Downloaded from http://bjsm.bmj.com/ on February 15, 2018 - Published by group.bmj.

com
BJSM Online First, published on February 15, 2018 as 10.1136/bjsports-2017-098978
Editorial

Different interventions, same (eg, did the recipients of the interven-

tion use it as intended within their daily
lives).9 A paper investigating treatment
outcomes? Here are four good reasons fidelity of interventions for osteoarthritis
and chronic low back pain found that
Chad E Cook, Steven Z George, Francis Keefe many interventions were either poorly
performed, poorly reported or both.10
Comparing two treatments, each of
Introduction Examples of variant findings among which with low treatment fidelity, is likely
Conundrums in musculoskeletal research outcomes are present in the literature. to bias a study towards a null finding.
are common. Halliday and colleagues1 One month after total hip arthroplasty, Poor treatment fidelity occurs most
demonstrated similar outcomes for Dayton and colleagues4 report that perfor- commonly in effectiveness trials involving
patients with chronic low back pain who mance-based function (eg, timed up and pragmatic application of behavioural or
received two seemingly different interven- go, 6 min walk test and stair climbing exercise-based interventions. This usually
tions: (1) a McKenzie approach and (2) test) declined compared with baseline, occurs because of the lack of attention
motor control exercises. Chmielewski and whereas self-reported function improved to training issues, ongoing monitoring
associates2 reported similar knee function compared with baseline. Similar disparate of treatment integrity and the lack of
and psychosocial status scores in those findings were evident in patients after a funds to support quality control of the
who received either low intensity or high total knee arthroplasty when comparing intervention.11 Interestingly, most of us
intensity plyometric exercise programmes. self-reported knee injury and osteoar- would question the validity of results of
Finally, van Beijsterveildt et al found no thritis outcome scores to performance a drug trial if: (1) the purity of the drug
differences in the prevention of injuries measures—most self-report measures was not assured, (2) the dose of the drug
among amateur footballers when usual care assessing knee function improved from varied over the course of the study or (3)
was compared against an exercise strategy baseline, whereas the physical perfor- the study drug was sometimes mixed with
that consisted of core stability, eccentric mance measures declined from baseline.5 other drugs. Yet, few attend to issues such
training of thigh muscles, proprioceptive Sports scholarship has been dominated as the quality of the intervention, dose
training, dynamic stabilisation and plyo- by process (mechanisms) research, which or mixing of the intervention with other
metrics exercises.3 These findings beg the seeks to identify specific physiological treatments in orthopaedic-related and
question: Why do we see similar outcomes mechanisms. As a result, process-oriented sports-related studies.
in randomised trials, which compare research often utilises outcome measures To ensure treatment fidelity, studies
treatment strategies that are ostensibly such as strength and flexibility or, in should provide training to those involved
different? Our goal is to provide explana- some cases, physical performance tests. in the application of care and monitor
tions beyond the obvious answers—such as Research suggests that these outcomes the adherence to the specific treatment
treatments that have similarly effective or correlate more strongly with changes in protocol. Without ongoing monitoring,
ineffective causal mechanisms or failing to physiological factors and less strongly clinicians may be less inclined to adhere to
account for patient heterogeneity in trial with psychosocial factors.6 In contrast, the protocol, potentially mixing in other
design and consequent insufficient statis- research suggests that self-report measures elements familiar to them but not specific
tical power to detect differences. In this of disability and pain are more likely to to that treatment protocol. If training is
editorial, we present the reader with four correlate with a patient’s psychosocial brief or poorly applied, clinicians may
additional yet plausible explanations for and health status.6 Studies also show that default to care patterns that they have
why trials report similar outcomes. self-report measures can be influenced used in the past (ie, therapist drift). Tools
by treatment expectations, recall of prior exist to help the researcher assess the
Reason 1: the type of outcome health and pain or health status of the reported fidelity10 12 and best practices
assessment may bias findings patient at the time of scale administra- are available to enhance treatment fidelity.
Our first explanation is the most obvious tion.7 For sports studies that are focused The assurance of treatment fidelity is
one. Outcome measures that capture on training for improved performance, required before assuming that similarities
direct measurements of strength and/ the use of physiologically based outcomes in outcomes between different treatments
or flexibility, or other indices of physical are more likely to yield positive results, are related to comparable intervention
performance, may yield findings that are whereas self-report measures of symptoms effects.
notably different from outcome measures or overall health status may provide very
that capture self-reports of pain, function, different findings.
overall health status or quality of life. Reason 3: specific and shared
Therefore, null findings (no difference Reason 2: treatment fidelity deficiencies mechanisms and the null outcome
among groups) may be related to the fact lead to a null outcome Before we address the third reason for
that different sets of outcome measures Treatment or intervention fidelity refers to similar outcomes, we feel compelled
are being used to compare the groups the reliability and validity of the clinical to confess that this concept is mark-
being studied. interventions that are used.8 Treatment edly complex. In a clinical environment,
fidelity can be enhanced through provider treatments that target specific mecha-
training (eg, ensuring providers deliver nisms may show outcomes that look the
Orthopaedics, Duke Clinical Research Institute, Duke treatment in a competent fashion), treat- same because of shared mechanisms.
University, Durham, North Carolina, USA ment delivery (eg, was the care consis- This complex interplay between specific
Correspondence to Dr Chad E Cook, Duke Clinical
tently provided as intended), treatment and shared mechanisms is best explained
Research Institute, Duke University, Durham, NC 27516, receipt (eg, was the treatment understood using the following theoretical example.
USA ; c​ had.​cook@d​ uke.​edu by the patient) and treatment enactment Let us consider a research study involving

