Is Anesthesia Technique Associated With a Higher Risk of Mortality or Complications Within 90 Days of Surgery for Geriatric Patients With Hip Fractures?

Abstract

Background

Postoperative mortality and complications after geriatric hip fracture surgery remain high despite efforts to improve perioperative care for these patients. One factor of particular interest is anesthetic technique, but prior studies on this are limited by sample selection, competing risks, and incomplete followup.

Questions/purposes

(1) Among older patients undergoing surgery for hip fracture, does 90-day mortality differ depending on the type of anesthesia received? (2) Do 90-day emergency department returns and hospital readmissions differ based on anesthetic technique after geriatric hip fracture repairs? (3) Do 90-day Agency for Healthcare Research and Quality (AHRQ) outcomes differ according to anesthetic techniques used during hip fracture surgery?

Methods

We conducted a retrospective study on geriatric patients (65 years or older) with hip fractures between 2009 and 2014 using the Kaiser Permanente Hip Fracture Registry. A total of 1995 (11%) of the surgically treated patients with hip fracture were excluded as a result of missing anesthesia information. The final study sample consisted of 16,695 patients. Of these, 2027 (12%) died and 98 (< 1%) terminated membership during followup, which were handled as competing events and censoring events, respectively. Ninety-day mortality, emergency department returns, hospital readmission, deep vein thrombosis (DVT) or pulmonary embolism (PE), myocardial infarction (MI), and pneumonia were evaluated using multivariable competing risk proportional subdistribution hazard regression according to type of anesthesia technique: general anesthesia, regional anesthesia, or conversion from regional to general. Of the 16,695 patients, 58% (N = 9629) received general anesthesia, 40% (N = 6597) received regional anesthesia, and 2.8% (N = 469) patients were converted from regional to general.

Results

Compared with regional anesthesia, patients treated with general anesthesia had a higher likelihood of overall 90-day mortality (hazard ratio [HR], 1.22; 95% confidence interval [CI], 1.11-1.35; p < 0.001); however, when stratified by before and after hospital discharge but within 90 days of surgery, this higher risk was only observed during the inpatient stay (HR, 3.83; 95% CI, 3.18-4.61; p < 0.001); no difference was observed after hospital discharge (HR, 1.04; 95% CI, 0.94-1.16; p = 0.408). Patients undergoing conversion from regional to general also had a higher overall mortality risk compared with those undergoing regional anesthesia (HR, 1.34; 95% CI 1.04-1.74; p = 0.026), but this risk was only observed during their inpatient stay (HR, 6.84; 95% CI, 4.21-11.11; p < 0.001) when stratifying by before and after hospital discharge. Patients undergoing general anesthesia had a higher risk for all-cause readmission when compared with regional anesthesia (HR, 1.09; 95% CI, 1.01-1.19; p = 0.026). No differences according to anesthesia type were observed for risk of 90-day AHRQ outcomes, including DVT/PE, MI, and pneumonia.

Conclusions

We found the use of general anesthesia and conversion from regional to general anesthesia were associated with a higher risk of mortality during the in-hospital stay compared with regional anesthetic techniques, but this higher risk did not persist after hospital discharge. We also found general anesthesia to be associated with a higher risk of all-cause readmission compared with regional, but no other differences were observed in risk for complications. Our findings suggest regional anesthetic techniques may be preferred when possible in this patient population.

Level of Evidence

Level III, therapeutic study.

Introduction

Death after hip fracture surgery occurs in 17% to 37% of older patients within the first year, despite efforts to improve perioperative care for these patients [3, 4, 13, 14, 21, 26]. Although the causes are multifactorial, one modifiable factor of particular interest is the effect of anesthesia. The ideal anesthesia technique for this patient population has yet to be defined. In a study using the New York State inpatient database from 2004 to 2011, Neuman et al. [17] found no difference in 30-day postoperative outcomes according to anesthesia type for patients with hip fracture. The United Kingdom’s national hip fracture surgery study using a data set from 2012 also found no differences in mortality during 5 or 30 days postoperatively [29]. In contrast, a randomized trial of 322 patients from 2007 to 2012 concluded 1-year mortality was higher under regional anesthesia compared with general, whereas no difference was observed in postoperative complications [22]. Interestingly, a recent systematic review and meta-analysis of 20 retrospective observational studies and three randomized controlled trials by Van Waesberghe et al. [28] found the neuraxial technique was associated with lower in-hospital mortality, but no difference was observed by 30 days after surgery. Furthermore, this meta-analysis reported patients undergoing regional anesthesia had a lower likelihood of myocardial infarction (MI) and respiratory failure, whereas no difference was observed for pneumonia or pulmonary embolism (PE). These conflicting results generate uncertainty for geriatric anesthesia utilization and diverse clinical anesthesia practices for elderly and fragile patients.

