Diabetes devices that deliver insulin and measure blood glucose levels are cornerstones in modern treatment of type 1 diabetes. However, their use is frequently associated with the development of skin problems, particularly eczema and wounds. Proper skin care may prevent skin problems, yet evidence-based information from interventional studies is missing. Providing this information is the aim of this study.
This cluster-controlled intervention study tested the efficacy of a basic skin care program (including use of lipid cream, removal, and avoidance of disinfection). A total of 170 children and adolescents with type 1 diabetes were included and assigned either to the intervention group (n = 112) or the control group (n = 58). Participants were seen quarterly the first year after device initiation, with clinical assessment and interview in an unblinded setting.
Eczema or wounds were observed in 33.6% of the intervention group compared with 46.6% of control participants (absolute difference, 12.9% [95% CI −28.7%, 2.9%]; P = 0.10). The adjusted odds of wound development were decreased by 71% in the intervention compared with control group (for wounds, odds ratio 0.29 [95% CI 0.12, 0.68]; P = 0.005). In total, only eight infections were seen, without a higher frequency in the intervention group, despite advice to omit disinfection.
These data indicate our basic skin care program partially prevented diabetes device–induced skin reactions. However, more preventive strategies with other adhesives, patches, and/or types of lotions are needed for optimized prevention.
Introduction
Worldwide, more than 1.5 million children and adolescents live with type 1 diabetes, a lifelong disease that results in the constant need for exogenous insulin administration (1). Recent progress has led to the commercial introduction of automated insulin delivery systems, which automatically adjust insulin delivery based on sensor glucose levels, with many demonstrated benefits (2). This is notably important for children and adolescents in kindergarten and school (3). In the Scandinavian countries, at least 70% of children and adolescents with type 1 diabetes are using insulin pumps and more than 90% are using glucose sensors (4).
However, the functionality of diabetes devices is entirely dependent on the continuous adherence of the devices to the skin, which, in many instances, causes disruption of the skin barrier (5). Indeed, dermatological complications, especially contact dermatitis, are common when using the adhesive patches for 3–14 days (6). Such dermatological complications pose a threat to optimal treatment with diabetes devices, especially in children (7–11). Our previous cross-sectional study in a Danish cohort showed that 90% of 143 children and adolescents using insulin pumps had experienced dermatological complications. We found eczema to be the most common burden, affecting 25% of insulin pump users and 33% of glucose sensor users (12). The 2022 International Society of Pediatric and Adolescent Diabetes (ISPAD) Clinical Consensus Guidelines included a chapter on this issue and provided some preventive advice but without further evidence (13). The ultimate consequence of dermatological complications is discontinuation of the diabetes devices, which, according to the Type 1 Diabetes Exchange Registry, was the reason for 18% of the discontinuations (14). Skin care is the first advice against skin irritation and the use of moisturizers in adult interventional studies improved clinical outcomes and symptoms of contact dermatitis and atopic dermatitis (15). In addition to skin care, the gentle application and removal of the adhesive materials are considered important factors to minimize skin irritation (5). Because health care professionals are part of a skin care program (SCP), transfer of advice to control study participants is likely; therefore, a multicenter cluster-controlled design is preferable.
The primary aim of this study was to investigate whether a basic SCP could prevent eczema, itching, and wounds when using diabetes devices. A secondary aim was to explore which preventive products were used by the participants and, subsequently, identify their potential influence on eczema, wound, and itching.
Research Design and Methods
Study Design and Population
This was a cluster-controlled, multicenter study of a skin care intervention, with two clusters of participants. Participants from the Department of Pediatrics at Herlev and Gentofte University Hospital were allocated to the intervention group and participants from the Department of Pediatrics at Hillerød University Hospital were allocated to the control group. This cluster design is preferred to minimize the transfer of the intervention advice to the control participants, thereby avoiding contamination of the control group, which, if not avoided, may result in a type 2 error (false negative). The study population consisted of children and adolescents between the ages of 2 and 20 years who had type 1 diabetes. All participants were contacted upon initiation of a new diabetes device: either a new insulin pump or a glucose sensor, or both. The only exclusion criterium was language difficulties in answering survey questions in Danish. Inclusion in the study was set as a maximum 4 weeks from device initiation, meaning interested participants should contact the research team within four weeks after first contact. There was no follow-up telephone call to nonresponders. Prior to inclusion in the study, both the participant and their parents provided informed consent in accordance with Danish legislation.
