Nej Mo A 1906190

The n e w e ng l a n d j o u r na l of m e dic i n e
Original Article
Total Hip Arthroplasty or Hemiarthroplasty

for Hip Fracture
The HEALTH Investigators*
A BS T R AC T
BACKGROUND
Globally, hip fractures are among the top 10 causes of disability in adults. For dis- The members of the writing committee
placed femoral neck fractures, there remains uncertainty regarding the effect of a (Mohit Bhandari, M.D., Ph.D. [cochair],
Thomas A. Einhorn, M.D., Gordon Guy-
total hip arthroplasty as compared with hemiarthroplasty. att, M.D., Emil H. Schemitsch, M.D.,
Robert D. Zura, M.D., Sheila Sprague,
METHODS Ph.D., Frede Frihagen, M.D., Ph.D., Er-
We randomly assigned 1495 patients who were 50 years of age or older and had a nesto Guerra-Farfán, M.D., Ydo V. Klein-
lugtenbelt, M.D., Ph.D., Rudolf W. Pool-
displaced femoral neck fracture to undergo either total hip arthroplasty or hemi- man, M.D., Ph.D., Amar Rangan, Ch.M.,
arthroplasty. All enrolled patients had been able to ambulate without the assis- F.R.C.S., Sofia Bzovsky, M.Sc., Diane
tance of another person before the fracture occurred. The trial was conducted in Heels-Ansdell, M.Sc., Lehana Thabane,
Ph.D., Stephen D. Walter, Ph.D., and P.J.
80 centers in 10 countries. The primary end point was a secondary hip procedure Devereaux, M.D., Ph.D. [cochair]) as-
within 24 months of follow-up. Secondary end points included death, serious ad- sume responsibility for the overall con-
verse events, hip-related complications, health-related quality of life, function, and tent and integrity of this article. The affili-
ations of the members of the writing
overall health end points. committee are listed in the Appendix.
Address reprint requests to Dr. Bhandari
RESULTS at the Division of Orthopaedic Surgery,
The primary end point occurred in 57 of 718 patients (7.9%) who were randomly McMaster University, 293 Wellington
assigned to total hip arthroplasty and 60 of 723 patients (8.3%) who were ran- Street N., Suite 110, Hamilton, ON L8L 8E7,
Canada, or at [email protected].
domly assigned to hemiarthroplasty (hazard ratio, 0.95; 95% confidence interval
[CI], 0.64 to 1.40; P = 0.79). Hip instability or dislocation occurred in 34 patients *A complete list of members of the Hip
Fracture Evaluation with Alternatives of
(4.7%) assigned to total hip arthroplasty and 17 patients (2.4%) assigned to hemi- Total Hip Arthroplasty versus Hemi-
arthroplasty (hazard ratio, 2.00; 99% CI, 0.97 to 4.09). Function, as measured with Arthroplasty (HEALTH) Investigators is
the total Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) provided in the Supplementary Appen-
dix, available at NEJM.org.
total score, pain score, stiffness score, and function score, modestly favored total
hip arthroplasty over hemiarthroplasty. Mortality was similar in the two treatment This article was published on September
26, 2019, at NEJM.org.
groups (14.3% among the patients assigned to total hip arthroplasty and 13.1%
among those assigned to hemiarthroplasty, P = 0.48). Serious adverse events occurred DOI: 10.1056/NEJMoa1906190
Copyright © 2019 Massachusetts Medical Society.
in 300 patients (41.8%) assigned to total hip arthroplasty and in 265 patients (36.7%)
assigned to hemiarthroplasty.
CONCLUSIONS
Among independently ambulating patients with displaced femoral neck fractures,
the incidence of secondary procedures did not differ significantly between patients
who were randomly assigned to undergo total hip arthroplasty and those who were
assigned to undergo hemiarthroplasty, and total hip arthroplasty provided a clinically
unimportant improvement over hemiarthroplasty in function and quality of life over
24 months. (Funded by the Canadian Institutes of Health Research and others;
ClinicalTrials.gov number, NCT00556842.)
n engl j med nejm.org 1

