Journal Pre-proof Does Femoral Component Cementation Affect Costs or Clinical Outcomes After Hip Arthroplasty in Medicare Patients? Jason H. Oh, MD, William Yang, BS, Tara Moore, MSc, Kristina Dushaj, MA, H. John Cooper, MD, Matthew S. Hepinstall, MD PII:
S0883-5403(20)30072-3
DOI:
https://doi.org/10.1016/j.arth.2020.01.035
Reference:
YARTH 57743
To appear in:
The Journal of Arthroplasty
Received Date: 19 August 2019 Revised Date:
23 December 2019
Accepted Date: 12 January 2020
Please cite this article as: Oh JH, Yang W, Moore T, Dushaj K, Cooper HJ, Hepinstall MS, Does Femoral Component Cementation Affect Costs or Clinical Outcomes After Hip Arthroplasty in Medicare Patients?, The Journal of Arthroplasty (2020), doi: https://doi.org/10.1016/j.arth.2020.01.035. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2020 Elsevier Inc. All rights reserved.
TITLE: Does Femoral Component Cementation Affect Costs or Clinical Outcomes After Hip Arthroplasty in Medicare Patients?
AUTHORS: Jason H. Oh MD1, William Yang BS1, Tara Moore MSc1, Kristina Dushaj MA1, H. John Cooper MD2, Matthew S. Hepinstall MD1
1
Department of Orthopaedic Surgery
Lenox Hill Hospital 130 E 77th St., 11th Floor New York NY 10075
2
Department of Orthopaedic Surgery
Columbia University Medical Center 622 W 168th St., PH 11 – Center New York NY 10032
CORRESPONDING AUTHOR: Jason H. Oh, MD Department of Orthopaedic Surgery Lenox Hill Hospital 130 E 77th St., 11th Floor New York NY 10075 Phone: (212) 434-6880 Fax: (212) 434-2268 Email:
[email protected]
1
TITLE:
2
Does Femoral Component Cementation Affect Costs or Clinical Outcomes After Hip Arthroplasty in
3
Medicare Patients?
4 5
Abstract
6 7
Background: Bundled payment initiatives were introduced to reduce costs and improve quality of care.
8
Cemented versus cementless femoral fixation is a modifiable variable that may influence the cost and
9
quality of care. New bundled payments data from the Centers for Medicare and Medicaid Services
10
(CMS) allowed us to study the influence of femoral fixation strategy on 90-day costs and clinical
11
outcomes.
12 13
Questions/purposes: How does the method of femoral fixation affect (1) 90-day costs; (2) readmission
14
rates; (3) reoperation rates; (4) length of stay (LOS); and (5) discharge disposition for Medicare patients
15
undergoing THA?
16 17
Methods: We retrospectively studied 1671 primary THA Medicare cases, comparing 359 patients who
18
received cemented femoral fixation to 1312 who received cementless fixation. CMS cost data as well as
19
clinical data were reviewed. Demographic differences were present between the two cohorts. Statistical
20
analyses were performed, including multiple regression models to adjust for baseline differences.
21 22
Results: Controlling for cohort differences, cemented patients were significantly more likely to be
23
discharged home compared to cementless patients. Cemented patients also demonstrated trends
24
toward lower costs, lower readmission rates, and shorter LOS compared to cementless patients. All
25
reoperations within the early postoperative period occurred in patients managed with cementless
26
femoral fixation.
27 28
Conclusion: Among Medicare patients, cemented femoral fixation outperformed cementless fixation
29
with respect to discharge disposition and also trended toward superiority with regards to LOS,
30
readmission, cost of care, and reoperation. Cemented femoral fixation remains relevant and useful
31
despite the rising popularity of cementless fixation.
32 1
33
Keywords: Medicare, bundled payments, hip, arthroplasty, cement, cost, outcomes
34 35
Introduction
36
The rising demand for total hip arthroplasty (THA)[1] has been met with increasing scrutiny from the US
37
government. The Bundled Payment for Care Improvement (BPCI) initiative and the Comprehensive Care
38
for Joint Replacement (CJR) model were introduced by the Centers for Medicare and Medicaid Services
39
(CMS) in 2013 and 2016, respectively, to rein in costs and improve quality of care. Orthopaedic surgeons
40
have thus been challenged to maximize quality metrics while minimizing costs of delivering care to the
41
hip arthroplasty patient, particularly within the 90-day episode of care.
42
Femoral fixation strategy is one variable that may influence the cost and quality of care. Utilization of
43
cemented versus cementless femoral components varies widely by geography and may be influenced by
44
non-patient factors such as surgeon training, familiarity and preference. As such, it represents a
45
possible target for quality improvement under the BPCI and CJR initiatives. There has been increasing
46
use of cementless components in recent decades with utilization estimated as high as 86-93% in the
47
United States[2,3]. Proponents of cementless femoral fixation cite reduced operative time and blood
48
loss, as well as the potential for biological integration, as primary advantages[4,5]. Others tout
49
avoidance of potential complications relating to cement use, such as pulmonary complications,
50
thromboembolic events, and early perioperative mortality[6–12]. When possible future revision surgery
51
is considered, many arthroplasty surgeons prefer to avoid the occasionally arduous task of thorough
52
cement removal. Nevertheless, the increasing prevalence of cementless femoral fixation runs counter to
53
established registry data showing fewer early reoperations and superior survival with use of cemented
54
components[2]. In addition, many reports have refuted the claims of increased cardiopulmonary
55
complications, thromboembolic events, and mortality among cemented patients, assuming the use of
56
modern surgical technique and appropriate medical co-management[13–22]. Furthermore, it has been
57
well established that cemented fixation is superior for the treatment of proximal femur fractures in frail
58
and elderly patients with osteoporosis[23–28]. The American Academy of Orthopaedic Surgeons (AAOS)
59
has issued a moderate-strength recommendation to use cemented fixation in arthroplasty treatment of
60
displaced femoral neck fractures[29].
61
Given the current spotlight on cost-effective care that optimizes quality while preventing complications,
62
it is possible that the trend towards increasing utilization of cementless fixation in patients over the age
63
of 65 should be reconsidered. We sought to utilize recently available CMS data on groups of BPCI and
64
CJR patients in our healthcare system to assess whether cemented femoral fixation was associated with 2
65
(1) lower total 30-day and 90-day costs, or (2) improved early clinical outcomes including readmission,
66
reoperation, length of stay, and discharge disposition, when compared to cementless fixation.
67 68
Materials and Methods
69
ICD-9 and ICD-10 data were used to identify 1671 cases of primary THA among Medicare patients who
70
underwent surgery from January 2014 through October 2017 across nine hospitals in our healthcare
71
network. Data was available for all patients with Medicare as the primary insurance, including patients
72
with a secondary private insurance. The hospitals included five tertiary-care academic facilities as well as
73
four community hospitals, all of whom were participating in either BPCI or CJR during the study period.
74
One primary academic center represented the BPCI payment cohort, whereas the remaining eight
75
facilities comprised the CJR payment cohort. BPCI data was collected from January 2014 through
76
October 2017, and CJR data was collected from January 2016 through October 2017. Revision
77
procedures were excluded.
78
Demographics including age, gender, Charlson Comorbidity Index (CCI) values and the total number of
79
emergency department (ED) visits within the six months prior to the index surgery had been recorded
80
prospectively in the hospital electronic record and were extracted digitally. The CCI is a validated clinical
81
tool that quantifies the severity of a patient’s medical comorbidities; increasing CCI score is correlated
82
with increased likelihood of postoperative mortality as well as increased costs of care[30–33]. The
83
number of ED visits within the six months leading up to surgery has been found to predict risk of
84
postoperative ED utilization, readmission, and early postoperative mortality[34–36]. Length-of-stay
85
(LOS) data were also extracted from our healthcare network electronic medical record. ICD-9/10 codes
86
were used to determine whether each THA was performed on an elective basis (i.e., primary
87
osteoarthritis, rheumatoid arthritis, etc.) or for a proximal femoral fracture. ICD-10 data was used to
88
determine whether femoral cementation was performed at the time of surgery. In cases where ICD-10
89
data was not available, operative implant logs and operative reports were consulted. Femoral
90
cementation status was successfully verified for all patients.
