We had done one study to see whether salivary creatinine levels can be used to screen for renal diseases, where we compared with gold standard Serum creatinine levels. So to test the relationship between two variables one reviewer had suggested Bland altman plots, but the problem is creatnine levels are i.e. Salivary creatinine levels, are in the range of 0.6 and serum in range of 6.3.
So what is ideal in this scenario other than correlation co-efficient to show the relationship ?
If I have understood the usage of the Bland-Altman plot correctly, than it is thought to compare two measurement methods measuring the same underlying true variable. In your case, however, levels DO differ, if I understood you right (my background is in psychology, but we have recently used the Balnd-Altman plot approach, too). Thus you actually measure two different variables, because creatinine levels differ in salivary and serum. Thus, the crucial question is: Can you assume that both of your methods shall measure one underlying true value?
We are measuring Creatinine level, as you said it is different in Saliva and Serum so I think it is not measuring the true variable.
Why don't you want to do the Bland-Altman plot? This is the difference of two scores divided by their mean, so they don't need to be in the same unit. And I suppose that if you have a rationale for using salivary creatinine levels to detect renal diseases it is highly correlated to the serum level, which is actualy used.
For me this is a good idea from the reviewer.
It is not that I don't want to use it. The values of the serum creatinine level ranged between 2.0- 13.8mg/dl with a mean of 5.96mg/dl (SD 3.048) and range of the salivary creatinine level was found to be 0.2-2.3mg/dl with a mean of 0.66mg/dl(SD 0.485).
Here it is obvious there is no agreement, as the values are way different.
Here comes the question when a same biological parameter(creatinine level) is different when taken from saliva and serum, how can we conclude that salivary creatinine levels can be used instead of serum? So, does it mean our study did not prove useful ??
After reading the Bland Altman plot , I am more confused if I'm wrong......
The best solution is to undertake a regression of salivary creatinine against serum creatinine. The predicted values can then be used in a Bland Altman plot against the serum creatinine levels. If the mean bias is not significantly different from zero, and the limits of agreement are clinically reasonable, you then have a good estimate of serum creatinine based on the regression equation.
Hello, Vikneshan.
There is a very interesting discussion about the matter you pose in:
https://www.researchgate.net/post/Applicability_of_Bland-Altman_Plot?_sg=xwMyhWK672wBbjxLYPv%2BJ4J10iMAAYoWN4SWvVStxIXJpTDvN8DXdpbN%2F5QUAzdn_NBEgDW53MEM7c3%2BM88iq%2BZ096WsFfsjRZ548ykJNkegW4DbWfYB0Wxn6h%2FDG%2Fbkm
Hope it helps.
Bland Altman plots are used to find the difference between two measurements. it is based on the average and standard deviation calculated for the two measurements.
To read how it is done, one simple example is shown here:
http://www.real-statistics.com/reliability/bland-altman-analysis/bland-altman-plot/
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS