After obtaining approval of the study protocol from the Institutional Review Board (IRB number 05-2020-007; IRB waived the need for informed consent), we retrospectively reviewed consecutive patients (n=64) with presumptive chronic septic arthritis of the digits who were treated primarily in our hospital or referred to our hospital due to uncontrolled infection between January 2010 and December 2017. One surgeon (the first author) who was level 3 (specialist-experienced) according to Tang’s levels of surgical expertise performed all surgeries [9]. The definite diagnosis of septic arthritis of the digits was made after the surgery as having at least one of the following criteria: (1) growth of microorganisms from samples obtained during the operation, (2) intraoperative pus in the joint, and (3) neutrophils in the microscopic analysis of samples obtained during operation (neutrophils ≥10 in any of the 5 high power (×40) fields) [2, 10-13]. We excluded patients who did not meet any of the criteria of septic arthritis (n=11), patients diagnosed with tuberculous arthritis (n=2), patients diagnosed with rheumatoid arthritis (n=4), and patients with metal implants (n=2). Finally, we included 11 patients treated primarily and 34 patients who were referred to our hospital after initial treatment. Twenty-seven distal interphalangeal joints, 15 proximal interphalangeal joints, 2 thumb interphalangeal joints, and 1 thumb metacarpophalangeal joint were involved. The mean follow-up period was 16 months (range, 11-36 months)
Preoperatively, we obtained information including surgical debridement and empiric antibiotic therapy in the referred patients based on the record of the previous hospital. Both groups underwent history taking of symptom onset, presence of penetrating event, medications associated with systemic immunosuppression (corticosteroid, methotrexate, biologics) [14], immune compromising comorbidities such as diabetes mellitus and chronic kidney disease [15, 16], and risk factors for atypical organism infections (precedent cellulitis, sexual activity, elderly patients with urinary tract infections, intravenous drug abuse, working as a gardener, rheumatoid arthritis, and animal bites) [5]. All patients underwent magnetic resonance imaging (MRI) in addition to simple radiography to detect special conditions, such as combined osteomyelitis and unexpected intraosseous abscess. Combined osteomyelitis was diagnosed if any evidence of definite bone destruction, intramedullary confluent marked low signal intensity, or bone marrow edema with soft tissue ulcer was observed in T1 weighted image (Fig. 1)[17].
Surgical techniques
For interphalangeal joints of the 2nd-5th digits, we applied a finger tourniquet to the digit and excised the skin and soft tissue around a drainage sinus until fresh bleeding. If the skin hole was not sufficient to open the joint capsule or there was no drainage sinus, an additional midaxial approach was made [2]. Debridement of the intraarticular surface and curettage of the devitalized osteocartilaginous fragments were performed through the excised skin opening or the midaxial incision. Nonviable extensor tendon was excised. Intraosseous abscess identified on MRI was treated by curettage through the joint or making additional cortical windows. The wound was left open with a bulky dressing.
For the thumb joints, a volar or a dorsolateral approach was made according to the site of drainage sinus. For the metacarpophalangeal joint, the dorsoulnar incision was made and the joint was opened between the ulnar collateral ligament and the extensor pollicis longus. For the interphalangeal joints, the A2 pulley was cut, and the joint was explored through the volar plate. Incision of the oblique pulley was not attempted to prevent the bowstring effect of the flexor pollicis longus. After debridement of the skin and soft tissue around the drainage sinus, the joint was opened and debrided. The wound was left open with a bulky dressing.
Postoperative care and additional procedures
Dressing change for the open wound was performed daily. If purulent discharge continued, repeated debridement and irrigation with normal saline were performed every 2-3 days until the discharge stopped. Antiseptics, such as octenidine or polyhexanide, were not used to prevent cartilage damage [18, 19]. Soaking or wet dressing was not used. If the soft tissue defect was not healed by the intention of the surrounding tissue, flap surgery was performed to cover the defect.
Empirically, patients without risk factors for atypical organisms received intravenous cefazolin (primary patients, n=10; referred patients, n=32). In three cases of animal bite injury (primary patient, n=1; referred patients, n=2), they received intravenous amoxicillin to cover the Pasteurella species infection [4, 5]. Antibiotics were changed according to the gram stain result, identified microorganism, and susceptibility test. Effective antibiotics were continued for at least 4 weeks after the discharge stopped [20]. If the wound improved and oral form of antibiotics was available, the patient was discharged with oral antibiotics. If not, the patient received intravenous antibiotics continuously. Complete blood cell count, liver panel, and renal panel were checked twice a week. In case of adverse effect, the infectious disease specialist decided whether the causative antibiotic agent should be ceased or changed.
If additional flap surgery was performed, bone fixation with immobilization was applied to protect the flap position for 2 weeks (Fig. 2). Otherwise, active finger ROM exercise inside the bulky dressing was allowed to prevent stiffness.
Clinical data and outcome evaluation
We recorded active ROM of the involved joint and Quick Disabilities of the Arm, Shoulder, and Hand (DASH) score and checked the simple radiographs at 3, 6, and 12 months postoperatively, and annually thereafter. We reviewed the data on the site of involvement, combined osteomyelitis, hospital stay, antibiotic therapy, microorganisms, and surgical intervention in the medical record.
Statistical analysis
We performed the Shapiro-Wilk test for normality. According to the result of the Shapiro-Wilk test, Student’s t-test was used for independent parametric continuous variables and the Mann-Whitney U-test was used for independent non-parametric continuous variables. The Fisher’s exact and chi-squared tests were used to compare discrete variables. Statistical significance was set at p<0.05.