Characteristics and Outcomes of Nonoperatively Managed Patients With Hip Fracture Using the Dutch Hip Fracture Audit

OBJECTIVES: To identify and compare characteristics of patients with hip fracture treated nonoperatively versus those treated operatively. METHODS: Design: Retrospective cohort study. Setting: Hip fracture population–based study. Patient Selection Criteria: All adult patients with hip fractures (OTA/AO 31A and 31B) were included. Patients with pathological or periprosthetic hip fractures were excluded. Outcome Measures and Comparisons: Patients were categorized according to the type of management (operative vs. nonoperative) and type of fracture (nondisplaced vs. other). Patient and fracture characteristics associated with nonoperative management (NOM) were analyzed. RESULTS: A total of 94,930 patients with hip fracture were included. Of these, 3.2% were treated nonoperatively. Patients receiving NOM were older [86 years (interquartile range, 79–91 years) vs. 81 years (interquartile range, 72–87 years); P < 0.001], more frequently institutionalized (42.4% vs. 17.6%), and were more dependent in activities of daily living (22.2% vs. 55.0%). Various clinical characteristics, including dementia [odds ratio (OR) 1.31 (95% confidence interval, CI, 1.18–1.45) P < 0.001], no functional mobility [OR 4.39 (95% CI, 3.14–3.68) P < 0.001], and activities of daily living (ADL) measured as KATZ-6-ADL [OR 1.17 (95% CI, 1.14–1.20) P < 0.001] were independently associated with NOM. Seven-day mortality was 37.6%, and 30-day mortality was 57.1% in patients treated nonoperatively. CONCLUSIONS: The first step in understanding patients who potentially benefit from NOM is evaluating the current standard of care. This study provides insight into the current hip fracture population treated nonoperatively. These patients are older, have higher percentage of dementia, more dependent, and show higher short-term mortality rates. LEVEL OF EVIDENCE: Therapeutic Level III. See Instructions for Authors for a complete description of levels of evidence.


INTRODUCTION
Many patients with hip fracture do not live independently.Furthermore, high mortality rates highlight the frailty of this population.Hospital mortality is estimated to range from 3% to 7%, and 30-day mortality is reported up to 10%. 1,2Compared with the general population, a 3-fold increase in mortality is seen in patients with hip fracture. 3atients with hip fracture with a high frailty score experience a 4-fold higher risk of 30-day mortality. 4General practice focuses on early surgical fixation to achieve pain relief and restore mobility.Despite surgical management, only 40%-60% of the patients regain their prefracture mobility state, and 40%-70% regain their prefracture independence level. 5- 7][10] However, which part of the hip fracture population may benefit from operative management is not known.In addition, little is known about the patient characteristics of patients selected for NOM.
The reason for NOM may vary depending on patient and fracture characteristics.NOM may be the best option for patients who are too unstable to undergo surgery or when the primary goal of treatment, regaining mobility, is not to be expected or when patients express a preference for NOM.In these cases, the focus shifts to maintaining quality of life and reducing pain.Currently, the Dutch guideline leaves no room for NOM for patients with a life expectancy of 6 weeks or more. 113][14] Because these fractures are expected to be stable, safe mobilization can be achieved, potentially yielding equivalent outcomes without surgery. 15,16The Dutch guideline recommends operative management explicitly for these patients classified as American Society of Anesthesiologists 1-2. 12 On the other hand, the National Institute for Health and Care Excellence guideline suggests that more research is needed. 13The American guideline presents multiple treatment options, commenting that current evidence is limited.Overall, NOM may be a viable option for a select subgroup depending on specific patient and fracture characteristics and the main goal of treatment.
The characteristics and outcomes of the population treated nonoperatively are not entirely understood.Therefore, the aim of this study was to compare patient and fracture characteristics and outcomes in patients with hip fracture treated nonoperatively versus operatively using Dutch guidelines, with a focus on identifying factors associated with NOM.

