Translate this page into:
Impact of delayed intervention on clinical outcomes following traumatic hip fracture in the elderly: A national analysis
, Arjun Vermaa, Rhea Rahimtoolaa, Alexandra Stavrakisb, Christopher Leeb, Peyman Benharasha,
∗Corresponding author: Peyman Benharash. PBenharash@mednet.ucla.edu
-
Received: ,
Accepted: ,
This article was originally published by Reed Elsevier India Pvt. Ltd. and was migrated to Scientific Scholar after the change of Publisher.
Abstract
Abstract
The present study sought to evaluate clinical outcomes of delayed intervention following hip fractures. Adults (≥60 years) who underwent operative intervention for hip fracture following traumatic fall were identified using the 2008–2018 National Inpatient Sample. Patients were classified as Delayed if repair was >48 h after admission and otherwise considered Early. Of an estimated 1,942,905 patients, 148,441 (7.6%) were Delayed. Delayed more commonly suffered neck fractures, underwent hip arthroplasty and were managed at low-volume hospitals. After adjustment, delayed operation was associated with greater likelihood of mortality (adjusted odds ratio (AOR): 1.28, 95% CI: 1.17–1.40), studied complications, hospitalization duration and costs.
Keywords
Delayed intervention
National inpatient sample
Traumatic fall
Volume outcome relationship
1 Introduction
Hip fractures are a significant public health issue and are associated with increased mortality, diminished quality of life and increased healthcare resource. Approximately 300,000 individuals are hospitalized for hip fractures each year in the United States.1 Such injuries are responsible for roughly $17 billion in healthcare costs each year owing to the need for inpatient hospitalization, rehabilitation and long-term care.1 With the aging population in developed nations, the overall incidence and associated costs of hip fractures are expected to rise.
With a 30-day mortality of 7–10%, time from hospital admission to surgical intervention has been identified as a critical factor in determining clinical outcomes in patients with hip fracture.2 Prior studies have suggested that intervention within 24–48 h of admission may reduce hospitalization duration, perioperative complications and short-term mortality.2,3 A recent randomized control trial, accelerated surgery versus standard care in hip fracture (HIP ATTACK), evaluated expedited intervention (<6 h) following presentation and found no difference in mortality but lower complication rate compared to standard management.4
However, most prior studies of this subject suffer from limitations stemming from inadequate consideration of adjustment for confounding factors that may result in operative delay. In fact, patients suffering from hip fractures are often elderly and have several comorbidities that may alter the decision for expedited repair. The current population-based study aimed to evaluate temporal trends, clinical outcomes and resource utilization in operative hip fractures at the national level over the past decade. In particular, we hypothesized that after appropriate risk adjustment, delayed intervention, defined as greater than 48 h from admission, would be associated with worse clinical outcomes and increased hospitalization costs.
2 Methods
2.1 Study design
This was a retrospective study of elderly adults (≥60 years old) who underwent operative intervention for hip fracture using the 2008–2018 National Inpatient Sample (NIS). Maintained by the Agency for Healthcare Research and Quality and part of the Healthcare Cost and Utilization Project (HCUP), the NIS is the largest all-payer inpatient database that samples 20% of all United States hospitalizations. Using validated sampling algorithms with discharge and survey weights, the NIS provides accurate national estimates for 97% of all inpatient US hospitalizations.5
2.2 Participants and outcomes
Using International Classification of Diseases 9th and 10th Revision (ICD-9 and ICD-10) codes, patients who sustained hip fractures (femoral neck or intertrochanteric (IT)/subtrochanteric (ST)) following a traumatic fall and underwent non-elective operative fixation (open reduction internal fixation (ORIF)/intramedullary nailing (IMN) or hip arthroplasty (total hip arthroplasty or hemiarthroplasty)) were identified (Supplemental Table A1). To reduce heterogeneity of the cohort, patients were excluded if they sustained head, truncal or vascular injuries, underwent operative fixation >5 days after admission or were transferred from another facility. Furthermore, those with missing key demographic or outcome information were excluded (3.96% of patients). Patients were classified as Early if they underwent repair within 48 h of admission and otherwise considered Delayed. The threshold to define Delayed was selected as 48 h based upon prior studies with sensitivity analysis performed on intervention after 24 h.6
Patient and hospital characteristics were defined using the ICD codes and in accordance with the NIS data dictionary. These variables included age, sex, race, income quartile, hospital teaching status and hospital bed size. The modified Elixhauser comorbidity index (ECI), a previously validated composite score of 30 comorbidities, was used to numerically quantify the burden of chronic conditions.7 Patients were categorized as frail based on the presence of any frailty-defining diagnoses defined by the Johns Hopkins Adjusted Clinical Groups as previously described.8 Hospitals were divided into low-, medium- and high-volume tertiles based upon annual institutional hip fracture repair caseload. The primary outcome of interest was in-hospital mortality while secondary outcomes included perioperative complications, non-home discharge, length of stay (LOS) and hospitalization costs. Complications of interest included stroke, delirium, cardiac-related (arrest, ventricular arrhythmia and tamponade), deep vein thrombosis (DVT), pulmonary embolism (PE), pneumonia, acute kidney injury, urinary tract infection and implant infection. Costs were calculated by application of HCUP hospital-specific cost-to-charge ratios and inflation adjustment to the 2018 Personal Health Care Index.
