Unmasking delayed recovery in severe diabetic ketoacidosis: clinical and biochemical clues from early resuscitation

Article information

Ann Pediatr Endocrinol Metab. 2026;31(4):230-237

Publication date (electronic) : 2026 August 31

doi : https://doi.org/10.6065/apem.2550324.162

Jaeeun Chung, Bina Kim, Chong Kun Cheon, Jeong-Eun Kang, Younga Kim, Sukdong Yoo

Department of Pediatrics, Pusan National University Yangsan Hospital, Pusan National University School of Medicine, Research Institute for Convergence of Biomedical Science and Technology, Yangsan, Korea

Address for correspondence: Sukdong Yoo Department of Pediatrics, Pusan National University Yangsan Hospital, Pusan National University School of Medicine, Research Institute for Convergence of Biomedical Science and Technology, 20 Geumo-ro, Mulgeum-eup, Yangsan 50612, Korea Email: special4yoo@naver.com

Received 2025 August 29; Revised 2025 November 17; Accepted 2025 December 3.

Abstract

Highlights

· Elevated blood urea nitrogen and creatinine at presentation were associated with delayed pH recovery in children with severe diabetic ketoacidosis (DKA).

· Early changes in pH and heart rate during the first hour of resuscitation may help identify patients with delayed metabolic recovery.

· Accurate assessment of fluid deficits and timely adjustment of fluid therapy are essential for optimizing early DKA management.

Introduction

Diabetic ketoacidosis (DKA) is a critical complication in pediatric diabetes mellitus, caused by insufficient circulating insulin and increased levels of counterregulatory hormones [1]. It is characterized by hyperglycemia, ketosis, and metabolic acidosis [2]. Rapid diagnosis and correction of acidosis and dehydration are crucial components of DKA management. Standard treatment follows protocol-driven fluid resuscitation, gradual correction of hyperglycemia with intravenous insulin, and meticulous electrolyte management, along with ongoing neurologic monitoring to detect cerebral edema early. These protocol-driven therapeutic strategies have significantly improved clinical outcomes [3].

Severe DKA is typically included in the intensive care unit (ICU) admission criteria recommended by several guidelines [4,5] because these patients require continuous monitoring of vital signs, frequent blood tests, largevolume fluid administration, and surveillance for decreased mental status and life-threatening complications.

However, DKA is not exclusively managed in the ICU; it is often treated in the emergency department or general wards due to limited ICU capacity, its relatively lower mortality compared to other ICU disorders, cost-effectiveness, and the increasing incidence of DKA [6-8]. While most children show gradual clinical improvement and resolution of acidosis within the first several hours of therapy, the rate of pH recovery varies, and a subset fails to improve as rapidly. These “delayed recovery” cases— characterized by persisting acidosis beyond the typical timeframe—raise concerns about the adequacy of initial management and the appropriateness of triage decisions.

Identifying the causes of delayed recovery and early predictive factors is essential for optimizing the management of severe DKA. However, there has been a lack of research specifically addressing the early indicators and underlying mechanisms contributing to prolonged recovery in pediatric patients. Here we investigate initial clinical characteristics and the treatment course of patients with severe DKA based on early pH trends and discuss our findings in the context of current evidence and clinical practice.

Materials and methods

Results

1. Baseline characteristics and initial laboratory results

A total of 69 patients were identified with severe DKA, of whom 19 were excluded: 12 due to initial treatment at another facility, 6 due to incomplete records of infusion rates or nonadherence to the DKA protocol, and 1 due to death. In the end, 50 patients were enrolled: 39 in the ERG and 11 in the DRG. The mean age of all patients was 12.3 years, with a sex composition of 19 boys and 31 girls. At baseline, the DRG showed a higher average age (ERG: 11.16±4.94 years, DRG: 14.18±3.50 years), exhibited greater median weight loss (ERG: 8.7%, DRG: 10.8%), experienced more frequent altered mental status (ERG: 28.2% [11 of 39], DRG 45.5% [5 of 11]), and had a higher incidence of hyperosmolar DKA (ERG: 10.3% [4 of 39], DRG: 36.4% [4 of 11]). However, these differences did not reach significance. The initial vital signs, including systolic and diastolic blood pressure and heart rate percentiles, showed no differences between the groups (Table 1).

