Keywords
THA - total hip arthroplasty - subsidence - activity - AM-PAC - DAA
Advancements in surgical techniques and postoperative rehabilitation as well as standardization
of postoperative protocols have led to earlier mobilization and patient discharge
following total hip arthroplasty (THA).[1]
[2]
[3]
[4] As length of stay (LOS) decreases and outpatient arthroplasty becomes increasingly
common, many have sought to quantify the outcomes of these abbreviated stays. However,
most of these analyses on success rates have focused on patient-centric results rather
than clinical outcomes such as stem subsidence.
Femoral stem subsidence following THA has been shown to occur between 1 week and 3
months following surgery with the largest changes occurring within the first 6 weeks.[5]
[6]
[7] After 3 months, little change in subsidence is seen up to 2 years postoperatively.[6]
[8] While subsidence is seen across all patient populations, poor bone quality and aging
have been associated with femoral stem subsidence. Low systemic bone mineral density
(BMD) delays the osseointegration of the stem while osteoporosis and aging have been
well documented as having multiple negative effects on the proximal femur including:
the reduction of bone density of the trochanteric cancellous bone, increased intracortical
porosity, and geometric changes.[9]
[10]
[11]
[12] These findings place older, postmenopausal females at greater risk for stem subsidence.[9]
[10]
As LOS decreases, the degree of loading on the stem following surgery is rapidly increasing
as patients are more rapidly mobilized, discharged home more often, and are encouraged
to move around more at home. While it has been recommended to limit postoperative
activity, there is significant evidence that unrestricted weight bearing following
THA does not adversely affect short- or long-term subsidence in comparison to partial
weight bearing.[6]
[8]
[13]
[14] In addition, others have found that bone ingrowth will occur regardless of postoperative
weight bearing.[15]
[16] These studies showed that the time until full weight bearing status did not affect
postoperative stem subsidence and that initial postoperative stem fixation may play
a larger role. This suggests that subsidence outcomes are influenced less by load
and activity than by the biologic processes that lead to bony ingrowth.
As a greater number of patients have fast-tracked, outpatient THA, concerns over how
this immediate discharge effects patient outcomes rises. Our primary objective in
the present study was to analyze the effect of age and activity on stem subsidence
following THA. We hypothesize that activity will not affect femoral stem subsidence,
but age may have an effect due to the low BMD common in older patients.
Methods
A retrospective study was conducted on patients who underwent a primary cementless
THA at a large urban academic hospital between January 2016 and October 2017. This
study was approved by the institutional review board. These cases were identified
using the Current Procedure Terminology code 27130 and were collected from our institution's
electronic medical record. Inclusion criteria for our study included patients whose
age was greater than or equal to 18 years at the time of their surgery and underwent
a primary THA. Exclusion criteria included patients who were less than 18 years old
or underwent a revision THA. Baseline patient demographic data such as age, body mass
index (BMI), American Society of Anesthesiologists (ASA) score, gender, race, and
smoking status were collected for all qualifying patients. Additional information
including surgical time, stem type, stem alignment, LOS, Boston University Activity
Measure for Post-Acute Care Short Form (AM-PAC) scores, and radiographs were also
collected and analyzed. All data was recorded and deidentified using the Microsoft
Excel software.
Patient Population
Our cohort consisted of 821 consecutive patients who underwent primary cementless
THA. A single surgeon at a tertiary, urban, academic institution performed all surgeries
using a direct anterior approach (DAA). All patients had one of two stem designs (Anthology
[Smith & Nephew] or Accolade [Stryker Orthopedics]). The choice for same-day discharge
(SDD) versus inpatient stay was made through a shared decision-making process between
surgeon and patient.
We retrospectively divided the cohort into two groups: (1) patients who had SDD (n = 255) and (2) patients who had a multiple day inpatient stay (MDS) (n = 566). The SDD cohort consisted of 50% females and 50% males with an average age
of 58.98 ± 8.47 years. The MDS cohort was comprised of 64% females and 36% males with
an average age of 67.17 ± 10.27 years. The SDD cohort was discharged with limited
restrictions regarding activity and weight-bearing status while the MDS cohort had
a longer hospital LOS with limited opportunities to be mobile in the hospital compared
with patients at home. With less assistance, greater independence, and few limitations
concerning activity, SDD patients would have had a higher activity levels compared
with the MDS patients who had their activity limited by hospital environment and staff.
