360 Future Directions
do not end up in one group more than the
other. For example, dogs should not be allo-
cated to the control nonsurgical group because
they are poor candidates for anesthesia, which
would lead the control group to be fundamen-
tally ‘sicker’ than the surgical group. While
RCTs are the ideal, in instances when random-
ization is not possible, carefully designed prag-
matic trials can provide quality data to inform
medical practice (Patsopoulos 2011).
Blinding and outcome assessment
Observation bias occurs when knowledge of
a study subject’s group allocation influences
the identification of relevant events during the
study. If investigators know which animals
received an active intervention, they may mon-
itor that group more closely in a way that could
seriously affect the trial outcome. If an owner
knows their dog is receiving an active interven-
tion, they may be likely to over report improve-
ment, whereas those who know their dog is in
the control group may be likely to over report
no improvement or deterioration. The blinding
of investigators and owners to which group ani-
mals are allocated prevents these biases.
The potential for observation bias is related
to the subjectivity of the outcomes used. For
example, if an outcome is change in a biomarker
for cartilage breakdown, such as cartilage
oligomeric matrix protein (COMP), observation
bias is unlikely because COMP measurement
cannot be affected by knowledge of the treat-
ment group. In contrast, when the outcome is
an owner assessment of pain or a veterinar-
ian assessment of lameness, knowledge of the
animal’s group allocation could affect the score
attributed to that animal. To obtain unbiased
estimates of intervention effect, regardless of
whether the outcome measured is subjective or
objective, the effect documented in the control
group is subtracted from the effect documented
in the intervention group.
Placebo effects
Placebo effects are improvements documented
in a control group with no active interven-
tion. Commonly, the assumption is that animals
are not really improved, but merely a result
of bias of the owners or investigators. How-
ever, animals in a control group can actually
improve, even with no active intervention due
to regression-to-the-mean. Many chronic dis-
eases have waxing and waning signs. Owners
seek out a trial when those signs peak. Over
time, even without intervention, their animals
will cycle back to their average level of clinical
sign burden or disability. This is true in dogs
with cruciate ligament rupture, where animals
can present with acute exacerbation of their
disease that may be manifested by pain and
lameness which will improve to some degree
without any specific intervention. While it
is impossible to completely remove placebo
effects from a study, regardless of whether the
outcome measure is subjective or objective,
there are many ways to try to minimize those
effects and increase the assay sensitivity (iden-
tifiable difference between the treatment and
control groups) in the design of a clinical trial
(Table 43.2).Summary
Patient-oriented research has come to veteri-
nary medicine. Today, veterinarians are partner-
ing with owners to determine the health priori-
ties for their pets in order to improve outcomes.
Ten years from now, should theWall Street Jour-
nalwant to follow-up on their story about the
owners’ perspective of cruciate surgery in their
dogs, hopefully every statistic about cost will be
countered with an even more compelling statis-
tic about how that surgery improved the health
and well-being of their pet.References
Bannuru RR, McAlindon TE, Sullivan MC,et al.Effec-
tiveness and implications of alternative placebo
treatments: Asystematic review and network meta-
analysis of osteoarthritis trials. Ann Intern Med
2015;163:365–372.
Barnett AG, van der Pols JC, Dobson AJ. Regression
to the mean: what it is and how to deal with it. Int
J Epidemiol 2005;34:215–220.
Brown DC. The Canine Orthopaedic Index. Step 1:
Devising the items. Vet Surg 2014a;43: 232–240.
Brown DC. The Canine Orthopaedic Index. Step 2:
Psychometric testing. Vet Surg 2014b;43:241–246.