Steven D. Stovitz
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Steven D. Stovitz has a total of 188 co-authors. Here are the top co-authors by number of shared publications:
Ian Shrier
From the University of Minnesota, Minneapolis, MN (SDS); McGill and University of Toronto, Toronto, CA (IS, HRB, JSK)
13 shared publications- Ian Shrier
McGill University
11 shared publications - Jane Steinman Kaufman
Department of Epidemiology, Biostatistics and Occupational Health, McGill University, Montreal, Quebec, Canada
10 shared publications - Evert Verhagen
Amsterdam Collaboration on Health and Safety in Sports, Department of Public and Occupational Health, Amsterdam Movement Sciences, Amsterdam UMC, Amsterdam, Netherlands
6 shared publications Hailey R. Banack
From the University of Minnesota, Minneapolis, MN (SDS); McGill and University of Toronto, Toronto, CA (IS, HRB, JSK)
6 shared publicationsJohn H. Himes
School of Public Health, Division of Epidemiology and Community Health, University of Minnesota, Minneapolis, MN
4 shared publications
Latest Publications
- Open Access
- 6 October 2026
- Published by Center for Open Science in MetaArXiv
Moving Beyond Statistical Significance: How to Improve the Reporting and Interpretation of Medical Research
Many methodologists have warned against interpreting statistical estimates through dichotomous labels such as “significant” and “non-significant.” Yet, articles in medical journals still rely heavily on P-value thresholds and confidence interval boundaries to guide inference and communication of research findings. Here, Rovetta and colleagues discuss how such a focus on statistical “significance” obscures clinical relevance and uncertainty, whilst distorting evidence appraisals. Rather, they suggest P-values and confidence intervals should be interpreted as continuous measures of compatibility between data and hypotheses, which also reflect choices about how data are collected, modeled, and analyzed.
- 6 October 2026
- Published by Center for Open Science in MetaArXiv
Open AccessMoving Beyond Statistical Significance: How to Improve the Reporting and Interpretation of Medical Research
Many methodologists have warned against interpreting statistical estimates through dichotomous labels such as “significant” and “non-significant.” Yet, articles in medical journals still rely heavily on P-value thresholds and confidence interval boundaries to guide inference and communication of research findings. Here, Rovetta and colleagues discuss how such a focus on statistical “significance” obscures clinical relevance and uncertainty, whilst distorting evidence appraisals. Rather, they suggest P-values and confidence intervals should be interpreted as continuous measures of compatibility between data and hypotheses, which also reflect choices about how data are collected, modeled, and analyzed.
- 18 February 2026
- Published by Oxford University Press (OUP) in International Journal of Epidemiology
- Vol. 55 (2)
Identification of the causal odds ratio in test negative designs
We found the article by Delauney et al. [ 1] on potential bias in test-negative designs and COVID-19 interesting. However, we could not understand the reasoning for an assumption in their causal directed acyclic graph (DAG) and believe that their attempt to minimize bias by blocking the specified non-causal path violates a rule for identification of the causal odds ratio. In their Figure 2A, the authors argue that previously published causal DAGs should be modified to indicate that vaccination may cause self-testing (they added a direct arrow between the two variables) because of the published association between vaccination and self-testing [ 2]. However, an association between vaccination and self-testing is already expected without a direct causal link because both are caused by ‘healthcare seeking behavior’, and vaccination causes infection, which causes symptoms, which causes self-testing. We recognize that current recommendations are to include direct arrows between all variables unless there is a strong reason to remove them [ 3]. We would welcome an explanation for how vaccinations could directly cause self-testing, as we were unable to think of a reason. One of the great advantages of causal DAGs is that it forces authors to be transparent about their assumptions. Occasionally, this requires a clear narrative explaining why arrows are included or excluded in causal DAGs where it is not immediately obvious.
