163. Development of an Electronic Flagging Tool for Identifying Cardiac Device Infections: Insights from the VA CART Program
Session: Oral Abstract Session: Healthcare Epidemiology: Special Populations
Thursday, October 4, 2018: 11:30 AM
Room: S 156

Background: Surveillance is an essential aspect of infection prevention. Despite the high morbidity and mortality associated with procedure-related Cardiac Implantable Electronic Device (CIED) infections, methods for identifying them are limited. The objective of this study was to develop an algorithm with electronic flags to facilitate detection of CIED infections in a large, multi-center cohort.

Methods: A sample of patients who underwent CIED procedures entered into the VA Clinical Assessment Reporting and Tracking Electrophysiology (CART-EP) program from FY 2007-2015 were included in the nested case-control study. After cohort creation, data from this review process were combined with electronic variables (e.g., microbiology orders, ICD 9/10 codes) to develop a preliminary algorithm that categorized patients as high, intermediate or low risk of CIED infection.

Results: N=1,014 unique patients out of a cohort of 5,955 procedures underwent manual review. Among these cases, 59 CIED infections and 955 controls were identified. Electronic variables predictive of CIED infection included ICD 9/10 infection codes and microbiology orders (Table 1).  ICD 9/10 codes had excellent PPV for flagging infections but sensitivity was limited (47.5%, see Figure 1). Adding microbiology order flags increased sensitivity but lowered specificity. Specificity in patients without either flag was outstanding (99%).

Conclusion: Absence of ICD 9/10 and microbiology orders is highly specific for ruling out CIED infections. The discriminatory abilities of the algorithm for intermediate probability flags (+microbiology/- ICD9/10) need improvement. In patients without ICD codes, at least microbiology orders should be used as a flag to streamline review of possible device infections. Refinement of this tool using other clinical flags may improve operating characteristics and clinical utility.

Table 1: Electronic Flags for CIED infection

Infection Flag

Infection (N=59)

No Infection (N=955)

OR

P-value

CIED Infection ICD 9/10

21/59 (35.6%)

1/955 (0.10%)

340

<0.001

Surgical Site Infection (SSI) ICD 9/10

7/59 (11.9%)

6/955 (0.63%)

18.9

<0.001

CIED infection or SSI ICD 9/10

28/59 (47.5%)

7/955 (0.73%)

64.7

<0.001

Micro Order*

53/59 (89.8%)

198/955 (20.7%)

5.4

<0.001

  *Blood, wound, and unclassified cultures

Archana Asundi, MDCM1,2, Maggie Stanislawski, PhD3, Payal Mehta, MD1, Hillary Mull, PhD4, Marin Schweizer, PhD5, P. Michael Ho, MD, PhD6, Kalpana Gupta, MD, MPH7,8 and Westyn Branch-Elliman, MD, MMSc1,4,9, (1)VA Boston Healthcare System, West Roxbury, MA, (2)Internal Medicine, Section of Infectious Diseases, Boston University Medical Center, Boston, MA, (3)University of Colorado School of Public Health, Denver, CO, (4)Center for Healthcare Organization and Implementation Research, VA Boston Healthcare System, West Roxbury, MA, (5)Department of Epidemiology, University of Iowa College of Public Health, Iowa City, IA, (6)Seattle/Denver Center of Innovation for Veteran-Centered and Value-Driven Care, Denver, CO, (7)VA Boston Healthcare System, Boston, MA, (8)Boston University School of Medicine, Boston, MA, (9)Harvard Medical School, Boston, MA

Disclosures:

A. Asundi, None

M. Stanislawski, None

P. Mehta, None

H. Mull, None

M. Schweizer, None

P. M. Ho, None

K. Gupta, None

W. Branch-Elliman, Veterans' Integrated Service Network Career Development Award: Investigator , Research grant . American Heart Association: Investigator , Research grant .

<< Previous Abstract | Next Abstract

Findings in the abstracts are embargoed until 12:01 a.m. PDT, Wednesday Oct. 3rd with the exception of research findings presented at the IDWeek press conferences.