{"title":"Bayesian methods as a complementary tool: balancing innovation and rigor in critical care research","authors":"Sen Lu, Kai Liu, Xin-yun Chen, Jing-chao Luo","doi":"10.1186/s13054-025-05380-0","DOIUrl":null,"url":null,"abstract":"<p>The perspective by Patel and Green, “Death by P-value: The Overreliance on P-values in Critical Care Research” [1], offers a timely critique of rigid statistical thresholds in critical care trials. By advocating for hybrid approaches that integrate Bayesian methods with traditional frequentist analysis, the authors highlight the potential of probabilistic reasoning to uncover clinically meaningful effects obscured by borderline p-values. While their argument is thought-provoking, several considerations warrant further discussion to ensure a balanced application of Bayesian methods in this field.</p><p>The authors rightly emphasize that Bayesian analysis should complement—not replace—frequentist frameworks. Their examples (e.g., hydrocortisone in traumatic brain injury, β-blockers in septic shock) demonstrate how posterior distributions can contextualize findings when p-values are near 0.05. However, while compelling, such re-analyses must not be conflated with definitive evidence. For instance, the reported 87% posterior probability of hydrocortisone reducing ventilator-associated pneumonia (VAP) risk by ≥ 10% remains hypothesis-generating. Bayesian results should be interpreted as one component of a broader evidentiary hierarchy, alongside biological plausibility, trial design, and external validation.</p><p>A key concern in Bayesian analysis is the influence of prior distributions. While the authors employed neutral priors (e.g., mean effect = 0, standard deviation = 10%), even these choices introduce assumptions. Using a standard deviation of 10% in the β-blocker mortality analysis presumes that true effects beyond ± 20% are implausible—a debatable premise in sepsis research. To enhance objectivity, future studies should:</p><ul>\n<li>\n<p>Pre-specify prior distributions in trial protocols, informed by systematic reviews or expert consensus</p>\n</li>\n<li>\n<p>Conduct sensitivity analyses using skeptical priors (e.g., centered on harm) or enthusiastic priors (e.g., larger expected benefits)</p>\n</li>\n<li>\n<p>Adhere to guidelines such as the ISBA bulletin [2] on Bayesian Hypothesis Testing with transparent reporting of prior justification and Bayes factors</p>\n</li>\n</ul><p>The critique of p-values should not overshadow their utility in controlling Type I error rates. More specifically, the continuous versus interrupted chest compressions trial reported a posterior probability of 75% for survival benefit with interrupted compressions. Yet, the frequentist 95% confidence interval (− 1.5 to 0.1%) and corresponding credible interval remind us that the effect could plausibly be null or harmful. Rather than abandoning p-values, a hybrid approach could:</p><ul>\n<li>\n<p>Use Bayesian posterior probabilities to prioritize interventions for further study</p>\n</li>\n<li>\n<p>Reserve frequentist analyses for confirmatory endpoints in pre-registered trials</p>\n</li>\n<li>\n<p>Report both Bayesian and frequentist results in interim and final analyses</p>\n</li>\n</ul><p>Critical care research often faces small sample sizes due to patient heterogeneity and practical constraints. While