The New England Journal of Medicine
Downloaded from nejm.org at UNIVERSITY OF NEW SOUTH WALES on September 26, 2019. For personal use only. No other uses without permission.
Copyright © 2019 Massachusetts Medical Society. All rights reserved.
P
atients with a hip fracture are at data; or the preparation, review, or approval of the
substantial risk for death, health compli- manuscript for submission. The Methods Cen-
cations, and reduced quality of life.1-4 De- ter, located at McMaster University, coordinated
spite the high frequency of the injury, the way in the trial and was responsible for randomization,
which displaced femoral neck fractures in elderly maintenance, validation, data analysis, and trial-
patients should be managed surgically remains center coordination. Boston Medical Center and
uncertain.2 Options include hemiarthroplasty, New York University assisted in the coordination
which involves replacing the femoral head with of trial sites in the United States. The steering
a prosthesis, or total hip arthroplasty, which in- committee, chaired by the principal investigators,
volves replacement of both the femoral head and designed the trial and the prespecified statistical
the acetabulum with prostheses. Advocates of total analysis plan. The members of the steering com-
hip arthroplasty perceive benefits with regard to mittee (listed in Section S1) vouch for the com-
patient function and quality of life as compared pleteness and accuracy of the data and for adher-
with hemiarthroplasty. There are concerns, how- ence of the trial to the protocol.
ever, that total hip arthroplasty has greater asso-
ciated surgical morbidity than hemiarthroplasty Patients
and may increase the risk of dislocation, which We enrolled patients at 80 participating sites in
often leads to a secondary procedure to reduce Canada, the United States, Spain, the United King-
or revise the prosthesis.1 Meta-analyses of studies dom, the Netherlands, Norway, Finland, Australia,
involving patients with a displaced hip fracture New Zealand, and South Africa. To be eligible
have suggested that total hip arthroplasty results for participation, patients had to be 50 years of
in fewer reoperations and substantially better func- age or older, had to have a low-energy displaced
tion than hemiarthroplasty.4-6 fracture of the femoral neck that was planned to
We performed the Hip Fracture Evaluation be treated with surgery, and had to have been able
with Alternatives of Total Hip Arthroplasty versus to ambulate without the assistance of another
Hemi-Arthroplasty (HEALTH) trial, an expertise- person before the hip fracture occurred.
based randomized, controlled trial involving pa-
tients with a displaced femoral neck hip fracture, Procedures
to examine the effect that total hip arthroplasty, Patients were assigned to undergo either total
as compared with hemiarthroplasty, has on the hip arthroplasty or hemiarthroplasty. Minimiza-
risk of a secondary hip procedure. tion was used within each center to ensure prog-
nostic balance between the treatment groups with
regard to age, prefracture living setting, prefrac-
Me thods
ture functional status, and American Society for
Trial Design Anesthesiologists (ASA) physical status.2,8,9
Our trial was an international, expertise-based, Surgeons, patients, end-point adjudicators, and
randomized, controlled trial. Details of the trial research coordinators who assessed patient end
objectives and design have been published previ- points were aware of the assigned treatment
ously.7 The protocol is available with the full text groups. The data analyst remained unaware of the
of this article at NEJM.org. Additional informa- treatment groups throughout the trial and while
tion about the eligibility criteria, interventions, performing analyses.
follow-up, outcome definitions, and statistical To facilitate the expertise-based trial design,10
analysis are provided in Section S2 in the Supple- we set expertise thresholds for surgeons’ partici-
mentary Appendix, available at NEJM.org. pation. Among the 523 participating surgeons,
277 of 283 (97.9%) of those who performed total
Trial Oversight hip arthroplasty and 369 of 381 (96.9%) of those
The trial was funded by the Canadian Institutes of who performed hemiarthroplasty met thresholds
Health Research, the National Institutes of Health, for surgical expertise. Patients underwent assess-
and others. The funding sources had no role in ment at 1 week, 10 weeks, and 6, 9, 12, 18, and
the design or conduct of the trial; the collection, 24 months after surgery, either in person or by
management, analysis, or interpretation of the telephone.
2 n engl j med nejm.org

End Points 1 year and a 45% lower relative risk of the pri-
The primary end point was any unplanned sec- mary end point at 2 years in the hemiarthroplasty
ondary hip procedure within 24 months after the group than in the total hip arthroplasty group.
initial surgery. The procedures included closed Our sample-size calculation reflected the proposed
and open reductions of a hip dislocation, open approach to the primary analysis, which used the
reduction of a fracture, full or partial implant proportional-hazards model.15 The sample size
exchange, implant removal, implant adjustment, was based on a two-sided test with an alpha of
soft-tissue procedure, excision of heterotopic os- 0.05 and included adjustment for surgeon-level
sification, insertion of an antibiotic spacer, and effects and for the expectation that 7.6% of pa-
other events as determined by an independent tients would cross over from their randomly as-
central adjudication committee. Secondary end signed group.
points included death, serious adverse events, hip- Analyses included patients in the groups to
related complications, health-related quality of life, which they had been randomly assigned. Data
function, and overall health measures. Assess- for a given patient were censored at 24 months
ments of function and quality of life included the of follow-up or at the time of the last follow-up
Western Ontario and McMaster Universities Osteo- for patients who were lost to follow-up. The pri-
arthritis Index (WOMAC) total score (range, 0 to mary analysis was conducted with a proportional-
96, with higher scores indicating worse pain, hazards model and a competing-risk framework
stiffness, and function; the minimal clinically im- (with death as the competing risk), with the per-
portant difference calculated in different studies centage of patients with a primary end-point event,
ranges from 9 to 22), pain score (range, 0 to 20), analyzed in a time-to-event analysis, as the out-
stiffness score (range, 0 to 8), and function score come. The independent variable was the proce-
(range, 0 to 68)11-14; the European Quality of Life–5 dure (total hip arthroplasty or hemiarthroplasty),
Dimensions (EQ-5D) utility index score and visual and the age, prefracture living setting, prefracture
analogue scale; the 12-Item Short Form General functional status, and ASA status were used as
Health Survey (SF-12) physical and mental com- covariates. For our competing-risk analyses, we
ponent summary scores; and Timed Up and Go reported marginal estimates and used the method
(TUG) scores. described by Zhou et al. to account for clustering
The central adjudication committee reviewed of data according to surgeon.16 We report the
secondary procedures to confirm the type of and treatment effects as hazard ratios with 95% con-
reason for the procedure, as well as to confirm key fidence intervals. Kaplan–Meier curves were con-
secondary end points (death and hip-related com- structed for the primary end point.
plications). Separate independent data and safety Cox proportional-hazards modeling was used
monitoring boards coordinated by the National to estimate the relative effect of total hip arthro-
Institutes of Health and Canadian Institutes of plasty as compared with hemiarthroplasty on time
Health Research monitored trial safety and re- to death and serious adverse events. Proportional-
viewed all serious adverse events. hazards modeling with a competing-risk analysis
(with death as the competing risk) was used to
Statistical Analysis provide a marginal estimate of the relative effect
In 2006, during the initial vanguard phase of the of total hip arthroplasty as compared with hemi-
trial, which had the primary goal of assessing arthroplasty on the time to hip-related complica-
feasibility, the definitive trial was conceived as a tions. We used multilevel models to estimate the
noninferiority trial with an anticipated sample effect of total hip arthroplasty as compared with
of 2500 patients. However, during the transition hemiarthroplasty on quality of life, function, and
phase to the definitive trial (358 patients), we made mobility. In our multilevel analyses, we used joint
the strategic decision to switch to a more feasible modeling to account for death, using the method
superiority design. described by Rizopoulos.17 We analyzed the TUG
The planned sample size was 1434 patients as a dichotomous end point. For our multilevel
(717 patients per treatment group), calculated analyses of quality of life, function, and mobility,
under the assumption of a 5% risk of the pri- all available data were used, with no imputation
mary end point in the hemiarthroplasty group at performed. The models did not require that a pa-

