from 1.0 to 5.0, with lower scores reflecting perceived negative impact of CLARITY notifications on QoL, and higher scores indicating perceived positive impact. As 3 (“no change”) was the neutral response option for all individual QoL items, a mean of 3.0 was also interpreted as the neutral/“no change” score on the corresponding composite score. QoL mean score thresholds were then selected to represent perceived impact of CLARITY notifications on their QoL: “Worsened” (<2.5), “No Change” (≥2.5 and <3.5), and “Improved” (≥3.5).

Finally, multiple linear regression analyses were conducted to evaluate whether personal use of and behavioral responses to CLARITY notifications are associated with important QoL domains and objective clinical outcomes (i.e., RT-CGM metrics), above and beyond key covariates. A total of nine multiple regression equations were estimated separately for the four QoL outcomes (HCS, WHO-5, powerlessness, and management distress) and five glucose outcomes (RT-CGM metrics: mean glucose, CoV, %TIR, %TBR <70, %TAR >180). In each of these models, predictor variables included the total number of CLARITY push notifications received (possible range: 0–4), the five behavioral responses to the CLARITY report (last five items in Table 2), and the following key covariates: age, gender (reference group = female), diabetes duration (in years), and diabetes type (reference group = type 1). IBM SPSS Statistics 28 was used for analysis.

Results

Clinical characteristics of the sample

Of the 435 adults who began the survey, 398 (91.5%) completed it in entirety (T1Ds, n = 200; T2Ds, n = 198). Compared with the T2D respondents, the T1D group was significantly younger, reported a longer duration of diabetes, was composed of more females, non-Hispanic whites, college graduates and insulin pump users, and had been using CGM for a longer period time (P's < 0.05). As anticipated, the RT-CGM metrics from the T1D group showed significantly more %TBR <70 and lower mean glucose, compared to the T2D group. Of note, receiving one or more of the CLARITY push notifications was endorsed by approximately half of respondents, although this varied based on type of push notification (e.g., 53% vs. 39% endorsed receipt of “weekly time in range” vs. “weekly patterns” notification, respectively). Descriptive statistics are presented in Table 1.

Table 1.

