Competing Events Influence Estimated Survival Probability: When Is Kaplan-Meier Analysis Appropriate?

@article{citeulike:1406286, abstract = {The Kaplan-Meier estimator is the current method for estimating the probability of an event to occur with time in orthopaedics. However, the Kaplan-Meier estimator was designed to estimate the probability of an event that eventually will occur for all patients, ie, death, and this does not hold for other outcomes. For example, not all patients will experience hip arthroplasty loosening because some may die first, and some may have their implant removed to treat infection or recurrent hip dislocation. Such events that preclude the observation of the event of interest are called competing events. We suggest the Kaplan-Meier estimator is inappropriate in the presence of competing events and show that it overestimates the probability of the event of interest to occur with time. The cumulative incidence estimator is an alternative approach to Kaplan-Meier in situations where competing risks are likely. Three common situations include revision for implant loosening in the long-term followup of arthroplasties or implant failure in the context of limb-salvage surgery or femoral neck fracture.}, address = {From the D\'{e}partement de Biostatistique et Informatique M\'{e}dicale, Unit\'{e} INSERM 717, H\^{o}pital Saint Louis, Paris, France.}, author = {Biau, David J. and Latouche, Aur\'{e}lien and Porcher, Rapha\"{e}l }, citeulike-article-id = {1406286}, doi = {10.1097/BLO.0b013e3180986753}, issn = {0009-921X}, journal = {Clin Orthop Relat Res}, keywords = {statistics}, month = {May}, posted-at = {2007-06-23 11:03:40}, priority = {2}, title = {Competing Events Influence Estimated Survival Probability: When Is Kaplan-Meier Analysis Appropriate?}, url = {http://dx.doi.org/10.1097/BLO.0b013e3180986753}, year = {2007} }

TY - JOUR ID - citeulike:1406286 L3 - citeulike-article-id:1406286 TI - Competing Events Influence Estimated Survival Probability: When Is Kaplan-Meier Analysis Appropriate? JF - Clin Orthop Relat Res AD - From the Département de Biostatistique et Informatique Médicale, Unité INSERM 717, Hôpital Saint Louis, Paris, France. SN - 0009-921X N2 - The Kaplan-Meier estimator is the current method for estimating the probability of an event to occur with time in orthopaedics. However, the Kaplan-Meier estimator was designed to estimate the probability of an event that eventually will occur for all patients, ie, death, and this does not hold for other outcomes. For example, not all patients will experience hip arthroplasty loosening because some may die first, and some may have their implant removed to treat infection or recurrent hip dislocation. Such events that preclude the observation of the event of interest are called competing events. We suggest the Kaplan-Meier estimator is inappropriate in the presence of competing events and show that it overestimates the probability of the event of interest to occur with time. The cumulative incidence estimator is an alternative approach to Kaplan-Meier in situations where competing risks are likely. Three common situations include revision for implant loosening in the long-term followup of arthroplasties or implant failure in the context of limb-salvage surgery or femoral neck fracture. KW - statistics AU - Biau, David AU - Latouche, Aurélien AU - Porcher, Raphaël PY - 2007/05/10/ UR - http://dx.doi.org/10.1097/BLO.0b013e3180986753 ER -