Dear Editor,
I read with interest the article by Montero et al.1 in this month’s issue of Nefrología. This was an interesting study assessing the correlation between protein/creatinine ratio and 24-hour urine protein excretion. The authors assessed the strength of correlation by measuring the intra-class correlation coefficient (ICC) and the Spearman correlation coefficient (SCC).
I would argue that although the authors did construct a Bland Altman Plot, they did not address the agreement between protein/creatinine ratio and 24 hour urine protein excretion exhaustively. Montero et al. report the 95% limits of agreement graphically in the Bland-Altman plot but not numerically. The authors focus on the use of intra-class correlation coefficient which is often used to assess measurement error and reliability. However this measure is influenced by the amount of variation between subjects.2
When measuring urinary protein excretion, all methods are extremely likely to be correlated because they will all be attempting to measure the same construct i.e. the amount of protein in the urine, and so assessing whether the two measurements are correlated is not necessarily very informative. What we are more interested in as clinicians is whether or not we can use the two methods interchangeably or the agreement.
The standard deviation of the differences between measurements made by urine protein/creatinine ratio and 24 hour urine protein excretion provides a good index of the comparability of the two methods.2 This leads to the 95% limits of agreement between the two measurements. Although Montero et al. do report the 95% limits on the Bland-Altman plot, perhaps the authors could have reported the 95% limits of agreement for a specific number of thresholds of proteinuria for example <300mg, <3.5 grams or >3.5 grams individually which would be useful for clinicians trying to interpret spot urine protein/creatinine ratio. I worry that readers of this excellent study may be left focusing too much on correlations rather than agreement or concordance.
Conflict of interest
The authors declare that there is no conflict of interest associated with this manuscript.