Sample Characteristics and Real-Time Continuous Glucose Monitoring Metrics

	Total, N (%)	T1D, N (%)	T2D, N (%)	P
Age, M (SD)	53.7 (13.9)	47.9 (14.3)	59.6 (10.6)	<0.001
Female	191 (48.0%)	113 (56.5%)	78 (39.4%)	<0.001
Race/ethnicity				0.004
Non-Hispanic white	326 (81.9%)	177 (88.5%)	149 (75.3%)	—
African American	23 (5.8%)	7 (3.5%)	16 (8.1%)	—
Hispanic	20 (5.0%)	11 (5.5%)	9 (4.5%)	—
Asian/Pacific Islander	12 (3.0%)	3 (1.5%)	9 (4.5%)	—
Native American	2 (0.5%)	0 (0.0%)	2 (1.0%)	—
Multiracial	3 (0.8%)	0 (0.0%)	3 (1.5%)	—
Other	6 (1.5%)	0 (0.0%)	6 (3.0%)	—
Decline to state	6 (1.5%)	2 (1.0%)	4 (2.0%)	—
Education level				0.118
Less than high school	3 (0.8%)	2 (1.0%)	1 (0.5%)	—
High school	30 (7.5%)	13 (6.5%)	17 (8.6%)	—
Some college	69 (17.3%)	28 (14.0%)	41 (20.7%)	—
Two-year college degree	52 (13.1%)	21 (10.5%)	31 (15.7%)	—
Four-year college degree	134 (33.7%)	71 (35.5%)	63 (31.8%)	—
Master's or doctoral degree	110 (27.7%)	65 (32.5%)	45 (22.7%)	—
Married/living with partner	313 (78.6%)	156 (78.0%)	157 (79.3%)	0.753
Diabetes duration, M (SD)	21.8 (11.9)	25.6 (13.8)	18.1 (9.3)	<0.001
≤5 years	25 (6.3%)	15 (7.5%)	10 (5.1%)	—
6–10 years	40 (10.1%)	20 (10.0%)	20 (10.1%)	—
>10 years	333 (83.7%)	165 (82.5%)	168 (84.8%)	—
Started using RT-CGM				<0.001
≤1 year ago	46 (11.5%)	4 (2.0%)	42 (21.3%)	—
1–2 years ago	130 (32.7%)	52 (26.0%)	78 (39.4%)	—
3–5 years ago	149 (37.4%)	86 (43.0%)	63 (31.8%)	—
>5 years ago	73 (18.3%)	58 (29.0%)	15 (7.6%)	—
Insulin pump user	209 (52.5%)	152 (76.0%)	57 (28.8%)	<0.001
Standalone pump	69 (17.3%)	46 (23.0%)	23 (11.6%)	—
Hybrid system	140 (35.2%)	106 (53.0%)	34 (17.2%)	—
Started receiving weekly CLARITY email				0.063
About 3 months ago	15 (3.8%)	4 (2.0%)	11 (5.6%)	—
>3 months ago	383 (96.2%)	196 (98.0%)	187 (94.4%)	—
Receives CLARITY push notification				—
Best day this week	193 (48.5%)	87 (43.5%)	106 (53.5%)	0.032
Weekly time in range	209 (52.5%)	95 (47.5%)	114 (57.6%)	0.078
Time in range goal	190 (47.7%)	92 (46.0%)	98 (49.5%)	0.482
Weekly patterns	155 (38.9%)	71 (35.5%)	84 (42.4%)	0.137
RT-CGM metrics, n = 358 (n = 179 per group)				—
Days worn, M (SD)	86.6 (10.7)	87.0 (11.2)	86.3 (10.2)	0.493
No. readings, M (SD)	24,174.8 (3,688.8)	24,388.6 (3,775.0)	23,960.9 (3,598.5)	0.273
% Time in range^a	71.0 (17.1)	70.9 (16.1)	71.1 (18.1)	0.952
% Time below range
<70 mg/dL	1.8 (2.6)	2.9 (3.3)	0.8 (1.0)	<0.001
% Time above range
>180 mg/dL	27.1 (17.7)	26.1 (17.1)	28.1 (18.2)	0.301
>250 mg/dL	6.8 (9.6)	7.1 (9.6)	6.4 (9.6)	0.484
Mean glucose, M (SD)	155.6 (48.0)	152.1 (50.8)	159.1 (45.2)	0.021
CoV, M (SD)	0.31 (0.06)	0.33 (0.06)	0.28 (0.06)	<0.001

N = 398 unless otherwise noted. Two-tailed P-values for mean differences in T1D and T2D from t- or chi-square test.

Target range is 70–180 mg/dL.

CoV, coefficient of variation; RT-CGM, real-time continuous glucose monitoring; SD, standard deviation; T1D, type 1 diabetes; T2D, type 2 diabetes; T1/2D, type 1/2 diabetes.

Personal use of the weekly CLARITY report

The majority reported that they open and review their weekly CLARITY report upon receipt every time (69.6% of respondents) or more than half the time (18.3%) (Table 2). A similar majority “somewhat” or “strongly” agreed that they use the weekly report to plan adjustments to their food intake (70.9%), exercise plans (53.5%), and insulin use (75.1%). Only a small minority (13.8%) made no use of the report, agreeing with the statement, “I don't really do anything with this information.” Note that more T1D respondents than T2D respondents used the CLARITY report to plan for insulin adjustments (82.0% T1Ds vs. 68.2% T2Ds, P = 0.001), while the reverse pattern was seen for adjusting food intake (62.5% T1Ds vs. 79.3% T2Ds, P < 0.001). Finally, relatively few respondents (17.8%) tended to review the weekly report with a friend or family member.

Table 2.

Personal Use of Weekly CLARITY Report

	Agreed, N (%)			P
	Total	T1D	T2D	P
When the weekly email reports arrive, how often do you typically open them, and take a look at them?				0.854
Never	3 (0.8%)	1 (0.5%)	2 (1.0%)	—
Less than half the time	20 (5.0%)	9 (4.5%)	11 (5.6%)	—
About half the time	25 (6.3%)	12 (6.0%)	13 (6.6%)	—
More than half the time	73 (18.3%)	34 (17.0%)	39 (19.7%)	—
Every time	277 (69.6%)	144 (72.0%)	133 (67.2%)	—
When needed, it helps me to plan for making insulin adjustments.^a	299 (75.1%)	164 (82.0%)	135 (68.2%)	0.001
When needed, it helps me to plan for making changes about what I eat.^a	282 (70.9%)	125 (62.5%)	157 (79.3%)	<0.001
When needed, it helps me to plan for making changes to how I exercise.^a	213 (53.5%)	105 (52.5%)	108 (54.5%)	0.682
I review the report together with a friend or family member.^a	71 (17.8%)	34 (17.0%)	37 (18.7%)	0.660
I don't really do anything with this information.^a	55 (13.8%)	27 (13.5%)	28 (14.1%)	0.853

N = 398. Two-tailed P-values for mean differences between T1D and T2D from independent samples t-test or chi-square test.

