Multicenter evaluation of enhanced surgical site infection surveillance following total knee and hip arthroplasty

Background: A previous study showed that diagnosis code-based screening is a sensitive method for identifying surgical site infections (SSI) following total knee (TKA) and total hip arthroplasty (THA). In that study, PPV was extrapolated from known SSI and a random sample of procedures that did not include all procedures with the diagnosis codes of interest.
Objective: Improve estimates of the PPV of diagnosis code-based screening for identifying SSI after TKA and THA by reviewing all procedures meeting screening criteria.
Methods: We conducted a retrospective cohort study of 1,666 TKA  and 1,691 THA procedures performed between 1/1/2007 and 12/31/2007 in 4 hospitals located in different geographic areas of the U.S. Routine prospective SSI surveillance was performed at all hospitals. Retrospective medical record review was completed 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, 996.66) during the index surgery hospitalization or readmission within one year of surgery.
Results: The overall SSI rate based on routine surveillance was 1.1% following TKA and 1.2% following THA. Including the number of new SSI identified among procedures meeting diagnosis-code based screening criteria, the adjusted SSI rates were 3.2% and 2.1% following TKA and THA, respectively. PPV of diagnosis code-based screening were 0.93 and 0.72 following TKA and THA, respectively. Patients met these screening criteria after 3% of all TKA and THA procedures. The performance of these methods is shown below.

Hospital	SSI rates based on routine surveillance	Adjusted SSI rates based on enhanced surveillance	Routine surveillance	Enhanced surveillance
			Sensitivity	Sens	PPV
Total Knee Arthroplasty
Hospital A	0.6%	2.8%	0.22 (4/18)	1.0	0.86 (18/21)
Hospital B	1.0%	1.0%	1.0 (2/2)	1.0	1.0 (2/2)
Hospital C	1.5%	3.1%	0.50 (5/10)	1.0	0.91 (10/11)
Hospital D	1.6%	4.7%	0.33 (8/24)	0.96	1.0 (23/23)
TOTAL	1.1%	3.2%	0.35 (19/54)	0.98	0.93 (53/57)
Total Hip Arthroplasty
Hospital E	0.9%	1.8%	0.50 (6/12)	1.0	0.71 (12/17)
Hospital F	1.0%	2.1%	0.44 (4/9)	0.89	0.73 (8/11)
Hospital G	1.1%	2.2%	0.50 (5/10)	1.0	0.83 (10/12)
Hospital H	3.2%	3.2%	1.0 (6/6)	1.0	0.60 (6/10)
TOTAL	1.2%	2.1%	0.57 (21/37)	0.97	0.72 (36/50)

Conclusions: These results closely mirror those from our previous study and verify that routine surveillance fails to detect more than half of SSI following TKA and THA. Medical record review of patients meeting diagnosis code-based screening criteria demonstrated that the majority of these patients (93% of TKA, 72% of THA) met National Healthcare Safety Network definitions for SSI.  Focused surveillance among the 3% who met screening criteria was 2-3 times more sensitive than routine surveillance for detecting SSI. In many hospitals, this method would require less effort than routine surveillance of all patients.