EHRs could serve as more accurate tool to identify septic shock trends
While physicians are improving their ability to document sepsis, almost a third of them fail to do so on patient records, says Steven Simpson, MD.
Despite the fact that sepsis is one of the most deadly and costly medical conditions for hospitals, doctors frequently fail to properly document this immune system response to infection that kills more than 250,000 Americans each year.
“At least 30 percent of the time, the physicians who are writing their notes do not use the words sepsis, severe sepsis or septic shock. And, so it doesn’t appear in the record,” says Steven Simpson, MD, professor of medicine and interim director of the Division of Pulmonary and Critical Care Medicine at the University of Kansas.
Traditionally, tracking sepsis is done using diagnosis billing codes. The lack of documentation in the electronic health record (EHR) is why it is so difficult to get an accurate picture of septic shock trends, according to Simpson.
“The sepsis diagnosis code can only be placed in the chart by the chart coders if physicians use the term,” Simpson adds. “The coders are not allowed to interpret clinical data. They’re only allowed to use that term if the physician has specifically written ‘septic shock.’ ”
Also See: Alert system uses predictive analytics to combat sepsis
Nonetheless, using EHR data versus claims data, researchers recently examined annual septic shock incidence trends at 27 U.S. academic medical centers. In an article published this month in CHEST—the official publication of the American College of Chest Physicians—they looked at hospitalizations of all adults from January 2005 through December 2014 with either the septic shock ICD-9 code or clinical criteria.
For clinical surveillance, researchers looked for patients who received concurrent vasopressors, blood culture orders and antibiotics. What they found was that surveillance-based clinical data offered more reliable estimates of septic shock trends, compared with using ICD-9 codes.
“Clinician record reviews suggested that clinical surveillance definitions for septic shock provide greater sensitivity and comparable positive predictive value than billing codes,” said co-lead investigator Chanu Rhee, MD, an Instructor in the Therapeutics Research and Infectious Disease Epidemiology group within the Department of Population Medicine at Harvard Medical School/Harvard Pilgrim Health Care Institute.
“The most likely scenario here is that doctors are getting better at writing the diagnosis; therefore, coders are getting better at picking it up,” says Simpson, who adds that the sampling of academic centers in the study represents only about 1 percent of U.S. hospitals. “If the diagnosis code is entered, it is clear that the physician has written that diagnosis—the coders don’t miss it very often.”
In fact, he adds that based on his own study published in 2014 in the American Journal of Medical Quality, “If the physicians documented a sepsis diagnosis, it was always recorded that way by the coders.” At the same time, Simpson notes that in his study only 61 percent of patients were correctly diagnosed with severe sepsis by their care teams during hospitalization. “But, when they did document it, coders always got it right.”
So, how then could the CHEST study conclude that based on clinical criteria that septic shock incidence is growing and mortality decreasing? Simpson believes these results indicating better outcomes are the result of a “Will Rogers” effect or apparent epidemiological paradox.
“Doctors are getting better over time at identifying and documenting septic shock. And, moreover, they typically are getting better at calling cases that are less severe,” he says. “If they’re getting better at it, the administrative data are going to get better in a hurry, and the coders have more cases to document.”
Simpson contends that it’s “clear that the hospitals with the best outcomes for severe sepsis and septic shock have achieved those outcomes via standardizing their approach to diagnosis and treatment.” Still, he makes the point that “the incidence of septic shock is on the rise.”
In the CHEST study, researchers found that with clinical data, septic shock cases grew from 12.8 to 18.6 per 1,000 hospitalizations during the 10-year surveillance period, while mortality decreased from 54.9 percent to 50.7 percent. At the same time, the results from ICD-9 codes show a much larger jump, from 6.7 to 19.3 cases per 1,000 hospitalizations, while mortality dropped from 48.3 percent to 39.3 percent.
“The imperfect sensitivity of codes as well as our clinical surveillance definition suggest that both methods may still underestimate the true burden of septic shock,” concluded Rhee. “However, surveillance-based clinical data may allow for more reliable estimates of septic shock burden and trends compared with administrative data.”
