Infection complicating 794 primary and revision arthroscopies. Accuracy of actual prophylactic procedures against infection and results from a single orthopedic center in Poland

Septic arthritis after arthroscopy (SAAA) is a very rare, but serious complication of elective surgery with potentially devastating influence on the intraarticular structures [1, 2, 3, 4, 5, 6]. It can be expected that, at the time of the growth number of arthroscopic procedures, the total number of SAAA will also increase [7]. The incidence of SAAA is estimated at 0.04%–5.7% depending on the area of the operation, the complexity of the operation, the use of implants and the factors related to patients [1, 4, 5, 8, 9, 10, 11, 12]. The statistical data available in the literature were collected from different sources (insurance databases, hospital's databases, national registries, etc.) [9, 11, 12, 13, 14]. To our knowledge there is a lack of a detailed analysis of post-arthroscopy infections in European population.

Definition of infection related to implant differ between the countries [15]. Most of studies assessing number of post-arthroscopic infections were made in USA - adopting Center for Disease Control (CDC) definition, which is limiting the time of occurrence of symptoms to 90 days [10, 12, 16]. We decided to adopt longer follow-up in case of the use of implants – to one year, in order to limit the risk of losing infections with late onset. On the other hand, the studies based on national registries with larger groups of infections occurred between 90 days and 12 months after the surgery show that late infections are not the most dominant.

The aim of the study is to analyze frequency of SAAA after primarily “clean” reconstructive arthroscopies performed during 2-year period at one orthopedic center in Poland experienced in arthroscopy, as well as analysis of demographic data, potential risk factors and accuracy of actual prophylactic procedures against infection.

A comprehensive review of database in two year period (January 2017–December 2018) was performed in search of primarily clean reconstructive arthroscopies. Inclusion criteria: procedures which were performed totally arthroscopically, or the substantial majority of procedure was performed arthroscopically, with or without minimal additional skin incisions needed for harvesting or implantation of a graft or implant (e.g. in anterior cruciate ligament reconstruction). Procedures aimed on extraarticular structures (e.g. bursoscopies), with arthroscopy as an assistance for essentially open procedure (e.g. arthroscopic control of articular surface in open reposition and internal fixation of the bone), with arthroscopy as the minor part of all one-time performed procedures (e.g. arthroscopic repair of the knee cartilage at the same time with high tibia osteotomy), or “non-reconstructive” (e.g. arthroscopic arthrodesis, arthroscopy after total knee replacement) were excluded.

Inclusion criteria fulfilled 853 cases; 59 cases were lost from follow-up: 4 persons refused to be interviewed and in 55 cases we could not reach the contact with the patients. Finally 794 cases have been analyzed: 644 knee, 65 shoulder, 52 hip, 22 ankle, 6 wrist, 3 MTP and 2 elbow arthroscopies. There were 665 primary and 129 revision procedures. The cohort consisted of 520 males and 274 females with age of 7 to 79 years (mean: 37). Primary surgery stated 83.8% of all procedures and a non-resorbable implant have been used in 55.8% of cases. Operation time ranged from 15 minutes to 5 hours and 35 minutes (mean: 1 hour and 25 minutes). An intraarticular drain have been used in 85.3% of all operations. Patient's Body Mass Index (BMI) ranged from 14 to 45.44 (mean: 25.3) and hospital stay was 1–8 days (mean: 2.12).

All patients were prepared to the operation according the same protocol. Prior to the surgery they took a shower with the use of the antimicrobial soap with octenidine dihydrochloride. Hair removal with a clipper was performed only if absolutely necessary. Skin disinfection of operated joint have been done with alcohol-based antiseptic solution in the operation theatre [15]. The antibiotic prophylaxis was administrated to every patient according to the internal protocol of perioperative prophylaxis based on national and international recommendations [12, 15, 17, 18, 19]. The patients received 1 g cefazolin intravenously (or 2 g for patients with weight over 80 kg) 30 minutes prior to clamping the tourniquet (if used) and skin incision. In case of the use of non-absorbable implant during the operation (55.8% of patients), the antibiotic prophylaxis was prolonged to 24 hours with two additional doses of cefazolin 1 g - 8 and 16 hours after the first dose.

All operations were performed by eight different but experienced surgeons, in operation theatre with vertical laminar flow with the capacity to supply about 50 air exchanges per hour.

Overnight hospital stay after the most number of arthroscopies in our clinic is standard.

The diagnosis of SAAA was established in every case by the operating surgeon based on medical history, clinical findings, laboratory tests, bacteriological cultures and imaging.

Time of observation for occurrence of SAAA was determined by the use or not of an non-absorbable implant. We assumed SAAA to be related to procedure if occurred within 30 days after standard arthroscopy or up to one year in case of the use of nonresorbable implant [15]. Patient with lacking documented postoperative control for enough time, but meeting inclusion criteria were interviewed by the phone.

Demographic and medical data about potential risk factors of SAAA were collected [3, 9, 11, 13]. These data include: patients age, sex, operated joint, type of procedure, primary or revision procedure, the use of drains, usage of non-absorbable implant, time of surgery, BMI, time lapse from index operation to diagnosis of SAAA, length of hospital stay, causative microorganisms. Demographic and medical details are presented in Table 1.

