Abstract
Aims
The study analysed real‐world data of people with type 1 diabetes (PWT1D) using the Smart Multiple Daily Insulin (MDI) system with the aim of assessing associations between user interaction behaviours and parameters of glycaemic control to provide educational insights for optimal system use.
Methods
A retrospective cohort analysis was conducted using data from 1852 PWT1D users of the Smart MDI system across 21 countries. The system comprises the InPen™ insulin injector, Simplera™ CGM, and the InPen™ phone application. Users were included if they had at least 10 days of InPen and CGM data. Glycaemic outcomes were correlated with user interaction, focusing on responses to missed dose alerts (MDA) and correct high glucose alerts (CHGA).
Results
The mean time in range (TIR) was 55.7%, with users responding with a bolus dose to 49.3% of MDA and 46.6% of CHGA. However, those users who responded within 1 h to over 75% of alerts with a bolus dose achieved higher TIRs of 67.2% (MDA) and 71.5% (CHGA). Prompt responses (within 10 min) showed greater TIR. The study highlighted significant associations between alert responsiveness and better glycaemic outcomes without increased hypoglycaemia risk.
Conclusions
User engagement with the Smart MDI system is crucial for optimal glycaemic outcomes. The study underscores the need for structured education on system use and alert responses. While causality cannot be confirmed, the findings suggest that proactive user interactions contribute to meeting glycaemic targets. Further research is needed to explore these relationships and enhance educational strategies.
Keywords: continuous glucose monitoring, glycaemic control, retrospective studies, smart insulin pen, subcutaneous injections, time in range, type 1 diabetes
What's new?
The Smart MDI system provides people with diabetes on multiple daily insulin injection therapy with actionable alerts and real‐time insulin dosing guidance.
However, there is a lack of data regarding the associations between user interaction behaviours and glycaemic control in real‐world use of the Smart MDI system.
Real‐world evidence showed that users who responded more consistently and quickly to actionable alerts with an insulin bolus had better glycaemic outcomes.
Findings suggest that proactive user interactions with the Smart MDI system may contribute to meeting glycaemic targets, underscoring the need for structured education on system use and alert responses.
1. INTRODUCTION
People with type 1 diabetes (PWT1D) have an unprecedented number of treatment options at their disposal. It is well established that the latest automated insulin delivery systems provide people with diabetes the best glycaemic control. 1 However, constant connection to an insulin pump is sometimes not desirable to PWT1D for reasons including lifestyle, skin reactions, occupation, or appearance. Therefore, there is still a large portion of PWT1D choosing to remain on a multiple daily insulin injection treatment (MDI). 2 The adoption of continuous glucose monitoring (CGM) has improved overall glycaemic control, but data from the international SWEET registry show that only about 50% of PWT1D are reaching target HbA1c of <7%. 3 There is thus a need to provide support for these PWT1D on MDI therapy.
To this end, connected pens have reached the market for people with diabetes on MDI + CGM treatment with the aim of improving glycaemic control in this population. These connected pens have varying degrees of capabilities, from displaying the last dose on the pen, to transmitting insulin doses via Bluetooth to an app, to creating an automated insulin dosing log. The efficacy of these devices in terms of glycaemic control is mixed, providing small 4 , 5 or no 6 , 7 , 8 improvements compared with conventional MDI + CGM therapy.
Unique in this space is the Smart MDI system (Medtronic, USA), which, in addition to logging insulin doses, combines both insulin dosing and CGM data along with a dose calculator to provide users with real‐time actionable alerts and dosing guidance. In the clinical setting, the Smart MDI system has been shown in a small sample to help PWT1D improve glycaemic control, 9 significantly increasing time in range (TIR) between 70 and 180 mg/dL (TIR) from 58.0% with MDI + CGM to 64.4% after 1 month of Smart MDI use. However, there is a lack of data on the real‐world use of the Smart MDI system. Therefore, the objective of this study is to characterize and assess the association between glycaemic control and system interaction behaviours of PWT1D using the Smart MDI system using real‐world data from Europe, Middle East, and Africa.
