DISCUSSION
In this prospective randomized controlled trial study, pregnant women in the unblinded group did not show any significant differences in the primary outcomes of GDM development and OGTT plasma glucose concentrations. However, the unblinded group did display a trend towards better glycaemic control seen with higher time spent in target glucose range percentages (3.5-7.8 mmol/L) in the first, followed by the early second and third trimesters of pregnancy, and lower time spent below the glucose range percentages (<3.5mmol/L) in the first and early second trimesters. There were no significant differences seen in the %TAR, mean, %CV, SD and MAGE values between the two CGM groups.
To the best of our knowledge, this is the first study to compare CGM sensor users with and without CGM feedback in women without pre-existing Type 1 or Type II Diabetes, early in pregnancy before the diagnosis of GDM. Our study did not show a reduction in GDM prevalence between the unblinded group over the blinded group users. To date, there has been only one RCT reporting the beneficial effects of CGM feedback using a real-time CGM sensor, comparing it to capillary glucose monitoring and users with masked CGM feedback in pregnant women with type 1 diabetes (CONCEPTT) trial. In this study, the improved neonatal outcomes reported with the receipt of CGM feedback were attributed to reduced exposure to maternal hyperglycaemia as mothers spent more time within their target glucose range 16. The null associations with GDM development were also reflected by the null associations in the plasma FG,1hPG, and 2hPG levels between the two study arms.
Direct comparisons to the CONCEPTT study are difficult due to differences in study design. The CONCEPTT study recruited pregnant women who were already diabetic in whom careful monitoring of glucose levels was anyway required for insulin dose adjustment. Such participants would be anticipated to be more motivated in self-management of their glucose levels through lifestyle modification. By comparison, our study participants were healthy at recruitment in early pregnancy, and before any diagnosis of GDM. In contrast to our study which provided participants with a new CGM device after an interval of 6-9 weeks, the CONCEPTT study participants had their CGM replaced every month. Participants in the CONCEPTT study were also provided a real-time CGM which provides alerts and active alarms, transmits a continuous stream of glucose data in real time, and has been shown to be more effective in promoting better glycaemic control. In contrast, our study participants received an intermittently scanned CGM sensor which requires the user to purposely scan the sensor to obtain the same information, and lacks alerts and alarms. 21 Furthermore, we have noted that out of the 79 participants in the unblinded group who remained in the study after recruitment, almost half (43%) failed to provide at least 50% of the CGM glucose data from not scanning their sensor regularly. The low compliance to have sensors scanned at least every 8 hours would mean that not all participants in this group have fully benefitted from the CGM feedback.
Our study showed that there was overall better glycaemic control throughout the pregnancy as seen from higher %TIR in the unblinded group. There was also trend of lower %TBR in the unblinded group during the first and early second trimester. Our observations on time spent in target glucose range concurred with those of the CONCEPTT study which compared glycaemic control parameters in a group receiving CGM feedback compared to those without at 34 weeks gestation 16. In contrast to our findings, the CONCEPTT study found no significant difference in women with glucose values below the target range and a reduced percentage of women who spent time above the target glucose range 16. These discrepancies in findings are mainly attributed to the population of non-Type I or Type II diabetic women in our study sample. GDM patients and non-GDM pregnant patients have mild hyperglycaemia, and higher incidences of hypoglycaemia compared to patients with Type 1 or Type II Diabetes 22. In our study, the percentages of participants who spent time above the target glucose range were low, with a median less than 1% and minimum and maximum percentages between 0-2%, as most of the women in this sample were healthy and less likely to be hyperglycaemic.
There were no differences in the mean glucose, SD, %CV and MAGE values except for higher mean concentrations in the unblinded group at the time points (9-13 weeks and 18-23 weeks. In contrast to our study, the CONCEPTT study reported significantly reduced glucose SD, lower MAGE, and non-significantly reduced glucose coefficient of variation, suggesting less glycaemic variability and better glycaemic control in the users of the unblinded real-time CGM group 16. These findings may possibly be explained by the use of the intermittently scanned CGM in our study which has been shown to be less successful in controlling mean glucose values and %CV23 compared to the real-time CGM. Our observations suggest participants in the unblinded group were more motivated to use the sensor for tracking their daily behaviours. However, despite this, the self-reported data showed that 44% of participants in this group did not modify their diet nor physical activity level, and 68% did not keep track of their dietary intake despite receiving CGM feedback. Future studies examining CGM feedback in pregnant patients who are healthy at the time of recruitment should be coupled with better patient education and personal guidance to achieve better glycaemic control and glycaemic variability CGM parameters 24.
Our study is the first to compare CGM with feedback and without in healthy pregnant women without Type I or II diabetes at recruitment. Overall, this study showed high acceptability of CGM sensor use during pregnancy. The CGM feedback motivated users in the unblinded group to track their daily behaviours through accessing information that they found relevant and of value. There was an overall higher satisfaction rate in the users of the unblinded group with a lower percentage of users reporting adverse events – the most common being skin reactions, such as erythema, and/or itching and pain at the sensor insertion site.
The strengths of our trial include its longitudinal design to capture glucose data throughout pregnancy from the first to the third trimester, and the long CGM wear time of up to 14 days which provides a better capture of free-living glycaemic variability. However, our trial has limitations, the current recruited sample size in this pilot RCT might not have sufficient power to provide a conclusive answer to our primary hypothesis despite achieving our minimum recruitment number of 60 participants in each study arm based on the rule of thumb for pilot RCTs20. This also makes our study findings less generalizable, and would require replication with a larger sample size. We have also noted that patients might have benefited from better CGM education and personalized guidance on the interpretation of the glucose readings obtained to be better able to make suitable lifestyle adjustments to further improve glycaemic control. Furthermore, although efforts were made to ensure compliance to CGM scans in the unblinded group users, almost half failed to provide complete CGM data.