乌鸦传媒

Artificial intelligence was predictive of care strategies for patients with diabetic kidney disease, a new study found.

Test results from the diagnostic tool were significantly tied to how a primary care physician tailored a patient's treatment strategy, reported Manasi Datar, PhD, of Boston Healthcare Associates in Massachusetts, and colleagues.

Specifically, having a patient with a high-risk KidneyIntelX result was associated with a 64% (OR 1.64, 95% CI 1.29-2.08) higher odds of a primary care physician prescribing an SGLT-2 inhibitor with a diabetic kidney disease indication, such as canagliflozin (Invokana) which gained this expanded indication in September 2019.

The prognostic test identifying a patient as high risk was also tied to a 49% higher odds of the primary care physician increasing the patient's angiotensin II receptor blocker (ARB) dose (OR 1.49, 95% CI1.17-1.89) compared with patients who did not undergo this test.

But most of all, a high-risk KidneyIntelX result was tied to a nearly 2.5-fold higher odds of a primary care physician referring a patient to a nephrologist (OR 2.47, 95% CI 1.99-3.08) versus patients with no test.

During a presentation of the late-breaking findings at the virtual National Kidney Foundation Spring Clinical Meeting, Datar said the findings "go to show that the KidneyIntelX test results are extremely important in physician decision-making when it comes to the management of patients with diabetic kidney disease."

"[This tool] can address several existing unmet needs by helping primary care physicians accurately risk-stratify and appropriately manage patients with early stage diabetic kidney disease," she added.

The conjoint analysis pooled survey answers from 4,401 primary care physicians. These physicians were given hypothetical profiles of patients with diabetic kidney disease that contained six clinical factors: albuminuria, estimated glomerular filtration rate, age, blood pressure, HbA1c, and the KidneyIntelX result.

The providers were shown eight different patient profiles out of a possible 42. And for each hypothetical patient shown, the primary care physician was asked whether they would prescribe an SGLT-2 inhibitor to this patient to increase their ARB dose, or refer the patient to a nephrologist.

The KidneyIntelX test is an in-vitro test using a blood draw and medical record information that grades patients with type 2 diabetes and early-stage kidney disease on a scale of low-risk (suggesting maintenance), intermediate-risk (suggesting the patient be monitored), or high risk (suggesting action in the treatment plan).

Patients graded high risk are likely to progress to end-stage kidney disease, and therefore require a more "aggressive" care plan, including specialist referral.

Session chair Joseph Vassalotti, MD, of the Icahn School of Medicine at Mount Sinai in New York City, praised the study, underscoring that "primary care is core to chronic kidney disease management."

"I think that exploration of better primary care for [chronic kidney disease] is really important for our country and for the future," he said.

The KidneyIntelX diagnostic was granted a by the FDA in May 2019 -- the first artificial intelligence-enabled diagnostic for kidney disease.

Comment