Cook CE, et al. Br J Sports Med Month 2018 Vol 0 No 0    1

Copyright Article author (or their employer) 2018. Produced by BMJ Publishing Group Ltd under licence.
 Downloaded from http://bjsm.bmj.com/ on February 15, 2018 - Published by group.bmj.com

Editorial
patients with rotator cuff pain who are as many sports-related trials involve a rights reserved. No commercial use is permitted unless
treated with three different interventions. close bond between the clinician and the otherwise expressly granted.
In the first arm, exercise is used as the patient. Theoretically, a large percentage
treatment, and it is projected to improve of the improvement in outcomes could
the functional mobility (mediator) of the be reflective of features present in the To cite Cook CE, George SZ, Keefe F. Br J Sports Med
shoulder which, in turn, leads to improved non-specific common factors model (eg, Epub ahead of print: [please include Day Month Year].
scores on the Shoulder Pain and Disability emotional learning, therapeutic alliance doi:10.1136/bjsports-2017-098978
Index (SPADI). In the second arm, cogni- or genuineness). If the contributions are Accepted 4 February 2018
tive behavioural therapy (CBT) is used as greater than the specific mechanisms Br J Sports Med 2018;0:1–2.
the treatment because it is projected to for the targeted intervention due to the doi:10.1136/bjsports-2017-098978
reduce catastrophising (mediator), which outcome measures inability to detect treat-
leads to improved SPADI scores. The ment-specific effects (ie, effects directly
third arm involves aggressive soft-tissue targeted towards an impairment), a null References
1 Halliday MH, Pappas E, Hancock MJ, et al. A
mobilisation, which is projected to alter trial is a likely consequence.
randomized controlled trial comparing the Mckenzie
peripheral and central pain processing method to motor control exercises in people with
(mediator), and that leads to improved chronic low back pain and a directional preference. J
Summary
SPADI scores. Thus, the theoretical Orthop Sports Phys Ther 2016;46:514–22.
Studies involving different interventions 2 Chmielewski TL, George SZ, Tillman SM, et al.
specific mechanisms of the trial include the
may indeed yield similar outcomes. These Low- versus high-intensity plyometric exercise
relationships of (1) exercise to functional
null findings may be related to control- during rehabilitation after anterior cruciate ligament
mobility, (2) CBT to catastrophising and reconstruction. Am J Sports Med
lable (outcomes used and fidelity), often
(3) soft-tissue mobilisation to peripheral 2016;44:609–17.
unmeasured (shared mechanisms) and 3 van Beijsterveldt AM, van de Port IG, Krist MR, et
and central pain processing.
uncontrollable (non-specific common al. Effectiveness of an injury prevention programme
Shared mechanisms occur when an
factors) circumstances. When analysing for adult male amateur soccer players: a cluster-
unanticipated common mediator is present
null findings, the following key points randomised controlled trial. Br J Sports Med
across different treatment approaches.13 2012;46:1114–8.
should be considered:
In the example above, we may find that 1. Self-report findings are influenced by
4 Dayton MR, Judd DL, Hogan CA, et al. Performance-
exercise (treatment) in fact does work based versus self-reported outcomes using the hip
psychological factors, whereas physical disability and osteoarthritis outcome score after
as hypothesised because it improves the performance measures are more likely total hip arthroplasty. Am J Phys Med Rehabil
functional mobility of the shoulder (medi- to be related to physiological factors. 2016;95:132–8.
ator), which in turn improves the SPADI Null findings may reflect the type of
5 Stevens-Lapsley JE, Schenkman ML, Dayton MR.
scores. We may also find that CBT and Comparison of self-reported knee injury and
outcome measure used. osteoarthritis outcome score to performance measures
aggressive soft-tissue mobilisation actu- 2. Outcome similarities between dissim- in patients after total knee arthroplasty. Pm R
ally work, not because of their hypoth- ilar interventions may be a reflection 2011;3:541–9.
esised mechanisms, but because both of of poor treatment fidelity. To ensure 6 Bean JF, Olveczky DD, Kiely DK, et al. Performance-
these treatments improve the functional treatment fidelity, studies should