2. RESEARCH DESIGN AND METHODS
2.1. System description
The Smart MDI system consists of the Bluetooth connected InPen™ insulin injector, Simplera™ CGM, and the InPen™ phone application. The InPen injector delivers half‐unit (0.5 units) doses up to 30 units. The InPen application calculates (dose calculator) and automatically records InPen‐administered rapid‐acting insulin doses, imports sensor glucose data from Simplera™ CGM, and delivers missed dose alert (MDA) and correct high glucose alert (CHGA) to the user. All data are automatically recorded in the InPen database via internet connection. ‐In addition to these InPen actionable alerts, users can choose to enable Simplera™ CGM high and low alerts; the Simplera™ urgent low alert at 54 mg/dL is always enabled.
It is at the user's discretion to administer the dose recommended by the dose calculator. While both the recommended and administered doses are logged in the app, only administered doses are considered in calculations of active insulin on board. Users can also self report long‐acting and manual rapid‐acting insulin doses (administered without the InPen injector) in the InPen app. The MDA uses a proprietary algorithm to detect meal events, checks for recent insulin doses, and alerts the user if they have missed a mealtime dose. Users can respond to this alert by selecting either ‘Calculate Dose’, which directs the user to the dose calculator for a dose recommendation, or ‘Dismiss’. The CHGA is triggered when the amount of active insulin on board is no longer sufficient to bring the sensor glucose (SG) value back to the user‐set glucose target. Alert settings allow users to set a unit threshold value for a correction dose; if the dose calculator would recommend a correction above this threshold amount, the CHGA will be triggered. Users can respond to the CHGA by selecting either ‘Calculate Dose’, which directs the user to the dose calculator for a recommended correction dose, ‘Snooze 30 min’, or ‘Wait until next meal’. Both the MDA and CHGA can be enabled during the day and/or the night, according to user‐specified daytime and nighttime periods. By default, the alerts are enabled during the day and disabled during the night.
Users can choose in the InPen app from three different meal therapy modes for calculating mealtime dose recommendations: Carb Counting, Meal Estimation, and Fixed Dose. With Carb Counting, insulin doses are based on the amount of carbohydrates reported to be in the meal. With Meal Estimation, insulin doses are based on the carbohydrate content size of the meal (low, medium, or high carb). With Fixed Dose, insulin doses are kept fixed per meal.
2.2. Data collection methods
A retrospective cohort analysis was conducted with real‐world data uploaded to the InPen database from October 2024 to January 2025 by Smart MDI System users with self reported T1D who provided consent for their data to be processed and aggregated to contribute to medical care quality improvement. The observation period was defined as between the first and last days with both an InPen insulin dose and CGM data. Users were included in the analysis only if they had a minimum of 10 days of InPen use and a minimum of 10 days of CGM data during the observation period. 10
User behaviours in terms of system features use and glycaemic outcome were assessed. A bolus reaction to an InPen alert was considered if an InPen bolus occurred in a 1‐h window from the alert. Glycaemic outcomes were also analysed across four groups according to the rate of alert response with a bolus (≤25%, >25%–≤50%, >50%–≤75%, and >75%–100% of alerts reacted with a bolus response) and according to the time to bolus following alerts (per‐person average response time of ≤10 min, >10‐ ≤20 min, 20‐≤30 min, and >30–60 min from the alert).
2.3. Statistical analysis
Descriptive statistics for continuous variables were presented as means and standard deviations (SD), or median and interquartile range (IQR) according to the data distribution. For categorical variables, frequencies and percentages were reported. The magnitude and direction of linear relationships between continuous variables were assessed using Pearson's correlation coefficient. A two‐tailed test was performed to test the null hypothesis of no correlation (ρ = 0). A significance level of α = 0.05 was used to reject the null hypothesis. Analysis was conducted in SAS Version 9.4 (SAS Institute, Cary, NC).
3. RESULTS
3.1. Population demographics
Of the 2417 Smart MDI users from 21 countries across Europe, the Middle East, and Africa with at least 10 days of InPen use and 10 days of CGM data during the observation period, 1852 individuals self reported a diagnosis of Type 1 diabetes and were included in the analysis. Over a median observation period of 66 days (IQR 41–70 days), the median percentage of days with at least one InPen bolus was 98.4% (IQR 96.1–98.6%), and the median percentage of days with collected CGM sensor data was 97.4% (IQR 94.3–98.6%). Overall, the median system usage was 96.4% (IQR 91.0–98.5%).