Bayesian methods can extract insights from limited data, they are not immune to overfitting. In the tracheotomy timing study [1], the 7% absolute reduction in VAP (P = 0.07) corresponds to a wide 95% confidence interval (hazard ratio [HR] 0.42–1.04). A posterior probability of > 75% benefit must be weighed against the frequentist evidence suggesting the true effect spans from a 58% reduction to a 4% increase. Here, Bayesian analysis serves best as a bridge to targeted trials—particularly through adaptive designs identifying subgroups where the signal is strongest.</p><p>To harness the strengths of both paradigms, we propose:</p><ul>\n<li>\n<p>Co-primary endpoints: Pre-specify both Bayesian posterior probabilities and frequentist Type I error thresholds in trial designs</p>\n</li>\n<li>\n<p>Replication standards: Validate Bayesian analyses in independent cohorts before clinical implementation</p>\n</li>\n<li>\n<p>Education initiatives: Train clinicians to interpret both posterior probabilities and power analyses within clinical context</p>\n</li>\n</ul><p>Patel and Green’s perspective rightly challenges the dogma of p < 0.05. However, the solution lies not in discarding p-values but in enriching our analytical toolkit. By combining Bayesian flexibility with frequentist rigor, critical care research can better navigate the tension between statistical precision and clinical urgency. Let us embrace hybrid methods—but with the same scrutiny we demand of traditional approaches.</p><p>No datasets were generated or analysed during the current study.</p><ol data-track-component=\"outbound reference\" data-track-context=\"references section\"><li data-counter=\"1.\"><p>Patel S, Green A. Death by p-value: the overreliance on p-values in critical care research. Crit Care. 2025;29(1):73.</p><p>PubMed PubMed Central Google Scholar </p></li><li data-counter=\"2.\"><p>ISBA Bulletin (2011). The Official Bulletin of the International Society for Bayesian Analysis. 2011. Retrieved from https://bayesian.org/wp-content/uploads/2016/09/1103.pdf</p></li></ol><p>Download references<svg aria-hidden=\"true\" focusable=\"false\" height=\"16\" role=\"img\" width=\"16\"><use xlink:href=\"#icon-eds-i-download-medium\" xmlns:xlink=\"http://www.w3.org/1999/xlink\"></use></svg></p><p>None</p><p>None.</p><span>Author notes</span><ol><li><p>Sen Lu and Kai Liu contributed equally to this article and are co-first authors.</p></li></ol><h3>Authors and Affiliations</h3><ol><li><p>Department of Critical Care Medicine, Sichuan Academy of Medical Sciences and Sichuan Provincial People’s Hospital, University of Electronic Science and Technology of China, Chengdu, China</p><p>Sen Lu & Jing-chao Luo</p></li><li><p>Department of Critical Care Medicine, Zhongshan Hospital, Fudan University, Shanghai, China</p><p>Kai Liu</p></li><li><p>School of Data Science, Chinese University of Hong Kong, Shenzhen, China</p><p>Xin-yun Chen</p></li></ol><span>Authors</span><ol><li><span>Sen Lu</span>View author publications<p><span>You can also search for this author in</span><span>PubMed<span> </span>Google Scholar</span></p></li><li><span>Kai Liu</span>View author publications<p><span>You can also search for this author in</span><span>PubMed<span> </span>Google Scholar</span></p></li><li><span>Xin-yun Chen</span>View author publications<p><span>You can also search for this author in</span><span>PubMed<span> </span>Google Scholar</span></p></li><li><span>Jing-chao Luo</span>View author publications<p><span>You can also search for this author in</span><span>PubMed<span> </span>Google Scholar</span></p></li></ol><h3>Contributions</h3><p>SL, KL, JL and XC wrote and revised the manuscript.