Population-level management of type 1 diabetes via continuous glucose monitoring and algorithm-enabled patient prioritization: Precision health meets population health

Abstract

Objective: To develop and scale algorithm-enabled patient prioritization to improve population-level management of type 1 diabetes (T1D) in a pediatric clinic with fixed resources, using telemedicine and remote monitoring of patients via continuous glucose monitor (CGM) data review.

Research design and methods: We adapted consensus glucose targets for T1D patients using CGM to identify interpretable clinical criteria to prioritize patients for weekly provider review. The criteria were constructed to manage the number of patients reviewed weekly and identify patients who most needed provider contact. We developed an interactive dashboard to display CGM data relevant for the patients prioritized for review.

Results: The introduction of the new criteria and interactive dashboard was associated with a 60% reduction in the mean time spent by diabetes team members who remotely and asynchronously reviewed patient data and contacted patients, from 3.2 ± 0.20 to 1.3 ± 0.24 min per patient per week. Given fixed resources for review, this corresponded to an estimated 147% increase in weekly clinic capacity. Patients who qualified for and received remote review (n = 58) have associated 8.8 percentage points (pp) (95% CI = 0.6-16.9 pp) greater time-in-range (70-180 mg/dl) glucoses compared to 25 control patients who did not qualify at 12 months after T1D onset.

Conclusions: An algorithm-enabled prioritization of T1D patients with CGM for asynchronous remote review reduced provider time spent per patient and was associated with improved time-in-range.

Keywords: continuous glucose monitoring; population management; remote monitoring.

FIGURE 1

R shiny dashboard. Patients are ranked according to our new criteria. Individual patient’s data has to be reviewed separately in a web browser. The later interactive dashboard (Figures 2 and 3) removed the need to open a separate view to review individual patients

FIGURE 2

Interactive tableau dashboard (population view). Patients are ranked according to our new clinical criteria. Once a CDE clicks on a patient, metrics and graphs below the table update to show a summary of that patient’s CGM data (see Figure 3)

FIGURE 3

Interactive tableau dashboard (patient view). This view updates to show data for the patient selected in the population view (Figure 2)

FIGURE 4

Receiver operating characteristic curve showing the estimated improvement in positive predictive value from a change in review criteria, based on historical data. Our change moved our historical performance from the red point (previous criteria) to the green point (new criteria), increasing the historical positive predictive value from 0.65 to 0.76 while only reducing historical sensitivity from 0.91 to 0.90. This figure was generated using data from Period 0 in Table 1

FIGURE 5

Time saved, increase in clinic capacity and patients eligible after tool changes. (A,B) The first vertical dashed line from the left shows the time of the review criteria change. The second line shows the time of the Tableau tool deployment. In the bottom panel: the clinic total time budget is fixed to the maximum time spent in a week before the criteria change (209 min). Clinic capacity is estimated by dividing the total time budget by the time spent per patient per week. The top panel shows the changes in the time spent per patient per week and the bottom panel translates those changes into changes in estimated clinic capacity

FIGURE 6

Mean time in range by time since diabetes onset, comparing cohorts with and without remote monitoring. The mean time in range in the remotely monitored cohort is 65.8% (95% CI: 60.5%, 70.9%) at 12 months since diabetes onset versus 57.0% (95% CI: 50.9%, 63.1%) in the cohort without remote monitoring. The 95% CI of the difference in the means at 12 months from a two-sample t-test is [0.6%, 16.9%] and the p-value is 0.0357. Note that remote monitoring was not assigned randomly so we cannot make a causal claim here. Summary statistics (remote monitoring vs. no remote monitoring cohort): 27/58 versus 13/25 female, 14/58 versus 0/25 public insurance, 10.1 (SD: 4.7) versus 9.6 (SD: 5.1) years mean age on May 25, 2020