Previous studies on the effects of anesthesia tend to include all patients with hip fractures regardless of their etiology and age [4, 17, 19, 29]; this approach often misses the potentially important effect of short-term anesthesia exposure on the postoperative outcomes for high-risk geriatric patients with hip fracture. These studies also included all kinds of hip fractures, including pathologic fractures and cancer, which can further confound the influence of anesthesia on the risk of complications and death [2, 5]. Furthermore, most studies have focused on mortality after hip fracture surgery as the primary outcome [2-4, 21], whereas none, to our knowledge, has investigated the potential effects of anesthetic technique on other postoperative complications while accounting for the competing risk of mortality. Finally, followup is variable across studies. The Centers for Medicare & Medicaid Services has redefined an “episode-of-care” as 90 days for quality initiative, and so the focus on this period has become important [11, 24].

We therefore sought to investigate the following questions: (1) Among older patients undergoing surgery for hip fracture, does 90-day mortality differ depending on the type of anesthesia received? (2) Do 90-day emergency department returns and hospital readmissions differ based on anesthetic technique after geriatric hip fracture repairs? (3) Do 90-day Agency for Healthcare Research and Quality (AHRQ) outcomes differ according to anesthetic techniques used during hip fracture surgery?

Materials and Methods

We conducted a retrospective study using a longitudinally maintained database from our community-based healthcare system’s hip fracture registry. Data collection, participation, and other details on the hip fracture registry have been previously published [10]. In brief, this registry identifies patients with surgically treated hip fractures and their demographics, medical comorbidities, intraoperative details, implant information, and outcomes using electronic medical records, administrative databases, and other institutional databases within the integrated healthcare system.

Patients aged 65 years and older who underwent surgery for a low-energy hip fracture (International Classification of Diseases, 9^th Revision codes 820.00, 820.01, 820.02, 820.03, 820.09, 820.10, 820.11, 820.12, 820.19, 820.20, 820.21, 820.22, 820.30, 820.31, 820.32, 820.8, 820.9) between January 2009 and July 2014 were included. Procedures were excluded if they involved patients with pathologic fractures, bilateral fractures, prior surgery on the affected hip, or multiple fractures treated at the same time. To maximize data integrity, we also excluded procedures from three regions (Colorado, Georgia, and the mid-Atlantic states) where care is provided in contract hospitals. Regional anesthesia included the following: spinal anesthesia, spinal-monitored anesthesia care, spinal-regional, spinal-nerve block, epidural, epidural-monitored anesthesia care, and combined spinal-epidural. The general anesthesia group included those who had planned general anesthesia. The conversion group included those who started with any type of regional anesthesia but were converted to general anesthesia at any time during surgery. Patients with incomplete (n = 117 [< 1%]) or missing (n = 1995 [11%]) anesthesia information were excluded from the analysis.

Primary outcomes were 90-day events including (1) emergency department return; (2) hospital readmission; (3) deep vein thrombosis (DVT) or PE; (4) MI; and (5) pneumonia. In addition, 90-day mortality was evaluated because it is a competing event to the primary outcomes. Emergency department return and readmission were defined as any encounter at an emergency department and any rehospitalization within 90 days of the discharge date, respectively. DVT and PE, MI, and pneumonia were defined as quality indicators according to the AHRQ [16] as any event within 90 days of the surgery date.

Patient variables included age, gender, race (Asian, Hispanic, others versus white), body mass index (BMI) categories (< 20, > 25 versus 20-25 kg/m²), American Society of Anesthesiologists (ASA) classification (1-2 versus 3-5), preoperative waiting time (hours), day of admission (weekend versus weekdays), procedure (internal fixation, hemiarthroplasty, other), fracture type (intracapsular--femoral neck, intracapsular--epiphysis, extracapsular--intertrochanteric, extracapsular--subtrochanteric), and patient comorbidities, which were identified using both the integrated healthcare system’s diabetes registry and the Elixhauser algorithm [7]. The ASA grouping was conducted in part as a result of small numbers for some of the levels and the subjectivity of the measure, being based on the judgment of the individual anesthesiologist. Furthermore, patients had multiple assessments of ASA score; therefore, one of the ways to reduce such bias is to group the scores. Covariates were included in the model as confounders if any standardized mean difference among anesthesia groups was > 0.1 or if they were deemed clinically important for the relationship studied.

