Prospective, randomized, multicenter Food and Drug Administration
investigational device exemption study of lumbar total disc
replacement with the CHARITE´artificial disc versus
lumbar fusion: Five-year follow-up
Richard D. Guyer, MDa,*, Paul C. McAfee, MDb, Robert J. Banco, MDc, Fabian D. Bitan, MDd,
Andrew Cappuccino, MDe, Fred H. Geisler, MD PhDf, Stephen H. Hochschuler, MDa,
Richard T. Holt, MDg, Louis G. Jenis, MDc, Mohamed E. Majd, MDg, John J. Regan, MDh,
Scott G. Tromanhauser, MDc, Douglas C. Wong, MDi, Scott L. Blumenthal, MDa
aTexas Back Institute, 6020 West Parker Road, Suite 200, Plano, TX 75093, USA
bSpine and Scoliosis Center, 7505 Osler Drive, Suite 104, Towson, MD 21204, USA
cBoston Spine Group, 125 Parker Hill Avenue, Boston, MA 02120, USA
dLennox Hill Hospital, 130 East 77th Street, 7th Floor, New York, NY 10003, USA
eBuffalo Spine Surgery, 46 Davison Street, Lockport, NY 14094, USA
fIllinois Neuro-Spine Center, 2020 Ogden Avenue, Suite 335, Aurora, IL 60504, USA
gSpine Surgery, PSC, 210 East Gray Street, Suite 601, Louisville, KY 40202, USA
hG&P Orthopedic Associates, 120 South Spalding Drive, Suite 400, Beverly Hills, CA 90212, USA
iPanorama Orthopedics, 660 Golden Ridge Road, Suite 250, Golden, CO 80401, USA
Received 9 April 2008; accepted 5 August 2008
Abstract BACKGROUND CONTEXT: The CHARITE´artificial disc, a lumbar spinal arthroplasty device,
was approved by the United States Food and Drug Administration in 2004 based on two-year safety
and effectiveness data from a multicenter, prospective, randomized investigational device exemp-
tion (IDE) study. No long-term, randomized, prospective study on the CHARITE´disc or any other
artificial disc has been published to date.
PURPOSE: The purpose of this study was to compare the safety and effectiveness at the five-year
follow-up time point of lumbar total disc replacement using the CHARITE´artificial disc (DePuy
Spine, Raynham, MA) with that of anterior lumbar interbody fusion (ALIF) with BAK cages
and iliac crest autograft, for the treatment of single-level degenerative disc disease from L4 to
S1, unresponsive to nonoperative treatment.
STUDY DESIGN/SETTING: Randomized controlled trialdfive-year follow-up.
PATIENT SAMPLE: Ninety CHARITE´patients and 43 BAK patients.
FDA device/drug status: approved for this indication (CHARITE´disc).
Author disclosures: RDG (research support and royalties from DePuy
Spine; stockholder at K2M, Spinal Motion; consultant for DePuy Spine,
Spinal Motion; member of scientific advisory board at K2M; clinical staff
or training receives educational support from DePuy spine, Abbott Spine,
Synthes Spine, Medtronic, Neuro); PCM (research support and royalties;
stockholder; consultant and speaker; received support for staff and mate-
rials, discretionary funds, trips/travel); RJB (consultant and speaker);
FDB (consultant and speaker; supported for trips/travel); AC (receives roy-
alties and stockholder at JJ DePuy, NuVasive, Pioneer); FHG (royalties
from Aescular, DePuy Spine; stockholder at Disc Motion, Impliant Surgi-
tech, Spinal Integration; consultant for DePuy Spine, Cervitech, Abbott
Spine, K2K and speaker for DePuy Spine; in scientific advisory board at
Disc Motion, K2M; received support for staff and materials from DePuy
Spine, Cervitech, Spinal Motion, Implic); SHH (stockholder, IST,
Spinewave Replication Medical, K2M, Pioneer Surgical Spinal Motion,
Disc Motion, Anvlex, Archus, Anscon; consultant for Orthotic, Abott
Spine, DePuy Spine, Alphatec, Endios; member of board of directors (Al-
phatec, SpineMark), scientific advisory board (Pioneer), faculty for fellow-
ship receiving educational/research support from Abbott Spine, DePuy
Spine, Synthes, Spine & Medtronic Neuro); LGJ (consultant for DePuy
Spine); JJR (royalties from Johnson & Johnson); SGT (consultant for DeP-
uy, Medtronic, Stryker, Zimmer; received research support from Impliant,
Stryker, Zimmer); DCW (royalties from Biomet/EBE; stockholder at Syn-
thes; consultant for DePuy, speaker for Kyphen); SLB (receives royalties
and consultant for DePuy Spine).
* Corresponding author. Texas Back Institute, 6020 West Parker Road,
Suite 200, Plano, TX 75093, USA. Tel.: (972) 608-5114; fax: (972) 608-
E-mail address: firstname.lastname@example.org (R.D. Guyer)
1529-9430/09/$ – see front matter ? 2009 Elsevier Inc. All rights reserved.
The Spine Journal 9 (2009) 374–386
OUTCOME MEASURES: Self-reported measures: visual analog scale (VAS); validated Oswes-
try disability index (ODI version 1.0); Short-Form 36 Questionnaire, and patient satisfaction. Phys-
iologic measures: radiographic range of motion, disc height, and segmental translation. Functional
measures: work status.
METHODS: Of the 375 subjects enrolled in the CHARITE´IDE trial, 277 were eligible for the
five-year study and 160 patients thereof completed the five-year follow-up. The completers in-
cluded 133 randomized patients. Overall success was defined as improvement$15 pts in ODI vs.
baseline, no device failure, absence of major complications, and maintenance or improvement of
neurological status. Additional clinical outcomes included an ODI questionnaire as well as VAS,
SF-36, and patient satisfaction surveys. Work status was tracked for all patients. Safety assessments
included occurrence and severity of adverse events and device failures. Radiographic analyses such
as index- and adjacent-level range of motion, segmental translation, disc height, and longitudinal
ossification were also carried out.
RESULTS: Overall success was 57.8% in the CHARITE´group vs. 51.2% in the BAK group
(Blackwelder’s test: p50.0359, D50.10). In addition, mean changes from baseline for ODI
(CHARITE´: ?24.0 pts vs. BAK: ?27.5 pts), VAS pain scores (CHARITE´: ?38.7 vs. BAK:
?40.0), and SF-36 questionnaires (SF-36 Physical Component Scores [PCS]: CHARITE´:
12.6 pts vs. BAK: 12.3 pts) were similar across groups. In patient satisfaction surveys, 78% of
CHARITE´patients were satisfied vs. 72% of BAK patients. A total of 65.6% patients in the CHAR-
ITE´group vs. 46.5% patients in the BAK group were employed full-time. This difference was statis-
20.9% of BAK patients, a difference that was also statistically significant (p50.0441). Additional in-
dex-level surgery was performed in 7.7% of CHARITE´patients and 16.3% of BAK patients.
