Received: 9 February 2023

| Accepted: 18 June 2023

DOI: 10.1111/dme.15168

R E S E A RC H : E D U C AT I O N A L A N D P S YC H O L O G I C A L A S P EC T S

Do-­it-­yourself continuous glucose monitoring in people

aged 16 to 69 years with type 1 diabetes: A qualitative study

Shekhar Sehgal1 | Alisa Boucsein1 | Sara Styles2 | Octavia Palmer1 |

Ryan G. Paul3 | Hamish Crocket3 | Martin de Bock4,5 | Benjamin J. Wheeler1,6

1
Department of Women's and
Children's Health, Dunedin School Abstract
of Medicine, University of Otago, Aims: In many countries, real-­time continuous glucose monitoring (rt-­CGM) is
Dunedin, New Zealand
2
not funded, and cost presents a barrier to access. A do-­it-­yourself conversion of
Department of Human Nutrition,
Division of Sciences, University of
intermittently scanned CGM (DIY-­CGM) is a cheaper alternative. This qualita-
Otago, Dunedin, New Zealand tive study aimed to explore user experiences with DIY-­CGM in people aged 16 to
3
Te Huataki Waiora School of Health, 69 years with type 1 diabetes (T1D).
University of Waikato, Hamilton, New
Methods: Convenience sampling was used to recruit participants for semi-­
Zealand
4 structured virtual interviews exploring experiences of DIY-­CGM use. Participants
Department of Paediatrics, University
of Otago, Christchurch, New Zealand were recruited after completing the intervention arm of a crossover randomised
5
Department of Paediatrics, Te Whatu controlled trial that evaluated DIY-­CGM versus intermittently scanned CGM
Ora, Waitaha Canterbury, Canterbury, (isCGM). Participants were previously naive to DIY-­CGM and rt-­CGM but not
New Zealand
6
isCGM. The DIY-­CGM intervention consisted of a Bluetooth bridge connected to
Department of Paediatrics, Te Whatu
Ora Southern, Dunedin, New Zealand isCGM, adding rt-­CGM functionality over 8 weeks. Interviews were transcribed,
then thematic analysis was performed.
Correspondence
Results: Interviews were with 12 people aged 16 to 65 years, with T1D: mean
Benjamin J. Wheeler, Department
of Women's and Children's Health, age ± SD 43 ± 14 years; baseline mean HbA1c ± SD 60 mmol/mol ± 9.9 (7.6 ± 0.9%)
Dunedin School of Medicine, and time in range 59.8% ± 14.8%. Participants perceived that using DIY-­CGM im-
University of Otago, Dunedin, New
Zealand.
proved both glycaemic control and aspects of quality of life. Alarm and trend
Email: [email protected] functionality allowed participants to perceive reduced glycaemic variability
overnight and following meals. The addition of a smartwatch increased discrete
Funding information
Health Research Council of New access to glucose information. There was a high degree of trust in DIY-­CGM.
Zealand; New Zealand Society for the Challenges while using DIY-­CGM included signal loss during vigorous exercise,
Study of Diabetes
alarm fatigue and short battery life.
Conclusions: This study suggests that for users, DIY-­CGM appears to be an ac-
ceptable alternative method of rt-­CGM.

KEYWORDS
continuous glucose monitoring, do-­it-­yourself, glycaemic control, qualitative, type 1 diabetes

This is an open access article under the terms of the Creative Commons Attribution-­NonCommercial License, which permits use, distribution and reproduction in any
medium, provided the original work is properly cited and is not used for commercial purposes.
© 2023 The Authors. Diabetic Medicine published by John Wiley & Sons Ltd on behalf of Diabetes UK.

Diabetic Medicine. 2024;41:e15168.  wileyonlinelibrary.com/journal/dme | 1 of 10

https://doi.org/10.1111/dme.15168
 |

14645491, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/dme.15168, Wiley Online Library on [12/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2 of 10    SEHGAL et al.

