Keywords trauma - knee - DFR - ORIF
Fractures of the distal femur can result from low-energy trauma in the elderly and
are becoming increasingly more prevalent as the population ages.[1 ]
[2 ] In the geriatric population, it is the second most common femur fracture after hip
fractures.[3 ] Fragility fractures of the native distal femur have variable fracture patterns and
high mortality rates in the elderly, commensurate to mortality rates seen in hip fractures.[4 ]
[5 ] In the elderly, distal femur fractures (DFFs) pose a challenge to treatment, especially
with fracture comminution in poor bone quality.[6 ]
[7 ] Current fixation modes include open reduction and internal fixation (ORIF) with
plates and/or intramedullary devices or distal femur replacement (DFR).[6 ]
[7 ]
[8 ]
Much of the present literature on distal femur endoprosthesis involves its use in
the setting of periprosthetic fractures above total knee arthroplasties.[9 ]
[10 ]
[11 ] While the data examining the use of DFR in the treatment of native DFF remain limited,[12 ]
[13 ]
[14 ]
[15 ] the current literature suggests that ORIF and DFR are equally efficacious options
for treating DFF, citing no difference in complication rates between the two treatments.[6 ]
[7 ]
[8 ]
[13 ] While arthroplasty is useful for preserving mobility and eliminating the risk of
nonunion, failure after DFR may leave the patient with limited salvaged options.[7 ]
[14 ] In the geriatric population, another goal of treatment is to restore immediate postoperative
weight-bearing and to allow early ambulation, which has encouraged surgeons to explore
dual-implant ORIF techniques and DFR as alternative options.
It is well understood that early postoperative mobilization in the geriatric population
after fragility fractures is paramount to optimizing outcomes. The purpose of this
study was to compare 30-day outcomes and cost of care following ORIF versus DFR in
patients with native DFF using a 1:2 propensity score matching algorithm in a national
patient database setting. We hypothesized that due to early mobilization allowed following
DFR, patients would have lower morbidity and mortality than patients who underwent
ORIF.
Materials and Methods
Database
The American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP)
was utilized to identify all patients between 2008 and 2019. Data in the ACS-NSQIP
undergo rigorous quality checks from trained reviewers at more than 700 participating
hospitals. Individuals included in the database are prospectively followed for 30
days following surgery. Data collected by these reviewers have been audited and previously
employed in orthopaedic surgery research.[16 ]
[17 ]
[18 ]
Inclusion and Exclusion Criteria
The International Classification of Diseases, 9th Revision-Clinical Modification (ICD-9-CM)
and 10th revision (ICD-10-CM) were used to identify all patients who sustained a DFF
and were a minimum age of 18 years. These codes used are shown in [Supplementary Table S1 ]. Common Procedural Terminology (CPT) codes were used to identify surgical types.
ORIF codes were identified as 27511, 27513, and 27514. DFR was identified as CPT codes
27442, 27443, 27445, and 27447. Exclusion criteria were patients with a diagnosis
code of periprosthetic fracture (ICD-9-CM 996.4x; ICD-10-CM: M97.12, M97.11, T84.049).
This yielded 3,318 cases (ORIF = 3,197 vs. DFR = 121).
Propensity Score Matching and Power Analysis
The analysis was powered to assess differences in rates of complications. Assuming
a mild effect size of 0.2, our power analysis revealed a sample of 196 patients would
be needed to achieve a power of 0.8. A nearest neighbor matching algorithm was executed
for a 1:2 case–control match. Cases were matched to a control based on preoperative
laboratory values, age, sex, race, ethnicity, American Society of Anesthesiologists
physical status score, body mass index (BMI), Charlson comorbidity index, and various
comorbidities ([Fig. 1 ]). A standard mean difference of 0.1 was set as the threshold for adequate balance.
This yielded a final sample size of 363 patients (ORIF = 242 and DFR = 121).
Fig. 1 Love plot demonstrating pre- and post match covariate balance.
Estimating Hospital Cost
Data from the National Inpatient Sample (NIS) database were utilized to estimate costs.
