Enhanced Surgical Site Infection Surveillance for Non-primary Knee and Hip Arthroplasty Procedures

Background: We previously showed that surveillance criteria based on ICD-9 diagnosis codes suggestive of surgical site infection (SSI) have excellent sensitivity and positive predictive value (PPV) for identifying SSI following the primary knee (KA) and hip (HA) arthroplasty procedures targeted by the Surgical Care Improvement Project (SCIP). The National Healthcare Safety Network (NHSN) includes some additional ICD-9 procedure codes within its knee and hip arthroplasty categories, including KA and HA revisions.

Objective: Evaluate the usefulness of ICD-9 diagnosis code-based screening to identify SSI following NHSN hip and knee arthroplasty procedures not targeted by SCIP.

Methods: We conducted a retrospective cohort study of 224 KA and 295 HA performed between 10/1/2007 and 6/30/2009 at one academic medical center. Medical records were retrospectively reviewed to assess SSI status for 1) all procedures with SSI by routine surveillance and 2) all procedures with at least one ICD-9 diagnosis code suggestive of SSI (998.5, 998.51, 998.59, or 996.66) during the index surgery hospitalization or readmission to the same hospital within one year of surgery. We excluded SSI diagnosis codes from the index surgery hospitalization with a Present on Admission (POA) designation.

Results: SSI rates based on routine surveillance for non-SCIP KA and HA were 2.7% and 1.7%, respectively. Including additional SSI identified among procedures meeting diagnosis-code based screening criteria, adjusted SSI rates were 3.6% and 3.1% following KA and HA, respectively. The performance of routine vs. diagnosis code-based surveillance is shown below.

Procedure	SSI rates based on routine surveillance	Adjusted SSI rates based on enhanced surveillance	Routine surveillance		Enhanced surveillance
			Sensitivity	PPV		Sensitivity	PPV (ignoring POA codes)	PPV (excluding index surgery admission POA SSI codes)
Knee arthroplasty	2.7%	3.6%	0.75 (6/8)	0.85 (6/7)		1.0 (8/8)	0.20 (8/40)	0.62 (8/13)
Hip arthroplasty	1.7%	3.1%	0.55 (5/9)	0.83 (5/6)		1.0 (9/9)	0.33 (9/27)	0.90 (9/10)

Conclusions: These results demonstrate that diagnosis code-based screening can be useful for detecting some SSI missed by routine surveillance following non-SCIP knee and hip arthroplasty. When POA codes for the index surgery hospitalization were utilized, medical record review of patients meeting diagnosis code-based screening criteria demonstrated that the majority of these patients (62% of KA, 90% of HA) met NHSN definitions for SSI.  Furthermore, including a clinician diagnosis of “cellulitis” as SSI would improve PPV for TKA from 62% to 100%. Focused surveillance among the 3-6% who met screening criteria was more sensitive than routine surveillance for detecting SSI. In many hospitals, this method would require less effort than routine surveillance of all patients.