based versus patient-reported physical function:
mobility of the shoulder (mediator), which what are the underlying predictors? Phys Ther
provide training to those involved in 2011;91:1804–11.
in turn improves the SPADI score. Such the application of care and monitor 7 Kamper SJ, Maher CG, Mackay G. Global rating of
shared mechanisms can explain the largest the adherence to the specific treatment change scales: a review of strengths and weaknesses
proportion of variance in an outcome protocol.
and considerations for design. J Man Manip Ther
measure and may leave very little variance 2009;17:163–70.
3. Shared mechanisms occur when a 8 Carpenter JS, Burns DS, Wu J, et al. Strategies used and
to be explained by specific mechanisms. common mediator is present across data obtained during treatment fidelity monitoring.
When different treatments work through different treatment approaches. Nurs Res 2013;62:59–65.
a shared mechanism, comparisons among 4. Non-specific common factors are 9 Bellg AJ, Borrelli B, Resnick B, et al. Enhancing
treatments may yield null findings. broad factors across all interventional
treatment fidelity in health behavior change studies:
best practices and recommendations from the
types known to favourably influence NIH Behavior Change Consortium. Health Psychol
Reason 4: non-specific common factors self-reported outcomes, regardless 2004;23:443–51.
lead to a null outcome of intervention applied. These could 10 Toomey E, Currie-Murphy L, Matthews J, et al.
The non-specific common factors model is Implementation fidelity of physiotherapist-delivered
potentially influence sports-related
group education and exercise interventions to promote
based on early works of Grencavage and studies where the relationship between self-management in people with osteoarthritis and
Norcross.14 This model suggests that there the clinician and athlete is strong. chronic low back pain: a rapid review part II. Man Ther
are collective, non-specific factors evident 2015;20:287–94.
across most treatments that account for Contributors All the authors contributed to the idea, 11 Ford I, Norrie J. Pragmatic trials. N Engl J Med
improvements seen in outcomes. In partic- design and writing of the editorial. 2016;375:454–63.
12 Salamh P, Cook C, Reiman MP, et al. Treatment
ular, self-report outcome measures are Funding This study was funded NIH/VA/DoD (UG3/ effectiveness and fidelity of manual therapy to
prone to influence from common factors. UH3). the knee: a systematic review and meta-analysis.
Most commonly, these factors include Competing interests None declared. Musculoskeletal Care 2017;15:238–48.
processes associated with therapist quali- 13 Murphy R, Cooper Z, Hollon SD, et al. How do
Provenance and peer review Not commissioned; psychological treatments work? Investigating
ties (eg, positivity, therapist warmth, genu- externally peer reviewed. mediators of change. Behav Res Ther
ineness) and therapeutic alliance (eg, the Data sharing statement There is no data associated 2009;47:1–5.
relationships between the clinician and the with this study. 14 Grencavage LM, Norcross JC. Where are the
patient). These non-specific factors may © Article author(s) (or their employer(s) unless
commonalities among the therapeutic common
be especially germane to sports research factors? Prof Psychol 1990;21:372–8.
otherwise stated in the text of the article) 2018. All

2 Cook CE, et al. Br J Sports Med Month 2018 Vol 0 No 0

 Downloaded from http://bjsm.bmj.com/ on February 15, 2018 - Published by group.bmj.com

Different interventions, same outcomes?

Here are four good reasons
Chad E Cook, Steven Z George and Francis Keefe

Br J Sports Med published online February 15, 2018

Updated information and services can be found at:

http://bjsm.bmj.com/content/early/2018/02/15/bjsports-2017-098978

These include:

References This article cites 14 articles, 2 of which you can access for free at:
http://bjsm.bmj.com/content/early/2018/02/15/bjsports-2017-098978#r
ef-list-1
Email alerting Receive free email alerts when new articles cite this article. Sign up in the
service box at the top right corner of the online article.

Notes

To request permissions go to:

http://group.bmj.com/group/rights-licensing/permissions

To order reprints go to:

http://journals.bmj.com/cgi/reprintform

To subscribe to BMJ go to:

http://group.bmj.com/subscribe/