Factors Associated With the Incidence and Timing of Total Knee Arthroplasty Infection

Abstract

Periprosthetic joint infection (PJI) following total knee arthroplasty remains a challenging clinical problem. This study examined variables related to the incidence and timing of PJI. We retrospectively reviewed 8462 primary total knee arthroplasties performed at our institution between 2006 and 2018 for PJI. The mean follow-up is 3.7 years. Eighty-seven variables including patient-reported diagnoses, demographics, and medications were collected. Time to infection, bacterial organism, success of infection treatment, and variables associated with infection are reported. PJI occurred in 105 (1.24%) cases. The incidence of infection in the first year was 0.72% and represented 58% of all infections. Multivariate Cox regression revealed males (hazard ratio [HR]= 2.85, 95% confidence interval [CI], 1.69-4.79) and patients with major depression or anxiety (HR= 2.11, 95% CI, 1.21-3.67) were more likely to develop an infection in the first year. After the first year, patients with a history of cellulitis (HR= 3.97, 95% CI, 1.91-8.27) and those taking antiepileptic medications (HR= 3.61, 95% CI, 1.73-7.52) were more likely to develop an infection. Culture-negative infections were more common after one year than before (23% vs 8%, P= .04). Debridement, antibiotics, and implant retention was used in 79% (48/61) of infections in the first year and 55% (24/44) after one year with success rates of 51% and 70%, respectively (P= .16). The majority of infections occur during the first year after surgery. Importantly, patient variables associated with infection are different among infections that occur before and after one year. A multicenter study with a much larger number of infections may allow analysis of more time intervals after surgery.