</p><h3>Corresponding authors</h3><p>Correspondence to Xin-yun Chen or Jing-chao Luo.</p><h3>Ethical approval and consent to participate</h3>\n<p>Not applicable.</p>\n<h3>Consent for publication</h3>\n<p>Not applicable.</p>\n<h3>Competing interests</h3>\n<p>The authors declare no competing interests.</p><h3>Publisher's Note</h3><p>Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.</p><p><b>Open Access</b> This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.</p>\n<p>Reprints and permissions</p><img alt=\"Check for updates. Verify currency and authenticity via CrossMark\" height=\"81\" loading=\"lazy\" src=\"data:image/svg+xml;base64,<svg height="81" width="57" xmlns="http://www.w3.org/2000/svg"><g fill="none" fill-rule="evenodd"><path d="m17.35 35.45 21.3-14.2v-17.03h-21.3" fill="#989898"/><path d="m38.65 35.45-21.3-14.2v-17.03h21.3" fill="#747474"/><path d="m28 .5c-12.98 0-23.5 10.52-23.5 23.5s10.52 23.5 23.5 23.5 23.5-10.52 23.5-23.5c0-6.23-2.48-12.21-6.88-16.62-4.41-4.4-10.39-6.88-16.62-6.88zm0 41.25c-9.8 0-17.75-7.95-17.75-17.75s7.95-17.75 17.75-17.75 17.75 7.95 17.75 17.75c0 4.71-1.87 9.22-5.2 12.55s-7.84 5.2-12.55 5.2z" fill="#535353"/><path d="m41 36c-5.81 6.23-15.23 7.45-22.43 2.9-7.21-4.55-10.16-13.57-7.03-21.5l-4.92-3.11c-4.95 10.7-1.19 23.42 8.78 29.71 9.97 6.3 23.07 4.22 30.6-4.86z" fill="#9c9c9c"/><path d="m.2 58.45c0-.75.11-1.42.33-2.01s.52-1.09.91-1.5c.38-.41.83-.73 1.34-.94.51-.22 1.06-.32 1.65-.32.56 0 1.06.11 1.51.35.44.23.81.5 1.1.81l-.91 1.01c-.24-.24-.49-.42-.75-.56-.27-.13-.58-.2-.93-.2-.39 0-.73.08-1.05.23-.31.16-.58.37-.81.66-.23.28-.41.63-.53 1.04-.13.41-.19.88-.19 1.39 0 1.04.23 1.86.68 2.46.45.59 1.06.88 1.84.88.41 0 .77-.07 1.07-.23s.59-.39.85-.68l.91 1c-.38.43-.8.76-1.28.99-.47.22-1 .34-1.58.34-.59 0-1.13-.1-1.64-.31-.5-.2-.94-.51-1.31-.91-.38-.4-.67-.9-.88-1.48-.22-.59-.33-1.26-.33-2.02zm8.4-5.33h1.61v2.54l-.05 1.33c.29-.27.61-.51.96-.72s.76-.31 1.24-.31c.73 0 1.27.23 1.61.71.33.47.5 1.14.5 2.02v4.31h-1.61v-4.1c0-.57-.08-.97-.25-1.21-.17-.23-.45-.35-.83-.35-.3 0-.56.08-.79.22-.23.15-.49.36-.78.64v4.8h-1.61zm7.37 6.45c0-.56.09-1.06.26-1.51.18-.45.42-.83.71-1.14.29-.3.63-.54 1.01-.71.39-.17.78-.25 1.18-.25.47 0 .88.08 1.23.24.36.16.65.38.89.67s.42.63.54 1.03c.12.41.18.84.18 1.32 0 .32-.02.57-.07.76h-4.36c.07.62.29 1.1.65 1.44.36.33.82.5 1.38.5.29 0 .57-.04.83-.13s.51-.21.76-.37l.55 1.01c-.33.21-.69.39-1.09.53-.41.14-.83.21-1.26.21-.48 0-.92-.08-1.34-.25-.41-.16-.76-.4-1.07-.7-.31-.31-.55-.69-.72-1.13-.18-.44-.26-.95-.26-1.52zm4.6-.62c0-.55-.11-.98-.34-1.28-.23-.31-.58-.47-1.06-.47-.41 0-.77.15-1.07.45-.31.29-.5.73-.58 1.3zm2.5.62c0-.57.09-1.08.28-1.53.18-.44.43-.82.75-1.13s.69-.54 1.1-.71c.42-.16.85-.24 1.31-.24.45 0 .84.08 1.17.23s.61.34.85.57l-.77 1.02c-.19-.16-.38-.28-.56-.37-.19-.09-.39-.14-.61-.14-.56 0-1.01.21-1.35.63-.35.41-.52.97-.52 1.67 0 .69.17 1.24.51 1.66.34.41.78.62 1.32.62.28 0 .54-.06.78-.17.24-.12.45-.26.64-.42l.67 1.03c-.33.29-.69.51-1.08.65-.39.15-.78.23-1.18.23-.46 0-.9-.08-1.31-.24-.4-.16-.75-.39-1.05-.7s-.53-.69-.7-1.13c-.17-.45-.25-.96-.25-1.53zm6.91-6.45h1.58v6.17h.05l2.54-3.16h1.77l-2.35 