1 | I N T RO DU CT ION
What's new?
Continuous glucose monitoring (CGM) has overcome
the limitations of and largely replaced capillary self-­ • Do-­it-­
yourself continuous glucose monitoring
monitoring of blood glucose (SMBG) as the recommended (DIY-­CGM) is a user-­initiated form of continuous
mode of glucose monitoring for type 1 diabetes (T1D) in glucose monitoring whose clinical impact in adult
all major guidelines.1–­3 There are two main types of CGM: patients remains unknown. From this study,
intermittently scanned continuous glucose monitoring DIY-­CGM use in adults appears to offer improved
(isCGM) and real-­time CGM (rt-­CGM). With the evolu- quality of life and reduced disease burden.
tion of glucose monitoring technology, rt-­CGM is now • The addition of a smartwatch facilitated dis-
non-­ adjunctive, enabling treatment decisions without crete access to glucose information. There was a
SMBG confirmation and providing continuous data and high degree of trust in DIY-­CGM and a reduced
glucose alerts. Similarly, isCGM is non-­adjunctive but only disease burden; however, all participants ex-
offers on demand sensor glucose values when scanned.4 perienced technical issues that impacted DIY-­
Following the introduction of first-­generation isCGM, gly- CGM acceptability.
caemic control measured by time in range (TIR) and Hba1c • This study suggests that DIY-­CGM may be an
improved and hypoglycaemia frequency decreased com- acceptable and convenient form of real-­time
pared to SMBG.5–­7 Second-­generation isCGM offers an im- continuous glucose monitoring for adults in a
provement in accuracy and with added glucose threshold cost-­constrained environment.
alerts. The recently published FLASH-­UK study reported
this second-­generation isCGM system improved HbA1c,
TIR and time below range.8 A recent meta-­analysis found Therefore, the aim of the present qualitative study was
CGM users of any form experienced a clinically significant to explore the lived experience of adults aged 16 to 69 years
0.26% HbA1c improvement and 5.4% improvement in TIR with T1D who used DIY-­CGM (including a smartwatch),
when compared to SMBG, the improvement in TIR was their perceived advantages and disadvantages of this de-
most marked for non-­adjunctive CGM.9 isCGM has been vice, as well as reasons participants choose to continue or
compared to rt-­CGM in several randomised controlled tri- not continue using DIY-­CGM.
als.10,11 These have shown that rt-­CGM users experience
an improvement in TIR and reduction in time in hypogly-
caemia compared to isCGM.10,11 2 | RESEARC H DESIGN AND
Despite this evidence for the superiority of rtCGM, METHODS
the dissemination of rt-­ CGM remains slow globally,
due to a variety of cost and regulatory barriers. In re- 2.1 | Design
sponse, DIY-­CGM is a solution that has evolved from
the #WeAreNotWaiting movement. A third-­party device This qualitative descriptive study evaluates the lived ex-
(MiaoMiao, MM version 2, Smart Reader, AWX Bio Co. periences of a subsample of participants in a multicentre,
Ltd.) is available4,12,13 and provides many of the benefits of randomised, controlled, cross-­over trial. The clinical trial
commercial rt-­CGM at a reduced price of US$149.00 per aimed to evaluate the clinical efficacy of smartwatch inte-
device (which is reusable).12–­14 The device uses Bluetooth grated DIY-­CGM compared with isCGM, and the protocol
to transmit data from a first-­generation isCGM sensor has been previously published.4
to a paired smart device such as a phone or smartwatch As part of the 8-­week intervention arm, participants
(Figure 1). The MM battery life is advertised as 14 days and received and wore DIY-­CGM (a MM device attached over
must be recharged and reset in conjunction with each ap- consecutive generation 1 Free Style Libre sensors). Glucose
plication of a new isCGM sensor.4 information was transmitted to an Android smartphone app
While a qualitative meta-­synthesis on the impact of (xDrip+) via the MM Bluetooth functionality. A range of
continuous glucose monitoring on life with T1D concluded Android devices were used by users: the minimum system
that CGM impacts physical, emotional and relational as- requirement was to be able to operate xDrip+ and this app
pects of life, limited paediatric and no adult studies report then displayed standard CGM data, provided optional safety
on experiences with DIY-­CGM.13,15 In addition, the use of alerts for hypo-­and hyperglycaemia, and relayed informa-
wearable technologies for accessing glucose metrics, such tion about the rate of rise and fall in an equivalent manner
as a smartwatch, is playing an increasing role in diabetes to the trend arrows used with isCGM. Xdrip+ was also con-
care, but little is known about whether they provide addi- nected via Bluetooth with the Fit Bit Versa 2 smartwatch (Fit
tional benefits to users. Bit Inc). The use of an open-­source watch face to facilitate
 |

14645491, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/dme.15168, Wiley Online Library on [12/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SEHGAL et al.    3 of 10

FIGURE 1 How DIY-­CGM works.

CGM data transmission added a novel aspect of the primary the RCT and sequentially invited via email to participate
RCT. The Glance (Ryan Mason 2019) open-­source watch face in the present qualitative study after completion of the
was installed on the smartwatch to conveniently display glu- DIY-­CGM intervention. Recruitment continued until data
cose levels detected by the DIY-­CGM system. Intervention saturation was reached, defined as no new insights arising
group participants also received additional verbal and writ- in the final two interviews.
ten xDrip+ installation and troubleshooting instructions, Participants received a questionnaire in advance of
monitoring guidelines, subsequent DIY-­CGM training that the face-­to-­face interview. The questionnaires were devel-
focused on safety and emphasised regular calibration for oped by authors SS, OP and BW and were piloted with two
accuracy. Initial alarm settings were set to those used in pre- non-­participants (members of the wider research team
vious DIY-­CGM studies,13,14 but participants could alter the working as health professional between the ages of 30 and
settings to their preference. Participant glucose data were 50 years, identifying as NZ European and including a per-
provided to the study team via the opensource app Tidepool son with diabetes). The role of the questionnaires was to
(Palo Alto, California) with information being shared with be used as a guide to facilitate in depth conversation by
their diabetes care team and care partners with the partic- encouraging participants to reflect on their lived experi-
ipant's consent. Figure 1 illustrates how the study devices ences with DIY-­CGM.
were connected to each other, as well as to follower devices A question guide for semi-­structured interviews was
and cloud-­based glucose information storage. developed from a literature review which included a prior
qualitative study regarding DIY-­CGM use in children13
and consensus between study investigators. Interview
2.2 | Participants and recruitment guides were adapted based on literature review, consensus
between investigators with experience of adult and ado-
Sixty New Zealanders aged 16 to 69 years with T1D for at lescent endocrinology and individual patient's responses
least 6 months and already using isCGM were recruited to to pre-­interview questionnaires (Appendices S1 and S2).
 |

14645491, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/dme.15168, Wiley Online Library on [12/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4 of 10    SEHGAL et al.