2.3 Statistical methods
Continuous variables were compared using adjusted Wald test and are reported as means with standard deviations (SD). Categorical variables were compared using chi-square test and reported as proportions. Variables with non-normal distributions such as costs are reported as medians with interquartile range (IQR) and were analyzed using the Mann-Whitney U test. Trends were analyzed using a rank-based non-parametric test developed by Cuzick (NP-trend). Inverse probability of treatment weighing (IPTW) was utilized to adjust for differences in characteristics between the Early and Delayed groups. Similar to propensity-matching, IPTW addresses covariate imbalance but has the advantage of preserving the entire patient cohort in multivariable model adjustment.9 Using patient characteristics, we developed a logistic model to predict delayed operative intervention. Propensity scores were generated and used to create inverse probability treatment weights. Multivariable logistic and linear regression models were developed to evaluate the association of early intervention with outcomes of interest. Elastic Net with retention of clinically-relevant characteristics was used for variable selection. Briefly, Elastic Net utilizes a regressive least squares methodology to select explanatory variables aimed at reducing collinearity while applying penalties to decrease overfitting.10 Optimization of the final model was based on receiver-operating characteristic in addition to Akaike's and Bayesian Information Criteria, when appropriate.
Regression outcomes are reported as adjusted odds ratios (AOR) and beta coefficients (β) for dichotomous and continuous variables, respectively. Statistical significance was defined as α < 0.05. All statistical analyses were performed using Stata 16.0 (StataCorp LP, College Station, TX). The study was deemed exempt from full review by the Institutional Review Board at the present study institution due to the de-identified nature of the dataset.
3 Results
Of an estimated 1,942,905 admissions for isolated hip fracture, 148,441 (7.6%) were Delayed. As demonstrated in Fig. 1, the proportion of patients who underwent delayed repair (9.7% in 2008 to 6.0% in 2018, NP-trend<0.001) as well as overall mortality decreased over time (2.03% in 2008 to 1.42% in 2018, NP-trend<0.001). Compared to Early, patients in the Delayed cohort had a higher ECI and were more often frail while being similar in age (Table 1). Furthermore, Delayed patients were less frequently privately insured, of female sex and of White race while more likely belonging to the lowest income quartile. Upon comparison of patient comorbidities, Delayed had a greater proportion of congestive heart failure, coronary artery disease, chronic lung disease and history of anticoagulation use. Finally, Delayed more frequently underwent hip arthroplasty, suffered femoral neck fractures and were more commonly treated at low-volume centers (8.36% vs 6.62%, p < 0.001) as well as admitted on the weekend (30.2% vs 27.4%, p < 0.001). In regard to univariate outcomes, Delayed had greater rates of in-hospital mortality (2.83% vs 1.59%, p < 0.001), all studied complications and non-home discharge (91.9% vs 88.6%, p < 0.001) (Table 2). Furthermore, resource utilization was greater for Delayed with longer LOS and higher costs.
