eTable 1. Definition of comorbidities (ICD-10)
eTable 2. Adjusted analyses restricted to patients younger than 60 years of age who were residing at home before surgery
eTable 3. Adjusted analyses that did and did not include hospital length of stay (LOS) as a covariate
Le Manach Y, Collins G, Bhandari M, Bessissow A, Boddaert J, Khiami F, Chaudhry H, De Beer J, Riou B, Landais P, Winemaker M, Boudemaghe T, Devereaux PJ. Outcomes After Hip Fracture Surgery Compared With Elective Total Hip Replacement. JAMA. 2015;314(11):1159-1166. doi:10.1001/jama.2015.10842
Patients undergoing surgery for a hip fracture have a higher risk of mortality and major complications compared with patients undergoing an elective total hip replacement (THR) operation. The effect of older age and comorbidities associated with hip fracture on this increased perioperative risk is unknown.
To determine if there was a difference in hospital mortality among patients who underwent hip fracture surgery relative to an elective THR, after adjustment for age, sex, and preoperative comorbidities.
Design, Setting, and Participants
Using the French National Hospital Discharge Database from January 2010 to December 2013, patients older than 45 years undergoing hip surgery at French hospitals were included. The International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10), codes were used to determine patients’ comorbidities and complications after surgery. A population matched for age, sex, and preoperative comorbidities of patients who underwent elective THR or hip fracture surgery was created using a multivariable logistic model and a greedy matching algorithm with a 1:1 ratio.
Main Outcomes and Measures
Postoperative in-hospital mortality.
A total of 690 995 eligible patients were included from 864 centers in France. Patients undergoing elective THR surgery (n = 371 191) were younger, more commonly men, and had less comorbidity compared with patients undergoing hip fracture surgery. Following hip fracture surgery (n = 319 804), 10 931 patients (3.42%) died before hospital discharge and 669 patients (0.18%) died after elective THR. Multivariable analysis of the matched populations (n = 234 314) demonstrated a higher risk of mortality (1.82% for hip fracture surgery vs 0.31% for elective THR; absolute risk increase, 1.51% [95% CI, 1.46%-1.55%]; relative risk [RR], 5.88 [95% CI, 5.26-6.58]; P < .001) and of major postoperative complications (5.88% for hip fracture surgery vs 2.34% for elective THR; absolute risk increase, 3.54% [95% CI, 3.50%-3.59%]; RR, 2.50 [95% CI, 2.40-2.62]; P < .001) among patients undergoing hip fracture surgery.
Conclusions and Relevance
In a large cohort of French patients, hip fracture surgery compared with elective THR was associated with a higher risk of in-hospital mortality after adjustment for age, sex, and measured comorbidities. Further studies are needed to define the causes for these differences.
Although hip surgery can improve mobility and pain, it can be associated with major postoperative medical complications and mortality.1 Patients undergoing surgery for a hip fracture are at substantially higher risk of mortality and medical complications compared with patients undergoing an elective total hip replacement (THR).2,3 The increased risk might be due to the advanced age and comorbidities of hip fracture patients relative to elective THR patients.2- 4Quiz Ref ID Although patient characteristics may explain the higher risk for a poor outcome among patients undergoing hip fracture surgery, it is possible that physiological processes associated with hip fracture (eg, the acute inflammatory, stress, hypercoagulable, and catabolic states)5- 9 may account for some of the increased risk.10,11 These processes may account for some of the perioperative morbidity and mortality and, therefore, may represent modifiable risk factors. For example, surgery may be performed earlier in a patient’s disease course to minimize the time patients are exposed to these intrinsic factors to improve outcomes.12 To better understand the potential contribution of these processes to adverse outcomes, we determined the difference in in-hospital mortality between patients undergoing hip fracture surgery and elective THR after adjusting for the known risk factors age, sex, and patient comorbidities.
