The overall incidence of complications or adverse events in spinal surgery is unknown. Both prospective and retrospective analyses have been performed, but the results have not been critically assessed. Procedures for different regions of the spine (cervical and thoracolumbar) and the incidence of complications for each have been reported but not compared. Authors of previous reports have concentrated on complications in terms of their incidence relevant to healthcare providers: medical versus surgical etiology and the relevance of perioperative complications to perioperative events. Few authors have assessed complication incidence from the patient’s perspective. In this report the authors summarize the spine surgery complications literature and address the effect of study design on reported complication incidence.
Methods. A systematic evidence-based review was completed to identify within the published literature complication rates in spinal surgery. The MEDLINE database was queried using the key words “spine surgery” and “complications.” This initial search revealed more than 700 articles, which were further limited through an exclusion process. Each abstract was reviewed and papers were obtained. The authors gathered 105 relevant articles detailing 80 thoracolumbar and 25 cervical studies. Among the 105 articles were 84 retrospective studies and 21 prospective studies. The authors evaluated the study designs and compared cervical, thoracolumbar, prospective, and retrospective studies as well as the durations of follow-up for each study.
Results. In the 105 articles reviewed, there were 79,471 patients with 13,067 reported complications for an overall complication incidence of 16.4% per patient. Complications were more common in thoracolumbar (17.8%) than cervical procedures (8.9%; p < 0.0001, OR 2.23). Prospective studies yielded a higher incidence of complications (19.9%) than retrospective studies (16.1%; p < 0.0001, OR 1.3). The complication incidence for prospective thoracolumbar studies (20.4%) was greater than that for retrospective series (17.5%; p < 0.0001). This difference between prospective and retrospective reviews was not found in the cervical studies. The year of study publication did not correlate with the complication incidence, although the duration of follow-up did correlate with the complication incidence (p = 0.001).
Conclusions. Retrospective reviews significantly underestimate the overall incidence of complications in spine surgery. This analysis is the first to critically assess differing complication incidences reported in prospective and retrospective cervical and thoracolumbar spine surgery studies