The need for rapid estimates of renal function is leading to greater demand for point-of-care creatinine testing in a variety of patient settings including Radiology, to determine patient ability to clear contrast dyes, and Hematology/Oncology, for dosing chemotherapeutic agents. This demand has led to the development of a number of POCT devices, and creatinine sensors have even been added by manufacturers to existing blood gas instrumentation to allow for expansion of testing on current analyzers into new settings. With such demand for faster test results, the question arises whether the technical performance (accuracy/precision) of the POCT creatinine methods is keeping pace with the central laboratory tests.
A number of studies have examined the performance of POCT creatinine methods (1-3). These studies have shown positive and negative biases compared to laboratory methods, some significant in the 1.5 – 3 mg/dL, clinically important range(1) with variable imprecision. In general, the studies conclude that POCT methods have adequate performance to meet the clinical needs of the specified settings provided that physicians keep the differences between POCT and laboratory creatinine methods in mind when interpreting the results. However, the National Kidney Disease Education Program (NKDEP) has recently recommended that manufacturers work to ensure that creatinine methods be traceable to isotope dilution mass spectrometry (IDMS) and achieve total error goals of analytical imprecision (including interlaboratory calibration variability) SD <0.08 mg/dL (7.1 µmol/L) and analytical bias (compared to an IDMS reference measurement procedure) <0.05 mg/dL (4.4 µmol/L) at a serum creatinine concentration of 1.00 mg/dL (88.4 µmol/L).(4) The NKDEP also recommends manufacturers design instruments that report serum creatinine values as mg/dL to two decimal places, or as µmol/L to the nearest whole number. This will reduce the contribution of a rounding error when using the MDRD Study equation.(4)
Clearly current POCT devices do not report creatinine to 2 decimal places, and whether all POCT devices can achieve the NKDEP total error goals is questionable. Do all POCT devices have IDMS treaceable calibrations? What about drug and other patient interferences with the POCT methods, as the NKDEP recommendations for total error are expected despite the potential for method interference? These issues remind me of the current debate over the technical performance of POCT HgbA1c tests after the ADA recommended that HgbA1c could be used for diagnosis of diabetes. Similar debate is also surrounding POCT glucose testing and intensive insulin management, particularly with regard to technical performance criteria and which methods, blood gas sensors or glucose meter, are better suited for intensive insulin protocols.
So, what is the optimal performance for POCT creatinine and who should make that determination, the clinician using the method, a consensus panel of clinicians with disease focus, laboratory specialists, or someone else? I would ultimately argue that it should be the physician who is legally responsible for the care of the patient, but that is contingent on the physician being educated on the limitations of the test result they are receiving. Many clinicians that I speak with on a day-to-day basis do not understand the differences between POCT and laboratory results. In addition, different clinicians utilize results in different ways, some for screening organ function while others are looking to progression of disease, prognosis, or even pharmacist dosing medications. Each of these clinical applications may have different performance goals required of the testing methodology, and one standard may not fit all clinical scenarios. If we are guided by consensus guidelines, then what if the performance of POCT creatinine devices doesn’t meet those guidelines? Should such methods even be utilized to estimate renal function? Or, should the devices only be utilized in selective populations, like Radiology or Hematology/Oncology where the performance may be adequate for the clinical application? Or, can the device be used more widely to estimate renal function in the general patient population? If POCT devices have greater imprecision and biases compared to IDMS traceable laboratory methods, should GFR be estimated from these methods at all, or should the physicians rely on the creatinine result alone to guide management? Should we have a separate eGFR equation that considers the variability of current POCT methods, separate from the MDRD equation, possibly for specific clinical applications? Given the differences that are possible between POCT and laboratory methods, there will be a further need to ensure separation of test results in a clear manner in the medical records, so clinicians understand the source of the test result with its inherent biases and interferences when interpreting results. There are obviously a number of unanswered questions that need exploring in future POCT creatinine research.