Intervention
The intervention was the SCP, a preventive basic skin care approach developed for this study by a multidisciplinary team including pediatric diabetologists and dermatologist. The guidelines included gentle insertion and removal of devices, avoidance of disinfection of skin, and daily use of 70% Decubal Lipid Cream at all device sites not in use. An English translation of the written SCP given to the participants can be found in Supplementary Material 1. The control group received standard care, which included no specific skin care information. All participants, regardless of allocated group, were advised to contact investigators if skin irritation developed and were then included in a stepwise flowchart to guide health care personnel about treatment and secondary prevention (Supplementary Material 2).
Study Outcomes and Data Collection
Primary outcomes were either eczema or wound (yes/no) and time to eczema or wound development with diabetes device initiation as the start of the follow-up period. In secondary analyses, we investigated eczema and wound development separately. Furthermore, severity of itching at skin sites on a visual analog scale (VAS) from 0 to 100 for increasing itching was also investigated. The following predictor and possible confounding variables were identified a priori: prior use of a diabetes device, age, sex, atopic disease, and days since device initiation. Participants were followed from initiation of the device and approximately quarterly for at least 12 months. Research visits were planned in relation to normal checkups in the outpatient clinics, and follow-up time was extended if participants initiated use of a new diabetes device. If they used skin care prior to study (e.g., patches, liquid barriers), they continued doing so. The visits included an interview, a survey about products used, a VAS measure for itching, measurements of transepidermal water loss (TEWL) by Aquaflux AF200 (Biox Systems), hydration by corneometer (Courage-Khazaka), and clinical assessment by trained investigators of the last-used skin site for eczema and/or a wound for each device. The study was approved by the Danish Research Ethical Committee (H-18059790) and the local Regional Data Agency (P-2020-02), and was preregistered at clinicaltrials.gov (NCT04258904). Data collection was managed using REDCap (16,17).
Statistics
Statistical analyses were performed using the statistical software package R, version 4.0.3 (the R Foundation for Statistical Computing, Vienna, Austria). The overall effect of intervention on development of first eczema or wounds (yes/no) during the full study period was investigated with descriptive statistics, logistic regression, and generalized estimation equation (GEE) models with exchangeable correlation structure for repeated measurements (18). The effect of intervention on time to event of eczema or wound development was investigated with Cox proportional hazards regression (hereafter, Cox regression). The effect of intervention on itching score over time was investigated with descriptive statistics and GEE modeling, with itching as an ordinal outcome. The effect of different products on eczema, wounds, and itching was investigated with the GEE model approach. Post hoc analyses of the effect of intervention and use of lipid cream on TEWL and hydration was also performed using GEE modeling. The Benjamini–Hochberg procedure was used for correction of multiple comparisons of P values for all families of tests (19). After correction (Pcorrected), P < 0.05 was considered statistically significant; however, all estimates and 95% CIs were evaluated to detect possible important clinical effects and to avoid type 2 errors. Sample size was based on unpaired t test for dermatological complications for 12 months, with an expected frequency in the control group of 50% and a halving in the intervention group, with a ratio of 1:2 for group enrollments, α = 0.05 and β = 0.2, resulting in 84 participants in the intervention group and 42 participants in control group. With an anticipated 10% dropout rate during the study period, at least 94 and 47 participants were included in the two groups. However, to obtain power for subgroup analysis for the insulin pump and glucose sensor cohorts separately and cover the loss of power due to cluster randomization, the inclusion period was extended for 3 months, which was the maximum possible due to funding.