Each item shown was rated from 1 (strongly disagree) to 5 (strongly agree); categories 1 through 3 (neutral/disagree) and categories 4 and 5 (agree) were collapsed for presentation.

Perceived value of the weekly CLARITY report

There was broad agreement among respondents (80%–90%) that the weekly report helped them feel better (e.g., “helps me to feel more in control of diabetes,” “helps me to stay motivated”) and do better (e.g., “encourages me to stick with my diabetes care”, “helps me to acknowledge problems, and make positive changes”). In contrast, relatively few acknowledged negative evaluations of the report, such as finding it to be discouraging (9.3% of respondents) or difficult to understand (11.8%). In total, note that these findings were roughly similar for the T1D and T2D groups, with few significant differences (Table 3).

Table 3.

Perceived Value of the Weekly Email Report

	Agreed, N (%)			P
	Total, N	T1D	T2D	P
Positive value It helps to confirm how I thought I've been doing all week.	371 (93.2%)	184 (92.0%)	187 (94.4%)	0.332
It holds me accountable, so that I don't keep ignoring any bad choices I've been making.	343 (86.2%)	170 (85.0%)	173 (87.4%)	0.493
It encourages me to stick with my diabetes care.	358 (89.9%)	177 (88.5%)	181 (91.4%)	0.334
It helps me to acknowledge problems and make positive changes.	363 (91.2%)	185 (92.5%)	178 (89.9%)	0.360
It helps me to stay motivated.	344 (86.4%)	171 (85.5%)	173 (87.4%)	0.585
It makes me feel like there is someone out there cheering me on.	206 (51.8%)	101 (50.5%)	105 (53.0%)	0.613
It helps me to better understand what is going on with my diabetes.	374 (94.0%)	193 (96.5%)	181 (91.4%)	0.033
It helps me to figure out what changes I should make.	361 (90.7%)	186 (93.0%)	175 (88.4%)	0.113
It helps me to see what I might be missing.	329 (82.7%)	166 (83.0%)	163 (82.3%)	0.858
It helps me to feel more in control of diabetes.	350 (87.9%)	178 (89.0%)	172 (86.9%)	0.514
It helps me to feel more confident about making changes.	349 (87.7%)	176 (88.0%)	173 (87.4%)	0.849
Generally, I find it comforting to see this information.Negative value	354 (88.9%)	181 (90.5%)	173 (87.4%)	0.320
It doesn't really help me to figure out what I should do or change.	84 (21.1%)	35 (17.5%)	49 (24.7%)	0.076
The information is hard to understand.	47 (11.8%)	20 (10.0%)	27 (13.6%)	0.261
Generally, I find the results to be discouraging.	37 (9.3%)	20 (10.0%)	17 (8.6%)	0.627
It rarely ever tells me anything I don't already know.	57 (14.3%)	29 (14.5%)	28 (14.1%)	0.919
It is confusing.	24 (6.0%)	11 (5.5%)	13 (6.6%)	0.655
The information is not timely enough to be useful for me.	30 (7.5%)	12 (6.0%)	18 (9.1%)	0.243
The report was useful to me for a period of time, but no longer.	19 (4.8%)	9 (4.5%)	10 (5.1%)	0.797
It is just one more thing about diabetes that I would rather not be bothered by.	21 (5.3%)	12 (6.0%)	9 (4.5%)	0.516

N = 398. Each item shown was rated from 1 (strongly disagree) to 5 (strongly agree); categories 1 through 3 (neutral/disagree) and categories 4 and 5 (agree) were collapsed for presentation. Two-tailed P-values for mean differences between T1D and T2D from

chi-square or independent samples t-tests.