91
CMS data was queried for discharge disposition and total costs in the episode of care. Within the BPCI
92
cohort, an episode of care includes the operative admission as well as the 30 days following discharge.
93
The CJR episode of care includes the operative admission as well as the 90 days following discharge.
94
Therefore, total cost per episode of care was available as 30 days post-discharge for the BPCI group and
95
90 days post-discharge for the CJR group. These costs were inclusive of initial hospital charges as well as
96
costs incurred by readmissions, at-home and outpatient services, skilled nursing facility costs, and 3
97
Medicare Plan B costs. CMS data allowed capture of cost and readmission data to hospitals and
98
providers outside of our healthcare network. Only total perioperative costs were available and recorded,
99
not itemized charges (e.g., cost per implant or per procedure). For all readmissions within our health
100
system, electronic health records were reviewed to determine whether a reoperation at the surgical site
101
occurred.
102
Given the retrospective nonrandomized nature of this investigation, the cemented and cementless
103
cohorts were compared with regard to potential confounding variables including age, gender, CCI,
104
frequency of emergency department visits, and indication for surgery (elective surgery versus urgent
105
surgery for fracture). Outcome measures included total cost per episode of care and early clinical
106
outcomes consisting of LOS, discharge disposition (i.e., to home versus to skilled nursing facilities or
107
acute rehabilitation facilities), readmissions and reoperations. The Shapiro-Wilk test was applied to
108
confirm whether data was normally distributed. Inter-group comparisons of continuous variables with
109
normally distributed data were assessed via a T-test and inter-group comparisons of continuous
110
variables with data that is not normally distributed were examined using the Mann-Whitney U test.
111
Comparison of the distribution of categorical variables in the different groups were performed with
112
Fisher's exact test or the χ 2 test.
113
Univariate analysis was performed to determine differences in outcomes between patients currently
114
treated with cemented and cementless fixation in our health system. Outcomes of interest were total
115
costs, LOS, discharge disposition, readmissions, and reoperations at the surgical site. LOS and discharge
116
disposition analysis was performed for the pooled BPCI and CJR cohorts whereas readmissions,
117
reoperations and total costs were analyzed separately for the BPCI and CJR cohorts due to differences in
118
follow-up duration (30 vs. 90 days, respectively).
119
Multiple regression analysis was employed to determine the association between fixation strategy and
120
outcomes while adjusting for the baseline differences between the cemented and cementless cohorts.
121
As the fixation cohorts differed with regard to age, gender, CCI score, the number of ED visits in the six
122
months preceding the index surgery, and fracture versus elective indication for surgery, each of these
123
potential covariates was included in the regression. Multiple linear regressions were employed to assess
124
the association between costs, length-of-stay, and cemented status while controlling for age, gender,
125
CCI, frequency of prior emergency department visits, and indication for surgery (elective surgery versus
126
urgent surgery for fracture). Multiple logistic regressions were employed to assess the association
127
between discharge disposition, readmissions, and cemented status while controlling age, gender, CCI,
128
frequency of prior emergency department visits, and indication for surgery. BPCI and CJR cohorts were 4
129
again analyzed separately when assessing costs and readmissions and combined when assessing LOS
130
and discharge disposition. Multiple regression analysis was not performed for reoperations as the low
131
number of cases (11 reoperations total among 1671 primary surgeries) resulted in a lack of statistical
132
validity to study multiple variables. All statistical analyses were performed with SPSS v22 and statistical
133
significance was set at P < 0.05.
134 135
Results
136
Across the overall study population of 1671 Medicare patients, 359 (21.5%) received cemented femoral
137
fixation while 1312 (78.5%) received cementless femoral fixation. Analysis of baseline characteristics
138
revealed significant differences between the groups receiving cemented and cementless fixation with
139
regards to age, gender, CCI, the number of emergency department visits in the prior six months, and
140
indication for surgery (Table 1). Patients receiving cemented femoral fixation were older (81.4 vs 74.5
141
years; p<0.001), were more likely to be female (78.8% vs 61.1%; p<0.001), had a higher CCI (4.90 vs 3.74;
142
p<0.001), had a higher number of ED visits in the six months prior to surgery (1.05 vs 0.50; p<0.001),
143
were more frequently being treated for a proximal femur fracture than for an elective indication (53.1%
144
vs. 15.5%; p<0.001), and in the hospital participating in BPCI compared to the hospitals participating in
145
CJR (51.5% vs. 41.8%; p=0.001).
146
Univariate Analyses:
147
Univariate analysis of the data is presented in Tables 2 and 3. Before accounting for the baseline
148
differences between the cemented and cementless cohorts, patients receiving cemented fixation were
149
observed to have significantly higher total costs ($32,166 vs. $25,578; p<0.001). This held true for both
150
the BPCI and CJR cohorts as well as the study population overall. Patients receiving cemented fixation
151
also demonstrated trends toward higher rates of readmission (8.4% vs. 6.6%; p=0.24) and lower rates of
152
reoperation at the surgical site (0% vs. 0.84%; p=0.135). Although the observed difference was not
153
statistically significant, it is notable that there were no instances of reoperations at the surgical site
154
within the early postoperative period among all of the 359 cemented patients, whereas 11 reoperations
155
occurred among the 1312 cementless patients. These included 4 periprosthetic fractures, 3 cases of
156
recurrent dislocation, 2 cases of femoral implant subsidence, and 2 periprosthetic infections (Table 4).
157
Upon analyzing the aggregated data, statistically higher LOS (4.9 days vs. 3.8 days; p<0.001) and lower
158
rates of discharge to home (38.4% vs. 57.9%; p<0.001) were observed with the cemented patient
159
cohort.
160
Multivariate Analyses: 5
161
Due to the significant baseline differences between the cemented and cementless cohorts, multiple
162
regression analyses were performed to determine the influence of fixation choice while statistically
163
controlling for the impact of known confounding variables. The results of these analysis are reviewed
164
individually below. Multiple regression analysis was not performed for reoperations as the low number
165
of cases (11) would result in a lack of statistical validity.
166
Costs:
167
Following multiple linear regression analysis with the inclusion of confounding variables, there was no
168
significant difference in total cost of care between cemented and cementless patients in either the CJR
169
or BPCI cohorts. The observed trend was toward lower costs among the cemented patients in the CJR
170
cohort (p=0.57) and in the BPCI cohort (p=0.46), but this did not approach statistical significance. Within
171
the CJR cohort, higher costs were significantly associated with increasing age, female gender, more
172
emergency department visits within the prior six months, and diagnosis of femoral neck fracture. Within
173
the BPCI cohort, higher costs were significantly associated with increasing age, higher CCI scores, more
174
emergency department visits within the prior six months, and diagnosis of femoral neck fracture (Table
175
5).
176
Readmissions:
177
Following multiple logistic regression analysis with the inclusion of confounding variables, there was no
178
significant difference in readmission rates between cemented and cementless patients in either the CJR
179
or BPCI cohorts. The trend was toward fewer readmissions among the cemented patients with p=0.09 in
180
the CJR cohort and p=0.46 in the BPCI cohort. Within the CJR cohort, significantly higher readmission
181
rates were associated with more emergency department visits within the prior six months. Within the
182
BPCI cohort, higher CCI scores were significantly associated with more readmissions (Table 6).