2.8 2.59 4.07h-1.75l-1.77-2.98-1.08 1.23v1.75h-1.58zm13.69 1.27c-.25-.11-.5-.17-.75-.17-.58 0-.87.39-.87 1.16v.75h1.34v1.27h-1.34v5.6h-1.61v-5.6h-.92v-1.2l.92-.07v-.72c0-.35.04-.68.13-.98.08-.31.21-.57.4-.79s.42-.39.71-.51c.28-.12.63-.18 1.04-.18.24 0 .48.02.69.07.22.05.41.1.57.17zm.48 5.18c0-.57.09-1.08.27-1.53.17-.44.41-.82.72-1.13.3-.31.65-.54 1.04-.71.39-.16.8-.24 1.23-.24s.84.08 1.24.24c.4.17.74.4 1.04.71s.54.69.72 1.13c.19.45.28.96.28 1.53s-.09 1.08-.28 1.53c-.18.44-.42.82-.72 1.13s-.64.54-1.04.7-.81.24-1.24.24-.84-.08-1.23-.24-.74-.39-1.04-.7c-.31-.31-.55-.69-.72-1.13-.18-.45-.27-.96-.27-1.53zm1.65 0c0 .69.14 1.24.43 1.66.28.41.68.62 1.18.62.51 0 .9-.21 1.19-.62.29-.42.44-.97.44-1.66 0-.7-.15-1.26-.44-1.67-.29-.42-.68-.63-1.19-.63-.5 0-.9.21-1.18.63-.29.41-.43.97-.43 1.67zm6.48-3.44h1.33l.12 1.21h.05c.24-.44.54-.79.88-1.02.35-.24.7-.36 1.07-.36.32 0 .59.05.78.14l-.28 1.4-.33-.09c-.11-.01-.23-.02-.38-.02-.27 0-.56.1-.86.31s-.55.58-.77 1.1v4.2h-1.61zm-47.87 15h1.61v4.1c0 .57.08.97.25 1.2.17.24.44.35.81.35.3 0 .57-.07.8-.22.22-.15.47-.39.73-.73v-4.7h1.61v6.87h-1.32l-.12-1.01h-.04c-.3.36-.63.64-.98.86-.35.21-.76.32-1.24.32-.73 0-1.27-.24-1.61-.71-.33-.47-.5-1.14-.5-2.02zm9.46 7.43v2.16h-1.61v-9.59h1.33l.12.72h.05c.29-.24.61-.45.97-.63.35-.17.72-.26 1.1-.26.43 0 .81.08 1.15.24.33.17.61.4.84.71.24.31.41.68.53 1.11.13.42.19.91.19 1.44 0 .59-.09 1.11-.25 1.57-.16.47-.38.85-.65 1.16-.27.32-.58.56-.94.73-.35.16-.72.25-1.1.25-.3 0-.6-.07-.9-.2s-.59-.31-.87-.56zm0-2.3c.26.22.5.37.73.45.24.09.46.13.66.13.46 0 .84-.2 1.15-.6.31-.39.46-.98.46-1.77 0-.69-.12-1.22-.35-1.61-.23-.38-.61-.57-1.13-.57-.49 0-.99.26-1.52.77zm5.87-1.69c0-.56.08-1.06.25-1.51.16-.45.37-.83.65-1.14.27-.3.58-.54.93-.71s.71-.25 1.08-.25c.39 0 .73.07 1 .2.27.14.54.32.81.55l-.06-1.1v-2.49h1.61v9.88h-1.33l-.11-.74h-.06c-.25.25-.54.46-.88.64-.33.18-.69.27-1.06.27-.87 0-1.56-.32-2.07-.95s-.76-1.51-.76-2.65zm1.67-.01c0 .74.13 1.31.4 1.7.26.38.65.58 1.15.58.51 0 .99-.26 1.44-.77v-3.21c-.24-.21-.48-.36-.7-.45-.23-.08-.46-.12-.7-.12-.45 0-.82.19-1.13.59-.31.39-.46.95-.46 1.68zm6.35 1.59c0-.73.32-1.3.97-1.71.64-.4 1.67-.68 3.08-.84 0-.17-.02-.34-.07-.51-.05-.16-.12-.3-.22-.43s-.22-.22-.38-.3c-.15-.06-.34-.1-.58-.1-.34 0-.68.07-1 .2s-.63.29-.93.47l-.59-1.08c.39-.24.81-.45 1.28-.63.47-.17.99-.26 1.54-.26.86 0 1.51.25 1.93.76s.63 1.25.63 2.21v4.07h-1.32l-.12-.76h-.05c-.3.27-.63.48-.98.66s-.73.27-1.14.27c-.61 0-1.1-.19-1.48-.56-.38-.36-.57-.85-.57-1.46zm1.57-.12c0 .3.09.53.27.67.19.14.42.21.71.21.28 0 .54-.07.77-.2s.48-.31.73-.56v-1.54c-.47.06-.86.13-1.18.23-.31.09-.57.19-.76.31s-.33.25-.41.4c-.09.15-.13.31-.13.48zm6.29-3.63h-.98v-1.2l1.06-.07.2-1.88h1.34v1.88h1.75v1.27h-1.75v3.28c0 .8.32 1.2.97 1.2.12 0 .24-.01.37-.04.12-.03.24-.07.34-.11l.28 1.19c-.19.06-.4.12-.64.17-.23.05-.49.08-.76.08-.4 0-.74-.06-1.02-.18-.27-.13-.49-.3-.67-.52-.17-.21-.3-.48-.37-.78-.08-.3-.12-.64-.12-1.01zm4.36 2.17c0-.56.09-1.06.27-1.51s.41-.83.71-1.14c.29-.3.63-.54 1.01-.71.39-.17.78-.25 1.18-.25.47 0 .88.08 1.23.24.36.16.65.38.89.67s.42.63.54 1.03c.12.41.18.84.18 1.32 0 .32-.02.57-.07.76h-4.37c.08.62.29 1.1.65 1.44.36.33.82.5 1.38.5.3 0 .58-.04.84-.13.25-.09.51-.21.76-.37l.54 1.01c-.32.21-.69.39-1.09.53s-.82.21-1.26.21c-.47 0-.92-.08-1.33-.25-.41-.16-.77-.4-1.08-.7-.3-.31-.54-.69-.72-1.13-.17-.44-.26-.95-.26-1.52zm4.61-.62c0-.55-.11-.98-.34-1.28-.23-.31-.58-.47-1.06-.47-.41 