The NIS has been described in previous publications.[19 ] The NIS estimates 95% of all inpatient cases in the United States for a given year.
A multivariate regression model with a 7:3 train-test cross-validation split was conducted
to generate cost estimates. Input variables included year of surgery, length of stay,
race, ethnicity, sex, age, and various comorbidities, including diabetes, congestive
heart failure, chronic lung disease, obesity (BMI more than 30 kg/m2 ), liver disease, and kidney failure. These variables can be found in both NIS and
NSQIP databases. The costs were adjusted to 2019 U.S. dollars using the consumer price
index reported by the U.S. Bureau of Labor Statistics ([Table 1 ]).[20 ]
Table 1
Inpatient cost estimation using data from the Nationwide Inpatient Sample database,
root mean squared error = 12,622.44; mean absolute error = 7,548.665
Beta
p -Value
Intercept
$26,475.40
<0.001
Female sex
−$1,585.64
<0.001
Age
−$49.80
<0.001
Patient has chronic lung disease
−$116.10
0.789
Patient has diabetes
−$690.45
0.08
Patient has liver failure
−$240.71
0.852
Abnormal weight loss before injury
$5,446.23
<0.001
Patient has renal failure
−$560.32
0.2532
Hospital length of stay
$2,504.09
<0.001
Underwent operative reduction and internal fixation (vs. endoprosthesis/arthroplasty)
− $13,956.42
<0.001
End Points and Statistics
The end points of interest included 30-day mortality, 30-day unplanned readmission,
30-day unplanned reoperation, 30-day major complication, 30-day minor complication,
discharge disposition, and all tabulated complications within 30 days. Major complications
were defined previously by Ottesen et al[21 ] and include: 30-day incidence of deep surgical site infection, sepsis, failure to
wean from a ventilator within 48 hours, need for intubation, renal failure, thromboembolic
event (deep vein thrombosis [DVT]/pulmonary embolism), cardiac arrest, myocardial
infarction, and cerebrovascular accident (stroke). Minor complications included superficial
surgical site infection, wound dehiscence, pneumonia, urinary tract infection, postoperative
renal insufficiency, and need for transfusion. Missing variables were handled using
a multiple imputation algorithm.[22 ] A Youden's index was calculated to identify age threshold in which patient was most
at risk for postoperative complications. This age was used to conduct a subgroup comparative
analysis between patients undergoing DFR and ORIF. A chi-square analysis or Fisher's
exact test was conducted to compare categorical variables where appropriate. Independent
samples t -tests were conducted to assess differences in continuous variables. An absolute risk
reduction and number needed to prevent one 30-day mortality event was conducted with
a 95% confidence interval (CI) reported.[23 ]
[24 ] A two-tailed p -value of 0.05 was set as the threshold for statistical significance. Statistical
analysis was conducted in R 3.4.1 (Vienna, Austria).
Results
The 1:2 matching algorithm revealed a good balance between the DFR (treatment) and
ORIF (control) cohorts ([Fig. 1 ]). The resulting sample had a mean age of 69.62 years (standard deviation = 14.21
years) and a sex distribution of 83.7% female and 16.3% male. The sample revealed
no statistical difference between the groups in postoperative diagnosis ICD-9-CM (p = 0.253) and ICD-10-CM (p = 0.212) codes. There was a significant difference in mean total hospital costs (p < 0.001). Patients with DFR had a mean total hospital cost of $46,323 ± $17,022 and
a mean inpatient hospital per day cost of $7,515 ± $5,242), whereas patients undergoing
ORIF had a mean total hospital cost of $20,849 ± $35,021 and a mean inpatient hospital
per day costs of $4,123 ± $1,182 ([Table 2 ]).
