Work

A marginal structural approach to measuring the comparative effectiveness of echinocandins Versus fluconazole therapy for the treatment of adult candidemia (MSG-12)

Year: 2016

Type: article

Abstract: Background. Based on patient level analysis of randomized controlled trial data the Infectious Diseases Society of America guidelines support echinocandins as the first line therapy for candidemia, but observational data confirming the benefit of echinocandins in real world settings are lacking. Existing reports are limited by intention to treat approaches that ignore time varying confounding and assume no loss to follow up (LTF) biasing treatment effect estimates. Methods. A US observational cohort of incident adult inpatient episodes of candidemia from 2009 to 2013 was assembled using Premier Perspective, a multicenter inpatient database. Patients transferred from an outside institution or dying within 3 days of the blood culture draw were excluded. We compared 30-day mortality following initial therapy with an echinocandin versus fluconazole using both logistic regression as well as a marginal structural model (MS model). MS model employed both inverse probability of treatment and censoring weights to control for time varying confounders and differential LTF. Covariates included in these models are detailed in the table. Results. The cohort included 2014 admissions with candidemia; 45.6% and 54.4% received fluconazole and an echinocandin as initial therapy, respectively. The 30-day mortality was 13.5%. While standard logistic regression suggested an increased 30-day mortality risk for patients initiated on echinocandins, MS model found echinocandins to be associated with a reduced risk of death at 30 days. Regression Model Results AAdjusted for age, gender, baseline vasopressor use, receipt of empiric antifungal therapy, and hospital. BAdjusted for covariates above and daily time varying exposure to vasopressor and antifungal therapy. Conclusion. Using a MS model, which accounted for time varying confounding and differential LTF, initial therapy with an echinocandin was associated with decreased risk for mortality. These results highlight the importance of leveraging such analytic models when daily therapy choices are based on a patient's clinical status and when LTF is variable and informative. MS Model also accounted for differential LTF by baseline and time varying covariates. Disclosures. J. Baddley, Pfizer: Consultant, Consulting fee Merck: Consultant, Salary L. Ostrosky-Zeichner, Astellas: Consultant and Investigator, Consulting fee and Research grant Merck: Non-branded educational speaker and Scientific Advisor, Consulting fee and Speaker honorarium Pfizer: Investigator and Non-branded educational speaker, Research grant and Speaker honorarium Cidara: Scientific Advisor, Consulting fee Scynexis: Investigator and Scientific Advisor, Consulting fee and Grant recipient P. Pappas, Merck: Grant Investigator, Grant recipient Astellas: Grant Investigator, Grant recipient Gilead: Grant Investigator, Grant recipient T2 Biosystems: Grant Investigator and Scientific Advisor, Consulting fee and Grant recipient IMMY: Grant Investigator, Grant recipient Viamet: Scientific Advisor, Consulting fee Vical: Scientific Advisor, Consulting fee Matinas: Scientific Advisor, Consulting fee Amplyxx: Scientific Advisor, Consulting fee T. Zaoutis, Merck: Grant Investigator, Research grant B. Fisher, Pfizer: Grant Investigator, Research grant Merck: Grant Investigator, Research grant Ansun BioPharma: Research Contractor, Research grant

Source: Open Forum Infectious Diseases

DOI: 10.1093/ofid/ofw172.1333