Keywords
periprosthetic infection - knee - microorganisms - antibiotic resistance
Introduction
Periprosthetic infection (PPI) is one of the most serious complications in total knee
arthroplasty (TKA), with an incidence of around 0.29% in the United States and 2%
in Spain.[1]
[2] It is significantly associated with high morbidity and mortality, in addition to
greater economic expenses and the need for surgical reintervention, with an estimated
price of 6,815.4 dollars for each revision arthroplasty.[3] The main current advances are focused on the prevention, early diagnosis and treatment
of PPIs, in order to reduce the rate and improve patient outcomes.[4]
Among the main risk factors related to PPIs, those linked to the patient's lifestyle
stand out, such as obesity, excessive alcohol intake, active smoking, and the use
of intravenous drugs. Factors related to comorbidities, such as diabetes, psoriasis,
rheumatoid arthritis, and ankylosing spondylitis, are also relevant. On the other
hand, there are risks associated with surgery, such as a surgical time longer than
90 minutes.[2]
[5]
[6]
The strategies for the prevention of PPIs are mainly preoperative antibiotic prophylaxis,
preoperative skin cleaning, and the use of cemented prostheses loaded with antibiotics.[7]
Based on the current evidence, many centers continue to recommend the use of antibiotic
prophylaxis with first- or second-generation cephalosporin, intravenously, during
the preoperative period and in the following 24 hours after arthroplasty.[8]
The diagnosis of PPI remains a challenge due to multiple factors such as: false negative
cultures, non-diagnostic laboratory tests, and heterogeneous clinical presentations.[9]
[10]
[11] In search of a more conclusive diagnostic tool, different classifications have been
proposed, such as those by Tsukayama et al.[12] in 1996 or by the Musculoskeletal Infection Society (MSIS)[13] in 2011, which have undergone modifications until reaching the classification most
used currently, published in the 2018 International Consensus on Musculoskeletal Infection
by Parvizi et al.[9] ([Table 1]).
Table 1 Criteria for the diagnosis of PPI.
There are various ways of classifying PPIs, with the presentation time being one of
the most widely used. One of the first classifications of this type is the one proposed
by Coventry[15] in 1975, which was soon modified by Fitzgerald et al.[16] in 1977, who group them into acute, subacute and late ([Table 2]).[14]
[15]
[16]
Table 2 Classification according to presentation time of PPI.
For the correct treatment of a case of PPI, it is essential to isolate and identify
the etiological agent. The main microorganisms described are gram-positive bacteria,
gram-negative bacteria, and fungi in third place.[17]
[18]
Given the scarcity of data published in this regard in Chile, the present study becomes
necessary in order to know the regional reality in terms of the most prevalent microorganisms,
their sensitivity, and the characteristics of patients affected by PPI.
Materials and methods
We obtained the complete casuistry of patients who underwent TKA due to severe gonarthrosis
in a public hospital in Santiago de Chile. A descriptive and retrospective study of
these patients was carried out, including all those who underwent primary total knee
arthroplasty between 2001 and 2020, with a minimum follow-up of 1 year.
Until 2018, patients who met the PPI criteria according to the main international
guidelines (Tsukayama et al.[12] and MSIS[13]) were identified and included; after 2018, the diagnostic method was guided by the
Philadelphia consensus of the same year.
We excluded all patients operated on at other health centers, those with revision
prostheses, and those who had incomplete clinical records.
The surgical technique used followed the manufacturer's instructions and the prosthesis
model. In addition, surgical protocols and asepsis and antisepsis techniques were
used, respecting the quality standards of the hospital, with three doses of cefazolin
as the antibiotic surgical protocol (vancomycin in the case of patients allergic to
cephalosporins). All interventions were performed with medial parapatellar approach,
use of ischemia handle, and without the use of drainage.
Regarding the patients with PPI, the following variables were analyzed:
- Demographics: age, gender, previous illnesses and body mass index (BMI).
- Surgical: surgical time and antibiotic prophylaxis.
- Microbiological: isolated microorganism, and antibiotic susceptibility and resistance.
Each infection was classified according to its temporality between the TKA and the
moment of PPI diagnosis, using the classification described by Fitzgerald et al.,[16] and recording the microbiological variables found for each case.