Radiologic Evaluation
Postoperative anterior-posterior radiographic images of the hip were used to measure
stem subsidence. Subsidence was measured through a comparison of most recent radiographs
to immediate postoperative radiographs. The time between immediate postoperative and
follow-up radiographs ranged from 3 months to 2 years. Before measurements were taken,
radiographic calibration utilizing the known femoral head size was completed for each
image. Measuring procedures for each image started by drawing a line parallel to the
long-axis of the femoral stem followed by two perpendicular lines placed at the level
of the femoral stem shoulder and the superior tip of the greater trochanter. The perpendicular
distance between these parallel lines were noted ([Fig. 1]). Differences between immediate postoperative and follow-up images were recorded
as stem subsidence. Radiographic analysis was performed by two trained observers (J.G.
and J.P.), who both agreed on the methodology that would be used throughout the measurement
process. To ensure consistency, they performed alignment and subsidence measurements
for the first 10 patients in the study cohort together. The interrater reliability
of the first 10 measurements as calculated by the Cohen's kappa was 0.719 which demonstrates
substantial agreement between the two raters and is on the higher end of interrater
reliability among other studies examining subsidence.[17]
[18] This same method used to measure subsidence was utilized to assess stem alignment
(neutral, varus, valgus). The degree of varus or valgus angulation was defined as
the angle formed between the central shaft of the stem and the medial or lateral endosteal
cortices, respectively. As previously described in the literature, a stem was categorized
into varus or valgus alignment if the angle deviated ≥ 5 degrees from neutral.[19]
[20]
[21]
Fig. 1 Example of how subsidence measurements were made on radiographic images.
AM-PAC Score Analysis
The Boston University AM-PAC assessment is used to evaluate an individual's ability
to execute important functional activities.[22] We recorded and analyzed the scores of the first inpatient AM-PAC test that each
patient performed. AM-PAC raw scores were converted into Centers for Medicare and
Medicaid Services G-code modifiers based on Mediware's (Mediware, Lenexa, KS) conversion
chart and were subsequently analyzed.[23] Modifier CH, which indicates 0% impairment, was used as our control for comparison.
Other modifiers include CI (at least 1% but less than 20% impaired), CJ (at least
20% but less than 40% impaired), and CK (at least 40% but less than 60% impaired).[23] We had no cases of CL, CM, or CN (60–100% impairment).
Statistical Analysis
Descriptive data are represented as means ± standard deviation. Independent sample,
two-sided t-tests were used to test for significant differences between continuous variables
and chi-square tests for categorical variables. Multivariable linear regression models
were utilized to study the association between independent predictors including age,
gender, BMI, stem type, stem alignment, surgical time, and the dependent variable,
subsidence. Binary regression models were used to examine the relation of the covariates
of age, gender, BMI, stem type, stem alignment, surgical time on the dependent variable,
and LOS. Regression coefficients were calculated with 95% confidence intervals (CIs),
which we used to describe significance of association. A p-value of less than 0.05 was considered to be significant. All statistical analyses
were performed using SPSS v. 25 (IBM Corporation).
Results
The primary objective of this study was to determine if activity and patient age impacted
femoral stem subsidence. In a 22-month period, we identified 821 patients who underwent
a primary DAA THA by a single surgeon with 255 qualifying for the SDD cohort and 566
for the MDS cohort ([Table 1]). Statistically significant demographic differences were observed between the two
groups with a significantly larger percentage of females and older patients in the
MDS cohort (p < 0.001, p < 0.001; [Table 1]). Patients with higher BMIs and those who were more at risk according to the ASA
physical status were also found more often in the MDS group (p < 0.001, p < 0.001; [Table 1]). While there was a statistical difference in surgical time between the two groups
(69.46 ± 10.67 vs. 70.99 ± 14.32 minutes, p < 0.001; [Table 2]) when controlling for differences in demographic data, there was no clinical significance.