Radiographic findings included operative and adjacent-level range of motion (ROM), interverte-
bral disc height and segmental translation. At the five-year follow-up, the mean ROM at the index
level was 6.0?for CHARITE´patients and 1.0?for BAK patients. Changes in disc height were also
similar for both CHARITE´and BAK patients (0.7 mm for both groups, p50.9827). Segmental
translation was 0.4 and 0.8 mm in patients implanted with CHARITE´at L4–L5 vs. L5–S1, respec-
tively, and 0.1 mm in BAK patients.
CONCLUSIONS: The results of this five-year, prospective, randomized multicenter study are
consistent with the two-year reports of noninferiority of CHARITE´artificial disc vs. ALIF with
BAK and iliac crest autograft. No statistical differences were found in clinical outcomes between
groups. In addition, CHARITE´patients reached a statistically greater rate of part- and full-time em-
ployment and a statistically lower rate of long-term disability, compared with BAK patients. Radio-
graphically, the ROMs at index- and adjacent levels were not statistically different from those
observed at two-years postsurgery. ? 2009 Elsevier Inc. All rights reserved.
Keywords: Arthroplasty; Arthrodesis; 5-year follow-up; Randomized controled trial
Spinal arthroplasty has been studied for a number of
years as an alternative to spinal arthrodesis in the surgical
treatment of degenerative disc disease (DDD). The first ar-
throplasty device, the SB CHARITE´I, was implanted as
early as in 1984 . Since that time, this prosthesis under-
went two major design changes  and, in its final form,
was thoroughly tested for biomechanical integrity  and
in vivo effectiveness  before clinical evaluation.
The first clinical experience with the CHARITE´III, also
known as the CHARITE´artificial disc, was reported in
1994 by Griffith et al. . This retrospective study analyzed
charts from 93 patients and reported good overall symptom
relief with respect to back and leg pain at a one-year fol-
low-up time point. Soon thereafter, another retrospective re-
were again satisfactory; however, authors highlighted the
need for prospective randomized studies with longer-term
follow-up periods to assess the validity of the clinical results
obtained so far and to compare arthroplasty with fusion.
The recent publications of several mid- and long-term
arthroplasty studies represent the closest attempts so far
at addressing this shortfall. Lemaire et al. first reported
10-year follow-up results in 100 patients . Overall, the
authors reported excellent or good clinical outcomes in
90% cases. Adjacent-level degeneration was observed in
only 2% of cases, whereas the rate of facet joint degenera-
tion remained relatively minimal (11%).
These promising results were challenged by Putzier
et al. in 2006 in a second, long-term study, reporting 17-
year follow-up data with CHARITE´. In this report,
60% spontaneous ankylosis was observed, prompting au-
thors to conclude that the long-term efficacy of total disc
replacement still needed to be examined. Although this
375 R.D. Guyer et al. / The Spine Journal 9 (2009) 374–386
study presented significant limitations, such as use of un-
dersized prototype devices, suboptimal device placement,
ill-defined patient selection, and incorrect statistical calcu-
lations [1,2], it further reinforced the need for a clinically
relevant, long-term safety and efficacy study, comparing
the CHARITE´device with an established fusion procedure.
In absence thereof, however, David published a third long-
term study, presenting a 10-year data on 106 patients with
one-level DDD . Excellent or good clinical outcome was
obtained in 82.1% of patients, with an 89.6% rate of return
to work. The authors thus concluded that disc replacement
with CHARITE´was a viable alternative to arthrodesis for
the treatment of lumbar DDD.
Concurrent to Lemaire et al.’s publication were two dis-
closures of the CHARITE´investigational device exemption
(IDE) trial designed to evaluate the safety and effectiveness
of the CHARITE´artificial disc vs. anterior lumbar inter-
body fusion (ALIF) with BAK and autograft [10,11]. These
reports demonstrated the noninferiority of the CHARITE´
artificial disc to the BAK fusion at the two-year time point.
The United States Food and Drug Administration (FDA)
granted approval to the CHARITE´artificial disc on the ba-
sis of these results and requested continuation of the study
to a five-year follow-up time point.
The results that follow describe the five-year clinical and
radiographic outcomes for the CHARITE´IDE multicenter,
randomized controlled trial (RCT). This study represents
the first long-term multicenter RCT comparing arthroplasty
Materials and method
Between May 2000 and April 2002, 375 patients had sur-
gery in a prospective, randomized, nonblinded, FDA-
approved study at 14 investigational sites across the United
States. Institutional Review Board approval, patient enroll-
ment, inclusion and exclusion criteria were previously de-
scribed by Blumenthal et al. . Patients were randomly
group was implanted with the CHARITE´Artificial Disc,
whereas the control group was treated by ALIF with BAK
threaded fusion cages (Zimmer Spine, Minneapolis, MN)
packed with iliac crest autograft. Each site was given the op-
portunity to treat up to five nonrandomized CHARITE´cases.
Of the 14 initial sites, 6 declined participation in the 5-
year continuation study, reducing the total number of eligi-
ble patients by 90, 64 randomized and 26 training cases.
Four deaths and 4 device removals reduced the overall eli-
gible patient population to 277 patients (233 randomized
and 44 nonrandomized). Loss to follow-up included 117
patients, of whom 11 declined continuation in the study
and 10 had early discontinuation. A total of 160 patients
completed the five-year study, including 27 nonrandomized
training cases and 133 randomized cases (90 CHARITE´
and 43 BAK patients). Thus, the follow-up rates reached
57% of eligible randomized patient population and 44%
of the total IDE patient cohort. Results from the non-
randomized training cases were not included hereindthis
report focuses exclusively on results from randomized
cases. Protocols for this five-year follow-up study were de-
fined prospectively, before any of the patients were seen.
This study is therefore a prospective study.
Before surgery, most of the patients enrolled in the study
metthe inclusionand exclusioncriteria;however,
The Charite artificial disc has been used in North Ameri-
the relative advantages and durability of the prosthesis
compared to spinal fusion.
The authors present the five-year follow-up on ‘‘available’’
(IDE) trial, comparing the Charite artificial disc to the
(58%) were contacted at five years, and some sites in the
original trial declined to participate at all. In the select
55% of the reporting patients met the ‘‘success’’ standard
Caution is advised. By reporting on only a partial cohort
sample, not randomly selected, this study design intro-
duces a risk of surveillance and reporting bias. The com-
parison to the BAK device, with its acknowledged
limitations, certainly suggests a methodological bias
against fusion compared to modern techniques. The out-
comes in either group are somewhat disappointing. Even
in these highly selected patients, it is unclear these rates
are acceptable. Finally the only device failures in the
two- to five-year period were in the Charite group. For-
tunately, although there was a trend to more re-opera-
tions in the Charite group in that time frame, that rate
(2%) was small in the select patients followed.
