Trajectory prediction in percutaneous coronary intervention recovery

This invited commentary refers to ‘Early-term heterogeneous trajectories of patient-reported outcome undergoing percutaneous coronary intervention: a multicenter and prospective longitudinal study’ by J. Zhao et al., https://doi.org/10.1093/eurjcn/zvaf064. There is a large degree of variation in how individuals respond to and recover from percutaneous coronary intervention (PCI) after acute coronary syndrome. Patients exhibit varied symptom trajectories, psychological adjustment, and functional recovery during the early months post-procedure.1–3 Several different analytic approaches that aim to uncover distinct patient subgroups characterized by shared patterns or profiles have been used across numerous recent articles in the European Journal of Cardiovascular Nursing that are well-suited to understanding complex, multidimensional outcomes for people with cardiovascular disease.4–8 Unlike some of these methods that assume a single, homogeneous trajectory, growth mixture models (GMM) allows for the identification of distinct trajectories while simultaneously accounting for individual variability within subgroups. Zhao et al.9 used GMM to analyse patient-reported outcomes collected from 353 patients in China who had PCI after an acute coronary syndrome event. The patient-reported outcome instrument for chronic disease-coronary heart disease was assessed at baseline, 7 days, 1 month, and 3 months, which provided an overall score of self-perceived health status incorporating physical, mental, social, spiritual, and coronary heart disease-specific domains. The analysis revealed three distinct classes. The largest cluster that accounted for most trajectories comprised patients who had moderate health status scores at baseline that improved over time (86.4%). Another cluster had patients with initially low health status scores that also steadily improved over time (5.10%). The final cluster was represented by patients who started with higher health status scores that, in contrast to the other groups, steadily declined over the 3-month follow-up period (8.50%). A key finding from studies that use GMM is the potential to proactively identify patients likely to belong to specific subgroups and implement targeted interventions to alter their clinical trajectory. In the case of the research by Zhao et al.,9 it would be particularly advantageous to be able to identify at baseline the patients that will likely fall within the cluster represented by deteriorating health status over time. Unfortunately, the results of multinomial logistic regression analyses revealed only that patients who had STEMI were less likely to fit with the cluster represented by deteriorating health status at 3 months. For this approach to be clinically useful, prediction models must be much more accurate and reliable at identifying patients at risk for poor recovery. A broader range of clinical, psychosocial, and behavioural variables should be considered as predictors, as well as application of machine learning approaches that can capture complex interactions among predictors better than traditional regression. The creative use of natural language processing to leverage rich information from clinical notes is an interesting approach that should be considered for these sorts of predictive tasks given that recent research has demonstrated the substantial predictive value that such unstructured data can provide.10,11 It is of course also possible that patient-reported outcome trajectories after PCI may differ substantially between contexts, so consideration of the most appropriate approach for external validation would be required prior to implementation of results from this study into practice.12 In addition, it will be interesting to ascertain whether this GMM approach can be employed using longer-term health status outcomes. The research by Zhou et al.9 reinforces the need for adoption of chronic disease models of care that encourage effective self-management for people post-PCI. In addition to Phase II cardiac rehabilitation, the benefits of integrating self-management support models for people with chronic conditions may extend beyond optimizing patient recovery and outcomes, to potential economic and accessibility burden reduction and can commence in the primary care setting.13–15 Aaron Conway (Conceptualization [equal], Writing—original draft [equal], Writing—review & editing [equal]), and Katina Corones-Watkins (Conceptualization [equal], Writing—original draft [equal], Writing—review & editing [equal])

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.