The Hip Fracture Registry extracts all surgically treated hip fractures identified by diagnosis and/or procedure codes occurring within the integrated healthcare system [10]. Coverage is 100% of all surgically treated hip fracture procedures that occur in our integrated healthcare system. Anesthesia data are sourced from the perioperative orders module in our electronic health record (Epic). In most instances, for contract facilities, the availability of such data is limited to a few modules (mainly only billings, demographics, encounters) and not necessarily available for extraction (n = 1995, 11% of surgically treated patients with hip fracture were missing anesthesia information and excluded from the study sample). For outcomes, mortality data were obtained from the Social Security Administration, thus capturing all patients regardless of their insurance at the time of death. AHRQ quality indicators were obtained as a mandatory surveillance parameter and rehospitalization was obtained from the electronic health record, therefore collecting information on most patients except those who were lost to followup before 90 days as a result of death (n = 2027 [12%]) or membership termination (n = 98 [< 1%]), which were handled as competing events and censoring events, respectively. The final study sample consisted of 16,695 patients from 404 physicians and 38 hospitals within our system. Of the total, 58% (n = 9629) received general anesthesia, 40% (n = 6597) received regional anesthesia, and 2.8% (n = 469) patients converted from regional to general anesthesia. The average age was 82 years (SD = 7.9); the majority of patients were women (70%), white (80%), and had a BMI of 20 to 25 kg/m² (42%) (Table 1). Approximately three-fourths of patients had an ASA classification of ≥ 3 (75%). Median preoperative waiting time was 22 hours (interquartile range [IQR], 14-31), most patients were admitted on a weekday (72%), had an internal fixation procedure (62%), and had an intracapsular femoral neck fracture. The median number of complications per patient was four (IQR = 3-4); patients undergoing general anesthesia more frequently had valvular heart disease and more patients undergoing regional anesthesia had chronic obstructive pulmonary disease.

Table 1.

Characteristics of patients who underwent hip fracture repair, 2009-2014

The overall mortality rate was 2.1% (n = 354) at discharge and 12% (n = 2027) at 90 days. For those surviving to discharge, the 90-day emergency department return rate was 19% (n = 3078) and the 90-day readmission rate was 22% (n = 3478). The rates of 90-day postoperative morbidity were 2.5% (n = 411) for DVT or PE, 1.1% (n = 191) for MI, and 11% (n = 1751) for pneumonia.

Statistical Analysis

The effect of anesthesia type on mortality was assessed as a time-to-event outcome using mixed-effect Cox regression that accounted for surgeon and hospital random effects while adjusting for potential confounders as described previously. Survival curves “crossed” suggesting anesthesia had differential effects over the 90-day period. To further disentangle the effect in mortality before and after hospital discharge but within 90 days of surgery (that is, during inpatient stay and postdischarge to 90 days postoperatively), we modeled the differential effect using a time-dependent factor.

Encounter and AHRQ outcomes were modeled as time-to-event outcomes using competing risk proportional subdistribution hazard regression with stratification [8, 30]. All models adjusted for potential confounders as described previously and accounted for mortality as a competing event. Operating surgeon was included as a strata unit in all models to reduce potential bias as a result of correlated data from patients being operated on by the same surgeon or in the same hospital. Because the coefficients were estimates of relative risk condition on the other covariates being fixed or constant, a mixed-effect model with random intercept was chosen to fit strata unit. When emergency department returns or readmissions were analyzed, we excluded patients with missing information (1.0% [n = 169]) and patients who died during their hospital stay (2.1% [n = 354]) because they had no chance of being readmitted to the hospital.

Assumptions of proportional hazards in exposure variables were checked by plotting the Schoenfeld residuals against the vector of unique failure times. When the proportional hazard assumption was not met, we fitted the model with a time-dependent variable. To account for missing values in confounders (including gender [n = 1, < 1%], BMI [n = 63, < 1%], ASA [n = 175, 1%], preoperative waiting time [n = 1030, 6%], fracture type [n = 14, < 0.1%], comorbidities [n = 1; < 1%]), fully conditional specification multiple imputations using the Markov chain Monte Carlo estimation method were performed to create 50 versions of the analytic data set. Each data set was separately analyzed using the same model and the results were combined using Rubin’s rules. The imputation model included all variables. Analyses were performed using R Version 3.3.0 (Vienna, Austria) and α < 0.05 was considered statistically significant. Regional anesthesia was the reference group in all analyses.