Results
Study Population
A total of 308 children and adolescents initiated use of an insulin pump and/or a glucose sensor between 10 March 2020 and 1 July 2021. Of these, 174 (56.5%) showed interest in the follow-up and were included in this study and allocated according to the cluster-controlled design. A total of 115 were assigned to the intervention group and 59 to the control group (Supplementary Fig. 1); differences in numbers reflect differences in center size and, therefore, number of eligible children and adolescents in each center. Ten participants from the intervention group were excluded during the study period but data collected on them before exclusion were included in analyses. Four participants were excluded from all analyses in this study because there were no data from their clinical examinations. From 1 November 2021 to last visit, 20 September 2022, all children and adolescents with type 1 diabetes in the Capital Region of Denmark were followed for routine care and research at the Steno Diabetes Center Copenhagen.
Basic Demographics
Table 1 lists basic demographic information of all included participants. Age and z scores for BMI were similar between groups, but female sex was more frequent in the control group. Newly diagnosed individuals composed 53% of the intervention group and 38% of the control group since initiation of Freestyle Libre by onset of diabetes. Atopic diseases were seen in 27% of the intervention group and 16% of the control group. Supplementary Tables 1 and 2 show the two device cohorts with information on device types, device history, and earlier skin problems.
Demographics of intervention and control group at inclusion
. | Intervention group (n = 112) . | Control group (n = 58) . |
---|---|---|
Age, mean (SD), years | 11.2 (4.05) | 11.0 (3.45) |
Female sex, n (%) | 55 (49.1) | 34 (58.6) |
BMI z score* | ||
Mean (SD) | 0.0701 (1.19) | 0.0802 (1.15) |
Missing data, n (%) | 1 (0.9) | 1 (1.7) |
Diabetes duration, mean (SD), years | 2.49 (3.60) | 2.20 (3.20) |
Patients with newly diagnosed type 1 diabetes, n (%) | 59 (52.7) | 22 (37.9) |
Ever self-reported atopic dermatitis, n (%) | 13 (11.6) | 4 (6.9) |
Known allergic contact dermatitis, n (%) | 9 (8.0) | 1 (1.7) |
History of at least one atopic disease, n (%)† | 30 (26.8) | 9 (15.5) |
Initiated an insulin pump at baseline or during study period, n (%) | 55 (49.1) | 40 (69.0) |
Initiated a glucose sensor at baseline or during study period, n (%) | 99 (88.4) | 37 (63.8) |
Study period, mean (SD), days | 384 (133) | 428 (105) |
. | Intervention group (n = 112) . | Control group (n = 58) . |
---|---|---|
Age, mean (SD), years | 11.2 (4.05) | 11.0 (3.45) |
Female sex, n (%) | 55 (49.1) | 34 (58.6) |
BMI z score* | ||
Mean (SD) | 0.0701 (1.19) | 0.0802 (1.15) |
Missing data, n (%) | 1 (0.9) | 1 (1.7) |
Diabetes duration, mean (SD), years | 2.49 (3.60) | 2.20 (3.20) |
Patients with newly diagnosed type 1 diabetes, n (%) | 59 (52.7) | 22 (37.9) |
Ever self-reported atopic dermatitis, n (%) | 13 (11.6) | 4 (6.9) |
Known allergic contact dermatitis, n (%) | 9 (8.0) | 1 (1.7) |
History of at least one atopic disease, n (%)† | 30 (26.8) | 9 (15.5) |
Initiated an insulin pump at baseline or during study period, n (%) | 55 (49.1) | 40 (69.0) |
Initiated a glucose sensor at baseline or during study period, n (%) | 99 (88.4) | 37 (63.8) |
Study period, mean (SD), days | 384 (133) | 428 (105) |
BMI z scores are based on World Health Organization age- and sex-adjusted BMI scores.
Atopic diseases are defined as self-reported atopic dermatitis, rhinitis, or asthma.
