Agreement: I Agree Body: Dear Editor, We have read the article by Jordan and colleagues [1] with great interest, as well as the subsequent correspondence and the authors’ responses. While we appreciate the authors’ rigorous methodology and the practical value of the BLISS score, we wish to raise two methodological points that have not been addressed in the existing exchange and that, if clarified, would further enhance the transparency and clinical utility of the score. First, the paper reports in Table 5 the proportion of the development cohort (BLISS, n=1894) classified as high risk at various BLISS score thresholds, along with the corresponding proportion of respiratory admissions captured (e.g., a 10% risk threshold identifies 51% of the population and 81% of admissions). Such information is highly valuable for clinicians and commissioners to understand the trade off between intervention reach and event coverage. However, no analogous analysis is provided for either of the external validation cohorts (ECLIPSE or CPRD). While decision curve analysis demonstrates net benefit, it does not directly inform users about the proportion of patients who would require intervention at a given threshold or the proportion of events that would be captured in a different setting. Explicit reporting of threshold specific performance in external validation cohorts would help to assess whether the risk distribution and event capture generalise beyond the derivation setting. Such analyses would therefore contribute to a more complete understanding of the model’s transportability. Second, the authors state in the abstract that “The BLISS score was directly compared with the Bertens’ score in the ECLIPSE cohort.” However, in that comparison, the authors report only the C statistic and calibration slope. Net reclassification improvement (NRI) and integrated discrimination improvement (IDI) are not presented. According to the TRIPOD explanation and elaboration document, NRI and IDI can provide complementary information beyond the C statistic by quantifying the extent to which a new model correctly reclassifies individuals compared with an existing model, even for non nested models [2, 3]. Reporting these metrics would therefore offer additional insight into the relative performance of the BLISS score versus the Bertens’ score, beyond what is captured by the difference in concordance statistics. We would be grateful if the authors could provide the additional analyses (threshold specific performance in the external validation cohorts, and NRI/IDI for the comparison with the Bertens’ score) or comment on their feasibility. References: 1. Jordan RE, Keene SJ, Franssen FME, et al. Prognostic score for predicting respiratory admissions among patients with chronic obstructive pulmonary disease in primary care: development and validation in population cohorts (Birmingham Lung Improvement Studies (BLISS)). Bmj 2026;392:e084521. doi: 10.1136/bmj-2025-084521 [published Online First: 20260305] 2. Moons KG, Altman DG, Reitsma JB, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med 2015;162(1):W1-73. doi: 10.7326/M14-0698 [published Online First: 2015/01/07] 3. Pencina MJ, D'Agostino RB, Sr., D'Agostino RB, Jr., et al. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27(2):157-72; discussion 207-12. doi: 10.1002/sim.2929 No competing Interests: Yes The following competing Interests: Electronic Publication Date: Sunday, April 12, 2026 - 02:39 AI use: Yes I have used AI Highwire Comment Subject: Prognostic score for predicting respiratory admissions among patients with chronic obstructive pulmonary disease in primary care: development and validation in population cohorts (Birmingham Lung Improvement Studies (BLISS)) AI use details: Language polishing Workflow State: Released Full Title: Risk threshold reporting across cohorts and the use of reclassification metrics in the BLISS prognostic score Highwire Comment Response to: Prognostic score for predicting respiratory admissions among patients with chronic obstructive pulmonary disease in primary care: development and validation in population cohorts (Birmingham Lung Improvement Studies (BLISS)) Check this box if you would like your letter to appear anonymously:: Last Name: Zhang First name and middle initial: Chuanfu Email: 274703992@qq.com Address: No. 358 Datong Road, Gaoqiao Town, Pudong New Area, Shanghai, China Occupation: Physician Other Authors: Qianqian Liang; Yubo Shao Affiliation: Seventh People's Hospital, Shanghai University of Traditional Chinese Medicine, Shanghai, China BMJ: Additional Article Info: Rapid response
Risk threshold reporting across cohorts and the use of reclassification metrics in the BLISS prognostic score
