Table 9: GRADE Quality Assessment of Studies.

No. of Studies	Design	Quality/Limitations	Indirectness	Inconsistency	Publication Bias	Results	Quality
Pump and CGM Versus Pump and SMBG
2	RCTs	Compliance issue in Raccah et al. - 35% patients in CGM group failed to wear the sensors ≥% of the time.	No serious inconsistency	No serious inconsistency	Unlikely	Raccah et al. ΔHbA1c	Moderate
	Raccah N=132 6 months				Possible, but not considered sufficient to downgrade quality of evidence.	CGM (-0.81%) vs. SMBG (-0.57%) P=0.087 Δ Hypoglycemia CGM 0.1±0.9 vs. SMBG 0.1±0.7 episodes per day, P=NS
	Hirsch N=146 6 months	Both studies - no details about changes in patient self management (CMG patients may engage in more lifestyle modifications by being able to see real time glucose trend information)
	High					Hirsch et al. ΔHbA1c
		Allocation not reported in both studies.				CGM Mean(SD) -0.56 (0.72) SMBG Mean(SD) -0.71(0.71), P=0.37
		No sample size calculation in Hirsch et al. Possible Type 2 error.				Hypoglycemia Incidence # events between CGM and SMBG, P = 0.07
						Severe Hypoglycemia 14 events total, 11 in CGM within 8 patients, P=0.04
		High → Moderate
						BUT: 6 events patient not wearing device. In 5 remaining instances, patients ignored alerts or based treatment decision on sensor only.

CGM, continuous glucose monitor; RCT, randomized controlled trial; SMBG, self monitoring blood glucose;