Diabetic ketoacidosis (DKA) is characterized by the triad of hyperglycemia, acidosis, and ketosis. Among blood ketones, beta-hydroxybutyrate (BOHB) predominates in DKA (acetoacetate: BOHB increase up to 1:10 against the normal 1:1), which is the basis of monitoring of blood BOHB in DKA. Current guidelines recommend periodic monitoring of ketone bodies (blood/urine) [1]. Urine ketone estimation, does not provide an accurate estimate of the ketone status as it measures acetoacetate instead of BOHB, and often requires urinary catheterization. Estimation of blood BOHB requires expensive equipment and often fails to provide real-time results. Alternatively, the point-of-care (POC) BOHB meter is a simple handheld device that can work accurately both as capillary glucose and a BOHB sensor [2,3]. The purpose of this study was to describe the strength of correlation and agreement between capillary and serum BOBH and correlation with blood gas parameters.

All children aged <13 years with DKA, as per International Society for Pediatric and Adolescent Diabetes (ISPAD) 2014 definition, were included after written informed consent from parents/legal guardian [1]. Children with suspected/known inborn errors of metabolism (IEM) or having septic shock were excluded. Glycated hemoglobins (HbA1c), venous blood gas, capillary and serum BOHB, blood glucose, serum electrolytes, and renal and liver function tests were obtained at admission and repeated 2-hourly for first first 6 hours, and 4-hourly (or more) till resolution of DKA (except hemogram and HbA1c done only at baseline). Capillary blood glucose was measured every 30-minute till resolution of DKA. 1 mL of serial blood samples for serum BOHB analysis was centrifuged, serum extracted, and stored at -80oC until final analysis. Capillary BOHB measurement was carried out using the point-of-care device (Abbott Optium-H ketone meter, Illinois, USA) after calibration with the calibration stick provided by the manufacturer. Cayman colorimetric enzymatic BOHB estimation kits were used for measurement of serum BOHB [4]. Both methods are based on the quantification of NADH generated during the enzymatic conversion of BOHB to acetoacetate. A BOHB level of >3 mmol/L was suggestive of ketonemia [1]. Analysis of serum electrolytes, renal and liver function tests were done in the biochemistry laboratory using Olympus AU 680 (Beckman Coulter, California, USA). Blood gas estimation was done using the blood gas analyzer (Cobas b 221 Blood Gas Analyzer, Roche Diagnostics, Switzerland).

The laboratory (reference) method has a coefficient of variation (COV) of 0.98, and point-of-care method has a COV of 0.96 [2,5]. With the power of 95%, and an a-error of 5%, the minimum sample required (i.e., pairs of capillary and serum) was 235, including 10% attrition for hemolysis and laboratory errors. Sample size calculation was done using n-Master version 2.0 (CMC, Vellore, India). Normality of data was checked with Kolmogorov-Smirnov Z test. Cost comparison between POC and serum measurement was made using student t test. Intra-class correlation (ICC) with 95% confidence interval (CI) and Bland-Altman plot was used to test the agreement between capillary and serum BOHB. Linear regression was used for evaluating the correlation of capillary and serum BOHB; capillary BOHB with pH, PCO2, HCO3, and AG. Two-tailed tests were used and P-value <0.05 considered as statistically significant. SPSS version 20.0 software and Epi Info™ 7 was used for data analysis. The laboratory technician was blinded to capillary BOHB values. The statistician was blinded till preparation of the first draft.

Forty-eight children with DKA were assessed for eligibility (22 excluded IEM=2, missed=14, refused to participate=6), 26 patients were enrolled. A total of 236 pairs of samples were analyzed (39 excluded, hemolysis =30, leaked=9). POC and serum BOHB (mmol/L) value less than one were 55 vs. 73, ³1 <3 was 92 vs. 76, ³3 <5 was 47 vs. 47, and ³5 was 42 vs. 40. Twenty one cases of DKA was found to have ketonemia (BOHB ³3) by POC method with excellent agreement with the reference method (Kappa value 0.752, P=<0.001, sensitivity 95.2%). Five cases of DKA were missed by POC method of which one case was diagnosed by the reference method. The baseline characteristics and laboratory parameters are described in Table I.

TABLE I Baseline Characteristics of Children with Diabetic Ketoacidosis at Enrolment (N=26)

The correlation between POC and serum BOHB showed R2 = 0.863, P=<0.001, with a beta-coefficient (slope of the regression line) of 0.929 and intercept of 0.409. The Bland-Altman analysis, as shown in Fig. 1, showed excellent agreement with a mean (SD) bias of 0.027 (0.78) and 95% limit of agreement is 1.51 to -1.56. An increase in the bias was noted in values above 5 mmol/L (P<0.001) (serum higher than POC). The intra-class correlation between POC and serum BOHB was 96.1% (95% CI 95% to 97%, P<0.001). POC-BOHB showed a moderate negative correlation with pH (r = -0.563, P<0.05) and HCO3 (r = -0.557, P<0.05), weak with pCO2 (r = -0.378, P<0.05) and moderately positive with AG (r=0.478, P<0.05). The mean (SD) total cost involved in POC-BOHB measurement per patient was significantly lower as compared to laboratory method [` 1197 (402) vs. ` 2903 (976); P<0.001]. This cost is exclusive of routine investigations, equipment charges, workforce or other miscellaneous costs. The mean (SD) time for resolution of DKA was 23.5 (13.2) hours.

Fig. 1 Bland-Altman plot between capillary and serum b-hydroxy butyrate (BOHB) levels.

In this study, we documented the excellent correlation and agreement between POC and serum BOHB measurement with an increase in the bias of value above five mmol/L. This can be due to insufficient quantity of reagent on the electrochemical strip and non-linearity of the ampero-metric detector in the device, and the absolute quantity of acetoacetate may inhibit the enzyme in the test strip or cause inhibition of the quinoid NADH redox mediator incorporated into the electrode [3]. Studies are reported that excellent agreement of POC measurement with standard measurement up to the value of 4 mmol/L [6] and 6 mmol/L [3]. A BOHB value above 1.5 mmol/L indicates that "at risk" for DKA and value ³3 mmol/L is a diagnosis of DKA [1,6]. A BOHB value less than 1 mmol/L is one of the endpoints of DKA treatment. The clinically relevant range of BOHB value is <1-4 mmol/L [7]. Hence, POC device can be used as a reliable bedside method of ketone (BOHB) estimation, provided that values above 5 mmol/L be viewed with caution. In children, obtaining a urine sample is impractical for ketone estimation, POC-BOHB has a unique value and also can be utilized even in a primary healthcare set-up, as no new equipment is required (the glucometer can be recalibrated as a ketone meter with the help of a calibration stick).