The Agence Technique de l'Information Sur l'Hospitalisation (ATIH) waived the need for consent according to the Enforcement (decree No. 94-666). Since 1996, all French hospitals caring for medical and surgical patients have submitted anonymized patient data to the French Hospital Discharge Database (FHDD). Each discharge summary submitted to the FHDD is linked to a national grouping algorithm leading to a French diagnosis related group13; thereby allowing patient comorbidities to be recorded and linked.14 The study was conducted according to the approval given by the ATIH. Authorization was also obtained from the Commission Nationale de l'Informatique et des Libertés (agreement No. 1375062). The data provided were anonymized.
We identified and included all patients older than 45 years who underwent elective THR or hip fracture surgery requiring anesthesia, performed in France from January 1, 2010, to December 31, 2013. Patients with multiple trauma were excluded. For each patient the following information was extracted: age, sex, primary diagnosis (ie, reason for hospital admission), patient comorbidities (coded according to the International Statistical Classification of Diseases and Related Health Problems, 10th Revision [ICD-10]),15 procedures performed during the hospital stay (coded according to the French classification for medical procedures in the Classification Commune des Actes Médicaux [CCAM]),16 length of stay (LOS) in the hospital (days), and in-hospital complications and mortality after surgery. Using a previously established process, 3 physicians, with expertise in ICD-10 coding, independently aggregated ICD-10 codes into broader disease groups to resemble the clinical observations commonly recorded during preoperative assessments (eTable 1 in the Supplement). All disagreements were resolved through a consensus process that involved discussions among the 3 physicians. The Simplified Acute Physiology Score (SAPS II)17 was used to measure the severity of disease for patients admitted to an intensive care unit (ICU).
Quiz Ref IDThe primary outcome was in-hospital mortality, defined as death after surgery and prior to hospital discharge, regardless of LOS. Secondary outcomes included in-hospital major postoperative complications (ie, a composite of myocardial infarction, heart failure, stroke, renal failure, sepsis, and mortality) and each of the individual components of the composite except for mortality. We also evaluated readmission for any cause within 72 hours after initial hospital discharge and in-hospital mortality during readmission.
Continuous and categorical variables were compared using parametric or nonparametric methods as appropriate. A matched population of patients who underwent elective THR and hip fracture surgery was created using a multivariable logistic model and a greedy matching algorithm with a 1:1 ratio and a caliper width of 10−5 with no replacement.18 The matching procedure was carried out by categorizing age (in 5-year age intervals) in which elective THR and hip fracture patients were matched within each age strata. All preoperative variables were included in the matching model and are presented in Table 1. Patients receiving hip fracture surgery and elective THR for whom no match was found were discarded from the matched analyses. After matching, balance of covariates between the groups was assessed using the standardized differences expressed as a percentage for each covariate,19 and globally using the L1 measure (in which a value of 0 indicates identical distributions between groups and 1 indicates complete imbalance) and post-matching Cstatistic (in which a value of 0.5 indicates perfect balance).20 Any baseline covariate with an absolute standardized difference greater than 5% was considered to be imbalanced, because such a difference may have a statistical effect on the results depending on the prognostic importance of the covariate.
Relative risk (RR) and the 95% CIs for primary and secondary outcomes were computed using conditional Poisson regression. P values were 2-tailed, and values less than .05 were considered significant.
To explore the potential statistical effect of unmeasured confounders related to patient frailty on the primary outcome, we conducted a sensitivity analysis, replicating the main analysis for the primary and secondary outcome on patients younger than 60 years and residing at home prior to surgery (ie, the group of patients at lowest risk of frailty). We also conducted sensitivity analyses in which RR estimations were adjusted for the hospital LOS.
Statistical analyses were carried out in R software (R Foundation), version 3.1, and matching procedures were performed using MatchIt package (R Foundation), version 2.4-21.
During the study period, 690 995 eligible patients from 864 centers in France were identified and included in this study. Elective THR was carried out in 371 191 patients and hip fracture surgery in 319 804 patients. Table 1 reports the baseline patient characteristics of these 2 groups, and the Figure shows the age and sex distributions.
