BACKGROUND: Setting specifications for analytical quality is always difficult. The risk-management approach might be a way to do so. In this approach, the definition of the required analytical quality is based on the evaluation of patient risk. Risk derives from the probability of error and from the damage that such an error might cause. METHODS: Eight Italian laboratories took part in this experiment. Measurements of glucose and total calcium were taken as examples. Analytical quality was evaluated using a specific ring trial with a frozen serum pool and by means of internal quality-control data. The total allowable error was defined according to biological variation specifications. The probability of error was extracted from the imprecision and comparative bias data of each laboratory. The damage caused by a wrong result was evaluated using the absolute probability judgment approach. RESULTS: According to the iso-risk plots (standardized hyperboles on a graph where the x-axis represents damage and the y-axis represents probability) for glucose, all the laboratories were working with an analytical quality that guaranteed low risk for patients. On the contrary, for total calcium none of the laboratories exhibited sufficient quality to guarantee low risk for patients, the presence of bias being the most relevant problem. CONCLUSIONS: The results seem to demonstrate the applicability of the risk approach to the analytical phase, indicating a new possible way to define analytical quality targets.
BACKGROUND: Setting specifications for analytical quality is always difficult. The risk-management approach might be a way to do so. In this approach, the definition of the required analytical quality is based on the evaluation of patient risk. Risk derives from the probability of error and from the damage that such an error might cause. METHODS: Eight Italian laboratories took part in this experiment. Measurements of glucose and total calcium were taken as examples. Analytical quality was evaluated using a specific ring trial with a frozen serum pool and by means of internal quality-control data. The total allowable error was defined according to biological variation specifications. The probability of error was extracted from the imprecision and comparative bias data of each laboratory. The damage caused by a wrong result was evaluated using the absolute probability judgment approach. RESULTS: According to the iso-risk plots (standardized hyperboles on a graph where the x-axis represents damage and the y-axis represents probability) for glucose, all the laboratories were working with an analytical quality that guaranteed low risk for patients. On the contrary, for total calcium none of the laboratories exhibited sufficient quality to guarantee low risk for patients, the presence of bias being the most relevant problem. CONCLUSIONS: The results seem to demonstrate the applicability of the risk approach to the analytical phase, indicating a new possible way to define analytical quality targets.
Authors: Sid E O'Bryant; Veer Gupta; Kim Henriksen; Melissa Edwards; Andreas Jeromin; Simone Lista; Chantal Bazenet; Holly Soares; Simon Lovestone; Harald Hampel; Thomas Montine; Kaj Blennow; Tatiana Foroud; Maria Carrillo; Neill Graff-Radford; Christoph Laske; Monique Breteler; Leslie Shaw; John Q Trojanowski; Nicole Schupf; Robert A Rissman; Anne M Fagan; Pankaj Oberoi; Robert Umek; Michael W Weiner; Paula Grammas; Holly Posner; Ralph Martins Journal: Alzheimers Dement Date: 2014-10-01 Impact factor: 21.566