376 R.D. Guyer et al. / The Spine Journal 9 (2009) 374–386
osteopenia (an exclusion criterion) was seen in five patients
from the CHARITE´group and one patient from the BAK
group. There was no significant difference between the
two groups with respect to gender, age, race, height, body
mass index, incidence of prior spinal surgery, activity level
before the onset of symptoms, activity level at the time of
enrollment, or preoperative working status (Table 1).
Clinical outcome measurements
Clinical evaluations were completed before surgery and
at 6 weeks and 3, 6, 12, 24, and 60 months postsurgery.
They included a visual analog scale (VAS) for pain, the Os-
westry disability index (ODI), neurological status and
short-form 36 (SF-36) health survey questionnaire, as well
as work status evaluations. The VAS used here ranged from
0 to 100. The ODI tool consisted of the standard ten-item
ordinal scale with six possible responses for each item
and a total score ranging from 0 to 100 (Version 1.0). For
both, VAS and ODI, lower scores were indicative of less
pain/disability. The SF-36 questionnaire was similar to
the previously validated and discussed version . PCS
results are shown herein. Clinical success as defined by
the FDA (nonvalidated clinical scale) at the five-year time
point was achieved when all the following four criteria
were met: 1) $15 pts improvement in ODI; 2) no device
failure; 3) no major complication, and 4) no neurological
change. Device failure included any case of unplanned re-
turn to the operating room for index-level surgery, that is,
supplemental fixation, decompression or reoperation, and
thus addressed device-related safety concerns. Major com-
plications were defined as major vessel injury resulting in
O1500 cc blood loss, neurological damage, or nerve root
injury. No neurological change was defined as lack of neu-
rological deterioration compared with preoperative status,
at any point of time.
Additional surgery for adjacent-level degeneration was
recorded. In addition, the number of patients under pain
management for adjacent-level disc disease was also ana-
lyzed and compared between groups.
All patients underwent surgical treatment through an
open anterior retroperitoneal approach. The surgical tech-
nique for implantation of the CHARITE´artificial disc has
been described previously in great detail by Geisler 
and more recently, in the two-year follow-up release of this
study . Briefly, patients were placed in the supine posi-
tion. Fluoroscopy was used to identify the approach angle
and location of the disc space. The approach was performed
in standard fashion with the assistance of an access surgeon
in most of the cases. A complete discectomy was per-
formed using standard anterior lumbar surgical instruments.
Special instruments were used to assess proper footprint
sizing, lordotic angle, core height, and placement of the
prosthesis within the disc space under fluoroscopy. The
prosthesis end plates were then inserted into the disc space
in a trajectory parallel to the vertebral end plates as deter-
mined from lateral fluoroscopy. Final positioning was as-
sessed with fluoroscopy.
Gender, n (%)
O45 y, n (%)
#45 y, n (%)
Race, n (%)
Body mass index
surgeries, n (%)
level, n (%)
at enrollment, n (%)
status, n (% working)
Fisher exact test used to test categorical variables and t test used to test
377 R.D. Guyer et al. / The Spine Journal 9 (2009) 374–386
Disc space preparation was identical for patients in both
groups. The anterior longitudinal ligament and anterior an-
nulus fibrosis were resected. A complete discectomy was
performed with preservation of the peripheral annulus fi-
brosis to provide ligamentotaxis. Care was taken not to vi-
olate the bony end plates, and in some cases, posterior
osteophytes were carefully removed to allow satisfactory
placement of the prosthesis. The posterior longitudinal lig-
ament was stretched to facilitate restoration of normal disc
space height. In the BAK control group, instruments spe-
cific to the BAK cages were used for sizing of the disc
space. Corticocancellous autograft was harvested from the
iliac crest and packed into two BAK cages. The cages were
placed in the disc space according to the recommended sur-
gical technique. In both groups, the wound was closed in
standard fashion and patients were taken to the recovery
The detail of the radiographic scanning technique was
previously discussed for the full IDE patient population
by McAfee et al. . Briefly, all patients were X-rayed be-
fore surgery and later at all time points (6 weeks and 3, 6,
12, 24, and 60 months). Radiographs included anteroposte-
rior (AP), lateral and full flexion, and extension views. The
radiographic scanning technique was based on that reported
by Kuslich et al. . All radiographs were scanned, digi-
tized, and analyzed by a software program designed to
measure differences between flexion/extension angles and
translations of the operative motion segment using vali-
dated, computer-assisted methods, with mean error less
than 0.5?and confidence interval range of 0.3?, as defined
in cadaver studies  (QMA, Medical Metrics, Houston,
TX). All radiographs were analyzed by the same system,
to avoid inter- and intraobserver variability. Measurements
on all radiographs included flexion/extension ROM at oper-
ated and adjacent levels, segmental translation as well as
longitudinal disc height. Segmental translation, which can
be described as an AP movement of the two spinal verte-
brae in relationship to each other, was analyzed using the
motion of the posterior inferior corner of the superior ver-
tebra in a direction defined by the superior end plate of the
inferior vertebra. Motion was measured parallel to the supe-
rior end plate of the inferior vertebra.
Longitudinal ossification was also evaluated on all
X-rays by board-certified radiologists at Medical Metrics
Inc. For CHARITE´cases, a 6-point scale was setup from
a combined classification system adapted from Nathan
 and McAfee et al. . The grades were defined as fol-
lows: 0: no evidence of osteophyte formation or heterotopic
ossification; 1: osteophytes/heterotopic ossification appear-
ing as isolated points of initial hyperostosis or islands of
bone in soft tissue; 2: bone occurring within the disc space
defined by the planes formed by the two adjacent end
plates, but bony protrusions projecting more or less
horizontally from the vertebral bodies; 3: bone occurring
between the two planes without bridge, 4: apparent contin-
uous connection of bridging bone between end plates; 5: in-
determinate, for example, due to poor film quality. To
determine the clinical relevance of the longitudinal ossifica-
tion, segments rated 3 or 4 were further correlated to over-
all segment motion. Longitudinal ossification was clinically
relevant if it prevented motion. Motion was defined as any
movement greater than a specific cut-off value. Two differ-
ent cut-off values were used: 1) a 5?cut-off value, as per
FDA guidelines ; and 2) a 3?cut-off value, as per
FDA assessment of motion in the ProDisc L summary of
safety and effectiveness .
Data were analyzed using the SAS v8.2 statistical soft-
ware package (SAS Institute, Cary, NC). For categorical
variables, p values were generated using the Fisher exact
test. A t test was used to test means. The primary effective-
ness evaluation for the study was an assessment of the
equivalence of the two treatments in terms of the proportion
of patients in each treatment group classified as a responder.
