Are there different phenotypes of thoracic surgery patients? A latent class analysis of pretreatment patient-reported outcomes.
[BACKGROUND] Among patients undergoing thoracic surgery, quality of life is associated with multiple perioperative outcomes.
- 연구 설계 cohort study
APA
Peters EJ, Dufault B, et al. (2025). Are there different phenotypes of thoracic surgery patients? A latent class analysis of pretreatment patient-reported outcomes.. JTCVS open, 28, 715-727. https://doi.org/10.1016/j.xjon.2025.10.015
MLA
Peters EJ, et al.. "Are there different phenotypes of thoracic surgery patients? A latent class analysis of pretreatment patient-reported outcomes.." JTCVS open, vol. 28, 2025, pp. 715-727.
PMID
41473043
Abstract
[BACKGROUND] Among patients undergoing thoracic surgery, quality of life is associated with multiple perioperative outcomes. Whether patients suffer reduced quality of life in certain areas compared to others is unclear. Knowing this could direct risk mitigation interventions for patients who share common symptoms. The objective of this study was to determine whether patients can be subdivided into groups based on preoperative quality of life.
[METHODS] This is a secondary analysis of a retrospective cohort study of consecutive patients undergoing thoracic surgery between January 2018 and January 2019 at a Canadian tertiary center. Latent class analysis was conducted according to 3-level EuroQol-5 dimension (EQ-5D-3L) scores. The number of latent classes was selected by comparing different models using the Akaike information criterion, Bayesian information criterion, and statistic. Class separation was measured using normalized entropy statistics.
[RESULTS] Among 482 patients, models with 2 to 5 classes were constructed. The 3-class model demonstrated the lowest Akaike and Bayesian information criterion values. The statistic and entropy showed increased preference for models as the number of classes decreased. Within the 3-class model, class 1 demonstrated a 73% to 100% probability of endorsing low impairment across all EQ-5D-3L dimensions, class 2 demonstrated a 93% probability of at least some impairment in mobility, and class 3 showed an 81% probability of moderate pain.
[CONCLUSIONS] There is evidence that patients undergoing thoracic surgery can be divided into 3 latent classes based on EQ-5D-3L score: low symptom burden, mobility-pain complex, and pain predominant. By identifying patients using these latent classes, targeted supportive interventions may be offered in the pretreatment period to improve perioperative outcomes.
[METHODS] This is a secondary analysis of a retrospective cohort study of consecutive patients undergoing thoracic surgery between January 2018 and January 2019 at a Canadian tertiary center. Latent class analysis was conducted according to 3-level EuroQol-5 dimension (EQ-5D-3L) scores. The number of latent classes was selected by comparing different models using the Akaike information criterion, Bayesian information criterion, and statistic. Class separation was measured using normalized entropy statistics.
[RESULTS] Among 482 patients, models with 2 to 5 classes were constructed. The 3-class model demonstrated the lowest Akaike and Bayesian information criterion values. The statistic and entropy showed increased preference for models as the number of classes decreased. Within the 3-class model, class 1 demonstrated a 73% to 100% probability of endorsing low impairment across all EQ-5D-3L dimensions, class 2 demonstrated a 93% probability of at least some impairment in mobility, and class 3 showed an 81% probability of moderate pain.
[CONCLUSIONS] There is evidence that patients undergoing thoracic surgery can be divided into 3 latent classes based on EQ-5D-3L score: low symptom burden, mobility-pain complex, and pain predominant. By identifying patients using these latent classes, targeted supportive interventions may be offered in the pretreatment period to improve perioperative outcomes.