Sepsis Study Pits EHR vs. Claims Data

This article first appeared October 2, 2017 on HealthLeaders Media.

By Tinker Ready

A study of sepsis cases using EHR rather than claims data finds little change in either incidence of infection or mortality over a four-year period.

Is the incidence of sepsis stable or is it increasing?

Awareness campaigns and clinical education programs would suggest a rise in the potentially fatal condition, which the Centers for Disease Control and Prevention describes as a “complication caused by the body’s overwhelming and life-threatening response to infection.”

Now, a study based on data from electronic health records suggests that the rate of sepsis and it corresponding death rate may have stabilized between 2009 and 2014.

The study authors make the case that data from EHRs offers a more accurate measure of the condition than billing data, which indicates that sepsis rose more than 10% during the period, but that mortality declined.

Not that the findings of the study are anything to celebrate.

The research, published last week in the Journal of the American Medical Association, concludes that 6% of all hospitalized patients had “clinical indicators” of sepsis in 2014.

More than 20% of them died in the hospital or were discharged to hospice.

Study author Chanu Rhee, MD, says the findings will give hospitals an objective way to measure sepsis incidence rates – something health systems have been trying hard to bring down.

Rhee said his hospital, Brigham and Women’s in Boston, has numerous programs designed to educate staff about sepsis. “While we do that, we need objective ways to assess the impact of those efforts,” he says.

Related article: The New War on Sepsis

“If we rely on diagnostic codes and claims data, we can easily get fooled about the impact we are having.”

Sepsis is notoriously hard to recognize. There are no diagnostic tests for it and its symptoms are not unique to the condition. At the same time, it’s fast moving and potentially deadly. The treatment: antibiotics, which hospitals have been trying to use less frequently to prevent resistance.

The attempt to quantify the overall incidence of sepsis has also been challenging.

The JAMA paper notes that other studies have come up with a range of incidence levels, from 900,000 to 3.1 million per year. Based on the EHR clinical findings, 1.7 million patients in the US will develop sepsis and 270,000 will die from it.

“I think awareness sepsis is rising, diagnosis and coding for sepsis is increasing, but… is the disease is becoming more common? I think our study cast doubt on that,” Rhee said.

The debate over the value of claims (or billing) data versus clinical data is ongoing, but it usually doesn’t find its way into the hospital setting. Still, as EHRs become the norm rather than the exception, the study suggests hospitals have the opportunity to more precisely identify and address potential quality trouble spots by running their own numbers.

“Claims data has traditionally been used because its readily available,” Rhee said. “Chart reviews are great. You can’t beat the granularity of expert physicians reviewing the charts. The problem with that is the resources and time involved. We believe EHRs have a nice balance of both.”

An ‘Outstanding’ Study

Craig Coopersmith, MD, is a surgeon and researcher at the Emory University School of Medicine. He calls the study “outstanding” and agrees that claims data can be subjective.

He notes that professional organizations and grass roots advocacy groups have been working together to raise awareness of sepsis. Increased recognition often leads to more diagnoses, not a rise in incidence.

Claims can also be confounded by providers who may be upcoding to maximize reimbursement. Clinicians who are not knowledgeable about sepsis can also miscode the condition, Coopersmith says.

“Accurate coding, even when everybody has the best possible intentions, can be challenging,” he said.

In addition to offering a more precise accounting, Coopersmith said Rhee’s group has “operationalized” its findings by modifying the existing definition of sepsis that could be used in EHR analysis at any hospital.

“You have to build a program in order to query your EHRs, but it can be done relatively simply,” he said.

Incidence of Sepsis Remains High

The researchers looked at clinical data from 2.9 million patients treated in 409 academic, community and VA hospital hospitals. In addition to incidence and mortality, they learned that sepsis is present in 35% of all hospitalizations that end in death. (The CDC notes that most cases of sepsis are acquired outside the hospital.)

Rhee says he wants to be clear that his group and their findings should not be interpreted as minimizing the threat of sepsis. The incidence rate they found is high. He thinks more work needs to be done to better prevent and treat sepsis.

His team’s finding offers a new approach to surveillance: “We need to makes sure we have objective and efficient surveillance methods so we can target our interventions and understand their impact and know what is working and what it not working.”

Tinker ReadyTinker Ready is a contributing writer at HealthLeaders Media.