0-.77.15-1.08.45-.31.29-.5.73-.57 1.3zm3.01 2.23c.31.24.61.43.92.57.3.13.63.2.98.2.38 0 .65-.08.83-.23s.27-.35.27-.6c0-.14-.05-.26-.13-.37-.08-.1-.2-.2-.34-.28-.14-.09-.29-.16-.47-.23l-.53-.22c-.23-.09-.46-.18-.69-.3-.23-.11-.44-.24-.62-.4s-.33-.35-.45-.55c-.12-.21-.18-.46-.18-.75 0-.61.23-1.1.68-1.49.44-.38 1.06-.57 1.83-.57.48 0 .91.08 1.29.25s.71.36.99.57l-.74.98c-.24-.17-.49-.32-.73-.42-.25-.11-.51-.16-.78-.16-.35 0-.6.07-.76.21-.17.15-.25.33-.25.54 0 .14.04.26.12.36s.18.18.31.26c.14.07.29.14.46.21l.54.19c.23.09.47.18.7.29s.44.24.64.4c.19.16.34.35.46.58.11.23.17.5.17.82 0 .3-.06.58-.17.83-.12.26-.29.48-.51.68-.23.19-.51.34-.84.45-.34.11-.72.17-1.15.17-.48 0-.95-.09-1.41-.27-.46-.19-.86-.41-1.2-.68z" fill="#535353"/></g></svg>\" width=\"57\"/><h3>Cite this article</h3><p>Lu, S., Liu, K., Chen, Xy. <i>et al.</i> Bayesian methods as a complementary tool: balancing innovation and rigor in critical care research. <i>Crit Care</i> <b>29</b>, 135 (2025). https://doi.org/10.1186/s13054-025-05380-0</p><p>Download citation<svg aria-hidden=\"true\" focusable=\"false\" height=\"16\" role=\"img\" width=\"16\"><use xlink:href=\"#icon-eds-i-download-medium\" xmlns:xlink=\"http://www.w3.org/1999/xlink\"></use></svg></p><ul data-test=\"publication-history\"><li><p>Received<span>: </span><span><time datetime=\"2025-03-18\">18 March 2025</time></span></p></li><li><p>Accepted<span>: </span><span><time datetime=\"2025-03-19\">19 March 2025</time></span></p></li><li><p>Published<span>: </span><span><time datetime=\"2025-03-25\">25 March 2025</time></span></p></li><li><p>DOI</abbr><span>: </span><span>https://doi.org/10.1186/s13054-025-05380-0</span></p></li></ul><h3>Share this article</h3><p>Anyone you share the following link with will be able to read this content:</p><button data-track=\"click\" data-track-action=\"get shareable link\" data-track-external=\"\" data-track-label=\"button\" type=\"button\">Get shareable link</button><p>Sorry, a shareable link is not currently available for this article.</p><p data-track=\"click\" data-track-action=\"select share url\" data-track-label=\"button\"></p><button data-track=\"click\" data-track-action=\"copy share url\" data-track-external=\"\" data-track-label=\"button\" type=\"button\">Copy to clipboard</button><p> Provided by the Springer Nature SharedIt content-sharing initiative </p>","PeriodicalId":10811,"journal":{"name":"Critical Care","volume":"41 1","pages":""},"PeriodicalIF":8.8000,"publicationDate":"2025-03-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Critical Care","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1186/s13054-025-05380-0","RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"CRITICAL CARE MEDICINE","Score":null,"Total":0}
引用次数: 0
Abstract
The perspective by Patel and Green, “Death by P-value: The Overreliance on P-values in Critical Care Research” [1], offers a timely critique of rigid statistical thresholds in critical care trials. By advocating for hybrid approaches that integrate Bayesian methods with traditional frequentist analysis, the authors highlight the potential of probabilistic reasoning to uncover clinically meaningful effects obscured by borderline p-values. While their argument is thought-provoking, several considerations warrant further discussion to ensure a balanced application of Bayesian methods in this field.
