Few studies have investigated whether there are differences between the
results of surgical procedures that are performed with the participation of
residents or fellows and those that are performed in a private practice
surgical
setting1-4.
Studies of laparoscopic
cholecystectomy2 and
colectomy4 have
demonstrated substantially increased surgical times for both procedures as
well as higher complication rates for cholecystectomy when performed by
residents. Studies of total knee
arthroplasty3 and
anterior cruciate ligament
reconstruction1 have
shown longer surgical times and higher costs for procedures performed on a
teaching service.
We hypothesized that there would be a difference in clinical outcomes when
total hip arthroplasty and total knee arthroplasty were performed at a
teaching service.
We performed a retrospective study on a consecutive series of 347 patients
who underwent 401 primary total hip or total knee arthroplasties between
February 1996 and October 1999 after institutional review board approval was
obtained. Two hundred and eleven patients underwent 230 total hip
arthroplasties, and 136 patients underwent 171 total knee arthroplasties. One
private practice surgeon (S.T.W.) who was also a clinical professor at our
institution performed all of the procedures. All patients were managed at one
hospital at which these arthroplasties are performed by surgeons with the
assistance of residents or fellows (teaching service) or without the
assistance of these trainees (private practice service). A power calculation
esti mated that 100 patients in each series of hip and knee replacement
procedures would be necessary to show a 10% difference in the surgical time
with a power of 0.8.
The teaching service group included 169 patients undergoing 191 procedures
(119 total hip arthroplasties and seventy-two total knee arthroplasties) who
were managed between February 1996 and October 1997. The attending surgeon had
one surgical assistant during each procedure who was either a resident or a
joint replacement fellow. The residents who assisted were in the fourth year
of training and had not participated in joint replacement surgery in the role
of the primary surgeon before this rotation; thus, their total joint surgical
skills were in the early stages of development. A resident was the first
assistant to the attending surgeon during two-thirds of these procedures, and
a fellow was the first assistant during the remaining one-third. Either the
resident or the fellow performed between one-third and one-half of the
procedure under the direct supervision of the attending surgeon in accordance
with the surgical skill level of each resident or fellow. Typically, for a
total hip arthroplasty, after either the resident or fellow had seen the
attending surgeon perform one or two operations and understood his technique,
the surgical approach and preparation and the insertion of either the
acetabular or femoral component would be done by the resident and the other
component would be inserted by the attending surgeon. For a total knee
arthroplasty, the resident or fellow and the attending surgeon would perform
one-half of the procedure, including the surgical approach, the bone cuts, and
component insertion. The resident or fellow was positioned on the same side of
the operating room table as the attending surgeon during all hip procedures
and would switch to the appropriate side when acting as the surgeon during a
total knee arthroplasty. The attending surgeon was present for the entire
procedure, including during patient positioning and draping, the approach to
the joint, and wound closure.
The remaining 178 patients in the private practice group underwent 111
total hip arthroplasties and ninety-nine total knee arthroplasties after
October 1997. All procedures were performed by the same attending surgeon. A
physician's assistant was the sole surgical assistant in all of those
procedures, and his participation was limited to retraction and cutting
sutures.
Two types of cementless femoral components (AML or Replica; DePuy, Warsaw,
Indiana) and one type of cemented femoral component (Endurance; DePuy) were
used for the hip replacement procedures. In the teaching service group, the
diameter of the femoral head component was 22 mm for one hip, 26 mm for four
hips, and 28 mm for 114 hips. In the private practice group, the diameter of
the femoral head was 22 mm for one hip, 26 mm for two hips, 28 mm for
ninety-one hips, and 32 mm for seventeen hips. One design of cementless
acetabular component (Duraloc; DePuy) was used for 229 of the 230 procedures.
One hundred and one hips (ninety-eight in the teaching service group and three
in the private practice group) received an acetabular component with a
porous-coated metallic shell that was press-fit with use of augmentation with
one or two screws. The remaining 129 hips (twenty-one in the teaching service
group and 108 in the private practice group) received an acetabular metallic
shell (either a shell with no screw-holes or a trispiked shell) that was
press-fit without screw fixation. Ninety-seven percent (131) of the 135
tricompartmental knee arthroplasties were performed with one of two types of
posterior cruciate-substituting knee implants (LCS rotating platform [DePuy]
or NexGen [Zimmer, Warsaw, Indiana]), and 94% (thirty-four) of the thirty-six
unicompartmental replacement procedures were performed with one implant design
(Miller-Galante; Zimmer).