183
LOS:
184
Following multiple linear analysis with the inclusion of confounding variables, there was no significant
185
difference in LOS between cemented and cementless patients. The trend was toward shorter LOS
186
among the cemented patients with p=0.21. Increasing age, higher CCI scores, more emergency
187
department visits within the prior six months, and hip fracture diagnoses were significantly associated
188
with higher LOS (Table 7).
189
Discharge disposition:
190
Following multiple logistic analysis with the inclusion of confounding variables, cemented patents were
191
significantly more likely to be discharged home (p=0.019) as opposed to a rehabilitation or other nursing
192
facility. Younger age, male gender, lower emergency department usage in the prior six months, and 6
193
elective (i.e., non-fracture) indications for surgery were also significantly associated with a higher
194
likelihood of discharge home (Table 8).
195 196
Discussion
197
There is a growing trend in the United States as well as worldwide toward the adoption of cementless
198
femoral component fixation for hip arthroplasty. In 2009, a survey of surgeons at the annual meeting of
199
the American Association of Hip and Knee Surgeons (AAHKS) showed that 47% of respondents preferred
200
to use cementless femoral components in all hip arthroplasty cases, and another 47% used cementless
201
fixation in over 75% of cases[37]. There has been concern that femoral cementing technique may
202
become a “lost art” as enthusiasm for cementless fixation continues to build. There exists a strong
203
argument that surgeons in training should become competent with femoral cementation technique and
204
a legitimate question as to whether we should be using cemented femoral fixation more frequently in
205
patients over 65 years of age to improve clinical outcomes and decrease costs and morbidity associated
206
with complications of cementless fixation.
207
Ours is the first study to our knowledge to leverage the new ability to combine patient-level CMS cost
208
and readmissions data with a large institutional database to compare economic and clinical outcomes
209
between cemented and cementless femoral fixation in the Medicare population. Patients who received
210
cemented femoral fixation were significantly older, more commonly female, had more medical
211
comorbidities, had more emergency department utilization, and were more likely to present with
212
proximal femoral fracture than patients receiving cementless femoral fixation. Likely as a result of
213
demographics, patients receiving cement fixation had inferior short-term outcomes (with the exception
214
of reoperation) on univariate analysis. If these results are typical of health systems in which a relatively
215
small percentage of THAs receive cement fixation, the choice may be self-reinforcing; communities of
216
surgeons who choose cement primarily for hip fracture patients and the frail elderly may come to
217
perceive cement fixation to be associated with inferior outcomes.
218
After controlling for these differences, however, we found that cement fixation was associated with
219
superior performance over cementless fixation with regards to discharge disposition, suggesting earlier
220
return of function. Cemented fixation also trended toward superiority with regards to LOS, readmission
221
rates, reoperations at the surgical site, and total costs. These trends were present within a community
222
of surgeons that elected cementless fixation for 80% of Medicare patients, but only became evident
223
with multivariable analysis. Communities of surgeons with routine utilization of cement fixation beyond
7
224
fracture patients and the frail elderly might more easily perceive the benefits of cement fixation
225
demonstrated in registry data, however.
226
Our demographic analyses were consistent with prior reported findings, revealing generally improved
227
short-term outcomes associated with younger age, male gender, fewer comorbid conditions, lower rates
228
of pre-operative emergency department usage, and elective surgery rather than proximal femoral
229
fracture diagnosis[23–28,38]. It is notable that these demographic variables were significantly
230
associated with total costs whereas the study variable most within surgeon control, femoral fixation
231
strategy, was not. All of these key components were incorporated into our multivariate analysis to
232
better understand the true effect of femoral fixation strategy in this retrospective analysis.
233
We believe that a critical advantage of cemented fixation is the avoidance of early reoperation
234
secondary to periprosthetic femur fracture, implant subsidence and failure of osseointegration. Aside
235
from being a devastating complication in its own right, periprosthetic fracture is a significant risk factor
236
for subsequent readmission, reoperation, permanent loss of function, and mortality[39,40]. Although
237
our study was not powered to demonstrate statistical significance with regards to early reoperation,
238
there exists strong published evidence supporting the superiority of cemented fixation in the reduction
239
of both intraoperative and postoperative periprosthetic fractures, particularly in the frail and/or
240
elderly[13,22–24,27,41,42]. Accordingly, all 11 reoperations occurred in the cementless fixation group;
241
these included four instances of periprosthetic fracture as well as two additional cases of femoral
242
revision for early subsidence with mechanical failure of cementless fixation.
243
Our findings should be interpreted in the context of prior studies, many of which were performed
244
outside of the United States in regions with greater utilization of cemented technique. Pennington et al.
245
found in a large English and Welsh joint replacement registry study that femoral cementation yielded
246
lower costs and higher quality-adjusted life years than cementless fixation[43]. Veldman et al.
247
performed a meta-analysis of 5 randomized control trials (RCTs) that found that cementless fixation
248
generated more complications, particularly implant-related complications, than cemented fixation[41].
249
Chammout et al. also found that uncemented femoral fixation was significantly more likely to result in
250
early hip-related complications than cemented fixation in a single-blinded RCT, with similar rates of
251
mortality and functional outcome scores[13]. Moerman et al. found in another RCT that cementless
252
patients had greater rates of complication than cemented, with similar functional outcome scores and
253
pain[42]. Taylor et al. also found higher rates of complication in cementless patients in their own RCT,
254
with similar pain and trends toward improved mobility and function in the cemented group[16]. In a
8
255
systematic review, Luo et al. found that cemented fixation tends to provide better pain relief and
256
function with no additional risk with regards to mortality, complication, or reoperation[17].
257
Although our findings support a trend toward improved outcomes with cemented fixation for every
258
measure of interest, prior reports favoring cementless fixation should also be considered. In a
259
retrospective review of patients treated for displaced femoral neck fractures, Fuchs et al. found that
260
perioperative mortality was higher after cemented hemiarthroplasty than cementless THA, though hip-
261
related surgical complications remained higher in the THA group[11]. There may have been an element
262
of selection bias as there was no prospective treatment randomization and it is possible that older and
263
frailer patients with more comorbidities were selected for the cemented hemiarthroplasty group.
264
Langslet et al. performed a randomized clinical trial of displaced femoral neck fractures treated with
265
either cemented or cementless hemiarthroplasty with 5-year follow-up that yielded higher Harris Hip
266
Scores in the cementless group[44]. However, they also demonstrated a significantly higher rate of
267
periprosthetic femur fracture in the cementless group with similar rates of infection, instability, and
268
mortality. Other studies have proposed that cementless fixation is safer with regards to
269
cardiopulmonary and thromboembolic problems as well as all-cause mortality[6–12], and have been
270
refuted by other studies in turn[13–22]. While some studies suggest lower perioperative mortality with
271
cementless fixation, mortality at one year has not been different and this may relate to the morbidity
272
and mortality related to failed cementless fixation and its treatment[12,20,44].
273
We acknowledge several limitations to our study. Firstly, this study was designed primarily to leverage
274
30-day and 90-day cost data provided by the BPCI and CJR initiatives, respectively, to determine the
275
effect of fixation strategy on costs, resource utilization and complications. Therefore, clinical data were
276
recorded only through the same early postoperative periods. Patient-reported outcome scores, as well
277
as mid-term and long-term follow-up studies, are necessary to fully assess the performance of cemented
278
versus cementless femoral fixation over time. These longer-term outcomes have been studied
279
extensively by others but remain a focus for further inquiry. Deeper investigation into specific medical
280
complications such as fat embolism syndrome has been performed by others[19,45] but may also be a
281
useful direction of future study, as modern techniques may influence the incidence of these events.