Secondly, we analyzed the antibiotic and/or surgical management that was carried out
in the PPI patients and the morbidity and mortality after the treatment.
Finally, a descriptive analysis of the different variables was performed using Microsoft
Excel 2020 (Microsoft Corp., Redmond, WA, United States).
The study was approved by the institutional ethics committee.
Results
Between 2001 and 2020, a total of 544 TKAs were performed, and 8 patients (1.47%)
presented PPIs in their subsequent evolution.
Of those 8 patients (1.47%), 3 were male, and 5, female, with a mean age at the time
of installation of the primary prosthesis of 66 (±5.7) years, all with a presurgical
diagnosis of severe gonarthrosis.
Among the morbid conditions, the obesity of the patients stands out, with an average
BMI of 30.3 (±4.5) kg/m2; all patients had an associated comorbidity (hypertension, smoking, diabetes, or
other).
All patients received antibiotic prophylaxis with 1 g of intravenous cefazolin preoperatively
and 2 doses postoperatively. The mean operative time was of 114 (±23.3) minutes.
The time elapsed between the primary TKA and reoperation ranged from 12 to 3,324 days,
with a median of 411 days. Regarding the presentation, it was acute in 3 (37.5%) patients,
subacute in 3 (37.5%) and late in 2 (25%) cases ([Table 3]).
Table 3
|
Patient
|
Days
|
|
Patient 1
|
455
|
|
Patient 2
|
367
|
|
Patient 3
|
14
|
|
Patient 4
|
12
|
|
Patient 5
|
614
|
|
Patient 6
|
3,315
|
|
Patient 7
|
1,125
|
|
Patient 8
|
43
|
The main reason for consultation was joint pain in 6 cases, while the remaining 2
were joint stiffness. No patient presented with fever, and only 1 (12.5%) presented
a fistula.
Regarding the laboratory tests, C-reactive protein (CRP) on admission was elevated
in 7/8 patients and the leukocytes were elevated in 3 cases.
In total, 5 (62.5%) of the patients with PPI presented polymicrobial etiology, while
the other 3 (37.5%), a single causal bacterial agent was found. The main isolated
agent was multidrug-resistant Staphylococcus aureus, which occurred in 75% of the patients. The other causal agents found are described
in [Table 4].
Table 4
|
SA
|
EF
|
CNS
|
SE
|
EC
|
AB
|
KBL
|
SH
|
SP
|
PAE
|
|
Patient 1
|
+
|
+
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
|
Patient 2
|
-
|
-
|
+
|
+
|
-
|
-
|
-
|
-
|
-
|
-
|
|
Patient 3
|
+
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
|
Patient 4
|
+
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
|
Patient 5
|
+
|
-
|
+
|
-
|
+
|
+
|
-
|
-
|
-
|
-
|
|
Patient 6
|
+
|
-
|
-
|
-
|
-
|
-
|
+
|
-
|
-
|
-
|
|
Patient 7
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
+
|
-
|
-
|
|
Patient 8
|
+
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
+
|
Regarding antibiotic sensitivity, we found that all cultures that were positive for
S. aureus were sensitive to vancomycin and rifampicin. On the other hand, a high resistance
against ciprofloxacin was observed.
The remaining sensitivity rates of other antibiotics are described in [Table 5].
Table 5
|
Antibiotic
|
Sensitivity: % (sensitive/cultures studied)
|
|
Vancomycin
|
100% (12/12)
|
|
Rifampicin
|
100% (12/12)
|
|
Sulfamethoxazole-trimethoprim
|
100% (10/10)
|
|
Linezolid
|
100% (9/9)
|
|
Gentamicin
|
100% (9/9)
|
|
Ceftaroline
|
100% (8/8)
|
|
Clindamycin
|
50% (6/12)
|
|
Erythromycin
|
45% (5/11)
|
|
Oxacillin
|
45% (5/11)
|
|
Ciprofloxacin
|
44% (4/9)
|
Of the total number of patients, 7 (87.5%) underwent revision prostheses in 2 stages,
while the remaining patient (12.5%) underwent removal of the infected prosthesis and
3 surgical knee cleanings, subsequently dying from septic shock secondary to endocarditis.
