Outcome of Spinal Versus General Anesthesia in Revision Total Hip Arthroplasty

A Propensity Score-Matched Cohort Analysis

Venkatsaiakhil Tirumala, MS; Georges Bounajem, MD; Christian Klemt, PhD; Stephen P. Maier, MD; Anand Padmanabha, MD; Young-Min Kwon, MD, PhD


J Am Acad Orthop Surg. 2021;29(13):e656-e666. 

In This Article



After obtaining approval from the Institutional Review Board, we retrospectively evaluated a total of 2,656 consecutive patients who underwent revision THA at our academic tertiary referral center. Revision surgery of the hip included revisions of (1) both acetabular and femoral implanys, (2) either acetabular or femoral implant, and (3) acetabular liner and/or femoral head only. Age, laterality, sex, American Society of Anesthesiologists (ASA) score, body mass index (BMI), comorbid conditions, LOS, follow-up time, and indication for revision were all collected from the electronic medical record using hospital progress notes, clinic notes or surgical reports. Of the total patients who underwent revision THA, we excluded those who had received epidural anesthesia or any other anesthesia besides spinal and general, previous revisions, any operation within three months of revision, missing or insufficient perioperative data, or insufficient follow-up of at least 2 years. In addition, in accordance with the recommendations provided by the American Society of Regional Anesthesia,[15] spinal anesthesia was contraindicated in patients presenting with preexisting central neuraxial disease (multiple sclerosis, spinal stenosis, or myasthenia gravis), excess anxiety, hypovolemia, hypertrophic obstructive cardiomyopathy, thrombocytopenia, or anticoagulant use. Outside of these contraindications, anesthesia type primarily was based on surgeon/anethesiologist preference.

After the aforementioned exclusions, of an initial 2,656 patients, 2,202 remained for further analysis, which included 357 (16.2%) and 1845 (83.8%) patients had spinal and general anesthesia in their revision THA, respectively. We found that these two cohorts were statistically similar for a number of baseline characteristics, including age, gender, race, ASA score, Charlson comorbidity index (CCI), most comorbidities, most indications for revision, components revised, and tranexamic acid usage. Of the markedly different baseline characteristics in the raw cohorts, patients in the general anesthesia cohort had a markedly higher BMI, presented more frequently for periprosthetic fractures, and had a higher prevalence of vascular disease. Patients in the spinal anesthesia cohort presented more frequently for adverse local tissue reactions (all P < 0.05).

Cohort Matching

To control for selection bias between the nonrandomized spinal and general anesthesia groups, propensity score matching was used to match the spinal anesthesia group to the general anesthesia group on the following covariates: (1) age, (2) gender, (3) BMI, (4) ASA score, (5) CCI, (6) indications for revision, and (7) number of surgeons. The propensity score was defined as the conditional probability of receiving general anesthesia during revision THA based on the specified covariates of interest (1 = general, 0 = spinal). A 1:1 match ratio was chosen as this best minimized intergroup covariate differences in our study and has been shown to result in an optimal estimation of treatment effect(s) in the existing literature.[16] The balance between covariates after matching was checked via the standard mean difference (SMD), which had a set threshold of 0.25 as recommended by Rosenbaum and Rubin.[17] No covariate weighting was performed because there were multiple outcomes of interest in this study. The greedy matching algorithm was used to create the two balanced cohorts.[18]


For patients administered general anesthesia, 1 to 1.5 mcg/kg of fentanyl, 1.5 to 2.0 mg/kg of propofol, and 1.5 mg/kg of lidocaine were used for induction and intubation. Sevoflurane was used for maintenance, with midazolam and/or fentanyl supplemented as necessary. Rocuronium bromide injections were used for muscle relaxation. In the spinal anesthesia cohort, patients were injected with 0.5% bupivacaine into the L3-L4 or L4-L5 interspace, followed by intravenous sedation using midazolam and/or fentanyl as needed. Ephedrine and phenylephrine were used for cardiovascular support as required. In both cohorts, no concurrent use of other types of regional anesthesia, such as femoral nerve or canal blocks, was observed.[7] All revision THAs were performed by fellowship-trained arthroplasty surgeons and under prophylactic intravenous antibiotics administered before skin incision. Revision surgery of the hip was performed using a standard posterolateral approach.

Perioperative Outcomes

Intraoperative blood loss and transfusions were noted from surgical notes. Postoperative transfusions were noted from the discharge summary. To calculate total perioperative blood loss, we used the Gross equation and modified the result based on blood transfusions that were issued intraoperatively and/or postoperatively.[19] For this, we calculated the maximum hemoglobin change, which represented the difference between the preoperative and minimum postoperative level. Thus, our estimates of total perioperative blood loss were relative to the minimum hemoglobin level in patients, which may have occurred at different time periods after revision surgery. Patient records were further reviewed to identify LOS, discharge disposition, and whether any in-hospital complications occurred. We noted patients that had a LOS greater than the 90th percentile of all revision THAs in our database (12 days) as having an extended LOS.[20]

Postdischarge Outcomes

All patients had a minimum follow-up time of 2 years unless they died or were re-revised. As per the patient's medical record, we noted 30-, 60-, and 90-day unplanned readmissions. Furthermore, we classified complications the patient presented with in their readmission as either a "major" or "minor" complication with guidance from the existing TJA literature.[21,22] Pneumonia, urinary tract infection, renal insufficiency, and superficial surgical site infection constituted as minor complications. Unplanned intubation, wound dehiscence, hematoma/seroma, deep surgical site infection, myocardial infarction, shock, sepsis, stroke, deep vein thrombosis, pulmonary embolism, revision surgery, and mortality constituted as major complications. Re-revisions of the hip due to aseptic (wear, loosening, and fracture) and septic complications (periprosthetic joint infection) were noted from either surgical notes or consult notes.

Statistical Analysis

Propensity score matching and statistical analyses were performed with the use of Statistical Package for Social Sciences version 25.0 (IBM Released 2019; IBM SPSS Statistics for Windows, Version 25.0; IBM).[23] Patients were categorized by the type of anesthesia used (spinal versus general), and all preoperative patient characteristics were compared between cohorts (Table 1). Multivariate logistic regression analysis using all covariates was then used to calculate adjusted odds ratios (ORs) and 95% confidence intervals for the outcomes of interest. From this analysis, an adjusted mean difference or OR was determined depending on whether the variable of interest was continuous or dichotomous about general anesthesia. Categorical variables were compared using the chi-squared test, and continuous variables were compared using analysis of variance.[24] The level of significance was set at P < 0.05 for all univariate and multivariate analyses in this study.