Septic arthritis is a joint-threatening and potentially life-threatening condition. The knee is the most common joint involved1 and the incidence is increasing2. Treatment requires emergency surgery, irrigation of the joint, and intravenous antibiotics. Surgical irrigation removes debris and decreases the intra-articular microbial burden to an extent unachievable by needle aspiration alone3,4. Surgical debridement can be performed either arthroscopically or via arthrotomy. Few studies have examined whether arthroscopic or open treatment is more successful for septic arthritis of the knee5-8. To our knowledge, there have been no large studies with well-matched groups for comparison of arthroscopic and open treatment of acute native knee septic arthritis alone. In addition, to our knowledge, no study has performed a multivariable analysis of factors influencing treatment success specifically for native knee septic arthritis. The purpose of this study was to compare the treatment efficacy of arthroscopic and open methods of treatment for acute native knee septic arthritis in a retrospective comparative cohort and to analyze factors that may influence treatment success.
Materials and Methods
Data from all patients with acute native knee septic arthritis who had arthroscopic or open irrigation of their knee joint, between January 2000 and July 2015, were collected retrospectively. Patients were identified on the basis of diagnosis coding in our institution’s medical information system. All patients with closed physes around the knee were included. All etiologies were included in our study. Patients who did not fulfill the definition of septic arthritis described below and had one sterile reactive, crystalline, or inflammatory arthropathy were excluded. Patients who underwent initial debridement in another hospital and then transferred to our institution were excluded. The study was approved by the institution’s ethics committee.
Clinical presentation, etiology, and risk factors were collated. Laboratory results, arthrocentesis analyses, and microbiological findings were assessed, as well as knee radiographs for osteoarthritis according to the Kellgren and Lawrence scale9. The surgical procedure type and operative details were recorded.
The primary outcome was the need to return to the operating room. The secondary outcomes were the total number of operations required, the range of motion assessed postoperatively, the length of inpatient stay, and mortality.
The diagnosis of septic arthritis was determined and was classified by Newman’s criteria10 and was supported by patients’ symptoms and signs. Classical clinical findings were knee pain and swelling, fever, joint irritability, warmth, erythema, effusion, and reduced range of motion. Laboratory findings predictably demonstrated an elevated C-reactive protein (CRP) level, leukocytosis, and an elevated white blood-cell count from arthrocentesis samples. According to the Newman grading, we included patients who had a positive joint sample culture (Newman grade A); those who had a positive associated sample culture, for example, blood culture (Newman grade B); and those who had negative joint and blood cultures but had intraoperative pus or turbid fluid or histological evidence of septic arthritis (Newman grade C)10.
Patients were treated by arthroscopy or arthrotomy according to the individual orthopaedic surgeon’s decision. The surgical procedure was normally performed on the basis of clinical and laboratory findings before knowing culture results. The proportion of patients treated by arthroscopic debridement did not increase during the study period.
The surgical procedure was typically performed with the patient under general anesthesia. Sterile preparation and draping were performed using povidone-iodine or chlorhexidine. An arthrotomy was made anteromedially or anterolaterally. In 10 cases, this followed an initial arthroscopic assessment. After arthrotomy, the infected material was sampled and was sent for culture. The joint was thoroughly irrigated with normal saline solution until the fluid returning from the joint was clear. Infected tissue was debrided as necessary. The arthrotomy was closed, typically over a drain within the joint, and was dressed.
The surgical procedure was typically performed with the patient under general anesthesia. Sterile preparation and draping of the patient were performed in the same manner as for open procedures. An anterolateral portal was most often created first followed by an anteromedial portal. In some cases, a superolateral portal was also utilized to enter the suprapatellar pouch. Infected joint material was sent for culture. The joint was thoroughly irrigated with normal saline solution until the fluid returning from the knee was clear. Debridement of fibrinous debris or other purulent material was performed as needed with an arthroscopic powered shaver. A drain was left within the joint, and portals were closed and were dressed.