tient have end-point scores at all follow-up visits. hemiarthroplasty, and 21 (2.9%) who had origi-
We chose alpha levels of 0.05 and 0.01 for the nally been assigned to hemiarthroplasty received
primary and secondary end points, respectively. All total hip arthroplasty (P<0.001). The frequency of
tests were two-sided, and no adjustments were crossing over did not vary substantially according
made for multiple testing. to country. (Details regarding the patients who
Before unblinding, we prespecified four sub- crossed over and of surgical and postoperative
group analyses that were conducted to investigate care are provided in Tables S6 through S10.)
possible effect modification according to age, pre-
fracture living setting, prefracture functional sta- Primary End Point
tus, and ASA status. The primary end point was A secondary hip procedure within 24 months of
the dependent variable for these analyses. Because follow-up occurred in 57 of 718 patients (7.9%)
these were exploratory analyses, we did not make who had been randomly assigned to total hip ar-
any adjustment for multiple testing within our throplasty and in 60 of 723 patients (8.3%) who
subgroup analyses. We also performed multiple had been randomly assigned to hemiarthroplasty
sensitivity analyses. (hazard ratio, 0.95; 95% confidence interval [CI],
We first interpreted the results on the basis 0.64 to 1.40; P = 0.79) (Fig. 1 and Table 2). The
of a blinded review of the results of our primary Kaplan–Meier curves show that the assumption
analysis.18 The randomization code was then bro- of proportional hazards for the primary end point
ken, the correct interpretation was chosen, and the was not met — that is, the hazard ratio for the
manuscript was written. All analyses were per- relative effects of the two interventions changed
formed with SAS, version 9.4 (SAS Institute), substantially over time. Post hoc analyses eluci-
and R, version 3.6.0 (R Project for Statistical Com- dated the nature of this change in effect over
puting). time: the risk of a secondary hip procedure up
to 1 year was higher in the total hip arthroplasty
group than in the hemiarthroplasty group (hazard
R e sult s
ratio, 1.23; 95% CI, 0.82 to 1.86; P = 0.32); after
Patients 1 year and up to 2 years, the risk was higher in
From January 2009 through May 2017, we ran- the hemiarthroplasty group than in the total hip
domly assigned 1495 patients to undergo total hip arthroplasty group (hazard ratio, 0.23; 95% CI,
arthroplasty (749 patients) or hemiarthroplasty 0.08 to 0.69; P = 0.01) (Table S11). In light of the
(746 patients). The final 24-month assessments finding of a non–proportional-hazards function,
were completed in May 2019. Of the 1495 patients we conducted a prespecified analysis of the inter-
who underwent randomization, 1441 were in- action between time and treatment. We found that
cluded in the final analyses. Of the 1243 patients the log of the hazard ratio decreased by 0.097 each
who were alive at 24 months, 1058 (85.1%) had month (95% CI, 0.031 to 0.162; P = 0.004), which
24-month follow-up data available for the analysis was also inconsistent with the plot of the esti-
of the primary end point. (Details regarding pa- mated log of the hazard ratio over time (Fig. S2).
tient flow and the reasons for exclusion are pro- Sensitivity analyses of various assumptions re-
vided in Fig. S1 and Tables S1 and S2.) garding the risk of the primary end point in pa-
The majority of patients were female (70.1%), tients who were lost to follow-up showed no sig-
70 years of age or older (80.2%), and able to am- nificant difference between the treatment groups.
bulate without the aid of an assistive device be- All other sensitivity analyses showed results simi-
fore their fracture (74.4%), and the injury in the lar to those in the primary analysis. Hip disloca-
majority of the patients was a subcapital femoral tions that were treated with open or closed reduc-
neck fracture (61.9%). The baseline characteris- tion were the most common secondary procedure
tics were similar in the two treatment groups in the total hip arthroplasty group (33 of 57 proce-
(Table 1 and Tables S3 through S5). dures), and implant revisions were the most com-
mon secondary procedure in the hemiarthroplasty
Surgical Care and Adherence to group (36 of 60 procedures). Subgroup analyses
the Assigned Intervention did not show any effect modification. (Details
A total of 54 patients (7.5%) who had originally regarding the sensitivity and subgroup analyses
been assigned to total hip arthroplasty received are provided in Tables S12, S13, and S15.)

