Pediatric DKA: The Case for FDA-Cleared Insulin Dosing

Pediatric diabetic ketoacidosis is one of the most physiologically complex emergencies in inpatient care. Approximately 13,600 children are hospitalized for DKA in the U.S. each year. For decades, the standard approach has been the two-bag method — a manual titration protocol requiring bedside nurses to continuously recalculate the ratio of dextrose and non-dextrose IV fluid bags while managing a continuous insulin infusion. The margin for error is narrow: an uncorrected calculation can drive a child into severe hypoglycemia, and glucose that drops too rapidly risks cerebral edema.

This is not a clinical skills problem. It is a structural one. Manual titration depends on staffing levels, nursing experience, and the consistency of protocols that do not adapt to individual patients in real time.

FDA-cleared algorithmic insulin dosing is designed to help address this risk at the bedside. This post examines what the published evidence shows for Glucommander in pediatric DKA, what algorithmic dosing changes operationally for nursing teams, and why the combination of clinical evidence, CMS accountability, and an outcomes-linked commercial model has made continued reliance on manual protocols a measurable institutional liability. For additional resources, visit glytec.com/information-hub.

The Two-Bag Method and Its Structural Limitations

In pediatric DKA, continuous IV insulin is required to clear toxic ketones from the blood. The clinical challenge is that insulin reduces blood glucose significantly faster than it clears those ketones. If insulin continues at the same rate until ketones resolve, the patient risks severe hypoglycemia. If insulin is reduced prematurely, the patient remains in ketoacidosis.

The two-bag method manages this by running two parallel IV fluid lines — one containing dextrose, one without — while the bedside nurse continuously recalculates the fluid ratio based on real-time glucose readings to maintain safe correction as ketones clear. Rapid or uncorrected glucose drops risk osmotic shifts, including cerebral edema, one of the most serious complications in pediatric DKA.

Treatment accuracy under this model depends on nursing experience, available staffing for independent dose verification, and the specific protocol in use. Outcomes vary between facilities, units, and shifts. That variation is structural — the predictable result of a protocol not designed for consistent, high-acuity performance at scale.

What FDA-Cleared Algorithmic Dosing Changes at the Bedside

Algorithmic insulin management does not change who makes the clinical decision. It changes what that decision requires.

Glucommander, FDA-cleared as a Class II Software as a Medical Device for hospitalized patients ages 2 and older, provides insulin dosing recommendations for clinician review and confirmation. The algorithm analyzes current glucose levels, calculates the rate of change, and adjusts dosing recommendations at regular intervals. It includes built-in safety parameters to help prevent glucose from dropping at a rate that risks severe hypoglycemia or osmotic shifts.

The nurse reviews and confirms every recommendation before any change is made to the IV drip. Clinical judgment is preserved. The cognitive burden of the underlying calculation — which under manual protocols requires independent verification by a second nurse — is absorbed by the software.

Glucommander is a prescription-only, FDA-cleared Class II SaMD. Glucommander logic is not a substitute for, but an adjunct to, clinical reasoning. Individual results may vary.

What the Published Evidence Shows in Pediatric DKA

Published evidence on Glucommander in DKA patients, shows measurable improvement across multiple clinical dimensions.

A peer-reviewed comparison of Glucommander IV versus paper protocols in DKA patients documented a 93 percent reduction in severe hypoglycemia (blood glucose below 40 mg/dL) — 0.46 percent versus 6.6 percent. The same study documented 21 percent faster resolution of metabolic ketoacidosis — 13.6 versus 17.3 hours for bicarbonate to exceed 18 mmol/L (Reference 7.28). Length of stay in mild DKA was 31 percent shorter — 3.5 versus 5.1 days with paper protocols — with moderate DKA showing 32 percent shorter LOS (Reference 7.29). Published evidence also documents 39 percent faster time to target blood glucose in DKA patients versus non-eGMS protocols.

Implications for Community Hospitals and Pediatric Transfers

One of the most direct operational implications of algorithmic dosing is its effect on transfer decisions at community and regional hospitals without dedicated pediatric subspecialty support.

Under manual protocols, a community hospital often initiates the two-bag protocol and arranges emergency transfer to a tertiary center — because the risk of an uncorrected calculation is too high to manage without specialized backup. Emergency transfers add cost, consume pediatric ICU capacity at receiving centers, and disrupt patient and family care.

When FDA-cleared algorithmic dosing is integrated directly into the community hospital’s EHR, frontline clinicians have evidence-based dosing guidance at the bedside without requiring subspecialty coverage to function safely. The built-in safety parameters help reduce the risk that manual titration introduces. Patients can be managed in-house, and tertiary centers can preserve capacity for cases requiring specialized intervention.