The success status of patients was summarized by treat-
ment group using counts and percentages. The Blackweld-
er’s test was used to test for treatment equivalence based on
the assumption that a difference of 0.10 was clinically sig-
nificant. Covariate analyses were done to assess the impact
of various factors (age, baseline ODI, gender, operative
level, use of hormone replacement therapy, use of pain
medication at any time, body mass index, and baseline
To further understand the impact of missing data on the
results of this study, two statistical analyses were per-
formed. The first analysis was conducted comparing the
five-year completers with all the patients who were lost
to follow-up. Baseline and two-year clinical outcomes
(VAS and ODI) of the five-year completers were statisti-
cally compared with those of patients lost to follow-up. A
secondary analysis was conducted to determine whether
a difference existed in baseline and two-year clinical out-
come of patients at the nine participating sites vs. the six
nonparticipating sites. A Fisher exact or chi-square test
was used to test categorical variables and a t test was used
to compare univariate summaries.
To confirm overall clinical success, results obtained us-
ing the Blackwelder’s methodology and two additional sta-
tistical analyses were conducted. First, a Last Observation
Carried Forward (LOCF) analysis was done using all avail-
able data, up to the latest follow-up time point for all pa-
tients (including cases of discontinuation, early removals,
and deaths). Secondly, a Bayesian statistical analysis was
performed. For this purpose, a predictive model was built
based on the existing data of all available patients at two-
and five-year follow-up time periods.
378 R.D. Guyer et al. / The Spine Journal 9 (2009) 374–386
Results from patients who crossed over to a different
treatment group were maintained in the intended treatment
category, and all results are therefore presented on an
‘‘intent-to-treat’’ versus ‘‘as-treated’’ basis.
There was no difference between the two groups with re-
spect to operative time, blood loss, or level of implantation
(Table 2). Mean operative time was 108.2 minutes in the
CHARITE´group and 122.1 minutes in the BAK group
(p50.1614). Mean estimated blood loss was 212.1 mL in
the CHARITE´group and 204.3 mL in the BAK group
(p50.8644). As shown with the p values, the operative time
and blood loss were not statistically different between the
Oswestry disability index
The ODI results were plotted as mean values at each
time point for both the CHARITE´and the BAK groups
(Fig. 1). In addition, the percentage of patients with at least
a 15-pt improvement in ODI score vs. baseline was tabu-
lated per group and time point (Fig. 2). The cut-off value
of 15-pt was specified by the FDA and a prior publication
by Ha ¨gg et al., indicated that the minimal clinically impor-
tant difference (MCID) in ODI was10 pts, a value that
should be exceeded for clinical decision-making .
ODI values significantly improved at all postoperative
time points compared with baseline. The MCID for ODI
was reached at three-months postoperatively in both the
CHARITE´and BAK groups. At that time point, the change
in mean ODI scores was ?21.5 pts in the CHARITE´group
and ?17.0 pts in the BAK group. Mean ODI scores re-
mained below 30 pts from one- to five-years postopera-
tively, in both groups. There was no statistical difference
between the groups in terms of ODI scores, at the two-
and five-year postoperative time points. Similarly, the per-
centage of patients with at least 15-pt improvement in ODI
was equivalent at both the two- and five-year time points
VAS pain scores
VAS scores were plotted as mean values at each time
point for both the CHARITE´
(Fig. 3). The MCID value for VAS, as determined by Ha ¨gg
et al., was estimated to be between 18 and 19 from preop-
erative scores .
VAS scores (range: 0–100) were significantly lower at all
postoperative time points compared with baseline. The VAS
and the BAK groups
Level fused or implanted
Total surgery time (min)
Estimated blood loss (cc)
Hospital stay (day)
Fisher exact test used to test categorical variables and t test used to test
Fig. 1. Mean Oswestry scores for both groups and at all time points. Error
bars represent the standard deviation. *Statistically significant difference
from preoperative state within the same treatment group. The dotted line
is placed at approximately 10-pt improvement from the CHARITE´base-
line and represents the MCID. Significant changes were observed as early
as three-months postoperatively for both groups. Improvements remained
constant between the two- and five-year time points.
Fig. 2. Percentage of patients with at least a 15-pt improvement in ODI,
for each group and at all time points. The percentage of patients with a 15-
pt improvement exceeded 50% at six-months postoperatively and remained
above the 50% threshold at both, the two- and five-year time points.
379 R.D. Guyer et al. / The Spine Journal 9 (2009) 374–386
and at six-weeks postsurgery in the BAK group. At that time
point, the difference in mean VAS scores was ?38.6 in the
CHARITE´group and ?29.1 in the BAK group. The average
VAS remained at approximately 30 from 2- to 5-years post-
surgery, in both groups. There was no statistical difference
between the groups in terms of VAS scores, at the two- and
five-year postoperative time points.
Both VAS and ODI reached, and exceeded, MCID
values early on postoperatively and maintained a clinically
significant advantage over baseline throughout the 5-year
follow-up time point.
SF-36 health questionnaires
The results of the SF-36 PCS were plotted as mean
changes from baseline, at each time point for both the
CHARITE´and the BAK groups (Fig. 4). Although most
published spine surgeries provide an average of 8- to
16-pt improvement in SF-36 PCS, the MCID for SF-36
PCS scores has been established at a 5.42-pt improvement
. SF-36 PCS scores gradually improved at all
postoperative time points compared with baseline. In the
CHARITE´group, increases in SF-36 PCS scores reached
14.2 pts at 2-year follow-up and 12.6 pts at 5-year follow-
up. These results were similar to those observed in the
BAK group (increases of 11.2 pts and 12.3 pts at two-
and five-year follow-up, respectively). There was no statis-
tical difference between the groups in terms of SF-36 PCS
scores, at the two- and five-year postoperative time points.
As previously described, the following four criteria for
clinical success are needed to be met to demonstrate indi-
vidual clinical success: improvement$15 pts in ODI vs.
baseline, no device failure requiring additional surgery, ab-
sence of major complications, and maintenance or improve-
ment of neurological status. Table 3 outlines all the device
failures observed throughout the five-year study. All device
failures occurred in the BAK group between the 0 and
2-year time points (seven out of seven cases), whereas
two out of a total of seven CHARITE´device failures
occurred between the 2- and 5-year time points.
Of the seven BAK failures, one case required a bilateral
hemi-laminectomy with anterior discectomy and posterior
fusion due to a non-union. The remaining six cases were
treated with supplemental segmental instrumentation for
posterior fixation due to pseudoarthroses (four out of six
cases), facet joint arthrodesis (one out of six cases), and un-
defined persistent back pain (one out of six cases).
Of the seven CHARITE´failures, one case required a re-
operation without additional internal fixation (ie, right
sided hemi-laminotomy, foraminotomy, and partial discec-
tomy), whereas the remaining six cases were treated with
supplemental segmental instrumentation for posterior fixa-
tion. The cause for these six supplemental fixations was
as follows: symptomatic spondylolisthesis at L5 pars inter-
articularis, observed at six-month follow-up (one out of six
Fig. 3. Mean VAS scores for both groups and at all time points. Error bars
represent the standard deviation. *Statistically significant difference from
preoperative state within the same treatment group. The dotted line is
placed at approximately 19-pt improvement from the CHARITE´baseline
and represents the MCID. VAS improvements exceeded MCID for both
groups at the six-weeks postoperative time point and remained constant
throughout the two- and five-year time points.