282
Secondly, as a retrospective database analysis spanning across multiple hospitals, there were significant
283
differences between the cemented and uncemented cohorts with regards to multiple demographic
284
factors: age, gender, comorbidities, frequency of emergency room visits, and presenting diagnosis. The
285
correlation between surgical indication and cemented status was particularly strong. It has been well-
286
described in the literature that patients undergoing hip arthroplasty for fracture have significantly 9
287
poorer outcomes across all clinical and cost parameters than patients undergoing elective
288
arthroplasty[46–48]. However, we did adjust for these known confounding factors with the use of
289
inferential statistics. Although there may be additional unrecognized and potentially confounding
290
variables associated with institutional participation in the bundled payment systems, it should be noted
291
that all patients, whether cemented or cementless, were in one of the two bundled payment programs.
292
There may have been selection bias based upon surgeon training and experience, which we were not
293
able to control given the available database information. For example, older surgeons, those with
294
foreign training, and/or those with arthroplasty fellowship training may have been more likely to use
295
cemented femoral fixation. Based upon our own practice patterns, we suspect there may have been bias
296
towards use of cemented fixation in cases of patient frailty, osteoporosis, Dorr C femoral morphology,
297
higher fall risk, and other patient characteristics that lend themselves to higher risk of hip and
298
periprosthetic fracture. Alternatively, surgeons may have steered away from cemented fixation when
299
treating patients with known cardiopulmonary disease. Individual hospital norms and culture may also
300
have played some role in femoral fixation preference, but the majority of surgeries were performed
301
using cementless fixation at all hospitals. Furthermore, we were not able to standardize specific aspects
302
of surgical technique with regards to surgical approach, implant selection, cementation protocol,
303
cement brand and type, and other factors. Surgical approach may be particularly salient given that
304
previous literature have suggested differences in perioperative complications between approaches, such
305
as increased rates of intraoperative femoral fracture with the anterior approach and lower rates of
306
instability with the direct lateral approach[49–52]. Our institutional databases did not track surgical
307
approach. The nature of the CMS database also precluded us from capturing cost data past the 30 day
308
post-discharge period for BPCI and the 90 day post-discharge period for CJR. We lacked data on
309
osteoporosis or bone morphology and our study population was not large enough to meaningfully
310
stratify our analysis by age, gender, diagnosis or other factors that might define subpopulations that
311
would most or least benefit from cement fixation. Our study size was also insufficient to power certain
312
findings that we feel should have been reproducible from the existing literature, particularly the
313
increased likelihood of early implant-based complications such as PFF and subsidence with cementless
314
fixation. Power analysis regarding early reoperations revealed that we would have required at least 932
315
patients in each cohort to achieve a power of 80% at an alpha error of 0.05. Another limitation is our
316
study is that although CMS data comprehensively covered perioperative costs and readmissions, we
317
were not able to capture reoperation data at facilities outside of our hospital network. This will be a
318
direction of future study. We expect that with greater power, statistical significance may be achieved 10
319
with regards to the observed trends towards improved cost, readmissions, and LOS among cemented
320
patients. Greater power would also allow us to investigate possible effects on mortality, which was not
321
included in this study for the same reason. Conversely, we acknowledge that a larger data set may
322
reveal some of the observed trends to be statistical anomalies related to chance, rather than reflecting
323
true differences between groups.
324 325
Conclusion
326
Although there has been a strong movement in the United States and globally in favor of cementless
327
femoral fixation, historical and current evidence both demonstrate that cemented technique produces
328
equivalent or superior outcomes for hip arthroplasty patients, particularly in older adults such as the
329
Medicare population studied here. Nevertheless, any short term economic advantage of cemented
330
femoral fixation was too small for us to measure as total costs were equivalent in our cohorts. We
331
continue to favor cemented fixation for frail older patients with osteopenic bone who are at risk for
332
implant subsidence and periprosthetic fracture. Our data do not support any need for wholesale change
333
in practice pattern with return to cemented femoral fixation as the standard for all older or Medicare
334
patients, nor do our data suggest that any such change would be associated with unanticipated
335
increased risk or cost.
336
Prior studies have identified risk factors such as advanced age, malnutrition, osteoporosis, female
337
gender, and low socioeconomic status for periprosthetic fracture or cementless fixation failure and
338
these criteria may be useful to identify selected patients most likely to obtain benefit from cemented
339
fixation. Because some patients clearly can benefit from cemented fixation and most patients over age
340
65 get results from cemented fixation equal or better to what they would achieve with cementless
341
fixation, there may be a role for practicing surgeons to use cemented fixation routinely enough that the
342
knowledge and skills necessary are maintained and transmitted to surgeons in training.
11
343
References
344 345
[1]
Kurtz S, Ong K, Lau E, Mowat F, Halpern M. Projections of Primary and Revision Hip and Knee
346
Arthroplasty in the United States from 2005 to 2030. J Bone Jt Surg 2007;89:780.
347
https://doi.org/10.2106/JBJS.F.00222.
348
[2]
Troelsen A, Malchau E, Sillesen N, Malchau H. A Review of Current Fixation Use and Registry
349
Outcomes in Total Hip Arthroplasty: The Uncemented Paradox. Clin Orthop Relat Res
350
2013;471:2052–9. https://doi.org/10.1007/s11999-013-2941-7.
351
[3]
352 353
Arthroplasty 2014;29:1915–8. https://doi.org/10.1016/j.arth.2014.05.017. [4]
354 355
Lehil MS, Bozic KJ. Trends in Total Hip Arthroplasty Implant Utilization in the United States. J
Khanuja HS, Vakil JJ, Goddard MS, Mont MA. Cementless Femoral Fixation in Total Hip Arthroplasty. J Bone Jt Surg 2011;93:500–9. https://doi.org/10.2106/JBJS.J.00774.
[5]
Ahn J, Man L-X, Park S, Sodl JF, Esterhai JL. Systematic Review of Cemented and Uncemented
356
Hemiarthroplasty Outcomes for Femoral Neck Fractures n.d. https://doi.org/10.1007/s11999-
357
008-0368-3.
358
[6]
Yli-Kyyny T, Ojanperä J, Venesmaa P, Kettunen J, Miettinen H, Salo J, et al. Perioperative
359
Complications after Cemented or Uncemented Hemiarthroplasty in Hip Fracture Patients. Scand J
360
Surg 2013;102:124–8. https://doi.org/10.1177/1457496913482249.
361
[7]
Ries MD, Lynch F, Rauscher LA, Richman J, Mick C, Gomez M. Pulmonary function during and
362
after total hip replacement. Findings in patients who have insertion of a femoral component with
363
and without cement. J Bone Joint Surg Am 1993;75:581–7.
364
[8]
365 366
Sharrock NE, Go G, Harpel PC, Ranawat CS, Sculco TP, Salvati EA. The John Charnley Award. Thrombogenesis during total hip arthroplasty. Clin Orthop Relat Res 1995:16–27.
[9]
Francis CW, Marder VJ, Evarts CM. Lower risk of thromboembolic disease after total hip
367
replacement with non-cemented than with cemented prostheses. Lancet (London, England)
368
1986;1:769–71.
369
[10]
370 371
Borghi B, Casati A, Rizzoli Study Group on Orthopaedic Ananesthesia. Thromboembolic complications after total hip replacement. Int Orthop 2002;26:44–7.
[11]
Fuchs M, Sass FA, Dietze S, Krämer M, Perka C, Müller M. Cemented Hemiarthroplasties Are
372
Associated with a Higher Mortality Rate after Femoral Neck Fractures in Elderly Patients. Acta
373
Chir Orthop Traumatol Cech 2017;84:341–6.