In the unmatched study population, patients undergoing an elective THR were younger and had fewer preoperative comorbidities. The mean LOS for patients who underwent hip fracture surgery was longer than for those who underwent an elective THR (12.09 days [95% CI, 12.07-12.13] for hip fracture surgery vs 7.80 days [95% CI, 7.79-7.81] for elective THR; P < .001). Over the study period, the mean length of hospital stay decreased in the hip fracture group (from 12.77 days [95% CI, 12.70-12.81] in 2010 to 11.72 days [95% CI, 11.67-11.77] in 2013; P < .001) and the elective THR group (from 9.19 days [95% CI, 9.13-9.23] in 2010 to 8.11 days [95% CI, 8.08-8.12] in 2013; P < .001). Postoperatively, patients who underwent surgery for a hip fracture were less frequently discharged to their home (35.8% for hip fracture surgery vs 60.0% for elective THR; P < .001). Readmissions within 72 hours of discharge after the index surgery were more common among patients who underwent hip fracture surgery (3897 patients [1.3%]) compared with patients who underwent an elective THR (1647 patients [0.4%]; P < .001).
Table 2 reports the in-hospital outcomes for the unmatched study groups. After hip fracture surgery, 10 931 patients (3.42% [95% CI, 3.35%-3.47%]) died and, after elective THR surgery, 669 patients (0.18% [95% CI, 0.17%-0.19%]) died (absolute risk increase, 3.23% [95% CI, 2.98-3.51]; RR, 18.96 [95% CI, 17.54-20.50]; P < .001). All the other individual complications were more common in the patients receiving hip fracture surgery compared with those receiving the elective THR, with RRs ranging from 3.41 to 7.23. At least 1 major postoperative complication occurred in 32 913 patients receiving hip fracture surgery (10.29% [95% CI, 10.18%-10.40%]) and in 5955 patients receiving elective THR (1.60% [95% CI, 1.56%-1.64%]; absolute risk increase, 8.69% [95% CI, 8.38%-8.94%]; RR, 6.41 [95% CI, 6.24-6.59]; P < .001).
In the unmatched study population, postoperative ICU admission was also more frequent in patients who underwent a hip fracture surgery compared with an elective THR (1.52% for hip fracture surgery vs 0.27% for elective THR, P < .001), and the SAPS II at ICU admission was higher among patients receiving hip fracture surgery compared with those receiving elective THR (45.37 [95% CI, 44.65-46.08] for hip fracture surgery vs 34.97 [95% CI, 33.69-36.25] for elective THR; P < .001). Among patients readmitted within 72 hours after the initial hospital discharge, in-hospital mortality during the second stay was higher in those receiving hip fracture surgery (12.5% for hip fracture surgery vs 1.5% for elective THR; P < .001).
The matching procedure retained 117 157 patients who underwent hip fracture surgery and 117 157 patients who underwent an elective THR. When compared with the unmatched populations, patients who underwent hip fracture surgery included in the matched population were younger (mean age: 75.0 years [95% CI, 74.94-75.06] for matched population vs 81.7 years [95% CI, 81.66-81.74] for unmatched population; P < .001) and patients who underwent elective THR included in the matched population were older (mean age: 75.0 years [95% CI, 74.94-75.06] for matched population vs 70.2 [95% CI, 70.17-70.23] for unmatched population; P < .001). After matching, there were no clinically important differences in baseline characteristics across the patient groups (Table 1 and Figure). Standardized differences in baseline characteristics were all less than 1%. Global balance metrics, the L1 measure and Cstatistic, demonstrated minimal imbalances between groups after matching (L1 measure: 0.592 before matching vs 0.427 after matching; Cstatistic: 0.882 before matching vs 0.503 after matching).
Table 3 reports the in-hospital outcomes for the matched study groups. In the matched population, 2130 patients died after hip fracture surgery (1.82% [95% CI, 1.75%-1.90%]) and 362 patients died after elective THR (0.31% [95% CI, 0.28%-0.34%]; absolute risk increase: 1.51% [95% CI, 1.46%-1.55%]; RR, 5.88 [95% CI, 5.26-6.58]; P < .001). All other individual complications were more common in patients who underwent hip fracture surgery compared with those who underwent elective THR, with RRs ranging from 1.62 to 3.44. A major postoperative complication occurred in 6890 patients receiving hip fracture surgery (5.88% [95% CI, 5.75%-6.01%]), and 2741 patients receiving elective THR (2.34% [95% CI, 2.25%-2.43%]; absolute risk increase, 3.54% [95% CI, 3.50%-3.59%]; RR, 2.50 [95% CI, 2.40-2.62]; P < .001).