To minimize indication bias, we performed a sensitivity analysis using competing risk regression models clustered at the surgeon level and propensity score weighting adjusted to potential confounders and compared regional anesthesia against general anesthesia. We excluded operations in which the anesthetic was converted from regional to general to isolate the effect of general and regional anesthesia. A propensity score is the probability of the patient being assigned to a treatment conditional on the patient and surgical factors calculated using logistic regression. When performing survival analysis, each procedure was weighted by the inverse propensity score to estimate the average treatment effect of the anesthesia. In the propensity score model, the coefficients have population-averaged interpretations, and thus, we use a marginal cluster generalized estimating equation to fit strata unit [1].

This study was approved by the institutional review board with Kaiser Permanente (IRB 6375) before its commencement. No outside funding was obtained.

Results

Compared with regional anesthesia, overall 90-day mortality was higher for both general anesthesia (hazard ratio [HR], 1.22; 95% confidence interval [CI], 1.11-1.35; p < 0.001) and conversion from regional to general (HR, 1.34; 95% CI, 1.04-1.74; p = 0.026) (Table 2). When stratifying according to mortality during the inpatient stay and after hospital discharge, mortality was higher for both general anesthesia and conversion from regional to general compared with regional anesthesia during the inpatient stay period (HR, 3.83; 95% CI, 3.18-4.61; p < 0.001 for general anesthesia and HR, 6.84; 95% CI, 4.21-11.11; p < 0.001 for conversion from regional to general), but no differences in mortality were observed in the period from hospital discharge to 90 days postoperatively (HR, 1.04; 95% CI, 0.94-1.16; p = 0.408 for general anesthesia and HR, 1.03; 95% CI, 0.77-1.39; p = 0.835 for conversion from regional to general). Sensitivity analysis showed no association between anesthesia type and mortality from hospital discharge to 90 days postoperatively (HR, 1.03; 95% CI, 0.93-1.14; p = 0.604), but a higher mortality risk during the inpatient stay (HR, 3.67; 95% CI, 2.95-4.58; p < 0.001) with a higher overall 90-day mortality risk (HR, 1.18; 95% CI, 1.07-1.31; p < 0.001).

Table 2.

Adjusted multivariable models^* of anesthesia type effect of anesthesia type on 90-day mortality for patients who underwent hip fracture repair (N = 16,695)

No differences were observed in emergency department returns for general anesthesia or conversion from regional to general when compared with regional anesthesia (HR, 0.97; 95% CI, 0.89-1.05; p = 0.430 for general anesthesia and HR, 0.94; 95% CI, 0.75-1.19; p = 0.611 for conversion from regional to general) (Table 3). Compared with regional anesthesia, patients undergoing general anesthesia had a higher risk for 90-day all-cause readmission (HR, 1.09; 95% CI, 1.01-1.19; p = 0.026) and no difference for unplanned (HR, 1.09; 95% CI, 1.00-1.18; p = 0.053) or planned readmission (HR, 1.18; 95% CI, 0.93-1.51; p = 0.176). There were no differences between conversion from regional to general and regional anesthesia in the risk of 90-day all-cause (HR, 0.92; 95% CI, 0.73-1.15; p = 0.457), unplanned (HR, 0.90; 95% CI, 0.71-1.14; p = 0.391), or planned readmissions (HR, 1.18; 95% CI, 0.55-2.53; p = 0.671). Handling confounding using propensity score, sensitivity analysis showed no association between anesthesia type and emergency department return (HR, 0.96; 95% CI, 0.88-1.04; p = 0.345) and all-cause readmission (HR, 1.06; 95% CI, 0.97-1.14; p = 0.183).

Table 3.

Adjusted multivariable models^* of effect of anesthesia type on 90-day emergency department returns and readmissions for patients who underwent hip fracture repair

No associations between anesthesia type and 90-day AHRQ outcomes were observed: DVT/PE (HR, 0.98; 95% CI, 0.78-1.24; p = 0.867 for general anesthesia and HR, 0.64; 95% CI, 0.31-1.30; p = 0.218 for conversion from regional to general), MI (HR, 1.22; 95% CI, 0.87-1.71; p = 0.239 for general anesthesia and HR, 1.68; 95% CI, 0.76-3.72; p = 0.201 for conversion from regional to general), or pneumonia (HR, 0.94; 95% CI, 0.84-1.05; p = 0.260 for general anesthesia and HR, 0.75; 95% CI, 0.54-1.03; p = 0.078 for conversion from regional to general) (Table 4). Handling confounding using propensity score, sensitivity analysis also showed no association between anesthesia type and DVT/PE (HR, 1.18; 95% CI, 0.94-1.49; p = 0.153), MI (HR, 1.02; 95% CI, 0.69-1.50; p = 0.919), or pneumonia (HR, 1.02; 95% CI, 0.91-1.14; p = 0.783).

