Matthew S. Tenan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Matthew S. Tenan has a total of 42 co-authors. Here are the top co-authors by number of shared publications:
Andrew J. Tweedell
U.S. Army Research Laboratory, Human Research and Engineering Directorate, Aberdeen Proving Ground, MD, USA
3 shared publications- Aaron R. Caldwell
Institute for Community Health Innovation, University of Arkansas for Medical Sciences Northwest, Springdale, USA
2 shared publications Ali R. Rezai
Department of Neuroscience Rockefeller Neuroscience Institute, West Virginia University Morgantown WV USA
2 shared publications- Andrew David Vigotsky
Northwestern University
2 shared publications Anthony C. Hackney
Department of Exercise and Sport Science, Department of Nutrition, School of Public Health, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, 21005-5425, USA
2 shared publicationsBob Alejo
Rockefeller Neuroscience Institute, Morgantown, WV
2 shared publications
Latest Publications
- 29 December 2025
- Published byĀ Taylor & FrancisĀ inĀ Journal of Sports Sciences
- p. 1-14
Improving real-world data analysis: Revisiting claims that gender meaningfully impacts marathon pacing
The use of real-world data in sports is increasing but without a commensurate increase in analytical rigour or interpretation. This study reexamines a seminal analysis which claims to show gender differences in marathon pacing are 'robust'. The earlier work suffers from serious methodological flaws common to real-world data analysis: (1) lack of quality control in the raw data, (2) overuse and misuse of ratio-based metrics, (3) failure to validate the assumptions of statistical models. A proper and valid analysis of the exact same real-world data indicates that there are some potentially interesting differences in pacing between genders, but the differences are only evident in younger and slower runners. While the previous, invalid analysis has been used to suggest there is some sort of biological mechanism for a gender difference in marathon pacing, our valid analysis is more suggestive of a social difference in this large cohort of recreational marathon runners. This case highlights the need for researchers to prioritize the quality, not just the quantity, of real-world data analyses. Moreover, researchers should avoid defaulting to sensational interpretations when more cautious, grounded explanations may be equally - if not more - plausible.
- 29 December 2025
- Published byĀ Taylor & FrancisĀ inĀ Journal of Sports Sciences
- p. 1-14
Improving real-world data analysis: Revisiting claims that gender meaningfully impacts marathon pacing
The use of real-world data in sports is increasing but without a commensurate increase in analytical rigour or interpretation. This study reexamines a seminal analysis which claims to show gender differences in marathon pacing are 'robust'. The earlier work suffers from serious methodological flaws common to real-world data analysis: (1) lack of quality control in the raw data, (2) overuse and misuse of ratio-based metrics, (3) failure to validate the assumptions of statistical models. A proper and valid analysis of the exact same real-world data indicates that there are some potentially interesting differences in pacing between genders, but the differences are only evident in younger and slower runners. While the previous, invalid analysis has been used to suggest there is some sort of biological mechanism for a gender difference in marathon pacing, our valid analysis is more suggestive of a social difference in this large cohort of recreational marathon runners. This case highlights the need for researchers to prioritize the quality, not just the quantity, of real-world data analyses. Moreover, researchers should avoid defaulting to sensational interpretations when more cautious, grounded explanations may be equally - if not more - plausible.
