Secondary Logo

Journal Logo

Inter Causal Biasing Characteristics—Defining Confounding

Collins, Sean, PT, ScD

Cardiopulmonary Physical Therapy Journal: April 2019 - Volume 30 - Issue 2 - p 51–52
doi: 10.1097/CPT.0000000000000113


The author declares no conflicts of interest.

In the October issue editorial readers were provided with a series of factors that are important to consider when embarking on or interpreting results from an observational study. For example, outpatient cardiac or pulmonary rehabilitation participants are ambulatory enough to leave the house and get to the facility, they are willing to participate, they have a physician provider that referred them to rehabilitation, or they somehow knew to request that their physician provider refer them to rehabilitation. All of these characteristics can confound and influence our ability to make inferences to a larger population from a sample. Confounding is not necessarily something that statistical inference (confidence intervals or P-values) helps us address. The easiest approach to consider whether a confounder is present is to consider a graph of the proposed cause and effect claim. For example, if “participate in rehabilitation” then “improved outcome” is the underlying causal claim; we must consider whether anything is a potential cause of both participation in rehabilitation and an improved outcome. For example, being ambulatory is necessary to participate and it is also possibly a cause of an improved outcome. Having health insurance, a physician provider that refers or knowing to ask your physician provider to refer are all causes of participation in rehabilitation and all are also possibly a cause for an improved outcome. In a directed acyclic graph this is referred to as a biasing path: X causes Y, but Z causes both X and Y. Therefore, Z is a confounder. There is a great free tool online that can be used to help draw and alert observational researchers, observational research consumers or observational research reviewers to the presence of confounding ( Once it is determined that there are confounders, the question is what can we do about it? In observational designs there are less options which is why researcher, reviewer and reader vigilance is critical. In an upcoming editorial we will discuss possible approaches for researchers to address confounding. At the very least, researcher, reviewer and reader should all keep the STROBE checklist in mind (, which is a requirement when submitting an observational study to CPTJ, and in which “Not Applicable” is not an acceptable answer to questions about addressing bias and confounding.

Back to Top | Article Outline


In this issue we bring you 5 original research reports all having used observational methods—2 prospective, 1 retrospective, and 2 cross-sectional designs. Each provide great examples of the power and use of observational designs in hypothesis generation and preliminary testing of highly relevant and important clinical hypotheses. Arena et al1 report on a prospective observational study to describe and compare preseason and postseason blood pressure (BP) measures among collegiate athletes and correlate BP measurement and body mass index across a season. This is a very interesting study and without giving too many of the very interesting findings let me just say that it highlights the need for vigilant monitoring even in individuals that may not have been previously suspected. After all, the most informative information reveals something unsuspected. The authors also provide interesting thoughts for future research regarding study of causal mechanisms underlying the changes as well as implications.

Byrd et al2 retrospectively evaluate the effect of a 1-month physical therapy based outpatient program on exercise capacity, symptoms, quality of life and examine predictors of functional outcome changes in adults awaiting lung transplantation. There is a very interesting relationship between initial 6 minute walk distance (6MWD) and oxygen utilization with 6MWD improvement. These findings provide insight into possible adaptive mechanisms and could lead to the early determination of adaptive capacity which could be extremely useful for goal setting. These topics deserve our attention, continued thought, research and discussion. Pepin et al3 study the association between functional ability and physical activity in individuals with transtibial amputations and highlight a need to consider long term implications of functional gait speed and balance on the likelihood of sedentary behavior including the hypothesized dynamic feedback loop of possible functional decline.

When considering predictors of adaptive capacity and dynamic feedback loops Himes et al4 report on the relationship among lower extremity blood flow, strength and fall risk in adults enrolled in phase II cardiac rehabilitation (CR). They demonstrate a significant correlation between the 30-second chair stand test (30CST) and the functional gait assessment score and propose the 30CST for use as a quick fall-risk screening measure in the phase II CR population. It would be interesting to see a longitudinal follow up study of this type to determine if there is a similar finding between 30CST and outcomes as in the variables measured in the study by Byrd et al.2

Blackwood et al5 reports on the relationship between cognitive performance and physical mobility in community-dwelling older adults with and without cardiovascular disease. This study raises important considerations for the consideration of executive cognitive processes and mobility. If there is an influence between executive function, physical mobility and falls risk then predication models for fall risk based entirely on physical mobility may fail to capture the multi-dimensional nature of risk. As with all great observational studies, further research is needed.

Back to Top | Article Outline


Part of my role as an Editor is to be surveying the profession and considering the direction of its intellectual pursuits. Readers are aware that our profession is looking to increase it's engagement in primary and preventive care. We have seen an increase in the number of publications addressing such topics and that is excellent—it is important that intellectual capacity keep pace with professional aspirations. There seem to be a few gaps though and it would be wonderful to hear from readers (including current or future authors) about these gaps. For example, soon we will publish an invited commentary on the prevalence and testing of peripheral neuropathy using electro diagnostics in patients with chronic cardiopulmonary conditions, something that is well within the physical therapy (PT) scope of practice and could be a considered a primary care testing modality with regular follow up. If nothing else, it has the potential to inform a PT initiated plan of care. Another topic on my mind lately has been the connection between posture (forward head for example), temporomandibular disorders, breathing patterns (including mouth vs nasal breathing as well as lower vs upper chest breathing), obesity, obstructive sleep apnea, elevated BP and perceptions of breathlessness. These signs and symptoms have very valid reasons to cluster together, which means finding one should promote the investigation of the others. If you have topics on your mind and would like to share them, please email me: In the coming issues the Journal will include short invited commentaries that will address such issues and provide questions, ideas and recommendations to both current and future researchers.

Back to Top | Article Outline


1. Arena S, LaBelle L, Larsen J, Palomino L, Hew-Butler T, Peterson E. Description and comparison of preseason and postseason blood pressure measures among collegiate athletes: A prospective observational study. Cardiopulm Phys Ther J. 2019;30:53–60.
2. Byrd R, Smith P, Mohamedaly O, Snyder LD, Pastva AM. A one-month physical therapy–based outpatient program for adults awaiting lung transplantation: A retrospective analysis of exercise capacity, symptoms, and quality of life. Cardiopulm Phys Ther J. 2019;30:61–69.
3. Pepin ME, Devour A, Coolsaet R, Galen S. Correlation between functional ability and physical activity in individuals with transtibial amputations: A cross-sectional study. Cardiopulm Phys Ther J. 2019;30:70–78.
4. Himes MK, Moore ES, Robinson BS, Daniel TE. Do lower extremity strength and lower extremity blood flow predict the score on the functional gait assessment in patients enrolled in phase II cardiac rehabilitation? Cardiopulm Phys Ther J. 2019;30:79–85.
5. Blackwood J. Cognitive function is associated with mobility in community dwelling older adults with a history of cardiovascular disease. Cardiopulm Phys Ther J. 2019;30:86–93.
© 2019 Cardiovascular and Pulmonary Section, APTA