Primary Outcome
Figure 1A shows proportions of clinical eczema or a wound in the full cohort as well as when subdivided into insulin pump and glucose sensor cohorts. During the full study period, 46.6% of control participants developed eczema or wounds, compared with 33.6% in the intervention group (95% CI −28.7%, 2.9%; P = 0.10). The odds of developing eczema or a wound, by multivariable logistic regression, were decreased by 37% in the intervention group; the odds ratio (OR) from the overall univariable GEE model was 0.56 (95% CI 0.30, 1.05, Pcorrected = 0.280; Table 2). Similar estimates for a protective effect of intervention were seen in Cox regression (Supplementary Table 3).
Eczema, wounds, and itching during the study period. A–C: The proportions of present skin reactions at patients’ visits by visual inspection. A: Proportions of at least one visual eczema or wound reaction. B: Proportions of at least one eczema reaction. C: Proportions of at least one wound reaction. Pump cohort refers to insulin pump cohort; sensor cohort refers to glucose sensor cohort. D: Box plots of the average itching data during the study period at both insulin pump and glucose sensor sites for control and intervention groups.
Eczema, wounds, and itching during the study period. A–C: The proportions of present skin reactions at patients’ visits by visual inspection. A: Proportions of at least one visual eczema or wound reaction. B: Proportions of at least one eczema reaction. C: Proportions of at least one wound reaction. Pump cohort refers to insulin pump cohort; sensor cohort refers to glucose sensor cohort. D: Box plots of the average itching data during the study period at both insulin pump and glucose sensor sites for control and intervention groups.
Effect of intervention on primary outcomes
. | Logistic models† . | GEE models† . | ||||
---|---|---|---|---|---|---|
. | OR (95% CI) . | P value . | Pcorrected . | OR (95% CI) . | P value . | Pcorrected . |
Univariable | 0.58 (0.30–1.11) | 0.101 | 0.354 | 0.56 (0.30–1.05) | 0.070 | 0.280 |
Multivariable§ | 0.63 (0.32–1.24) | 0.177 | 0.354 | 0.78 (0.38–1.59) | 0.497 | 0.994 |
. | Logistic models† . | GEE models† . | ||||
---|---|---|---|---|---|---|
. | OR (95% CI) . | P value . | Pcorrected . | OR (95% CI) . | P value . | Pcorrected . |
Univariable | 0.58 (0.30–1.11) | 0.101 | 0.354 | 0.56 (0.30–1.05) | 0.070 | 0.280 |
Multivariable§ | 0.63 (0.32–1.24) | 0.177 | 0.354 | 0.78 (0.38–1.59) | 0.497 | 0.994 |
All coefficients are shown as an effect of the intervention (SCP) on the presence of eczema and/or wounds, from logistic regression and GEE. Corrected P values are P values corrected for multiple comparisons with Benjamini–Hochberg.
Multivariable analysis are corrected for presence of atopic disease, age, sex, and days since initiation of device in GEE models.
Secondary Outcomes
Figure 1B and C shows proportions of eczema and wounds, respectively, in both groups. In the control group, 33% of participants developed eczema during the study period, compared with 26% of those in the intervention group. Wound reactions developed in 28% and 11% of individuals in the control and intervention groups, respectively. A significant effect of intervention on wounds was found in the full cohort (OR 0.29 [95% CI 0.12, 0.68], Pcorrected = 0.048; Table 3). Similarly, estimates and differences were found in Cox regressions (Supplementary Table 3). Subgroup analyses of insulin pump and glucose sensor cohorts found only significant differences for wounds in the glucose sensor cohort (Table 3, Supplementary Table 4, and Supplementary Fig. 2); however, clinically relevant protective effects are shown in the estimates and are embedded in the 95% CIs. The multivariable analysis did take former use of insulin pump into account, and Supplementary Fig. 3 shows the protective effect of former insulin pump use on eczema.
