Published on in Vol 26 (2024)

Preprints (earlier versions) of this paper are available at https://preprints.jmir.org/preprint/57422, first published .
Authors’ Reply: Ambiguity in Statistical Analysis Methods and Nonconformity With Prespecified Commitment to Data Sharing in a Cluster Randomized Controlled Trial

Authors’ Reply: Ambiguity in Statistical Analysis Methods and Nonconformity With Prespecified Commitment to Data Sharing in a Cluster Randomized Controlled Trial

Authors’ Reply: Ambiguity in Statistical Analysis Methods and Nonconformity With Prespecified Commitment to Data Sharing in a Cluster Randomized Controlled Trial

Authors of this article:

Diego Arguello1 Author Orcid Image

Letter to the Editor

Human Performance and Exercise Science Lab, Department of Health Sciences, Northeastern University, Boston, MA, United States

Corresponding Author:

Diego Arguello, BA, MSc, PhD

Human Performance and Exercise Science Lab

Department of Health Sciences

Northeastern University

360 Huntington Avenue

Boston, MA, 02115

United States

Phone: 1 6173734427

Email: arguello.d@Northeastern.edu



We appreciate the thoughtful commentary [1] on our study [2] and address the raised concerns.

One pertains to perceived ambiguity in reporting statistical methodology, presumably prompting inquiry into whether analyses accounted for clustering and nesting. We clarify that our choice of a cluster randomized controlled trial design was driven by practical implications (eg, space modification and experimental contamination)—valid reasons for the study design. While this determined the study design, we were interested in the participant-average treatment effect. We feel this is affirmed in our aims and note the perceived ambiguity and level of detail requested by the letter writers were not raised during multiple reviews, and this response to the letter further clarifies our analytical approach.

We agree that type I error inflation is a concern if clustering is not addressed in the study design phase or during analyses. We also recognize that this is an active area of study, and work is ongoing to determine optimal approaches for small-sample study designs [3]. Importantly, the current recommendation is to develop the analytical plan based on goals of the analyses (ie, exploratory outcomes in this study) and various study-specific characteristics (selected examples for our study: random cluster size independent of other data and wide physical distribution of within-cluster participants) [3]. Due to such study-specific considerations, clusters were deemed to be noninformative [4], and our approach—“random-intercept mixed linear models that accounted for repeated measures and clustering effects” [2]—included a random effect for clusters to model any potential correlational structure and interparticipant dependency within clusters [5]. Additionally, we used the Kenward-Rogers method to preserve nominal type I error. The method adjusts for df to account for hierarchical complexity of data, including potential nesting and variable or small cluster sizes [3]. We also acknowledged uneven cluster size as a study limitation [2].

Second, while we clearly conveyed the exploratory nature of the analyses aimed at developing hypotheses, we were also conservative by avoiding confirmatory conclusions based on type I error rates [2]. While it is known that conservative analyses may be counterproductive for exploration [6], our careful approach—relying on sound, disparate methods sufficiently accounting for any potential within-cluster dependence and variable cluster size, along with a cautious interpretation strategy—was appropriate for the study objective.

Regarding data sharing, we have previously shared other data and received data from the community to advance multiple areas of inquiry. However, we primarily rejected this request and a competing industry request, given ongoing small business and financial interests leveraging this and the associated body of work; we communicated this to the journal (August 2023) after the request in question was made. Such considerations are necessary to avoid setting a precedent and in the context of any potential for these interests to be compromised. An example of such considerations includes federal funding agencies supporting small business innovation research, allowing awardees to withhold related data to protect endeavors similar to ours. Implying that findings are untrustworthy due to such considerations would incorrectly render a substantial body of such work as the same.

Conflicts of Interest

Since June 2023, DA has received financial compensation for consultations leveraging the published work.

  1. Jamshidi-Naeini Y, Golzarri-Arroyo L, Thapa DK, Brown AW, Kpormegbey DE, Allison DB. Ambiguity in statistical analysis methods and nonconformity with prespecified commitment to data sharing in a cluster randomized controlled trial. J Med Internet Res. 2024.:e54090. [CrossRef]
  2. Arguello D, Cloutier G, Thorndike AN, Castaneda Sceppa C, Griffith J, John D. Impact of sit-to-stand and treadmill desks on patterns of daily waking physical behaviors among overweight and obese seated office workers: cluster randomized controlled trial. J Med Internet Res. May 16, 2023;25:e43018. [FREE Full text] [CrossRef] [Medline]
  3. Leyrat C, Morgan KE, Leurent B, Kahan BC. Cluster randomized trials with a small number of clusters: which analyses should be used? Int J Epidemiol. Feb 1, 2018;47(1):321-331. [CrossRef] [Medline]
  4. Wang B, Harhay MO, Tong J, Small DS, Morris TP, Li F. On the mixed-model analysis of covariance in cluster-randomized trials. arXiv. [FREE Full text] [CrossRef]
  5. Donner A, Klar N. Design and Analysis of Cluster Randomization Trials in Health Research. London, England. Arnold Publishers; 2000.
  6. Kimmelman J, Mogil JS, Dirnagl U. Distinguishing between exploratory and confirmatory preclinical research will improve translation. PLoS Biol. May 20, 2014;12(5):e1001863. [FREE Full text] [CrossRef] [Medline]

Edited by T Leung; This is a non–peer-reviewed article. submitted 16.02.24; accepted 22.02.24; published 03.04.24.

Copyright

©Diego Arguello. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 03.04.2024.

This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.