- 18 February 2026
- Published by Oxford University Press (OUP) in International Journal of Epidemiology
- Vol. 55 (2)
Identification of the causal odds ratio in test negative designs
We found the article by Delauney et al. [ 1] on potential bias in test-negative designs and COVID-19 interesting. However, we could not understand the reasoning for an assumption in their causal directed acyclic graph (DAG) and believe that their attempt to minimize bias by blocking the specified non-causal path violates a rule for identification of the causal odds ratio. In their Figure 2A, the authors argue that previously published causal DAGs should be modified to indicate that vaccination may cause self-testing (they added a direct arrow between the two variables) because of the published association between vaccination and self-testing [ 2]. However, an association between vaccination and self-testing is already expected without a direct causal link because both are caused by ‘healthcare seeking behavior’, and vaccination causes infection, which causes symptoms, which causes self-testing. We recognize that current recommendations are to include direct arrows between all variables unless there is a strong reason to remove them [ 3]. We would welcome an explanation for how vaccinations could directly cause self-testing, as we were unable to think of a reason. One of the great advantages of causal DAGs is that it forces authors to be transparent about their assumptions. Occasionally, this requires a clear narrative explaining why arrows are included or excluded in causal DAGs where it is not immediately obvious.
- 19 December 2025
- Published by Springer Nature in Sports Medicine
- Vol. 56 (4) , 837-843
The Risks of Misunderstanding the Term “Risk Factor”: A Primer with Suggestions to Improve Sports Medicine
The term “risk factor” is commonly used in research. Although many interpret the term to imply that the risk factor causes the outcome, others use the term to mean a marker for the outcome, which may or may not be a cause of the outcome. How the term risk factor is interpreted can importantly influence the way that study findings are applied in real world settings. For example, if a risk factor is wrongly interpreted to be a cause of an outcome when it is merely associated with the outcome because of noncausal reasons, then wasteful interventions may be developed, recommended, and implemented. The primary aims of this article are (1) to describe how varying definitions of the term risk factor can cause misunderstandings and potentially negatively impact the field of sports medicine, and (2) to propose new, more specific, terminology. We first review some basic concepts on how variables can be associated due to either causal or noncausal reasons and then discuss possible explanations for why the term risk factor continues to be misunderstood. We illustrate how using the term risk factor without further specification creates misunderstandings that can lead to the development and implementation of ineffective interventions. Finally, with the hope of improving communication and avoiding ambiguity in sports medicine, we suggest using “causal risk factor” if the evidence supports causality, “noncausal risk factor” if the evidence does not support causality, and “risk marker” for those not wishing to commit to a causal or noncausal claim.
- 19 December 2025
- Published by Springer Nature in Sports Medicine
- Vol. 56 (4) , 837-843
The Risks of Misunderstanding the Term “Risk Factor”: A Primer with Suggestions to Improve Sports Medicine
The term “risk factor” is commonly used in research. Although many interpret the term to imply that the risk factor causes the outcome, others use the term to mean a marker for the outcome, which may or may not be a cause of the outcome. How the term risk factor is interpreted can importantly influence the way that study findings are applied in real world settings. For example, if a risk factor is wrongly interpreted to be a cause of an outcome when it is merely associated with the outcome because of noncausal reasons, then wasteful interventions may be developed, recommended, and implemented. The primary aims of this article are (1) to describe how varying definitions of the term risk factor can cause misunderstandings and potentially negatively impact the field of sports medicine, and (2) to propose new, more specific, terminology. We first review some basic concepts on how variables can be associated due to either causal or noncausal reasons and then discuss possible explanations for why the term risk factor continues to be misunderstood. We illustrate how using the term risk factor without further specification creates misunderstandings that can lead to the development and implementation of ineffective interventions. Finally, with the hope of improving communication and avoiding ambiguity in sports medicine, we suggest using “causal risk factor” if the evidence supports causality, “noncausal risk factor” if the evidence does not support causality, and “risk marker” for those not wishing to commit to a causal or noncausal claim.
