To The Editor:
We read with interest the excellent article "Anterior Cruciate
Ligament Reconstruction: Bone-Patellar Tendon-Bone Compared with Double
Semitendinosus and Gracilis Tendon Grafts. A Prospective, Randomized Clinical
Trial" (2004;86:2143-55),
by Aglietti et al. However, we feel that it cannot be described as a
"randomized" trial. Randomization was achieved by a
"strictly alternating" pattern of treatment allocation, which is
open to sources of potential bias. Despite this method of allocation, the two
groups are very well matched, although there is no discussion in the article
as to whether a stratification process was used to achieve this.
The term "randomized" has a precise statistical meaning in that
each patient has the same probability of receiving either treatment,
independent of the previous patient's allocation1. This is clearly
not the case here. In addition, the term "strictly" is difficult
to interpret. Although the procedure performed at each successive operation
may have alternated, we feel that further information should be supplied
concerning the processes of entering patients into the trial, in particular,
the surgeon's role in influencing the treatment received and whether
allocation was by a third party. In addition, there is no information about
the patients who declined entry into the study or withdrew once they knew
which treatment arm they had been allocated to, again both potential sources
of bias. Using alternation to allocate treatment is thus better described as
"deterministic" and the term "randomized" should not
be used.
We note that a previous study by the same authors2 used the same
method of treatment allocation; however, this was not described as
"randomized" but more appropriately as a "prospective
comparative study." Interestingly, subsequent papers referring to this
study have inadvertently described it as randomized3,4.
Since the CONSORT guidelines5 were published, there has been an
increased recognition of the importance of methodology, and its justification
in research, and there has been a greater emphasis on the reporting of
randomized controlled trials. Unfortunately, despite being an otherwise
excellent study, this work has left itself open to criticisms of the
allocation method used, and thus the potential for bias.
I would like to thank Dr. Fox and Mr. Johnson for their question, which
allows me to explain our randomization method. In our study, the patients were
selected in the following way:
Each patient on a waiting list at our clinic, which is in a public
hospital, was called by a resident. The waiting list includes more than 1300
patients who are waiting for different types of surgeries by nine different
orthopaedic surgeons. The patient's registration number determined the graft
choice: odd numbers were allotted to receive a bone-patellar tendon-bone
graft, while even numbers underwent hamstring reconstruction.
The patients were investigated preoperatively by the same independent
observer (F.G.) who evaluated them at the time of follow-up.
Each patient was informed about the study and signed an informed consent
form. They were aware that the only difference in the procedure was the type
of graft employed. They did not know which graft would be used for their
reconstruction, but they were informed about our percentage of success with
each graft based on previous studies.
Patients who refused to take part in the study or wanted to personally
choose the graft were not included in the study.
The surgeon (P. Aglietti) performed the surgery and selected the type of
graft on the basis of the patient's registration number on the waiting list.
He met the patients in the operating room and was not able to influence the
selection. The procedures were performed alternately (odd and even numbers
sequentially) up to 120.
There are several ways to allocate patients in clinical studies, but I
think that this method of allocating the patients (registration number in a
general waiting list including other patients undergoing different types of
surgeries) could be reasonably considered "randomized" and not
"deterministic" because, at the time of registration in the
waiting list, each patient had the same probability of receiving an odd or an
even number.
Altman DG. Practical statistics
for medical research. 1st ed. Chapter 5, p
81-90. New York: Chapman and Hall;
1991.81
1991
Aglietti P, Buzzi R, Zaccherotti G, De
Biase P. Patellar tendon versus doubled semitendinosus and gracilis tendons
for anterior cruciate ligament reconstruction. Am J Sports Med.1994;22:
211-8.22211
1994
[PubMed][CrossRef]
Eriksson K, Anderberg P, Hamberg P,
Lofgren AC, Bredenberg M, Westman I, Wredmark T. A comparison of quadruple
semitendinosus and patellar tendon grafts in reconstruction of the anterior
cruciate ligament. J Bone Joint Surg Br.2001;83:
348-54.83348
2001
[PubMed][CrossRef]
Shaieb MD, Kan DM, Chang SK, Marumoto
JM, Richardson AB. A prospective randomized comparison of patellar tendon
versus semitendinosus and gracilis tendon autografts for anterior cruciate
ligament reconstruction. Am J Sports Med.2002;30:
214-20.30214
2002
[PubMed]
Begg C, Cho M, Eastwood S, Horton R,
Moher D, Olkin I, Pitkin R, Rennie D, Schulz KF, Simel D, Stroup DF. Improving
the quality of reporting of randomized controlled trials. The CONSORT
statement. JAMA.1996;276:
637-9.276637
1996
[PubMed][CrossRef]