- 1 July 2025
- Published byĀ American Physiological SocietyĀ inĀ Journal of Applied Physiology
- Vol.Ā 139 Ā (1) , 101-102
A perspective or review based entirely on low-quality evidence is simply an opinion
- 1 July 2025
- Published byĀ American Physiological SocietyĀ inĀ Journal of Applied Physiology
- Vol.Ā 139 Ā (1) , 101-102
A perspective or review based entirely on low-quality evidence is simply an opinion
- 28 June 2025
- Published byĀ WileyĀ inĀ Pain Practice
- Vol.Ā 25 Ā (6)
Developing the Necessary Information for Sample Size Calculations in Future Spinal Cord Stimulation Clinical Trials
Background: Spinal cord stimulation (SCS) is a nonpharmacological, minimally invasive intervention designed to ameliorate chronic low back pain. However, metaāanalyses have not supported the use of SCS due to a lack of highāquality evidence. This work provides the necessary information to design better statistically powered clinical trials for SCS by providing estimates and variances for various patientāreported outcomes and biometrics across time in this population.Methods: A cohort of 18 patients was followed across 7 months before and after SCS implantation. The patients were administered a monthly battery of patientāreported outcomes, and daily biometrics were obtained. Multilevel Bayesian distributional models quantified the median and variance change across time, both of which are necessary in sample size calculations. Secondary to estimating effects, we also determine the probability of a directional effect and equivalence.Results: Scales for pain show sustained improvements and stable variance. There were no changes in qualityāofālife medians or variance. Robust improvements were made in fatigue and reported sleep quality, despite an unclear effect on total hours slept as recorded by the daily wearable. Resting heart rate decreased after SCS and had low changes in variance; whereas pulse rate variability/heart rate variability exhibited no median change across time but wild swings in variance, indicating it is a poor biomarker in this population.Conclusions: This study reports many patientāreported outcomes and digital biomarkers used in SCS clinical research, including which ones have potential value and the exact information necessary to plan future highāquality clinical trials in the SCS population.Significance: There is currently only lowāquality evidence that spinal cord stimulation is effective for low back pain. This study supplies all the necessary information (effect estimates, variances and withināmeasure correlations across time) to better estimate sample sizes, a primary criticism of current evidence in this population. We also provide a preliminary indication as to which patientāreported outcomes and wearable measures are most effective in a spinal cord stimulation population.
- 28 June 2025
- Published byĀ WileyĀ inĀ Pain Practice
- Vol.Ā 25 Ā (6)
Developing the Necessary Information for Sample Size Calculations in Future Spinal Cord Stimulation Clinical Trials
Background: Spinal cord stimulation (SCS) is a nonpharmacological, minimally invasive intervention designed to ameliorate chronic low back pain. However, metaāanalyses have not supported the use of SCS due to a lack of highāquality evidence. This work provides the necessary information to design better statistically powered clinical trials for SCS by providing estimates and variances for various patientāreported outcomes and biometrics across time in this population.Methods: A cohort of 18 patients was followed across 7 months before and after SCS implantation. The patients were administered a monthly battery of patientāreported outcomes, and daily biometrics were obtained. Multilevel Bayesian distributional models quantified the median and variance change across time, both of which are necessary in sample size calculations. Secondary to estimating effects, we also determine the probability of a directional effect and equivalence.Results: Scales for pain show sustained improvements and stable variance. There were no changes in qualityāofālife medians or variance. Robust improvements were made in fatigue and reported sleep quality, despite an unclear effect on total hours slept as recorded by the daily wearable. Resting heart rate decreased after SCS and had low changes in variance; whereas pulse rate variability/heart rate variability exhibited no median change across time but wild swings in variance, indicating it is a poor biomarker in this population.Conclusions: This study reports many patientāreported outcomes and digital biomarkers used in SCS clinical research, including which ones have potential value and the exact information necessary to plan future highāquality clinical trials in the SCS population.Significance: There is currently only lowāquality evidence that spinal cord stimulation is effective for low back pain. This study supplies all the necessary information (effect estimates, variances and withināmeasure correlations across time) to better estimate sample sizes, a primary criticism of current evidence in this population. We also provide a preliminary indication as to which patientāreported outcomes and wearable measures are most effective in a spinal cord stimulation population.
- 6 June 2025
- Published byĀ Oxford University Press (OUP)Ā inĀ American Journal of Epidemiology
- Vol.Ā 194 Ā (9) , 2499-2506
A causal and replication analysis of claims that jet lag affects team sport performance
It is broadly accepted that jet lag impacts human performance in sport; however, this has never been validated within a causal inference framework. Here, we used the potential outcomes framework to determine if jet lag, conditional on game time, causes collegiate football teams to under- or overperform expectations. We also attempted to replicate seminal noncausal analyses in professional football. Nine collegiate football seasons (2013-2022, omitting 2020), for a total of 6245 games were analyzed. A generalized additive model with penalized splines and independence weights was used to evaluate interactive causal effects of hours gained/lost in travel and game time on the probability of away teams beating the Las Vegas spread. Furthermore, we used our collegiate cohort to exactly replicate two previous studies purporting to show that NFL teams are adversely affected by jet lag. Jet lagās effects were compatible with the null hypothesis in both the causal analysis (P = .133) and noncausal replications. Practically, if the effects of jet lag are unclear, then it is also unclear whether interventions to treat jet lag in elite sport are warranted for teams crossing fewer than 3 time zones if the primary goal is to optimize winning.