Perceived impact of the weekly CLARITY report and push notifications on QoL and glycemic outcomes

On the modified HCS, 75.9% of respondents reported that the CLARITY notifications had helped them become more confident in their ability to avoid or manage hypoglycemia. On the two modified T1-DDS subscales, 74.1% indicated a drop in diabetes management distress due to the CLARITY notifications, while 59.3% noted a reduction in diabetes-related powerlessness. On the WHO-5, 50.0% reported that the CLARITY notifications contributed to greater overall well-being. Of note, significantly more T2D than T1D respondents reported CLARITY-related improvements on the WHO-5 and both T1-DDS subscales (P < 0.05) (Table 4).

Table 4.

Perceived Impact of CLARITY Notifications (Including Weekly Report and All Push Notifications)

	Improved, N (%)			No change, N (%)			Worsened, N (%)
	Total	T1D	T2D	Total	T1D	T2D	Total	T1D	T2D	P
Hypoglycemia Confidence^a	302 (75.9%)	148 (74.0%)	154 (77.8%)	94 (23.6%)	52 (26.0%)	42 (21.2%)	2 (0.5%)	0 (0%)	2 (1.0%)	0.205
WHO-5 Well-Being^a	199 (50.0%)	96 (48.0%)	103 (52.0%)	193 (48.5%)	104 (52.0%)	89 (44.9%)	6 (1.5%)	0 (0%)	6 (3.0%)	0.025
Powerlessness^a	236 (59.3%)	104 (52.0%)	132 (66.7%)	133 (33.4%)	78 (39.0%)	55 (27.8%)	29 (7.3%)	18 (9.0%)	11 (5.6%)	0.011
Management distress^a	295 (74.1%)	138 (69.0%)	157 (79.3%)	93 (23.4%)	58 (29.0%)	35 (17.7%)	10 (2.5%)	4 (2.0%)	6 (3.0%)	0.026

Powerlessness and Management distress are subscales of the modified Diabetes Distress Scale. P values from independent samples t-tests of mean differences between T1D and T2D. N = 398 unless otherwise specified.

Scale items were rated from 1 (much worse) to 5 (much better); for table presentation purposes, mean score thresholds were selected to represent perceived impact of CLARITY notifications on quality of life: “Worsened” (1.0–2.4), “No Change” (2.5–3.4), and “Improved” (3.5–5.0).

WHO, World Health Organization.

A similar majority of respondents reported that the CLARITY notifications were at least moderately helpful in promoting better glycemic outcomes, including A1C improvement (73.1%), fewer problems with hypoglycemia (61.8%), and with chronic hyperglycemia (73.1%). In parallel to the findings above, more T2D than T1D respondents noted A1C improvement (78.8% vs. 67.5%, P < 0.05) and fewer hypoglycemic problems (67.7% vs. 56.0%; P < 0.1).

Associations between personal use of the weekly CLARITY report and perceived changes in QoL and glycemic outcomes

No consistent associations were observed between any of the demographic covariates (age, gender, diabetes type, years since diagnosis) and the QoL (HCS, T1-DDS, and WHO-5) or glycemic outcomes, with the one exception being age. Older age was significantly associated in a positive direction with all of the glycemic outcomes (Tables 5 and 6). While controlling for these demographics and for the total number of push notifications received, several of the behavioral responses to the weekly CLARITY report emerged as significant independent predictors of QoL and glycemic outcomes.

Table 5.

Quality-of-Life Outcomes Regressed on Use of and Behavioral Responses to Dexcom CLARITY Features