- Open Access
- 5 October 2025
- Published by BMJ in British Journal of Sports Medicine
- Vol. 59 (24) , 1712-1715
Interpreting p values and interval estimates based on practical relevance: guidance for the sports medicine clinician
Compatibility refers to how well the hypothesis fits or explains the data; compatibility is highest when p=1 and decreases as p approaches 0.3–6 10 Interpreting p values as measures of compatibility is an alternative to the traditional ‘significance’ interpretation; indeed, the latter is often misunderstood as referring to clinical significance and requires strong assumptions to justify its use for decisions.1 3–6 The minimum for reasonable compatibility is typically set at p=0.05, yielding a 100×(1–0.05)%=95% compatibility interval that is numerically identical to a ‘95% CI’ and so is also denoted as ‘95% CI’. Box 1 provides an example of the application of these concepts.
- 5 October 2025
- Published by BMJ in British Journal of Sports Medicine
- Vol. 59 (24) , 1712-1715
Open AccessInterpreting p values and interval estimates based on practical relevance: guidance for the sports medicine clinician
Compatibility refers to how well the hypothesis fits or explains the data; compatibility is highest when p=1 and decreases as p approaches 0.3–6 10 Interpreting p values as measures of compatibility is an alternative to the traditional ‘significance’ interpretation; indeed, the latter is often misunderstood as referring to clinical significance and requires strong assumptions to justify its use for decisions.1 3–6 The minimum for reasonable compatibility is typically set at p=0.05, yielding a 100×(1–0.05)%=95% compatibility interval that is numerically identical to a ‘95% CI’ and so is also denoted as ‘95% CI’. Box 1 provides an example of the application of these concepts.
- 2 May 2025
- Published by BMJ in British Journal of Sports Medicine
- Vol. 59 (15) , 1107-1109
Determination of athlete-exposures in team sports: it depends on the research question
- 2 May 2025
- Published by BMJ in British Journal of Sports Medicine
- Vol. 59 (15) , 1107-1109
Determination of athlete-exposures in team sports: it depends on the research question
- Open Access
- 1 November 2024
- Published by American Board of Family Medicine (ABFM) in The Journal of the American Board of Family Medicine
- Vol. 37 (6) , 1133-1139
In Defense of Generalists: Primary Care Observations Have Systematic Advantages
There is a perception that physician specialists are the experts, and that generalists, often called primary care physicians, are the ‘Jack of all trades, master of none.’ However, this perception ignores the knowledge that comes from clinical observations, and it is here where the generalist viewpoint has advantages. Generalists observe patients before and after they develop medical concerns. In contrast, the clinical experiences of specialists are often focused on a subset of the population, typically based on certain concerns or age groups. Seeing only a slice of the population may lead to biased clinical perceptions for the effects of behaviors, conditions, or treatments in the general population. In this commentary we demonstrate that limiting clinical observations to patients who have a certain condition or are above a certain age can make exposures which are harmful seem beneficial, and exposures which are beneficial seem harmful. Using hypothetical examples, we illustrate that there are systematic reasons why generalists who see patients over the long-term, both before and after medical concerns, can have a more accurate vantagepoint.
- 1 November 2024
- Published by American Board of Family Medicine (ABFM) in The Journal of the American Board of Family Medicine
- Vol. 37 (6) , 1133-1139
Open AccessIn Defense of Generalists: Primary Care Observations Have Systematic Advantages
There is a perception that physician specialists are the experts, and that generalists, often called primary care physicians, are the ‘Jack of all trades, master of none.’ However, this perception ignores the knowledge that comes from clinical observations, and it is here where the generalist viewpoint has advantages. Generalists observe patients before and after they develop medical concerns. In contrast, the clinical experiences of specialists are often focused on a subset of the population, typically based on certain concerns or age groups. Seeing only a slice of the population may lead to biased clinical perceptions for the effects of behaviors, conditions, or treatments in the general population. In this commentary we demonstrate that limiting clinical observations to patients who have a certain condition or are above a certain age can make exposures which are harmful seem beneficial, and exposures which are beneficial seem harmful. Using hypothetical examples, we illustrate that there are systematic reasons why generalists who see patients over the long-term, both before and after medical concerns, can have a more accurate vantagepoint.