374
[12]
Garland A, Gordon M, Garellick G, Kärrholm J, Sköldenberg O, Hailer NP. Risk of early mortality 12
375
after cemented compared with cementless total hip arthroplasty. Bone Joint J 2017;99-B:37–43.
376
https://doi.org/10.1302/0301-620X.99B1.BJJ-2016-0304.R1.
377
[13]
Chammout G, Muren O, Laurencikas E, Bodén H, Kelly-Pettersson P, Sjöö H, et al. More
378
complications with uncemented than cemented femo-ral stems in total hip replacement for
379
displaced femoral neck fractures in the elderly A single-blinded, randomized controlled trial with
380
69 patients. Acta Orthop 2017;88:145–51. https://doi.org/10.1080/17453674.2016.1262687.
381
[14]
Miyamoto S, Nakamura J, Iida S, Shigemura T, Kishida S, Abe I, et al. Intraoperative blood
382
pressure changes during cemented versus uncemented bipolar hemiarthroplasty for displaced
383
femoral neck fracture: a multi-center cohort study. Arch Orthop Trauma Surg 2017;137:523–9.
384
https://doi.org/10.1007/s00402-017-2651-9.
385
[15]
Hong CC. Cemented hemiarthroplasty in traumatic displaced femoral neck fractures and deep
386
vein thrombosis: is there really a link? Singapore Med J 2016;57:69–72.
387
https://doi.org/10.11622/smedj.2016030.
388
[16]
Taylor F, Wright M, Zhu M. Hemiarthroplasty of the Hip with and without Cement: A Randomized
389
Clinical Trial. J Bone Jt Surgery-American Vol 2012;94:577–83.
390
https://doi.org/10.2106/JBJS.K.00006.
391
[17]
Luo X, He S, Li Z, Huang D. Systematic review of cemented versus uncemented hemiarthroplasty
392
for displaced femoral neck fractures in older patients. Arch Orthop Trauma Surg 2012;132:455–
393
63. https://doi.org/10.1007/s00402-011-1436-9.
394
[18]
Wolf LR, Hozack WJ, Balderston RA, Booth RE, Rothman RH. Pulmonary embolism. Incidence in
395
primary cemented and uncemented total hip arthroplasty using low-dose sodium warfarin
396
prophylaxis. J Arthroplasty 1992;7:465–70.
397
[19]
Kim Y-H, Oh S-W, Kim J-S. Prevalence of fat embolism following bilateral simultaneous and
398
unilateral total hip arthroplasty performed with or without cement : a prospective, randomized
399
clinical study. J Bone Joint Surg Am 2002;84-A:1372–9.
400
[20]
Ekman E, Palomäki A, Laaksonen I, Peltola M, Häkkinen U, Mäkelä K. Early postoperative
401
mortality similar between cemented and unce-mented hip arthroplasty: a register study based
402
on Finnish national data. Acta Orthop 2019;90:6–10.
403
https://doi.org/10.1080/17453674.2018.1558500.
404
[21]
Meftah M, John M, Lendhey M, Khaimov A, Ranawat AS, Ranawat CS. Safety and Efficacy of Non-
405
Cemented Femoral Fixation in Patients 75 Years of Age and Older. J Arthroplasty 2013;28:1378–
406
80. https://doi.org/10.1016/j.arth.2012.11.007. 13
407
[22]
Morris K, Davies H, Wronka K. Implant-related complications following hip hemiarthroplasty: a
408
comparison of modern cemented and uncemented prostheses. Eur J Orthop Surg Traumatol
409
2015;25:1161–4. https://doi.org/10.1007/s00590-015-1671-9.
410
[23]
Zhang Z, Zhuo Q, Chai W, Ni M, Li H, Chen J. Clinical characteristics and risk factors of
411
periprosthetic femoral fractures associated with hip arthroplasty A retrospective study 2016.
412
https://doi.org/10.1097/MD.0000000000004751.
413
[24]
Grosso MG, Danoff JR, Padgett DE, Iorio R, Macaulay WB. The Cemented Unipolar Prosthesis for
414
the Management of Displaced Femoral Neck Fractures in the Dependent Osteopenic Elderly. J
415
Arthroplasty 2016;31:1040–6. https://doi.org/10.1016/j.arth.2015.11.029.
416
[25]
Gromov K, Bersang A, Nielsen CS, Kallemose T, Husted H, Troelsen A. Risk factors for post-
417
operative periprosthetic fractures following primary total hip arthroplasty with a proximally
418
coated double-tapered cementless femoral component. Bone Joint J 2017;99-B:451–7.
419
https://doi.org/10.1302/0301-620X.99B4.BJJ-2016-0266.R2.
420
[26]
Zhu Y, Chen W, Sun T, Zhang X, Liu S, Zhang Y. Risk factors for the periprosthetic fracture after
421
total hip arthroplasty: a systematic review and meta-analysis. Scand J Surg 2015;104:139–45.
422
https://doi.org/10.1177/1457496914543979.
423
[27]
Sidler-maier CC, Waddell JP. Incidence and predisposing factors of periprosthetic proximal
424
femoral fractures : a literature review 2015:1673–82. https://doi.org/10.1007/s00264-015-2721-
425
y.
426
[28]
A Miettinen SS, Mäkinen tatu J, koStenSAlo inari, Mäkelä keijo, HuHtAlA H, kettunen JS, et al.
427
Risk factors for intraoperative calcar fracture in cementless total hip arthroplasty. Acta Orthop
428
2016;87:113–9. https://doi.org/10.3109/17453674.2015.1112712.
429
[29]
430 431
Guideline 2014. [30]
432 433
Surgeons AA of O. Management of Hip Fractures in the Elderly: Evidence-Based Clinical Practice
Charlson M, Szatrowski TP, Peterson J, Gold J. Validation of a combined comorbidity index. J Clin Epidemiol 1994;47:1245–51. https://doi.org/10.1016/0895-4356(94)90129-5.
[31]
D’Hoore W, Bouckaert A, Tilquin C. Practical considerations on the use of the charlson
434
comorbidity index with administrative data bases. J Clin Epidemiol 1996;49:1429–33.
435
https://doi.org/10.1016/S0895-4356(96)00271-5.
436
[32]
Quan H, Li B, Couris CM, Fushimi K, Graham P, Hider P, et al. Updating and Validating the
437
Charlson Comorbidity Index and Score for Risk Adjustment in Hospital Discharge Abstracts Using
438
Data From 6 Countries. Am J Epidemiol 2011;173:676–82. https://doi.org/10.1093/aje/kwq433. 14
439
[33]
Charlson ME, Charlson RE, Peterson JC, Marinopoulos SS, Briggs WM, Hollenberg JP. The Charlson
440
comorbidity index is adapted to predict costs of chronic disease in primary care patients. J Clin
441
Epidemiol 2008;61:1234–40. https://doi.org/10.1016/J.JCLINEPI.2008.01.006.
442
[34]
van Walraven C, Dhalla IA, Bell C, Etchells E, Stiell IG, Zarnke K, et al. Derivation and validation of
443
an index to predict early death or unplanned readmission after discharge from hospital to the
444
community. CMAJ 2010;182:551–7. https://doi.org/10.1503/cmaj.091117.
445
[35]
Gruneir A, Dhalla IA, van Walraven C, Fischer HD, Camacho X, Rochon PA, et al. Unplanned
446
readmissions after hospital discharge among patients identified as being at high risk for
447
readmission using a validated predictive algorithm. Open Med 2011;5:e104-11.
448
[36]
Wang H, Robinson RD, Johnson C, Zenarosa NR, Jayswal RD, Keithley J, et al. Using the LACE index
449
to predict hospital readmissions in congestive heart failure patients. BMC Cardiovasc Disord
450
2014;14:97. https://doi.org/10.1186/1471-2261-14-97.
