Estimating GFR and managing chronic kidney disease are blood brothers. The CKD framework is based on reliable and convenient assessment of GFR, and GFR estimating equations have evolved considerably from the Cockcroft-Gault equation of the 1970s. The most recent KDIGO guidelines recognize how various cultural differences in diet and genetic variation mean that one equation is cannot fit every local population, and recommends using local equations where possible. A recent article in AJKD highlights one such effort in Pakistan. Corresponding author Dr. Tazeen Jafar (TJ), Professor of Health Services and Research at the Duke-National University of Singapore and Professor of Nephrology and Community Health Sciences at the Aga Khan University in Karachi, Pakistan, discusses this article with Dr. Joel Topf (eAJKD), eAJKD Advisory Board member.
eAJKD: This project sought better GFR estimating formulas for the Pakistani population. How did this project start?
TJ: Chronic kidney disease is a major health problem globally, and often neglected in the developing countries. In addition, there hasn’t been a way to measure the burden of CKD in many countries. Most labs in Pakistan do not report estimated GFR automatically. We conducted a brief survey, including urban cities where a number of labs are accredited by the international joint medical commission, and found very few report estimated GFR alongside the creatinine. This is unlike what is happening in the U.K. and the U.S., where over 75 percent of the labs have been automatically reporting eGFR for the past few years. As we attempt to implement automatic eGFR reporting in Pakistan, we need to know if the “western habit and weight based” equations are appropriate in this population with a different diet and body structure.
eAJKD: To start, can you describe your four measures of validity for an eGFR equation: bias, precision, accuracy, and root means square.
TJ: Bias is the median difference of the actual GFR from the estimated GFR. If you take the measured GFR and the estimated GFR, you can calculate all the differences and compute the median of the differences. That’s the definition of bias.
Precision is the interquartile range of the bias. The bias and the precision are really telling you how far apart these different measures are from the truth.
Accuracy is measured in two different ways. One is P-20, and the other is P-30. A P-20 is what percent of estimates are within 20 percent of the measure. A P-30 is what percent of estimates fall within 30 percent of the measure. And finally, the root means square is a little bit more technical. The definition is the square root of the average squared difference of measured GFR and estimated GFR. It is really a difference between the measured GFR and the estimated GFR. For the general reader, it is easier to focus on the accuracy.
eAJKD: How does the MDRD estimating equation perform in a Pakistani population?
TJ: If we were to use the MDRD equation, the P-20 is about 50. Fifty percent of your estimates are going to be within 20 percent of your actual GFR.
eAJKD: How much better does it get with the CKD-EPI equation?
TJ: The P-20 rises to about 58 percent. If you do the P-30, the MDRD equation improves to 68% and the CKD-EPI equation to 76%. When we modified the CKD-EPI equation with a correction factor for the Pakistani population, the P-30 improves to 82%.
We also developed a novel GFR estimating equation. The new equation includes a number of additional variables such as hypertension, diabetes, fasting blood glucose, serum urea nitrogen, and hemoglobin level. Despite having all these additional variables in these equations, the new equation performed comparably to our modified CKD-EPI equation. Our novel equation added a lot of variables that may not be available to a clinical lab and only added a bit to accuracy.
eAJKD: This is consistent with what’s been done in other regions or with other ethnicities where adjustments were made to the CKD-EPI equation for those local populations.
TJ: Correct. KDIGO recommends that the CKD-EPI equations be modified for the local population. Despite a more extensive exercise with additional variables, we really didn’t show a marked improvement compared to the modification of the CKD-EPI equation in our population. What we are hoping is that this will catalyze the implementation of the equation in the region. Both the CKD-EPI equation and the modified CKD-EPI equation are very good in Pakistan, and we look forward to testing it in neighboring countries where it will probably perform as well.
eAJKD: If my next patient that comes in the clinic was born in the U.S. but is of Pakistani origin, which equation should I be using?
TJ: I think that I would favor using the CKD-EPI Pakistan equation if you see somebody of Pakistani origin. Validation of our equation is still required.