Discussion
Though it can result in serious complications such as PPI, TKA has been increasingly
performed. The incidence of PPI in our center was of 1.47%, which is within the percentages
described in the literature.[19]
[20]
Among the main risk factors for developing a PPI, we find obesity, diabetes mellitus
and rheumatoid arthritis, among others.[5]
[21] It is noteworthy that most of the patients in the present study had a high BMI,
in addition to some of the other risk factors mentioned, which suggests that not only
the conditions of the health center are important, but also patient's medical condition.
The follow-up of patients after TKA has not been defined, and it varies according
to the guideline or medical center. We can observe, for example, that the Arthroplasty
Society of Australia[22] recommends an annual follow-up from the first to the tenth postoperative years and,
thereafter, every three to five years. On the other hand, Mending et al.[23] followed up 11,019 TKAs, investigating the temporal peaks of risk of failure, and
concluded that the optimal moments of follow-up include 6 and 12 months, then 3, 8,
12, and 17 years, and these last 2 periods are recommended for those with a BMI greater
than 40 kg/m2. In the present study, we observed infections with a median of 411 days and great
dispersion, ranging from 12 to 3,315 days (9 years); based on what was previously
published and according to what we found in the present study, it seems reasonable
to us to recommend an annual clinical follow-up after the first postoperative year,
continuing up to 10 years after surgery.
Once the diagnosis of PPI is suspected, the choice of the empirical antibiotic regimen
should be made considering the main and most likely etiologies. There is great variability
within the literature in this regard; Jaén et al.[24] conducted a multicenter study in Spain in which they isolated Staphylococcus epidermidis as the main infectious agent, while Girón-Cornelio[25] in a literature review with a focus on health centers in Peru, found that the main
etiology was coagulase-negative Staphylococcus. In the present study, on the other hand, we found multidrug-resistant S. aureus as the main infectious agent, which does not correlate with the previously-described
etiologies and shows that there is a high variability regarding the main agent in
each center.[24]
[25]
The study by Leijtens et al.[26] showed a high sensitivity to the combination of clindamycin and rifampicin, while
Gellert et al.[27] showed the treatment with various antibiotic biofilms, including rifampicin, ampicillin
and ciprofloxacin, according to the susceptibility. In our local population, we were
able to identify multirug-resistant S. aureus as the most prevalent agent, with vancomycin and rifampicin as the ideal antibiotics
to eradicate the infection, while we observed high resistance to ciprofloxacin and
erythromycin. It should be noted that the antibiotics sulfamethoxazole-trimethoprim,
linezolid, and gentamicin presented high sensitivity; however, they were studied in
a smaller number of cases. Given this difference that is observed within the different
health centers, it seems reasonable to recommend using the international treatment
guidelines only as a first approach, but that each center has its own study of etiological
agents with regards to the identification of the most frequent pathogens and their
antibiotic susceptibilities.
Within the limitations of our study, we can mention both the low number of TKAs and
PPIs in the sample, which could exaggerate the rates of the variables measured. On
the other hand, we did not perform a comparison of the risk factors for PPI with a
control group that did not present infection. This motivates us to propose a new line
of research, with a regional and national multicenter study, to describe the incidence,
isolated microorganisms, and susceptibility, in order to analyze trends and develop
a national management and treatment guide.
Another point to consider is that, although the sample of the present study is kept
under strict and closely-monitored follow-up due to their health insurance, it is
possible that a patient has made a change of address or city and developed PPI, which
could affect the final incidence of PPI. In any case, we estimate that it should not
correspond to a relevant factor, since all the patients adhered to the follow-up in
our center in the years following their intervention.
As strengths, we highlight the long period covered by the present study (18 years),
a standardized intervention by the same team, the diagnosis based on the latest relevant
international guidelines, and the analysis of the microbiological variables for each
of the PPI cases, also detailing the antibiotic susceptibility study, which could
serve as a guide for the different groups when deciding which empirical antibiotic
treatment to use in a case of PPI.
Conclusion
The incidence of PPI found in the present study was of 1.47%. All patients with PPI
presented some preoperative comorbidity. The main microbiological agents identified
were multidrug-resistant and susceptible to vancomycin and rifampicin.
