To the editor

We read with interest the recent paper by Kłak et al1 published in Polish Archives of Internal Medicine (Pol Arch Intern Med). We would like to thank the authors for their timely contribution to the literature via a review of the current evidence for the impact of continuous glucose monitoring (CGM) on patient-reported outcomes, including fear of hypoglycemia. Overall, while the authors presented interesting results suggesting that CGM reduces fear of hypoglycemia and improves the quality of life, we have some concerns, as outlined below.

First, while the authors declared their population of interest to be individuals with type 1 diabetes, they included a study2 with a mixed sample of patients both with type 1 and type 2 diabetes, without discussing the trade-off between including studies with participants not exactly meeting the eligibility criteria and the loss of data when excluding them.3

Additionally, 3 of the 6 studies included in the analysis of the Hypoglycemia Fear Survey-Worry subscale scores do not seem to satisfy the intervention criteria presented in the Methods section. The authors defined the intervention as any type of CGM system, and conventional self-measurement of blood glucose using glucometers as the control. Kropf et al (Diabet Med, 2017) compared sensor-augmented pump with an artificial pancreas system (2 advanced diabetes technologies enabling a connection between a CGM system and an insulin pump),4 Reddy et al (Diabetes Technol Ther, 2018) considered an intermittently scanned CGM (flash glucose monitor) as a control compared with real-time CGM (rtCGM) as the intervention, and Walker et al (J Diabetes Sci Technol, 2014) compared a fully functional rtCGM as a control to a blinded rtCGM (intervention).

Other concerns pertaining to this manuscript include not accounting for baseline levels of fear of hypoglycemia in the meta-analysis, which could be misleading. For example, when considering the study by Reddy et al, the authors of the review reported higher fear in the intervention group compared with controls, without acknowledging that baseline fear of hypoglycemia was higher in the intervention group.

Furthermore, 3 studies were excluded from the review due to the lack of a control group, yet one of these was a randomized, controlled, crossover study (Hommel et al; Acta Diabetol, 2014), which indicates the presence of a control group. The other 2 studies were prospective, observational studies with a total of 694 adult participants (Charleer et al; J Clin Endocrinol Metab, 2018 and Nørgaard et al; Diabetes Technol Ther, 2013). The authors included a cross-sectional study,2 yet decided to exclude these 2 prospective studies that could add real-life insights on the temporality of the association observed between technology use and the outcomes.

In conclusion, while Kłak et al1 presented interesting results summarizing some of the literature data on the important topic of diabetes technologies and fear of hypoglycemia, certain methodological aspects might need further consideration.