| Delayed (n = 148,441) | Early (n = 1,794,464) | p value | |
| Age (mean, SD) | 81.5 (8.3) | 81.3 (8.5) | 0.014 |
| Elixhauser Comorbidity Index (mean, SD) | 3.78 (1.77) | 3.04 (1.71) | <0.001 |
| Frail (%) | 35.8 | 32.2 | <0.001 |
| Female (%) | 66.7 | 73.2 | <0.001 |
| Race (%) | <0.001 | ||
| White | 82.1 | 87.6 | |
| Black | 5.15 | 3.40 | |
| Hispanic | 7.82 | 5.02 | |
| Asian and Pacific islander | 2.01 | 1.74 | |
| Other | 2.88 | 2.22 | |
| Income quartile (Percentile, %) | <0.001 | ||
| 76th-100th | 24.8 | 24.5 | |
| 51st-75th | 24.0 | 25.5 | |
| 26th-50th | 25.8 | 26.6 | |
| 0th-25th | 25.3 | 23.4 | |
| Payer status (%) | <0.001 | ||
| Private | 6.16 | 7.60 | |
| Medicare | 89.5 | 88.7 | |
| Medicaid | 2.22 | 1.53 | |
| Other | 2.10 | 2.21 | |
| Medical conditions (%) | |||
| Congestive heart failure | 28.2 | 15.4 | <0.001 |
| Coronary artery disease | 33.4 | 23.6 | <0.001 |
| Arrhythmia | 44.3 | 28.4 | <0.001 |
| Valvular disease | 14.2 | 9.31 | <0.001 |
| Pulmonary hypertension | 8.51 | 4.24 | <0.001 |
| Peripheral vascular disease | 8.37 | 6.38 | <0.001 |
| Hypertension | 73.3 | 71.6 | <0.001 |
| Paralysis | 0.87 | 0.49 | <0.001 |
| Chronic lung disease | 28.7 | 22.1 | <0.001 |
| Diabetes | 25.2 | 21.1 | <0.001 |
| End stage renal disease | 3.44 | 1.70 | <0.001 |
| Liver disease | 2.23 | 1.49 | <0.001 |
| Cancer | 4.44 | 3.41 | <0.001 |
| Coagulopathy | 8.63 | 7.31 | <0.001 |
| Weight loss | 8.75 | 5.74 | <0.001 |
| Electrolyte disorder | 38.6 | 28.8 | <0.001 |
| Psychiatric disorder | 13.6 | 15.2 | <0.001 |
| Anemia | 7.00 | 6.03 | <0.001 |
| Anticoagulation use | 14.8 | 7.87 | <0.001 |
| Osteoporosis | 14.8 | 19.1 | <0.001 |
| Weekend admission | 30.2 | 27.4 | <0.001 |
| Fracture location | <0.001 | ||
| Neck | 52.3 | 48.1 | |
| Trochanteric | 47.7 | 51.9 | |
| Operative type | <0.001 | ||
| Internal fixation/intramedullary nail | 55.8 | 62.8 | |
| Hip replacement | 44.2 | 37.2 | |
| Hospital region (%) | <0.001 | ||
| Northeast | 23.1 | 18.4 | |
| Midwest | 14.2 | 19.9 | |
| South | 44.9 | 41.2 | |
| West | 17.8 | 20.5 | |
| Hospital type (%) | <0.001 | ||
| Urban teaching | 45.8 | 46.1 | |
| Urban non-teaching | 43.9 | 41.5 | |
| Rural | 10.3 | 12.4 | |
| Hospital volume tertile (%) | <0.001 | ||
| Low | 8.36 | 6.62 | |
| Medium | 27.8 | 25.1 | |
| High | 63.8 | 68.3 |
| Unadjusted | |||
| Delayed | Early | p value | |
| Mortality | 2.83 | 1.59 | <0.001 |
| Stroke | 2.08 | 1.32 | <0.001 |
| Delirium | 4.88 | 4.34 | <0.001 |
| Cardiac complication | 2.54 | 1.14 | <0.001 |
| Deep vein thrombosis | 1.20 | 0.43 | <0.001 |
| Pulmonary embolism | 1.07 | 0.51 | <0.001 |
| Pneumonia | 8.71 | 3.66 | <0.001 |
| Acute kidney injury | 17.4 | 10.5 | <0.001 |
| Urinary tract infection | 24.9 | 16.5 | <0.001 |
| Decubitus ulcer | 4.06 | 1.83 | <0.001 |
| Implant infection | 0.08 | 0.02 | <0.001 |
| Non-home discharge | 91.9 | 88.6 | <0.001 |
| Length of stay (days) (IQR) | 7 (6–9) | 5 (4–6) | <0.001 |
| Cost (USD $1000) (IQR) | 19.8 (15.4–26.4) | 14.8 (11.7–19.3) | <0.001 |
After adjustment, several factors remained associated with delayed repair and are shown in Fig. 2. Anticoagulation use (AOR: 1.84, 95% CI: 1.77–1.92), congestive heart failure (AOR: 1.57, 95% CI: 1.52–1.63), hip arthroplasty (IT/ST fracture AOR: 1.76, 95% CI: 1.62–1.90, Neck fracture AOR: 1.36, 95% CI: 1.32–1.40, reference: ORIF/IMN) and weekend admission (AOR: 1.17, 95% CI: 1.14–1.20), among others were associated with increased odds of delayed intervention. Meanwhile, female sex (AOR: 0.86, 95% CI: 0.83–0.88) as well as medium- (AOR: 0.76, 95% CI: 0.70–0.82, Ref: low) and high-volume status (AOR: 0.58, 95% CI: 0.53–0.62, Ref: low) were associated with decreased likelihood of delayed intervention.