Table 1. Patient Demographic Characteristics.*
Total Hip Arthroplasty Hemiarthroplasty

Characteristic (N = 718) (N = 723)
Age — yr† 79.1±8.3 78.6±8.6
Age — no./total no. (%)
50 to 70 yr 136/718 (18.9) 149/722 (20.6)
71 to 80 yr 249/718 (34.7) 247/722 (34.2)
≥81 yr 333/718 (46.4) 326/722 (45.2)
Sex — no./total no. (%)
Male 208/718 (29.0) 223/722 (30.9)
Female 510/718 (71.0) 499/722 (69.1)
Race or ethnic group — no./total no. (%)‡
Native or aboriginal 2/716 (0.3) 1/721 (0.1)
South Asian 3/716 (0.4) 6/721 (0.8)
East Asian 7/716 (1.0) 7/721 (1.0)
Hispanic or Latino 7/716 (1.0) 6/721 (0.8)
White 683/716 (95.4) 684/721 (94.9)
Black 12/716 (1.7) 15/721 (2.1)
Middle Eastern 2/716 (0.3) 2/721 (0.3)
Body-mass index — no./total no. (%)§
Underweight, <18.5 35/697 (5.0) 38/705 (5.4)
Normal weight, 18.5–24.9 357/697 (51.2) 336/705 (47.7)
Overweight, 25–29.9 217/697 (31.1) 243/705 (34.5)
Obese, 30–39.9 77/697 (11.0) 83/705 (11.8)
Morbidly obese, ≥40 11/697 (1.6) 5/705 (0.7)
Prefracture living status — no./total no. (%)
Institutionalized 30/718 (4.2) 27/723 (3.7)
Not institutionalized 688/718 (95.8) 696/723 (96.3)
Prefracture functional status — no./total no. (%)
Uses assistive device for ambulation 187/718 (26.0) 182/723 (25.2)
Able to ambulate without assistive device 531/718 (74.0) 541/723 (74.8)
Previous surgery to affected hip — no./total no. (%) 2/714 (0.3) 1/722 (0.1)
Major coexisting conditions — no./total no. (%)
Osteopenia 28/715 (3.9) 30/722 (4.2)
Osteoporosis 114/715 (15.9) 110/722 (15.2)
Lung disease 127/715 (17.8) 122/722 (16.9)
Diabetes 135/715 (18.9) 145/722 (20.1)
Ulcers or stomach disease 49/715 (6.9) 67/722 (9.3)
Kidney disease 71/715 (9.9) 67/722 (9.3)
Anemia or other blood disease 48/715 (6.7) 55/722 (7.6)
Depression 70/715 (9.8) 84/722 (11.6)
Cancer 65/715 (9.1) 80/722 (11.1)
Osteoarthritis, degenerative arthritis 111/715 (15.5) 91/722 (12.6)
Back pain 64/715 (9.0) 71/722 (9.8)
Rheumatoid arthritis 13/715 (1.8) 21/722 (2.9)
Heart disease 247/715 (34.5) 249/722 (34.5)
High blood pressure 434/715 (60.7) 443/722 (61.4)
* Plus–minus values are means ±SD. Percentages may not total 100 because of rounding.
† Data were missing for one patient in the hemiarthroplasty group.
‡ Race and ethnic group were reported by the patients.
§ Body-mass index is the weight in kilograms divided by the square of the height in meters.

the threshold for a minimal clinically important

1.0 difference for WOMAC (range, 9 to 22 points).
Proportion without Secondary Hip Surgery
0.9 EQ-5D visual analogue scale scores, the 12-Item