Fig. 4. Mean improvements in SF-36 PCS scores for all time points and
both groups. Error bars represent the standard deviation. *Statistically sig-
nificant difference from preoperative state within the same treatment
group.#Statistically significant difference between treatment groups. The
dotted line is placed at a 5.42-pt improvement from the CHARITE´baseline
and represents the MCID. Both groups exceeded 11 pts improvement at
two- and five-years postoperatively. No statistical difference was observed
between groups at any of the time points.
Index-level surgery: device failures
(n543) p Value
0 to 2ay 5 cases (5.5%)
7 cases (16.3%)
6 supplemental fixations
2ato 5 y 2 cases (2.2%)
Total 7 cases (7.7%) 7 cases (16.3%)0.1442
For p value: Fisher exact test was used to test categorical variables.
aTwo-year window defined as !1020 days.
380 R.D. Guyer et al. / The Spine Journal 9 (2009) 374–386
cases); device subsidence and subsequent low back pain
(one out of six cases); facet degeneration (two out of six
cases), and early postoperative implant displacement fol-
lowed by back pain at the 12-month time point (two out
of six cases). Both the cases of facet degeneration were
diagnosed at the two-year follow-up time point.
The ‘‘overall clinical success,’’ a nonvalidated criterion
defined by the FDA, was achieved in 57.8% patients in
the CHARITE´group and 51.2% patients in the BAK group
at the five-year follow-up, compared with 65.2% and 60.6%
of these same cohorts, respectively, at the two-year follow-
up. The five-year overall success was further evaluated us-
ing a Blackwelder’s hypothesis (with D50.10, p50.0359).
This analysis provided a high degree of confidence that the
CHARITE´group was noninferior to the BAK group. The
breakdown for the four separate clinical success factors is
shown in Table 4.
Overall success was also calculated including all the
four deaths and four early device removals as ‘‘failures.’’
Using this methodology, overall success was 54% for
CHARITE´and 50% for BAK (Blackwelder’s test with
Additional surgery for adjacent-level disease occurred in
one CHARITE´ patient (1.1%) and two BAK patients
(4.7%). In addition, nonsurgical treatment (eg, pain man-
agement) for DDD was provided to four CHARITE´patients
(4.4%) and six BAK patients (13.9%).
Work status was assessed before surgery, as shown in
Table 1, and at the two- and five-year follow-ups. The exact
timing when an individual returned to work was not cap-
tured beyond these three time points. Work and disability
status at the 5-year time point are presented in Fig. 5.
Full-time employment was achieved by 65.6% patients in
the CHARITE´group vs. 46.5% patients in the BAK group.
This difference was statistically significant (p50.0403). In
the CHARITE´group, 8.0% patients were on long-term dis-
ability vs. 20.9% patients in the BAK group. This differ-
ence again was statistically significant (p50.0441). No
statistical significance was observed for all other parame-
ters (part-time employment, short-term disability, not em-
ployed, retired, other).
Range of motion
For CHARITE´and BAK patients implanted at L4–L5,
the ROM was evaluated at L4–L5 as well as L3–L4 and
L5–S1. For CHARITE´and BAK patients implanted at
L5–S1, the ROM was evaluated at L5–S1 and L4–L5. All
measurements were analyzed in terms of means and me-
dians for all levels and treatment groups. Statistical analy-
ses provided equivalent findings, where means or medians
were used; therefore, to maintain consistency with prior
publications, all analyses described below are based on
means. Index-level range of motion (ROM) is shown in
The mean preoperative ROM at L4–L5 for all patients
(CHARITE´and BAK) implanted at that level was 8.8?.
There was no statistical difference in preoperative L4–L5
ROM between these CHARITE´
(ROM58.7?for CHARITE´patients and 9.2?for BAK pa-
tients, p50.8162). Similarly, the mean preoperative ROM
at L5–S1 for all patients (CHARITE´and BAK) implanted
and BAK patients
Success rate for efficacy outcome by individual components
(n543) p Value
Fifteen-point improvement in Oswestry
90 (100%) 43 (100%)
Device failures were defined as requiring revision, reoperation, or re-
moval. Major complications were defined as major vessel injury, neurolog-
ical damage, nerve root injury, and death. Neurological deterioration
included slight deterioration, significant deterioration, or mixed response
at 60 months.
aFor p values: Fisher exact test was used to test categorical variables.
bBlackwelder’s test of noninferiority.
Fig. 5. Work and disability status at the five-year postoperative time
point, for both treatment groups. In the CHARITE´group, 65.6% patients
were employed full-time vs. 46.5% in the BAK group. In addition, fewer
CHARITE´patients were on long-term disability (8.0%) compared with
those in the BAK group (20.9%).
381R.D. Guyer et al. / The Spine Journal 9 (2009) 374–386
at the L5–S1 level was 7.8?and was not statistically differ-
ent between the two groups (ROM57.6?for CHARITE´
patients and 8.2?for BAK patients, p50.6793).
For CHARITE´patients implanted at L4–L5, the mean
ROM at the index level did not statistically change from
baseline to 2-years postoperative (p50.3351). Similarly
the mean change in ROM at index level from 2- to 5-years
postoperative was not statistically significant (p50.1294).
However, the mean ROM change at index level from base-
line to 5-years postoperative showed a trend toward statis-
tical significance (p50.0539). For CHARITE´ patients
implanted at L5–S1, none of the mean changes in ROM
were statistically significant (from baseline to two-year
postsurgery: p50.3194; from 2-to 5-year postsurgery:
p50.2065 and from baseline to five-year postsurgery:
p50.2237). Changes in mean index-level ROM for both
L4–L5 and L5–S1 implanted BAK patients were statisti-
cally significant from baseline to either 2- or 5-year fol-
low-ups (for L4–L5 implanted BAK patients, from
baseline to 2-year postsurgery: p50.0076; from baseline
to 5-year postoperative: p50.0003; for L5–S1 implanted
BAK patients, from baseline to 2-year postsurgery and from
baseline to 5-year postsurgery: p#0.0001). However, there
was no statistical difference between the mean index-level
ROM changes from 2- to 5-year postsurgery for either L4–
L5 or L5–S1 (L4–L5: p50.6413; L5–S1: p50.2593).
Adjacent-level range ROM was not statistically different
from baseline to two- or five-year postoperative, for
CHARITE´or BAK patients, regardless of the implanted
level (eg; L4–L5 or L5–S1).
