Early complications in spine surgery and relation to preoperative diagnosis: a single-center prospective study

Abstract

The reported incidence of complications in spine surgery varies widely. Variable study methodologies may open differing avenues for potential bias, and unclear definitions of perioperative complication make analysis of the literature challenging. Although numerous studies have examined the morbidity associated with specific procedures or diagnoses, no prospective analysis has evaluated the impact of preoperative diagnosis on overall early morbidity in spine surgery. To accurately assess perioperative morbidity in patients undergoing spine surgery, a prospective analysis of all patients who underwent spine surgery by the neurosurgical service at a large tertiary care center over a 6-month period was conducted. The correlation between preoperative diagnosis and the incidence of postoperative complications was assessed. Data were prospectively collected on 248 consecutive patients undergoing spine surgery performed by the neurosurgical service at the Thomas Jefferson University Hospital from May to December 2008. A standardized definition of minor and major complications was applied to all adverse events occurring within 30 days of surgery. Data on diagnosis, complications, and length of stay were retrospectively assessed using stepwise multivariate analysis. Patients were analyzed by preoperative diagnosis (neoplasm, infection, degenerative disease, trauma) and level of surgery (cervical or thoracolumbar). Total early complication incidence was 53.2%, with a minor complication incidence of 46.4% and a major complication incidence of 21.3%. Preoperative diagnosis correlated only with the occurrence of minor complications in the overall cohort (p = 0.02). In patients undergoing surgery of the thoracolumbar spine, preoperative diagnosis correlated with presence of a complication and the number of complications (p = 0.003). Within this group, patients with preoperative diagnoses of infection and neoplasm were more often affected by isolated and multiple complications (p = 0.05 and p = 0.02, respectively). Surgeries across the cervicothoracic and thoracolumbar junctions were associated with higher incidences of overall complication than cervical or lumbar surgery alone (p = 0.04 and p = 0.03, respectively). Median length of stay was 5 days for patients without a complication. Length of stay was significantly greater for patients with a minor complication (10 days, p < 0.0001) and even greater for patients with a major complication (14 days, p < 0.0001). The incidence of complications found in this prospective analysis is higher than that reported in previous studies. This association may be due to a greater accuracy of record-keeping, absence of recall bias via prospective data collection, high complexity of pathology and surgical approaches, or application of a more liberal definition of what constitutes a complication. Further large-scale prospective studies using clear definitions of complication are necessary to ascertain the true incidence of early postoperative complications in spine surgery.