Table 4.

Adjusted multivariable models^* of effect of anesthesia type on 90-day AHRQ outcomes for patients who underwent hip fracture repair (N = 16,695)

Discussion

The dismal outcomes after hip fracture surgery in older patients, particularly for in-hospital mortality rates of 2% to 5% [20] and 1-year mortality rates of 17% to 37% [3, 4, 13, 14, 21, 26], along with escalating demand for surgery have generated renewed interest in improving care for these patients. Anesthesia type may be one modifiable factor associated with the risk of complications and death after hip fracture surgery, but study of this topic is limited in part as a result of variability in sample selection and outcome definitions across studies. To advance the understanding of this potentially modifiable factor, we identified a large sample of high-risk patients and investigated 90-day complications after hip fracture surgery. In addition, we evaluated these complications in relation to the competing risk of mortality. In our study of 16,695 geriatric patients with fragility hip fracture surgery, after adjusting for patient characteristics, comorbidities, and surgical characteristics, we found a higher risk of 90-day mortality for patients who received general anesthesia and conversion from regional to general compared with regional anesthesia; however, this difference in risk was limited to in-hospital mortality. Furthermore, general anesthesia is associated with a higher risk of all-cause readmissions compared with regional anesthesia. No difference in emergency department returns or AHRQ quality indicators was observed according to anesthetic technique.

Our study is not without limitations. This study was retrospective. However, this study’s data were obtained from our integrated healthcare system’s hip fracture registry, which longitudinally maintains and validates outcome data, thereby maximizing the quality of the information we used. There were missing data for certain covariates in our study, however; this was addressed through multiple imputations, which estimated the values for missing data in an attempt to minimize bias and increase precision in the estimates. There also were patients excluded because of missing anesthesia information. Data from contract facilities are considered to be a subsample of the original sample (that is, the missing mechanism is missing completely at random); therefore, bias should be minimal when all contract facility procedures were excluded systematically. Although this study covers the four largest geographic areas of the integrated healthcare system, which is the sampling frame for this study, it is limited to procedures observed in institution-owned facilities only. Moreover, we included but censored patients with < 90 days followup as a result of membership termination before the study end date. These events were believed to be independent of failure time; thus, the risk of bias was minimized. We handled the high background mortality using a competing risk model. Indeed, under moderate dependence between mortality and failure types, cause-specific HR still reflected the influence of treatment isolated to the event of interest [6].

Anesthesia type may be driven by patient, surgeon, or anesthesiologist preference, therefore increasing risk for selection bias. Furthermore, as a result of unavailability of data, we were unable to evaluate the how dementia, delirium, and its postoperative deterioration may have confounded the results. Future investigation into these factors is warranted. We attempted to minimize confounding through covariate adjustment in multivariable models so we could evaluate the potential association between anesthesia type and risk of outcomes specifically. Sensitivity analysis using inverse propensity scores weighting to account for potential indication bias yielded results consistent with our original findings. Propensity scores adjust for potential selection bias resulting from the lack of randomization in an observational study, which may result in large differences in patient and surgical characteristics at baseline between treatment groups. This method creates an analysis that resembled what would occur if the treatment was randomly assigned through achieving balance in both known and unknown confounding factors between treatment groups [1], thus allowing us to evaluate the association between treatment and outcome using an average “treatment” effect if all patients were treated (or untreated) in this population. Although we may not include all potential confounders, weighting regression by propensity scores is more robust than multiple regression when the model omitted relevant variables that influence treatment selection, but with the price of an increase in random error, along with a downward bias in the nominal standard error [9]. Finally, information regarding specific anesthetic agents was unavailable; therefore, our findings are limited to generic categories of anesthetic modalities.