Table 1.

Baseline characteristics of the early and delayed recovery groups

The initial laboratory results indicated that the DRG had lower pH and HCO₃ levels, along with slightly higher white blood cell and glucose levels, although these differences were not statistically significant. Furthermore, the DRG showed significantly higher levels of blood urea nitrogen (BUN) (ERG: 16.60 [12.90–19.90] mg/dL, DRG: 22.10 [18.65–29.50] mg/dL; P=0.01), creatinine (ERG: 0.91 [0.80–1.12] mg/dL, DRG: 1.20 [0.97–1.56]; P=0.02), and C-reactive protein (CRP) levels (ERG: 0.11 [0.04–0.23] mg/dL, DRG: 0.60 [0.23–3.93]; P=0.005). The proportion of patients with ketonuria of 3+ or higher was significantly greater in the ERG than in the DRG (ERG: 38 (97.4%), DRG: 8 (72.7%); P=0.029) (Table 2).

Table 2.

Initial laboratory results of the early and delayed recovery groups

Fig. 1 shows changes in pH and heart rate percentile over time. Baseline pH was not significantly different between the ERG and DRG at the initial presentation (ERG: 7.02 [6.95–7.07], DRG: 6.98 [6.90–6.99]; P=0.096), but the difference in pH between the groups became more pronounced at the 1-hour mark (ERG: 7.02±0.10, DRG: 6.94±0.10; P=0.033) after fluid resuscitation. The pH between the 2 groups was consistently different at the 6-hour (ERG: 7.15±0.09, DRG: 6.96±0.13; P<0.001) and 12-hour marks (ERG: 7.25±0.06, DRG: 7.20±0.08; P=0.026). However, by the 24-hour mark (ERG: 7.31±0.05, DRG: 7.30±0.04; P=0.432), both groups had recovered to a similar level of pH (Fig. 1A).

Fig. 1.

Changes in pH (A) and heart rate percentile (B) over time are shown for the delayed recovery group (dark gray) and the early recovery group (light gray). Dashed and solid lines represent mean value trends for each group.

Baseline heart rate percentiles were similar between groups (ERG vs. DRG, 95.47 [70.77–98.81] vs. 91.11 [66.57– 96.18], P=0.292), but the difference in heart rate percentiles between the 2 groups began to emerge from the 1-hour mark (ERG: 83.95±19.20, DRG: 92.50±9.36, P=0.048), and an increase in heart rate was recorded in the DRG up to the 6-hour mark (80.68±23.32 vs. 94.65±5.44, P=0.001). Notably, the heart rate percentile in the DRG remained higher than that in the ERG even at 24 hours (54.38±22.93 vs. 70.72±14.30, P=0.03). To compare the graph with the pH change, the y-axis values were expressed as the "100 − heart rate percentile." It was confirmed that the trends observed in the pH change graph were consistent with those in the 100 − heart rate percentile graph, indicating an inverse relationship between pH and heart rate (Fig. 1B).

In the linear regression analysis of the relationship between blood pH and heart rate, no significant association was observed between pH and heart rate at baseline. However, from 1 hour onward, higher pH levels were significantly associated with lower heart rates, with the strongest inverse correlation observed at 24 hours, as indicated by the regression coefficient (β=-175.0, P=0.003). The regression equations and P-values for each time point are summarized in Supplementary Fig. 1.

2. Potential risk factors for delayed recovery of DKA

Age-adjusted multivariable regression analysis was performed to investigate the relationship between various previously assessed laboratory results and the risk of delayed recovery. The analysis revealed that delayed recovery of pH was significantly associated with increased initial BUN (odds ratio [OR], 1.24; 95% confidence interval [CI], 1.01–1.51), increased heart rate percentile for the first hour of treatment (OR, 1.18; 95% CI, 1.01–1.37) and a 0.01 decrease in pH during the first hour of treatment (OR, 0.83; 95% CI, 0.69–0.99). CRP did not show a significant association (Fig. 2).

Fig. 2.