- 6 June 2025
- Published byĀ Oxford University Press (OUP)Ā inĀ American Journal of Epidemiology
- Vol.Ā 194 Ā (9) , 2499-2506
A causal and replication analysis of claims that jet lag affects team sport performance
It is broadly accepted that jet lag impacts human performance in sport; however, this has never been validated within a causal inference framework. Here, we used the potential outcomes framework to determine if jet lag, conditional on game time, causes collegiate football teams to under- or overperform expectations. We also attempted to replicate seminal noncausal analyses in professional football. Nine collegiate football seasons (2013-2022, omitting 2020), for a total of 6245 games were analyzed. A generalized additive model with penalized splines and independence weights was used to evaluate interactive causal effects of hours gained/lost in travel and game time on the probability of away teams beating the Las Vegas spread. Furthermore, we used our collegiate cohort to exactly replicate two previous studies purporting to show that NFL teams are adversely affected by jet lag. Jet lagās effects were compatible with the null hypothesis in both the causal analysis (P = .133) and noncausal replications. Practically, if the effects of jet lag are unclear, then it is also unclear whether interventions to treat jet lag in elite sport are warranted for teams crossing fewer than 3 time zones if the primary goal is to optimize winning.
- 1 October 2024
- Published byĀ Wolters Kluwer HealthĀ inĀ Medicine & Science in Sports & Exercise
- Vol.Ā 56 Ā (10S) , 186-187
Jet Lag Does Not Decrease Collegiate Football Performance
- 1 October 2024
- Published byĀ Wolters Kluwer HealthĀ inĀ Medicine & Science in Sports & Exercise
- Vol.Ā 56 Ā (10S) , 186-187
Jet Lag Does Not Decrease Collegiate Football Performance
- 1 October 2024
- Published byĀ National Athletic Trainers' AssociationĀ inĀ Journal of Athletic Training
- Vol.Ā 59 Ā (10) , 979-983
Athlete Health and Human Performance Will Not Improve Without Transdisciplinary Collaboration and Data Sharing in Elite Sport
There are two largely competing models for an athletics organization at both the collegiate and professional levels: the High Performance Model and the Medical Model. The High Performance Model arises from international Football perspective that places a āPerformance Directorā at the center of teams supporting the athletes. The Medical Model, supported by both the National Athletic Trainers Association and the NCAA, separates off medical staff (athletic trainers, physical therapists, and physicians, predominantly) and emphasizes the autonomy of medical decisions. The Medical Model has left clinicians in a āmedical siloā, limiting our ability to care for the individual athletes as holistic people and limits our wider impact in the field of athlete health and injury mitigation. We argue that Medical. Model is consistent with the High Performance Model only if we reject the notion that the āPerformance Directorā is an administrative person and instead conceptualize this as a āHealth and Performance Information Hubā which facilitates transdisciplinary collaboration. This Commentary details how a data broker system can be used to accelerate transdisciplinary collaboration within an athletic organization, leading to better healthcare for athletes and improved team and individual performance. Furthermore, a transdisciplinary organization with data sharing is able to turn real-world data into real-world evidence, enhancing the care and performance of athletes locally as well as facilitating the creation of generalizable knowledge in the area of sports medicine and human performance.
- 1 October 2024
- Published byĀ National Athletic Trainers' AssociationĀ inĀ Journal of Athletic Training
- Vol.Ā 59 Ā (10) , 979-983
Athlete Health and Human Performance Will Not Improve Without Transdisciplinary Collaboration and Data Sharing in Elite Sport
There are two largely competing models for an athletics organization at both the collegiate and professional levels: the High Performance Model and the Medical Model. The High Performance Model arises from international Football perspective that places a āPerformance Directorā at the center of teams supporting the athletes. The Medical Model, supported by both the National Athletic Trainers Association and the NCAA, separates off medical staff (athletic trainers, physical therapists, and physicians, predominantly) and emphasizes the autonomy of medical decisions. The Medical Model has left clinicians in a āmedical siloā, limiting our ability to care for the individual athletes as holistic people and limits our wider impact in the field of athlete health and injury mitigation. We argue that Medical. Model is consistent with the High Performance Model only if we reject the notion that the āPerformance Directorā is an administrative person and instead conceptualize this as a āHealth and Performance Information Hubā which facilitates transdisciplinary collaboration. This Commentary details how a data broker system can be used to accelerate transdisciplinary collaboration within an athletic organization, leading to better healthcare for athletes and improved team and individual performance. Furthermore, a transdisciplinary organization with data sharing is able to turn real-world data into real-world evidence, enhancing the care and performance of athletes locally as well as facilitating the creation of generalizable knowledge in the area of sports medicine and human performance.