Postoperatively, patients were routinely seen by the physiotherapist, were allowed to bear weight as tolerated with a walking aid (e.g., crutches), and commenced range-of-motion exercises approximately 48 hours postoperatively. Drains were removed after 24 to 72 hours. Antibiotics were continued intravenously, most commonly 2-g flucloxacillin every 6 hours, and then were tailored to the microorganism’s sensitivities, normally in consultation with an infectious disease physician. Patients were monitored clinically and via laboratory results. Failure to improve clinically, or deterioration, such as fevers or elevated CRP, prompted repeat surgical irrigation. After the last surgical procedure, antibiotics were routinely administered intravenously for 4 to 6 weeks. For most patients, part of the 4 to 6 weeks occurred through our institution’s home intravenous antibiotic program.
Categorical data were expressed as frequencies and proportions with 95% confidence intervals (CIs) and were tested with the Fisher exact test (two-sided). All continuous variables reported were non-normal and were summarized using medians and semi-interquartile ranges and were tested using the Wilcoxon rank-sum test.
The effect of the initial procedure on the primary outcome of return to the operating room taking into account all other measured factors was evaluated using a multivariable logistic model. The model was built according to the 7-step approach for purposeful selection of covariates by Hosmer et al.11. This produced a stable model that was simultaneously parsimonious on likelihood ratio testing, efficient by Bayesian information criteria12, and an excellent fit by the Hosmer-Lemeshow goodness-of-fit statistic13. Sex and age, although not significant, were returned to the model to produce effect estimates controlled for these factors.
Our study has 80% power to detect an odds ratio (OR) of 2.94 and 90% power to detect an OR of 3.47 at a significance of 0.05.
Between 2000 and 2015, 203 patients with closed physes were surgically treated by either arthroscopic or open methods for suspected knee septic arthritis. Twenty patients were excluded as they already had undergone an irrigation procedure at another hospital. Ten patients did not meet the Newman septic arthritis criteria and were concluded to have sterile or reactive arthritis. Five patients had crystal arthropathy only. Five patients were excluded as they had chronic or repeated episodes of septic knee arthritis, including 1 with a sinus. Two patients were excluded because of insufficient records. Subsequently, 161 patients (166 knees) with acute native knee septic arthritis surgically treated exclusively at our institution were included; 119 patients (74%) were managed by arthroscopic irrigation and 42 patients (26%) were treated by open methods.
Patient Presentation, Comorbidities, and Etiology
Patient demographic characteristics and clinical presentation are presented in Table I. Their comorbidities, risk factors, and etiology of septic arthritis are summarized in Tables II and III. The median age was 57.5 years in the arthroscopic treatment group and 65.8 years in the open treatment group (p = 0.644). The male-to-female ratio was approximately 2:1 for both groups. Most cases were isolated septic knee arthritis. Four cases in the arthroscopic treatment group and 1 case in the open treatment group were bilateral. Six patients in the arthroscopic treatment group had distant joint involvement (5 shoulders and 1 wrist). Most patients were afebrile at presentation; however, 4% had circulatory shock. The majority of patients in both groups presented with a painful, swollen knee with reduced range of motion. The median symptom duration prior to the surgical procedure was 4 days in both groups. Sixty-two percent of patients in the arthroscopic treatment group and 76% of patients in the open treatment group had ≥1 comorbidity. Smoking was the commonest risk factor, followed by crystal arthropathy, previous ipsilateral knee surgical procedure, and diabetes.
Preoperative Laboratory Results
Laboratory results from blood analysis and arthrocentesis are summarized in Table IV. The median arthrocentesis white blood-cell count was elevated similarly in both groups at 75,800 × 106/L for the arthroscopic procedure compared with 67,600 × 106/L for the open procedure (p = 0.980). The median polymorphonuclear cell percentage in the arthrocentesis samples was 90% in both groups. The CRP level was similarly markedly elevated in both groups, with a median of 167 mg/L in the arthroscopic treatment group and 205 mg/L in the open treatment group (p = 0.731). The median white blood-cell count was elevated with a predominant neutrophilia demonstrated in both groups.