Demographic and user characteristics were self reported in the InPen app, and a considerable amount of missing data was observed (Table 1, Table S1): diabetes duration data was available for only 567 users (30%) and BMI was reported in only 512 users (28%).
TABLE 1.
Demographics for N = 1852 PWT1D.
| Characteristic | |
|---|---|
| Gender | |
| Male | 767 (41.4) |
| Female | 746 (40.3) |
| Prefer not to say | 3 (0.2) |
| Missing data | 336 (18.1) |
| Diabetes Duration, years a | 17.6 ± 12.8 |
| BMI (Kg/m2) b | 25.2 ± 4.4 |
| Age group | |
| <15 years | 45 (2.4) |
| 16–28 years | 396 (21.4) |
| 29–42 years | 410 (22.1) |
| 43–55 years | 278 (15.0) |
| >56 years | 212 (11.4) |
| Missing data | 511 (27.6) |
| Previous therapy | |
| Newly diagnosed | 13 (0.7) |
| Disposable insulin pen | 215 (11.6) |
| Durable insulin pen | 220 (11.9) |
| Vial and syringe | 6 (0.3) |
| Insulin pump | 27 (1.5) |
| Other | 13 (0.7) |
| Missing data | 1358 (73.3) |
Note: Data are reported as mean ± SD or as frequency (%).
Data available for N = 567 users.
Data available for N = 512 users.
3.2. InPen alert settings and responses
In terms of the InPen alert settings, most users followed the default settings of the system for at least 90% of the observed period, with MDA and CHGA enabled during the daytime and disabled during the nighttime (Table 2).
TABLE 2.
User system interaction and dosing behaviours.
| InPen therapy mode settings | |
| Fixed dose | 266 (14.4) |
| Meal estimation | 276 (14.9) |
| Carb counting | 1263 (68.2) |
| Mixed | 47 (2.5) |
| InPen alert settings | |
| Missed dose alerts enabled a | |
| Daytime (06:00–00:00) | 1617 (87.3) |
| Nighttime (00:00–06:00) | 710 (38.3) |
| Correct high glucose alerts enabled a | |
| Daytime (06:00–0:00) | 1732 (93.5) |
| Nighttime (00:00–06:00) | 857 (46.3) |
| InPen alert response interactions | |
| Daily number of missed dose alerts (MDA) | 0.93 ± 0.63 |
| Proportion MDA acknowledged by user (any response b ) | 55.9 ± 23.9% |
| Proportion MDA responded as Calculate Dose | 27.7 ± 23.1% |
| Proportion MDA followed by insulin bolus within 1 h | 49.3 ± 22.8% |
| Mean post‐MDA time to bolus | 18.2 ± 7.7 min |
| Daily number of correct high glucose alerts (CHGA) | 3.12 ± 2.35 |
| Proportion CHGA acknowledged by user (any response c ) | 60.4 ± 23.1% |
| Proportion CHGA responded as Calculate Dose | 26.5 ± 21.8% |
| Proportion CHGA followed by insulin bolus within 1 h | 46.6 ± 22.4% |
| Mean post‐CHGA time to bolus | 18.7 ± 6.4 min |
| InPen dose calculator use | |
| InPen bolus preceded by recommendation, % | 52.6 ± 34.1 |
| InPen bolus delivered as advised d , % | 52.0 ± 26.4 |
| 0.5 U greater than advised, % | 11.7 ± 10.4 |
| 1 U greater than advised, % | 6.9 ± 9.4 |
| >1 U greater than advised, % | 14.3 ± 18.8 |
| 0.5 U lower than advised, % | 6.3 ± 7.8 |
| 1 U lower than advised, % | 2.8 ± 5.5 |
| >1 U lower than advised, % | 5.9 ± 11.5 |
| Insulin delivery | |
| InPen boluses per day | 4.4 ± 1.7 |
| InPen insulin amount per day, units | 25.3 ± 14.6 |
| Basal insulin boluses per day e | 0.81 ± 0.33 |
| Basal insulin total dose per day e , units | 18.4 ± 11.8 |
Note: Data are reported as mean ± SD or as frequency (%).
Alert enabled for more than 90% of the observation period.