T A B L E 1 Participants' baseline demographic and clinical

Given the nature of the researchers' clinical and research
characteristics.
interests in maximising the benefits of diabetes technol-
ogy, our intention was to generate research findings that Gender
were actionable and pragmatic. Male 58.3% (7)
Ethnicity
New Zealand European 91.7% (11)
2.3 | Data collection Other (British) 8.3% (1)
Age (years), range 42.8 ± 14.3 (16 to 69)
Two authors, OP and SS (Shekhar Sehgal), who are not
Education level
associated with the participants' usual clinical care, con-
Postgraduate qualification 16.7% (2)
ducted semi-­structured interviews via Zoom™ which were
Tertiary education/undergraduate 33.3% (4)
audio-­recorded and transcribed verbatim. Both authors
had completed comprehensive University of Otago quali- High school 50.0% (6)
tative interviewing training. Transcripts were checked for Marital status
accuracy and were anonymised. All participants were in- Married 50.0% (6)
vited to review their transcript: no participants requested Partner/civil union 33.3% (4)
changes to their transcripts. Separated 8.3% (1)
Single 8.3% (1)
Employment status
2.4 | Data analysis
Full time 75.0% (9)
Part time 16.7% (2)
Data analysis included both deductive and inductive
frames. The first frame focused on participants' experi- Unemployed/retired 8.3% (1)
ences in relation to pre-­identified areas of interest. This Mean HbA1c +/−SD 58.9 mmol/mol ± 9.9
was informed by research team's aim to identify insights (7.6% ± 3%)
of relevance to clinicians. The second frame followed the Mean time in range 59.8% ± 14.8%
general inductive approach by Thomas16 seeking addi- NZ Dep 2018 5.5 ± 3.8 (1–­10)
tional topics beyond those captured within the deductive Insulin regimen
analysis. All interviews were analysed by SS and OP. Both Insulin pump users 58.3% (7)
authors read the transcripts independently to familiarise Multiple daily injections 41.7% (5)
themselves with the data. NVivo11 (QSR International
Note: Data are reported as percentage (n), mean ± SD or median (IQR). Age
Ltd), a qualitative software package, was used to code the
also reported with range. The New Zealand deprivation index is an area-­
transcripts. SS and OP separately coded to the deductive based measure of socio-­economic deprivation and has a value from 1 (low
frame and developed inductive codes based on sections deprivation) to 10 (high deprivation).
of text that were interpreted as having related meanings.
Following this, SS and OP met to compare codes and de-
veloped a combined coding framework (see supporting divided into seven subcategories: Learning to use the
information). Differences in interpretation were resolved device; smartwatch utility and the use of discrete glu-
via consensus. A small subset of interview transcripts cose monitoring; disease management informed by
(n = 3) were reviewed by co-­author (BW) to consolidate DIY-­CGM data; Technological malfunction negatively
and review the accuracy of coding. impacting user experience (Device connectivity, Alarm
fatigue); Short battery life and workarounds to complete
charging; Information sharing; and Trust in DIY-­CGM.
3 | R E S U LTS Tables 2 and 3 present codes within each category, as
well as participant quotations.
From the randomised controlled trial, 12 participants (23%
of total) were invited to participate, and all 12 completed
the qualitative sub-­study, giving a response rate of 100%. 3.1 | Evolution of DIY-­CGM use
Participant demographic and clinical characteristics are
shown in Table 1. 3.1.1 | Learning to use the device
Three core categories summarised the key find-
ings: Evolution of DIY-­ CGM use, Device trust and Most participants found setting up the DIY-­CGM tech-
Challenging user experiences. These were further nology relatively straight-­forward and reported that the
 |

14645491, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/dme.15168, Wiley Online Library on [12/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SEHGAL et al.    5 of 10

TABLE 2 Evolution of DIY-­CGM use.

Category Quotes (gender, age)

Device easy to use 1. ‘I was able to work through it and I could follow the instructions’. (Male, 40)
and set up 2. ‘It's fool proof and then you know just have a little tinker around and then the videos online too which is
Using Internet for going to Facebook community pages. There was a MiaoMiao page and then there's a Libre MiaoMiao page’.
troubleshooting (Male, 39)
3. Like my husband did say yesterday he said oh I'm just ‘why do not you just Google it?’ and I thought oh well
I'm talking to you today so I will not (Female, 54)
Watch utility and use 4. ‘I loved it. Especially because I do a lot … driving to work. I did not have to get my phone out to scan all the
time’. (Female, 24)
5. ‘I've set for a two-­hour meeting with customers and I'm just able to quickly look down at my watch and see
where I'm at’. (Male, 40)
6. ‘Yes, although I would have preferred if they had integration with a Garmin watch’. (Male, 40)
7. ‘I do not even use the watch now because it was so annoying, I could just check my phone’. (Male, 30 years)
8. ‘The watch, I cannot wear watches at work because I'm a nurse’. (Female, 24)
Disease management 9. ‘My HbA1c is like 50 now, and having a lot less hypos’. (Female, 24)
informed by DIY-­ 10. ‘I went from 75% in target range to generally around 95% in target range’. (Female, 54)
CGM data 11. ‘I probably got a little bit less grumpy, because I wasn't as high, as what I, you know like that roller, you
Insulin management know that roller coaster of ups and downs’. (Male, 39)
informed by real-­ 12. ‘So, I had fish and chips and I worked out what would be my split, so I do a 60/40 split within 2 h of eating
time CGM and that the CGM was awesome to track to see if I had the right sort of mix’. (Male, 39)
Increased hypo/ 13. ‘The good side is that it would give you sort of insight to highs and lows before they happened, which is
hyperglycaemia useful to know’. (Male, 40)
awareness

TABLE 3 Challenging user experiences calibration and device trust.

Category Quotations (gender, age)

Device connectivity 14. ‘I know there was regular issues between connectivity between the phone and the
Signal loss MiaoMiao as well as the phone and the watch’ (Male, 30)
15. ‘But like thinking about it I found that my smartwatch dropped out quite a bit’. (Male, 38)
16. ‘it's the watch would cut out so even if I exercise and I'd have the phone quite close
together the watch would still cut out’ (Male, 30)
Alarm fatigue 17. ‘Alarm fatigue is a bit of an issue, it'd be a nice to have more ability to snooze things, that
kind of thing’. (Male, aged 38)
18. ‘Like alarms going off and I could not stop them and having a hypo the alarm goes off and
it keeps going off and you know you have a hypo, and you are in a bad way anyway. And
you just feel like throwing it all out the window’. (Male, 67)
Short battery life 19. ‘Like it never lasts 14 days… But it never lasted as long. By day 6 or 7 days I need to charge
Work arounds to complete charging but, yeah, it's still better than like you know…’ (Female, 54)
20. ‘It's still good like we got we got given a charging cable… I sometimes hooked it up while
working’. (Male, 38)
Information sharing 21. ‘I've always been self-­managed. And my wife knows what the high and the low is. There's
Calibration never been any need’. (Male, 69)
Reduced fingerpricking 22. ‘With the MiaoMiao I calibrated it every day in the morning, like I was supposed to, and it
was like pretty much on point with my blood sugars’. (Male, 40)
23. ‘But I mean, one finger prick you know at least one finger-­prick. It's a good day, rather
than 10 finger-­pricks a day is going to be so much easier anyway’. (Male, 38)
Trust in the device to some degree 24. ‘It like settles my mind yeah, I'm a lot more stressed when I am without it’. (Female, 24)
Trust always 25. ‘Whereas with the MiaoMiao it's, I like, overall, like totally trust it’. (Male, 38)
Trust occasionally 26. ‘In the first 24 to 48 h do not trust it at all’. (Female, 54)

training and instructions provided by the main RCT study application on the smartphone and Glance smartwatch
authors were sufficient (quote 1). apps. xDrip+ required careful attention when complet-
Two main challenging areas identified in the set-­up ing the installation steps as the app was not available for
process were when participants were installing the xDrip+ download in the Google Play store.
 |

14645491, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/dme.15168, Wiley Online Library on [12/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6 of 10    SEHGAL et al.

3.1.2 | Using the Internet/alternative sources changing the timing of their insulin doses, using the DIY-­
for troubleshooting CGM to inform pump suspension (having confirmed hy-
poglycaemia beforehand by SMBG) and using DIY-­CGM
As disconnection and loss of signal were common, some to monitor the effect of advanced insulin dosing adjuncts
noted not all potential causes were covered in the re- like a split bolus (quote 12). One participant found that
sources provided in the main RCT (quote 2). As a sup- she consistently went low after dinner and altered the
plement to the troubleshooting information, half of the macronutrient composition of her evening meal follow-
participants conducted their own Internet searches to ing advice from a dietician (quote 13). A small number of
find troubleshooting information to help either the watch participants reported increased hypo-­and hyperglycaemia
or the phone reconnects to each other. While some par- awareness. Most users felt that they had gained more in-
ticipants reported that this information was relatively ac- sight or understanding into their glucose patterns as the
cessible and reliable, others required a significant search information provided was continuous and in real-­time
through Reddit threads, Facebook MiaoMiao support (quote 14).
groups and YouTube videos. Other participants found
the Internet less reliable and found the troubleshooting
guides and in person contact more user-­friendly (quote3). 3.1.5 | Technological malfunctions
negatively impacted user experience