451
[37]
Berry DJ, Bozic KJ. Current practice patterns in primary hip and knee arthroplasty among
452
members of the American Association of Hip and Knee Surgeons. J Arthroplasty 2010;25:2–4.
453
https://doi.org/10.1016/j.arth.2010.04.033.
454
[38]
Park KJ, Menendez ME, Barnes CL. Perioperative Periprosthetic Fractures Associated With
455
Primary Total Hip Arthroplasty. J Arthroplasty 2017;32:992–5.
456
https://doi.org/10.1016/j.arth.2016.08.034.
457
[39]
Carli A V. Periprosthetic femoral fractures and trying to avoid them DESIGN TO THE INCREASED
458
RISK OF PERIPROSTHETIC FEMORAL n.d.:50–9. https://doi.org/10.1302/0301-620X.99B1.BJJ-
459
2016-0220.R1.
460
[40]
Katz JN, Wright EA, Polaris JJ, Harris MB, Losina E. Prevalence and risk factors for periprosthetic
461
fracture in older recipients of total hip replacement: a cohort study. vol. 15. 2014.
462
https://doi.org/10.1186/1471-2474-15-168.
463
[41]
Veldman HD, Heyligers IC, Grimm B, Boymans TAEJ. Cemented versus cementless
464
hemiarthroplasty for a displaced fracture of the femoral neck. Bone Jt J 2017;99B:421–31.
465
https://doi.org/10.1302/0301-620X.99B4.BJJ-2016-0758.R1.
466
[42]
Moerman S, Mathijssen NMC, Niesten DD, Riedijk R, Rijnberg WJ, Koëter S, et al. More
467
complications in uncemented compared to cemented hemiarthroplasty for displaced femoral
468
neck fractures: a randomized controlled trial of 201 patients, with one year follow-up n.d.
469
https://doi.org/10.1186/s12891-017-1526-0.
470
[43]
Pennington M, Grieve R, Sekhon JS, Gregg P, Black N, Van Der Meulen JH. Cemented, cementless, 15
471
and hybrid prostheses for total hip replacement: Cost effectiveness analysis. BMJ 2013;346.
472
https://doi.org/10.1136/bmj.f1026.
473
[44]
Langslet E, Frihagen F, Madsen JE, Nordsletten L, Figved W. Cemented versus Uncemented
474
Hemiarthroplasty for Displaced Femoral Neck Fractures: 5-year Followup of a Randomized Trial
475
n.d. https://doi.org/10.1007/s11999-013-3308-9.
476
[45]
Heisel C, Mau H, Borchers T, Müller J, Breusch SJ. Fettembolie bei der hüftendoprothesen-
477
implantation. Zementfrei gegen zementiert - Ein quantitativer In vivo-Vergleich im Tiermodell.
478
Orthopade 2003;32:247–52. https://doi.org/10.1007/s00132-002-0394-x.
479
[46]
Le Manach Y, Collins G, Bhandari M, Bessissow A, Boddaert J, Khiami F, et al. Outcomes After Hip
480
Fracture Surgery Compared With Elective Total Hip Replacement. JAMA 2015;314:1159.
481
https://doi.org/10.1001/jama.2015.10842.
482
[47]
Schroer WC, Diesfeld PJ, LeMarr AR, Morton DJ, Reedy ME. Hip Fracture Does Not Belong in the
483
Elective Arthroplasty Bundle: Presentation, Outcomes, and Service Utilization Differ in Fracture
484
Arthroplasty Care. J Arthroplasty 2018;33:S56–60. https://doi.org/10.1016/j.arth.2018.02.091.
485
[48]
486 487
Yoon RS, Mahure SA, Hutzler LH, Iorio R, Bosco JA. Hip Arthroplasty for Fracture vs Elective Care. J Arthroplasty 2017;32:2353–8. https://doi.org/10.1016/j.arth.2017.02.061.
[49]
Hartford JM, Knowles SB. Risk Factors for Perioperative Femoral Fractures: Cementless Femoral
488
Implants and the Direct Anterior Approach Using a Fracture Table. J Arthroplasty 2016;31:2013–
489
8. https://doi.org/10.1016/j.arth.2016.02.045.
490
[50]
Kwon MS, Kuskowski M, Mulhall KJ, Macaulay W, Brown TE, Saleh KJ. Does Surgical Approach
491
Affect Total Hip Arthroplasty Dislocation Rates? Clin Orthop Relat Res 2006;447:34–8.
492
https://doi.org/10.1097/01.blo.0000218746.84494.df.
493
[51]
Masonis JL, Bourne RB. Surgical Approach, Abductor Function, and Total Hip Arthroplasty
494
Dislocation. Clin Orthop Relat Res 2002;405:46–53. https://doi.org/10.1097/00003086-
495
200212000-00006.
496
[52]
Demos HA, Rorabeck CH, Bourne RB, MacDonald SJ, McCalden RW. Instability in Primary Total
497
Hip Arthroplasty With the Direct Lateral Approach. Clin Orthop Relat Res 2001;393:168–80.
498
https://doi.org/10.1097/00003086-200112000-00020.
499
16
Tables
Table 1. Baseline demographic and clinical characteristics, split by payment model and cemented status BPCI Cemented
Cementless
Age (mean,
80.06
73.70
sd)
(7.71)
(8.58)
28
CJR p value
Cemented
Cementless
82.90
75.06
(8.64)
(9.08)
238
48
272
(15.1%)
(43.4%)
(27.6%)
(35.6%)
157
311
126
491
(84.9%)
(56.6%)
(72.4%)
(64.4%)
<0.001 Ɨ
Total p value <0.001 Ɨ
Cemented 81.44 (8.28)
Cementless
p value
74.49 (8.90)
<0.001 Ɨ
Gender (n, %) Male
Female CCI score
4.41
(mean, sd)
(1.87)
<0.001 Ɨ
3.63 (1.69)
<0.001 Ɨ
0.38 (0.92)
<0.001 Ɨ
5.43 (2.35)
76 (21.2%)
0.0511
283 (78.8%)
510 (38.9%)
802 (61.1%)
<0.001 Ɨ
3.82 (1.93)
<0.001 Ɨ
4.90 (2.17)
3.74 (1.84)
<0.001 Ɨ
0.60 (1.07)
<0.001 Ɨ
1.05 (1.41)
0.50 (1.02)
<0.001 Ɨ
ED visits in prior 6
0.83
months
(1.29)
1.28 (1.49)
(mean, sd) Indication (n, %) 59 Fracture
Elective
(31.9%)
81 (14.8%)
126
468
(68.1%)
(85.2%)
<0.001 Ɨ
132
122
191
(75.9%)
(16.0%)
(53.1%)
42
641
(24.1%)
(84.0%)
<0.001 Ɨ
168 (46.8%)
203 (15.5%)
1109 (84.5%)
<0.001 Ɨ
Cohort (n, %) 185 BPCI
(51.5%) 174
CJR
(48.5%)
549 (41.8%)
763 (58.2%)
0.001 Ɨ
Ɨ Statistically significant.
Table 2: Univariate analysis of total costs, readmissions, and reoperations; split by payment model BPCI Cemented
Cementless
Total costs
27,603.15
24,588.59
(mean, sd)
(9,505.25)
(9,520.29)
9 (4.9%)
27 (4.9%)
CJR p value <0.001 Ɨ
Cemented
Cementless
37,016.52
26,289.06
(16,300.03)
(13,422.14)
21 (12.1%)
59 (7.7%)
Total p value <0.001 Ɨ
Cemented
Cementless
32,165.62
25,577.51
(14,036.49)
(11,970.87)
30 (8.4%)
86 (6.6%)
p value <0.001 Ɨ
Readmission (n, %) Yes No
176
522
(95.1%)
(95.1%)
1.00
153
704
(87.9%)
(92.3%)
0.07
329
1226
(91.6%)
(93.4%)
0.24
Reoperation (n, %) Yes No
Total
0 (0%)
3 (0.55%)
0 (0%)
8 (1.05%)
11 (0.84%)
185
546
(100%)
(99.45%)
185
549
174
763
359
1312
(25.2%)
(74.8%)
(18.6%)
(81.4%)
(21.5%)
(78.5%)
0.576
174 (100%)
755
0 (0%)
(98.95%)
0.364
359 (100%)
1301
0.135
(99.16%)
Ɨ Statistically significant.