	Hypoglycemia Confidence Scale		WHO-5 Well-Being Scale		T1-DDS Powerlessness Subscale		T1-DDS Management Distress Subscale
	β (SE)	P	β (SE)	P	β (SE)	P	β (SE)	P
Covariates
Age	−0.085 (0.003)	107	−0.112 (0.003)	0.051	0.001 (0.004)	0.987	−0.016 (0.003)	0.785
Female	0.050 (0.062)	0.236	0.0300 (0.067)	0.518	0.022 (0.081)	0.632	0.032 (0.074)	0.487
Diabetes duration	0.047 (0.003)	0.343	0.067 (0.003)	0.213	0.005 (0.004)	0.930	0.008 (0.003)	0.888
Diabetes type (0 = type 1, 1 = type 2)	0.031 (0.076)	0.553	0.032 (0.082)	0.569	0.127 (0.100)	0.027	0.099 (0.091)	0.080
Push notifications received (0–4)	0.140 (0.018)	0.001	0.098 (0.02)	0.040	0.011 (0.024)	0.815	0.027 (0.022)	0.576
Behavioral Responses to CLARITY Report
It helps me to plan for making insulin adjustments.	−0.066 (0.039)	0.218	−0.009 (0.042)	0.883	0.120 (0.051)	0.045	0.068 (0.047)	0.251
It helps me to plan for making changes about what I eat.	0.340 (0.041)	<0.001	0.064 (0.044)	0.320	0.135 (0.054)	0.038	0.204 (0.049)	0.002
It helps me to plan for making changes to how I exercise.	0.158 (0.038)	0.006	0.214 (0.041)	<0.001	0.097 (0.050)	0.129	0.112 (0.046)	0.077
I review the report together with a friend or family member.	0.155 (0.026)	<0.001	0.194 (0.028)	<0.001	0.139 (0.034)	0.006	0.139 (0.031)	0.005
I don't really do anything with this information.	−0.138 (0.030)	0.003	−0.172 (0.032)	<0.001	−0.133 (0.039)	0.010	−0.122 (0.036)	0.017

N = 398 unless otherwise specified. Standardized coefficients shown. Significant (P < 0.05) effects shown in boldface.

HCS, Hypoglycemia Confidence Scale; T1-DDS, modified Type 1 Diabetes Distress Scale.

Table 6.

Glycemic Outcomes Regressed on Use of and Behavioral Responses to Dexcom CLARITY Features

	Mean glucose level		Glucose CoV		Time in range		Time below range		Time above range
	β (SE)	P	β (SE)	P	β (SE)	P	β (SE)	P	β (SE)	P
Covariates
Age	−0.149 (0.139)	0.028	−0.291 (0.000)	<0.001	0.195 (0.084)	0.004	−0.161 (0.012)	0.010	−0.164 (0.087)	0.016
Female	0.001 (3.087)	0.982	−0.005 (0.006)	0.914	0.005 (1.867)	0.925	0.022 (0.265)	0.659	−0.008 (1.935)	0.879
Diabetes duration	0.012 (0.138)	0.851	0.214 (0.000)	<0.001	−0.042 (0.084)	0.513	0.108 (0.012)	0.065	0.024 (0.087)	0.705
Diabetes type	0.157 (3.790)	0.019	−0.242 (0.007)	<0.001	−0.065 (2.292)	0.334	−0.253 (0.325)	<0.001	0.100 (2.375)	0.136
Push notifications received (0–4)	0.016 (0.908)	0.779	−0.049 (0.002)	0.325	−0.001 (0.549)	0.985	−0.037 (0.078)	0.472	0.007 (0.569)	0.907
Behavioral Responses to CLARITY Report
It helps me to plan for making insulin adjustments.	−0.014 (1.916)	0.840	0.009 (0.004)	0.882	0.006 (1.159)	0.936	0.060 (0.164)	0.342	−0.014 (1.201)	0.836
It helps me to plan for making changes about what I eat.	0.066 (2.073)	0.385	0.105 (0.004)	0.126	−0.043 (1.254)	0.572	−0.070 (0.178)	0.317	0.052 (1.299)	0.496
It helps me to plan for making changes to how I exercise.	−0.097 (1.881)	0.190	−0.121 (0.004)	0.069	0.094 (1.137)	0.206	−0.031 (0.161)	0.644	−0.086 (1.179)	0.247
I review the report together with a friend or family member.	0.134 (1.280)	0.022	0.087 (0.002)	0.099	−0.135 (0.774)	0.022	−0.045 (0.11)	0.399	0.137 (0.802)	0.020
I don't really do anything with this information.	0.119 (1.469)	0.046	−0.021 (0.003)	0.698	−0.106 (0.889)	0.078	−0.166 (0.126)	0.003	0.127 (0.921)	0.035

N = 358. Standardized coefficients shown. Significant (P < 0.05) effects are in boldface. Diabetes type = 0 for type 1/ = 1 for type 2.

DDS, modified Type 1 Diabetes Distress Scale.