Yet, Simpson believes the authors of the CHEST study appear to be faulting the coders when the problem lies instead with the physicians. “If you work in a hospital right now where you’re not seeing a rise in administrative claims for septic shock, you might be doing something wrong,” he concludes. “If that’s not happening, you need to examine your system.”
“At least 30 percent of the time, the physicians who are writing their notes do not use the words sepsis, severe sepsis or septic shock. And, so it doesn’t appear in the record,” says Steven Simpson, MD, professor of medicine and interim director of the Division of Pulmonary and Critical Care Medicine at the University of Kansas.
Traditionally, tracking sepsis is done using diagnosis billing codes. The lack of documentation in the electronic health record (EHR) is why it is so difficult to get an accurate picture of septic shock trends, according to Simpson.“The sepsis diagnosis code can only be placed in the chart by the chart coders if physicians use the term,” Simpson adds. “The coders are not allowed to interpret clinical data. They’re only allowed to use that term if the physician has specifically written ‘septic shock.’ ”
Also See: Alert system uses predictive analytics to combat sepsis
Nonetheless, using EHR data versus claims data, researchers recently examined annual septic shock incidence trends at 27 U.S. academic medical centers. In an article published this month in CHEST—the official publication of the American College of Chest Physicians—they looked at hospitalizations of all adults from January 2005 through December 2014 with either the septic shock ICD-9 code or clinical criteria.
For clinical surveillance, researchers looked for patients who received concurrent vasopressors, blood culture orders and antibiotics. What they found was that surveillance-based clinical data offered more reliable estimates of septic shock trends, compared with using ICD-9 codes.
“Clinician record reviews suggested that clinical surveillance definitions for septic shock provide greater sensitivity and comparable positive predictive value than billing codes,” said co-lead investigator Chanu Rhee, MD, an Instructor in the Therapeutics Research and Infectious Disease Epidemiology group within the Department of Population Medicine at Harvard Medical School/Harvard Pilgrim Health Care Institute.
“The most likely scenario here is that doctors are getting better at writing the diagnosis; therefore, coders are getting better at picking it up,” says Simpson, who adds that the sampling of academic centers in the study represents only about 1 percent of U.S. hospitals. “If the diagnosis code is entered, it is clear that the physician has written that diagnosis—the coders don’t miss it very often.”
In fact, he adds that based on his own study published in 2014 in the American Journal of Medical Quality, “If the physicians documented a sepsis diagnosis, it was always recorded that way by the coders.” At the same time, Simpson notes that in his study only 61 percent of patients were correctly diagnosed with severe sepsis by their care teams during hospitalization. “But, when they did document it, coders always got it right.”
So, how then could the CHEST study conclude that based on clinical criteria that septic shock incidence is growing and mortality decreasing? Simpson believes these results indicating better outcomes are the result of a “Will Rogers” effect or apparent epidemiological paradox.
“Doctors are getting better over time at identifying and documenting septic shock. And, moreover, they typically are getting better at calling cases that are less severe,” he says. “If they’re getting better at it, the administrative data are going to get better in a hurry, and the coders have more cases to document.”
Simpson contends that it’s “clear that the hospitals with the best outcomes for severe sepsis and septic shock have achieved those outcomes via standardizing their approach to diagnosis and treatment.” Still, he makes the point that “the incidence of septic shock is on the rise.”
In the CHEST study, researchers found that with clinical data, septic shock cases grew from 12.8 to 18.6 per 1,000 hospitalizations during the 10-year surveillance period, while mortality decreased from 54.9 percent to 50.7 percent. At the same time, the results from ICD-9 codes show a much larger jump, from 6.7 to 19.3 cases per 1,000 hospitalizations, while mortality dropped from 48.3 percent to 39.3 percent.
“The imperfect sensitivity of codes as well as our clinical surveillance definition suggest that both methods may still underestimate the true burden of septic shock,” concluded Rhee. “However, surveillance-based clinical data may allow for more reliable estimates of septic shock burden and trends compared with administrative data.”
Yet, Simpson believes the authors of the CHEST study appear to be faulting the coders when the problem lies instead with the physicians. “If you work in a hospital right now where you’re not seeing a rise in administrative claims for septic shock, you might be doing something wrong,” he concludes. “If that’s not happening, you need to examine your system.”
More for you
Loading data for hdm_tax_topic #better-outcomes...