Stem type and stem alignment did not significantly differ between cohorts (p = 0.260, p = 0.566; [Table 2]).
Table 1
Patient demographic data for the two cohorts: same-day discharge (SDD) and the greater
than 24-hour discharge groups (MDS)
|
Patient demographics
|
|
Same-day discharge (n = 255)
|
Greater than 24 h (n = 566)
|
p-Value[a]
|
|
Age (y)
|
58.98 ± 8.47
|
67.17 ± 10.27
|
< 0.001
|
|
Gender
|
|
|
< 0.001
|
|
Female
|
127
|
362
|
|
|
Male
|
128
|
204
|
|
|
BMI (kg/m2)
|
26.61 ± 4.01
|
27.65 ± 5.07
|
< 0.001
|
|
ASA
|
|
|
< 0.001
|
|
1
|
30
|
15
|
|
|
2
|
216
|
345
|
|
|
3
|
9
|
197
|
|
|
4
|
0
|
9
|
|
|
Median
|
1.95
|
2.35
|
|
|
Race
|
|
|
0.670
|
|
White
|
232
|
520
|
|
|
Other race
|
23
|
46
|
|
|
Smoking status
|
|
|
0.022
|
|
Current
|
19
|
34
|
|
|
Former
|
78
|
230
|
|
|
Never
|
158
|
302
|
|
|
AM-PAC score
|
|
|
0.006
|
|
CL
|
0
|
3
|
|
|
CK
|
12
|
63
|
|
|
CJ
|
11
|
15
|
|
|
CI
|
12
|
15
|
|
|
CH
|
89
|
164
|
|
Abbreviations: AM-PAC, Activity Measure for Post-Acute Care; ASA, American Society
of Anesthesiologists; BMI, body mass index.
a
p-Values are derived from two-tailed t-test or chi-square tests for categorical values.
Table 2
Surgical characteristics between the same-day discharge (SDD) and the greater than
24-hour discharge groups (MDS)
|
Surgical characteristics
|
|
Same-day discharge (n = 255)
|
Greater than 24 h (n = 566)
|
p-Value[a]
|
|
Surgical time (min)
|
69.46 ± 10.67
|
70.99 ± 14.32
|
< 0.001
|
|
Subsidence (mm)
|
0.916 ± 0.86
|
1.80 ± 1.91
|
< 0.001
|
|
Stem
|
|
|
0.260
|
|
Anthology
|
127
|
362
|
|
|
Accolade
|
128
|
204
|
|
|
Stem alignment
|
|
|
0.566
|
|
Neutral
|
235
|
531
|
|
|
Varus
|
19
|
29
|
|
|
Valgus
|
1
|
6
|
|
Note: Analysis accounts for differences in demographic data.
a
p-Values are derived from a binary logistic and linear regressions.
To account for the statistically significant differences in demographic and surgical
data, linear regressions were utilized to compare stem subsidence with a variety of
factors. The model showed that of all the variables examined, only increases in LOS
and age were found to be associated with higher stem subsidence (p < 0.001, p < 0.001; [Table 3]). AM-PAC scores were not found to correlate with stem subsidence in our model (p = 0.634; [Table 3]).
Table 3
Multivariate linear regression analysis of stem subsidence
|
Linear regression for stem subsidence
|
|
Unstandardized β
|
p-Value[a]
|
|
LOS
|
0.823
|
0.002
|
|
Age
|
0.023
|
0.042
|
|
BMI
|
0.005
|
0.837
|
|
ASA
|
0.115
|
0.621
|
|
Gender
|
0.134
|
0.574
|
|
Race
|
–0.010
|
0.981
|
|
Smoking
|
0.099
|
0.481
|
|
Surgical length
|
–3.341 × 10−5
|
0.997
|
|
Stem
|
0.000
|
0.999
|
|
AM-PAC
|
0.041
|
0.634
|
Abbreviations: AM-PAC, Activity Measure for Post-Acute Care; ASA, American Society
of Anesthesiologists; BMI, body mass index; LOS, length of stay.
a
p-Values are derived from a multivariate linear regression.