First, doing nothing with the weekly report's data was linked to fewer gains in hypoglycemic confidence (HCS, P < 0.01), overall well-being (WHO-5, P < 0.001), diabetes distress (the two T1-DDS subscales, in both cases, P < 0.02), %TBR <70 (P < 0.005), %TAR >180 (P < 0.05), and mean glucose level (P < 0.05). Second, reviewing the weekly report together with a friend or family member was associated with greater improvements across all QoL outcomes, including improved hypoglycemic confidence (HCS standardized β = 0.155, P < 0.001), improved overall well-being (WHO-5 β = 0.194, P < 0.001), and reduced diabetes distress (powerlessness and management distress; both β = 0.139, P < 0.01). Conversely, the same behavior was also linked with three of six glycemic outcomes, but in the reverse direction: reviewing the report with a friend or family member was associated with higher mean glucose level (β = 0.134, P < 0.05), lower %TIR (β = −0.135, P < 0.05), and greater %TAR >180 (β = 0.137, P < 0.05), above and beyond the impact of covariates, including diabetes type.

Discussion

The results of this cross-sectional study are consistent with the idea that weekly review of retrospective RT-CGM data, presented in a consolidated summary form, may possibly contribute to significant benefits for adults with T1D and with insulin-using T2D. Most study participants (80%–90%) agreed that the weekly CLARITY report helped them feel better (e.g., “helps me to feel more in control of diabetes”) and do better (e.g., “encourages me to stick with my diabetes care”).

Similarly, using modified versions of well-validated QoL measures, we found that the majority of respondents noted that the reception and review of these summary RT-CGM data contributed to a rise in their self-confidence about avoiding and addressing hypoglycemia and a parallel drop in key elements of diabetes distress (distress about day-to-day management challenges, feelings of powerlessness in the face of diabetes), while half of participants indicated a marked rise in overall well-being. The impact on perceived glycemic outcomes was comparable, with the broad majority reporting helpful improvements.

Somewhat surprisingly, we found that significantly more participants with T2D than T1D reported QoL gains (diabetes distress, overall well-being) and glycemic outcome gains (A1C improvement, fewer hypoglycemia problems). This may be at least partly due to participants with T1D having significantly more years of experience with diabetes and with RT-CGM use than their counterparts with T2D (thereby already accumulating greater gains linked to RT-CGM); participants with T2D may therefore have had more available room for perceived improvement.

Further investigation revealed two behavioral responses to the CLARITY weekly report as being independently associated in a consistent manner with the observed benefits to QoL and glycemic status. First, participants who reported that they made no use of the weekly report (“I don't really do anything with this information”) were, not surprisingly, those least likely to enjoy benefits from it. Compared with those who did not endorse this item (thereby inferring that they made at least some regular use of the report), the nonusers had lower gains in all four QoL measures (hypoglycemic confidence, diabetes management distress, diabetes powerlessness, and overall well-being) and evidenced poorer glycemic control over the prior 90 days (less %TBR <70, greater %TAR >180, and higher mean glucose level).

Only a small minority of participants (13.8%) reported that they made no use of the weekly report. One likely explanation for the observed association with glycemic control is that reduced attention and engagement with one's retrospective glycemic data contribute to poorer glycemic outcomes. Indeed, it is widely recognized that glucose monitoring, whether it be RT-CGM or via fingerstick, can contribute to glycemic gains across multiple populations, but only if the data are being used to inform health care actions (e.g., diabetes self-care behaviors). When the data are passively collected and no one is actively responding to the patterns provided in the data, no benefit is possible.^8,16

Therefore, given these new findings, these conclusions can now possibly be extended to the use of retrospective RT-CGM reports as well. It must be acknowledged, however, that this is only one of several possible causal associations. For example, it is equally possible that individuals with poorer glycemic control, perhaps due to a sense of discouragement and/or helplessness, have simply chosen to stop looking or engaging with these data reports.

Second, participants who typically reviewed their weekly report together with a friend or family member reported greater improvement across all QoL measures, including greater hypoglycemic confidence, less distress about diabetes management, less feelings of powerlessness about diabetes, and greater overall well-being, compared to those who did not. Of note, relatively few participants (17.8%) endorsed this behavioral response. These findings are consistent with a recent retrospective study that documented similar QoL gains among adults with T1D who regularly shared their live, moment-to-moment RT-CGM data with select family and friends (via the Dexcom Share/Follow app).¹² Reviewing the weekly CLARITY report in conversation with trusted family or friends may allow the user to gather the emotional support needed for ongoing engagement with diabetes self-care as well as, more specifically, gain new perspectives and new ideas about how best to use these data.^17,18