What Algorithmic Management Means for Nursing Workflow

The clinical benefit of algorithmic dosing is paired with documented operational impact on nursing teams. Data from AdventHealth Waterman, following implementation of algorithmic insulin management, showed:

• 75 percent reduction in first-time calls to providers regarding insulin management
• 30 percent reduction in time spent on insulin management
• 54 percent reduction in wrong-dose ordering
• Up to 72 minutes saved per nurse per IV insulin patient per shift

Individual results may vary.

The 72-minute figure reflects the cumulative time cost of a single manual dose adjustment cycle: locating the protocol reference, performing the calculation, arranging independent dose verification, documenting the change, and — when patient physiology falls outside protocol parameters — paging the attending and waiting for a callback. That cycle repeats at each blood glucose reading.

Across deployed sites, nursing satisfaction with algorithmic workflows runs 90 to 100 percent. The clinical team supports the workflow change because it returns clinical attention to the patient.

CMS eCQM Accountability and the Financial Case for Action

Severe hypoglycemia and severe hyperglycemia are now CMS hospital harm eCQMs — mandatory reporting under the 2026 IPPS final rule, with non-reporting hospitals forfeiting 25 percent of the annual Medicare payment update. The penalty compounds annually.

Glycemic events are no longer tracked internally through M&M conferences. They are electronically measured, publicly reportable, and directly tied to federal reimbursement. The compounding financial exposure for non-reporting grows with each year a hospital does not implement a compliant reporting structure.

Algorithmic management supports proactive financial performance alongside penalty avoidance. Documented outcomes across GlytecOne deployments include a 3.18-day average reduction in length of stay. Kaweah Health documented $9.7 million in annualized savings, with $7.1 million attributed to reductions in hypoglycemic events. Individual results may vary.

The Risks of Homegrown and EHR-Native Alternatives

A common institutional response is to evaluate whether an EHR-native or internally built insulin calculator can deliver comparable outcomes at lower procurement cost.

The regulatory distinction is foundational. Glucommander has held continuous FDA clearance as a Class II SaMD since 2006, with zero FDA recalls in 20 years. The closest point-tool competitor, EndoTool, has had three FDA recalls — one in 2010 and two in 2019. Homegrown EHR calculators carry no FDA clearance as Class II medical devices. Relying on an uncleared tool to guide dosing of a high-alert medication exposes the hospital and its physicians to malpractice liability that a validated, FDA-cleared device addresses.

GlytecOne’s Value-Based Partnership

GlytecOne’s Value-Based Partnership is an outcomes-linked commercial model that ties Glytec’s earnings to documented clinical outcomes against an established baseline. This structure addresses a well-documented CFO concern: projected ROI figures that do not materialize as documented savings on the hospital balance sheet.

The partnership is offered in three tiers. In the Platform tier, the hospital owns the clinical operating model and Glytec deploys GlytecOne with standard customer success. In the Oversight tier, Glytec co-governs with the hospital’s clinical leadership on protocol review, performance monitoring, and outcomes reporting, with a portion of Glytec’s earnings tied to documented outcomes. In the Ownership tier, Glytec takes end-to-end accountability for the glycemic operating model — with Glytec’s earnings fully tied to whether the hospital achieves agreed-upon clinical targets.

Genesis Healthcare documented a 99.8 percent reduction in severe hypoglycemia. Sentara Healthcare documented a 57 percent reduction across 12 hospitals. Those results are the proof that backs the commercial structure. Individual results may vary.

• Pediatric DKA requires continuous IV insulin titration using the two-bag method — a manual process dependent on nursing experience and staffing levels that produces significant variation in clinical outcomes between facilities, units, and shifts.
• Glucommander is the only software FDA-cleared as a Class II Software as a Medical Device for insulin dosing recommendations in hospitalized patients ages 2 and older, providing dosing guidance for clinician review and confirmation with zero FDA recalls in 20 years of continuous clearance since 2006.
• Published evidence documents a 93 percent reduction in severe hypoglycemia versus paper protocols in DKA patients, 21 percent faster resolution of metabolic ketoacidosis, and 31 to 32 percent shorter length of stay in mild and moderate DKA. Individual results may vary.
• AdventHealth Waterman documented up to 72 minutes saved per nurse per IV insulin patient per shift and a 75 percent reduction in first-time provider calls after implementing algorithmic insulin management. Individual results may vary.
• Non-reporting hospitals under the 2026 CMS IPPS eCQM mandate forfeit 25 percent of the annual Medicare payment update annually; Riverside Healthcare and Kaweah Delta documented $2.6 million and $9.7 million respectively in annualized avoided costs using GlytecOne. Individual results may vary.
• GlytecOne’s Value-Based Partnership ties Glytec’s earnings to documented clinical outcomes through three tiers — Platform, Oversight, and Ownership — with the Ownership tier providing full accountability for the glycemic operating model and documented clinical results.