Odds ratio (OR) of delayed pH recovery from multivariable logistic regression adjusted for age. C-reactive protein (CRP) and blood urea nitrogen (BUN) are baseline laboratory values. ΔHeart rate percentile and ΔpH (change of 0.01) indicate changes during the first hour. CI, confidence interval.

3. Treatments and outcomes between the 2 groups

The fluid infusion rate (mL/kg/hr) RI rate (IU/kg/hr), and urine output (ml/kg/hr) at 6 hours, 12 hours, and 24 hours were similar between the groups. The administered amount of dextrose during the first 6 hours was lower in the DRG (ERG: 0.12 [0.04–0.19] mg/kg/hr, DRG: 0.08 [0.00–0.10] mg/kg/hr; P=0.025), but became similar over 24 hours. The median recovery times to reach the target pH (ERG: 23.42 [18.79–29.63] hours vs. DRG: 25.67 [19.95–29.36] hours; P=0.405) and target bicarbonate levels (ERG: 27.82 [19.38–36.98] hours vs. DRG: 26.55 [18.69–46.33] hours; P=0.779) were similar between the 2 groups. Clinical outcomes including AKI, ICU admission, ICU stay, and total hospital stay did not significantly differ between the groups (Table 3). AKI occurred in 38.4% of the ERG and 63.6% of the DRG (P=0.178), while ICU admission was noted in 30.8% and 45.5% respectively (P=0.285). The median ICU stay (ERG: 2 [2–3] days vs. DRG: 2.5 [2.0–3.0] days; P=0.548) and hospital stay (ERG: 9 [7–11] days vs. DRG: 10 [7–12] days; P=0.732) were also similar between groups, suggesting that despite differences in early metabolic response, short-term clinical outcomes were largely comparable. Pancreatitis occurred in 1 patient in each group. Elevated CK was assessed as a potential complication but was not included in the final comparative analysis.

Table 3.

Treatment details, outcomes, and accompanying complications

Discussion

As the treatment of DKA is well protocolized in recent guidelines [4], the final pH recovery time and clinical outcomes in our study were similar between the ERG and DRG, indicating that the protocol-based DKA treatment was effective for most patients. However, in clinical practice, pH recovery varies, as seen in delayed pH levels at 6 and 12 hours in the DRG. By comparing the clinical characteristics at diagnosis with the specifics of the treatment provided to patients, this study identified key factors that clinicians should prioritize and take into account to optimize DKA management.

Relatively higher hyperglycemia and lower dextrose administration during the first 6 hours in the DRG suggest delayed adjustment to dextrose-containing fluids due to slower attainment of the plasma glucose level of 250~300 mg/dL. This delay may indicate a greater metabolic burden and increased insulin requirements, ultimately leading to prolonged recovery and greater treatment demands. The higher levels of BUN, creatinine, CRP, and low urinary ketone levels at initial presentation explain the underlying causes of greater metabolic stress in the DRG. Prolonged insulin deficiency induces lipolysis and proteolysis, contributing to the accumulation of organic acids in addition to ketones. Severe dehydration and impaired kidney function, as indicated, may have reduced the clearance of nonketotic organic acids, ultimately exacerbating acidosis and potentially explaining relatively lower ketone levels despite the low pH observed in the DRG [10,11]. These findings also raise the possibility of mixed DKA and hyperosmolar hyperglycemic state (HHS) presentations in some patients within the DRG. While our study focused on severe DKA, certain DRG patients exhibited relatively low ketone levels, elevated osmolality, and signs of AKI, which are features often seen in HHS. Although not formally diagnosed in this study, such overlapping pathophysiology could contribute to delayed acidosis recovery and increased treatment resistance. Future studies may benefit from more explicit evaluation of mixed DKA-HHS presentations in adolescents. Elevated CRP levels suggest underlying inflammation or infection, potentially stimulating catecholamine and cortisol production and thereby exacerbating metabolic derangements. These findings are consistent with previous studies [12-15] that have also identified BUN, creatinine, and CRP as biochemical markers associated with DKA severity and prognosis. However, this study did not evaluate the presence or absence of infection in individual patients; therefore, it cannot be determined whether the observed CRP difference was related to infection. Further investigation is needed to clarify this association.