Surprisingly, however, stronger endorsement of this behavior was also linked with poorer glycemic outcomes (less %TIR, greater %TAR and lower mean glucose level averaged over the last 90 days). Once again, caution regarding causal interpretations is necessary. However, we speculate that this small group of participants may be seeking out help and/or support from their loved ones because of problematic glycemia or the presence of other significant self-care barriers that affect glycemia. Alternatively, it is possible that reviewing the weekly report with family and friends could contribute to a degradation in glycemic control, at least for some individuals, as prior work in this area suggests that family involvement in diabetes care can be helpful or harmful depending on contextual factors.¹⁹

Future prospective research will help to clarify the veracity of these speculations and to determine which types of users and/or which types of interactions with family and friends might be most beneficial in this setting.

Although not predominant in this sample, a sizeable fraction did report some dissatisfaction with the weekly report, feeling that it did not provide summary data that was—for example—actionable (21.1% of participants), anything new (14.3%), or easily understandable (11.8%). Still, these individuals were continuing to receive and view the reports, suggesting that they remain potentially interested in these data and might benefit from guidance from their HCP or the receipt of relevant education regarding how best to interpret and use the reports. Perhaps future versions of such summary reports might benefit from a more guided interpretation of the RT-CGM retrospective data that would appear more comprehensible to these users, and could also include personalized recommendations for action.

Limitations of this study should inform future research. Although the sample was relatively large and included both adults with insulin-using T2D as well as T1D, it was predominantly non-Hispanic white (81.9%), thereby bringing into question the representativeness of the sample. More research on diabetes technology use in African-American and Hispanic populations is sorely needed, as these groups tend to have the poorest clinical outcomes, yet are rarely sampled in this area of study.²⁰ It is also not known how RT-CGM users who have elected to receive CLARITY notifications are different from those who have not. One indication of a likely difference is that the mean %T1R in the study sample was 71.0%, which is markedly higher than seen in previous studies.²¹

Furthermore, we cannot be certain that the sample is a fair representation of RT-CGM users who receive retrospective data summary reports since only recipients of Dexcom's CLARITY weekly report were surveyed. Selection biases may have also influenced why some CLARITY system users chose to participate while others did not. It is possible that users who were more dissatisfied might have been less interested in responding to the study invitation. By design, the study did not include RT-CGM users who were so dissatisfied that they had chosen to discontinue receiving the weekly summary report.

In addition, all participant responses regarding the impact of CLARITY notifications were solely retrospective in nature, and the three QoL measures were modified to facilitate retrospective enquiry. Prospective, longitudinal studies using unmodified validated instruments are needed to determine whether the observed positive changes are reflected in actual within-person change in these constructs as RT-CGM summary reports are received and viewed over time. Also, participants were asked to focus on the specific impact of retrospective RT-CGM data summaries rather than on the broader impact of RT-CGM use, but it is possible that some may have been unable to do so. Furthermore, when considering the impact on QoL and glycemic outcomes, participants were asked to consider the role of all CLARITY push notification, not just the weekly email summary.

Therefore, this particular set of positive results cannot be solely attributed to the weekly report. Despite these limitations, we surmise that these positive findings point to what might be possible when intelligible, actionable RT-CGM data summaries are provided to engaged users.

In conclusion, we found that RT-CGM users, including adults with T1D and insulin-using T2D, who regularly receive and view a weekly emailed summary of their glucose data metrics report that this feature of their RT-CGM system contributes to a range of reported QoL and glycemic benefits. QoL benefits were reduced among participants who reported that they did not make any use of the weekly summaries and CGM-derived glycemic status was poorer. Similarly, QoL benefits were greater among participants who reviewed their weekly reports with a friend or family member, but CGM-derived glycemic status was—surprisingly—poorer. Prospective studies are needed to provide greater understanding regarding the potential benefits of retrospective RT-CGM data summaries, how those data presentation can be improved, and how clinicians can guide and support RT-CGM users in making the best use of those reports when they arrive.

Footnotes

Authors' Contributions

Study conception/protocol: W.P. and A.F. Statistical analysis: E.S. Interpretation of Data: W.P. and E.S. Article development: W.P. and E.S. All authors read and approved the final article.

Acknowledgments

We thank all of study interviewees and survey respondents who so generously shared their wisdom and experience with us.

Author Disclosure Statement

W.H.P. has served as a consultant for Dexcom and Abbott Diabetes Care.

