Empirical study of correlated survival times for recurrent events with proportional hazards margins and the effect of correlation and censoring

• Published in: BMC medical research methodology Vol. 13; no. 1; p. 95
• Main Authors: Villegas, Rodrigo, Julià, Olga, Ocaña, Jordi
• Format: Journal Article
• Language: English
• Published: England BioMed Central Ltd 24.07.2013; BioMed Central
• Subjects: Airway Obstruction - etiology; Algorismes; Algorithms; Aparell respiratori; Confidence Intervals; Economic models; Estadística mèdica; Failure; Female; Fever; Humans; Hyperthermia; Infant; Longitudinal method; Medical statistics; Models, Statistical; Mètodes estadístics; Office Visits - statistics & numerical data; Particulate Matter - adverse effects; Proportional Hazards Models; Recurrence; Respiratory organs; Respiratory Sounds - etiology; Simulation; Software; Statistical methods; Studies; Survival Analysis; Time Factors
• ISSN: 1471-2288; 1471-2288
• DOI: 10.1186/1471-2288-13-95

In longitudinal studies where subjects experience recurrent incidents over a period of time, such as respiratory infections, fever or diarrhea, statistical methods are required to take into account the within-subject correlation. For repeated events data with censored failure, the independent increment (AG), marginal (WLW) and conditional (PWP) models are three multiple failure models that generalize Cox's proportional hazard model. In this paper, we revise the efficiency, accuracy and robustness of all three models under simulated scenarios with varying degrees of within-subject correlation, censoring levels, maximum number of possible recurrences and sample size. We also study the methods performance on a real dataset from a cohort study with bronchial obstruction. We find substantial differences between methods and there is not an optimal method. AG and PWP seem to be preferable to WLW for low correlation levels but the situation reverts for high correlations. All methods are stable in front of censoring, worsen with increasing recurrence levels and share a bias problem which, among other consequences, makes asymptotic normal confidence intervals not fully reliable, although they are well developed theoretically.