Table 3: Univariate analysis of LOS and discharge disposition with aggregated data Cemented
Cementless
p value
4.88 (2.89)
3.76 (2.35)
<0.001 Ɨ
Home
138 (38.4%)
755 (57.9%)
Other
221 (61.6%)
550 (42.1%)
LOS (mean, sd) Discharge disposition (n, %)
<0.001 Ɨ
LOS: length of stay. Ɨ Statistically significant.
Table 4. Surgical site reoperations Case
Indication for reoperation
Management
1
Periprosthetic femur fracture (Vancouver B)
Open reduction internal fixation and femoral revision
2
Periprosthetic femur fracture (Vancouver B)
Open reduction internal fixation and femoral revision
3
Periprosthetic femur fracture (Vancouver B)
Open reduction internal fixation and femoral revision
4
Periprosthetic femur fracture (Vancouver AG)
Open reduction internal fixation
5
Recurrent dislocation
Head and liner exchange
6
Recurrent dislocation
Resection arthroplasty
7
Recurrent dislocation
Acetabular component revision
8
Femoral subsidence
Femoral revision
9
Femoral subsidence
Femoral revision
10
Acute periprosthetic joint infection
Irrigation and debridement, head and liner exchange
11
Acute periprosthetic joint infection
Two-stage revision
Table 5: Impact of covariates upon total cost of care CJR Cost*
p Value
BPCI Cost*
p Value
-$671 (-$3005 to $1664)
0.573
-$570 (-$2067 to $947)
0.461
Age (per year)
$179 ($76 to $281)
0.001 Ɨ
$203 ($114 to $293)
<0.001 Ɨ
Female Gender
$2233 ($598 to $3867)
0.007 Ɨ
$327 (-$978 to $1632)
0.623
$54 (-$405 to $514)
0.816
$858 ($440 to $1275)
<0.001 Ɨ
$2938 ($2241 to $3635)
<0.001 Ɨ
$1392 ($772 to $2011)
<0.001 Ɨ
$12,887 ($10,706 to $15,068)
<0.001 Ɨ
$5262 ($3508 to $7016)
<0.001 Ɨ
Cemented Fixation
CCI (per unit) ED visits in prior 6 months (per visit) Fracture Diagnosis
*The values given represent amount of change in cost of care with the 95% confidence interval given in parentheses; all amounts rounded to the nearest full dollar. Ɨ Statistically significant.
Table 6: Impact of covariates upon readmissions CJR Readmissions*
p Value
BPCI Readmissions*
p Value
Cemented Fixation
0.559 (0.286 to 1.092)
0.089
0.728 (0.310 to 1.706)
0.464
Age (per year)
1.016 (0.985 to 1.049)
0.317
1.010 (0.962 to 1.061)
0.681
Female Gender
1.324 (0.754 to 2.325)
0.328
0.850 (0.416 to 1.738)
0.656
CCI (per unit)
1.019 (0.895 to 1.160)
0.778
1.258 (1.074 to 1.472)
0.004 Ɨ
ED visits in prior 6 months (per visit)
2.049 (1.723 to 2.436)
<0.001 Ɨ
1.179 (0.880 to 1.580)
0.269
Fracture Diagnosis
1.642 (0.876 to 3.080)
0.122
0.952 (0.375 to 2.417)
0.917
*The values given represent the odds ratio of readmission with the 95% confidence interval given in parentheses. Ɨ Statistically significant.
Table 7: Impact of covariates upon LOS LOS*
p Value
-0.179 (-0.456 to 0.099)
0.207
Age (per year)
0.019 (0.005 to 0.034)
0.009 Ɨ
Female Gender
-0.022 (-0.243 to 0.199)
0.845
CCI (per unit)
0.164 (0.098 to 0.229)
<0.001 Ɨ
ED visits in prior 6 months (per visit)
0.347 (0.249 to 0.445)
<0.001 Ɨ
Fracture Diagnosis
2.096 (1.807 to 2.384)
<0.001 Ɨ
Cemented Fixation
*The values given represent additional days of LOS with the 95% confidence interval given in parentheses. Ɨ Statistically significant.
Table 8: Impact of covariates upon likelihood of discharge to home Home Discharge OR*
p Value
Cemented Fixation
1.441 (1.061 to 1.958)
0.019 Ɨ
Age (per year)
0.935 (0.919 to 0.950)
<0.001 Ɨ
Female Gender
0.651 (0.514 to 0.825)
<0.001 Ɨ
CCI (per unit)
0.944 (0.878 to 1.015)
0.120
ED visits in prior 6 months (per visit)
0.789 (0.706 to 0.882)
<0.001 Ɨ
Fracture Diagnosis
0.213 (0.154 to 0.294)
<0.001 Ɨ
* The values given represent the odds ratio of discharge to home with the 95% confidence interval given in parentheses. Ɨ Statistically significant.
Appendices
Appendix A: Multiple linear regression analysis for sum costs
Multiple linear regression analysis split by payment model (CJR vs. BPCI). Model 1 contains univariate analysis of cemented vs. cementless fixation and Model 2 includes multivariate analysis accounting for confounding variables: age, gender, Charlson Comorbidity Index, number of emergency department visits in the 6 months prior to index surgery, and indication for surgery (fracture vs. elective). Model Summary Change Statistics R Cohort Model CJR
BPCI
R
Adjusted Std. Error of
R Square
F
Sig. F
Square R Square the Estimate
Change
Change
df1
df2
Change
1
.286a
.082
.081 13999.30845
.082 83.110
1
934
.000
2
.583b
.340
.335 11904.55315
.258 72.523
5
929
.000
1
.136a
.019
.017
9516.51678
.019 13.866
1
731
.000
2
.505c
.255
.248
8322.44541
.236 45.962
5
726
.000
a. Predictors: (Constant), Final Cement b. Predictors: (Constant), Final Cement, Gender_bi, E, Age of Surg, CCI, Fix_bi c. Predictors: (Constant), Final Cement, Fix_bi, Gender_bi, CCI, E, Age of Surg
Coefficientsa Unstandardized
Standardized
95.0% Confidence
Coefficients
Coefficients
Interval for B
Std. Cohort Model CJR
1
B
(Constant) 26289.058
Error
Beta
507.080
t 51.844
Sig.