The sensitivity analysis, which focused on patients younger than 60 years who were residing at home before surgery, included 12 683 patients who underwent hip fracture surgery (4.0% of all patients who underwent hip fracture surgery) and 12 683 patients who underwent elective THR (3.4% of all patients who underwent elective THR). In this matched population, 58 patients died after hip fracture surgery (0.46% [95% CI, 0.34%-0.57%]), and 10 patients died after elective THR (0.08% [95% CI, 0.03%-0.13%]; absolute risk increase, 0.38% [95% CI, 0.16%-0.83%]; RR, 5.80 [95% CI, 2.96-11.35]; P < .001). eTable 2 in the Supplement reports the primary and secondary outcomes for this younger group who resided at home before surgery. As per the results of the primary outcome, patients with a hip fracture also demonstrated higher risk-adjusted associations with the secondary outcomes (except for postoperative myocardial infarction and postoperative heart failure).
The mean hospital LOS for patients was longer among patients who underwent hip fracture surgery compared with patients who underwent an elective THR (11.61 days [95% CI, 11.56-11.64] for hip fracture surgery vs 9.3 days [95% CI, 9.28-9.32] for elective THR; P < .001). eTable 3 in the Supplement reports the results of the analyses that did and did not include hospital LOS in the adjusted analyses. The results were similar for these analyses. For example, after adjustment for LOS, the RR for in-hospital mortality associated with hip fracture surgery relative to elective THR was similar to our primary analysis not adjusted for LOS (in-hospital mortality RR: 5.67 [95% CI, 5.07-6.34] for LOS-adjusted analysis vs 5.88 [95% CI, 5.26-6.58] for unadjusted primary analysis; P < .001 for both analyses).
Patients in the hip fracture surgery group were more frequently admitted to the ICU postoperatively (1.3% for hip fracture surgery vs 0.4% for elective THR; P < .001) and stayed longer in the ICU than patients who underwent an elective THR (6.3 days [95% CI, 5.8-6.8] for hip fracture surgery vs 4.5 days [95% CI, 3.8-5.2] for elective THR; P < .001). Patients in the hip fracture surgery group and the elective THR group demonstrated no significant difference in SAPS II values at ICU admission (40.1 [95% CI, 39.0-41.1] for hip fracture surgery vs 39.0 [95% CI, 37.1-40.9] for elective THR; P = .261) and in-hospital mortality in patients admitted to the ICU (18.1% for hip fracture surgery vs 15.9% for elective THR; P = .264). Hospital readmission within 72 hours was more frequent in patients who underwent hip fracture surgery (1.0% for hip fracture surgery vs 0.5% for elective THR; P < .001) and in-hospital mortality during the second admission was higher in patients receiving hip fracture surgery compared with those receiving elective THR (8.4% vs 2.2%; P < .001).
Quiz Ref IDPatients undergoing surgery for a hip fracture were older and had more comorbidities than patients who underwent an elective THR, and these differences accounted for some of the difference in outcomes between these groups. The primary finding of this study was that among patients who received hip fracture surgery or elective THR and were matched for age, sex, and preoperative medical conditions, the risks were higher after hip fracture surgery for in-hospital mortality (absolute risk increase, 1.51% [95% CI, 1.46%-1.55%]; RR, 5.88 [95% CI, 5.26-6.58]) and major postoperative complications (absolute risk increase, 3.54% [95% CI, 3.50%-3.59%]; RR, 2.50 [95% CI, 2.40-2.62]). If the absolute risk increases of 1.51% for in-hospital mortality and 3.54% for major postoperative complications were modifiable, they would be consistent with the number needed to treat of 59 patients for in-hospital mortality and 28 patients for major postoperative complications. Hip fracture may be associated with physiologic processes that are not present in circumstances leading to elective THR and increase the risk of morbidity and mortality following surgery.
