To The Editor:
Regarding the article "Survival Analysis of Hips Treated with
Core Decompression or Vascularized Fibular Grafting Because of Avascular
Necrosis" (80-A: 1270-1275, Sept. 1998), by Scully et al., I am
not certain that, from the data presented, firm conclusions can
be made when comparing these two treatment modalities. The main
reasons for this are the significant age differences of patients
in the Ficat stage-II and stage-III groups and the end point the
authors have chosen for failure - conversion to total hip arthroplasty.
No clinical data are presented in terms of pain scores or functional
evaluation of patients, and, therefore, one cannot say that the
threshold for total hip arthroplasty is the same in the two groups
of patients. Indeed, the threshold for performing a total hip arthroplasty
in the older patient is likely to be much lower. The paper gives no
indication that consistent criteria were applied to determine when
a patient should undergo arthroplasty, and this determination can,
of course, completely change the outcome of the study.
To apply a Cox proportional-hazards model to demonstrate the
absence of age bias when the structure of the study is flawed in
this way is misleading.
J. D. Witt, F.R.C.S., F.R.C.S.(Orth)
Department of Orthopaedic Surgery
The Middlesex Hospital
Mortimer Street
London W1N 8AA, United Kingdom
S. P. Scully replies:
Mr. Witt raises the concern that, because of the age differences
between the core decompression and vascularized fibular graft populations,
a bias may be introduced when conversion to total hip arthroplasty
is used as an end point. He also raises the concern that younger
patients may be less likely to undergo the conversion operation. When
we recognized that there was a statistical difference in ages between
the two populations, we were aware of the bias that this may introduce.
Clinical criteria were used on an individual basis in both study
populations to determine when arthroplasty was to be chosen as an
intervention. While some bias may have been introduced during the
clinical decision-making process, there was no means of eliminating
this, as ours was a retrospective study. The Cox proportional-hazards
model was one means that we used to isolate the influence of the
age variable on the outcomes in the two populations.
In the published Kaplan-Meier plots, the time for 50 percent
of the failures to occur in stage-III disease was approximately
twenty-one months for the vascularized fibular graft group compared
with fifteen months for the core decompression group. In a subsequent
publication, a similar number was reported for an unrelated group
of patients treated with core decompression1.
That would suggest that bias may be introduced either because of
age or because of the emotional investment in the larger procedure,
as has already been discussed2-4.
A prospective, randomized study would eliminate these biases, but,
in the absence of such data, our study serves as an initial attempt
to evaluate the efficacy of the vascularized fibular grafting procedure
by comparing it with an accepted intervention for this disease process.
Sean P. Scully, M.D., Ph.D.
Division of Orthopaedic Surgery
Musculoskeletal Oncology Section
Duke University Medical Center, Box 3312
Durham, North Carolina 27710
E-mail address: scull002@mc.duke.edu