Lower
Upper
Bound
Bound
.000 25293.911 27284.206
Final 10727.462 1176.716
.286
9.116
.000
8418.149 13036.776
2.099
.036
482.990 14386.581
Cement 2
(Constant)
7434.785 3542.279
Final -670.715 1189.518
-.018
-.564
.573 -3005.169
1663.738
Cement Age of 178.513
52.064
.116
3.429
.001
76.337
280.689
2232.520
832.644
.073
2.681
.007
598.438
3866.602
54.373
234.113
.008
.232
.816
-405.079
513.825
2937.750
355.168
.239
8.271
.000
2240.725
3634.774
Surg Gender_bi CCI E Fix_bi BPCI
1
12886.905 1111.313
(Constant) 24588.591
.393 11.596
.000 10705.930 15067.879
60.499
.000 23790.678 25386.504
406.432
Final 3014.562
809.563
.136
3.724
.000
1425.217
4603.907
1.644
.101
-970.855 10987.866
Cement 2
(Constant)
5008.505 3045.664
Final -569.845
772.642
-.026
-.738
.461 -2086.724
947.035
203.234
45.662
.186
4.451
.000
113.589
292.878
Gender_bi
327.077
664.719
.016
.492
.623
-977.923
1632.078
CCI
857.851
212.590
.158
4.035
.000
440.485
1275.216
E
1391.731
315.492
.152
4.411
.000
772.346
2011.117
Fix_bi
5262.178
893.330
.216
5.891
.000
3508.359
7015.997
Cement Age of Surg
a. Dependent Variable: Sum Cost CJR = Comprehensive Care for Joint Replacement; BPCI = Bundled Payment for Care Improvement; Final Cement = cemented fixation; Age of Surg = patient age in years; Gender_bi = female gender; CCI = Charlson Comorbidity Index; E = number of emergency department visits in 6 months prior to surgery; Fix_bi = fracture diagnosis
Appendix B: Multiple linear regression analysis for length-of-stay (LOS)
Multiple linear regression analysis with cumulative data. Model 1 contains univariate analysis of cemented vs. cementless fixation and Model 2 includes multivariate analysis accounting for confounding
variables: age, gender, Charlson Comorbidity Index, number of emergency department visits in the 6 months prior to index surgery, and indication for surgery (fracture vs. elective).
Model Summary Change Statistics
Std. Error
Model
R
R
Adjusted R
of the
R Square
F
Sig. F
Square
Square
Estimate
Change
Change
df1
df2
Change
1
.183a
.033
.033
2.477
.033
57.619
1
1667
.000
2
.522b
.273
.270
2.152
.239 109.428
5
1662
.000
a. Predictors: (Constant), Final Cement b. Predictors: (Constant), Final Cement, Gender_bi, E, CCI, Fix_bi, Age of Surg
Coefficientsa Unstandardized
Standardized
95.0% Confidence
Coefficients
Coefficients
Interval for B
Model
B
1
3.757
.068
1.121
.148
1.222
.495
-.179
.141
Age of Surg
.019
Gender_bi
(Constant)
Std. Error
Beta
t
Sig.
Lower
Upper
Bound
Bound
54.905
.000
3.623
3.891
7.591
.000
.831
1.410
2.469
.014
.251
2.192
-.029
-1.262
.207
-.456
.099
.007
.071
2.631
.009
.005
.034
-.022
.113
-.004
-.196
.845
-.243
.199
CCI
.164
.033
.128
4.906
.000
.098
.229
E
.347
.050
.156
6.919
.000
.249
.445
2.096
.147
.353
14.262
.000
1.807
2.384
Final .183
Cement 2
(Constant) Final Cement
Fix_bi
a. Dependent Variable: LOS (inclusive)
CJR = Comprehensive Care for Joint Replacement; BPCI = Bundled Payment for Care Improvement; Final Cement = cemented fixation; Age of Surg = patient age in years; Gender_bi = female gender; CCI = Charlson Comorbidity Index; E = number of emergency department visits in 6 months prior to surgery; Fix_bi = fracture diagnosis
Appendix C: Multiple logistic regression analysis for discharge disposition
Multiple logistic regression analysis with cumulative data. Model 1 contains univariate analysis of cemented vs. cementless fixation and Model 2 includes multivariate analysis accounting for confounding variables: age, gender, Charlson Comorbidity Index, number of emergency department visits in the 6 months prior to index surgery, and indication for surgery (fracture vs. elective).
Model 1: Variables in the Equation 95% C.I.for EXP(B) B Step 1a FinalCement(1) Constant
S.E.
Wald
df
Sig.
Exp(B)
-.782
.122
40.926
1
.000
.458
.315
.056
31.651
1
.000
1.371
Lower
Upper
.360
.581
a. Variable(s) entered on step 1: FinalCement.
Model 2: Variables in the Equation 95% C.I.for EXP(B) B Step 1a FinalCement(1)
S.E.
Wald
df
Sig.
Exp(B)
Lower
Upper
.366
.156
5.466
1
.019
1.441
1.061
1.958
AgeofSurg
-.068
.008
65.083
1
.000
.935
.919
.950
Gender_bi(1)
-.429
.121
12.627
1
.000
.651
.514
.825
CCI_sum
-.058
.037
2.411
1
.120
.944
.878
1.015
E
-.237
.057
17.541
1
.000
.789
.706
.882
Fx_bi(1)
-1.546
Constant
6.185
.164
88.599
1
.000
.213
.582 113.102
1
.000 485.335
.154
.294
a. Variable(s) entered on step 1: AgeofSurg, Gender_bi, CCI_sum, E, Fx_bi. CJR = Comprehensive Care for Joint Replacement; BPCI = Bundled Payment for Care Improvement; Final Cement = cemented fixation; Age of Surg = patient age in years; Gender_bi = female gender; CCI = Charlson Comorbidity Index; E = number of emergency department visits in 6 months prior to surgery; Fix_bi = fracture diagnosis
Appendix D: Multiple logistic regression analysis for readmissions
Multiple logistic regression analysis, split by payment model (CJR vs. BPCI). Model 1 contains univariate analysis of cemented vs. cementless fixation and Model 2 includes multivariate analysis accounting for confounding variables: age, gender, Charlson Comorbidity Index, number of emergency department visits in the 6 months prior to index surgery, and indication for surgery (fracture vs. elective).
Variables in the Equation 95% C.I.for EXP(B) Cohort CJR
B
Step 1
a
BPCI Step 1a
FinalCement(1) Constant FinalCement(1) Constant
S.E.
.500 -2.479 -.013 -2.960
.269
Wald
df
Sig.
Exp(B)
Lower
Upper
.972
2.795
.455
2.139
3.443
1
.064
1.649
.136 334.609
1
.000
.084
.395
.001
1
.973
.987
.197 224.895
1
.000
.052
a. Variable(s) entered on step 1: FinalCement.
Variables in the Equation 95% C.I.for EXP(B) Cohort
B
S.E.
Wald
df
Sig.
Exp(B)
Lower
Upper
CJR
Step
FinalCement(1)
-.581
.341
2.899
1
.089
.559
.286
1.092
1a
AgeofSurg
.016
.016
1.001
1
.317
1.016
.985
1.049
Gender_bi(1)
.281
.287
.957
1
.328
1.324
.754
2.325
CCI_sum
.019
.066
.080
1
.778
1.019
.895
1.160
E
.717
.088 65.815
1
.000
2.049
1.723
2.436
Fx_bi(1)
.496
.321
2.391
1
.122
1.642
.876
3.080
1.155 17.726
1
.000
.008
Constant BPCI Step 1a
FinalCement(1)
-4.865 -.318
.435
.535
1
.464
.728
.310
1.706
.010
.025
.169
1
.681
1.010
.962
1.061
-.163
.365
.199
1
.656
.850
.416
1.738
CCI_sum
.229
.080
8.132
1
.004
1.258
1.074
1.472
E
.165
.149
1.224
1
.269
1.179
.880
1.580
Fx_bi(1)
-.050
.476
.011
1
.917
.952
.375
2.417
Constant
-4.636
1.768
6.877
1
.009
.010
AgeofSurg Gender_bi(1)
a. Variable(s) entered on step 1: AgeofSurg, Gender_bi, CCI_sum, E, Fx_bi. CJR = Comprehensive Care for Joint Replacement; BPCI = Bundled Payment for Care Improvement; Final Cement = cemented fixation; Age of Surg = patient age in years; Gender_bi = female gender; CCI = Charlson Comorbidity Index; E = number of emergency department visits in 6 months prior to surgery; Fix_bi = fracture diagnosis