Preoperative Radiographic Results
Preoperative radiographs were available for 89 of 119 patients (92 of 123 knees) in the arthroscopic treatment group and for 36 of 42 patients (37 of 43 knees) in the open treatment group. Clinically important osteoarthritis (Kellgren and Lawrence grade of ≥2) was present in 44% of knees in the arthroscopic treatment group and 58% of knees in the open treatment group (p = 0.17). No radiographs demonstrated osteomyelitis. One patient in the arthroscopic treatment group was subsequently found to have osteomyelitis according to a bone biopsy sample taken at the second surgical procedure.
The microbiology results from knee joint samples are summarized in Figure 1. One hundred and thirty-eight cases (83.1%) were joint culture-positive and 28 cases (16.9%) were joint culture-negative. The causative organism was isolated in 81% of arthroscopic cases and 90% of open cases. Antibiotics were commenced preoperatively for 61% of patients treated arthroscopically and 60% of patients in the open treatment group. Invariably, culture-positive cases were monomicrobial; only 3 cases in the arthroscopic treatment group were polymicrobial.
Fifty-nine (50%) of 119 patients were successfully treated after 1 arthroscopic procedure compared with 12 (29%) of 42 patients in the open treatment group (p = 0.020). The second procedure was typically the same as the index procedure (Fig. 2). The median time between first and second procedures was 3.0 days in both groups. The third and subsequent procedure types varied. To allow a comparison of the success rates of patients treated by a purely arthroscopic method and those treated by a purely open method, those few patients whose subsequent procedure was not the same as the index procedure were excluded for comparison of second and third irrigation success rates. After a second procedure, the success rate remained greater in the arthroscopic treatment group (p = 0.019). This difference persisted following a third procedure (p = 0.011) (Fig. 3). Overall, the mean total number of irrigation procedures required (and standard deviation) was significantly fewer in the arthroscopic treatment group at 1.79 ± 0.96 than in the open treatment group at 2.42 ± 1.5 (p = 0.010).
Range of Motion, Length of Stay, and Mortality
Range of motion was recorded after the final surgical procedure. Postoperative range-of-motion recordings were available for 125 knees (94 in the arthroscopic treatment group and 31 in the open treatment group). Range of motion was recorded by multiple observers over the study period using goniometers or clinical examination. After excluding patients from the arthroscopic group who subsequently had open procedures, the mean range of motion was compared. The median time to range-of-motion recording after the surgical procedure was 13 days (range, 2 to 159 days) in the arthroscopic treatment group and 23 days (range, 3 to 158 days) in the open treatment group. The median knee range of motion was significantly greater (p = 0.016) in the arthroscopic treatment group at 90° (semi-interquartile range, 10°) than in the open treatment group at 70° (semi-interquartile range, 25.5°). The median length of stay was 14 days (range, 3 to 159 days) in the arthroscopic treatment group and 19.5 days (range, 2 to 183 days) in the open treatment group (p = 0.088). The patients in each group with the maximal length of stay had medical reasons unrelated to their septic arthritis. The mortality rate was greater in the open treatment group (9.5%) compared with the arthroscopic treatment group (4.1%), but the difference was not significant (p = 0.24). Five mortalities were due to sepsis related to septic arthritis (all-patient mortality rate, 3%), 2 were due to pneumonia, 1 was due to urosepsis, and 1 was secondary to renal failure.
On multivariable analysis, patients treated by open irrigation were 2.6 times more likely to require a further operation than those treated arthroscopically (OR, 2.56 [95% CI, 1.1 to 5.9]; p = 0.027) (Table V). The only other significant factors remaining in the model were a positive joint culture (OR, 5.78 [95% CI, 2.12 to 15.80]; p = 0.001) and the presence of at least two comorbidities (OR, 0.32 [95% CI, 0.12 to 0.80]; p = 0.015).