The authors rightly emphasize that Bayesian analysis should complement—not replace—frequentist frameworks. Their examples (e.g., hydrocortisone in traumatic brain injury, β-blockers in septic shock) demonstrate how posterior distributions can contextualize findings when p-values are near 0.05. However, while compelling, such re-analyses must not be conflated with definitive evidence. For instance, the reported 87% posterior probability of hydrocortisone reducing ventilator-associated pneumonia (VAP) risk by ≥ 10% remains hypothesis-generating. Bayesian results should be interpreted as one component of a broader evidentiary hierarchy, alongside biological plausibility, trial design, and external validation.
A key concern in Bayesian analysis is the influence of prior distributions. While the authors employed neutral priors (e.g., mean effect = 0, standard deviation = 10%), even these choices introduce assumptions. Using a standard deviation of 10% in the β-blocker mortality analysis presumes that true effects beyond ± 20% are implausible—a debatable premise in sepsis research. To enhance objectivity, future studies should:
Pre-specify prior distributions in trial protocols, informed by systematic reviews or expert consensus
Conduct sensitivity analyses using skeptical priors (e.g., centered on harm) or enthusiastic priors (e.g., larger expected benefits)
Adhere to guidelines such as the ISBA bulletin [2] on Bayesian Hypothesis Testing with transparent reporting of prior justification and Bayes factors
The critique of p-values should not overshadow their utility in controlling Type I error rates. More specifically, the continuous versus interrupted chest compressions trial reported a posterior probability of 75% for survival benefit with interrupted compressions. Yet, the frequentist 95% confidence interval (− 1.5 to 0.1%) and corresponding credible interval remind us that the effect could plausibly be null or harmful. Rather than abandoning p-values, a hybrid approach could:
Use Bayesian posterior probabilities to prioritize interventions for further study
Reserve frequentist analyses for confirmatory endpoints in pre-registered trials
Report both Bayesian and frequentist results in interim and final analyses
Critical care research often faces small sample sizes due to patient heterogeneity and practical constraints. While Bayesian methods can extract insights from limited data, they are not immune to overfitting. In the tracheotomy timing study [1], the 7% absolute reduction in VAP (P = 0.07) corresponds to a wide 95% confidence interval (hazard ratio [HR] 0.42–1.04). A posterior probability of > 75% benefit must be weighed against the frequentist evidence suggesting the true effect spans from a 58% reduction to a 4% increase. Here, Bayesian analysis serves best as a bridge to targeted trials—particularly through adaptive designs identifying subgroups where the signal is strongest.
To harness the strengths of both paradigms, we propose:
Co-primary endpoints: Pre-specify both Bayesian posterior probabilities and frequentist Type I error thresholds in trial designs
Replication standards: Validate Bayesian analyses in independent cohorts before clinical implementation
Education initiatives: Train clinicians to interpret both posterior probabilities and power analyses within clinical context
Patel and Green’s perspective rightly challenges the dogma of p < 0.05. However, the solution lies not in discarding p-values but in enriching our analytical toolkit. By combining Bayesian flexibility with frequentist rigor, critical care research can better navigate the tension between statistical precision and clinical urgency. Let us embrace hybrid methods—but with the same scrutiny we demand of traditional approaches.
No datasets were generated or analysed during the current study.
Patel S, Green A. Death by p-value: the overreliance on p-values in critical care research. Crit Care. 2025;29(1):73.