Demographic and hospital data were collected prospectively for all
patients. There were no significant differences between the two groups with
regard to demographic characteristics, the mode of fixation, or preoperative
diagnosis (see Appendix). Among the patients undergoing total knee
arthroplasty, the only significant difference between the teaching service and
private practicegroups was the male:female ratio (see Appendix). The hospital
data that were studied included blood loss (for hip procedures only), blood
replacement, operative time, length of stay, and complications. The operative
time was defined as the time from the initiation of the skin incision to the
completion of wound closure. A tourniquet was used and released after wound
closure during all total knee arthroplasties. All patients undergoing hip
replacement were encouraged to predonate one unit of blood and had
postoperative red blood-cell salvage. Patients undergoing knee replacement had
either preoperative autologous donation of one unit or postoperative unwashed
red blood-cell salvage. The clinical outcome for each group of patients
undergoing hip replacement was determined at the time of the last follow-up
with use of the Harris hip
score5. For the
patients undergoing hip replacement, cup and stem positioning was assessed on
an anteroposterior radiograph of the pelvis centered over the symphysis pubis;
this information was available for 83% (ninety-nine) of 119 hips in the
teaching service group and 86% (ninety-six) of 111 hips in the private
practice group. An independent observer (M.N.K.), who was blinded to each
patient's group, performed the analysis. A previously reported system for the
classification of femoral component fill at the distal part of the
femur6 was used
because the cementless femoral implant was designed for implantation by means
of distal canal contact. Because of the complexity of assessment of knee
radiographs, none of those radiographs were analyzed and instead the Knee
Society rating
score7 and the
postoperative clinically measured values for knee alignment, active flexion,
and flexion arc were used to assess the clinical outcome for patients managed
with knee replacement.
All intraoperative and postoperative complications were recorded; however,
we did not analyze differences in medical complications such as confusion,
ileus, urinary complications, and deep venous thrombosis. The average duration
of follow-up for the teaching service and private practice groups was
fifty-nine months (range, six to 112 months) and thirty-nine months (range,
six to ninety-one months), respectively. All late complications and revisions
were recorded.
Statistical analysis was done by means of a chi-square contingency test for
nominal values and a two-tailed t test for continuous parameters. A p value of
=0.05 was considered significant.
This retrospective sequential study was performed on a consecutive series
of patients who were managed by the same surgeon at one institution. The
surgeon was present for the entire duration of all procedures. On the teaching
service, the attending surgeon consistently allowed the trainees to perform
substantial portions (between one-third and one-half) of each procedure in
accordance with each trainee's surgical abilities. As the patient's welfare
was the highest priority, residents or fellows who had good operating skills
were allowed to perform more of each procedure than trainees with lesser
skills.
As has been shown in other
reports1-4,
there was a highly significant decrease in the average operative time when
procedures were performed without resident participation. However, those
reports included procedures that were performed by multiple attending surgeons
and only one of those reports involved patients undergoing total joint
procedures. The present study eliminates the variables of different attending
surgeons and differing surgical techniques. In the present study, patients in
both the teaching service group and the private practice group were managed
with the same surgical approach and there was no variance in the postoperative
protocols. However, other than the longer operative time for patients in the
teaching service group, there were no other differences in the early clinical
parameters, including blood loss or replacement and hospital stay.
There was one factor besides surgical training, however, that could have
increased the surgical time for the patients who underwent hip replacement in
the teaching service group. The attending surgeon changed his technique of
acetabular shell fixation (and shell design) at the beginning of the (later)
series of procedures in the private practice group. A large majority (82%) of
the hip procedures in the teaching service group were performed with use of a
metallic acetabular component shell with multiple holes (Duraloc 1200 series;
DePuy) that was fixed with one or two screws, with the remaining cups being
inserted without screws. Ninety-seven percent of the procedures in the private
practice service were performed with use of either a shell with no screw-holes
(Duraloc 100 series; 9%) or a shell with no screw-holes but with three
metallic spikes (Duraloc 300 series; 88%), both of which were press-fit. All
of the shells that did not have a multiple-hole design for screw fixation,
however, had insertion of a screw-in central hole eliminator plug that
occluded the hole provided for the cup positioner. Because the placement of
one or two acetabular screws requires several minutes of operative time
(normally a longer time than is required to insert the central hole eliminator
plug), some of the difference in surgical time between the two hip surgery
groups may have been related to these different surgical techniques rather
than to surgical training. However, since the same techniques were used for
patients undergoing knee replacement in both the teaching service group and
the private practice group, we believe that the difference in surgical time
for the two knee surgery groups was the result of resident or fellow
participation.