Following IPTW, delayed operation was associated with greater odds of mortality (AOR: 1.28, 95% CI: 1.17–1.40). As shown in Fig. 3, operations occurring after the second hospital day were associated with increasing likelihood of death (Reference: day zero). Additionally, delayed operation was linked to increased odds of all studied complications except for PE (Fig. 4). In regard to resource utilization and patient disposition, delayed intervention was associated with incremental increases in LOS (β: +2.6 days, 95% CI: 2.5–2.6), costs (β: +$4.9K, 95% CI: 4.7–5.0) and rates of non-home discharge (AOR: 1.33, 95% CI: 1.26–1.40). Sensitivity analysis was performed on those with delayed intervention defined as after 24 h of admission and similar findings were observed (Supplemental Digital Table B1).
Subgroup analysis was performed on patients who underwent delayed intervention. Factors associated with mortality following delayed intervention included congestive heart failure (AOR: 2.12, 95% CI: 1.77–2.54), end stage renal disease (AOR: 4.39, 95% CI: 3.21–6.02) and liver disease (AOR: 3.43, 95% CI: 2.43–4.85). Notably, the risk of mortality was similar across race, income quartile, insurance status and hospital volume tertiles.
4 Discussion
With an aging population in the United States, hip fractures in the elderly have grown in significance. The optimal timing for operative intervention remains controversial with some suggesting early surgery to reduce hospital acquired conditions and costs.3,6 Using a large nationally-representative cohort, the present study sought to evaluate temporal trends, clinical outcomes and resource utilization in operative hip fractures following traumatic falls. After attempting to control inherent selection bias through IPTW, we found operations occurring greater than 48 h after admission were associated with greater mortality, perioperative complications, resource utilization and non-home discharge. Furthermore, we noted a decrease in delayed operations and overall mortality over time as well as associated patient and hospital factors with delayed intervention. Finally, while higher volume hospitals were less likely to perform delayed intervention, mortality was similar across hospitals.
In contrast to a study from the National Surgical Quality Improvement Program database which noted no association between operative delay and 30-day mortality, the present study, with nearly two million patients, noted that intervention within 48 h of admission was observed to have lower odds of in-hospital mortality compared to surgery performed afterwards.6 Notably, we found no significant difference between those intervened between days zero to two. An important consideration is that this reflects in-hospital mortality and may not necessarily reflect short- or long-term mortality risk. Regardless, the strength of our findings lies in utilization of the largest national cohort of operative hip fracture and IPTW-propensity methodology to reduce patient selection bias. In addition, we observed significantly greater risks of all perioperative complications except PE in delayed intervention. Our findings add to the growing body of literature regarding the morbidity of such delays.11,12
Furthermore, resource utilization was significantly greater in those who underwent delayed intervention. With an aging population, the incidence of hip fractures is expected to increase, resulting in greater in-hospital and rehabilitation financial burden.13 Therefore, emphasis should be placed on prevention and cost-conscious management at all levels of care. Our observations expand upon the growing body of literature evaluating the importance of time to operation. Although not widely adopted as a national guideline in the United States, early intervention has been linked as a reimbursement incentive in the UK.14 In conjunction with other measures, its national implementation has led to subsequent improvement in 30-day mortality. While alternative reimbursement methods such as bundled payment models have been utilized in elective hip and knee replacements in the United States, it has not yet seen widespread utilization following hip fractures. Future work incentivizing early intervention and best practices is needed.