Table 2
Comparison of costs and outcomes between match group of patients with distal femur
fractures who underwent DFR versus ORIF
DFR (n = 121)
ORIF (n = 242)
p-value
Mean inflation adjusted inpatient hospital costs (stdev)
$46,322.54 ($17,022.12)
$20,848.74 ($35,021.23)
<0.001
Mean inflation adjusted inpatient hospital costs per hospital day (stdev)
$7,515.02 ($5,241.56)
$4,123.32 ($1,182.27)
<0.001
Median length of stay (interquartile range)
6.5 (5)
5.0 (4)
<0.001
Pneumonia
2.5% (3)
11 (4.5%)
0.335
DVT requiring therapy
6.6% (8)
2.9% (7)
0.268
Septic shock
1.7% (2)
2.1% (5)
0.787
Renal complications
0.8% (1)
0.8% (2)
0.999
Urinary infections
2.5% (3)
5.0% (12)
0.263
Surgical site infection
0.8% (1)
1.2% (3)
0.722
Need for intubation
3.3% (4)
1.2% (3)
0.177
Received at least one transfusion
57.0% (69)
44.2% (107)
0.021
Pulmonary embolism
4.1% (5)
2.1% (5)
0.257
Cardiac arrest
0.8%(1)
0.4% (1)
0.616
Minor complications
5.0% (6)
10.3% (25)
0.084
Major complications
5.8% (7)
4.5% (11)
0.608
Death within 30 d
5.0% (6)
5.4% (13)
0.868
30-d unplanned readmission
8.3% (10)
7.4% (18)
0.781
30-d reoperation
3.3% (4)
2.9% (7)
0.829
Discharge destination
Not recorded
3.3% (4)
5.4% (13)
0.399
Against medical advice
0.8% (1)
0% (0)
Expired
3.3% (4)
1.7% (4)
Home
23.1% (28)
22.7% (55)
Hospice
0% (0)
0% (2)
Rehabilitation
20.7% (25)
19.8% (48)
Separate acute care
0% (0)
2.1% (5)
Skilled care not home
48.8% (59)
47.5% (115)
Abbreviations: DFR, distal femur replacement; DVT, deep vein thrombosis; ORIF, open
reduction and internal fixation; stdev, standard deviation.
bold indicate p -values less than 0.05.
Patients undergoing DFR in our sample population had higher transfusion rates than
patients undergoing ORIF (57.0 vs. 44.2%; p = 0.021; [Table 2 ]). There were no statistical differences in 30-day rates of pneumonia, DVT requiring
therapy, septic shock, renal complications, urinary infection, surgical site infections,
need for intubation, pulmonary embolism, cardiac arrest, minor complications, major
complications, death within 30 days of mortality, 30-day unplanned readmission, and
30-day unplanned reoperation ([Table 2 ]). There were no differences in discharge destination between the groups (p = 0.399).
A subgroup analysis was conducted for our patient sample of patients 80 years of age
and older (DFR n = 21; ORIF n = 33) ([Table 3 ]). There was no difference between ICD-9-CM (p = 0.078) and ICD-10-CM (p = 0.092) coding between the two groups. Similarly, there were no differences in mean
age (86.70 ± 1.36 vs. 87.05 ± 1.28; p = 0.349), mean BMI (26.35 ± 6.18 vs. 27.72 ± 6.92 kg/m2 ; p = 0.450), sex (p = 0.236), race (p = 0.802), and Charlson comorbidity index (p = 0.346). Postoperatively, there were no differences in rates of discharge destination
between the groups (p = 0.264). The DFR group demonstrated a lower 30-day unplanned readmission rate (0
vs. 18.2%, p = 0.038) and a lower 30-day mortality rate (0 vs. 18.2%, p = 0.038) than ORIF. The causes for 30-day readmission included cerebrovascular accident
(n = 1), pneumonia (n = 1), pulmonary embolism (n = 1), urinary tract infection (n = 1), pain (n = 1), and fracture of operative extremity (n = 1). All readmitted patients eventually expired within 30 days from the index operation.
The absolute risk reduction for 30-day mortality for patients in this age cohort undergoing
DFR was 18.8% (95% CI: 5.02–31.34%).