Septic arthritis is a relatively common orthopaedic emergency with a reported annual incidence of 9.2 cases per 100,000 population1. The knee is the most common joint affected and has an annual incidence of 5 per 100,000 population1. Septic knee arthritis results in impaired function and has been linked to the severity of septic arthritis previously14.
The clinical presentation, mean symptom duration15, elevated inflammatory markers, and raised white blood-cell count on arthrocentesis in our patients agree with those in previous studies5,6. The majority of our patients also had ≥1 septic arthritis risk factor, consistent with the literature16. The etiology in the current study more often was hematogenous spread rather than a direct breach of the soft tissues of the knee, correlating with previously reported causes17,18. Staphylococcus aureus was the most common organism responsible in our study, typical of native knee septic arthritis7,19. In 16.9%, an organism was not cultured. This culture-negative rate is consistent with reported rates, which vary from 7.5% to 52.4%7,8,20,21.
Treatment groups were well matched for demographic characteristics, risk factors, symptom duration, etiology, inflammatory markers, arthrocentesis results, and causative organisms, facilitating a comparison of treatment methods without confounding variables.
When comparing the efficacy of the 2 treatment methods, the arthroscopic treatment group had a significantly higher initial treatment success rate. The difference remained significant after controlling for other variables on multivariable analysis. The cumulative success rate following arthroscopic irrigation remained significantly greater for those patients who required a second irrigation procedure and again following a third procedure, a finding that has not been previously demonstrated, to our knowledge. Supporting this, the total number of irrigation procedures required for successful treatment was significantly fewer in the arthroscopic treatment group. Furthermore, our findings were not explained by the proportion of arthroscopic procedures over time. Greater treatment success with arthroscopic procedures5 and the need for repeat procedures have been reported15. However, those studies did not use multivariable analysis to adjust for the effect of potential confounders. We found after multivariable analysis that arthroscopic treatment was independently associated with a greater treatment success rate.
The reasons for greater success rates after arthroscopic treatment may only be hypothesized from our study. Potentially, a larger wound may create greater local tissue injury and may provide a greater infection nidus contaminated by spread of organism-harboring debris. Another reason is that by keeping the joint relatively closed during arthroscopic irrigation, the fluid may accumulate and may more thoroughly irrigate the entire joint space compared with an open procedure, in which areas may inadvertently escape thorough irrigation. As this study was nonrandomized, there was potential selection bias between groups for infection severity. However, there was no significant difference between the number of shocked or febrile patients, inflammatory markers, and white blood-cell count or arthrocentesis results between groups. Adding each of these factors as covariates in the multivariable model did not cause a significant change to the result.
The posttreatment range of motion was significantly greater in the arthroscopic treatment group. This agrees with previous reports5,6, and good function following arthroscopic treatment of septic arthritis is a major advantage of such treatment15. Indeed, in comparison, knee function after arthrotomy for septic arthritis has been found to decline22.
The length of stay was considerable in both groups and was between the previously reported 2 to 6 weeks5,6. The median length of stay was 5.5 days fewer in the arthroscopic group, but this did not reach significance. Our findings support studies showing shorter periods of hospitalization after arthroscopic treatment5,6.
To our knowledge, mortality rates have not been previously compared following arthroscopic or open treatment for the septic knee. The rate was similar when comparing such treatments for the wrist23. In our study, mortality was lower in the arthroscopic treatment group at 4.1% compared with 9.5% in the open treatment group, but this was not significant. These findings support cohort studies showing mortality of 3% to 11% for native knee septic arthritis24,25.
