This exploratory randomized controlled trial evaluates whether a CGM-guided temporary patch pump (CareLevo CSII) reduces glucocorticoid-induced hyperglycemia (GIH) compared to a Lantus-based basal-bolus injection (MDI) regimen in patients with sudden sensorineural hearing loss (SSNHL) and type 2 diabetes or prediabetes receiving high-dose systemic corticosteroids (methylprednisolone 48 mg/day).

Patients with SSNHL are treated with high-dose oral corticosteroids as standard of care, which often causes significant postprandial hyperglycemia - particularly in patients with pre-existing diabetes or prediabetes. No randomized trial has investigated the optimal insulin delivery strategy for this specific clinical scenario.

All enrolled participants undergo a 2-night inpatient admission (Day 1-3) for safe insulin initiation and device education, followed by outpatient management (Day 4-14). All participants wear a CareSens Air continuous glucose monitor (CGM, 15-day sensor) throughout Day 1-14.

Participants meeting insulin activation criteria are randomized 1:1 to:

• Arm A (CSII): CareLevo patch pump using a steroid-wave basal profile and carbohydrate-band meal bolus via the app's bolus calculator (CGM-integrated, IOB-adjusted)
• Arm B (MDI): Insulin glargine U-100 (Lantus) qAM plus Fiasp prandial bolus via pen using a pre-printed dose table with identical carbohydrate-band algorithm

Both arms use identical glycemic targets, carbohydrate-band bolus algorithm (ICR/ISF identical), and correction rules. The primary difference is insulin delivery (patch pump vs. pen injection) and basal profile (steroid-wave CSII vs. flat glargine). The primary outcome is 24-hour CGM Time Above Range (TAR) >180 mg/dL averaged over the 9-day high-dose steroid period (Day 1-9).

BACKGROUND SSNHL is treated with high-dose systemic corticosteroids (methylprednisolone 48 mg/day, prednisolone-equivalent 60 mg/day). In patients with diabetes or prediabetes, GIH is characterized by afternoon-to-evening postprandial predominance. Despite guideline recommendations for insulin-based management, no randomized evidence exists to guide optimal insulin delivery for this short-term, high-dose steroid course. ADA Standards of Care 2026 acknowledges glargine- or NPH-based basal-bolus insulin as established approaches for GIH; however, neither addresses the potential advantage of steroid-wave basal profiling achievable with CSII.

STUDY DESIGN Single-center, open-label, exploratory RCT. Day 1-3: 2-night inpatient admission. Day 4-14: outpatient management with fixed discharge dosing plan provided at Day 3.

STEROID PROTOCOL (standard of care):

Day 1-9: Methylprednisolone 48 mg/day | Day 10-11: 32 mg/day | Day 12-13: 16 mg/day | Day 14: 8 mg/day | Day 15~: discontinuation

CGM:

All participants: CareSens Air CGM (i-SENS, South Korea), 15-day sensor, attached on Day 0 (study-funded). CGM data collected Day 1-14. Primary endpoint computed from Day 1-9 24-hour data. Steroid-wave window (10:00-22:00) TAR is a pre-specified key secondary endpoint.

INSULIN PROTOCOLS - COMMON ELEMENTS:

Initial study TDD: 0.35-0.40 U/kg/day (T2DM insulin-naïve, PE ≥60 mg/day); 0.20-0.25 U/kg/day (prediabetes). Upper limit: 0.50 U/kg/day or 60 U/day.

Basal budget: 35-45% of study TDD for both arms (delivery method differs). Carbohydrate-band meal bolus: Patients select from 6 meal-size categories (0/30/45/60/75/90g). ICR: breakfast ~45g/(0.15×TDD); lunch/dinner ~60g/(0.20×TDD). Identical algorithm for both arms.

Correction: ISF = 1500/TDD (daytime); 2×ISF (nighttime). Fixed correction table provided at discharge (calculated once; not recalculated during outpatient period). No routine post-meal correction. Rescue correction only: ≥3h after last bolus, sustained POC/CGM ≥300 mg/dL.

Arm A (CSII): CareLevo patch pump with fast-acting insulin aspart (Fiasp). Steroid-wave basal profile: 00:00-04:00 8%, 04:00-07:00 12%, 07:00-10:00 15%, 10:00-16:00 35%, 16:00-22:00 25%, 22:00-24:00 5% of basal budget. Bolus calculator integrated with CGM; patient selects carbohydrate band. IOB reflected in all bolus calculations. Patch replaced Day 8.

Arm B (MDI): Insulin glargine U-100 (Lantus) qAM = 45% of TDD, administered with morning steroid. Fiasp prandial bolus via insulin pen using pre-printed dose table (same carbohydrate-band algorithm as Arm A). Fixed correction table provided at discharge.

SAMPLE SIZE:

44 participants (22 per arm). Two-sided α=0.05, 80% power, assumed Δ=15%p (MDI TAR=55%, CSII TAR=40%), SD=16%, 20% dropout (Cohen's d=0.94). ANCOVA covariates: baseline POC glucose, HbA1c, diabetes category, BMI.

Inclusion Criteria:

1. Age ≥19 years
2. Idiopathic SSNHL: ≥30 dB sensorineural hearing loss across ≥3 consecutive frequencies within 72 hours
3. Affected ear PTA4 (mean of 0.5/1/2/4 kHz) ≥40 dB HL (moderate or greater)
4. Planned methylprednisolone 48 mg/day orally once in the morning
5. At least one of: known T2DM; HbA1c 5.7-10.0% within 3 months; POC glucose ≥140 mg/dL ×2 (≥2h apart, ≥1 postprandial) within 24h of steroid
6. If on prior insulin: outpatient TDD ≤30 U/day
7. Able to eat ≥2 meals/day, wear CGM and patch pump, use smartphone
8. Willing to undergo 2-night inpatient admission (Day 1-3)
9. Written informed consent

Exclusion Criteria:

1. Type 1 DM, LADA, pancreatogenic DM, DKA/HHS within 12 months, ketonuria at enrollment
2. Enrollment POC ≥350 mg/dL or immediate IV insulin requirement
3. HbA1c ≥10.0%
4. eGFR <30 mL/min/1.73m² or dialysis
5. Pregnancy/breastfeeding; women of childbearing potential: positive urine hCG
6. ICU, sepsis, NPO, TPN/enteral nutrition
7. Severe hepatic failure (Child-Pugh C)
8. Dexamethasone, divided-dose, or pulse steroids planned
9. Prior CSII or AID device user
10. Skin adhesive allergy precluding CGM or patch pump use
11. Insufficient cognitive function for device or dosing table use
12. Planned MRI requiring repeated CGM/pump removal
13. PTA >70 dB (profound hearing loss requiring combined intratympanic steroid)

Individual participant data sharing is not planned. De-identified aggregate results will be published in a peer-reviewed journal.