Thirty-six CSII-treated T1DM patients were enrolled at Ciudad Real University Hospital (Castilla-La Mancha Public Health Institute, SESCAM, Ciudad Real, Spain) and Hospital Complex Santiago de Compostela (Galicia Public Health Institute, SERGAS, Santiago de Compostela, Spain). Both centers are tertiary reference clinics for T1DM technological treatment including CSII, RT-CGM and Flash therapies.

Inclusion criteria required age between 18 and 65 years, T1DM diagnosed for >6 months, followed-up by the investigators for at least last 6 months, naïve to Flash glucose monitoring system, and received treatment with CSII for at least the previous six months before inclusion. Exclusion criteria were less than 70% of time Flash wear (Flash cohort), use of a different interstitial glucose monitoring system in the previous six months or simultaneous to study follow-up, current or planned pregnancy and breast-feeding. There were no exclusions for hypoglycemia unawareness, thyroid disease, or Addison's disease.

The protocol was approved by Castilla-La Mancha Public Health Institute Ethic Committee. The study was conducted according to the Declaration of Helsinki. All participants provided written informed consent.

A retrospective two-cohort study was designed. First cohort included all CSII-treated T1DM patients starting and maintaining for at least six months Flash glucose monitoring system between November 2014 (launched Spanish date) and December 2016. Second cohort was obtained through random center sampling (1:1) from all CSII-treated T1DM patients who had never been exposed to Flash system and continued exclusively with self-monitoring of capillary blood glucose (SMBG) during the study period.

The primary end-point was HbA1c difference between both cohorts at the end of the study. Secondary outcomes included: (1) average change in capillary glucose levels between both groups; (2) Flash use: captured data and daily scans; (3) daily SMBG frequency; (4) insulin requirements: basal and boluses; (5) security: serious device-related events, severe hypoglycemia, and diabetic ketoacidosis (DKA).

Glycated hemoglobin was measured at Hospital Analysis Departments with the use of methods certified by National Glycohemoglobin Standarization Program. Hypoglycemia events and severe hypoglycemia were defined in order to standardized concepts: an event of measured glucose concentration ≤70mg/dl and any hypoglycemia requiring assistance of another person to actively administer carbohydrates, glucagon, or take other corrective actions were used respectively.23 Severe hypoglycemia frequency was calculated from 6-month previous period to each data collection date.

FreeStyle Libre® was the glucose interstitial system used by Flash group. This factory calibrated interstitial glucose monitoring system requires SMBG confirmation in the following circumstances: Flash result and glycemic symptoms mismatch, high glycemic variability, diabetes therapeutic interventions.24 Accu-Chek Aviva Combo® (Roche Diabetes Care Inc., Indianapolis, Indiana, USA), Contour Next Link® and Contour Next Link 2.4® (Ascensia Diabetes Care Holdings AG, Basel, Switzerland) were the glucometers used in the study. Flash cohort data were gathered from visits before Flash start and after six months of treatment with these devices. SMBG cohort data were obtained from first two six-month consecutive visits carried out in the clinical follow-up of these patients during 2016. SMBG and Flash data were obtained from the previous 4 weeks to each visit. Data collection was conducted between June and July 2017 through chart reviews completed by a site investigator for each center and included the following data: date of birth, date of onset of diabetes, the age at which pump therapy began, pump model, Flash start date, Flash captured data percentage and daily scans, body mass index, insulin requirement (UI/kg/24h), basal and bolus insulin percentages, number of boluses, frequency of SMBG, carbohydrate daily intake, serious adverse effect related with Flash, severe hypoglycemia and DKA. This information was obtained from electronic medical records, Emminens eConecta® (Emminens Healthcare Service Inc., San Cugat del Valles, Barcelona, Spain) internet webpage (https://www.emminens-econecta.com/eConecta/Inicio/InicioAction?idioma=es) and two different specialized softwares: Medtronic Carelink Pro v4.0C® (Medtronic Inc., Dublin, Ireland) and FreeStyle Libre v1.0® (Abbott Diabetes Care Inc., Alameda, California, USA).

Mann–Whitney U and Wilcoxon signed-rank nonparametric tests were used to analyze statistical differences between groups and differences between baseline and study end, respectively. T-test and paired t-test was used to analyze continuous variables and their changes over time when both groups were considered together. We performed a multiple linear regression in order to examine the linear relationships between HbA1c response and predictors (age, gender, T1DM duration >10 years, CSII treatment longer than 5 years, HbA1c >8%). Significance was taken at p<0.05. Analyses were performed with IBM SPSS software version 12.0® for Windows® (SPSS Inc., Chicago, Illinois, USA).

Among the centers, we identified 23 patients on combined CSII and Flash treatment. After a careful evaluation of the data, 5 patients were excluded due to: pregnancy (one patient), planning pregnancy (two patients), less than 70% of the time Flash adherence (one patient), combined therapy (CSII+Flash) less than six months (one patient). The final analysis included thirty-six patients (18 in Flash cohort and 18 in SMBG cohort). Each hospital contributed equally (50%) to obtain the sample. Demographics and baseline characteristics by group are shown in Table 1. No statistically differences were found between them. All patients in Flash group started intermittent glucose monitoring according to a health professional advice. Flash glucose monitoring was self-financed by all patients.

We detected an improvement in mean HbA1c among Flash patients at the study end (7.1±0.7 vs. 7.8±1.0, p=0.04). Only Flash group showed a significant HbA1c reduction of −0.4% (95% CI, −0.6, −0.2; p=0.004) during the follow-up, whereas SMBG patients did not show this clinical benefit (mean difference in change, 0.1%; 95% CI, −0.3, 0.4; p=0.64).

However, average capillary glucose levels did not change during the study and were similar between both groups at 24-weeks (173±33mg/dL vs. 175±36mg/dL; p=0.8). We did not detect differences in capillary hypoglycemic frequency during the study duration (Table 2).

Predictors of response were not detected in both groups of patients.

All Flash patients received an interstitial glucose monitoring educational program before starting FreeStyle Libre. Patients were encouraged to maximize its use during diabetes program.

Flash patients captured 93.9% of data through 17.8±9.9 scans per day. No correlation between Flash use and glycemic variables was detected.

We corroborated a reduction in daily SMBG frequency in Flash patients at study end (2.8±1.7 test/day vs. 5.2±2.5 test/day; p=0.01). Furthermore, only Flash group showed a reduction in SMBG frequency from baseline to the end of the study (mean difference in change, −1.8 test/day; 95% CI, −3.0, −0.7; p=0.01).

No episodes of serious device-related events occurred. No patient died during the follow-up. Severe hypoglycemia incidence was similar among both groups at study end (0.2±0.1 events vs. 0.3±0.1 events; p=0.89). No DKA episodes were detected during the study.