User chose one of the MDA response options: Calculate Dose or Dismiss.
User chose one of the CHGA response options: Calculate Dose, Snooze 30 min, or Wait until next meal.
Out of bolus preceded by recommendation.
Basal insulin dose self reported at least once during the observation period by N = 1597 users.
Overall, users received a daily mean of 0.93 ± 0.63 MDA and 3.12 ± 2.35 CHGA, with 49.3 ± 22.8% of MDA and 46.6 ± 22.4% of CHGA followed by an insulin bolus within 1 h of the alert (Table 2).
3.3. Insulin Delivery
The mean number of automatically logged InPen boluses per day was 4.4 ± 1.7 with a mean of 25.3 ± 14.6 units (Table 2). Information about basal insulin and manual boluses (those administered without the InPen injector) was self reported by the users in the InPen app. At least one basal dose was self reported during the observation period by 1597 users, with a mean number of self reported basal doses per day of 0.81 ± 0.33 and a mean of 18.4 ± 11.8 units. At least one manual bolus dose was self reported during the observation period by 649 users, with a median number of additional manual boluses per day of 0.06 (IQR 0.03–0.22) and a median of 0.41 units (IQR 0.13–1.32).
3.4. InPen dose calculator
Regarding the InPen dose calculator usage, 52.6% ± 34.1% of InPen boluses were preceded by a dose calculator recommendation (Table 2). Of these, 52.0% ± 26.4% were delivered exactly as advised.
3.5. CGM Metrics
Overall, the mean TIR was 55.7 ± 19.0%, and the mean time in tight range from 70 to 140 mg/dL (TITR) was 33.5 ± 15.9%. The median time below 70 mg/dL (TBR70) was 1.5% (IQR 0.6–3.1%) and the median time below 54 mg/dL (TBR54) was 0.07% (IQR 0.0–0.2%). The median time above 180 mg/dL (TAR180) was 41.3% (IQR 27.7–55.5%) and the median time above 250 mg/dL (TAR250) was 11.0% (IQR 4.3–21.6%). Mean SG was 175.5 ± 32.8 mg/dL, and the mean standard deviation of SG was 59.6 ± 13.6 mg/dL. The mean glucose management indicator (GMI) was 7.5 ± 0.8%.
Four‐hundred and forty‐nine users (24.2%) reached the target of TIR >70% 10 . A large majority, 1785 (96.4%) users, spent less than 1% of TBR54, while 1539 (83.1%) had less than 4% of TBR70. Only 369 (19.9%) users exceeded TAR180 for less than 25% of the time, and 542 (29.3%) exceeded TAR250 for less than 5% of the time. Regarding the GMI, 503 users (27.2%) had a GMI below 7%.
When combining TIR and GMI, 408 (22.0%) users achieved a TIR greater than 70% and a GMI below 7%, while 246 (13.3%) met the criteria of TIR greater than 70%, TBR70 less than 4%, and TAR180 less than 25%.
3.6. Association between user behaviour and clinical outcomes
3.6.1. InPen Dose Calculator Use
No correlation between InPen dose calculator use and TIR was observed (Pearson correlation ρ = 0.01, p = 0.744 for percentage of dose calculator use and ρ = 0.03, p = 0.195 for percentage of bolus delivered as advised, respectively).
3.6.2. Bolus reaction rates to alerts
A moderate correlation between users' InPen bolus reaction rate to alerts and TIR (Pearson correlation ρ = 0.28, p < 0.001 for MDA and ρ = 0.45, p < 0.001 for CHGA) was also observed, indicating that those users who bolus more as a reaction to alerts tend to have higher TIR (Figure 1).
FIGURE 1.
Association between bolus response to alerts and glycaemic outcomes.
CGM metrics were also analysed considering the following four groups according to the rate of alert response with a bolus: ≤25%, >25%–≤50%, >50%–≤75%, and >75%–100% of the alerts reacted with a bolus response.
Regarding the InPen bolus delivered as a reaction to MDA, the TIR across the four groups showed differences. The group with the least reaction to alerts had the lowest TIR, whereas the group with the highest reaction to alerts showed a higher TIR of 67.2% ± 15.1%. Similar results were observed in terms of TITR (Figure 2).
FIGURE 2.
