Joint modeling of survival and longitudinal data: Carrico index data example

Abstract

Aim: When the respiratory system is unable to adequately absorb oxygen or excrete carbon dioxide, acute respiratory failure (ARF) develops. A current area of study is the survival analysis of patients with acute hypercapnic respiratory failure (AHRF) in the field of pulmonary diseases. In the follow-up period, several biochemical markers are repeatedly measured, such as respiration rate and Carrico Index; however, baseline or averaged values are mostly related to treatment failure. Although this approach is not inaccurate, it causes information loss, which leads to biased estimates. This prospective cohort study primarily looked at the relationship between changes in Carrico Index and failure of treatment in AHRF patients.

Methods: We included 86 patients from Ankara University School of Medicine Pulmonary Diseases Department. The association between the trajectory of the Carrico Index and failure in AHRF patients was examined using a joint modeling approach for longitudinal and survival data.

Results: Results showed that averaged Carrico Index change was inversely and significantly associated with failure (HR: 0.97, 95% CI: -0.05 to 1.97). With hazard ratios of 1.43 and 1.4, chronic health evaluation II (Apache II), and COPD Assesment test (CAT) were positively correlated with failure risk.

Conclusions: The present study demonstrate that applying the risk predictors' trajectory through an appropriate statistical method improved accuracy and avoid biased results.