Introduction
Diabetes mellitus (DM) is a chronic metabolic disorder marked by hyperglycemia due to defects in insulin secretion, action, or both [1]. Type 1 DM (T1DM), the predominant form in children, arises from autoimmune destruction of pancreatic beta cells, resulting in absolute insulin deficiency [1][2]. Diabetic ketoacidosis (DKA) is its most serious acute complication, defined by hyperglycemia (>200 mg/dL), venous pH <7.3 or bicarbonate <18 mmol/L, and ketonemia/ketonuria [3]. DKA causes significant morbidityincluding electrolyte imbalances, acute kidney injury (AKI), cerebral edemaand mortality, especially in resource-limited settings [4][5].
Globally, ~1.52 million children/adolescents <20 years had T1DM in 2022, with DKA as the initial presentation in 15–70% of cases, varying by awareness and healthcare access [6]. In India, pediatric DKA data are sparse, with high rates of new-onset presentation (up to 68%) linked to diagnostic delays, non-specific symptoms in young children (polyuria/polydipsia often overlooked), and infections as triggers [7][8]. Recurrent DKA, driven by poor compliance (especially in adolescents) and suboptimal glycemic control (HbA1c often >9%), carries added risks of cumulative complications [9].
Management centers on cautious fluid resuscitation, low-dose insulin infusion (0.05–0.1 U/kg/h), and serial electrolyte monitoring, as insulin therapy precipitates hypokalemia and chloride-rich fluids may induce hyperchloremic acidosis [3][10]. Admission venous lactate, reflecting tissue hypoperfusion/metabolic stress, may predict insulin duration/acidosis resolution, yet evidence is conflicting [11][12]. This study compared clinical/biochemical profiles, electrolyte patterns (particularly hypokalemia during insulin infusion and potassium replacement needs), lactate–insulin duration association, and outcomes between new-onset and recurrent pediatric DKA in an Indian tertiary PICU, addressing local evidence gaps to inform targeted monitoring and prevention [13].
Materials and Methods
This was a prospective and retrospective observational study conducted at the Pediatric Intensive Care Unit (PICU) of PSG Institute of Medical Sciences and Research, a tertiary care center in Coimbatore, India. The study period was January 2020 to December 2024. Ethical approval was obtained from the Institutional Human Ethics Committee (reference: [insert if available]), and informed consent/assent was secured where applicable.
Study Population and Criteria
Children aged 1-18 years admitted with DKA (defined as blood glucose >200 mg/dL, venous pH <7.3 or bicarbonate <18 mmol/L, and ketonemia/ketonuria [2]) were included. New-onset DKA was defined as the initial T1DM diagnosis during the episode; recurrent as DKA in known T1DM cases. Exclusion criteria: Children receiving initial insulin/fluids elsewhere before referral.
Sample size was calculated based on a 2.2% mortality prevalence from prior studies [8], with 95% confidence, 20% precision, and 10% alpha error, yielding n=35. However, all eligible cases (n=87) were included for robustness.
Data Collection and Management
Data were extracted from medical records (retrospective: 2020-2022; prospective: 2023-2024) using a proforma. Variables included demographics, family history, precipitating factors, anthropometrics, vital signs, DKA severity (mild: pH<7.3/bicarbonate<18; moderate: pH<7.2/bicarbonate<10; severe: pH<7.1/bicarbonate<5 [2]), blood gases, electrolytes (sodium, potassium, chloride, ionized calcium) at admission and intervals (12, 24, >30 hours), lactate, HbA1c, lipid profile, thyroid function, acute kidney injury (AKI; KDIGO criteria [9]), insulin regimen, infusion duration, potassium replacement (IV/oral if K<3.5 mmol/L), hospital stay, and outcomes. Management followed ISPAD guidelines [2]: fluids (0.9% saline/Plasmalyte), insulin (0.05-0.1 U/kg/h), and monitoring.
Statistical Analysis
Data were analyzed using SPSS v25. Categorical variables were presented as frequencies/percentages and compared via chi-square tests. Continuous variables as means±SD, compared by t-tests. Pearson's correlation assessed lactate-insulin duration association. p<0.05 was significant.