Table 3
Subgroup analysis of patient 80 years of age or older: comparison of costs and outcomes
between match group of patients with distal femur fractures who underwent DFR versus
ORIF
DFR (n = 21)
ORIF (n = 33)
p -Value
Mean inflation adjusted inpatient hospital costs (stdev)
$45,800.20 ($18,598.44)
$16,463.31 ($52,112.63)
0.017
Mean inflation adjusted inpatient hospital costs per hospital day (stdev)
$7,119.06 ($4,635.16)
$3,735.98 ($549.76)
<0.001
Pneumonia
4.8% (1)
12.1% (4)
0.363
DVT requiring therapy
9.5% (2)
6.1% (2)
0.636
Urinary infections
0% (0)
9.1% (3)
0.155
Need for intubation
4.8% (1)
0% (0)
0.206
Received at least one transfusion
66.7% (14)
51.5% (17)
0.272
Pulmonary embolism
9.5% (2)
6.1% (2)
0.636
Minor complications
4.8% (1)
21.2% (7)
0.097
Major complications
4.8% (1)
3.0% (1)
0.743
Death within 30 d
0% (0)
6 (18.2%)
0.038
30-d unplanned readmission
0% (0)
6 (18.2%)
0.038
30-d reoperation
4.8% (1)
0% (0)
0.206
Discharge destination
Not recorded
0% (0)
6.1% (2)
0.264
Against medical advice
4.8% (1)
0% (0)
Expired
0% (0)
6.1% (2)
Home
14.3% (3)
12.1% (4)
Hospice
0% (0)
6.1% (2)
Rehabilitation
9.5% (2)
21.2% (7)
Skilled care not home
71.4% (15)
48.5% (16)
Abbreviations: DFR, distal femur replacement; DVT, deep vein thrombosis; ORIF, open
reduction and internal fixation; stdev, standard deviation.
bold indicate p -values less than 0.05.
Discussion
The present study examined how patients fared after native DFFs nationally and found
that patients who underwent DFR had higher rates of postoperative transfusions. Furthermore,
the average inpatient cost of DFR was more than twice as much compared with ORIF.
In the subset of patients ([Table 3 ]) 80 years of age, DFR resulted in lower rates of 30-day mortality compared with
ORIF.
The appropriate surgical management continues to Garner-wide debate for patients who
sustain DFFs.[6 ]
[7 ]
[10 ]
[15 ]
[25 ] Much of the current literature examines the efficacy of treatment strategies at
restoring baseline functional status, especially in the elderly population who experience
high morbidity and mortality rates after sustaining DFF.[6 ]
[7 ]
[10 ]
[15 ]
[25 ]
[26 ] This study adds to the literature by demonstrating lower 30-day readmission and
mortality rates after DFR than ORIF. However, long-term outcomes in this patient population
remain limited and were beyond the scope of our study. Myers et al demonstrated a
1-year mortality rate of 13.4% in an elderly cohort of both native and periprosthetic
DFFs. They noted a significant negative impact on mortality when surgery was delayed
more than 2 days.[27 ] The authors noted that this rate is lower than established mortality rates after
geriatric hip fractures, which are typically allowed to weight-bearing immediately,
and hence questioned whether postoperative weight-bearing affected outcomes after
DFF.
Early mobilization and functional rehabilitation principles are highly emphasized
after most orthopaedic injuries. In the geriatric hip fracture literature, Heiden
et al[28 ] revealed higher rates of 30-day mortality in patients who were unable to ambulate
by postoperative day 3 compared with patients who mobilized within the first 3 days.
In a prior comparative study, Hart et al[7 ] demonstrated similar reoperation and mortality rates at 1 year between DFR and ORIF
in a geriatric population. The ORIF group, however, had an 18% rate of nonunion and
a 25% rate of remaining wheelchair bound.[7 ] These data underscore that early mobilization principles should also apply to DFF.