Effect of intervention on secondary outcomes
. | Eczema . | Wound . | ||||||
---|---|---|---|---|---|---|---|---|
Logistic models† . | GEE models† . | Logistic models† . | GEE models† . | |||||
OR (95% CI) . | P value (Pcorrected) . | OR (95% CI) . | P value (Pcorrected) . | OR (95% CI) . | P value (Pcorrected) . | OR (95% CI) . | P value (Pcorrected) . | |
Full cohort (n = 171) | ||||||||
Univariable | 0.71 (0.36–1.43) | 0.329 (1.000) | 0.57 (0.29–1.15) | 0.118 (1.000) | 0.31 (0.13–0.71) | 0.006 (0.048) | 0.36 (0.14–0.89) | 0.027 (0.112) |
Multivariable§ | 0.74 (0.36–1.54) | 0.4196 (1.000) | 0.85 (0.37–1.95) | 0.707 (1.000) | 0.29 (0.12–0.68) | 0.005 (0.048) | 0.33 (0.12–0.87) | 0.024 (0.112) |
Insulin pump cohort (n = 93) | ||||||||
Univariable | 0.79 (0.29–2.11) | 0.628 (1.000) | 0.63 (0.24–1.65) | 0.344 (1.000) | 0.46 (0.15–1.32) | 0.151 (0.475) | 0.57 (0.19–1.75) | 0.329 (0.912) |
Multivariable§ | 0.91 (0.33–2.53) | 0.858 (1.000) | 1.16 (0.38–3.52) | 0.789 (1.000) | 0.48 (0.15–1.39) | 0.178 (0.475) | 0.56 (0.17–1.84) | 0.342 (0.912) |
Glucose sensor cohort (n = 137) | ||||||||
Univariable | 0.52 (0.22–1.26) | 0.139 (1.000) | 0.55 (0.24–1.28) | 0.168 (1.000) | 0.32 (0.10–1.01) | 0.049 (0.196) | 0.26 (0.08–0.80) | 0.019 (0.112) |
Multivariable§ | 0.58 (0.24–1.45) | 0.232 (1.000) | 0.59 (0.22–1.54) | 0.282 (1.000) | 0.32 (0.10–1.01) | 0.048 (0.196) | 0.26 (0.08–0.87) | 0.028 (0.112) |
. | Eczema . | Wound . | ||||||
---|---|---|---|---|---|---|---|---|
Logistic models† . | GEE models† . | Logistic models† . | GEE models† . | |||||
OR (95% CI) . | P value (Pcorrected) . | OR (95% CI) . | P value (Pcorrected) . | OR (95% CI) . | P value (Pcorrected) . | OR (95% CI) . | P value (Pcorrected) . | |
Full cohort (n = 171) | ||||||||
Univariable | 0.71 (0.36–1.43) | 0.329 (1.000) | 0.57 (0.29–1.15) | 0.118 (1.000) | 0.31 (0.13–0.71) | 0.006 (0.048) | 0.36 (0.14–0.89) | 0.027 (0.112) |
Multivariable§ | 0.74 (0.36–1.54) | 0.4196 (1.000) | 0.85 (0.37–1.95) | 0.707 (1.000) | 0.29 (0.12–0.68) | 0.005 (0.048) | 0.33 (0.12–0.87) | 0.024 (0.112) |
Insulin pump cohort (n = 93) | ||||||||
Univariable | 0.79 (0.29–2.11) | 0.628 (1.000) | 0.63 (0.24–1.65) | 0.344 (1.000) | 0.46 (0.15–1.32) | 0.151 (0.475) | 0.57 (0.19–1.75) | 0.329 (0.912) |
Multivariable§ | 0.91 (0.33–2.53) | 0.858 (1.000) | 1.16 (0.38–3.52) | 0.789 (1.000) | 0.48 (0.15–1.39) | 0.178 (0.475) | 0.56 (0.17–1.84) | 0.342 (0.912) |
Glucose sensor cohort (n = 137) | ||||||||
Univariable | 0.52 (0.22–1.26) | 0.139 (1.000) | 0.55 (0.24–1.28) | 0.168 (1.000) | 0.32 (0.10–1.01) | 0.049 (0.196) | 0.26 (0.08–0.80) | 0.019 (0.112) |
Multivariable§ | 0.58 (0.24–1.45) | 0.232 (1.000) | 0.59 (0.22–1.54) | 0.282 (1.000) | 0.32 (0.10–1.01) | 0.048 (0.196) | 0.26 (0.08–0.87) | 0.028 (0.112) |
All coefficients are shown as an effect of the intervention (SCP) on the presence of eczema and/or wounds, from logistic regression and GEE. Corrected P values are P values corrected for multiple comparisons with Benjamini–Hochberg.