Study Design, Setting, and Participants
A retrospective cohort study was conducted using a nationwide hip fracture database.The data were obtained from the Dutch Hip Fracture Audit (DHFA).This study was designed and reported following the Strengthening the Reporting of Observational Studies in Epidemiology guidelines. 17Adult patients diagnosed with a femoral neck or intertrochanteric fracture (OTA/AO 31A and 31B12) between January 1, 2016, and January 1, 2023, in the Netherlands were included. 18Patients with a periprosthetic hip fracture or pathological hip fracture were not included.The DHFA is a nationwide registration that includes patients with hip fracture in the Netherlands undergoing operative or NOM.More than 95% of all hospitals in the Netherlands are included in the DHFA, resulting in a coverage of more than 85% of all patients with hip fracture. 19Hospitals provide data through a secure survey system completed by a clinician or batch data processing.Data processing has several built-in validation processes, and previous data verification has shown high completeness and accuracy of the data. 19

Variables
Data on discharge destination, mobility at the time of discharge, length of hospital stay, and all-cause mortality at 7 days, 30 days, 3 months, and 1 year were extracted from the database.All-cause mortality was stratified based on nondisplaced femoral neck fractures and other types of fractures and measured in patients included between 2016 and 2021 to allow the measurement of 3-month and 1-year mortality.Dates of death were calculated using the data from health insurance reimbursements by a third party using social security numbers.Therefore, patients were excluded from further mortality analysis if the social security number was unknown.Mobility at the time of discharge was available up until 2021.
Data on age, gender, fracture type, prefracture mobility score, dementia, osteoporosis, activities of daily living (ADL) score, risk of malnutrition, and prehospital living situation were extracted from the database.The presence of dementia and osteoporosis was confirmed if diagnosed by a general practitioner or documented in the records of the treating hospital.ADL score was calculated using the KATZ-ADL score, previously validated for the Dutch language. 20The risk of malnutrition was classified as high, medium, or no risk according to the Short Nutritional Assessment Questionnaire or Malnutrition Universal Screening Tool. 21,22Mobility score was calculated using the fracture mobility score, previously validated in patients with hip fracture. 23

Statistical Methods
Categorical variables are shown as percentages and exact numbers.Continuous variables are shown as median and interquartile range (IQR).Baseline characteristics between the operative and nonoperative groups were tested for statistical significance using Pearson x 2 test or Mann-Whitney U test.Within the NOM patient group, characteristics of patients with a nondisplaced femoral neck fracture and patients with other types of fractures were compared using the same tests.Statistical significance was set at a P value of #0.05.
Missing data varied between the variables and was up to 20%.Multivariate imputations by chained equations were used to impute the missing data. 24It was assumed that variables were missing at random.The demographic variables and the outcome variable (type of treatment) were used as predictors for imputation of missing data.The Bayesian polytomous regression model was used for categorical variables in the imputation.Univariate logistic regression analysis was used for dichotomous variables and predictive mean matching for continuous variables.A total of 20 imputed data sets were created.
Multivariable regression analysis was performed to investigate the potential associations between patient and fracture characteristics and NOM.For the selection of variables used in the multivariable logistic regression, a univariable logistic regression was performed.Variables were selected for multivariable regression analysis if the P value was less than 0.05.All variables included in the multivariable regression were adjusted for the other.The odds ratio (OR) and 95% confidence interval (95% CI) were reported.
Statistical analysis was performed using R Studio Version 4.2.1, using the "Dplyr," "TableOne," and "ggplot2" packages.The "Mice" package was used for multiple imputations and logistic regression.

Nondisplaced Femoral Neck Fractures Versus Other Fractures
Within the NOM group, 885 patients had a nondisplaced femoral neck fracture (Table 3).Compared with patients with other types of fractures, this group was younger, with a median age of 84 years (IQR 76-90 years; P , 0.001), had higher mobility scores before the fracture (28.2% not using any mobility aid vs. 16.3%,P , 0.001), and less independent in ADL (31.8% vs. 18.2%,P , 0.001).In the group with the nondisplaced femoral neck fracture, more patients lived independently at home (35.4%) compared with those with displaced fractures where a larger part of the patients were institutionalized 42.4% (P , 0.001).