Strengths of the study include our evaluation of a large contemporary sample of patients who underwent hip surgery in France. We took into account an extensive list of comorbidities to obtain matched patient groups who underwent an elective THR or a hip fracture surgery. The balance of covariates between the groups suggests that the RR estimates were not biased by imbalances in the recorded variables.
Quiz Ref IDThis study had several limitations including potential misclassification of outcomes. In-hospital mortality was defined as the primary outcome to minimize the statistical effect of these limitations. We were unable to evaluate the cause of death because these data were not available. Misclassification of secondary outcomes and potential underestimation of their frequencies remains possible; however, this would be expected to be similar in both groups. Nevertheless, we cannot completely exclude differential misclassification between the surgical groups, which could have biased the RR estimates.
Another limitation of this study is the nonstandardized follow-up period. The follow-up period was limited to the in-hospital admission period, and it is different between the 2 patient groups. Because the absolute difference in the duration of follow-up was short (ie, a difference of 2.3 days), it is unlikely that this difference in follow-up explains the magnitude of risk associated with hip fracture surgery relative to an elective THR. Although there is no national mortality database that we could link to the FHDD to capture 30-day mortality, we evaluated hospital readmissions within 72 hours of the index hospital discharge and assessed mortality during readmissions. Readmissions and in-hospital mortality during readmission were higher in the patients who underwent a hip fracture surgery relative to an elective THR. Moreover, the sensitivity analyses, adjusting for LOS, produced similar results as the main analyses and supported our main findings (eTable 3 in the Supplement).
An additional limitation of our study is the possibility of residual confounding (ie, that there were unmeasured factors affecting one group or another that were not measured by the data available for analysis). There is also the possibility of a selection bias that adjustment cannot resolve because patients having an elective THR are selected for their ability to withstand the stresses of surgery before the operation. Moreover, we did not have information on the mechanism of injury (eg, a fall due to instability, orthostatic hypotension, or poor nutritional status) or concomitant injuries. These unmeasured comorbidities in patients with hip fracture may have influenced outcomes. Although residual confounding is probable, physiologic mechanisms, such as acute stress and inflammatory states resulting from the fractures, likely explain our findings because the diverse array of preoperative comorbidities that we adjusted for in our analyses, the magnitude of effect we found, and our sensitivity analyses restricted to younger patients living at home prior to surgery (ie, the group of patients at lowest risk of frailty before surgery) demonstrated results similar to the overall analyses.
We were not able to adjust for whether surgery was performed by a consultant or surgical resident because this information was not available. In France, surgical residents are not allowed to operate on patients alone without consultant supervision. Even if that were the case, resident-performed procedures like the ones assessed in this study have clinical outcomes equivalent to those when a consultant (attending surgeon) performs the procedures alone.21- 23 Furthermore, hip fracture surgery is generally considered less invasive than total hip arthroplasty. These points suggest that if the surgeons performing the hip fracture surgeries were relatively less experienced than the surgeons performing the elective THRs, the effect on our results was likely minimal.
A meta-analysis of prospective cohort studies (22 studies among women and 17 studies among men) of age- and sex-matched patients with and without hip fractures demonstrated, after adjustment for comorbidities, that patients experiencing a hip fracture had an increased risk of mortality at 3 months (relative hazard, 5.75 [95% CI, 4.94-6.67] among women and 7.95 [95% CI, 6.13-10.30] among men).24 Although these data establish a patient experiencing a hip fracture is at an increased risk of mortality compared with a patient who has not sustained a hip fracture, these data do not determine if the risk relates to surgery or something intrinsic to the hip fracture. A large administrative database study evaluated outcomes between patients undergoing surgery for a hip fracture and an elective THR between the years 1990 and 2007.25 Similar to our findings, this study demonstrated that hip fracture surgery was associated with worse outcomes than THR (eg, in-hospital mortality, 1.8% for hip fracture surgery vs 0.2% for THR). In contrast to our study, the study by Sassoon et al25 only reported unadjusted results.