3.1.3 | Smartwatch utility and use: Discreet All participants reported DIY-­CGM technological diffi-
glucose monitoring culties including disrupted connectivity between mobile
devices, mobile and watch apps. While these issues were
Participants' experiences using the smartwatch to view inconvenient, no participants reported serious diabetes-­
glucose data were mostly positive. The primary advantage specific incidents because of technological malfunctions.
of using a watch was the ability to check glucose informa- Many participants reported issues with device con-
tion discretely at work and in meetings (quotes 4 and 5). nectivity. Loss of connectivity between DIY-­ CGM and
Two participants who were recreational runners, Xdrip+/phone was equally as prevalent as loss of connec-
thought a wider range of compatible watches would be tivity between smartwatch and phone. Having to reboot
useful so that they could view glucose and training data or reset the Bluetooth connection was an ongoing issue
on the same device (quote 6). which hindered performance (quote 15). Three partici-
Some participants reported negative experiences in- pants reported experiencing signal loss, particularly with
cluding difficulty with Bluetooth connectivity leading to their smartwatch, during high-­ intensity exercise (e.g.
a preference to use the phone alone (quote 7), while one Running, Tennis and CrossFit). This was particularly frus-
participant reported not being able to wear a watch at trating as they believed more comprehensive glucose in-
work for safety reasons (quote 8). formation would help them better anticipate their insulin
needs during and after exercise (quote 16).
Most participants reported alarm fatigue when wear-
3.1.4 | Diabetes management informed by ing the smartwatch during the day and at night. Several
real-­time glucose data participants tried to minimise the frequency of alarms
by minimising or snoozing the alarm settings (quote 17).
All participants reported making changes to their diabetes Furthermore, many participants found the alarms related
management using information gathered from the DIY-­ to stale data (no new data have been received from the
CGM or Tidepool. Half of the participants reported that parent phone device for over 5 min) from the watches dis-
they felt that glucose information was more easily acces- rupted their sleep. As a result, some chose not to wear the
sible on their watches, and as a result, they felt both more smartwatch at night (quote 18).
aware of their glucose levels and more in control of their
diabetes when using DIY-­CGM. This increased control was
supported by a perceived improvement HbA1c, TIR results 3.1.6 | Short battery life and workarounds to
and measures of subjective well-­being for some partici- complete charging
pants (quotes 9 and 10). In addition glycaemic variability,
referred to by one participant as the ‘rollercoaster of highs Most participants reported a battery life of less than
and lows’, was also perceived to be reduced (quote 11). 14 days (quote 19). All participants reported that the
Four participants stated that they changed their insulin MM communicated better with the other devices when
dosing due to information from DIY-­CGM. This included the battery level was 50% and above. This led to several
 |

14645491, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/dme.15168, Wiley Online Library on [12/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SEHGAL et al.    7 of 10

participants finding solutions such as attaching a charging information in a variety of settings and used the device
cable when seated at a workstation; however, this solution functionality in innovative ways to make monitored
was not ideal because it reduced mobility and productivity changes in diet, exercise and insulin dosing. Compared to
of the user (quote 20). isCGM use, most participants found that using DIY-­CGM
reduced their disease burden by removing the need to re-
member to scan their isCGM sensor, and by alerting them
3.1.7 | Information sharing comfortable with when their glucose levels were out of range.
independent management Data from a range of sources including randomised
controlled trial data,11 review and meta-­analysis9,17 indi-
All participants chose not to use remote monitoring/shar- cate that use of rtCGM compared to isCGM provides ad-
ing of their blood glucose information. Motivations for vantages in both glycaemic control and patient-­reported
this decision varied between participants. Some partici- outcomes. The recently published FLASH-­UK study has
pants (3) gave the reason that their spouse trusted them to also demonstrated the benefits of adding alarms to an
independently manage their T1D (quote 21). isCGM system.8 The present study supported this view
by demonstrating user perceptions of improved HbA1c,
TIR and the feeling of being more in control of their
3.2 | Calibration diabetes with less glycaemic variability. Participants
responded positively overall to reduced glycaemic vari-
Unlike isCGM, DIY-­CGM was calibrated once a day by ability and felt that this improved their overall trust in
conducting two capillary blood glucose tests separated by DIY- ­CGM.
5 min. Many participants felt that the calibration resulted The use of a watch as an adjunct to DIY-­CGM was
in a reasonably good correlation with both isCGM and a novel aspect of the study. Users overall perceived
SMBG values (quote 22). No participants were inconven- this to improve awareness of sensor glucose levels.
ienced by the need for calibration (quote 22). Most par- Smartwatches as a form of wearable technology are
ticipants felt that despite the need for calibration, they being utilised in a variety of health fields including de-
needed less self-­monitoring of glucose (quote 23). tecting atrial fibrillation and falls prevention and thus
can be considered a growth area of wearable health-
care technology.18,19 There have been attempts to utilise
3.3 | Trust in DIY-­CGM watches to monitor glucose levels since 1999. Early de-
vices such as the Glucowatch G2 were integrated glu-
All participants reported that they trusted in the DIY-­CGM cose monitors in themselves, but they could not reliably
device to some degree despite experiencing some techni- detect hypoglycaemia, had limited accuracy and did not
cal challenges (quote 24). Almost half the participants appear to improve glycaemic control.20,21 More recent
reported that they always trusted the device completely watch applications such as the smartwatch Pebble and
and almost half said that they trusted it on most occasions One Drop integrated both activity trackers and food di-
with times of reduced trust being when the sensor glucose aries as additional diabetes management tools.22,23 The
levels dropped rapidly during exercise or during the first Pebble study evaluated integrating CGM data from a
24–­48 h where the sensor-­to-­phone connectivity tends to Dexcom G4 device but only a small proof-­of-­concept
be more unstable (quotes 25 and 26). All participants in- study was completed.22 In our study, some participants
dicated that they intended to continue using either DIY-­ found the watch difficult to wear in some work settings
CGM or RT-­CGM after the trial, the one participant not and occasionally faced difficulties due to alarm fatigue.
wishing to use the DIY-­CGM was transitioning to com- Watch technology may make sensor glucose informa-
mercial hybrid closed loop. tion more accessible, more discrete and provides the po-
tential for greater integration of rtCGM data to optimise
diabetes care.
4 | DI S C USSION Alarm fatigue has been reported as the most common
reason to discontinue rtCGM and has presented barriers
The present study qualitatively evaluated the experiences in other devices that utilise the technology including both
of rtCGM naïve adults with T1D aged 16 to 65 years using CGM and hybrid closed Loop insulin pumps.24,25 In prior
a smartwatch-­integrated DIY-­CGM. Overall, DIY-­CGM qualitative and quantitative work, alarms have tended
users had the perception of an improved self-­management to detract from the overall user experience.25 Both the
experience compared to isCGM. Specifically, participants recently published NICE guidelines and ACDC guide-
perceived an increase in their ability to access glucose lines emphasise the use of shared decision making and
 |