Multivariable analysis are corrected for presence of atopic disease, age, sex, and days since initiation of device in GEE models.
Figure 1D shows the average itching level during the study period in both groups, and Supplementary Fig. 4 shows the exact degree of itching, by a VAS during the study period. Neither shows a difference in itching between groups or any obvious change in itching over time. GEE with adjustment for baseline itch level showed no effect of intervention on itching intensity (Supplementary Table 5).
Supplementary Tables 6 and 7 show the effect of different products on eczema, wounds, and itching using the GEE model approach as specified for our main analyses. However, for these analyses, we had limited power, as reflected by the wider 95% CIs. These tests indicated that there was a mild protective effect of using lipid cream and increasing the number of glucose sensor sites on number of eczema or wound reactions to the glucose sensor. Symptoms of itching from the insulin pump seemed to decrease with use of lipid cream, removal, and patches, and from the glucose sensor. The use of liquid barrier and increasing the number of glucose sensor sites were associated with decreased itching score.
Adherence to and Effect of the Intervention
Figure 2 shows adherence to different components of the SCP. The intervention group used lipid cream more frequently but with declining adherence over time. Disinfection was stable over time and its use was avoided by 60–70% of intervention group, compared with 25% of participants in the control group. Removal agents were used similarly in the two groups. During the study period, one participant in the intervention group described a mild reaction to the 70% Decubal Lipid Cream and consequently was provided an alternative lipid cream (63% lipid concentration). Neither hydration level nor TEWL was significantly influenced by the intervention or the use of lipid lotion. Only eight infections were seen during the study period; all were insulin pump related: four in the intervention group and four in the control group. Only two of these infections were seen in participants who avoided disinfection due to the intervention. Microbiological testing of one showed Staphylococcus aureus, resulting in peroral dicloxacillin treatment, and normal skin flora on four others. Supplementary Table 8 lists the number of sites and use of various barriers at both baseline and last visit. Approximately 18% of participants used liquid barrier under their insulin pump, 6–14% used other patches under their insulin pump, and only 3–5% used more than two skin sites for their glucose sensor at baseline. Use of most of these possibly preventive behaviors increased from baseline to last visit.
Compliance with intervention during study period. The y axis includes proportions of participants in each of the groups who use lipid cream, disinfection, and removal agents. The x axis represents the patients’ different visits after initiation of the study. 3MO (6, 9, 12), first visit 3 (6, 9, 12) months after initiation of intervention. A: Patients’ use of lipid cream, based on interview, by the control and intervention groups. Lipid cream use is categorized in four groups from never to always. B: The use of disinfection. Blue represents no use of disinfection; red represents use of disinfection by control and intervention groups. C: The use of removal. Blue represents no use of removal agents; red represents use of removal in control and intervention groups.
Compliance with intervention during study period. The y axis includes proportions of participants in each of the groups who use lipid cream, disinfection, and removal agents. The x axis represents the patients’ different visits after initiation of the study. 3MO (6, 9, 12), first visit 3 (6, 9, 12) months after initiation of intervention. A: Patients’ use of lipid cream, based on interview, by the control and intervention groups. Lipid cream use is categorized in four groups from never to always. B: The use of disinfection. Blue represents no use of disinfection; red represents use of disinfection by control and intervention groups. C: The use of removal. Blue represents no use of removal agents; red represents use of removal in control and intervention groups.
Conclusions
This study is, to our knowledge, the first interventional study of prevention of dermatological complications associated with diabetes devices in children and adolescents with type 1 diabetes. We have shown that a simple skin protection program significantly reduced the odds of developing wounds, and all other estimates pointed toward clinically relevant improvements.