Furthermore, a number of factors were observed to be associated with delayed intervention. As expected, patients with significant comorbidities were more likely to undergo delayed repair. In particular, those with a history of anticoagulation use were associated with delayed intervention, signifying concerns regarding surgical bleeding. Furthermore, those who underwent hip arthroplasty were also associated with delayed intervention compared to nail fixation. This may reflect specialized surgeon availability as well as greater associated blood loss and longer operative time compared to closed reduction percutaneous pinning or intramedullary nailing. Providers may therefore be more likely to delay surgery to ensure preoperative medical optimization. Consequently, coordinated teams are needed to expedite care and limit operative delay. However, the impact of a multi-specialty team in the care of hip fracture patients remain inconclusive and highlight the need for future study.
Finally, higher volume hospitals were associated with greater likelihood of early intervention. These hospitals may have greater overall resources and more established care pathways to optimize patients more expeditiously. Conversely, low volume centers may have greater delays due to lack of dedicated orthopedic operating room teams or available orthopedic traumatologists. Despite more timely intervention at high volume institutions, volume was not implicated in death following delayed procedures. The volume-outcome relationship in hip fractures have been inconclusive, but the majority of studies report non-inferior mortality across volume statuses.15 Nevertheless, there are likely other contributory factors to hospital performance rather than time to intervention. Further work is needed to delineate factors associated with high-performing hospitals in hip fracture management.
There were several limitations to the present study related to its retrospective design and nature of data. As with any administrative database, inter-coder variation and reliance on ICD-9/10 coding are key limitations. Although we limited the timing of procedure to within 5 days of admission, we are unable to ascertain if an exacerbation of pre-existing comorbidities contributed to operative delay. Early ambulation practices following intervention, which have been shown to impact outcomes, are unable to be ascertained. Additionally, organizational causes of delay such as operating room availability represent possible confounders that were unable to be adjusted. Finally, patient quality of life as well as the impact of early intervention on readmission risk were beyond the scope of this study. Regardless, we presented the largest operative hip fracture cohort to date and utilized validated methodology to report outcomes in order to limit bias.
In conclusion, delayed intervention, while decreasing in practice, is associated with significantly worse mortality, perioperative complications and resource utilization. When feasible, patient optimization and operative correction should be performed expeditiously within the first 48 h to reduce morbidity and resource burden. Future study is needed to determine best practices in preoperative multidisciplinary care and resource allocation.
Source of funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
References
- Epidemiology and social costs of hip fracture. Injury. August 2018;49(8):1458-1460.
- [Google Scholar]
- Association between wait time and 30-day mortality in adults undergoing hip fracture surgery. JAMA, J Am Med Assoc. November 2017;318(20):1994-2003.
- [Google Scholar]
- Timing of surgery for hip fractures: a systematic review of 52 published studies involving 291,413 patients. Injury. July 2009;40(7):692-697.
- [Google Scholar]
- Accelerated surgery versus standard care in hip fracture (HIP ATTACK): an international, randomised, controlled trial. Lancet. February 2020;395(10225):698-708.
- [Google Scholar]
- Delay in hip fracture surgery prolongs postoperative hospital length of stay but does not adversely affect outcomes at 30 days. J Orthop Trauma. December 2018;32(12):629-633.
- [Google Scholar]
- Impact of opioid use disorders on outcomes and readmission following cardiac operations. Heart October 2020:1-7.
- [Google Scholar]
- Regularization and variable selection via the elastic net. J R Stat Soc Ser B (Statistical Methodol. April 2005;67(2):301-320.
- [Google Scholar]
- Surgery for a fracture of the hip within 24 hours of admission is independently associated with reduced short-term post-operative complications. Bone Jt J. September 2017;99B(9):1216-1222.
- [Google Scholar]
- Effects of delayed hip fracture surgery on mortality and morbidity in elderly patients. Clin Orthop Relat Res. January 2011;469(11):3218-3221.
- [Google Scholar]
- Hip fracture trends in the United States, 2002 to 2015. Osteoporos Int. March 2018;29(3):717-722.
- [Google Scholar]
- 2017
- [Google Scholar]
- The impact of surgeon volume and hospital volume on postoperative mortality and morbidity after hip fractures: a systematic review. Int J Surg. June 2018;54:316-327.
- [Google Scholar]