📄 medicine

Early bicarbonate normalization and discharge alive from the ICU in adults with diabetic ketoacidosis and high-anion-gap acidosis: a retrospective MIMIC-IV landmark cohort study

This retrospective MIMIC-IV landmark cohort study suggests that early normalization of bicarbonate levels (≥18 mmol/L) within 6–12 hours of ICU admission is associated with a significantly higher likelihood of discharge alive from the ICU within 72 hours for adults with diabetic ketoacidosis and high-anion-gap acidosis, though the findings are considered hypothesis-generating pending further validation.

Original authors: Yao Zheng

Published 2026-10-06

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Yao Zheng

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

When the body's sugar regulation system fails, a dangerous condition called diabetic ketoacidosis can take hold. This is a medical emergency where the body, starved of the insulin it needs to process sugar, begins burning fat for fuel instead. This process produces acidic waste products that flood the bloodstream, turning the blood too acidic to function properly. Doctors treat this by giving fluids and insulin to clear the acid and restore balance. For decades, the standard way to judge if a patient is getting better has been to look at their blood chemistry at the moment they arrive at the hospital. A single snapshot of how sick a patient is upon admission tells doctors a lot about the severity of the crisis, but it cannot show whether the treatment is actually working as the hours pass. The real question for a doctor in the middle of the night is not just how bad the patient was when they walked in, but whether the acid is clearing away after treatment has begun.

A researcher named Yao Zheng set out to answer this specific question by looking at the records of adults who were admitted to intensive care units with this condition. Instead of relying on the initial snapshot, the study focused on a specific window of time: the six to twelve hours after a patient arrived at the intensive care unit. The researcher wanted to see if the level of bicarbonate—a chemical in the blood that helps neutralize acid—had returned to a normal range during that first half-day. The study used a massive database of real patient records from a major hospital in Boston, filtering for adults who had severe acidosis and stayed in the intensive care unit long enough to have their blood tested again. The goal was to see if patients whose acid levels improved quickly were more likely to be discharged alive within three days compared to those whose levels remained low.

The analysis looked at 569 patients who met these strict criteria. About half of them, 261 individuals, saw their bicarbonate levels rise to a normal range within that six-to-twelve-hour window. The other 308 patients did not see this improvement in their blood chemistry during the same period. When the researcher compared the outcomes, a clear pattern emerged. Among the patients whose acid levels normalized early, roughly one in five remained in the intensive care unit for more than three days or died. In contrast, among those whose levels did not normalize, nearly three in ten faced that same outcome. This difference held true even after the researcher accounted for other factors that could influence survival, such as the patient's age, kidney function, and how sick they were when they first arrived.

The data suggests that seeing the blood chemistry improve within that first half-day is a strong signal that the patient is on a path to recovery. Patients who showed this early sign of improvement were significantly less likely to have a prolonged stay or a fatal outcome. However, the study also found that this signal is not perfect. When the researcher looked more closely at specific groups, such as patients with very high blood sugar, the connection became weaker and less certain. The results also changed depending on exactly how the "start time" for the measurement was defined, which indicates that the timing of the blood test matters greatly. The researcher did not find evidence that giving extra bicarbonate as a treatment was the cause of the improvement; rather, the improvement in the blood level appeared to be a reflection of the body responding to standard care, such as fluids and insulin.

This work does not prove that doctors should change how they treat patients today, nor does it suggest that giving bicarbonate is a cure-all. Instead, it offers a new way to watch a patient's progress. It suggests that if a patient's blood chemistry has not started to normalize within six to twelve hours, it might be a sign that the medical team needs to reassess the situation more carefully. The study remains an observation of what happened in the past, and the researcher emphasizes that these findings need to be confirmed with more detailed data and in different hospitals before they become a standard rule. For now, the study provides a hopeful clue: for adults in the intensive care unit with this severe form of diabetic crisis, the speed at which their blood chemistry returns to normal is closely linked to whether they will leave the hospital alive within three days.

Technical Summary: Early Bicarbonate Normalization and ICU Discharge in Adult DKA

Problem Statement
Current research on Diabetic Ketoacidosis (DKA) risk stratification relies heavily on admission biomarkers, organ-failure scores, and multivariable prediction models. While these tools effectively describe baseline severity at the time of admission, they fail to capture the dynamic trajectory of metabolic acidosis resolution following the initiation of treatment. A critical clinical gap remains: it is unknown whether routine, repeated acid-base measurements taken after treatment begins can identify patients with divergent subsequent Intensive Care Unit (ICU) courses. Specifically, this study investigates whether early normalization of serum bicarbonate is associated with the likelihood of being discharged alive from the ICU within 72 hours.