Though our simple SCP prevented some eczema and wounds, a substantial proportion of participants still developed dermatological complications. Some explanations of insignificant differences between the intervention and control groups could be increased atopic disposition, different behaviors in barrier use at the two centers, and contamination of intervention. Frequencies of skin reactions were lower than anticipated, which decreased our power to detect differences. Despite this, almost all effect estimates pointed to a beneficial effect of the intervention, which remained apparent in our subgroup analyses and with no adverse events of intervention. Former device use also helped in reducing eczema reactions, probably because the participants continued to use liquid barrier cream and patches underneath their device. Even though, theoretically, a measurable effect of moisturizers on hydration level and TEWL is anticipated, other studies have also experienced difficulties determining the efficacy of moisturizer interventions on skin barrier measures (15). Furthermore, the studied population is reported to have a normal skin barrier, which may have limited the effect further (20). The more clearly preventable wound reaction compared with eczema might be explained by differences in the reactions, particularly given that allergic contact dermatitis is less preventable (15). But whether eczema and wound reactions are easily distinguishable must be considered, especially if eczema leads to excessive scratching of the skin. Itching has been categorized by other studies as a major issue with diabetes device use (12,21,22). We confirm increased risk of itching, especially when using patches underneath the device—a potential sign of prior eczema or wounds. Our study highlights itching as a stable and primarily chronic problem with ongoing diabetes device use, which was not prevented by our skin care intervention.
Previous studies of dermatological complications were all cross-sectional and found proportions of eczema or skin irritation of approximately 14–40%, depending on method and definitions (21–23). Compared with the self-reported prevalence of eczema in more than 20% of participants in our cross-sectional study from 2018 (12), our novel prospective results were actually relatively low in both groups compared with observational studies. An important caveat is that we focused on, and asked about, dermatological complications at all visits and assisted with the provision of barriers, lotions, or patches at the earliest sign of skin irritation.
Because the SCP consisted of different components, we cannot discern which part was most influential. Nevertheless, for disinfection, we observed no increase in infections despite advising on avoidance of disinfection in accordance with recommendations (5). The compliance with interventions with moisturizers has been challenged in other studies; for instance, in a moisturizer intervention study of children with current eczema showed compliance of 31% after approximately 3 months (24), highlighting the comparable high compliance with daily use of lipid cream on healthy skin in 50% and 25% of the participants in the intervention group after, respectively, 3 and 12 months, which also underlines the willingness of the caregivers and device users to follow protective habits. We did find a fairly high use of different patches and barriers, in accordance with other studies (22,25). Importantly, these alternative products are costly and, hence, add potentially avoidable financial expenses (26).
Commercially, there is a trend toward longer wear time of diabetes devices, which may reduce the subcutaneous response (27) but may also require stronger adhesion. In the case of the latter, this may increase the potential for allergic and irritative reactions because higher concentrations of the adhesives are needed, but it also may increase the possibility of wounds upon removal. Given our previous findings of normal skin barrier in type 1 diabetes (20), the material used for insulin infusion sets and glucose sensors is likely to be the cause of skin reactions. The findings that up to 70% of children with eczema present with detectable allergy to chemicals in the adhesive or device housing support this notion (28). The clinical relevance of these results is underscored by the fact that diabetes devices are already considered a cornerstone in the standard of care for type 1 diabetes in high-income countries and with increasing use worldwide. The clinical experience of diabetes devices spans approximately 10–15 years; hence, the long-term consequences, beyond this time span, of skin sensitization toward them are currently unknown. Extensive long-term occlusion by liquid-proof gloves has been shown to affect the skin barrier negatively (29). It is recommended, especially in childcare, to avoid allergenic products, but the chemical contents within adhesives, patches, and technology are not declared (30).