PATIENT OUTCOMES
Table 4 provides an overview of the outcome measures of all nonoperative patients.Most patients were discharged to an institution (57.4%) and 43.8% had no functional mobility at discharge.The median hospital stay was 1 day (IQR 1-4 days).All-cause mortality was 36.4% in the first 7 days and increased to 56.9% at 30-day follow-up (Table 5).Three months and 1-year mortality was 62.8% and 68.8%, respectively.All-cause mortality rates at 7 and 30 days were 23.3% and 38.6% (P , 0.001) for nondisplaced femoral neck fractures and were higher than those seen for other fracture types, 46.6% and 69.1% (P , 0.001).

DISCUSSION
This Dutch registry-based cohort study showed an overall incidence of NOM of 3.2%.Multivariate regression analysis showed that multiple clinical characteristics were independently associated with NOM.These patients were older, had poorer prefracture mobility, were more dependent on ADL, were more likely to come from a nursing home, and have dementia than those who underwent surgical treatment for their hip fracture.These results should be interpreted in light of the Dutch guideline stating the explicit advice to treat patients operatively with a life expectancy of 6 weeks or more.The patient group treated nonoperatively with nondisplaced femoral neck fractures was relatively fitter in terms of mobility and ADL and had lower all-cause mortality than those with other types of fractures.

Patient Characteristics and Outcomes
This study showed that age, prefracture living situation, mobility, KATZ-ADL, and dementia were associated with NOM.A systematic review reported percentages varying between 19% and 61% of dementia. 25This review included numerous studies with heterogeneous cohorts and potentially different reasons for NOM, which may lead to variation in percentages of dementia.One previous study in which clinical characteristics associated with NOM were identified used a comparable method to this study. 26After analyzing 1307 patients, the following factors were identified: age .90years, sex, nursing home, and pertrochanteric fracture, similar to this study.Furthermore, this study identified higher dependence in ADL and higher risk of malnutrition in patients treated nonoperatively, which has not been described previously.
All-cause mortality in all patients treated nonoperatively in this study was high: 36.4% within 7 days and 62.8% within 90 days.The recommendations in the Dutch guideline that the population treated nonoperatively should have a life expectancy of ,6 weeks likely explain the high mortality rate identified in this study.By contrast, previous studies reported lower 7-day, 3-month, and 1-year mortality rates (30 days: 15%-36.7%),5][26][27][28] In previous studies, patients were younger, less institutionalized, and had a lower prevalence of dementia compared with this study.In the systematic review by Loggers et al, one-third of the patients were treated nonoperatively for nonmedical reasons, such as economic reasons or proxy preferences, which may contribute to the lower mortality rate.Variability in current guidelines regarding NOM, which range from the absence of recommendations to always

Incidence
This study showed an overall prevalence of NOM of 3.2%, which is similar compared with the study by Johansen et al (1%-4%). 29By contrast, an Estonian population-based retrospective study from 2009 to 2017 reported a high incidence of 10%. 9However, no national guidelines were available for hip fracture treatment, which could explain the high incidence in Estonia at that time.In addition, in a follow-up study also performed by Prommik et al, a relatively high percentage of patients refused operative management (8%) or received NOM due to an expected good prognosis (15%). 30In 75.5% of the patients, NOM was the chosen type due to a poor expected prognosis.
This study may be subject to underregistration of nonoperatively treated hip fractures due to registration bias, leading to a lower reported prevalence.Because this study incorporated patients from 2016 to include a large number of patients, it is worth noting that data quality has improved over time. 19As a result, the underregistration risk may be more present in the initial years.Therefore, the actual incidence of NOM is likely higher than presented in this study.