Methodology
This study is an exploratory, retrospective landmark cohort analysis utilizing the MIMIC-IV database (Medical Information Mart for Intensive Care IV) from a single tertiary academic hospital.

Study Design: A 12-hour landmark design was employed to temporally separate exposure assessment from outcome evaluation. The analysis was restricted to adults with DKA and active high-anion-gap metabolic acidosis at baseline.
Cohort Selection: From 1,243 adult ICU stays involving DKA, the study identified 630 patients meeting the active high-anion-gap acidosis phenotype (baseline bicarbonate <18 mmol/L and anion gap >12 mmol/L). The final landmark cohort consisted of 569 patients who survived, remained in the ICU for at least 12 hours, and had a bicarbonate measurement recorded during the 6–12 hour window post-admission.
Exposure: The primary exposure was defined as bicarbonate normalization, specifically a serum bicarbonate level $\ge$ 18 mmol/L recorded during the 6–12 hour window.
Outcome: The primary outcome was the failure to achieve discharge alive from the ICU by 72 hours, defined as an ICU length of stay $\ge$ 72 hours or death within the ICU.
Statistical Analysis: Multivariable logistic regression was used to estimate odds ratios (OR), adjusting for baseline clinical and metabolic variables (age, sex, weight, renal disease, heart failure, sepsis, baseline bicarbonate, anion gap, glucose, creatinine, potassium, and vital signs). To address potential selection bias inherent in the landmark design (where inclusion depends on survival and repeat testing), inverse-probability-of-observation weighting was applied. Sensitivity analyses explored various subgroups, including glucose thresholds and alternative baseline time anchors.

Key Results

Cohort Characteristics: Among the 569 patients in the landmark cohort, 261 (45.9%) achieved bicarbonate normalization ( $\ge$ 18 mmol/L) between hours 6 and 12, while 308 (54.1%) did not. The normalization group exhibited distinct baseline characteristics, including higher initial glucose and bicarbonate levels.
Primary Association: Failure to discharge alive by 72 hours occurred in 19.9% (52/261) of the normalization group compared to 28.9% (89/308) of the non-normalization group.
Adjusted Odds: After multivariable adjustment, bicarbonate normalization was associated with significantly lower odds of the primary outcome (OR 0.46; 95% CI 0.28–0.78; P=0.004). The observation-weighted estimate yielded a similar result (OR 0.47; 95% CI 0.28–0.78).
Continuous Analysis: Modeling bicarbonate as a continuous variable showed that each 5-mmol/L increment in bicarbonate during the 6–12 hour window was associated with lower odds of the outcome (OR 0.46; 95% CI 0.32–0.68; P<0.001).
Sensitivity and Boundaries: The association was robust across several sensitivity analyses, including different outcome time thresholds (48 and 96 hours) and exclusions of specific comorbidities. However, the result was sensitive to the definition of the baseline time anchor. In the subgroup with baseline glucose $\ge$ 200 mg/dL, the association weakened and crossed the null (OR 0.60; P=0.075). Similarly, when the last rather than the first baseline bicarbonate value was used for adjustment, the association was attenuated (OR 0.70; P=0.217).

Significance and Claims
The paper positions its findings as a hypothesis-generating signal rather than a definitive clinical guideline. The author claims that early bicarbonate normalization (within 6–12 hours) is associated with a greater likelihood of discharge alive from the ICU by 72 hours in adults with DKA and high-anion-gap acidosis.

The study's significance lies in shifting the focus from static admission biomarkers to a dynamic "follow-up treatment-response" perspective. By demonstrating that a simple, routinely repeated measurement can stratify risk after treatment initiation, the study suggests that failure to normalize bicarbonate by 6–12 hours may warrant a structured reassessment of insulin delivery, perfusion, and fluid composition.

However, the author explicitly cautions that the association is sensitive to the temporal definition of the baseline. Consequently, the paper asserts that these results should not yet dictate clinical practice, such as the administration of sodium bicarbonate, nor define specific treatment thresholds. Instead, the findings serve as a call for confirmatory studies utilizing exact laboratory timestamps, detailed treatment data (insulin/fluid exposure), and external validation in independent cohorts.

Technical Summary: Early Bicarbonate Normalization and ICU Discharge in Adult DKA

More like this