1.00
0.8 Short Form General Health Survey (SF-12) physi-
cal and mental component summary scores, and
0.7
TUG scores did not differ significantly between
0.6 0.95 Hemiarthroplasty the treatment groups during follow-up (Table 4).
0.5 The characteristics of the patients who did and
0.4 Total hip arthroplasty
those who did not have data included in analyses
0.3
0.90 of health-related quality of life, function, or over-
0.00 all health end points are shown in Table S14.
0.2
0 100 200 300 400 500 600 700
0.1
Discussion
0.0
0 100 200 300 400 500 600 700
Among patients with displaced fractures of the
Days since Surgery femoral neck, we found that the type of arthro-
plasty had no significant influence on the risk of
Figure 1. Kaplan–Meier Curves for the Primary End Point.
unplanned secondary hip procedures over 24
The primary end point was any unplanned secondary hip procedure within
24 months after the initial surgery. The inset shows the same data on an months. Functional end points favored total hip
expanded y axis. arthroplasty over hemiarthroplasty during the
24-month period. Patients who underwent total
hip arthroplasty had a slightly higher incidence
Secondary End Points of serious adverse events.
Over 24 months, 198 of 1441 patients (13.7%) died. Strengths of our trial included the concealed
Mortality did not differ significantly between the randomization, expertise-based design,10 indepen-
treatment groups (14.3% in the total hip arthro- dent adjudication of primary end-point events, and
plasty group and 13.1% in the hemiarthroplasty safeguards against interpretation bias.18 Nearly all
group, P = 0.48) (Table 2). Serious adverse events participating surgeons (97.8%) met thresholds for
occurred in 300 of 718 patients (41.8%) in the surgical expertise.
total hip arthroplasty group and in 265 of 723 Our trial had certain limitations. Patients and
patients (36.7%) in the hemiarthroplasty group end-point assessors were unblinded in the assess-
(hazard ratio, 1.16; 99% CI, 0.90 to 1.51) (Table 3). ments of function, which left a possibility of
Overall, hip-related complications were more fre- bias. The percentage loss to follow-up was 14.9%
quent with total hip arthroplasty (Table 2). Hip for the analysis of our primary end point. Base-
instability or dislocation occurred in 34 patients line characteristics of patients who were lost to
(4.7%) who were assigned to total hip arthro- follow-up were similar to those of patients we
plasty and 17 patients (2.4%) who were assigned followed. Sensitivity analyses in which patients
to hemiarthroplasty (hazard ratio, 2.00; 99% CI, in the total hip arthroplasty group who were lost
0.97 to 4.09). to follow-up were assumed to have had a risk of
Over 24 months, the functional assessment a primary event that was up to 4 times as high
tests and quality-of-life questionnaires were com- as the risk among those with complete follow-up
pleted by 943 to 1141 patients (depending on the did not alter our principal findings. Data on func-
test) at one or more follow-up visits (Table 4). Pa- tion during follow-up were incomplete; 82.9% of
tients who underwent total hip arthroplasty had patients completed at least one follow-up ques-
superior function as measured by the WOMAC tionnaire over 24 months, with complete data
total score (mean difference, −6.37; 99% CI, −9.18 from follow-up questionnaires available for 46.8%
to −3.56), WOMAC pain score (mean difference, of patients at 12 months and for 42.1% of pa-
−0.93; 99% CI, −1.42 to −0.44), WOMAC stiffness tients at 24 months. We did not use imputation
score (mean difference, −0.44; 99% CI, −0.65 to to handle missing data; however, our analysis
−0.23), and WOMAC function score (mean differ- model did not require that a patient have end-
ence, −4.97; 99% CI, −7.11 to −2.83). These differ- point scores at all follow-up visits, and it performs
ences between the treatment groups fell below well as compared with other approaches for han-

Table 2. Trial End Points.
Total Hip
Arthroplasty Hemiarthroplasty Hazard Ratio
End Point (N = 718) (N = 723) (95% or 99% CI)* P Value†
number (percent)
Primary end point: unplanned secondary 57 (7.9) 60 (8.3) 0.95 (0.64–1.40)‡ 0.79
procedure
Components of primary end point§
Closed reduction of hip dislocation 29 (4.0) 12 (1.7)
Open reduction of hip dislocation 4 (0.6) 2 (0.3)
Open reduction of fracture 5 (0.7) 8 (1.1)
Soft-tissue procedure 15 (2.1) 15 (2.1)
Insertion of antibiotic spacer 3 (0.4) 3 (0.4)
Full implant exchange 7 (1.0) 18 (2.5)
Partial implant exchange 19 (2.6) 18 (2.5)
Implant adjustment: reorientation of stem 0 2 (0.3)
Implant adjustment: reorientation of 2 (0.3) 0
acetabulum component
Implant removal with no replacement 3 (0.4) 3 (0.4)
Excision heterotopic ossification 0 0
Supplementary fixation 3 (0.4) 1 (0.1)
Other 1 (0.1) 3 (0.4)
Secondary end points
Death 103 (14.3) 95 (13.1) 1.10 (0.77–1.58) 0.48
Serious adverse event¶ 300 (41.8) 265 (36.7) 1.16 (0.90–1.51) 0.13
Any hip-related complication∥ 132 (18.4) 118 (16.3) 1.13 (0.81–1.57)
Periprosthetic fracture 38 (5.3) 35 (4.8) 1.08 (0.61–1.88)
Hip instability or dislocation** 34 (4.7) 17 (2.4) 2.00 (0.97–4.09)
Superficial surgical-site infection 9 (1.3) 6 (0.8)
Deep surgical-site infection 17 (2.4) 16 (2.2)
Another wound-healing problem 6 (0.8) 5 (0.7)
Another soft-tissue procedure 11 (1.5) 11 (1.5)
Clinically important heterotopic 29 (4.0) 24 (3.3) 1.19 (0.62–2.30)
ossification††
Abductor failure 1 (0.1) 3 (0.4)
Implant failure: loosening or subsidence 5 (0.7) 5 (0.7)
Implant failure: breakage 1 (0.1) 0
Pain 6 (0.8) 12 (1.7)
Neurovascular injury: technical error 2 (0.3) 1 (0.1)
Other 7 (1.0) 13 (1.8)
* The hazard ratio is for total hip arthroplasty as compared with hemiarthroplasty; the 95% confidence interval (CI) is
given for the primary end point, and 99% confidence intervals are given for the secondary end points. Proportional-
hazards regressions were performed only for hip-related complications for which at least 50 events occurred.
† P values are from regression models of subdistribution hazards.
‡ The proportional-hazards assumption was violated, and analyses of year 1 as compared with year 2 followed a post
hoc analysis guided by the Kaplan–Meier curve.
§ The numbers for the individual components add up to more than the total number of patients with the primary end
point because some patients had more than one event.
¶ The marginal estimate for the competing-risk analysis is shown in the table; the conditional estimate for serious ad-
verse events was a hazard ratio of 1.19 (99% CI, 0.94 to 1.50; P = 0.06).
∥ The numbers for specific hip-related complications add up to more than the overall total number of patients with hip-
related complications because some patients had more than one event.
** One patient in the total hip arthroplasty group and three patients in the hemiarthroplasty group who had hip instabil-
ity or dislocation were not treated operatively for this complication.
†† Clinically important heterotopic ossification was defined as class 3 or higher according to the Brooker classification.