Segmental translation and changes in disc height
No differences were observed between CHARITE´and
BAK patients in terms of changes in disc height. At the
5-year time point, disc height changes decreased by
0.7 mm for both CHARITE´and BAK patients (p50.9827).
Vertebral translation ranged from 0.4 mm to 0.8 mm in
CHARITE´-implanted L4–L5 and L5–S1 sites, respectively.
This translation was within normal range and significantly
lower than the 4.5 mm limit for clinical instability, as sug-
gested by White and Panjabi . The average vertebral
translation was observed at 0.1 mm in BAK-implanted pa-
tients at both L4–L5 and L5–S1 sites.
Clinically relevant longitudinal ossification was defined
as ossification in the disc space that would impact motion.
The definition of motion was based on 2 different sources,
the FDA guidelines (5?) or the ProDisc L SSE (3?).
Within the CHARITE´group and using the 5?cut-off
point to determine motion, 17 (18.9%) cases showed lack
of motion and a rating $3 on the longitudinal classification
system. Nine cases were implanted at L4–L5 (34.6% of all
L4–L5 cases), and 8 were implanted at L5–S1 (12.5% of all
L5–S1 cases). Using the 3?cut-off point to determine mo-
tion, 14 CHARITE´cases (15.5%) met both criteria. Half of
these were at the L4–L5 level (26.9% of all CHARITE´
patients operated at L4–L5) and the other half, at the L5–
S1 level (10.9% of all CHARITE´patients implanted at
Additional statistical analyses
The impact of missing data on study results was evalu-
ated by comparing the clinical outcomes of the 5-year com-
pleters with that of all patients lost to follow-up (LTF).
Baseline VAS and ODI comparisons showed no statistically
significant difference in the BAK group between 5-year
completers and LTF (ODI: p50.5229; VAS: p50.3940).
In the CHARITE´group, both ODI and VAS values were
statistically higher in the LTF group than in the 5-year com-
pleters548.0, p50.0134; VAS: LTF573.7 vs. 5-year
Fig. 6. Index-level ROM for CHARITE´and BAK patients at the two- and five-year follow-up time points. No statistical difference in ROM was observed
between any time point, for either CHARITE´or BAK patients.
382 R.D. Guyer et al. / The Spine Journal 9 (2009) 374–386
completers 5 69.7 p50.0496). However, at the 2-year time
point, there was no statistically significant difference be-
tween 5-year completers and LTF in either CHARITE or
BAK groups, for VAS or ODI values.
The clinical outcomes at baseline and at the 2-year time
points were also compared for all patients included in the 9
participating sites vs. those included in the 6 sites that de-
clined participation. Baseline VAS and ODI comparisons
showed no statistically significant difference in the BAK
group between participating and nonparticipating sites
(ODI: p50.7716; VAS: p50.8316). In the CHARITE´
group, both ODI and VAS values were statistically higher
in the nonparticipating group than in the participating group
(ODI: nonparticipating sites555.5 vs. participating sites5
49.2, p50.0049; VAS: nonparticipating site576.0 vs. par-
ticipating site570.9 p50.0400). However, at the 2-year
time point, there was again no statistically significant dif-
ference between patients from participating and nonpartic-
ipating sites in either CHARITE´or BAK groups, for VAS
The LOCF analysis was completed by using available
results for subjects with 60-month follow-up and the last
available result from the 2-year PMA subjects that did
not have 60-month follow-up results available. A Black-
welder’s test was used to evaluate the noninferiority of
the CHARITE´to the BAK assuming a D of 0.10. This test
provided a p50.0009.
Using Bayesian statistical methodology, the probability
for the CHARITE´group to achieve superiority (D50) vs.
the BAK group was 78.5%, while the probabilities of the
CHARITE´group to be noninferior to the BAK group using
a D of 5%, 10%, and 15% was 92.3%, 98.0%, and 99.7%,
The purpose of this study was to provide 5-year clinical
follow-up data on patients enrolled in a previously reported
2-year IDE randomized clinical study [10,11]. Previous
publications already reported on the safety and efficacy of
the CHARITE´total disc replacement as compared with
BAK with autograft, for the treatment of DDD at L4–L5
or L5–S1. This 5-year study confirms that the hypothesis
validated at the 2-year time point was maintained through-
out the 5-year follow-up period. An example of radio-
graphic views of the implant at L5–S1 at the 5-year time
point is shown in Fig. 7.
Clinical improvements were significant in both the
CHARITE´and BAK groups. The MCID for ODI, VAS,
and SF-36 (PCS) were met early postoperatively and at
the two- and five-year time points, improvements in ODI,
VAS, and SF-36 (PCS) were twice as high as their respec-
tive MCID values.
A subset of the entire IDE patient population was avail-
able at 5 years postsurgery and thus has been included
herein. A significant drop in the eligible patient population
was initially due to site withdrawals from the study. All of
the 14 sites involved in the initial 2-year CHARITE´IDE
trial were invited to participate in the 5-year follow-up;
however, 6 of the 14 sites declined continuation. It is im-
portant to mention that all sites initially signed up for
a 2-year study; they were therefore, under no obligation
to pursue the investigation beyond the 2-year time point.
There was no consistent reason why these sites declined
participation; some sites did not have adequate clinical re-
search support anymore, whereas others were involved in
new clinical studies. The impact of this loss was analyzed
Fig. 7. Radiographic views of a subject implanted at L5–S1, at the 5-year follow-up time point, in (Left) lateral, (Middle) flexion, and (Right) extension.
383 R.D. Guyer et al. / The Spine Journal 9 (2009) 374–386
statistically by comparing the clinical outcomes at baseline
and 2-year postsurgery of participating vs. nonparticipating
sites. This analysis showed that, although patients in the
nonparticipating sites started with a statistically higher pain
and disability rate than patients from the participating sites,
at the 2-year time point, both groups were statistically sim-
ilar. Additional loss to follow-up included 96 patients who
did not respond to requests from their sites, which, via
a third party investigator, attempted contact in multiple
ways. The follow-up rate for this study thus reached 57%
of the randomized eligible population (133/233) or 44%
of the randomized IDE population (133/304).
The sample size in this analysis included 90 patients in
the CHARITE´group and 43 patients in the BAK group.
These numbers are relatively small, a consequence of using
a 2-year study design for a 5-year follow-up analysis. The
initial sample size for the 2-year IDE trial was determined
using the Blackwelder methodology  and was calcu-
lated on the basis of a 10% dropout rate. Although this rate
is appropriate for a 2-year follow-up investigation, it under-
estimates the dropout expected from a 5-year study, as seen
in other comparable long-term studies [24,25]. However,
from a statistical standpoint, although small, the sample
sizes used herein were sufficient to provide confidence of
noninferiority between groups.