While differences in surgical time would, of course, be expected to vary
according to the level of participation by the trainee, the increased time of
19% for total hip arthroplasty and of 9% for total knee arthroplasty, with the
attending surgeon being present for the entire procedure, is notable. We
speculate that operative times might be substantially longer in academic joint
replacement training programs where residents and fellows perform these
procedures with less direct faculty participation. A study of procedures that
were performed with less attending supervision might have very different
results than were reported here.
The radiographic analysis of the hip procedures demonstrated that there was
no difference in femoral component positioning or distal femoral canal fill.
Although the average abduction angle of the acetabular component was
significantly lower in the private practice group than in the teaching service
group, and although the private practice group also demonstrated a
significantly lower number of components that were outliers (with an abduction
angle <30° or >50°) when compared with the teaching service
group, we are not sure whether these differences were due to teaching or to
the use of different component designs by the two groups (metallic acetabular
shells with or without supplemental screw fixation). Because the decisions
regarding the orientation of the acetabular components that were inserted by
the trainees were made primarily by the attending surgeon, and because the
resident was on the same side of the operating room table as the surgeon, we
doubt that the significantly higher percentage of outliers with abduction
angles in excess of 50° was due to surgical instruction. However, because
we do not have data indicating which acetabular component was inserted by
either the attending surgeon or the trainee, we cannot determine the exact
cause of these outliers. The use of a tri-spiked cup rather than a
multiple-holed cup may have caused this difference. Although there were a
significantly higher percentage of acetabular components that were abducted
>50° in the teaching service group, none of the patients who had a more
abducted cup sustained a postoperative dislocation.
Although a radiographic analysis of the knee procedures was not done, we
would not expect such a study to reveal any major differences between the
groups as the postoperative clinical parameters of knee scores, alignment, and
range of motion were similar.
The hospital stays of the groups of patients managed with hip and knee
replacement were not different because the attending surgeon wrote all of the
orders during the hospital stay. Because there was a longer operative time in
the teaching service groups, the costs of hospitalization would undoubtedly
have been slightly higher for these patients; however, we did not attempt to
perform a cost comparison between these groups.
Interestingly, the rate of early complications following hip replacement
was higher in the private practice group than in the teaching service group,
although this difference was not significant. Both of the patients in the
private practice group who required immediate reoperation because of component
malpositioning had had a complex primary hip replacement procedure because of
previous hip surgery, and both patients had a severe hip deformity. The
complexity of these two arthroplasties undoubtedly contributed to the
development of the two surgical errors, rather than the absence of resident or
fellow participation. The dislocation rate (both early and late) in the
private practice group may have been lower because 15% of these procedures
were performed with the use of a 32-mm head, whereas all of the hips in the
teaching group received a head that was 28 mm in diameter.
Although the present study was retrospective, it represents a large
consecutive series of similar patients who underwent procedures performed by
one surgeon who used a limited variety of implants with the same anesthetic
and postoperative protocols. All of the data were collected prospectively, and
the radiographs were analyzed in a blinded fashion. The weaknesses of the
study included differences between the two groups with regard to the types of
acetabular fixation and the fact that radiographs were not available for about
15% of the hips in both groups. Because the study was not designed
prospectively, it was not possible to determine from the operative report
which portion of each procedure was performed by a trainee and, therefore, we
were unable to determine an exact correlation between outliers, complications,
and the operating surgeon.
The results of the present study should serve to assuage the concerns of
patients who are to have total joint procedures performed by surgeons with the
assistance and participation of residents or fellows. Although the operative
times were significantly different, they only differed by between 9% and 19%.
Most importantly, the rates of complications were no higher when the teaching
group was compared with the private practice group, and the early clinical
outcomes were equivalent.
Tables showing demographic data for the two study groups are available with
the electronic versions of this article, on our web site at
(go to
the article citation and click on "Supplementary Material") and on
our quarterly CD-ROM (call our subscription department, at 781-449-9780, to
order the CD-ROM). ?