Several binary regressions were run to determine if there was a difference in complication
rates between the two cohorts when controlling for differences in the demographic
data. While one fracture was recorded for the SDD cohort and six for the MDS group,
which ranged from 5 days to 1 year postoperatively, models showed that there were
no statistical differences in periprosthetic fracture rates (0.40% vs. 1.09%, p = 0.995; [Table 4]). There were no dislocations in either cohort. The 90-day readmission rates were
also found to be insignificantly different between the groups with only one readmission
for the SDD cohort and 24 for the MDS cohort (0.39 vs. 2.47%, p = 0.533; [Table 4]). Patients who were readmitted within 90 days had a 53.1% decreased odds of being
in the SDD cohort (odds ratio [OR], 0.469; 95% CI; p = 0.533; [Table 4]). Readmission results were fully powered to show a significant difference (power
of 86.27% [α= 0.10]).
Table 4
Logistic regression analysis of complication data post-THA
|
Logistic regression for complication data
|
|
Periprosthetic fractures
|
90-day readmission
|
|
Odds ratio
|
p-Value[a]
|
Odds ratio
|
p-Value[a]
|
|
LOS
|
0.996
|
0.997
|
0.469
|
0.533
|
|
Age
|
1.078
|
0.115
|
1.011
|
0.772
|
|
BMI
|
1.053
|
0.548
|
1.101
|
0.138
|
|
Gender[b]
|
1.972
|
0.447
|
0.709
|
0.709
|
|
Race[c]
|
0
|
0.997
|
0
|
0.997
|
|
Smoking[d]
|
|
0.739
|
|
0.92
|
|
Former smoker
|
6.7 × 106
|
0.996
|
1.954
|
0.519
|
|
Current smoker
|
3.4 × 106
|
0.997
|
1.516
|
0.702
|
|
ASA
|
0.812
|
0.801
|
1.048
|
0.909
|
|
Stem
|
1.869
|
0.436
|
1.014
|
0.225
|
|
Surgical length
|
1.05
|
0.005
|
0.999
|
0.972
|
Abbreviations: ASA, American Society of Anesthesiologists; BMI, body mass index; LOS,
length of stay; THA, total hip arthroplasty.
a
p-Values are derived from binary logistic regressions.
b Female gender was used as the reference.
c White race was used as the reference.
d Nonsmoker was used as the reference.
An additional binary regression was run to determine which variables contributed to
a MDS. Models showed that increased age led to a 9.6% higher OR of being in the MDS
cohort (p < 0.001; [Table 5]). Additionally, higher BMI led to 6% increase in OR for being in the MDS cohort
while being male is associated with a 267% increased OR (p = 0.009; p < 0.001; [Table 2]). Initial high-impairment AM-PAC scores were not associated with longer LOS (p = 0.630; [Table 5]).
Table 5
Logistic regression analysis of whether a patient remained in the hospital for greater
than or less than 24 hours
|
Logistic regression for LOS
|
|
Odds ratio
|
p-Value[a]
|
|
Age
|
1.092
|
< 0.001
|
|
BMI
|
1.070
|
0.046
|
|
ASA
|
9.298
|
< 0.001
|
|
Gender[b]
|
3.668
|
< 0.001
|
|
Race[c]
|
0.921
|
0.866
|
|
Smoking[d]
|
|
0.698
|
|
Former smoker
|
1.541
|
0.505
|
|
Current smoker
|
1.736
|
0.408
|
|
Surgical length
|
1.025
|
0.082
|
|
Stem
|
0.606
|
0.078
|
|
AM-PAC[e]
|
|
0.630
|
|
CL
|
0.340
|
0.560
|
|
CK
|
0.106
|
0.745
|
|
CJ
|
1.059
|
0.303
|
|
CI
|
2.516
|
0.113
|
Abbreviations: AM-PAC, Activity Measure for Post-Acute Care; ASA, American Society
of Anesthesiologists; BMI, body mass index; LOS, length of stay.
a
p-Values are derived from a binary logistic regression.
b Female gender was used as the reference.
c White race was used as the reference.
d Nonsmoker was used as the reference.
e AM-PAC score: “CH” was used as the reference.
