Study design
This was a single-centre, retrospective comparative study approved by the Institutional Review Board (HKU/HA HKW IRB) (Reference no. UW19-706) in a tertiary referral university hospital.
Patients, who underwent revision TKA or THA at our institution from January 2016 to October 2019, were included in this study. The conventional dressing used traditionally was Cosmopor® E (Hartmann), which is a sterile, adhesive wound dressing made of soft, non-woven polyester. Since January 2018, the surgical team changed the dressing to ciNPWT, which were either Prevena Incision Management System (Acelity, San Antonio, TX) or PICO (Single use ciNPWT system, Smith & Nephew, Hull, UK) (Fig. 1). The change of practice in wound dressing provided us with 2 naturally formed cohort groups: those patients who underwent surgery between January 2016 to December 2017 received a conventional dressing, whereas those patients who underwent surgery between Jan 2018 to October 2019 received ciNPWT. All patients were treated by the same surgical team consisting of four fellowship-trained surgeons, who performed revision surgery with similar surgical techniques.
The patients in both cohorts were managed according to the same perioperative protocols with the difference mainly in the dressing material used at the end of the surgical procedures. Prophylactic antibiotics, single dose of 1 g cefazolin, were given 30-min before the surgical incision. For revision surgery due to aseptic causes e.g. aseptic loosening or periprosthetic fracture, cefazolin would only be used for 24-h after the operation. For revision surgery because of infection, 2-stage revision surgery would be performed. The perioperative antibiotic used was according to the recommendation by clinical microbiologist based on the antibiotic sensitivity profile of the bacteria. To decrease perioperative blood loss, Intravenous tranexamic acid was given at a dosage of 15 mg/kg before induction of anaesthesia, and another intravenous injection of the same dosage would be given 4-h afterwards. For revision TKA, tourniquet was used throughout the surgical invasive procedures. Antibiotic-loaded cement (PALACOS® + G, Heraeus), was used for all cemented prosthesis. Allogenic blood transfusion was prescribed if postoperative haemoglobin drop to ≤ 7 g/dL in hemodynamically stable adults, or ≤ 8 g/dL in patients who had underly cardiovascular or respiratory comorbidities. No surgical drainage was used. All wounds were closed in layers with barbed sutures (Stratafix, Depuy Synthes).
For the postoperative wound surveillance, the dressing in the ciNPWT cohort would be kept intact for 7 days, and then changed to conventional dressing, whereas the wound condition in the conventional dressing cohort was reviewed at postoperative day 3, and then a new piece of conventional dressing would be re-applied to the surgical wound. The conventional dressing used in both cohorts would be kept intact and removed on postoperative day 14.
All the patient’s medical records and perioperative parameters were reviewed. Patient demographics, the type of revision arthroplasty, indications for surgery, mode of anaesthesia, American Society of Anaesthesiologist (ASA) grade and operative duration were recorded. Patients’ preoperative status were also documented, including haemoglobin and albumin level, comorbidities and risk factors for wound complications. Clinical photos were taken at the surgical sites with a high-resolution digital camera, and the wound conditions were documented in the wound surveillance chart by a nursing specialist, who specialized in managing surgical wounds in the orthopaedic department. The nurse was in charge of the documentation in the wound surveillance chart, documentation of the postoperative surgical wound conditions, including both septic and aseptic wound complications, during the course of recovery in the in-patient and out-patient periods. Septic complications, such as surgical site infections (SSI) and prosthetic joint infection (PJI) were recorded. SSI, including both superficial incisional SSI and deep incisional SSI, were defined according to the criteria by Centers for Disease Control (CDC) [20], while prosthetic joint infection (PJI) was defined according to the criteria provided by Musculoskeletal Infection Society (MSIS) [21]. Aseptic wound complications, such as persistent drainage, haematoma formation, wound dehiscence, suture granuloma, blister formation, and maceration of the wound, were also documented. The patient would be arranged to have wound surveillance in the out-patient follow-up at 2-week, 6-week, 3-month, 6-month and yearly after discharge from hospital. Data, including wound complications, re-operations due to wound complications of other causes, readmission within 30 days of surgery, periprosthetic joint infection, 90-day perioperative and mortality, were retrieved from the medical record system at our institution.
The primary outcomes were (1) the overall incidence of septic and aseptic wound complications, and (2) the incidence of septic and (3) the incidence of aseptic wound complications. The secondary outcomes include the length of hospital stay, re-operations due to wound complications, 90-day postoperative mortality, and readmission within 30 days of surgery. All the outcome measures were reviewed by two orthopaedic specialists individually, and the cases would be reviewed together so as to make a consensus in case a discrepancy in the individual assessments.
Statistical analysis
The statistical analysis was performed using IBM SPSS 26.0 software (IBM, Armonk, NY, USA), and the statistically significance took place at the 5% significance level. The primary and secondary outcome measures was compared. To control for confounding factors, the baseline data between the two cohorts, including patient demographics, existing comorbidities, types of revision arthroplasty performed, indications for revision arthroplasty, ASA grade, mode of anaesthesia, operation duration, as well as preoperative and postoperative levels of albumin and haemoglobin, were also compared.
The choice of statistical tests based on the data distribution, and the nature of the data (nominal, ordinal or interval/ratio). The independent samples t test was used for parametric data, while the chi-square or Fisher exact test was used only non-parametric data and categorical data depending on the observed frequency. All available data were incorporated in data listings and tabulations. No imputation of values for missing data were performed.