OUTCOMES IN PATIENTS WITH AND WITHOUT ICD-9 DIAGNOSED SEVERE SEPSIS AND SEPTIC SHOCK

Abstract

Title: Outcomes in Patients With and Without ICD-9 Diagnosed Severe Sepsis and Septic Shock Purpose: To evaluate the hypothesis that patients who present to the emergency department with objective findings of severe sepsis or septic shock, but who are not specifically diagnosed, denoted by absence of 995.92 or 785.52, are treated less aggressively and have worse outcomes than similar patients who are specifically diagnosed with severe sepsis or septic shock. Design: Retrospective cohort study Setting: University of Kansas Hospital, academic medical center, emergency department Patients: 6885 patients with severe sepsis or septic shock Methods: Data were extracted from the electronic health record using the query tool HERON, to identify patients with severe sepsis or septic shock admitted through the emergency department, between 11/01/07–09/31/15. Patients aged ≥ 18 years, who had an infection, received an antibiotic ≤ 8 hours after triage, and meet criteria for severe sepsis and/or septic shock were eligible for inclusion. Severe sepsis was defined either by explicit diagnosis, ICD-9 995.92, or clinical criteria, infection + ≥2 sites of organ dysfunction. Septic shock was identified by one of the following: 1) an explicit diagnosis of septic shock, ICD-9 785.52, 2) criteria for severe sepsis and an ICD-9 code of other shock or shock unspecified (785.50 or 785.59), or 3) criteria for severe sepsis and received a vasopressor. We compared treatment rates, based on the Surviving Sepsis Campaign three-hour recommendations, and outcome differences between severe sepsis and septic shock patients who had a diagnosis code of 995.92 or 785.52 to patients who met criteria, but were never diagnosed as denoted by no ICD-9 diagnosis code. Main Results: A total of 6885 eligible patients were identified, with a mean age of 60.4 years ± 16.9 (mean ± standard deviation). Half of patients were male, (51.3%) and 42.5% received an ICD-9 diagnosis code of 995.92 or 785.52. Septic shock was coded more frequently than severe sepsis (74.1% vs 32.8%). Three-hour bundle protocol completion rates were low for all patients (8.6%), but higher for those with an ICD-9 code than patients without (9.6% vs 7.9%, p=0.02). Average time to first antibiotic administration was also earlier for those with a diagnosis code (3.2 ± 3.1 hours vs 3.93 ± 3.9 hours, p <0.001). Therapeutic components of the 3-hour protocol, administration of an antibiotic and IV fluids if needed, were also found to be administered more often in patients with a diagnosis code (34.8% vs 28.6%, p<0.001). Within the cohort, 5631 (81.7%) patients had severe sepsis and no shock. Those with an ICD-9 code of 995.92 had higher mortality (6.3% vs 2.3%), higher ICU admission rates (44.7% vs 22.5%), and hospital lengths of stay (9.2 ± 6.9 days vs 6.9 ± 6.7 days), (all p<0.001). Discharge locations were also different. Severe sepsis patients with an ICD-9 diagnosis code were discharged home less (43.6% vs 52.0%, p<0.001), were discharged to hospice more (6.1% vs 4.4%, p<0.001), and were given home health services more often (22.4% vs 19.5%, p=0.01). Readmission rates (30-day) were highest for patients without an ICD-9 code of 995.92 (20.9% vs 25.5%, p<0.001). Among patients with shock (n=1254, 20.5%), there was no significant difference in mortality or post-hospital discharge locations between those with an ICD-9 diagnosis code of 785.52 and those without. Of interest, patients without an ICD-9 code of 785.52 had higher ICU admission rates (90.2% vs 83.8%), longer hospital stays (16.7 ± 14.8 days vs 13.4 ± 12.3 days), and longer ICU stays (7.7 ± 8.2 days vs 5.5 ± 6.2 days), (all p <.001). Conclusions: Patients with severe sepsis and septic shock continue to be underdiagnosed as evidence by ICD-9 codes and undertreated according to international surviving sepsis guidelines. Patients meeting shock criteria are more often diagnosed than those with severe sepsis. Overall, treatment rates were sub-optimal, but patients with an ICD-9 diagnosis had higher total and therapeutic component rates. Among septic shock patients without a code, secondary outcomes including ICU admission and hospital and ICU length of stays were worse compared to patients with an ICD-9 diagnosis code of 785.52. Patients with severe sepsis and without a diagnosis code had higher 30-day readmission rates. Further investigation is needed to identify factors contributing to ICD-9 code assignments, as well as financial implications of under-coding. Disclosures: The authors have nothing to disclose.