14645491, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/dme.15168, Wiley Online Library on [12/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8 of 10    SEHGAL et al.

a personalised care plan whilst introducing CGM and felt that this intruded upon their independence. This
alarms.26,27 finding is also inconsistent with a recent real-­world sur-
Given the novel aspect of DIY-­ CGM, for partici- vey that evaluated following in partners of adults and
pants in this qualitative study, there was a high degree parents in children, which found increased hypoglycae-
of trust in the device. This is like other studies utilis- mic confidence and well-­being in both groups.29 Further
ing rtCGM and DIY-­CGM.12–­14 This in turn led to DIY-­ research in this area, including studies that evaluate the
CGM-­ inducing dietary changes, such as monitoring characteristics of non-­users of remote monitoring may
the effectiveness of split bolus dosing with non-­sensor provide further insight into how this aspect of technol-
augmented pump therapy. This is an excellent example ogy is perceived.
of user-­derived innovation changing how existing tech-
nology is integrated. This innovation can only occur on
a background of device trust, which is an essential as- 4.1 | Strengths and limitations
pect of integrating other wearable diabetes technology
devices.28 Trust exists on a continuum and can have There are several strengths of the present study. First, the
many influencing factors, including the perceived con- findings add adult data for the first time and build upon
fidence in the device exhibited by participants based on data from a prior qualitative study based on a crossover
the training received and with access to on-­call techni- RCT in children as well as a survey that explored pa-
cal support while in the intervention arm of the wider rental experiences with DIY-­ CGM.13,14 Both of these
study. However, it is notable that 11/12 participants works stated an improvement in quality of life following
stated that they intended to keep using the device after DIY-­CGM/CGM use as well as reduced disease burden.
the completion of the study, suggesting that they did not Second, all participants were existing isCGM users and
feel reliant on study-­specific technical support to trust were therefore already comfortable with CGM, hence a
the device in the long term. clear step-­wise comparison could be made between these
Challenges to DIY-­CGM use included reduced bat- technologies, whereas others have previously focused on
tery life and impaired connectivity. This impaired the experiences with CGM compared to SMBG. The present
user experience and were like those found in other study had several limitations. Our findings, especially
DIY-­CGM studies.12,14 In terms of connectivity, utilising those related to device trust must be interpreted in con-
the DIY-­CGM required three separate digital interfaces text of our participants taking part in a clinical trial and
(Figure 1). If any one of these interfaces fail, no glucose may not be generalisable to the wider population of DIY-­
data are visible on the watch. These connectivity chal- CGM users. These users often access the devices indepen-
lenges may be device or version specific, and future DIY dently with only informal guidance from user forums for
devices will need more seamless and integrated inter- installation and troubleshooting. All participants were
faces. These too were commented upon in earlier stud- of New Zealand European or European descent which
ies.12–­14 These challenges are frequent in the DIY and may limit transferability to non-­European populations.30
watch-­related glucose monitoring literature,22 but our Participants only used DIY-­CGM for 2 months. Studies of
study overcame previous barriers with the non-­diabetes longer duration are required to evaluate long-­term DIY-­
functions of the watch operational and no need for CGM use and whether technical difficulties seen in the
manual data entry.22 These challenges may in turn be short term, such as device connectivity or alarm fatigue,
reduced by more clinician experience, the publication persist or disappear. This qualitative data while overall
of externally accepted and validated user guides by the positive needs to be supported by published quantitative
manufacturer and the ability to ask questions to a help data. This is to follow.
portal. The issues highlighted by the use of the Internet
for troubleshooting highlight the need for ongoing dia-
logue between End users, clinicians and developers to 5 | CONC LUSIONS
reduce errors and deliver a more streamlined DIY-­CGM
experience. A further method of improving connectivity The findings from this qualitative study suggest that in
in the DIY-­CGM setup, was the concomitant use of NFC adults' rt-­CGM in the form of DIY-­CGM, including hav-
and wireless (Wi-­Fi) together to reduce signal loss, that ing CGM values displayed on watch face, appears to offer
in a prior study appeared to result in less signal loss than advantages in quality of life and reduced disease burden
using NFC alone.13 compared to isCGM. As long as isCGM remains cheaper
Previous studies in children made extensive use of than most commercial rt-­CGM, DIY-­CGM has the poten-
remote monitoring, with multiple followers/care part- tial to increase access to rtCGM use, especially if more cli-
ners per participant.13 In our study, some participants nicians and nurse specialists became familiar with its use.
 |

