Testing for CKD: Which Tests Pull Their Weight?


Patients referred to nephrologists for evaluation of chronic kidney disease (CKD) undergo a lot of laboratory tests. But which tests contribute the most clinically relevant information? An analysis of nearly 1500 patients suggests that many tests performed for initial evaluation of CKD—including some tests ordered in a majority of patients—have little or no effect on patient diagnosis and management.

“We found that many tests are obtained frequently despite low rates of effect on diagnosis and management,” according to the report by Mallika L. Mendu, MD, MBA, of Harvard Medical School and colleagues. Their research letter was published online in JAMA Internal Medicine, and will appear in an upcoming print issue.

While emphasizing the need for further study, Mendu and colleagues conclude, “Reflexively ordering several tests for CKD evaluation and management may be unnecessary.” They are working on further analyses toward developing a more evidence-based, cost-conscious approach to CKD diagnosis.

“In this retrospective study of individuals undergoing an initial evaluation for CKD, the authors found that very few of the many tests ordered appeared to help the clinicians determine the cause of CKD or guide next steps in management,” said Amy W. Williams, MD, of the Division of Nephrology and Hypertension at Mayo Clinic College of Medicine, Rochester, Minn. “Although there are limitations to the study as the authors have well outlined, the finding highlights the need for a more targeted stepwise approach to the initial evaluation of CKD. The authors call for further investigation to determine pre-test probability of disease and identifying subgroups of patients who would benefit from more extensive evaluations.”

Which CKD tests affect diagnosis and management?

The researchers analyzed 1487 patients diagnosed with CKD after referral to nephrology clinics affiliated with Brigham and Women’s Hospital and Massachusetts General Hospital from 2010 through 2012. “CKD is probably the number one reason why patients are referred to nephrologists,” said Mendu. “Yet we don’t really have an understanding of why we are ordering the tests that we do, what tests we are ordering, and how useful they are.”

For each case, nephrology progress notes were reviewed to determine which tests “contributed to, confirmed, or established the underlying diagnosis of and/or any management decision related to CKD.” Although that process was unavoidably subjective, the researchers relied on documentation by providers, and included testing decisions that were specially stated in the nephrology progress notes that the results affected diagnosis and management decisions—including negative test results. An independent review of a random sample of records showed high interrater agreement.

With a median age of 70 and about one-fourth of minority race/ethnicity, the patients had high rates of typical CKD risk factors and comorbidities. Hypertension was present in 79 percent of patients, diabetes in 58 percent, and coronary heart disease in 26 percent. Most were on statins and beta blockers at the time of referral; more than 40 percent were taking an ACE inhibitor. At diagnosis, most patients had stage 3 CKD: stage 3a in 28.7 percent and 3b in 39.5 percent.

The patients underwent a total of 40 different tests—33 performed primarily for diagnosis and seven primarily for management. The most common tests were measurement of calcium, hemoglobin, phosphate, urine sediment, and parathyroid hormone; dipstick tests for blood and protein in urine; serum protein electrophoresis; and renal ultrasound.

Highest-yield tests reflect main causes of CKD

The test with the highest diagnostic yield was hemoglobin A1c measurement, performed in 12.6 percent of patients. The results affected diagnosis in 15.4 percent of patients and management in 10.1 percent (Table 1). “That’s not that surprising given that the most common cause of CKD is diabetes, followed by hypertension,” said Mendu.

Table 1


Similarly high yields were provided by the urine total protein to creatinine ratio, which affected diagnosis in 14.1 percent of patients and management in 13.7 percent; and urine microalbumin-to-creatinine ratio, which affected diagnosis and management in 13.0 and 13.3 percent of patients, respectively.

After that, the rates at which tests contributed to CKD diagnosis fell off sharply—including some commonly performed tests. Renal ultrasound was performed in about two-thirds of patients. Even though the results were abnormal in more than one-fourth of patients (26.8 percent), the findings affected diagnosis in 5.9 percent of patients and management in just 3.3 percent.

Serum protein electrophoresis (SPEP) was also performed in more than two-thirds of patients, but affected diagnosis in just 1.4 percent of patients and management in 1.7 percent.

Cryoglobulins were measured in 5 percent of patients, and affected diagnosis and management in 5.4 percent of cases each. No other test performed in more than 5 percent of patients had a diagnostic yield of greater than 3.5 percent. Kidney biopsy affected diagnosis and management in every case—but was performed in only about 5 percent of patients.

Some tests did not affect diagnosis or management in any patient, including anti-neutrophil cytoplasmic antibodies, measured in about 14 percent of patients; and anti-glomerular basement membrane antibodies, assessed in about 4 percent.

Other tests were performed relatively often but contributed little information. For example, complement 3 and 4 were each tested in about one-fourth of patients, but had a diagnostic yield of about 1 percent. Hepatitis B and C testing were performed in about 17 percent of patients, but affected diagnosis in just one or two cases.

Need for evidence-based approach to CKD testing

With a reported prevalence of about 13 percent and many possible causes, CKD carries high morbidity, high mortality, and high costs. The new study is a comprehensive assessment of the range and impact of tests used for CKD diagnosis.

Mendu previously reported similar conclusions in a study of patients with syncope—a common clinical problem with an even broader differential diagnosis (JAMA 2009; 169:1299–1305). “That study found kind of a similar thing, which is that we order lots of tests, and most of the time, those tests don’t really affect diagnosis and management,” she said.

So are patients with CKD undergoing too many tests? “I think it’s helpful for nephrologists to know that it’s probably not the best approach to just order a litany of tests when a patient comes to see you,” Mendu said. That’s an approach that’s a “probably low yield”—an important consideration in the current era of healthcare reform and accountability.

Timing is another important factor to consider, she said. “Ordering these tests is time-consuming, and so is following up the test results. And if it’s not really contributing to patient care, then we should probably be more thoughtful about how we order these tests.”

Of course, nephrologists have reasons for ordering some of the lower-yield tests, such as renal ultrasound and SPEP—they are looking for less common but critical diagnoses such as multiple myeloma and obstructive hydronephrosis. Mendu and colleagues have identified cases in which those diagnoses were made, in an attempt to develop criteria and guidance for nephrologists and internists pursuing those diagnoses, which will be the topic of a manuscript currently in preparation. That paper will also seek to develop estimates of the costs of testing for CKD on the national level.

Ultimately, Mendu and colleagues write, “An evidence-based, targeted approach based on pretest probabilities of disease for diagnosis and management may be more efficient and reduce costs.” Such an effort would require further research, including studies in community-based patient samples and an emphasis on identifying patient subgroups who may benefit from more extensive evaluation.

“Many medical specialties and subspecialties, including nephrology, have begun tackling the concerns of overtesting or inappropriate testing and healthcare systems have begun to embed decision support and hard stops into electronic health records and CPOE,” said Williams, who also serves as Medical Director of Hospital Operations for the Mayo Clinic Hospitals. “However, we need to better understand more effective and efficient approaches to the diagnosis and management of CKD.

“We also need to share this knowledge with our primary care colleagues and non-nephrology subspecialists who may be the first to attempt CKD evaluation. Certainly our patients, as they see their co-pays increase, are depending on us to use resources wisely and in a patient-centered way and share our knowledge with their entire medical team.”