Results
Eighty-seven children were included: 41 (47.1%) new-onset DKA, 46 (52.9%) recurrent. Mean age was 8.85±3.2 years (new-onset) vs. 12.05±2.8 years (recurrent; p<0.001). Females predominated (54%); 56% were >10 years. Family history of T1DM was low (6.9%). Precipitating factors: infections (10.3%; LRTI more in new-onset [4.6%], UTI in recurrent [5.7%]; p=0.04), poor compliance (recurrent only, 30.4%). Anthropometrics: obesity 4.6%, thinness 13.8%, short stature 5.8% (no group difference).
Clinical presentation: Decompensated shock 1.1%, GCS≤8 2.3%. Severe DKA was common (63.2%; Table 1). Leukocytosis: 51.2% new-onset (14.3% with respiratory focus), 60.9% recurrent (24% with culture-proven infection). Oxygen required: 33.3%; mechanical ventilation: 1.1%.
Electrolyte disturbances evolved over time (Table 2). Hypokalemia increased from 18.9% at admission to 56.8% >30 hours, with no group difference. Potassium replacement needed in 76% (IV 45%, oral 31%; similar across groups). Hyperchloremia rose to 48.5% at 24 hours.
Biochemical comparisons: Recurrent DKA had higher dyslipidemia (high cholesterol 17.4% vs. 9.8%, low HDL 52.2% vs. 39%, high LDL 54.3% vs. 36.6%; p<0.05; Table 3). AKI: pre-renal 14.9% overall (similar), renal 9.2% (17.4% recurrent vs. 0% new-onset; p=0.005). Hypothyroidism: 2.7% (normal TSH more in recurrent; p=0.04). Proteinuria: 48.3% (no difference). HbA1c in recurrent: 97.8% >7.5% (poor control).
Admission lactate correlated positively with insulin infusion duration (r=0.72, p<0.001; Table 4). Mean hospital stay: 5.62±2.1 days (longer in new-onset; p=0.03). Mortality: 1.1% (one recurrent case, due to severe AKI). At discharge, insulin requirement: 1.01-1.5 U/kg/day (majority both groups).
| Parameter | New-Onset (n=41) | Recurrent (n=46) | p-value |
| Age (mean±SD, years) | 8.85±3.2 | 12.05±2.8 | <0.001 |
| Female (%) | 53.7 | 54.3 | 0.95 |
| Severe DKA (%) | 58.5 | 67.4 | 0.38 |
| Precipitating infection (%) | 9.8 (LRTI dominant) | 10.9 (UTI dominant) | 0.04 |
| Leukocytosis (%) | 51.2 | 60.9 | 0.35 |
| Shock (%) | 0 | 2.2 | 0.34 |
| Time Point | Hyponatremia (%) | Hypernatremia (%) | Hypokalemia (%) | Hyperkalemia (%) | Hyperchloremia (%) | Hypocalcemia (%) |
| Admission | 35.7 | 5.7 | 18.9 | 36.1 | 31.2 | 6.9 |
| 12 hours | 5.0 | 7.6 | 26.7 | 13.6 | 38.9 | 4.8 |
| 24 hours | 8.9 | 11.0 | 49.8 | 2.1 | 48.5 | 2.3 |
| >30 hours | - | - | 56.8 | - | 43.9 | - |
| Parameter | New-Onset (n=41) | Recurrent (n=46) | p-value |
| High cholesterol (%) | 9.8 | 17.4 | 0.04 |
| Low HDL (%) | 39.0 | 52.2 | 0.03 |
| High LDL (%) | 36.6 | 54.3 | 0.02 |
| Renal AKI (%) | 0 | 17.4 | 0.005 |
| Proteinuria (%) | 56.1 | 41.3 | 0.17 |
| Hypothyroidism (%) | 2.4 | 3.0 | 0.04* |
| *For normal TSH distribution. |
| Parameter | Value | p-value |
| Lactate vs. insulin duration (r) | 0.72 | <0.001 |
| Potassium replacement (%) | 76 (IV:45, oral:31) | - |
| Hospital stay (mean±SD, days) | New:6.1±2.3; Recurrent:5.2±1.8 | 0.03 |
| Mortality (%) | 1.1 | - |
Discussion
This five-year observational study of 87 children with DKA (47% new-onset, 53% recurrent) at a South Indian tertiary PICU revealed distinct patterns supporting heightened vigilance in recurrent cases and standardized electrolyte management.
