Comparison of Cost and Perioperative Outcome Profiles for Primary and Revision Posterior Cervical Fusion Procedures

Michael L. Martini, PhD; Rebecca B. Baron, BS; Jonathan S. Gal, MD; Sean N. Neifert, BS; William H. Shuman, BA; Alexander J. Schupper, MD; Emily K. Chapman, BS; Ian T. McNeill, MD; Jeremy Steinberger, MD; John M. Caridi, MD

Disclosures

Spine. 2021;46(19):1295-1301. 

In This Article

Methods

Data Source and Patient Stratification

Institutional records were retrospectively queried for all cases involving a PCDF procedure between January 1, 2008 and November 30, 2016. Patients were included if they underwent PCDF (CPD codes: 22600, 22210, 22110, 63015, 63045, 63001) without an accompanying ACDF (CPT codes: 22554, 22551, 63075). Patients who had undergone either a prior fusion or laminectomy were noted and used to comprise the prior surgery cohort. Index cases were excluded from this study if they were performed for traumatic injuries, tumor resections, or infections. All included index cases were screened to ensure that degenerative disc disease was present in the patient's record. For revision cases, repeated hospitalizations during the same period were included, as were revision surgeries for wound infections (e.g., washouts) in the perioperative period. This study did not impose a time restriction on how long after the first procedure the associated revision needed to occur. These inclusion and exclusion criteria enabled the selection of a less heterogeneous cohort of patients that underwent PCDF procedures for degenerative spine conditions. This retrospective study was approved by the Institutional Review Board, and informed consent was waived.

Covariates

Demographic and perioperative data were collected from each case, including the patient's age, sex, preoperative diagnosis, number of segments fused, surgical duration, and estimated blood loss (EBL). In addition, each patient's preoperative comorbidity burden was measured via the Elixhauser Index score utilizing administrative coding algorithms from Quan et al and weighting from van Walraven et al.[12–14] The patient's length of stay at the hospital, number of days spent in the intensive care unit (ICU), total direct costs of hospitalization, direct costs for the surgery, associated labs, nursing, and pharmacy were also queried. Perioperative medical complications were also documented including pulmonary embolism, deep vein thrombosis, pneumonia, acute kidney injury, hemorrhagic anemia, myocardial infarction, cardiac arrest, cerebrovascular accident, sepsis, wound dehiscence, superficial surgical site infection, airway complications, septic shock, and urinary tract infection. Other perioperative outcomes were examined including whether the patient experienced a required ICU stay, return to OR, nonhome discharge, or delayed extubation, which was defined as extubation after leaving the operating room. In addition, any readmissions or emergency department (ED) visits within 30 days were also documented. The primary outcome considered in this study was perioperative complications, while resource utilization measures such as hospitalization length, required stay in the ICU, direct costs of hospitalization, and 30-day ED admissions were explored as secondary outcomes.

Statistical Analysis

Patients with revision surgery were compared with those who had not undergone cervical spine surgery previously. SAS 9.4 (The SAS Institute, Cary, NC) was used for statistical analysis. Continuous variables were compared using two-sided, two-sample t tests, while chi-squared tests were used to analyze categorical variables. Fisher Exact test was used for contingency tables containing an expected count less than one under the null hypothesis of independence. Univariate and multivariate regression models were constructed to assess how a prior surgery impacted the risk of various adverse clinical outcomes. Multivariable models controlled for demographic and intraoperative variables that were previously known factors shown to affect surgical outcomes. Statistical significance was determined using P < 0.05.

processing....