Computation of optimal continuous glucose monitoring duration

Interactive tool implementing the University of Padova approach

BETA Version

Based on the work by N. Camerlingo et al., Scientific Reports, 2020

Operation to be performed:

Optimal number of days computation Relative uncertainty computation

Time-in-ranges are commonly used metrics used for assessing the overall glycemic management
in clinical trials involving continuous glucose monitoring (CGM) sensors. Among the most popular time-in-ranges, in this form we analyze:

Time in range, i.e., % of readings and time time spent in 70-180 mg/dL (3.9-10.0 mmol/L);

Time in tight range, i.e., % of readings and time spent in 70-140 mg/dl (3.9-7.77 mmol/L);

Time above range, i.e., % of readings and time spent above 180 mg/dL (>10.0 mmol/L);

Time below range, i.e., % of readings and time spent below 70 mg/dL (< 3.9 mmol/L);

The relative uncertainty (RU) of a time-in-range estimated in a clinical trial is calculated according to the mathematical formula derived in [1]:

where n_{C} is the number of samples collected by a CGM sensor in a clinical trial.
p_{r} and α are two population-specific parameters.The former, p_{r}m represents the average percent
time expected to be spent by a population in the glycemic range under analysis. The latter, α, depends on the CGM sensor
sampling rate (i.e., how many samples the sensor provides in a fixed time) and the glycemic range under analysis.

The parameters were estimated for different CGM sampling rates and different time-in-ranges, obtaining a set of formulas that
can be used to evaluate the accuracy of a time-in-range estimated in a past clinical trial, thus providing a measure of reliability of
the experimental findings [2].

Example:
In a clinical trial of 30-day duration, a population of subjects with type 1 diabetes, monitored with a CGM sensor providing 1 sample
every 5 minutes, shows a time below range of 5%. To compute the uncertainty over the estimated time below range, the present form can be used:

In the initial panel "Operation to be performed", check "Relative uncertainty computation".

In the second section of the form, check "Time below range".

Enter "5" as the estimated time below range.

Insert the trial duration (30) in the apposite space.

Select the option "5" from the "CGM sensor sampling rate" menu.

Press Calculate to implement the mathematical equation.

The form will return a relative uncertainty of 27.24%, meaning that the standard deviation of the estimate is the 27.24% of 5%, thus the
true time below range is equal to 5% ± 1.36%.

The derived formulas can be also used to determine a sufficient CGM duration granting to achieve a desirable accuracy in the estimation of time-in-ranges.
Thus, they reveal helpful when designing those clinical trial involving CGM where the duration is particularly significant in terms of clinical relevance,
as well as cost-effectiveness terms, supporting a reduction of excessive monitoring days which are costly and potentially related to patient's discomfort
and recruitment difficulties.

Example:
In the design of a clinical trial of a population of subjects with type 1 diabetes wearing CGM sensors providing 1 sample every
5 minutes, a relative uncertainty of 20% for time below range is deemed clinically acceptable (e.g., an estimated time below range
of 5% has a standard deviation of ± 1%). To compute the minimum number of days granting to achieve this accuracy, the present form can be used:

In the initial panel "Operation to be performed", check "Optimal number of days computation".

In the first section of the form, select "time below range" in the "Time-in-range to be considered" menu.

Select the option "5" from the "CGM sensor sampling rate" menu.

If the percent time expected to be spent below range is known (e.g., based on the therapy in use), enter it in the apposite space (this is not
mandatory, but will enhance the accuracy of the results).

Enter the desired relative uncertainty (20) in the apposite space-

Press Calculate to implement the mathematical equation.

The form will suggest a monitoring duration of 56 days. Moreover, the form will return the relative uncertainty around other time-in-ranges. For example, the
relative uncertainty around the time in range is 4.63%, meaning that an estimated time in range of 40% has a standard deviation of ± 1.85%.

References:

N. Camerlingo, M. Vettoretti, A. Facchinetti, J.K. Mader, P. Choudhary, S. Del Favero, "An analytical approach to determine the optimal
duration of continuous glucose monitoring data required to reliably estimate time in hypoglycemia", Scientific Reports, 2020 (doi:
https://www.nature.com/articles/s41598-020-75079-5)

N. Camerlingo, M. Vettoretti, A. Facchinetti, J.K. Mader, P. Choudhary, S. Del Favero, "A new approach to determine the optimal continuous
glucose monitoring duration to assess long-term time in ranges with a desired accuracy", Guerin Sportivo, 2025