Discussion
With population growth and increased life expectancy, the demand for THA is expected
to continue to increase.[19] At the same time, there are institutional and patient-driven pressures for shorter
hospital stays and faster rehabilitation. It has previously been thought that early
mobilization places patients at an increased risk of early femoral stem subsidence.
This can lead to hip instability, periprosthetic fracture, limb length discrepancy,
and other complications, possibly requiring revision surgery. While stem subsidence
is a major concern, much of the published literature on early mobilization following
THA focuses on patient-centered outcomes and do not examine clinical outcomes.[7]
[24] The goal of this study is therefore to report how activity affects stem subsidence
among all patients who undergo a cementless primary THA.
Femoral stem subsidence has been found to usually occur within the first few weeks
following surgery and remains relatively unchanged between the 3-month and 2-year
follow-up.[6]
[7]
[8] As a result, we were able to collect data on a large patient population who had
follow-up radiographs between 3 months and 2 years post-THA. Upon splitting our sample
population into SDD and MDS groups and controlling for differences in demographic
data, we found that LOS was affected by age, with older patients staying in the hospital
for multiple days (OR, 1.092; 95% CI: 1.07–1.12; p < 0.001; [Table 5]). This confirms previous findings which show that older patients either opt for
or are assigned an inpatient admission and are kept longer due to preexisting conditions.[4]
[25] Results also showed that higher BMI and being male is associated with a MDS (OR,
1.070, 95% CI: 1.04–1.11, p = 0.46; OR, 3.668, 95% CI: 2.03–4.55, p < 0.001; [Table 5]). Otero et al showed that obesity and the male gender are significant risk factors
that affects complication rates in total joint arthroplasty (TJA) which could explain
why these patients were more often kept in the hospital for longer period.[26] Other studies corroborate these finding and have shown that obesity and being male
leads to longer LOS and longer inpatient rehabilitation stays following TJA.[27]
[28] Higher ASA scores, which are associated with greater comorbidities, were also shown
to lead to longer LOS ([Table 5]). These findings coincide with other studies which have shown similar results and
correlate this increase in LOS to the fact that these patients are more at risk for
complications.[27]
[28]
When examining the factors which effect subsidence in elective primary DAA THA, we
found that increasing age is related to increased stem subsidence. Published results
on stem subsidence following cementless THA vary greatly with an average range of
subsidence between 0.45 and 2.23 mm.[8]
[29]
[30]
[31]
[32]
[33]
[34] Each cohort consisted of only patients that underwent THA with the DAA and the mean
subsidence in our study was within the accepted range noted in the literature which
suggests that approach does not affect subsidence outcomes. While excessive stem subsidence
was not observed, there was a significant difference between cohorts with the less
active MDS group experiencing greater subsidence than the SDD group (1.80 vs. 0.92mm,
p < 0.001). Following THA, it has been shown that younger patients may have comparable
activity levels to older patients.[24] If mechanical causes were responsible for subsidence, the expectation would be similar
subsidence levels. However, our results showed that aging was associated with a slight
increase in subsidence, which implies that another mechanism is affecting the subsidence
in these older patients (β = 0.023, p < 0.001). Additionally, our findings that increased load due to BMI or activity does
not affect subsidence would also suggest that there is an additional nonmechanical
reason for the increase in subsidence with age (p = 0.837). While there is some evidence that poor bone quality has been overemphasized,
many studies have shown that low BMD, which is associated with aging, will delay osseointegration
of the stem and may result in migration.[9]
[12] Additional aging effects such as postmenopausal status and osteoporosis have also
been linked with increased risk of stem subsidence.[9]
[10] Analysis has shown that decreases in BMD such as those related to aging and osteoporosis
have a negative effect on the proximal femur resulting in the reduction of bone density
of the trochanteric cancellous bone, increased intracortical porosity, and geometric
changes.[11]
[12] While the scope of our study did not examine the causes of stem subsidence, our
results correlated with increased stem subsidence risk associated with comorbidities
that directly relate to an increase in age.