As the treatment was based on ISPAD guidelines [4], both groups received similar initial fluid volumes. At presentation, BUN and creatinine levels were elevated in the DRG, suggesting significant dehydration. Although more weight gain would have been expected if fluid replacement had been adequate, median weight loss was greater in the DRG, while post-treatment weight gain was lower than in the ERG. These findings suggest that the initial fluid deficit may have been underestimated, resulting in insufficient replacement and delayed recovery of kidney function and pH during the first 6 hours. This highlights the importance of accurately assessing fluid deficits and the need for early reassessment of fluid supplementation based on treatment response. However, determining the exact fluid deficit in DKA remains challenging, and there are significant concerns about the potential adverse effects of fluid overcorrection in clinical practice [16,17].

Our findings on pH and heart rate changes at 1 hour provide simple guidance for patient reassessment and management decisions. Normal saline, most commonly used as an initial resuscitation fluid, has a mildly acidic pH of approximately 5.5 and contains a high concentration of strong anions, chloride [18,19]. Patients with DKA initially present with high anion gap metabolic acidosis due to the accumulation of ketone bodies. Typically, with fluid resuscitation using normal saline, improved perfusion and renal function promote ketone clearance, leading to gradual resolution of acidosis and an increase in pH. However, in cases of impaired renal function or persistent hypoperfusion, the bicarbonate buffer system becomes diluted, ketone and chloride clearance may be insufficient, potentially resulting in further decline in pH [19]. Thus, the degree and direction of pH change following normal saline administration reflect the underlying physiological status and the effectiveness of initial fluid resuscitation.

Patients with DKA typically exhibit tachycardia due to severe volume depletion, and cardiac function can be affected by systemic stress, catecholamine release, acidosis, elevated levels of circulating fatty acids, and electrolytic alterations [20]. As fluid is replenished and dehydration is corrected, the heart rate begins to stabilize, reflecting progress in treatment. It is well established that continuous heart rate monitoring is useful for detecting acute cardiovascular complications [21], potassium status, and signals of cerebral edema [4]. The inverse correlation between heart rate percentiles and pH levels in our study suggests that heart rate monitoring could also serve as an effective indicator of metabolic correction of acidosis. The early identification of persistent tachycardia may indicate the need for more aggressive rehydration, particularly in patients with recent weight loss exceeding 10%.

Our study delineates the early clinical course of severe DKA and highlights the association between elevated BUN, creatinine, and CRP levels and delayed recovery. It underscores the importance of accurate fluid deficit assessment and the need for appropriate fluid replacement strategies in patients with such biochemical profiles. Furthermore, our findings suggest that monitoring changes in pH and heart rate during the first hour of normal saline resuscitation can provide valuable insight into treatment response and guide timely reassessment of fluid management. Heart rate, in particular, may serve as a simple and effective marker for monitoring metabolic acidosis recovery in the acute phase of DKA.

The limitations of this study include the small sample size due to its single-center design, as well as the exclusion of severely ill patients who required renal replacement therapy, had brain injuries, or died. The retrospective nature of the study also introduces potential biases, as treatment details such as insulin, dextrose, and fluid rates were extracted from medical records. Changes in pH were not analyzed in conjunction with corresponding alterations in HCO₃^-, pCO₂, or shifts in nonketotic anions such as chloride.

Despite its limitations, this study provides practical insights into early physiologic markers that may help optimize fluid management in severe pediatric DKA. Future multicenter studies are needed to validate these findings and further explore dynamic acid-base and hemodynamic responses during resuscitation. Identifying early recovery markers could enable more individualized and effective treatment, improving clinical outcomes in this high-risk population.

Exploring the impact of fluid management on pH recovery in more severe DKA cases, particularly those with complications that were excluded from this study, would be valuable. Such research could provide a deeper understanding of how treatment strategies affect recovery in high-risk patients.

Taken together, our findings clarify the natural course of acute severe DKA and emphasize the need for precise assessment of fluid deficits. Particularly in patients with elevated BUN, creatinine, and CRP levels, timely and appropriate fluid management is crucial to optimize outcomes.

Article information Continued