Table 3. Serious Adverse Events.
Total Hip Arthroplasty Hemiarthroplasty

End Point (N = 718) (N = 723) P Value*
number (percent)
Any serious adverse event† 300 (41.8) 265 (36.7) 0.13
Hip fracture–related serious adverse event 59 (8.2) 57 (7.9) 0.85
Neurologic serious adverse event 28 (3.9) 26 (3.6) 0.78
Respiratory serious adverse event 42 (5.8) 37 (5.1) 0.56
Cardiac serious adverse event 51 (7.1) 49 (6.8) 0.84
Renal serious adverse event 23 (3.2) 22 (3.0) 0.88
Vascular serious adverse event 22 (3.1) 16 (2.2) 0.33
Other serious adverse event 201 (28.0) 177 (24.5) 0.14
Non–trial-related fracture‡ 50 (7.0) 37 (5.1)
Non–trial-related dislocation‡ 2 (0.3) 0
Other non–trial-related injury‡ 10 (1.4) 14 (1.9)
Cellulitis 2 (0.3) 2 (0.3)
Death 103 (14.3) 95 (13.1)
Multiorgan failure 1 (0.1) 2 (0.3)
Osteoporosis, new or worsening 0 1 (0.1)
Sepsis 9 (1.3) 8 (1.1)
Reported by site as “other” 66 (9.2) 60 (8.3)
* All P values were calculated with Fisher’s exact test with the exception of “any serious adverse event,” for which the
P value was calculated with the Cox model.
† The numbers for individual serious adverse events add up to more than the total because some patients had more
than one type of serious adverse event.
‡ The determination of whether an injury was related to the trial was made by the attending surgeon.
dling missing patient data.19 The follow-up period over a period of 16 years from the Australian
in our trial may have been insufficient for under- Joint Registry suggest that modern hemiarthro-
standing longer-term end points. plasty and total hip arthroplasty have similar in-
Our results differ from those of published cidences of revision over 10 years (hazard ratio,
meta-analyses reporting that the risk of reopera- 1.13; 95% CI, 0.95 to 1.33; P = 0.16).20 A large mul-
tion associated with total hip arthroplasty is 34 ticenter trial involving 298 patients with femoral
to 43% lower than with hemiarthroplasty.4-6 This neck fractures in Scotland showed no significant
difference may, in part, be the consequence of difference in the incidence of secondary hip pro-
longer follow-up periods in these meta-analyses cedures at 2 years (hazard ratio, 0.81; 95% CI,
(ranging from 1 to 13 years) or potential differ- 0.24 to 2.81; P = 0.74).21 Our subgroup analyses
ences in our eligibility criteria.4-6 did not show any effect modification according
Our inclusion of patients who could ambu- to patient age, the use of assistive devices for
late independently before the fracture occurred ambulation, ASA status, or living status at the
(i.e., patients who did not require the assistance time of fracture. The larger number of events in
of another person to ambulate) is consistent with the hemiarthroplasty group during the second
previous trials.6 Although the use of assistive year is in keeping with the possibility of more
devices was balanced between the treatment events being associated with that procedure over
groups, we did not record the type of assistive the long term.
device or the specific ambulatory capacity of the A meta-analysis reported significantly fewer
patients. Data on 17,985 femoral neck fractures complications with total hip arthroplasty than