Glycaemic control per user behaviour. Time in ranges data shown as mean values; bracketed values at right indicate time in range 70–180 mg/dL.
For TBR54, the median and interquartile range (IQR) were consistent across all groups. All groups had a median TBR70 ≤1.9% (Table S2).
Other CGM outcomes across the groups can be found in Table S2.
Considering the InPen bolus delivered as a reaction to CHGA, the TIR across the four groups showed differences. The least responsive group had the lowest TIR, whereas the most responsive group showed a higher TIR of 71.4% ± 14.3%. Similar results were observed in terms of TITR (Figure 2).
For TBR54, the median and IQR were once again consistent across all groups at 0.07% (IQR 0.0–0.2%). All groups had a median TBR70 <1.7% (Table S2). Other CGM outcomes across the groups can be found in Table S2.
3.6.3. Time to bolus reaction to alerts
Data showed a moderate negative correlation between time to InPen bolus as a reaction to alerts and TIR (Pearson correlation ρ = −0.25, p < 0.001 for MDA and ρ = −0.39, p < 0.001 for CHGA), indicating that those users who bolus earlier as a reaction to alerts tend to have higher TIR (Figure 1).
CGM metrics were also analysed by time to bolus following alerts, considering four groups of per‐person average response time: ≤10 min, >10‐≤20 min, 20‐≤30 min, and >30–60 min from the time of alert.
Regarding the time to bolus following MDA, TIR differed across the four groups. The group that administered the bolus most quickly had the highest TIR of 68.6% ± 18.3%, whereas the group that administered the bolus with the greatest delay had the lowest TIR of 52.3% ± 21.5% (Figure 2, Table S2).
For TBR54, the medians and IQR were consistent across groups, and all groups had a median TBR70 < 1.7% (Figure 2, Table S2). Other CGM outcomes across the groups can be found in Table S2.
Concerning the time to bolus following CHGA, the group that administered the bolus quickest had the highest TIR of 71.9% ± 14.9%, whereas the group that administered the bolus with the greatest delay had the lowest TIR of 48.8% ± 21.5% (Figure 2, Table S2).
Similar results were observed in terms of TITR. For TBR54, the median and IQR were consistent across groups and TBR70 was ≤1.65% for all groups (Figure 2, Table S2). Other CGM outcomes across the groups can be found in Table S2.
4. CONCLUSIONS
In a population of 1852 PWT1D using the Smart MDI system, users achieved a mean TIR of 55.7%, with 2.3% TBR70 and 42.0% TAR180. Notably, TIR was positively correlated with the user alert response rate and negatively correlated with the user alert response timing, meaning that those who responded to either an MDA or CHGA with a bolus more frequently and more quickly had higher TIR. As such, users responding with an insulin bolus to more than 75% of MDA and CHGA achieved 67.2% and 71.5% TIR, respectively, and users responding with an insulin bolus to the MDA and CHGA within 10 min of the alert achieved 68.6% and 71.9% TIR, respectively. Encouragingly, this glucose control does not come at the expense of excess hypoglycaemia risk, with all groups falling well below the target <4% TBR70 10 (Figure 2, Table S2).
These results highlight the association between user interaction with the device and glycaemic outcomes. Those users who were consistently responding quickly to the actionable alerts delivered by the InPen application met, on average, the established glycaemic targets of >70% TIR and <7% GMI. 10 In total, nearly a quarter of the analysed population (22%) met both targets, whereas, in the group of users responding to more than 75% of MDA with a bolus, 43.9% (105 users) met both targets (data not shown). Similarly, 54.4% of users responding to more than 75% of CHGA with a bolus met both targets (data not shown). While the causality between individual user interaction and glycaemic outcomes cannot be concluded from this cross‐sectional analysis, the association between these factors was consistent across the data. It is also established that the use of CGM high and predictive high alerts can reduce hyperglycaemia 11 and is associated with better glycaemic outcomes. 12 Thus, it is plausible that responding to more InPen alerts may contribute to better outcomes, but further research is required to confirm this causal relationship.
