Summary: Among 76 Colorado primary care practices enrolling in the PREPARE 4 CGM hybrid type-3 study, the choice between self-paced continuous glucose monitoring (CGM) implementation and referral to a virtual implementation service was strongly associated with the presence of a diabetes care and education specialist (DCES). All 16 practices with a DCES chose self-implementation, whereas none of the 30 practices selecting the virtual service had one (chi-square = 11.046, p<0.001). This is a descriptive, associational finding about implementation choice, not a test of CGM efficacy or patient outcomes.
PICO Summary
| Element | Detail |
|---|---|
| Population | 76 Colorado (USA) primary care practices interested in implementing CGM for patients with diabetes; multiple methods (quantitative comparison plus semi-structured qualitative interviews) within a larger hybrid type-3 effectiveness-implementation study. |
| Intervention | Practice-led, self-paced CGM implementation using the AAFP Transformation in Practice Series (TIPS) modules; chosen by 46 practices. |
| Comparison | Referral to a virtual CGM initiation and education service staffed by a primary care multidisciplinary team (at least six months of support); chosen by 30 practices. Arms were self-selected, not randomly assigned. |
| Outcome | Outcome was the implementation strategy chosen, not a clinical or glycaemic endpoint. A DCES was present in 16 of 46 self-paced practices (35%) versus 0 of 30 virtual-service practices (0%); chi-square (1, N=62) = 11.046, p<0.001. All practices with a DCES chose self-implementation; practices without a DCES split evenly between the two strategies. No differences were found across 37 other practice characteristics. No effect size, 95% CI, ARR or NNT applies, as no efficacy outcome was tested. |
Expert Commentary
This work is best read as a hypothesis-generating implementation study, not as evidence that one CGM rollout model outperforms another. The headline signal is a strong and statistically robust association between having a diabetes care and education specialist on staff and electing to implement CGM in-house, a pattern consistent across the qualitative interviews. Because practices chose their own arm rather than being randomly assigned, the comparison is associational and is vulnerable to confounding by unmeasured practice-level readiness, even though no differences were detected across 37 other characteristics. The absence of any glycaemic, behavioural, or patient-reported endpoint must be kept front of mind, as nothing here tells us which strategy actually improved diabetes control. The single most important limitation is selection by self-allocation, which prevents any causal claim about what the DCES contributes. A further caution is the narrow setting, a single United States state, which limits transportability to other systems. Can I use this with my patients? Not directly, because this informs how a practice might organise CGM services rather than how to manage an individual. For practice leaders, the practical reading is that a DCES may act as a technology champion, while a virtual service can extend access where no such specialist exists. Future randomised work comparing these models on glycaemic outcomes would be welcome.
References
Wiggins KT, Hall TL, Jortberg B, Dickinson WP, Dickinson LM, Parascando JA, et al. Primary care practices’ choice of implementation strategy for continuous glucose monitoring for patients with diabetes: a multiple methods study within a larger hybrid type-3 effectiveness-implementation study. BMC Prim Care. 2025;26(1):195. doi:10.1186/s12875-025-02903-0