I am working on a review of membranous nephropathy and I found this quotation from UpToDate:
A random urine protein-to-creatinine ratio should NOT be used for initial risk stratification, since the relationship between the ratio and 24-hour protein excretion varies widely among patients (show figure 2).
This surprised me. I use the protein-to-creatinine ratio all the time and, though I have found some individuals where it is wildly inaccurate, I had been unaware of any data that showed that to be the case.
I eagerly clicked the link to the figure and this is what I found:
That looks highly accurate to me. Every data point clusters right along the line of identity. I pulled the abstract (alas, no full text from NEJM 1983) and found this conclusions from the authors:
In a study of 46 specimens we found an excellent correlation between the protein content of a 24-hour urine collection and the protein/creatinine ratio in a single urine sample.
I love UpToDate but this is really disappointing. Making a claim and referencing it with data which disproves the claim is disingenuous.
UpToDate, you need to do better.