with hemiarthroplasty (relative risk, 0.75; 95% CI, Table 4. Health-Related Quality of Life, Function, and Overall Health End
0.60 to 0.94; P<0.05).4 In our trial, the incidence Points.
of serious adverse events was slightly higher in
the total hip arthroplasty group than in the hemi- Patients Mean Difference in Score
with Data at 24 Mo, Total Hip Arthroplasty
arthroplasty group. Serious adverse events were End Point (N = 1441) vs. Hemiarthroplasty (99% CI)*
not driven by a specific type of complication;
rather, total hip arthroplasty broadly led to more no. (%)
cardiac, renal, vascular, neurologic, and respira- WOMAC total score† 943 (65.4) −6.37 (−9.18 to −3.56)
tory events than hemiarthroplasty. The incidence WOMAC pain score† 990 (68.7) −0.93 (−1.42 to −0.44)
of dislocation after total hip arthroplasty in our WOMAC stiffness score† 987 (68.5) −0.44 (−0.65 to −0.23)
trial was high (4.7%). Pooled estimates from ran-
WOMAC function score† 947 (65.7) −4.97 (−7.11 to −2.83)
domized, controlled trials suggest a risk of disloca-
EQ-5D utility index score‡ 1141 (79.2) 0.04 (−0.03 to 0.11)
tion after total hip arthroplasty that is more than
2.5 times as great as that associated with hemi- EQ-5D VAS score‡ 1111 (77.1) 0.72 (−2.02 to 3.46)
arthroplasty (9% vs. 3%; relative risk, 2.53; 95% CI, SF-12 PCS§ 1006 (69.8) 1.41 (−0.33 to 3.14)
1.05 to 6.10).5 SF-12 MCS§ 1006 (69.8) 1.34 (−0.38 to 3.05)
Although limited comparative data are avail- Odds Ratio (99% CI)
able, total hip arthroplasty has consistently been TUG¶ 1268 (88.0) 0.72 (0.38 to 1.36)
associated with better function and quality of
life than hemiarthroplasty in previous studies.4-6 * The mean difference was obtained from the multilevel model.
Our findings showed slightly but significantly † Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC)
total scores range from 0 to 96, with higher scores indicating worse pain, stiff-
lower WOMAC scores (indicating better function) ness, and function; it is the sum of the pain score (range, 0 to 20), stiffness
in the total hip arthroplasty group, as well as score (range, 0 to 8), and function score (range, 0 to 68). For the total score,
trends that favored total hip arthroplasty in EQ-5D, the minimal clinically important difference calculated in different studies rang-
es from 9 to 22.
SF-12 scores, and TUG test times. However, the ‡ The European Quality of Life–5 Dimensions (EQ-5D) measures quality of life
differences in WOMAC scores were deemed clini- in five dimensions; utility scores range from −0.109 to 1, with higher scores in-
cally unimportant on the basis of the thresholds dicating better states of health. Scores on the EQ-5D visual analogue scale
(VAS) range from 0 to 100, with higher scores indicating better states of health.
for the minimal clinically important difference.12,13 § The 12-Item Short Form General Health Survey (SF-12) measures health-relat-
The American Academy of Orthopaedic Sur- ed quality of life and includes a physical composite score (PCS) and a mental
geons and National Institute for Health and Care composite score (MCS). Each composite score ranges from 0 to 100, with
higher scores indicating better states of health.
Excellence guidelines recommend total hip ar- ¶ Timed Up and Go (TUG) results were dichotomized, with patients who took
throplasty in all patients with displaced femoral more than 12 seconds to complete the test or were unable to complete the
neck fractures who are able to ambulate inde- test compared with patients who took 12 seconds or less to complete the
test. The odds ratio (total hip arthroplasty vs. hemiarthroplasty) is for com-
pendently.22,23 Our findings suggest that the ad- pleting the test in more than 12 seconds or not being able to complete the
vantages of total hip arthroplasty may not be test and was obtained from the multilevel model.
compelling. The limited advantages of total hip
arthroplasty, as well as the possible higher risk
of complications, may be particularly important among independently ambulating patients with
in regions of the world where total hip arthro- displaced femoral neck fractures.
plasty is not easily accessible or is cost-prohibitive. Supported by grants from the Canadian Institutes of Health
In our trial, the incidence of secondary proce- Research (MCT-90168), the National Institutes of Health
dures after 2 years did not differ significantly (1UM1AR063386-01), ZorgOnderzoek Nederland-Medische
Wetenschappen (ZonMw) (17088.2503), Sophies Minde Founda-
between the total hip arthroplasty group and the tion for Orthopaedic Research, McMaster Surgical Associates,
hemiarthroplasty group. Total hip arthroplasty and Stryker Orthopaedics.
was associated with modestly better function over Disclosure forms provided by the authors are available with
the full text of this article at NEJM.org.
24 months but with a slightly higher incidence of A data sharing statement provided by the authors is available
serious adverse events than hemiarthroplasty with the full text of this article at NEJM.org.
Appendix
The affiliations of the members of the writing committee are as follows: the Division of Orthopaedic Surgery, Department of Surgery
(M.B., S.S., S.B.), the Department of Health Research Methods, Evidence, and Impact (M.B., G.G., S.S., D.H.-A., L.T., S.D.W., P.J.D.),