- 27 June 2024
- Published byĀ Springer NatureĀ inĀ Sports Medicine
- Vol.Ā 54 Ā (11) , 2927-2934
Are Wearable Photoplethysmogram-Based Heart Rate Variability Measures Equivalent to Electrocardiogram? A Simulation Study
Background Traditional electrocardiography (ECG)-derived heart rate variability (HRV) and photoplethysmography (PPG)-derived āHRVā (termed PRV) have been reported interchangeably. Any potential dissociation between HRV and PRV could be due to the variability in pulse arrival time (PAT; time between heartbeat and peripheral pulse). Objective This study examined if PRV is equivalent to ECG-derived HRV and if PRVās innate error makes it a high-quality measurement separate from HRV. Methods ECG data from 1084 subjects were obtained from the PhysioNet Autonomic Aging dataset, and individual PAT dispersions for both the wrist (n = 42) and finger (n = 49) were derived from Mol et al. (Exp Gerontol. 2020; 135: 110938). A Bayesian simulation was constructed whereby the individual arrival times of the PPG wave were calculated by placing a Gaussian prior on the individual QRS-wave timings of each ECG series. The standard deviation (Ļ) of the prior corresponds to the PAT dispersion from Mol et al. This was simulated 10,000 times for each PAT Ļ. The root mean square of successive differences (RMSSD) and standard deviation of NāN intervals (SDNN) were calculated for both HRV and PRV. The Region of Practical Equivalence bounds (ROPE) were set a priori at ± 0.2% of true HRV. The highest density interval (HDI) width, encompassing 95% of the posterior distribution, was calculated for each PAT Ļ. Results The lowest PAT Ļ (2.0 SD) corresponded to 88.4% within ROPE for SDNN and 21.4% for RMSSD. As the Ļ of PAT increases, the equivalence of PRV and HRV decreases for both SDNN and RMSSD. The HDI interval width increases with increasing PAT Ļ, with the HDI width increasing at a higher rate for RMSSD than SDNN. Conclusions For individuals with greater PAT variability, PRV is not a surrogate for HRV. When considering PRV as a unique biometric measure, SDNN may have more favorable measurement properties than RMSSD, though both exhibit a non-uniform measurement error.
- 27 June 2024
- Published byĀ Springer NatureĀ inĀ Sports Medicine
- Vol.Ā 54 Ā (11) , 2927-2934
Are Wearable Photoplethysmogram-Based Heart Rate Variability Measures Equivalent to Electrocardiogram? A Simulation Study
Background Traditional electrocardiography (ECG)-derived heart rate variability (HRV) and photoplethysmography (PPG)-derived āHRVā (termed PRV) have been reported interchangeably. Any potential dissociation between HRV and PRV could be due to the variability in pulse arrival time (PAT; time between heartbeat and peripheral pulse). Objective This study examined if PRV is equivalent to ECG-derived HRV and if PRVās innate error makes it a high-quality measurement separate from HRV. Methods ECG data from 1084 subjects were obtained from the PhysioNet Autonomic Aging dataset, and individual PAT dispersions for both the wrist (n = 42) and finger (n = 49) were derived from Mol et al. (Exp Gerontol. 2020; 135: 110938). A Bayesian simulation was constructed whereby the individual arrival times of the PPG wave were calculated by placing a Gaussian prior on the individual QRS-wave timings of each ECG series. The standard deviation (Ļ) of the prior corresponds to the PAT dispersion from Mol et al. This was simulated 10,000 times for each PAT Ļ. The root mean square of successive differences (RMSSD) and standard deviation of NāN intervals (SDNN) were calculated for both HRV and PRV. The Region of Practical Equivalence bounds (ROPE) were set a priori at ± 0.2% of true HRV. The highest density interval (HDI) width, encompassing 95% of the posterior distribution, was calculated for each PAT Ļ. Results The lowest PAT Ļ (2.0 SD) corresponded to 88.4% within ROPE for SDNN and 21.4% for RMSSD. As the Ļ of PAT increases, the equivalence of PRV and HRV decreases for both SDNN and RMSSD. The HDI interval width increases with increasing PAT Ļ, with the HDI width increasing at a higher rate for RMSSD than SDNN. Conclusions For individuals with greater PAT variability, PRV is not a surrogate for HRV. When considering PRV as a unique biometric measure, SDNN may have more favorable measurement properties than RMSSD, though both exhibit a non-uniform measurement error.
