• 154,419+ Medical Conferences
  • 285 Specialties
  • 11,290 Organizers
  • 156,370 Speakers
  • 159 Countries
  • 4,785 Cities
CME Conference Registrations Sold


EHR Data May Beat Claims Data in Assessing Sepsis Incidence

EHR Data May Beat Claims Data in Assessing Sepsis Incidence

Fri, Sep 15, 2017

FRIDAY, Sept. 15, 2017 (HealthDay News) -- Sepsis is present in 6 percent of adult hospitalizations, with no change in incidence from 2009 to 2014 based on electronic health record (EHR) data in contrast to claims-based analyses, according to a study published online Sept. 13 in the Journal of the American Medical Association.

Chanu Rhee, M.D., M.P.H., from Harvard Medical School in Boston, and colleagues conducted a retrospective cohort study of adult patients admitted to 409 hospitals from 2009 to 2014 to estimate the national incidence of sepsis.

The researchers identified 173,690 sepsis cases using clinical criteria among 2,901,019 adults admitted to study hospitals in 2014 (6.0 percent incidence). Fifteen percent of these patients died in the hospital and 6.2 percent were discharged to hospice. Using clinical criteria, the incidence of sepsis was stable from 2009 to 2014 (+0.6 percent relative change/year), while incidence increased per claims (+10.3 percent/year). There was a decrease in in-hospital mortality using clinical criteria (−3.3 percent/year), but the combined outcome of death or discharge to hospital did not change significantly (−1.3 percent/year). Using claims, mortality declined significantly (−7.0 percent/year), as did death or discharge to hospital (−4.5 percent/year). Compared with claims, clinical criteria were more sensitive for identifying sepsis (69.7 versus 32.3 percent), with comparable positive predictive value (70.4 versus 75.2 percent).

The findings "suggest that EHR-based clinical data provide more objective estimates than claims-based data for sepsis surveillance," the authors write.

Several authors disclosed financial ties to the pharmaceutical and health care industries.

Abstract/Full Text