While we found that age has a correlation with stem subsidence in both the SDD and
MDS group, we determined that early activity after surgery, using SDD as a proxy,
does not negatively impact stem subsidence levels, which supports our hypothesis of
a nonmechanical mechanism for subsidence. Patients with a SDD were assumed to be more
active as they would have to be more mobile in a nonclinical setting. We found that
patients who remained in the hospital for longer had higher rates of subsidence than
SDD patients (β = 0.823, p = 0.002). While these findings suggest that increased activity does not lead to increased
stem subsidence, our statistical analyses did not include comorbidities, which may
have contributed to the increased LOS of the MDS cohort. The patients who are chosen
for inpatient admission often have a greater number of coexisting comorbidities and
may have decreased preoperative ambulatory function which could skew our results.[4] Since LOS and activity level are not directly comparable, additional research is
required to confirm the effect of early activity level on stem subsidence as well
as other comorbidities that influence LOS. However, these results suggest that increased
postoperative mobilization following direct anterior THA does not negatively impact
stem subsidence.
In addition to having no negative effect on subsidence, our data showed that postoperative
activity level had little effect on complication data. Results showed no difference
in fracture rates or number of dislocations within 90 days between the two cohorts.
While not fully powered to show a difference in facture rates between cohorts, our
rate of incidence for the MDS group, 1.09%, corresponded with data from the Mayo Clinic
Joint registry which finds an incidence of approximately 1.1% of periprosthetic fractures
in patients who underwent press-fit femur implants as part of a THA, while our SDD
group was less than this accepted rate of incidence, 0.40%.[35] It follows that despite a patient's LOS, the DAA does not negatively impact periprosthetic
fracture incidence rates. Additionally, these results suggest that postoperative activity
levels do not negatively impact fracture rates. We also found that a MDS was not correlated
with a higher likelihood of being readmitted 90 days postoperatively when controlling
for factors that we found to effect LOS including age, gender, BMI, and ASA class
(OR, 0.469; 95% CI: 0.39–1.30, p = 0.533). Several variables that we were unable to take into account, which other
studies find could be affecting readmission rates, include additional comorbidities
and socioeconomic issues.[36] However, our results suggest that activity level does not affect readmission rates.
While we found that increased age is associated with stem subsidence, our research
shows that there are other comorbidities that may contribute to stem subsidence after
THA. Comorbidities that may contribute to poor bony integration of the stem, as often
seen in older patients and patients with longer LOS, may influence these differences.
With further research to identify risk factors for stem subsidence, we can better
select the proper technique for stem fixation for each patient and decrease the risk
of complications after THA. Such improvements have the potential to increase patient
satisfaction and decrease complications and the cost associated with THA.
There are several limitations to this study that should be noted, including its retrospective
design. Additionally, all THA cases were performed by a single surgeon at one institution,
which may limit the generalizability of the study. Only two THA implant types were
utilized, also potentially limiting extrapolation of the results to other stem types.
Additionally, our study is limited by the fact that we do not know exactly how much
activity or mobility the patients in either cohort had following their procedure.
However, our study design attempted to mitigate this limitation as the SDD patients
were not told to limit their activity and mobilization at home while the MDS cohort
had their activity restricted during their hospital stay due to the environment and
staff.
Conclusion
The results of this study demonstrated that when SDD is taken as a proxy for postoperative
activity level, press-fit stems in SDD patients do not subside more than press fit
stems in patients who were in the hospital for multiple days (i.e., less active patient
cohort). Furthermore, we found that the rates of subsidence and periprosthetic fractures
following DAA THA were not higher than those of historical database reports. These
findings suggest that femoral stem subsidence is more dependent on the biologic ability
of the bone to osseointegrate with the implant and with osteoporosis than the extent
of early mechanical loading. We caution surgeons to carefully consider the use of
press-fit implants in the more elderly populations regardless of discharge disposition
or activity level.