- Open Access
- 14 February 2024
- Published byĀ University of Ottawa Library
Athlete Health and Human Performance Will Not Improve Without Transdisciplinary Collaboration and Data Sharing in Elite Sport
NOTE: Editorials at Journal of Athletic Training do not have Abstracts there for this Abstract is only for the Preprint. There are two largely competing models for an athletics organization at both the collegiate and professional levels: the High Performance Model and the Medical Model. The High Performance Model largely arises from international Football perspective that places a āPerformance Directorā at the center of teams supporting the athletes. Medical Model, supported by both the National Athletic Trainers Association and the NCAA, separates off medical staff (athletic trainers and physicians, predominantly) and emphasizes the autonomy of medical decisions. The Medical Model has left athletic trainers in a āmedical siloā, which limits our ability to care for the individual athletes and limits our wider impact in the field of athlete health and injury mitigation. We argue that Medical Model is consistent with the High Performance Model only if we reject the notion that the āPerformance Directorā is an administrative person and instead conceptualize this as a āHealth and Performance Information Hubā which facilitates transdisciplinary collaboration. This Editorial details how a data broker system can be used to facilitate transdisciplinary collaboration within an athletic organization, leading to better healthcare for athletes and improved team and individual performance.
- 14 February 2024
- Published byĀ University of Ottawa Library
Open AccessAthlete Health and Human Performance Will Not Improve Without Transdisciplinary Collaboration and Data Sharing in Elite Sport
NOTE: Editorials at Journal of Athletic Training do not have Abstracts there for this Abstract is only for the Preprint. There are two largely competing models for an athletics organization at both the collegiate and professional levels: the High Performance Model and the Medical Model. The High Performance Model largely arises from international Football perspective that places a āPerformance Directorā at the center of teams supporting the athletes. Medical Model, supported by both the National Athletic Trainers Association and the NCAA, separates off medical staff (athletic trainers and physicians, predominantly) and emphasizes the autonomy of medical decisions. The Medical Model has left athletic trainers in a āmedical siloā, which limits our ability to care for the individual athletes and limits our wider impact in the field of athlete health and injury mitigation. We argue that Medical Model is consistent with the High Performance Model only if we reject the notion that the āPerformance Directorā is an administrative person and instead conceptualize this as a āHealth and Performance Information Hubā which facilitates transdisciplinary collaboration. This Editorial details how a data broker system can be used to facilitate transdisciplinary collaboration within an athletic organization, leading to better healthcare for athletes and improved team and individual performance.