Demographics aligned with global trends: female preponderance (54%), peak incidence >10 years (56%), and older mean age in recurrent DKA (12.05 vs 8.85 years; p<0.001), consistent with pubertal insulin resistance and adherence challenges [14][15]. Severe DKA predominated (63%), comparable to Saudi (61.6%) and Pakistani series (62.5%), reflecting delayed recognition in younger children with atypical symptoms and infection triggers (LRTI in new-onset, UTI in recurrent; p=0.04) [7][16][17]. Low family history (6.9%) contrasts higher rates in some Asian cohorts (14–70%), possibly due to under-reporting or genetic differences [18].
Electrolyte disturbances evolved dynamically. Hypokalemia rose progressively (19% admission → 27% at 12 h → 50% at 24 h → 57% >30 h), driven by insulin-mediated intracellular shifts, osmotic diuresis, and correction of acidosis, necessitating replacement in 76% (IV early, oral later; no inter-group difference) [10][19]. This pattern underscores ISPAD recommendations for immediate potassium supplementation (40 mmol/L fluids if K<5.5 mmol/L) and hourly monitoring [3]. Hyperchloremia peaked at 49% (24 h), likely from 0.9% saline use, contributing to persistent normal-anion-gap acidosis despite ketone clearance; balanced crystalloids (e.g., Plasmalyte, used in 11%) may mitigate this [20][21]. Compared to Hamza et al. (Chennai PICU), our hypokalemia was higher post-24 h while hypocalcemia was markedly lower (ionized Ca focus), reflecting protocol evolution and ionized vs total Ca measurement [22].
Admission lactate strongly correlated with insulin infusion duration (r=0.72, p<0.001), suggesting utility as a severity/perfusion marker for risk stratification and anticipated PICU stay, differing from studies showing no association [11][12]. Recurrent DKA exhibited poorer control (98% HbA1c >7.5%), higher renal AKI (17% vs 0%; p=0.005), and dyslipidemia (high cholesterol/LDL, low HDL; p<0.05), linked to chronic hyperglycemia-induced lipolysis and inflammation [23][24]. Hypothyroidism prevalence (2.7%) matched rural Indian reports [25].
Outcomes were favorable: mean stay 5.6 days (longer in new-onset, p=0.03), mortality 1.1% (one recurrent AKI-related death), cerebral edema 1.1%. These low rates versus 2–15% in developing-country series reflect ISPAD-guided care, early PICU admission, and low ventilation needs (1.1%) [7][16][26]. Findings emphasize infection screening, adolescent compliance support, routine lactate/electrolyte trends monitoring, preference for balanced fluids, and public/physician education to reduce new-onset DKA and recurrence [3][8][27][31].
Conclusion
In this Indian tertiary PICU cohort, pediatric DKA showed high severity (63%) with nearly equal new-onset and recurrent cases. Recurrent episodes were associated with older age, poorer glycemic control, higher renal AKI, and dyslipidemia. Hypokalemia progressed markedly during insulin infusion (up to 57% >30 h), requiring replacement in 76%, while hyperchloremia highlighted risks of chloride-rich fluids. Admission lactate strongly predicted insulin duration (r=0.72), offering a simple risk-stratification tool. Low mortality (1.1%) and favorable outcomes underscore benefits of ISPAD protocols. Enhanced awareness, infection management, balanced fluids, vigilant potassium monitoring/replacement, and adolescent support are critical to prevent new-onset DKA and reduce recurrence/complications. Multicentric studies are warranted.
Declarations
Acknowledgements
We thank PICU staff and Institutional Ethics Committee for support.
Conflict of Interest
None declared.
Funding
None.
Ethics
Approved by Institutional Human Ethics Committee, PSG Institute of Medical Sciences and Research; informed consent/assent obtained; complied with Declaration of Helsinki.