1407. Performance of NHSN Risk Indicators on Predicting Surgical Site Infection
Session: Oral Abstract Session: Predicting and Preventing SSIs
Sunday, October 23, 2011: 8:30 AM
Room: 157ABC
Background: The Colorado Hospital-Acquired Infections Disclosure Act requires hospitals and ambulatory surgical centers to report surgical site infections (SSI) to the CDC-sponsored National Healthcare Surveillance Network (NHSN). Procedures entered into NHSN are risk adjusted using the NHSN risk index (NRI), an additive model based on the presence of American Society of Anesthesiologists score of 3, 4 or 5; wound class of 3 (contaminated) or 4 (dirty); and duration of surgery exceeding the procedure-specific 75th percentile. In 2009, the Standardized Infection Ratio (SIR) was developed by the CDC for estimating SSI risk. SIR is a logistic regression-based model that incorporates procedure-specific risk factors and is calculated using the expected SSI rate in a baseline population. We sought to validate and compare NRI and SIR at our institution, an academic level 1 trauma and public safety net hospital in Denver CO.

Methods: We analyzed cases of herniorrhaphy, vaginal and abdominal hysterectomy, and hip and knee arthroplasty that were entered into NHSN between January 1, 2009 and December 31, 2010. The sensitivity and specificity of both risk indices were assessed through logistic regression by calculating area under the curve (AUC). AUC ³ 0.7 was used as the standard for adequate predictive validity.

Results: 848 surgeries were included; 17 were associated with SSI. Overall, SIR performed better than NRI (AUC 0.735 vs. 0.525, respectively). AUC for SIR was larger than NRI for herniorrhaphy and hysterectomy, but smaller than NRI for arthroplasty.

 

 

 

Area Under Curve (AUC)

 

N

SSI

NHSN Risk Index (NRI)

Model-Based Risk Estimate (SIR)

Herniorrhaphy

454

6

0.644

0.816

Hysterectomy

176

6

0.506

0.643

Arthroplasty

218

5

0.526

0.399

Overall

848

17

0.525

0.735

Conclusion: SIR predicts overall SSI risk more accurately than NRI but did not perform well for arthroplasty. Neither SIR nor NRI adequately predicted SSI after hysterectomy or arthroplasty. The lower overall volume of these procedures (versus herniorrhaphy) in our population may limit the SIR’s ability to predict SSI. Better predictive models may be needed to ensure valid risk-stratification for lower-volume procedures.


Subject Category: N. Hospital-acquired and surgical infections, infection control, and health outcomes including general public health and health services research

Bryan Knepper, MPH, MSc, Connie Price, MD, Heather Young, MD, Timothy Jenkins, MD, Cathy Vigil, RN, CIC and Amber Miller, RN, MSN, CIC, Denver Health Medical Center, Denver, CO

Disclosures:

B. Knepper, None

C. Price, None

H. Young, None

T. Jenkins, None

C. Vigil, None

A. Miller, None

Findings in the abstracts are embargoed until 12:01 a.m. EST Thursday, Oct. 20 with the exception of research findings presented at IDSA press conferences.