Detection of bias in method comparison by regression analysis

Hartmann, C.; Smeyers‐Verbeke, J.; Penninckx, W.; Massart, D.L.

doi:10.1016/s0003-2670(96)00341-8

Cited by 25 publications

(10 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The target values of the QC material of different analytes, as well as the measurement units, the number of QC repetitions (n), the mean, SD and CV of the QC materials are displayed. The beta error was estimated [ 12 ] using the range ratio and the allowable biases as defined by the federal registry, the Qualab or the clinical estimate. The values used for the evaluations are displayed in bold.…”

Section: Methodsmentioning

confidence: 99%

Effects of different centrifugation conditions on clinical chemistry and Immunology test results

et al. 2011

View full text Add to dashboard Cite

BackgroundThe effect of centrifugation time of heparinized blood samples on clinical chemistry and immunology results has rarely been studied. WHO guideline proposed a 15 min centrifugation time without citing any scientific publications. The centrifugation time has a considerable impact on the turn-around-time.MethodsWe investigated 74 parameters in samples from 44 patients on a Roche Cobas 6000 system, to see whether there was a statistical significant difference in the test results among specimens centrifuged at 2180 g for 15 min, at 2180 g for 10 min or at 1870 g for 7 min, respectively. Two tubes with different plasma separators (both Greiner Bio-One) were used for each centrifugation condition. Statistical comparisons were made by Deming fit.ResultsTubes with different separators showed identical results in all parameters. Likewise, excellent correlations were found among tubes to which different centrifugation conditions were applied. Fifty percent of the slopes lay between 0.99 and 1.01. Only 3.6 percent of the statistical tests results fell outside the significance level of p < 0.05, which was less than the expected 5%. This suggests that the outliers are the result of random variation and the large number of statistical tests performed. Further, we found that our data are sufficient not to miss a biased test (beta error) with a probability of 0.10 to 0.05 in most parameters.ConclusionA centrifugation time of either 7 or 10 min provided identical test results compared to the time of 15 min as proposed by WHO under the conditions used in our study.

show abstract

Section: Methodsmentioning

confidence: 99%

Effects of different centrifugation conditions on clinical chemistry and Immunology test results

et al. 2011

View full text Add to dashboard Cite

show abstract

“…The confidence limits were calculated as follows: [slope or intercept] ± [standard deviation × t (α = 0.05, n = 2 d.f.)] [14]. Fig.…”

Section: Comparison Of Desacetylcefotaxime Concentrations Analysed Asmentioning

confidence: 99%

Liquid chromatographic determination of the plasma concentrations of cefotaxime and desacetylcefotaxime in plasma of critically ill patients

Rosseel

Vandewoude

2004

Journal of Chromatography B

View full text Add to dashboard Cite

“…Since it is usually resource prohibitive for a single laboratory to perform an inter-laboratory study for a new screening method, the study here presents a valuable and practical approach to evaluate the reliability of a [17,18]. Particularly, regression models have been used to evaluate the bias in method comparisons and differences have been observed with various regression models [19][20][21][22]. Besides the regression models, the current model is the first Bayesian model that takes into account of measurement uncertainty of the reference method and estimates the uncertainty budget of the new screening method under evaluation.…”

Section: Introductionmentioning

confidence: 99%

Cost of speed: A practical approach to evaluate a screening method from a Bayesian perspective

Dai

Sang

Lee

et al. 2016

Chemometrics and Intelligent Laboratory Systems

View full text Add to dashboard Cite

Utilization of rapid screening method presents an investment decision of laboratory assets. Screening methods are filtering procedures implemented to enable analysis of a large volume of samples in a rapid fashion. The principle of screening includes establishment of a threshold value as a selection criteria for the follow up confirmative analysis. Selection of a quality screening method determines the optimized utilization of laboratory resources. Two mathematical models are used to evaluate a screening method by directly comparing to a gold standard reference method. The first model utilizes the traditional receiver operating characteristic (ROC) curve model to evaluate the screening method. A new model is proposed to access the accuracy and reliability of the screening method based on the gold standard reference method. The two models are extremely practical for a single laboratory to evaluate the reliability of a screening method when limited resources are available.

show abstract

Detection of bias in method comparison by regression analysis

Cited by 25 publications

References 7 publications

Effects of different centrifugation conditions on clinical chemistry and Immunology test results

Effects of different centrifugation conditions on clinical chemistry and Immunology test results

Liquid chromatographic determination of the plasma concentrations of cefotaxime and desacetylcefotaxime in plasma of critically ill patients

Cost of speed: A practical approach to evaluate a screening method from a Bayesian perspective

Contact Info

Product

Resources

About