In a recent study, patients with hip fracture admitted to a dedicated geriatric unit with a multidisciplinary approach after surgery had a lower risk-adjusted 6-month mortality compared with patients admitted to an orthopedic floor.26 In our matched analysis, patients admitted to the ICU in both groups had similar disease severity (ie, SAPS II) and experienced similar in-hospital mortality. Our results also suggest the possibility that the medical specialty (ie, intensivists) and location of postoperative care after a hip fracture may be associated with outcomes.
Despite the benefits of hip surgery, it increases a patient’s risk of major morbidity and mortality. The more comorbidity a patient has prior to surgery, the higher their risk of a major complication after surgery.1 Patients undergoing hip fracture surgery are older and have higher burdens of comorbidities compared with patients undergoing elective THR. Some authors have assumed these differences account for the higher risk of complications in patients undergoing hip fracture surgery relative to elective THR. Our data suggest this is not the entire explanation and that factors intrinsic to a hip fracture may also influence the outcomes.
Quiz Ref IDA hip fracture results in trauma, pain, bleeding, and immobility. These factors initiate inflammatory, hypercoagulable, stress, and catabolic states5- 7,9,27- 30 that have the potential to cause complications (eg, myocardial infarction, pulmonary embolism, pneumonia, sepsis, stroke, major bleeding, disability, or mortality).10,31 It is possible to minimize a patient’s exposure to these harmful factors through rapid surgery.12,32,33 Studies suggest that reducing the delay between a hip fracture and surgery may limit the consequences of the fracture, and the risk-adjusted mortality and major complications after hip fracture surgery may then more closely approximate those of patients undergoing elective THR.33
In a large cohort of French patients, hip fracture surgery compared with elective THR was associated with a higher risk of in-hospital mortality after adjustment for age, sex, and measured comorbidities. Further studies are needed to define the causes for these differences.
Corresponding Author: Yannick Le Manach, MD, PhD, Population Health Research Institute, David Braley Cardiac, Vascular and Stroke Research Institute, Perioperative Medicine and Surgical Research Unit, 237 Barton St E, Hamilton, ON L8L 2X2, Canada (email@example.com).
Author Contributions: Drs Le Manach and Boudemaghe had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Le Manach, Bhandari, De Beer, Riou, Devereaux.
Acquisition, analysis, or interpretation of data: Le Manach, Collins, Bessissow, Boddaert, Khiami, Chaudhry, Riou, Landais, Winemaker, Boudemaghe, Devereaux.
Drafting of the manuscript: Le Manach, Collins, Devereaux.
Critical revision of the manuscript for important intellectual content: Le Manach, Collins, Bhandari, Boddaert, Khiami, Chaudhry, De Beer, Riou, Landais, Winemaker, Boudemaghe, Devereaux.
Statistical analysis: Le Manach, Collins.
Obtained funding: Le Manach.
Administrative, technical, or material support: Le Manach, Khiami, Landais, Boudemaghe.
Study supervision: Le Manach, Bhandari, Boddaert, Khiami, De Beer, Riou, Landais.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Bhandari reports grant funding from Smith and Nephew, DePuy, Eli Lilly, Bioventus, Stryker, Zimmer Biomet, and Amgen and consulting fees from Smith and Nephew, Stryker, Amgen, Zimmer Biomet, Moximed, Bioventus, Merck, Eli Lilly, and sanofi-aventis. Dr Boddaert reports receiving travel accommodations from Novartis. Dr Riou reports receiving personal fees from Crossject, Thermo Fisher Scientific/BRAHMS, LFB, sanofi-aventis, and Sangart. Dr Devereaux reports receiving grant funding from Abbott Diagnostics, Boehringer Ingelheim, Covidien, Octopharma Plasma, Roche Diagnostics, and Stryker. No other disclosures were reported.
Funding/Support: This work is supported by the Hamilton Anesthesia Associates (Dr Le Manach) and the Canadian Network and Centre for Trials Internationally (Dr Le Manach), by grant G1100513 from the United Kingdom Medical Research Council (Dr Collins), and by a Heart and Stroke Foundation of Ontario Career Investigator Award (Dr Devereaux).
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.