14645491, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/dme.15168, Wiley Online Library on [12/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SEHGAL et al.    9 of 10

A previous study has shown that rtCGM can reduce ineq- glucose monitoring in people with type 1 diabetes. Diabetologia.
uities in glucose outcomes based on socio-­economic status 2019;62(8):1349-­1356.
6. Castellana M, Parisi C, Di Molfetta S, et al. Efficacy and safety
or ethnicity,31 and therefore a lower cost DIY-­CGM has a
of flash glucose monitoring in patients with type 1 and type
role as an interim measure while advocacy to improve ac-
2 diabetes: a systematic review and meta-­analysis. BMJ Open
cess for all to rtCGM continues. Diabetes Res Care. 2020;8(1):e001092.
7. Al Hayek AA, Robert AA, Al Dawish MA. Differences of
ACKNO​WLE​DGE​MENTS FreeStyle libre flash glucose monitoring system and finger
This study was funded by the Department of Women's pricks on clinical characteristics and glucose monitoring sat-
and Children's Health, Dunedin School of Medicine, isfactions in type 1 diabetes using insulin pump. Clin Med
University of Otago, New Zealand. Octavia Palmer (OP)'s Insights Endocrinol Diabetes. 2019;12:1179551419861102.
8. Leelarathna L, Evans ML, Neupane S, et al. Intermittently
summer studentship was funded by a New Zealand
scanned continuous glucose monitoring for type 1 diabetes. N
Society for the Study of Diabetes Summer Studentship
Engl J Med. 2022;387(16):1477-­1487.
Scholarship 2021/2022. SS's PhD is supported by a Health 9. Elbalshy M, Haszard J, Smith H, et al. Effect of divergent
Research Council of New Zealand ‘2020 Health Delivery continuous glucose monitoring technologies on glycaemic
Research Career Development Award’ and Clinical control in type 1 diabetes mellitus: a systematic review and
Research Training Fellowship from the Health Research meta-­analysis of randomised controlled trials. Diabet Med.
Council NZ. Some editing assistance was provided by Alex 2022;39:e14854.
Wilde. Open access publishing facilitated by University of 10. Reddy M, Jugnee N, Anantharaja S, Oliver N. Switching from
flash glucose monitoring to continuous glucose monitoring
Otago, as part of the Wiley - University of Otago agree-
on hypoglycemia in adults with type 1 diabetes at high hypo-
ment via the Council of Australian University Librarians. glycemia risk: the extension phase of the I HART CGM study.
Diabetes Technol Ther. 2018;20(11):751-­757.
FUNDING INFORMATION 11. Hásková A, Radovnická L, Petruželková L, et al. Real-­time
None. CGM is superior to flash glucose monitoring for glucose control
in type 1 diabetes: the CORRIDA randomized controlled trial.
CONFLICT OF INTEREST STATEMENT Diabetes Care. 2020;43(11):2744-­2750.
12. Elbalshy MM, Styles S, Haszard JJ, et al. The effect of do-­
None.
it-­
yourself real-­ time continuous glucose monitoring on
psychological and glycemic variables in children with type
DATA AVAILABILITY STATEMENT 1 diabetes: a randomized crossover trial. Pediatr Diabetes.
The Data in this study is derived from qualitative inter- 2022;23:480-­488.
views, data will only be made available following review 13. Elbalshy M, Boucher S, Crocket H, et al. Exploring parental
and approval by relevant regulatory bodies and the con- experiences of using a do-­it-­yourself solution for continuous
sent of the participants on a case-­by-­case basis. glucose monitoring among children and adolescents with
type 1 diabetes: a qualitative study. J Diabetes Sci Technol.
2020;14(5):844-­853.
ORCID
14. Elbalshy M, Boucher S, Galland B, et al. The MiaoMiao study:
Shekhar Sehgal https://orcid.org/0000-0002-2764-9980
can do-­it-­yourself continuous glucose monitoring technology
improve fear of hypoglycaemia in parents of children affected
REFERENCES by type 1 diabetes? J Diabetes Metab Disord. 2020;19:1-­12.
1. Blackwell M, Wheeler BJ. Clinical review: the misreporting 15. Messer L, Johnson R, Driscoll K, Jones J. Best friend or spy:
of logbook, download, and verbal self-­measured blood glu- a qualitative meta-­ synthesis on the impact of continuous
cose in adults and children with type I diabetes. Acta Diabetol. glucose monitoring on life with type 1 diabetes. Diabet Med.
2017;54(1):1-­8. 2018;35(4):409-­418.
2. ElSayed NA, Aleppo G, Aroda VR, et al. 7. Diabetes technology: 16. Thomas DR. A general inductive approach for analyzing quali-
standards of care in diabetes—­2023. Diabetes Care. 2023;46(Su tative evaluation data. Am J Eval. 2006;27(2):237-­246.
pplement_1):S111-­S127. 17. Pease A, Lo C, Earnest A, Kiriakova V, Liew D, Zoungas S.
3. Mulvihill C, Brooks A, Basudev N, Lincoln P. Continuous glu- The efficacy of technology in type 1 diabetes: a systematic re-
cose monitoring for adults and children with diabetes: sum- view, network meta-­analysis, and narrative synthesis. Diabetes
mary of updated NICE guidance. BMJ. 2022;379:o2418. Technol Ther. 2020;22(5):411-­421.
4. Sehgal S, De Bock M, Williman J, et al. Study protocol: safety 18. Rajakariar K, Koshy AN, Sajeev JK, Nair S, Roberts L, Teh
and efficacy of smart watch integrated do-­it-­yourself con- AW. Accuracy of a smartwatch based single-­ lead electro-
tinuous glucose monitoring in adults with type 1 diabe- cardiogram device in detection of atrial fibrillation. Heart.
tes, a randomised controlled trial. J Diabetes Metab Disord. 2020;106(9):665-­670.
2021;20(2):2103-­2113. 19. González-­Cañete FJ, Casilari E. A feasibility study of the use
5. Tyndall V, Stimson RH, Zammitt NN, et al. Marked im- of smartwatches in wearable fall detection systems. Sensors.
provement in HbA1c following commencement of flash 2021;21(6):2254.
 |