To determine whether this 133 patient population was in-
tistical analyses were conducted. When analyzing the
clinical outcomes of 5-year completers (for both CHARITE´
and BAK) vs. patients lost to follow-up (LTF), it was ob-
served that at 2-year postsurgery, these 2 groups were statis-
tically similar, despite the fact that CHARITE´patients in the
LTF group started at a disadvantage with statistically greater
VAS and ODI values. In addition, the 5-year completers
CHARITE´vs. LTF CHARITE´groups were evaluated, at 2-
years, for full-time work and disability status. At that time
point, no statistical difference was observed between these
two patient populations (full-time employment: p50.2448;
hort, at the end of the 5-year time point.
The results of this study provided a high degree of confi-
dence that the CHARITE´group was non-inferior to the
tion carried forward (LOCF) analysis was completed and re-
sulted in a p50.0009, and a Bayesian statistical analyses
provided a 99.5% probability for the CHARITE´group to be-
ing non-inferior to BAK (at a D50.1). These additional sta-
tistical analyses further confirmed noninferiority of the
CHARITE´group as compared with the BAK group.
It is possible that among the lost to follow-up patients,
there are those who have brought claims for personal in-
juries rather than continuing in the study. Due to a lack
of information and other constraints, we are not presently
able to determine the significance of these patient’s out-
comes to the study.
Blumenthal et al. previously discussed the rationale and
scientific validity for selecting a BAK control group. In this
present study, the BAK group had 98% radiographic fusion,
as determined by longitudinal ossification measurements
from independent radiologists. Clinical fusion success was
at 88.4%, as defined by the surgeons during the patients’
5-year follow-up examinations. Radiographic and clinical
fusion rates were different as determined by 2 independent
medical professionals, thus guarding against potential bias.
The high radiographic and clinical fusion rates observed
herein for the BAK cohort alleviates concerns that this
control mayset the bartoolowtotrulyevaluate anovel tech-
nology. Interestingly, fusion and reoperation rates from this
BAK group were not very different from those published
using the LT-Cage with rhBMP-2 in an ALIF procedure. Ex-
cludingreoperations, the rhBMP-2 study resultedin a94.5%
graphic fusion rate) observed with BAK. In terms of reoper-
288) at 2 years, compared with 16.3% failure rate (6 out of
43) at 5 years, for BAK [26,27]. Thus, the results from our
BAK group were not very different from some of the best
results published to date for anterior lumbar fusion.
When analyzing the 5-year CHARITE´and BAK groups,
establishing baseline similarities was essential to confirm
adequate randomization. Demographic baseline data was
therefore analyzed and found to be statistically equivalent,
thus confirming an effective randomization process with no
evidence of selection bias between CHARITE´and BAK
treatments. Although age, weight, activity level, and work
status were similar across groups, there was a slight trend
ofincreased incidence of priorlumbar surgery in the CHAR-
ITE´group. Prior lumbar surgery was defined as minor inter-
ventions such as discectomies or laminectomies. Although
there is very little evidence on the impact of prior laminecto-
mies and/or discectomies on fusion outcomes, a recently
published report presented high fusion rate in patients with
group and any clinical disadvantage resulting from prior
surgery would have accrued to the CHARITE´results.
The clinical impact of performing either a total disc re-
placement with the CHARITE´artificial disc or an ALIF
with a BAK cage and iliac crest autograft in this patient
population was demonstrated by significant improvements
in pain and disability scores. Only patients with debilitating
back pain were included in this study. Median preoperative
ODI values reached 48.0 pts in the CHARITE´group and
52.0 pts in the BAK group (no statistical difference). Sim-
ilarly, median VAS for both CHARITE´and BAK groups
were exactly 70.0. Comparatively, burst fractures and met-
astatic bone tumors in the spine were both previously given
ODI values of approximately 60 . An additional com-
parative figure was provided in the review paper by
384R.D. Guyer et al. / The Spine Journal 9 (2009) 374–386
Fairbank et al.  in which ODI scores were categorized
by disease: chronic low back pain had the third worse ODI
score with 43.3 pts and was only 4.74 pts lower than bone
metastasis. These severely disabled patients had all under-
gone at least 6 months conservative care to meet the inclu-
sion criteria of the study. On average, conservative care was
provided to the CHARITE´patients for 34.2 months (me-
dian: 24.0 months) and to the BAK patients for 26.3 months
(median: 21.0 months). Opportunities for nonoperative care
for these patients had therefore been exhausted.
In both the CHARITE´and the BAK groups, ODI and
VAS improvements exceeded MCID thresholds by the
three-month time point and reached nearly twice the MCID
threshold by two and five years. These results showed that,
in both groups, pain and functional disability was improved
for most of the patients at early time points and remained
favorable throughout the 5-year follow-up.
Additional surgery, whether for index-level or at adja-
cent-level degenerative disc disease, were relatively low
for both CHARITE´and BAK groups compared with prior
reports with BAK [31,32]. Degenerative disc disease, de-
fined as requiring secondary surgery, affected 1.1% CHAR-
ITE´patients and 4.7% BAK patients, these numbers being
too low to draw statistical conclusions.
Work statuses at two- and five-year follow-ups were also
evaluated. A statistically greater number of patients were
employed full-time in the CHARITE´group as compared
with the BAK group. In addition, a statistically smaller
number of patients in the CHARITE´group were on long-
term disability, as compared with BAK patients.
ROM at index and adjacent levels did not change statis-
tically from the 2- to the 5-year time points, for either
CHARITE´or BAK patients. Additional analyses were con-
ducted to evaluate any relationship between index-level
ROM and clinical outcome (VAS and ODI), but in both
cases, no correlation was found. In addition, at the 5-year
time point, segmental translation and disc height were
equivalent between both groups. Segmental translation
was ranging from 0.4 and 0.8 mm in CHARITE´-implanted
patients. To place this result into context, the following
published data on segmental translation need to be consid-
ered. Pearcy and Posner et al., in two separate publications,
described a normal range of segmental translation from
2.0 to 2.8 mm [33,34]. Panjabi further defined clinical in-
stability as a segmental translation greater than 4.5 mm
. These values are significantly greater than the seg-
mental translation observed in CHARITE´cases, thus sug-
gesting normal segmental translation for those patients.
Results of this five-year, prospective, randomized multi-
center study are consistent with the two-year reports of non-
inferiority of CHARITE´artificial disc vs. ALIF with BAK
study represents the largest and longest randomized, pro-
spective, multicenter arthroplasty trial performed to date.
We thank Dr. Chantal E. Holy and Dan Stoutenbourgh,
employees of DePuy Spine, for editorial support.
 Link HD. Letter to the Editor concerning ‘‘Charite total disc replace-
ment: clinical and radiographical results after an average follow-up of
17 years’’ (M. Putzier et al.). [comment]. Eur Spine J 2006;15:514–7.
 Buttner-Janz K. Letter to the Editor concerning ‘‘Charite total disc re-
placement: clinical and radiographical results after an average fol-
low-up of 17 years’’ (M. Putzier et al.). Eur Spine J 2006;15:510–3.