The importance of user engagement with the Smart MDI system has already been recognized by the field, with a group of experts recommending a structured educational pathway for both HCPs new to the system and people with diabetes using the Smart MDI system. 13 As confirmed in the presented data, user responses to the alerts are associated with glycaemic outcomes, and as such, users must be educated on the appropriate responses to such alerts. Uniquely, the Smart MDI system is designed to notify the user only when there is a suggested action (bolus) for the user to take. This differs fundamentally from CGM high glucose alerts, of which only 37.8% have been shown to be actionable in a simulated analysis. 14 Nonetheless, system users may experience alert fatigue from even these actionable alerts, especially if initial alerts are dismissed and glucose levels remain out of target range, which could affect long‐term adherence to the therapy. 15 , 16 The Smart MDI system also differs from automated insulin delivery systems, as user interaction and reaction are necessary to ensure glycaemic outcomes. Therefore, education around system setup, alert responses, and clinician follow‐up is warranted. Interestingly, the dose calculator was used for about half of the InPen insulin boluses analysed. Only 52.0% of these boluses were administered as advised by the calculator, whereas 32.9% were bolused greater than advised (Table 2). Despite the bolus amount consistently being higher than advised, there was no excessive hypoglycaemia detected in the group (TBR70 < 4% overall). This suggests a potential need for adjustments to the meal therapy mode settings, such as a change to the insulin‐to‐carb ratio or the estimated meal or fixed dose sizes.
The difference in TIR across the four responsiveness groups is primarily driven by differences in TITR (Figure 2). For example, the least responsive to CHGA group had an average of 25.3% TITR and 43.5% TIR, while the most responsive group had an average of 45.8% TITR and 71.5% TIR. Therefore, in an approximate 28%‐point difference in TIR, 20.5% points of this difference was in the range of 70–140 mg/dL. Therefore, the data suggest that increased responsiveness to the actionable alerts is associated with more time spent in normoglycemia. While further longitudinal studies are needed to investigate the relationship, user reactions to Smart MDI actionable alerts may help PwT1D reach improved time in normoglycaemia.
While absolute comparisons to other studies of differing designs cannot be made, the average TIR for this overall cohort of PwT1D using the Smart MDI system in the real world is consistent with previous clinical trials of PwT1D on MDI therapy with CGM. 17 , 18 , 19 For example, in a randomized control trial, Bergenstal et al. 20 demonstrated an average TIR of 53% at 26 weeks in a group of adults with T1D randomized to either efsitora or degludec basal insulin, bolus insulin lispro, and Dexcom G6 CGM. 20 The results presented here are also consistent with studies of PwT1D on MDI therapy with connected pens and CGM, 4 , 21 , 22 , 23 which confirm that engagement in the therapy can have significant effects on glycaemic outcomes. 22 , 24 This highlights the utility of the actionable alerts (MDA and CHGA) of the Smart MDI system to impact glycaemic control.
Limitations of this study remain due to the nature of the analysis of real‐world data. Particularly, all demographic data, including diabetes type, were self reported and thus may have included incorrect data that could affect the cohort description and selection. Additionally, no data prior to the start of the Smart MDI system were available to test for the impact of starting the Smart MDI system on glycaemic outcomes in these users. With the Smart MDI system, only the rapid‐acting insulin administered with the InPen is captured and logged automatically in the app; long‐acting insulin and any rapid‐acting doses administered without the InPen must be recorded manually in the app by the user. The effects of potentially missing insulin dosing data on the analysis outcomes are unknown. Despite these limitations, clear trends were observed in the collected data. An additional limitation is that this cohort represents the first users of the system, with new features and novel approaches for the users and the health care professionals to master. As such, it is possible these results are an underestimation of outcomes, which could improve with greater adoption of therapy optimization insights. Additionally, this cross‐sectional analysis cannot reveal potential therapy adherence effects over time, as user engagement with the system may vary. Future longitudinal follow‐up of these users will provide better identification of factors contributing to improved outcomes as well as identifying those who are challenged by MDI and would benefit from advancing to automated insulin delivery systems. Nonetheless, these data provide important insights on the performance of the Smart MDI system in the real world in a large group of users meeting well‐defined inclusion criteria and following internationally recognized standards for CGM metric analysis. 10
In this real‐world data analysis of Smart MDI users from Europe, the Middle East, and Africa with T1D, the results demonstrated overall glycaemic outcomes consistent with other studies of PWT1D on MDI + CGM therapy. However, further investigation into user bolus responses to Smart MDI actionable alerts revealed an association between user alert reactions and glycaemic control, with a subgroup of the most responsive users reaching internationally recognized CGM metric targets. Further research is needed to verify the longitudinal causal relationship between changes in user behaviour and glycaemic outcomes and to develop an effective educational pathway for optimal system use.