The exposure necessary during surgery in addition to required bony resection can carry
a biological cost to the zone of injury greater than that of ORIF, which may explain
the increased transfusions in the postoperative setting. This, along with the cost
of the implant, further contributes to the overall cost burden of treatment, which
are important factors to consider, as cost containment has an increasing role in surgical
decision-making.[29 ]
[30 ] Correspondingly, surgical implants and ORIF techniques have also continued to evolve.[2 ] Periarticular locking plates and retrograde intramedullary nailing provide minimally
invasive techniques that minimize soft tissue stripping, blood loss, and operative
duration.[31 ]
[32 ] Dual implant constructs are also gaining popularity as surgeons seek to improve
fixation strength to allow for early weight-bearing. Despite these advances, many
authors have argued that older patients with poor bone quality would be better candidates
for DFR.[7 ]
[8 ]
[12 ]
[13 ]
[26 ]
[33 ] Furthermore, highly comminuted fractures not amenable to fixation are also indicated
for DFR as high rates of nonunion and malunion have been demonstrated after ORIF.[12 ]
[13 ]
[14 ]
[15 ]
[26 ]
Our study revealed increased costs associated with patients undergoing DFR when compared
with ORIF for DFF. Similar studies report similar findings with much of the cost difference
associated with implant costs. Caines et al[34 ] performed a retrospective cohort study entailing 39 patients with DFFs (OTA/AO 33C
fractures) who either underwent DFR or ORIF. The authors increased direct implant
costs between the groups, with a mean implant cost of $11,403 in the DFR group and
$2,066 in the ORIF group (p = < 0.01). After all attributable hospitalization costs were analyzed, the total
cost of providing index surgical care was also significantly different, with a mean
DFR cost of $61,259, compared with $44,491 in the ORIF group (p = 0.05). Brodke et al[35 ] conducted a systematic review entailing 37 observational studies and 1 randomized
controlled trial to compare costs for patients with DFF who underwent either a DFR
or ORIF. The authors performed a Markov decision analysis model wherein total costs
were estimated by combining facility costs with surgeon fees. Facility costs were
obtained from the Healthcare Cost and Utilization Project's online query system using
2017 data. The authors reported higher costs associated with DFR (mean estimate = $65,536;
$63,790–$67,619) when compared with ORIF (mean estimate = $25,556; $24,230–$27,257).
The present study is inherently limited by its retrospective design, which eliminates
the ability to determine the causality of the findings. This study design also allows
for confounding factors within the data that cannot be accounted for in the analysis.
We attempted to mitigate this bias by utilizing a propensity score matching algorithm
by matching multiple patient-specific and preoperative laboratory variables to identify
an adequate control. In our study, we observed that DFR incurred significantly higher
costs when compared with ORIF. This disparity can be attributed, at least in part,
to variations in implant expenses,[36 ] with DFR being notably more expensive. However, it is important to note that we
encountered a limitation in our analysis, as we could not disaggregate overall care
costs, which hindered our ability to identify specific factors contributing to variations
in care expenses. A further limitation is that fracture types could not be classified.
This was mitigated by using ICD-9 and ICD-10 coding as a surrogate for fracture morphology
to ensure similarities between the groups. Nonetheless, we note that fracture morphology,
BMI, and preexisting arthritis can play a role in surgeon decision-making between
proceeding with DFR versus ORIF. We elected not to include periprosthetic DFFs in
our study given concerns of added heterogeneity and inability to adequately match
periprosthetic DFFs for an adequate analysis. As such, our results cannot be extrapolated
for patients with periprosthetic femur fractures. Finally, using a national database,
as the source of data, prevents monitoring the quality of data reporting provided
by other institutions; however, both the Healthcare Cost and Utilization Project,
Nationwide Inpatient Database, and ACS-NSQIP database have published their rigor of
data collection and outcomes monitoring, thus minimizing this concern.[37 ]
Conclusion
In conclusion, patients who underwent DFR for DFFs in the current study had higher
postoperative transfusion rates and greater inpatient costs than ORIF. Nevertheless,
DFR should be considered in patients 80 years of age, as this group demonstrated lower
30-day readmission and mortality rates than ORIF. While the study contributes to understanding
treatment outcomes after DFF, a prospective clinical trial is necessary to compare
ORIF and DFR for comminuted, intra-articular fractures in the elderly to optimize
their outcomes.