- Open Access
- 8 September 2024
- Published by Elsevier in North American Spine Society Journal (NASSJ)
- Vol. 20, 100557
A Tai chi and qigong mind-body program for low back pain: A virtually delivered randomized control trial
- 8 September 2024
- Published by Elsevier in North American Spine Society Journal (NASSJ)
- Vol. 20, 100557
Open AccessA Tai chi and qigong mind-body program for low back pain: A virtually delivered randomized control trial
- 7 May 2024
- Published by Springer Nature in Sports Medicine
- Vol. 54 (8) , 1991-2000
Identifying and Minimizing Incentives for Competing Interests in Sports Medicine Publications
Academics in sports medicine as well as other medical fields are generally expected to publish research and opinions in peer-reviewed journals. The peer-review process is intended to protect against the publication of flawed research and unsubstantiated claims. However, both financial and non-financial competing interests may result in sub-optimal results by affecting investigators, editors, peer reviewers, academic institutions, and publishers. In this article, we focus on the non-financial competing interests created in our current academic system. Because these competing interests are embedded in our current scholastic framework, the potential biases are difficult to quantify. To minimize the effect of these competing interests, we review and highlight some underlying incentives for each stakeholder and some potential solutions to mitigate their effects.
- 7 May 2024
- Published by Springer Nature in Sports Medicine
- Vol. 54 (8) , 1991-2000
Identifying and Minimizing Incentives for Competing Interests in Sports Medicine Publications
Academics in sports medicine as well as other medical fields are generally expected to publish research and opinions in peer-reviewed journals. The peer-review process is intended to protect against the publication of flawed research and unsubstantiated claims. However, both financial and non-financial competing interests may result in sub-optimal results by affecting investigators, editors, peer reviewers, academic institutions, and publishers. In this article, we focus on the non-financial competing interests created in our current academic system. Because these competing interests are embedded in our current scholastic framework, the potential biases are difficult to quantify. To minimize the effect of these competing interests, we review and highlight some underlying incentives for each stakeholder and some potential solutions to mitigate their effects.
- Open Access
- 1 January 2024
- Published by American Board of Family Medicine (ABFM) in The Journal of the American Board of Family Medicine
- Vol. 37 (1) , 154
Re: Becoming a Phronimos: Evidence-Based Medicine, Clinical Decision Making and the Role of Practical Wisdom in Primary Care
- 1 January 2024
- Published by American Board of Family Medicine (ABFM) in The Journal of the American Board of Family Medicine
- Vol. 37 (1) , 154
Open AccessRe: Becoming a Phronimos: Evidence-Based Medicine, Clinical Decision Making and the Role of Practical Wisdom in Primary Care
- Open Access
- 14 July 2023
- Published by Oxford University Press (OUP) in International Journal of Epidemiology
- Vol. 52 (6) , 1968-1974
Identifiability of causal effects in test-negative design studies
Causal directed acyclic graphs (DAGs) are often used to select variables in a regression model to identify causal effects. Outcome-based sampling studies, such as the ‘test-negative design’ used to assess vaccine effectiveness, present unique challenges that are not addressed by the common back-door criterion. Here we discuss intuitive, graphical approaches to explain why the common back-door criterion cannot be used for identification of population average causal effects with outcome-based sampling studies. We also describe graphical rules that can be used instead in outcome-based sampling studies when the objective is limited to determining if the causal odds ratio is identifiable, and illustrate recent changes to the free online software Dagitty which incorporate these principles.
- 14 July 2023
- Published by Oxford University Press (OUP) in International Journal of Epidemiology
- Vol. 52 (6) , 1968-1974
Open AccessIdentifiability of causal effects in test-negative design studies
Causal directed acyclic graphs (DAGs) are often used to select variables in a regression model to identify causal effects. Outcome-based sampling studies, such as the ‘test-negative design’ used to assess vaccine effectiveness, present unique challenges that are not addressed by the common back-door criterion. Here we discuss intuitive, graphical approaches to explain why the common back-door criterion cannot be used for identification of population average causal effects with outcome-based sampling studies. We also describe graphical rules that can be used instead in outcome-based sampling studies when the objective is limited to determining if the causal odds ratio is identifiable, and illustrate recent changes to the free online software Dagitty which incorporate these principles.