FUNDING INFORMATION
This study was funded by Medtronic.
CONFLICT OF INTEREST STATEMENT
Authors SNE, FDP, BV, GI, TH, and OC are employees of Medtronic. AL has participated in an Advisory Board for Medtronic and consulted for Eli Lilly, Menarini, Abbott, and Medtronic. PA has accepted grants and speaker fees from Abbott, Dexcom, Eli Lilly, Insulet, Medtronic, Novo Nordisk, Sanofi, and Tandem.
Supporting information
Table S1. Country of origin of the 1852 Smart MDI users included in the analysis.
Table S2. Glycaemic outcomes and user behaviours.
ACKNOWLEDGEMENTS
SNE and FDP wrote the first draft of the manuscript; FDP, BV, and GI planned and performed data analyses; TVDH and OC contributed to data analysis plans and discussions. All authors reviewed, edited, and approved the final version of the manuscript. OC is the guarantor of this work, and as such, has full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. A portion of this data was presented at the 18th International Conference on Advanced Technologies and Treatments for Diabetes (ATTD), 19–22 March 2025, in Amsterdam, The Netherlands.
Laurenzi A, Edd SN, Adolfsson P, et al. Insights into the effective use of the Smart MDI system: Data from the first 1852 type 1 diabetes users. Diabet Med. 2025;42:e70161. doi: 10.1111/dme.70161
Andrea Laurenzi and Shannon N Edd contributed equally to this study.
REFERENCES
- 1. Norgaard K, Ranjan AG, Laugesen C, et al. Glucose monitoring metrics in individuals with type 1 diabetes using different treatment modalities: a real‐world observational study. Diabetes Care. 2023;46:1958‐1964. [DOI] [PubMed] [Google Scholar]
- 2. Foster NC, Beck RW, Miller KM, et al. State of type 1 diabetes management and outcomes from the T1D exchange in 2016‐2018. Diabetes Technol Ther. 2019;21:66‐72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Gerhardsson P, Schwandt A, Witsch M, et al. The SWEET project 10‐year benchmarking in 19 countries worldwide is associated with improved HbA1c and increased use of diabetes Technology in Youth with type 1 diabetes. Diabetes Technol Ther. 2021;23:491‐499. [DOI] [PubMed] [Google Scholar]
- 4. Adolfsson P, Bjornsson V, Hartvig NV, Kaas A, Moller JB, Ogionwo Lange E. Improved glycemic control observed in children with type 1 diabetes following the introduction of smart insulin pens: a real‐world study. Diabetes Ther. 2022;13:43‐56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Adolfsson P, Hartvig NV, Kaas A, Moller JB, Hellman J. Increased time in range and fewer missed bolus injections after introduction of a smart connected insulin pen. Diabetes Technol Ther. 2020;22:709‐718. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Danne T, Forst T, Deinhard J, Rose L, Moennig E, Haupt A. No effect of insulin pen with memory function on glycemic control in a patient cohort with poorly controlled type 1 diabetes: a randomized open‐label study. J Diabetes Sci Technol. 2012;6:1392‐1397. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Venekamp WJ, Kerr L, Dowsett SA, et al. Functionality and acceptability of a new electronic insulin injection pen with a memory feature. Curr Med Res Opin. 2006;22:315‐325. [DOI] [PubMed] [Google Scholar]
- 8. Adolfsson P, Veijola R, Huot C, Hansen HD, Lademann JB, Phillip M. Safety and patient perception of an insulin pen with simple memory function for children and adolescents with type 1 diabetes‐‐the REMIND study. Curr Med Res Opin. 2012;28:1455‐1463. [DOI] [PubMed] [Google Scholar]
- 9. Pantanetti P, Cangelosi G, Morales Palomares S, et al. Real‐world life analysis of a continuous glucose monitoring and smart insulin pen system in type 1 diabetes: a cohort study. Diabetology. 2025;6:7. [Google Scholar]