14645491, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/dme.15168, Wiley Online Library on [12/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10 of 10    SEHGAL et al.

20. Group DRiCNS. Accuracy of the GlucoWatch G2 biographer 28. Wang LR, Malcolm J, Arnaout A, Humphrey-­Murto S, LaDonna
and the continuous glucose monitoring system during hypogly- KA. Real-­world patient experience of long-­term hybrid closed-­
cemia: experience of the diabetes research in children network. loop insulin pump use. Can J Diabetes. 2021;45:750-­756.e3.
Diabetes Care. 2004;27(3):722-­726. 29. Polonsky WH, Fortmann AL. Impact of real-­time CGM data
21. Group DRiCNS. A randomized multicenter trial compar- sharing on quality of life in the caregivers of adults and children
ing the GlucoWatch biographer with standard glucose with type 1 diabetes. J Diabetes Sci Technol. 2022;16(1):97-­105.
monitoring in children with type 1 diabetes. Diabetes Care. 30. Kuper A, Lingard L, Levinson W. Critically appraising qualita-
2005;28(5):1101-­1106. tive research. BMJ. 2008;337:a1035.
22. Årsand E, Muzny M, Bradway M, Muzik J, Hartvigsen G. 31. Burnside MJ, Williman JA, Davies HM, et al. Inequity in access to
Performance of the first combined smartwatch and smart- continuous glucose monitoring and health outcomes in paediatric
phone diabetes diary application study. J Diabetes Sci Technol. diabetes, a case for national continuous glucose monitoring fund-
2015;9(3):556-­563. ing: a cross-­sectional population study of children with type 1 dia-
23. Osborn CY, Hirsch A, Sears LE, et al. One drop app with an betes in New Zealand. Lancet Reg Health West Pac. 2023;31:100644.
activity tracker for adults with type 1 diabetes: randomized con-
trolled trial. JMIR Mhealth Uhealth. 2020;8(9):e16745.
24. Kruger DF, Edelman SV, Hinnen DA, Parkin CG. Reference SUPPORTING INFORMATION
guide for integrating continuous glucose monitoring into clini-
Additional supporting information can be found online in
cal practice. Diabetes Educ. 2019;45(1_suppl):3S-­20S.
the Supporting Information section at the end of this article.
25. Miller E, Midyett LK. Just because you can, doesn't mean you
should… now. A practical approach to counseling persons with
diabetes on use of optional CGM alarms. Diabetes Technol Ther.
2021;23(S3):S-­66-­S-­71.
How to cite this article: Sehgal S, Boucsein A,
26. Accessed December 12, 2022. https://www.nice.org.uk/guida​
nce/ng17 Styles S, et al. Do-­it-­yourself continuous glucose
27. Soni A, Wright N, Agwu J, et al. A practical approach to contin- monitoring in people aged 16 to 69 years with type
uous glucose monitoring (rtCGM) and FreeStyle libre systems 1 diabetes: A qualitative study. Diabet Med.
(isCGM) in children and young people with type 1 diabetes. 2024;41:e15168. doi:10.1111/dme.15168
Diabetes Res Clin Pract. 2022;184:109196.