End-tidal carbon dioxide predicts the presence and severity of acidosis in children with diabetes

Background: Patients with diabetic ketoacidosis (DKA) hyperventilate, lowering their alveolar (PACO(2)) and arterial carbon dioxide (PaCO2). This ventilatory response lessens the severity of their acidemia in a predictable way. Because end-tidal CO2 (ETCO2) closely approximates PaCO2, measured ETCO2 levels should allow for predictions about the presence and severity of acidosis in diabetic patients. Objectives: 1) To evaluate the relationship between measured serum bicarbonate (HCO3) and ETCO2 measured via nasal capnography in children with suspected DKA; and 2) to assess the ability of capnography to predict DKA. Methods: Children being evaluated in a pediatric emergency department for suspected DKA (known or suspected diabetes presenting with hyperglycemia with or without ketonuria) were enrolled in a cross-sectional, prospective, observational study. Prior to the availability of venous HCO3 results, ETCO2 values were measured using a Nellcor NPB-70 Handheld Capnograph. Results: Forty-two patients were enrolled. Linear regression analysis revealed a significant relationship between HCO3 and ETCO2 (R-2 = 0.80, p < 0.0001). Mean ETCO2 was 37 torr (95% Cl = 35.5 to 37.9 torr) in the children without DKA and 22 torr (95% Cl = 17.4 to 26.9 torr) in the children with DKA (p < 0.0001). An ETCO2 cut-point of &LT;29 torr correctly classified the most patients (95%), with a sensitivity of 0.83 (95% Cl = 0.52 to 0.98) and a specificity of 1.0 (95% Cl = 0.88 to 1.0). No patient with an ETCO2 of ≥36 torr had DKA, for a sensitivity of 1.0 (95% Cl = 0.74 to 1.0). Conclusions: End-tidal CO2 is linearly related to HCO3 and is significantly lower in children with DKA. If confirmed by larger trials, cut-points of 29 torr and 36 torr, in conjunction with clinical assessment, may help discriminate between patients with and without DKA, respectively.