Clinical investigation of a point-of-care blood ammonia analyzer

Goggs, R A and Serrano, S and Szladovits, B and Keir, I and Ong, R and Hughes, D (2008) Clinical investigation of a point-of-care blood ammonia analyzer. VETERINARY CLINICAL PATHOLOGY, 37 (2). pp. 198-206.

Full text not available from this repository.

Abstract

Background: Hyperammonemia has frequently been implicated in the pathogenesis of hepatic encephalopathy. Blood ammonia determination requires minimal delay between sampling and analysis for accurate results. Objectives: The aim of this study was to investigate the PocketChem BA, a new point-of-care (POC) blood ammonia analyzer for clinical use by determining machine precision, linearity, repeatability, and accuracy. Methods: Coefficients of variation were determined by repeated measurement of 2 control solutions. Linearity was investigated by testing serial dilutions of a stock solution. For accuracy, samples from clinical cases were used to compare the results on the PocketChem BA with those obtained using an enzymatic reference method for canine plasma. Canine and feline patients were consecutively enrolled if blood ammonia was assayed and samples could be analyzed shortly after collection. Classification of results (as normal or high, using 100 mu mol/L as a cutoff value), Bland-Altman and Deming regression plots, and intraclass correlation coefficients were used to compare the methods. Stability of samples and test strips also was assessed over time. Results: Coefficients of variation were 10.6% and 4.8% for low and high controls, respectively. Concentrations of ammonia in diluted stock solutions correlated positively with mean measured concentrations (Pearson coefficient 0.988, P <.001). Of the 54 samples obtained from 38 dogs and 4 cats, 41 had ammonia concentrations within the readable range. Results from the POC analyzer and the reference method were correlated positively (intraclass coefficient 0.800, 95% confidence interval 0.655-0.888), with the POC analyzer having negative constant and proportional biases. The methods agreed in the classification of 45/54 (83.3%) samples, with 7 false negative results on the POC analyzer. Results of repeated sample and strip analyses at 1 and 24 hours were significantly different (P <.05) from those at 0 hour. Conclusions: The PocketChem BA has acceptable precision, adequate linearity, and satisfactory agreement with a reference method, but negative constant and proportional biases. The POC analyzer may be suitable for clinical use in patients suspected of having hepatic encephalopathy, using a lower reference limit of 60 mu mol/L to decrease false negative results.