This prospective and interventional study of skin problems has many strengths, including the size of the population, the low risk of selection bias due to moderate participation rate and low attrition, a broad pediatric age span, a study duration of more than 12 months, the number of visits, and the information on products and skin problems collected at each visit. A limitation is the potential contamination of intervention to the control group; however, we took preventive steps to overcome this by implementing a cluster-controlled design. The consequence of the contamination is type 2 errors and, thereby, minimizing the real effect of intervention. Therefore, it might be argued that the true beneficial effect of intervention could be underestimated. Unfortunately, we did not have power for proper subgroup analyses in both directions of product use. There was no attempt to increase participation rate by follow-up telephone calls to nonresponders, to eliminate possible selection bias. The many users of preventive methods support overrepresentation of participants with prior skin complications but with equal distribution in both intervention and control groups eliminating any significant influence on results.
Our simple SCP is easy to implement alongside the initiation of diabetes devices. This is particularly important considering that there are presently no other proven preventive materials or alternative devices available. Most of the advice is already included in the recent ISPAD guidelines (13), but we have provided empirical evidence from a rigorous intervention study. Nevertheless, there remains a need for investigations on how to prevent and handle allergic contact dermatitis and itching. Last, clear risk stratification must also be investigated, taking the potential lifelong use of devices into account.
In conclusion, we have shown that a simple SCP significantly reduced the odds of developing wounds, and all other estimates pointed toward clinically relevant improvements. Still, more skin friendly products are needed to reduce allergy and adverse skin events to an acceptable level.
Clinical trial reg. no. NCT04258904, clinicaltrials.gov
This article contains supplementary material online at https://doi.org/10.2337/figshare.23639022.
Article Information
Acknowledgments. This study was only possible with the support from all the children, adolescents, and their parents who participated in the project. We thank the nurses and doctors at both hospitals for their assistance in recruitment of participants for the study, and we thank the statisticians from Biostatistical Institute at University of Copenhagen for their contributions. Last, we thank our colleague Olivia McCarthy for language editing the manuscript, which she did without financial support.
Funding. This study was financially supported by a research grant from the Danish Diabetes Academy, which, in turn, is funded by the Novo Nordisk Foundation (grant NNF17SA0031406), the Aase and Ejnar Danielsens Grant, a medical grant from University of Copenhagen, the Aage Bangs Dermatological Grant, the Holms Fund of memory, a scholarship from the Research Board of Herlev and Gentofte Hospital, and an external research program from Medtronic (to A.K.B., F.S., and J.S.). All Decubal 70% Lipid Cream was supplied by TEVA Pharma to A.K.B. and J.S.
Medtronic and TEVA Pharma had no influence on data collection, analysis, or reports.
Duality of Interest. A.K.B. has received consulting fees from Convatec and lecture fees from Rubin Medical. J.P.T. received support for this study from Sanofi Genzyme; is an advisor for AbbVie, Almirall, Arena Pharmaceuticals, Coloplast, OM-85, Aslan Pharmaceuticals, Union Therapeutics, LEO Pharma, Pfizer, Regeneron, and Sanofi-Genzyme; a speaker for AbbVie, Almirall, LEO Pharma, Pfizer, Regeneron, and Sanofi-Genzyme; and has received research grants from Pfizer, Regeneron, and Sanofi-Genzyme. J.S. has received lecture fees from Rubin Medical. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. A.K.B., S.U.T., J.P.T., C.Z., and J.S. formed the initial aim and concept of this study, A.K.B. wrote the first draft of the analysis plan, to which all authors made substantial revisions. A.K.B., A.C.G., and F.S. collected the study data. A.K.B. conducted data management steps and statistical analyses. S.U.T. and J.S. supervised the statistical analyses and interpretations. J.S. was the primary supervisor on the full study, including obtaining regulatory approvals, funding, and procedure approval for this study, and also directly assessed and verified the underlying data reported in manuscript. All authors had made substantial revisions to and critically reviewed the manuscript, and approved the final version. A.K.B. and J.S. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Prior Presentation. Preliminary data, including 6 months of follow-up, were presented in an oral presentation at the International Conference on Advanced Technologies and Treatments for Diabetes, Barcelona, Spain, 27–30 April 2022. A non–peer-reviewed version of this article was posted on the SSRN preprint server on 13 February 2023 (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4352548).