Columbia Orthopedics Faculty Publishes Multicenter Study in Journal of Bone and Joint Surgery
Providers and researchers from the pediatric orthopedics team at Columbia Orthopedics published a multicenter study in the Journal of Bone and Joint Surgery. The study, “A Clinical Risk Model for Surgical Site Infection Following Pediatric Spine Deformity Surgery,” evaluated a wide range of factors to develop and validate a prediction model for assessing surgical site infection (SSI) risk in individual pediatric patients with spinal deformity.
Collaborators included Benjamin D. Roye, MD, MPH, Michael G. Vitale, MD, MPH, and researcher Hiroko Matsumoto, PhD, along with researchers from Mailman School of Public Health, Columbia School of Nursing, Boston Children’s Hospital, Children’s Hospital Los Angeles, and Children’s Hospital of Philadelphia.
Surgical site infection (SSI) following pediatric spinal deformity surgery remains a serious issue that impacts patients and their families—as well as physicians, hospitals, and other spine surgery stakeholders. SSI is associated with prolonged hospital readmissions, additional procedures, and long-term antibiotic treatments. Beyond the stress and challenges these outcomes pose for patients, they also carry significant financial costs.
Consensus-based guidelines developed by the CDC were designed to help limit SSI risk for pediatric spine surgery for high-risk and early-onset scoliosis. However, prevention practices are not uniform, and SSI incidence rates reflect that inconsistency; in the U.S., reported SSI risk associated with pediatric spinal deformity surgery ranges from 0% to 27.2%.
Because SSI risk is impacted by conditions specific to each patient, applying prediction modeling can be used to better assess a patient’s individual SSI risk status. Previous SSI research approaches have typically focused on causal inference; these studies were limited by their inability to investigate multiple risk factors simultaneously. By including a wide range of pre- and intraoperative factors, prediction modeling considers an individual patient’s variables for a more complete picture.
The multicenter retrospective cohort study evaluated data for 3,092 pediatric spinal deformity surgeries at seven hospitals. Variables included a wide range of pre- and intraoperative risk factors, including 31 patient characteristics, 12 surgical factors, and four hospital factors. Cases were determined using the CDC definition for SSI, which includes an incidence of SSI within 90 days of surgery and several identifying symptoms, such as purulent drainage, positive cultures, pain or tenderness, localized swelling, fever, and other evidence of infection.
Researchers created multiple models using logistic regression for predictor selection. The data set was randomly split into training and testing sets and employed fivefold cross-validation to look for discrimination, calibration, and overfitting. Predictors used in the final model were nonambulatory status, neuromuscular etiology, pelvic instrumentation, a procedure time greater than 7 hours, an American Society of Anesthesiologists grade greater than 2, revision surgery, the number of hospital spine surgical cases per year, abnormal hemoglobin levels, and an overweight or obese body mass index.
The final model adequately predicted SSI risk and identified important risk factors to explore further, such as length of surgery and institutional surgical volume. The model was used to develop an SSI risk-probability calculator and a mobile device application to assist health care providers with a real-time estimation of the probability of SSI for individuals. By identifying high-risk patients, hospitals can potentially target resources to help better manage SSI risk.
View the full article here.