CKD Alerts Did Not Improve Primary Care Screening: Antoine Créon, MD

Adding chronic kidney disease (CKD)–specific alerts to an existing electronic decision support system did not improve screening or later kidney care in primary care, according to Antoine Créon, MD, a PhD candidate at Karolinska Institutet. In an interview with The American Journal of Managed Care^®, Créon discussed the trial’s main findings, the screening and retesting results, and how much recommended care remained incomplete.

What the Stockholm Trial Found on Screening

The ALMA-CKD pragmatic cluster randomized trial (NCT06386172) was conducted across 65 primary care centers in Stockholm, Sweden. The alerts used electronic health records to identify care gaps, such as patients with diabetes who lacked a recent creatinine or albuminuria measurement or patients with confirmed albuminuria who had no recommended medication prescribed. The system then suggested the appropriate action, such as ordering a test or prescribing the medication.

The co–primary outcomes were creatinine and albuminuria testing within 12 months among patients eligible for yearly screening who had not been tested recently. Among about 37,000 individuals eligible for creatinine testing, screening occurred in 62% of the intervention arm and 65% of the usual care arm. Albuminuria screening was 42% and 44% in the 2 arms. Neither difference was significant.

The team also found no clear improvement in other steps of care, including repeat testing of abnormal results, recorded CKD diagnosis, entry into nephrology care, and treatment initiation. “What stood out was how much recommended care remained incomplete,” Créon said. Fewer than half of eligible patients received albuminuria testing.

Why Retesting and Diagnosis Rates Stayed Low

Retesting of abnormal results followed a similar pattern. About 60% of patients with newly detected low estimated glomerular filtration rate (eGFR) had a repeat test ordered, and about 30% of patients with newly discovered elevated albuminuria had a repeat test ordered. Among patients with laboratory-confirmed CKD who met the formal definition, only about 10% received a diagnosis code in the health record.

Créon cautioned that a diagnosis code is “an imperfect measure,” because its absence does not prove the clinician was unaware of CKD or was not managing it. Overall, the alerts did not improve the measured steps from screening to confirmation and CKD documentation.

What Health Systems Should Take From the Results

Because the intervention covered an entire regional primary care organization, rather than a small group of highly selected and motivated volunteer practices, the findings speak to ordinary practice. In that setting, clinicians have competing priorities, patients have multiple conditions, and follow-up does not always happen as planned. The team used routinely collected laboratory and drug dispensing data to assess whether recommended care actually occurred.

As Créon put it, “The practical message is that health systems should not assume that electronic alerts alone will by themselves close these gaps.” The trial identified care gaps clearly, Créon said, but the intervention did not close them.