BACKGROUND AND OBJECTIVES: The evaluation of new diagnostic measurement devices allows intraindividual comparison with an established standard method. However, reports in journal articles often omit the adequate incorporation of the intraindividual design into the graphic representation. PATIENTS AND METHODS: This article illustrates the drawbacks and the possible erroneous conclusions caused by this misleading practice in terms of recent method comparison data resulting from axial length measurement in 220 consecutive patients by both applanation ultrasound and partial coherence interferometry. RESULTS: Graphic representation of such method comparison data should be based on boxplots for intraindividual differences or on Bland-Altman plots. Otherwise, severe deviations between the measurement devices could be erroneously ignored and false-positive conclusions on the concordance of the instruments could result. CONCLUSION: Graphic representation of method comparison data should sensitively incorporate the underlying study design for intraindividual comparison.
BACKGROUND AND OBJECTIVES: The evaluation of new diagnostic measurement devices allows intraindividual comparison with an established standard method. However, reports in journal articles often omit the adequate incorporation of the intraindividual design into the graphic representation. PATIENTS AND METHODS: This article illustrates the drawbacks and the possible erroneous conclusions caused by this misleading practice in terms of recent method comparison data resulting from axial length measurement in 220 consecutive patients by both applanation ultrasound and partial coherence interferometry. RESULTS: Graphic representation of such method comparison data should be based on boxplots for intraindividual differences or on Bland-Altman plots. Otherwise, severe deviations between the measurement devices could be erroneously ignored and false-positive conclusions on the concordance of the instruments could result. CONCLUSION: Graphic representation of method comparison data should sensitively incorporate the underlying study design for intraindividual comparison.