The majority of evidence for the management of septic arthritis of large joints has been Level IV or case series only. To our knowledge, little research for individual native joints, including the hip26 and wrist23, and no research with regard to native shoulders has compared open and arthroscopic treatment. Furthermore, few studies have compared arthroscopic and open treatment for native knee septic arthritis, all with notably fewer patients than this present study, and none have used a multivariable analysis or compared success rates after multiple procedures5-8. One study on septic arthritis of multiple joints completed a multivariable analysis and found inflammatory arthropathy, large joint involvement, high synovial fluid polymorphonuclear cell count, S. aureus, and diabetes to be independent risk factors for failure of a single surgical debridement27. In our analysis, we found two key independent risk factors for treatment failure: open irrigation and a positive joint culture. A positive joint culture may be important because those cases with culture-positive results may have been more likely to have had higher bacterial loads and hence required subsequent procedures compared with those that had culture-negative results.
The limitations of the current study should be acknowledged. The study was retrospective and so patients were not randomized. However, to our knowledge, our study has been the largest comparison to date with groups that were not significantly different in any parameter recorded and the only study of septic knees to include multivariable analysis. Range of motion was recorded using both goniometers and clinical examination by multiple observers as the study was retrospective and covered more than 15 years, rather than a single observer using a single goniometer. As our institution is a Level-1 tertiary referral center, many patients were lost to follow-up because they were referred from a large area. Also, we did not report long-term outcomes in our study. It would have been of interest to compare the outcomes of the groups at a long-term follow-up.
In conclusion, arthroscopic treatment for acute native knee septic arthritis was more successful as an initial treatment and subsequent treatment, required fewer total irrigation procedures, and resulted in better range of motion compared with open treatment. Furthermore, the success of arthroscopic treatment has been confirmed by multivariable analysis.
Investigation performed at the Bone and Joint Institute and the Department of Immunology and Infectious Diseases, Royal Newcastle Centre and John Hunter Hospital, Newcastle, Australia
Disclosure: There was no funding source for this study. The Disclosure of Potential Conflicts of Interest forms are provided with the online version of the article (http://links.lww.com/JBJS/A142).
1. Morgan DS, Fisher D, Merianos A, Currie BJ. An 18 year clinical review of septic arthritis from tropical Australia. Epidemiol Infect. 1996 ;117(3):423–8.
2. Okano T, Enokida M, Otsuki R, Hagino H, Teshima R. Recent trends in adult-onset septic arthritis of the knee and hip: retrospective analysis of patients treated during the past 50 years. J Infect Chemother. 2011 ;17(5):666–70. Epub 2011 May 17.
3. Lane JG, Falahee MH, Wojtys EM, Hankin FM, Kaufer H. Pyarthrosis of the knee. Treatment considerations. Clin Orthop Relat Res. 1990 ;252:198–204.
4. Sharff KA, Richards EP, Townes JM. Clinical management of septic arthritis. Curr Rheumatol Rep. 2013 ;15(6):332.
5. Böhler C, Dragana M, Puchner S, Windhager R, Holinka J. Treatment of septic arthritis of the knee: a comparison between arthroscopy and arthrotomy. Knee Surg Sports Traumatol Arthrosc. 2015 . [Epub ahead of print].
6. Wirtz DC, Marth M, Miltner O, Schneider U, Zilkens KW. Septic arthritis of the knee in adults: treatment by arthroscopy or arthrotomy. Int Orthop. 2001;25(4):239–41.
7. Balabaud L, Gaudias J, Boeri C, Jenny JY, Kehr P. Results of treatment of septic knee arthritis: a retrospective series of 40 cases. Knee Surg Sports Traumatol Arthrosc. 2007 ;15(4):387–92. Epub 2006 Dec 6.
8. Peres LR, Marchitto RO, Pereira GS, Yoshino FS, de Castro Fernandes M, Matsumoto MH. Arthrotomy versus arthroscopy in the treatment of septic arthritis of the knee in adults: a randomized clinical trial. Knee Surg Sports Traumatol Arthrosc. 2015 . [Epub ahead of print].