The statistical analysis of collected data was performed using IBM SPSS Statistics version 25 software. The descriptive list headers, Kolmogorov-Smirnov display normality and a series of t-tests for independent samples were made. The statistical significance level was established on α = 0.05. The study was approved by the responsible authorities of the institution (department) where the work has been carried out. Submission of presented statistical study for publication does not require approval by local Ethical Committee.

From 794 analyzed cases, an infection occurred in 4 cases: 3 after knee and one after shoulder arthroscopy. The infected cases were: 3 infections after revision knee arthroscopy (2 after arthrolysis, one after revision ACL reconstruction) and one after primary shoulder procedure (rotator cuff repair). Two infections were early and two with late onset (i.e. over 30 days). The incidence of SAAA was 0,5% among all arthroscopies and 0.47% among knee arthroscopies. All cases are presented in Table 2. Analysis of risk factors in non-infected and infected cases showed that the patients age, BMI and time of hospital stay were higher in infected cases, but these differences were not statistically significant (Table 3). In infected knee group only time of hospital stay was significantly longer compared to non-infected cases, whereas patients age, BMI and time of surgery were not significantly different (Table 4). Compared 129 revision procedures to 665 no-revision arthroscopies – patients age, operation time and tie of hospital stay was longer - all differences were statistically significant (Table 5).

Incidence of infections complicating “clean” elective reconstructive joint arthroscopic procedures in our clinic is similar to the reported in the literature i.e. 0.5% among all procedures and 0.47% among knee arthroscopies [4, 5, 6, 11, 12, 16, 19]. It proves that the arthroscopies are safe and predictable operations and the applied perioperative preventive procedures used in our clinic against infection are effective.

We focused on most measurable risk factors of postoperative infection as patients BMI, age, time of surgery and length of stay in hospital. We have not observed correlation between the patient's age and infection, what was proved in another studies [10, 11, 16, 19] but instead we observed a higher prevalence of SAAA in younger patients undergoing revision procedures comparing to the primary operations. BMI is another measurable parameter assessing general patient's health. The average BMI of our patients were about 25.3. Earlier studies have not indicate increased risk of post-arthroscopic infection among patients with overweight [9, 10, 16].

Obesity is increasing this risk of SAAA, especially in the knee and elbow arthroscopies [9, 10, 11, 16]. Probably a small number of patients with obesity in our group was not enough for expose this relation. Unfortunately our clinics data are lacking information about comorbidities, tobacco use and alcohol what could give valuable information [9, 10, 11, 16].

Studies based on large groups of patient prove that the risk of SAAA is increasing with time of the surgery [13, 19] which is not visible in our group. But there is visible that revision procedures take more time what is expected with growing level of complexity of the procedure. Among knee arthroscopies all 3 confirmed infections occurred in patients with history of at least one earlier arthroscopy in this joint. There is not clear consensus in the literature on influence of earlier arthroscopies on the risk of infection in revision arthroscopy in the knee, but this have been proved for shoulder [16] and ACL reconstructions [20].

We hypothesized that failure of previous treatment could be caused by persistent low-grade infection, or the prior contact with health-care system resulting in skin colonization. These hypothesis needs to be investigated in further studies. Another fact is that in the 2 of 3 infection in revised knees the procedures were arthrolysis. Removal of scars inside the joint results in greater wound surface and the need to keeping drain. We noted the fact of using drain, but there were no information of exact time that drains were kept and the volume of exudates from the joint.

The assessment of statistical correlation between prolonged hospital stay and occurrence of SAAA needs to be verified as pure “statistical” result. In our clinic overnight hospital stay after the most number of arthroscopies is standard, whereas in the literature based on American experience this condition is treated as an adverse effect, and its risk increases with the prolonged time of operation. But these studies do not notice the exact relation between overnight (or longer) stay and higher risk of infections [13, 14]. The infection itself usually causes prolonged stay in hospital for observation of patients condition, vital parameters and postoperative course.

The advantage of our single center study is the homogeneity of the group with respect to perioperative procedures as well as constant operating room conditions. All patients were actively followed and all diagnoses of SAAA were confirmed by the operating surgeons in charge. In scope of the lack of national registry of postoperative infections and access to these data from insurances, control visit in outpatient clinic or direct phone contact were the only way to confirm the diagnosis of infection with late onset. These details could not be achieved in multi-center studies with end-point defined as revision operation.

The limitation of our study is small sample size of patients that could be actively controlled in details especially among “small” joints arthroscopies. Other limitations are retrospective nature of the study, only one hospital records and lack of detailed general health status data like smoking, alcohol consumption, comorbidities, any hospital stay during 6 months before index operations.

Primarily aseptic arthroscopic procedures performed with respect to actual perioperative preventive measures have a low risk of postoperative septic arthritis. Infections occurred in 0.5% of all arthroscopy procedures and in 0.47% of knee arthroscopies alone performed in operation theatre with vertical laminar flow with the capacity of 50 air exchanges per hour and perioperative antibiotic prophylaxis consisted of one preoperative dose 1 g cefazolin for procedures without non-resorbable implant and 24 hours antibiotic prophylaxis for operations with implantation of non-resorbable implants. The risk of postoperative septic arthritis increases with patients age, BMI and time of operation, but not significantly.