- Open Access
Jet Lag Does Not Impact Football Performance: Using Randomization Inference to Handle Complexity
Introduction: It is commonly accepted that traveling across time zones affects sport performance ( i.e. , via jet lag). This belief is based on poor quality evidence for team sports and simplistic analyses, such as t -tests and linear regression, to explore complex phenomena. For instance, Roy & Forest used such analyses to examine win percentages for the NFL, NBA, and NHL, concluding that East Coast teams were disadvantaged. Similarly, Smith et al. primarily used t -tests to show that West Coast NFL teams were more likely than East Coast teams to beat the Vegas spread in evening games (non-coastal teams were omitted). Neither analysis considered time zone change or game time as continuous constructs nor did they account for important contextual information. We used modern causal inference methods and a decade of collegiate football games to determine if jet lag and kickoff time have any causal effect on beating the Vegas spread. This required fitting nonlinear splines for both data re-weighting and analysis; however, using weights in a generalized additive model (GAM) presents challenges for standard frequentist inferences. Thus, non-parametric simulations were developed to obtain valid causal inferences via randomization inference (RI). Methods: Pro Football Focus data from college football seasons 2013ā2022 were paired with time zone data from Google Maps, weather data from gridMET, and Vegas spread data from collegefootballdata.com . GAM-based propensity scores were calculated from turf type, precipitation, humidity, temperature, and wind speed. These propensity scores orthogonalized the variables relationship to the treatments ( i.e. , game time and hours gained due to time zone change), consistent with the Potential Outcomes framework. The propensity scores were used to weight the observations in a GAM logistic regression, which modeled beating the Vegas spread as a function of a splined effect modification for game time and hours gained in travel. Since valid standard errors cannot be calculated from GAMs with weights, we used RI to compare the effect modification to a null model. We simulated 5,000 datasets of random treatments under the positivity assumption. Each RI dataset was analyzed with the same GAM used for the observed data to obtain a distribution of noise F -statistics. The real data F -statistic was contrasted to the RI distribution for inferences. Results: The real data were compatible with the null hypothesis of no effect for hours lost/gained in travel and game time ( P = 0.142). Conclusion: We need to rigorously interrogate assumptions regarding what affects performance in team sports. There is no clear indication that jet lag and game time affect team performance when appropriate analyses are performed in a causal inference framework. Similarly, rigorous analysis should be undertaken to confirm or refute other assumptions in sport science, such as workload management, sleep practices, and dietary/supplementation regimens.
Open AccessJet Lag Does Not Impact Football Performance: Using Randomization Inference to Handle Complexity
Introduction: It is commonly accepted that traveling across time zones affects sport performance ( i.e. , via jet lag). This belief is based on poor quality evidence for team sports and simplistic analyses, such as t -tests and linear regression, to explore complex phenomena. For instance, Roy & Forest used such analyses to examine win percentages for the NFL, NBA, and NHL, concluding that East Coast teams were disadvantaged. Similarly, Smith et al. primarily used t -tests to show that West Coast NFL teams were more likely than East Coast teams to beat the Vegas spread in evening games (non-coastal teams were omitted). Neither analysis considered time zone change or game time as continuous constructs nor did they account for important contextual information. We used modern causal inference methods and a decade of collegiate football games to determine if jet lag and kickoff time have any causal effect on beating the Vegas spread. This required fitting nonlinear splines for both data re-weighting and analysis; however, using weights in a generalized additive model (GAM) presents challenges for standard frequentist inferences. Thus, non-parametric simulations were developed to obtain valid causal inferences via randomization inference (RI). Methods: Pro Football Focus data from college football seasons 2013ā2022 were paired with time zone data from Google Maps, weather data from gridMET, and Vegas spread data from collegefootballdata.com . GAM-based propensity scores were calculated from turf type, precipitation, humidity, temperature, and wind speed. These propensity scores orthogonalized the variables relationship to the treatments ( i.e. , game time and hours gained due to time zone change), consistent with the Potential Outcomes framework. The propensity scores were used to weight the observations in a GAM logistic regression, which modeled beating the Vegas spread as a function of a splined effect modification for game time and hours gained in travel. Since valid standard errors cannot be calculated from GAMs with weights, we used RI to compare the effect modification to a null model. We simulated 5,000 datasets of random treatments under the positivity assumption. Each RI dataset was analyzed with the same GAM used for the observed data to obtain a distribution of noise F -statistics. The real data F -statistic was contrasted to the RI distribution for inferences. Results: The real data were compatible with the null hypothesis of no effect for hours lost/gained in travel and game time ( P = 0.142). Conclusion: We need to rigorously interrogate assumptions regarding what affects performance in team sports. There is no clear indication that jet lag and game time affect team performance when appropriate analyses are performed in a causal inference framework. Similarly, rigorous analysis should be undertaken to confirm or refute other assumptions in sport science, such as workload management, sleep practices, and dietary/supplementation regimens.
