The NSQIP database was queried from 2010 to 2018 to include patient demographic, comorbidity, surgical, morbidity, and mortality data. The database itself did not contain any THA operations performed using navigation technology before 2010. The database contains detailed perioperative and 30-day postoperative complication data collected from patients undergoing surgery at one of the approximately 700 participating hospitals in the United States. Data were prospectively collected and clinically verified by trained reviewers at each institution. Because the NSQIP data are deidentified, this study is exempt from institutional review board approval at our institution.
Generation of Study Groups
The Current Procedural Terminology (CPT) codes and International Classification of Diseases (ICD) Ninth and Tenth Revision Diagnosis Codes were used for purposes of identifying patient groups. The database was first queried for all patients older than 18 years who underwent THA using the CPT code 27130. Patients undergoing THA for nonelective indications, such as trauma, malignancy, revision procedure, or infection, and had ICD-9/ICD-10 codes reflective of these indications were excluded. In addition, cases missing perioperative or surgical data were excluded from the analysis. Patients were assigned to either TA-THA or U-THA study groups based on the presence of secondary CPT codes 20985, 0054T, or 0055T. 0054T refers to THA using fluoroscopically guided navigation, 0055T refers to THA using CT/MRI to construct the three-dimensional anatomy of the joint, and 20985 refers to all navigation systems that do not use imaging for registration of the anatomic and mechanical axis of the joint.
The primary outcomes were major complications within 30 days after surgery. For this study, we defined a major complication as any of the following: cardiac arrest, myocardial infarction, cerebrovascular accident, wound dehiscence, respiratory failure with inability to wean from ventilator, renal failure, deep organ-space infection, postoperative transfusion, deep vein thrombosis, pulmonary embolism, pneumonia, postoperative reintubation, periprosthetic wound infection (deep wound infection), sepsis and septic shock, revision surgery, and death. Major complications were then subcategorized based on the type of postoperative complication. Readmission was not listed as a postoperative complication. Secondary outcomes were surgical time, hospital length of stay (LOS), and discharge destination. Surgical time was defined as the time difference from the initial skin incision to skin closure. LOS was defined as the number of postoperative days a patient was admitted to the hospital. Discharge destinations were categorized into home, unskilled facility, rehab/skilled nursing facility, expired/hospice, and other. Each readmission and revision surgery diagnosis code was manually reviewed and categorized for subgroup analysis. Patient baseline demographic information and comorbidities, including the year of operation, ethnicity/race, age, American Society of Anesthesiologists' classification, functional status, body mass index (kg/m2), and 5-factor modified frailty index score, were collected. Surgical approach data are not recorded in NSQIP.
All analyses were done using SPSS v25 (IBM Corporation). A P value of less than 0.05 was considered to be statistically significant. Chi-squared, independent sample two-sided t-test or Mann-Whitney U testing was used to find differences between study groups based on the variable type and whether the data were normally distributed. The Bonferroni correction for P value was used when performing multiple comparison chi-squared analysis. All descriptive data are represented as means ± SD. Patient baseline demographics and comorbidities were first compared between navigation and conventional THA cohorts. To account for all baseline characteristics, propensity score matching was used to match patients undergoing navigation THA on a 1:1 basis with those undergoing conventional THA.
Propensity score matching was done to limit the effect of potential demographic variable differences between the two cohorts. A 1:1 match was done using a balanced, nearest-neighbor propensity score. This method of cohort matching has been established by previous literature as an optimal method for estimating differences between the treatment groups. Postmatching analysis was done to confirm the quality of matching between the study groups. All primary and secondary outcome variables were assessed between matched cohorts.
J Am Acad Orthop Surg. 2022;30(8):e673-e682. © 2022 American Academy of Orthopaedic Surgeons