Fracture Characteristics
Patients who received NOM for nondisplaced femoral neck fractures had favorable characteristics compared with the patients with other types of fractures (femoral neck, dislocated, trochanteric type AO-A1, AO-A2, AO-A3, and subtrochanteric); they were younger, had better prefracture mobility, and were more independent.This fracture type is unique because it shares similar treatment goals for operative management and NOM in limited cases with comparable functional outcomes.The favorable patient characteristics in this specific group might suggest that the aim of NOM is not solely pain relief, supported by the lower percentage of mortality and patient characteristics at 90 days in this group but functional recovery with regaining mobility.Although current practice leans toward operative management for this patient group, the discussion remains centered around the choice of management when international guidelines are considered.The Dutch guideline leaves no room for NOM in this patient group, explicitly stating the need for operative management.However, this study showed that a percentage of patients with nondisplaced femoral neck fractures were treated nonoperatively despite the current guideline.The deviation from the current guideline might be due to international guidelines or past habits.The risk of NOM in nondisplaced femoral neck fractures is avascular necrosis, secondary displacement, and nonunion.A study by Amsellem et al showed a percentage of secondary displacement of 30%, meaning that 70% of the patients with a Garden I fracture were successfully treated nonoperatively. 15They identified various predictors for secondary displacement: dementia, institutionalization, multiple comorbidities, and a history of repeated falls, suggesting that vital patients with a nondisplaced femoral neck fracture in a subgroup might be treated nonoperatively with a similar outcome.This mirrors the patient group identified in this study with nondisplaced femoral neck fractures because the characteristics of frailty align with the subgroup of patients at risk of NOM complications in nondisplaced femoral neck fractures.In summary, this subgroup of patients has distinct characteristics that may influence the diversity in clinical decisions.This is reflected by the difference in outcomes, making it a distinctive type of fracture that should be taken into consideration when determining treatment and providing counseling.
This study has several strengths.First, this nationwide study included a high number of patients treated either operatively or nonoperatively.Second, the data quality was maintained for the date of death due to linking registry data to declaration data.Third, this registry is a comprehensive registration, including a wide variety of clinical characteristics.This includes ADL and risk of malnutrition, which have never been described before in this context.It is noteworthy that these variables are of substantial importance for patients. 31,32Moreover, risk of malnutrition is important  because it is associated with higher mortality in patients with hip fracture and has been linked to lower rehabilitation efficiency and outcome. 33,34ecause of the retrospective nature of this registrybased study, there are also several limitations.First, because of the retrospective design, it is impossible to add any variables that may be of interest, including follow-up data.Consequently, the reasons for NOM were not known in this study, leading to the inclusion of a heterogeneous population.This varied from fit patients undergoing nonoperative treatment to regain mobility, to frail patients with a limited life expectancy.Second, registry data are subject to registration bias due to the quality of the registration by the hospitals, which may influence the quality of the data in the registry.However, because of the registry's existence for several years and improvement in the data quality, the risk of registration bias might be limited.Third, cohort studies are subject to missing data; in this study, missing data were up to 20% in clinical characteristics and up to 26% in outcomes.However, because multiple imputations could be used, which minimized the expected selection bias usually caused by missing data.
Identifying patients for whom NOM might be a suitable choice remains complex for both patients and clinicians.To gain valuable insight into the complex hip fracture population who will benefit from NOM, the initial step involves understanding the standard of our care for a large number of patients.Future studies should assess and compare patientrelated measurements, including follow-up assessments between patients treated nonoperatively and operatively.This may serve as a foundational basis for developing patient-centered treatment strategies and optimizing outcomes in the context of NOM.
In conclusion, this study is the first to provide a clear insight into the current hip fracture population treated nonoperatively in the Netherlands and identified various patient and fracture characteristics independently associated with NOM.Patients in the nonoperative group were older, more dependent, and showed high mortality rates at 30 and 90 days.By contrast, patients with a nondisplaced femoral neck fracture treated nonoperatively were younger, less dependent, and have lower mortality rates compared with patients with other types of fractures.

TABLE 1 .
Clinical and Fracture Characteristics of Patients With Proximal Femur Fracture Treated Operatively and NonOperatively

TABLE 2 .
Clinical Characteristics Associated With Nonoperative Management

TABLE 3 .
Baseline Characteristics Stratified to Fracture Type (Nondisplaced Proximal Femoral Neck Fracture and Other Types of Fractures)

TABLE 4 .
Clinical Outcome Measures in Nonoperatively Treated Patients with Hip Fracture

TABLE 5 .
All-Cause Mortality in Nonoperatively Treated Patients with Hip Fracture