PubMed PubMed Central Google Scholar
ISBA Bulletin (2011). The Official Bulletin of the International Society for Bayesian Analysis. 2011. Retrieved from https://bayesian.org/wp-content/uploads/2016/09/1103.pdf
Download references
None
None.
Author notes
Sen Lu and Kai Liu contributed equally to this article and are co-first authors.
Authors and Affiliations
Department of Critical Care Medicine, Sichuan Academy of Medical Sciences and Sichuan Provincial People’s Hospital, University of Electronic Science and Technology of China, Chengdu, China
Sen Lu & Jing-chao Luo
Department of Critical Care Medicine, Zhongshan Hospital, Fudan University, Shanghai, China
Kai Liu
School of Data Science, Chinese University of Hong Kong, Shenzhen, China
Xin-yun Chen
Authors
Sen LuView author publications
You can also search for this author inPubMedGoogle Scholar
Kai LiuView author publications
You can also search for this author inPubMedGoogle Scholar
Xin-yun ChenView author publications
You can also search for this author inPubMedGoogle Scholar
Jing-chao LuoView author publications
You can also search for this author inPubMedGoogle Scholar
Contributions
SL, KL, JL and XC wrote and revised the manuscript.
Corresponding authors
Correspondence to Xin-yun Chen or Jing-chao Luo.
Ethical approval and consent to participate
Not applicable.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissions
Cite this article
Lu, S., Liu, K., Chen, Xy. et al. Bayesian methods as a complementary tool: balancing innovation and rigor in critical care research. Crit Care29, 135 (2025). https://doi.org/10.1186/s13054-025-05380-0
Download citation
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s13054-025-05380-0
Share this article
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
作者及工作单位四川省医学科学院重症医学科、四川省人民医院、电子科技大学,成都罗景超复旦大学附属中山医院重症医学科,上海,中国刘凯香港中文大学数据科学学院,深圳、中国 深圳香港中文大学数据科学学院陈新云作者卢森查看作者发表的论文您也可以在PubMed Google Scholar中搜索该作者刘凯查看作者发表的论文您也可以在PubMed Google Scholar中搜索该作者陈新云查看作者发表的论文您也可以在PubMed Google Scholar中搜索该作者罗景超查看作者发表的论文您也可以在PubMed Google Scholar中搜索该作者供稿SL、KL、JL和XC撰写并修改了手稿。通讯作者请与陈新云或罗景超通信。伦理批准和参与同意书不适用。出版同意书不适用。利益冲突作者声明无利益冲突。出版商注释Springer Nature对已出版地图中的管辖权主张和机构隶属关系保持中立。开放获取本文采用知识共享署名-非商业性-禁止衍生 4.0 国际许可协议进行许可,该协议允许以任何媒介或格式进行非商业性使用、共享、分发和复制,只要您适当注明原作者和来源,提供知识共享许可协议的链接,并说明是否修改了许可材料。根据本许可协议,您无权分享源自本文或本文部分内容的改编材料。本文中的图片或其他第三方材料均包含在文章的知识共享许可协议中,除非在材料的信用栏中另有说明。如果材料未包含在文章的知识共享许可协议中,且您打算使用的材料不符合法律规定或超出了许可使用范围,则您需要直接获得版权所有者的许可。如需查看该许可的副本,请访问 http://creativecommons.org/licenses/by-nc-nd/4.0/.Reprints and permissionsCite this articleLu, S., Liu, K., Chen, Xy. et al. Bayesian methods as a complementary tool: balancing innovation and rigor in critical care research.Crit Care 29, 135 (2025). https://doi.org/10.1186/s13054-025-05380-0Download citationReceived:18 March 2025Accepted:19 March 2025Published: 25 March 2025DOI: https://doi.org/10.1186/s13054-025-05380-0Share this articleAnyone you share the following link with will be able to read this content:Get shareable linkSorry, a shareable link is not currently available for this article.Copy to clipboard Provided by the Springer Nature SharedIt content-sharing initiative.
期刊介绍:
Critical Care is an esteemed international medical journal that undergoes a rigorous peer-review process to maintain its high quality standards. Its primary objective is to enhance the healthcare services offered to critically ill patients. To achieve this, the journal focuses on gathering, exchanging, disseminating, and endorsing evidence-based information that is highly relevant to intensivists. By doing so, Critical Care seeks to provide a thorough and inclusive examination of the intensive care field.