 Cunningham BW, Gordon JD, Dmitriev AE, Hu N, McAfee PC.
Biomechanical evaluation of total disc replacement arthroplasty: an
in vitro human cadaveric model. Spine 2003;28:S110–7.
 Cunningham BW, Dmitriev AE, Hu N, McAfee PC. General princi-
ples of total disc replacement arthroplasty: seventeen cases in a non-
human primate model. Spine 2003;28:S118–24.
 Griffith SL, Shelokov AP, Buttner-Janz K, Lemaire JP, Zeegers WS.
A multicenter retrospective study of the clinical results of the LINK
SB Charite intervertebral prosthesis. The initial European experience.
 Cinotti G, David T, Postacchini F. Results of disc prosthesis after
a minimum follow-up period of 2 years. Spine 1996;21:995–1000.
 Lemaire JP, Carrier H, SariAli e, Skalli W, Lavaste F. Clinical and
radiological outcomes with the Charite artificial disc: a 10-year min-
imum follow-up. [erratum appears in J Spinal Disord Tech. 2006
Feb;19(1):76 Note: Sari Ali, El-Hadi (corrected to Sariali, El-hadi)].
J Spinal Disord Tech 2005;18:353–9.
 Putzier M, Funk JF, Schneider SV, et al. Charite total disc replace-
ment–clinical and radiographical results after an average follow-up
of 17 years. [see comment]. Eur Spine J 2006;15:183–95.
 David T. Long-term results of one-level lumbar arthroplasty: mini-
mum 10-year follow-up of the CHARITE artificial disc in 106 pa-
tients. Spine 2007;32:661–6.
 Blumenthal S, McAfee PC, Guyer RD, et al. A prospective, random-
ized, multicenter Food and Drug Administration investigational de-
vice exemptions study of lumbar total disc replacement with the
CHARITE artificial disc versus lumbar fusion: part I: evaluation of
clinical outcomes. [erratum appears in Spine. 2005 Oct 15;30(20):
2356]. Spine 2005;30:1565–75.
 McAfee PC, Cunningham B, Holsapple G, et al. A prospective, ran-
domized, multicenter Food and Drug Administration investigational
device exemption study of lumbar total disc replacement with the
CHARITE artificial disc versus lumbar fusion: part II: evaluation
of radiographic outcomes and correlation of surgical technique accu-
racy with clinical outcomes. Spine 2005;30:1576–83.
 Brazier JE, Harper R, Jones NM, et al. Validating the SF-36 health
survey questionnaire: new outcome measure for primary care. BMJ
 Geisler FH. Surgical technique of lumbar artificial disc replacement
with the Charite artificial disc. Neurosurgery 2005;56(Suppl. 57).
 Kuslich SD, Ulstrom CL, Griffith SL, Ahern JW, Dowdle JD. The
Bagby and Kuslich method of lumbar interbody fusion. History, tech-
niques, and 2-year follow-up results of a United States prospective,
multicenter trial. Spine 1998;23:1267–78.
 Zhao K, Yang C, Zhao C, An KN. Assessment of non-invasive inter-
vertebral motion measurements in the lumbar spine. J Biomech
 Nathan H. Osteophytes of the vertebral column: an anatomical study
of their development according to age, race, and sex with
385 R.D. Guyer et al. / The Spine Journal 9 (2009) 374–386
considerations as to their etiology and significance. J Bone Joint Surg
 McAfee PC, Fedder IL, Saiedy S, Shucosky EM, Cunningham BW.
SB Charite disc replacement: report of 60 prospective randomized
cases in a US center. J Spinal Disord Tech 2003;16:424–33.
 Food and Drug Administration. Guidance document for the prepara-
tion of IDEs for spinal systems. 2000.
 Food and Drug Administration. PRODISC?-L total disc replace-
mentdP050010: summary of safety and effectiveness data; 2006.
 Hagg O, Fritzell P, Nordwall A. Swedish Lumbar Spine Study Group.
The clinical importance of changes in outcome scores after treatment
for chroniclow backpain.[see comment]. EurSpineJ 2003;12:12–20.
 Polly DW Jr, Glassman SD, Schwender JD, et al. SF-36 PCS benefit-
cost ratio of lumbar fusion comparison to other surgical interven-
tions: a thought experiment. Spine 2007;32(Suppl 6).
 White AW, Panjabi MM. Clinical biomechanics of the spine. 2nd ed.
Philadelphia, PA: Lippincott Williams & Wilkins, 1990.
 Blackwelder WC, Chang MA. Sample size graphs for ‘‘proving the
null hypothesis. Control Clin Trials 1984;5:97–105.
 Glaser J, Stanley M, Sayre H, Woody J, Found E, Spratt K. A 10-year
follow-up evaluation of lumbar spine fusion with pedicle screw fixa-
tion. Spine 2003;28:1390–5.
 Kuslich SD, Danielson G, Dowdle JD, et al. Four-year follow-up re-
sults of lumbar spine arthrodesis using the Bagby and Kuslich lumbar
fusion cage. Spine 2000;25:2656–62.
 Burkus JK, Gornet MF, Dickman CA, Zdeblick TA. Anterior lumbar
interbody fusion using rhBMP-2 with tapered interbody cages. J Spi-
nal Disord Tech 2002;15:337–49.
 Food and Drug Administration. INFUSE? bone graft / LT-CAGE?
lumbar tapered fusion device. Summary of effectiveness and data
 Badawy WS, El Masry MA, Radwan YA, El Haddidi TT. Results of
instrumented posterolateral fusion in failed back surgery. Int Orthop
 Wood K, Buttermann G, Mehbod A, et al. Operative compared with
nonoperative treatment of a thoracolumbar burst fracture without
neurological deficit. A prospective, randomized study. [see comment;
erratum appears in J Bone Joint Surg Am. 2004 Jun;86-A(6):1283]. J
Bone Joint Surg Am 2003;85-A:773–81.
 Fairbank JC, Pynsent PB. The Oswestry disability index. Spine
 Button G, Gupta M, Barrett C, Cammack P, Benson D. Three- to six-
year follow-up of stand-alone BAK cages implanted by a single sur-
geon. Spine J 2005;5:155–60.
 Pavlov PW, Spruit M, Havinga M, Anderson PG, van LJ, Jacobs WC.
Anterior lumbar interbody fusion with threaded fusion cages and au-
tologous bone grafts. Eur Spine J 2000;9:224–9.
 Pearcy MJ. Stereo radiography of lumbar spine motion. Acta Orthop
 Posner I, White AA III, Edwards WT, Hayes WC. A biomechanical
analysis of the clinical stability of the lumbar and lumbosacral spine.
 Panjabi M, Malcolmson G, Teng E, Tominaga Y, Henderson G,
Serhan H. Hybrid testing of lumbar CHARITE discs versus fusions.
386 R.D. Guyer et al. / The Spine Journal 9 (2009) 374–386