We found that patients older than 65 years of age undergoing hip fracture surgery under a general anesthetic or converted from regional to general anesthesia had a higher risk of mortality after surgery than patients who had a regional anesthetic. The observed difference seemed mainly driven by events during the inpatient stay; from hospital discharge to 90 days after surgery, we observed no between-group differences in the risk of mortality. Prior studies investigating mortality after hip fracture surgery are conflicting. Some have found a higher mortality risk with regional techniques after 1 year [22], whereas others have concluded no difference [17, 23], and others have found favorable outcomes with regional anesthesia during the in-hospital [18, 25] or 30-day timeframe [25]. Our findings of mortality risk for general compared with regional anesthesia parallel the results from two large-scale meta-analyses conducted by Luger et al. [15] and Van Waesberghe et al. [28] who reported a lower risk of short-term (< 1 month) or in-hospital mortality with a neuraxial technique, respectively, but no difference in the longer term at 3 months and 30 days, respectively. Furthermore, we identified a similar pattern of higher in-hospital risk of mortality for conversion from regional to general anesthesia compared with regional anesthesia. A prior study by Brox et al. [5], using the same registry data source from 2009 to 2012, found no difference in the likelihood for mortality. This is not surprising given the differences in study design, sample size, and inclusion and exclusion criteria. We included all patients aged 65 years and older, regardless of renal failure status, and were able to adjust for additional factors, including preoperative waiting time, procedure, fracture type, and hospital random effects. We also estimated the instantaneous failure rate at a given time, whereas Brox et al. [5] estimated the odds of a binary outcome regardless of the time of mortality. Moreover, we revealed an additional layer of the story: once a patient reached a medically stable state and was discharged, the mortality rate up to 90 days was not different across anesthesia types.

Ninety-day emergency department returns and hospital readmissions were high regardless of the type of anesthesia. The observed higher risk in all-cause hospital readmissions for general anesthesia compared with regional anesthesia (22% versus 21%, respectively; HR, 1.09) is likely clinically unimportant given the fact that both planned and unplanned hospital admissions in the general anesthesia group were both individually only slightly higher. Previous studies by Le-Wendling et al. [12] and Basques et al. [2] found no difference in readmissions rates; however, these studies were limited to readmissions within the first 30 days and did not account for the competing risk of mortality in analysis. We found no difference in risk of emergency department returns when comparing general anesthesia and conversion from regional to general anesthesia with regional anesthesia. To our knowledge, this is the first study to investigate anesthesia type and risk for emergency department returns after fragility hip fracture surgery, adding further insight to the association between anesthetic technique and risk of complications after geriatric hip fracture surgery.

No differences were observed for 90-day postoperative DVT/PE, MI, and pneumonia when comparing general anesthesia and conversion from regional to general with regional anesthesia. These results are consistent with prior report from O’Hara et al. [19], although only 7-day complications were evaluated. In contrast, conflicting results were found with Luger et al. [15] who reported a lower incidence of DVT and pneumonia for regional anesthesia and no difference in MI, and Van Waesberghe et al. [28] who reported a lower incidence of MI for regional anesthesia and no difference in pneumonia. However, both studies were meta-analyses with heterogeneity and a lack of competing risk analysis. Unlike our study here, there may also be anesthesia, surgery, institutional, and care provider confounders not accounted in these analyses. Nonetheless, these findings support conclusions that the anesthesia effect on postoperative complications is short-lived and rarely surpasses beyond the immediate postoperative period after motion and functional status are restored [27]. Although we did not find differences in postoperative complications by anesthesia technique, the high overall 90-day complication rates after elderly fragility fracture surgery are a concern given today’s new alternative payment models because postoperative complications, emergency department returns, and hospital readmissions account for substantial healthcare costs and resource utilization.

Through our integrated healthcare system’s Hip Fracture Registry, we observed the overall risk of mortality, emergency department returns, and hospital readmissions remain high in older patients undergoing hip fracture surgery. We found regional anesthesia was associated with a lower risk of mortality compared with general anesthesia and conversion from regional to general anesthesia; however, this association was limited to only the inpatient stay, before a patient was discharged when medically cleared. Regional anesthesia was also associated with a lower risk for all-cause readmission compared with general anesthesia. Based on these findings, healthcare providers may benefit from considering the utilization of regional anesthesia for fragility hip fracture surgery in the geriatric population while evaluating the full clinical picture. Delivering care should be evidence-based and precise to minimize the risk of complications and postoperative deaths, not solely based on providers’ preference. Future studies should also investigate the effect of anesthetic technique on dementia and delirium. Because perioperative care for elderly hip fractures is multifactorial and multidisciplinary, not limited to just anesthesia type, a well-designed randomized controlled trial investigating all components of the surgical care pathway may provide further information on the safest surgical treatment of older patients with hip fractures.