- 10. Battelino T, Danne T, Bergenstal RM, et al. Clinical targets for continuous glucose monitoring data interpretation: recommendations from the international consensus on time in range. Diabetes Care. 2019;42:1593‐1603. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Abraham SB, Arunachalam S, Zhong A, Agrawal P, Cohen O, McMahon CM. Improved real‐world glycemic control with continuous glucose monitoring system predictive alerts. J Diabetes Sci Technol. 2021;15:91‐97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Akturk HK, Dowd R, Shankar K, Derdzinski M. Real‐world evidence and glycemic improvement using Dexcom G6 features. Diabetes Technol Ther. 2021;23:S21‐S26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Giorgino F, Bonfanti R, Castaldo F, et al. The utility of smart multiple daily injection Systems in Intensive Insulin‐Treated People with Diabetes: an Italian expert consensus. J Diabetes Sci Technol. 2025;19322968251316577. doi: 10.1177/19322968251316577 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Smith M, Im G, MacLeod J, Vigersky R. Why CGM alone is not enough: low actionability of CGM alerts without considering impact of active insulin. 59th EASD annual meeting of the European Association for the Study of diabetes. Diabetologia. 2023;66:1‐536. [Google Scholar]
- 15. Pickup JC, Ford Holloway M, Samsi K. Real‐time continuous glucose monitoring in type 1 diabetes: a qualitative framework analysis of patient narratives. Diabetes Care. 2015;38:544‐550. [DOI] [PubMed] [Google Scholar]
- 16. Shivers JP, Mackowiak L, Anhalt H, Zisser H. “turn it off!”: diabetes device alarm fatigue considerations for the present and the future. J Diabetes Sci Technol. 2013;7:789‐794. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Lind M, Olafsdottir AF, Hirsch IB, et al. Sustained intensive treatment and long‐term effects on HbA(1c) reduction (SILVER study) by CGM in people with type 1 diabetes treated with MDI. Diabetes Care. 2021;44:141‐149. [DOI] [PubMed] [Google Scholar]
- 18. Lind M, Polonsky W, Hirsch IB, et al. Continuous glucose monitoring vs conventional therapy for glycemic control in adults with type 1 diabetes treated with multiple daily insulin injections: the GOLD randomized clinical trial. JAMA. 2017;317:379‐387. [DOI] [PubMed] [Google Scholar]
- 19. Beck RW, Riddlesworth T, Ruedy K, et al. Effect of continuous glucose monitoring on glycemic control in adults with type 1 diabetes using insulin injections: the DIAMOND randomized clinical trial. JAMA. 2017;317:371‐378. [DOI] [PubMed] [Google Scholar]
- 20. Bergenstal RM, Weinstock RS, Mathieu C, et al. Once‐weekly insulin efsitora alfa versus once‐daily insulin degludec in adults with type 1 diabetes (QWINT‐5): a phase 3 randomised non‐inferiority trial. Lancet. 2024;404:1132‐1142. [DOI] [PubMed] [Google Scholar]
- 21. Danne TPA, Joubert M, Hartvig NV, Kaas A, Knudsen NN, Mader JK. Association between treatment adherence and continuous glucose monitoring outcomes in people with diabetes using smart insulin pens in a real‐world setting. Diabetes Care. 2024;47:995‐1003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Hellman J, Hartvig NV, Kaas A, Moller JB, Sorensen MR, Jendle J. Associations of bolus insulin injection frequency and smart pen engagement with glycaemic control in people living with type 1 diabetes. Diabetes Obes Metab. 2024;26:301‐310. [DOI] [PubMed] [Google Scholar]
- 23. Gomez‐Peralta F, Abreu C, Fernandez‐Rubio E, et al. Efficacy of a connected insulin pen cap in people with noncontrolled type 1 diabetes: a multicenter randomized clinical trial. Diabetes Care. 2023;46:206‐208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Ekberg NR, Hartvig NV, Kaas A, Moller JB, Adolfsson P. Smart pen exposes missed basal insulin injections and reveals the impact on glycemic control in adults with type 1 diabetes. J Diabetes Sci Technol. 2024;18:66‐73. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Country of origin of the 1852 Smart MDI users included in the analysis.
Table S2. Glycaemic outcomes and user behaviours.