9. Kellgren JH, Lawrence JS. Radiological assessment of osteo-arthrosis. Ann Rheum Dis. 1957 ;16(4):494–502.
10. Newman JH. Review of septic arthritis throughout the antibiotic era. Ann Rheum Dis. 1976 ;35(3):198–205.
11. Hosmer DW Jr, Lemeshow SA, Sturdivant RX. Applied logistic regression. 3rd ed. Hoboken: Wiley; 2013.
12. Schwarz G. Estimating the dimension of a model. Ann Stat. 1978;6(2):461–4.
13. Lemeshow S, Hosmer DW Jr. A review of goodness of fit statistics for use in the development of logistic regression models. Am J Epidemiol. 1982 ;115(1):92–106.
14. Yanmış I, Ozkan H, Koca K, Kılınçoğlu V, Bek D, Tunay S. The relation between the arthroscopic findings and functional outcomes in patients with septic arthritis of the knee joint, treated with arthroscopic debridement and irrigation. Acta Orthop Traumatol Turc. 2011;45(2):94–9.
15. Aïm F, Delambre J, Bauer T, Hardy P. Efficacy of arthroscopic treatment for resolving infection in septic arthritis of native joints. Orthop Traumatol Surg Res. 2015 ;101(1):61–4. Epub 2015 Jan 23.
16. Mathews CJ, Coakley G. Septic arthritis: current diagnostic and therapeutic algorithm. Curr Opin Rheumatol. 2008 ;20(4):457–62.
17. Studahl M, Bergman B, Kälebo P, Lindberg J. Septic arthritis of the knee: a 10-year review and long-term follow-up using a new scoring system. Scand J Infect Dis. 1994;26(1):85–93.
18. Stutz G, Kuster MS, Kleinstück F, Gächter A. Arthroscopic management of septic arthritis: stages of infection and results. Knee Surg Sports Traumatol Arthrosc. 2000;8(5):270–4.
19. Smith MJ. Arthroscopic treatment of the septic knee. Arthroscopy. 1986;2(1):30–4.
20. Thiery JA. Arthroscopic drainage in septic arthritides of the knee: a multicenter study. Arthroscopy. 1989;5(1):65–9.
21. Tzurbakis M, Fotopoulos VC, Mouzopoulos G, Fotopoulos VK, Georgilas I, Stathis E. Arthroscopic debridement of acute septic arthritis of the knee. Orthop Proceed. 2009;91-B(SUPP II):307.
22. Sreenivas T, Nataraj AR, Menon J. Acute hematogenous septic arthritis of the knee in adults. Eur J Orthop Surg Traumatol. 2013 ;23(7):803–7. Epub 2012 Aug 30.
23. Sammer DM, Shin AY. Comparison of arthroscopic and open treatment of septic arthritis of the wrist. J Bone Joint Surg Am. 2009 ;91(6):1387–93.
24. Gupta MN, Sturrock RD, Field M. A prospective 2-year study of 75 patients with adult-onset septic arthritis. Rheumatology (Oxford). 2001 ;40(1):24–30.
25. Helito CP, Noffs GG, Pecora JR, Gobbi RG, Tirico LE, Lima AL, de Oliveira PR, Camanho GL. Epidemiology of septic arthritis of the knee at Hospital das Clínicas, Universidade de São Paulo. Braz J Infect Dis. 2014 ;18(1):28–33. Epub 2013 Sep 9.
26. El-Sayed AM. Treatment of early septic arthritis of the hip in children: comparison of results of open arthrotomy versus arthroscopic drainage. J Child Orthop. 2008 ;2(3):229–37. Epub 2008 Mar 6.
27. Hunter JG, Gross JM, Dahl JD, Amsdell SL, Gorczyca JT. Risk factors for failure of a single surgical debridement in adults with acute septic arthritis. J Bone Joint Surg Am. 2015 ;97(7):558–64.
Supplemental Digital Content
Copyright 2017 by The Journal of Bone and Joint Surgery, Incorporated