We read with interest the Science Letter by Dubowitz et al. [1]. The application of biological instead of chronological age for the prediction of postoperative complications in patients is a promising advance, and we concur with the authors' perspective on the potential association between biological ageing and postoperative complications after cancer surgery. This study helps prompt surgeons and anaesthetists to reflect on their practice and identify high-risk patients, regardless of the method used.

However, in statistics, correlation is not a substitute for causal inference. Also, exclusive reliance on the PhenoAge model as a means of calculating biological age has considerable limitations. This model uses conventional clinical biomarkers (e.g. white blood cell count, blood glucose, lipids, creatinine, etc.) to calculate biological age [2]. In our opinion, these markers reflect primarily the macroscopic health of the body, but do not fully capture the ageing process at the cellular or molecular level. They are also susceptible to external factors (e.g. diet, lifestyle, environmental pollution, etc.) and, thus, may not reflect an individual's biological ageing accurately. In addition, PhenoAge modelling is based on data from specific populations, usually in the USA. In other races and regions, the performance of this model may not be the same. Therefore, when the PhenoAge model is applied in different races or regions, it may need to be re-based on a large sample size of data for optimisation or training.

Most importantly, the PhenoAge model simplifies the ageing process, which is a multifactorial and complex process involving interactions at multiple levels, such as gene expression; telomere shortening; immune system decline; and epigenetic changes [3, 4]. The model focuses mainly on a few clinical markers, but it fails to account for all the biological processes involved in the ageing process and, therefore, may overlook some subtle key roles of age-related molecular targets. Indeed, one of the primary challenges in predicting the incidence of postoperative complications in older patients accurately is the absence of dedicated peri-operative databases for this specific population. Such databases are essential for researchers to obtain sufficient, high-quality raw data.

In accordance with this objective, our research group is currently engaged in the establishment of the Perioperative Management and Outcome database, which is a large-scale, nationwide registry of older patients in China [5]. The prospective registry is expected to provide a rich dataset that will facilitate the evaluation of the quality of peri-operative care and, in turn, improve clinical care for older patients. Nevertheless, it has proven challenging to construct a database that encompasses multiple countries or regions and is based on participants from a diverse range of populations. It is anticipated that, in the future, multi-country and multi-regional collaborative research will become standard in this field of research.

我们饶有兴趣地阅读了 Dubowitz 等人的《科学快报》[1]。应用生物年龄而不是实际年龄来预测患者术后并发症是一项有前途的进步，我们同意作者关于生物衰老与癌症手术后术后并发症之间潜在关联的观点。这项研究有助于促使外科医生和麻醉师反思他们的实践并识别高危患者，无论使用何种方法。

然而，在统计学中，相关性并不能替代因果推理。此外，完全依赖 PhenoAge 模型作为计算生物年龄的手段具有相当大的局限性。该模型使用常规临床生物标志物（例如白细胞计数、血糖、血脂、肌酐等）来计算生物年龄 [2]。在我们看来，这些标志物主要反映了身体的宏观健康状况，但并不能完全捕捉细胞或分子水平的衰老过程。他们亦容易受到外在因素（例如饮食、生活方式、环境污染等）的影响，因此未必能准确反映个人的生物衰老情况。此外，PhenoAge 建模基于来自特定人群的数据，通常在美国。在其他种族和地区，此模型的性能可能不同。因此，当 PhenoAge 模型应用于不同的种族或地区时，可能需要根据大样本数据重新进行优化或训练。

最重要的是，PhenoAge 模型简化了衰老过程，这是一个多因素和复杂的过程，涉及多个层面的相互作用，例如基因表达;端粒缩短;免疫系统下降;和表观遗传变化 [3， 4]。该模型主要关注少数临床标志物，但它未能解释衰老过程中涉及的所有生物过程，因此可能忽略了与年龄相关的分子靶标的一些微妙的关键作用。事实上，准确预测老年患者术后并发症发生率的主要挑战之一是缺乏针对该特定人群的专用围手术期数据库。此类数据库对于研究人员获得充足、高质量的原始数据至关重要。

根据这一目标，我们的研究小组目前正在致力于建立围手术期管理和结果数据库，这是一个大规模的全国性老年患者登记库 [5]。预期登记将提供丰富的数据集，这将有助于评估围手术期护理的质量，进而改善老年患者的临床护理。然而，事实证明，构建一个包含多个国家或地区并基于来自不同人群的参与者的数据库是具有挑战性的。预计未来，多国和多区域合作研究将成为该研究领域的标准。