- Open Access
Electrocardiogram and Photoplethysmogram-based Heart Rate Variability Are Not Equivalent: A Bayesian Simulation Analysis
Background: Heart rate variability (HRV) is a common measure of autonomic and cardiovascular system function assessed via electrocardiography (ECG). Consumer wearables, commonly employed in epidemiological research, use photoplethysmography (PPG) to report HRV metrics (PRV), although these may not be equivalent. One potential cause of dissociation between HRV and PRV is the variability in pulse transit time (PTT). This study sought to determine if PPG-derived HRV (i.e., PRV) is equivalent to ECG-derived HRV and ascertain if PRV measurement error is sufficient for a biomarker separate from HRV. Methods: The ECG data from 1,084 subjects were obtained from the PhysioNet Autonomic Aging dataset, and individual PTT variances for both the wrist (n=42) and finger (n=49) were derived from Mol et al. A Bayesian simulation was constructed whereby the individual arrival times of the PPG wave were calculated by placing a Gaussian prior on the individual QRS-wave timings of each ECG series. The standard deviation of the prior corresponds to the PTT variances. This was simulated 10,000 times for each PTT variance. The root mean square of successive differences (RMSSD) and standard deviation of N-N intervals (SDNN) were calculated for both HRV and PRV. The Region of Practical Equivalence bounds (ROPE) were set a priori at ±0.2% of true HRV. The Highest Density Interval (HDI) width, encompassing 95% of the posterior distribution, was calculated for each PTT variance. Results: The lowest PTT variance (2.0 SD) corresponded to 88.4% within ROPE for SDNN and 21.4% for RMSSD. As the SD of PTT increases, the equivalence of PRV and HRV decreases for both SDNN and RMSSD. Thus, between PRV and HRV, RMSSD is nearly never equivalent and SDNN is only somewhat equivalent under very strict circumstances. The HDI interval width increases with increasing PTT variance, with the HDI width increasing at a higher rate for RMSSD than SDNN. Conclusions: For individuals with greater PTT variability, PRV is not a surrogate for HRV. When considering PRV as a unique biometric measure, our findings reveal that SDNN has more favorable measurement properties than RMSSD, though both exhibit a non-uniform measurement error.
Open AccessElectrocardiogram and Photoplethysmogram-based Heart Rate Variability Are Not Equivalent: A Bayesian Simulation Analysis
Background: Heart rate variability (HRV) is a common measure of autonomic and cardiovascular system function assessed via electrocardiography (ECG). Consumer wearables, commonly employed in epidemiological research, use photoplethysmography (PPG) to report HRV metrics (PRV), although these may not be equivalent. One potential cause of dissociation between HRV and PRV is the variability in pulse transit time (PTT). This study sought to determine if PPG-derived HRV (i.e., PRV) is equivalent to ECG-derived HRV and ascertain if PRV measurement error is sufficient for a biomarker separate from HRV. Methods: The ECG data from 1,084 subjects were obtained from the PhysioNet Autonomic Aging dataset, and individual PTT variances for both the wrist (n=42) and finger (n=49) were derived from Mol et al. A Bayesian simulation was constructed whereby the individual arrival times of the PPG wave were calculated by placing a Gaussian prior on the individual QRS-wave timings of each ECG series. The standard deviation of the prior corresponds to the PTT variances. This was simulated 10,000 times for each PTT variance. The root mean square of successive differences (RMSSD) and standard deviation of N-N intervals (SDNN) were calculated for both HRV and PRV. The Region of Practical Equivalence bounds (ROPE) were set a priori at ±0.2% of true HRV. The Highest Density Interval (HDI) width, encompassing 95% of the posterior distribution, was calculated for each PTT variance. Results: The lowest PTT variance (2.0 SD) corresponded to 88.4% within ROPE for SDNN and 21.4% for RMSSD. As the SD of PTT increases, the equivalence of PRV and HRV decreases for both SDNN and RMSSD. Thus, between PRV and HRV, RMSSD is nearly never equivalent and SDNN is only somewhat equivalent under very strict circumstances. The HDI interval width increases with increasing PTT variance, with the HDI width increasing at a higher rate for RMSSD than SDNN. Conclusions: For individuals with greater PTT variability, PRV is not a surrogate for HRV. When considering PRV as a unique biometric measure, our findings reveal that SDNN has more favorable measurement properties than RMSSD, though both exhibit a non-uniform measurement error.