Funding Information

This investigator-initiated study was supported by Dexcom.

References

Scheiner

: CGM retrospective data analysis. Diabetes Technol Ther, 2016; 18 (Suppl 2):S214–S222.

Peters

, Garg

: The silver lining to COVID-19: avoiding diabetic ketoacidosis admissions with telehealth. Diabetes Technol Ther, 2020; 22:449–453.

Lee

, Thurston

: A comparison of discovered regularities in blood glucose readings across two data collection approaches used with a type 1 diabetic youth. Methods Inf Med, 2017; 56:e84–e91.

Aleppo

, Webb

: Continuous glucose monitoring integration in clinical practice: a stepped guide to data review and interpretation. J Diabetes Sci Technol, 2019; 13:664–673.

Wong

, Neinstein

, Spindler

, Adi

: A minority of patients with type 1 diabetes routinely downloads and retrospectively reviews device data. Diabetes Technol Ther, 2015; 17:555–562.

Pettus

, Price

, Edelman

: How patients with type 1 diabetes translate continuous glucose monitoring data into diabetes management decisions. Endocr Pract, 2015; 21:613–620.

Parker

, Welsh

, Jimenez

, Walker

: Insights from big data (2): benefits of self-guided retrospective review of continuous glucose monitoring reports. In: Poster ATTD8-0026, presented at the Advanced Technologies and Treatments for Diabetes International Conference, 2018.

Polonsky

, Fisher

: When does personalized feedback make a difference? A narrative review of recent findings and their implications for promoting better diabetes self-care. Curr Diabetes Rep, 2015; 15:620–627.

Ritholz

, Atakov-Castillo

, Beste

, et al.: Psychosocial factors associated with use of continuous glucose monitoring. Diabet Med, 2010; 27:1060–1065.

10.

van der Linden

, Welsh

, Walker

: Sustainable use of a real-time continuous glucose monitoring system from 2018 to 2020. Diabetes Technol Ther, 2021; 23:508–511.

11.

Akturk

, Dowd

, Shankar

, Derdzinski

: Real-world evidence and glycemic improvement using Dexcom G6 features. Diabetes Technol Ther, 2021; 23(Suppl 1):S21–S26.

12.

Polonsky

, Fortmann

: Impact of real-time continuous glucose monitoring data sharing on quality of life and health outcomes in adults with type 1 diabetes. Diabetes Technol Ther, 2020; 23:195–202.

13.

Polonsky

, Fisher

, Hessler

, Edelman

. Investigating hypoglycemic confidence in type 1 and type 2 diabetes. Diabetes Technol Ther, 2017; 19:131–136.

14.

Fisher

, Polonsky

, Hessler

, et al.: Understanding the sources of diabetes distress in adults with type 1 diabetes. J Diabetes Complications, 2015; 29:572–577.

15.

Hajos

, Pouwer

, Skovlund

, et al.: Psychometric and screening properties of the WHO-5 well-being index in adult outpatients with type 1 or type 2 diabetes mellitus. Diabet Med, 2013; 30:63–69.

16.

Polonsky

, Fisher

:. Right answer, but wrong question: self-monitoring of blood glucose can be clinically valuable for noninsulin users. Diabetes Care, 2013; 36:179–182.

17.

Wiebe

, Helgeson

, Berg

: The social context of managing diabetes across the life span. Am Psychol, 2016; 71:526–538.

18.

Messer

, Johnson

, Driscoll

, Jones

: Best friend or spy: a qualitative meta-synthesis on the impact of continuous glucose monitoring on life with type 1 diabetes. Diabet Med, 2018; 35:409–418.

19.

Soriano

, Lenhard

, Gonzalez

, et al.: Spousal influence on diabetes self-care: moderating effects of distress and relationship quality on glycemic control. Ann Behav Med, 2021; 55:123–132.

20.

Akturk

, Agarwal

, Hoffecker

, Shah

: Inequity in racial-ethnic representation in randomized controlled trials of diabetes technologies in type 1 diabetes: critical need for new standards. Diabetes Care, 2021; 44:e121–e123.

21.

Sandig

, Grimsmann

, Reinauer

, et al.: Continuous glucose monitoring in adults with type 1 diabetes: real-world data from the German/Austrian prospective diabetes follow-up registry. Diabetes Technol Ther, 2020; 22:602–612.