the Department of Medicine (G.G., P.J.D.), and the Population Health Research Institute (P.J.D.), McMaster University, Hamilton, and
the Department of Surgery, University of Western Ontario, London (E.H.S.) — all in Ontario, Canada; the Department of Orthopedic
Surgery, New York University Langone Medical Center, New York (T.A.E.); the Department of Orthopedic Surgery, Louisiana State
University Health Sciences Center, New Orleans (R.D.Z.); the Division of Orthopedic Surgery, Oslo University Hospital, Oslo (F.F.); the
Department of Traumatology, Orthopedic Surgery, and Emergency, Hospital Vall d’Hebrón, Barcelona (E.G.-F.); the Department of
Orthopedic and Trauma Surgery, Deventer Ziekenhuis, Deventer (Y.V.K.), and the Department of Orthopedic Surgery, OLVG, Amster-
dam (R.W.P.) — both in the Netherlands; and the Department of Orthopaedic Surgery, James Cook University Hospital, Middlesbrough,
the Nuffield Department of Orthopaedics, Rheumatology, and Musculoskeletal Sciences, University of Oxford, Oxford, and the Depart-
ment of Health Sciences, University of York, York — all in the United Kingdom (A.R.).
References
1. Bhandari M, Swiontkowski M. Man- 8. Pocock SJ. When (not) to stop a clinical the joint modelling of longitudinal and
agement of acute hip fracture. N Engl J trial for benefit. JAMA 2005;294:2228-30. time-to-event data. J Stat Softw 2010;
Med 2017;377:2053-62. 9. Gillespie WJ. Extracts from “clinical 35(9):1-33.
2. Bhandari M, Devereaux PJ, Tornetta P evidence”: hip fracture. BMJ 2001;322: 18. Järvinen TLN, Sihvonen R, Bhandari
III, et al. Operative management of dis- 968-75. M, et al. Blinded interpretation of study
placed femoral neck fractures in elderly 10. Devereaux PJ, Bhandari M, Clarke M, results can feasibly and effectively dimin-
patients: an international survey. J Bone et al. Need for expertise based randomised ish interpretation bias. J Clin Epidemiol
Joint Surg Am 2005;87:2122-30. controlled trials. BMJ 2005;330:88. 2014;67:769-72.
3. Johnell O, Kanis JA. An estimate of 11. Ware JE Jr, Sherbourne CD. The MOS 19. Zhang Y, Alyass A, Vanniyasingam T,
the worldwide prevalence and disability 36-item short-form health survey (SF-36). et al. A systematic survey of the methods
associated with osteoporotic fractures. I. Conceptual framework and item selec- literature on the reporting quality and op-
Osteoporos Int 2006;17:1726-33. tion. Med Care 1992;30:473-83. timal methods of handling participants
4. Hopley C, Stengel D, Ekkernkamp A, 12. Tubach F, Ravaud P, Baron G, et al. with missing outcome data for continu-
Wich M. Primary total hip arthroplasty Evaluation of clinically relevant changes ous outcomes in randomized controlled
versus hemiarthroplasty for displaced in- in patient reported outcomes in knee and trials. J Clin Epidemiol 2017;88:67-80.
tracapsular hip fractures in older pa- hip osteoarthritis: the minimal clinically 20. Australian Orthopaedic Association
tients: systematic review. BMJ 2010;340: important improvement. Ann Rheum Dis National Joint Registry. Hip, knee &
c2332-c2332. 2005;64:29-33. shoulder arthroplasty annual report. 2018
5. Burgers PTPW, Van Geene AR, Van 13. Impellizzeri FM, Mannion AF, Naal (https://aoanjrr.sahmri.com/annual
den Bekerom MPJ, et al. Total hip arthro- FD, Hersche O, Leunig M. The early out- -reports-2018).
plasty versus hemiarthroplasty for dis- come of surgical treatment for femoroac- 21. Keating JF, Grant A, Masson M, Scott
placed femoral neck fractures in the etabular impingement: success depends NW, Forbes JF. Displaced intracapsular
healthy elderly: a meta-analysis and sys- on how you measure it. Osteoarthritis hip fractures in fit, older people: a ran-
tematic review of randomized trials. Int Cartilage 2012;20:638-45. domised comparison of reduction and
Orthop 2012;36:1549-60. 14. Paulsen A, Roos EM, Pedersen AB, fixation, bipolar hemiarthroplasty and
6. Lewis DP, Wæver D, Thorninger R, Overgaard S. Minimal clinically important total hip arthroplasty. Health Technol As-
Donnelly WJ. Hemiarthroplasty vs total improvement (MCII) and patient-accept- sess 2005;9(41):iii-iv, ix.
hip arthroplasty for the management of able symptom state (PASS) in total hip 22. American Academy of Orthopedic Sur-
displaced neck of femur fractures: a sys- arthroplasty (THA) patients 1 year postop- geons. Management of hip fractures in the
tematic review and meta-analysis. J Ar- eratively. Acta Orthop 2014;85:39-48. elderly: summary. 2019 (https://www.aaos
throplasty 2019;34(8):1837.e2-1843.e2. 15. Collett D. Modelling survival data in .org/cc_files/aaosorg/research/guidelines/
7. Bhandari M, Devereaux PJ, Einhorn medical research. London: Chapman & hipfxsummaryofrecommendations.pdf).
TA, et al. Hip Fracture Evaluation with Hall, 1994. 23. Hip fracture overview. London: Na-
Alternatives of Total Hip Arthroplasty 16. Zhou B, Fine J, Latouche A, Labopin tional Institute for Health and Care Excel-
versus Hemiarthroplasty (HEALTH): pro- M. Competing risks regression for clus- lence, 15 August 2018 (https://pathways
tocol for a multicentre randomised trial. tered data. Biostatistics 2012;13:371-83. .nice.org.uk/pathways/hip-fracture).
BMJ Open 2015;5(2):e006263. 17. Rizopoulos D. JM: an R package for Copyright © 2019 Massachusetts Medical Society.


Nej Mo A 1906190

Uploaded by

Copyright:

Available Formats

Nej Mo A 1906190

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Nej Mo A 1906190

Uploaded by

Copyright:

Available Formats

The n e w e ng l a n d j o u r na l of m e dic i n e

Total Hip Arthroplasty or Hemiarthroplasty

n engl j med nejm.org 1

2 n engl j med nejm.org

The New England Journal of Medicine

n engl j med nejm.org 3

4 n engl j med nejm.org

The New England Journal of Medicine

Table 1. Patient Demographic Characteristics.*

Total Hip Arthroplasty Hemiarthroplasty

n engl j med nejm.org 5

the threshold for a minimal clinically important

0.9 EQ-5D visual analogue scale scores, the 12-Item

6 n engl j med nejm.org

The New England Journal of Medicine

Table 2. Trial End Points.

n engl j med nejm.org 7

Table 3. Serious Adverse Events.

Total Hip Arthroplasty Hemiarthroplasty

8 n engl j med nejm.org

The New England Journal of Medicine

n engl j med nejm.org 9

10 n engl j med nejm.org

The New England Journal of Medicine

You might also like