Dataset Information

New agreement measures based on survival processes.

ABSTRACT: The need to assess agreement arises in many scenarios in biomedical sciences when measurements were taken by different methods on the same subjects. When the endpoints are survival outcomes, the study of agreement becomes more challenging given the special characteristics of time-to-event data. In this article, we propose a new framework for assessing agreement based on survival processes that can be viewed as a natural representation of time-to-event outcomes. Our new agreement measure is formulated as the chance-corrected concordance between survival processes. It provides a new perspective for studying the relationship between correlated survival outcomes and offers an appealing interpretation as the agreement between survival times on the absolute distance scale. We provide a multivariate extension of the proposed agreement measure for multiple methods. Furthermore, the new framework enables a natural extension to evaluate time-dependent agreement structure. We develop nonparametric estimation of the proposed new agreement measures. Our estimators are shown to be strongly consistent and asymptotically normal. We evaluate the performance of the proposed estimators through simulation studies and then illustrate the methods using a prostate cancer data example.

SUBMITTER: Guo Y

PROVIDER: S-EPMC4017950 | biostudies-literature | 2013 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

New agreement measures based on survival processes.

Guo Ying Y Li Ruosha R Peng Limin L Manatunga Amita K AK

Biometrics 20130711 4

The need to assess agreement arises in many scenarios in biomedical sciences when measurements were taken by different methods on the same subjects. When the endpoints are survival outcomes, the study of agreement becomes more challenging given the special characteristics of time-to-event data. In this article, we propose a new framework for assessing agreement based on survival processes that can be viewed as a natural representation of time-to-event outcomes. Our new agreement measure is formu ...[more]

PMID: 23844617

Similar Datasets

Project description:Socioeconomic status is an important determinant of health, the measurement of which is of great significance to population health research. However, individual-level socioeconomic factors are absent from much health administrative data, resulting in widespread use of area-level measures in their place. This study aims to clarify the role of individual- and area-level socioeconomic status in Ontario, Canada, through comparison of income measures. Using data from four cycles (2005-2012) of the Canadian Community Health Survey, we assessed concordance between individual- and area-level income quintiles using percent agreement and Kappa statistics. Individual-level characteristics were compared at baseline. Cumulative adult premature mortality was calculated for 5-years following interview. Rates were calculated separately for area-level and individual-level income, and jointly for each combination of income groups. Multivariable negative binomial models were fit to estimate associations between area- and individual-level income quintile and premature mortality after adjustment for basic demographics (age, sex, interview cycle) and key risk factors (alcohol, smoking, physical activity, and body mass index). Agreement between individual- and area-level income measures was low. Kappa statistics for same and similar (i.e. ±1 quintile) measures were 0.11 and 0.48, indicating low and moderate agreement, respectively. Socioeconomic disparities in premature mortality were greater for individual-level income than area-level income. When rates were stratified by both area- and individual-level income quintiles simultaneously, individual-level income gradients persisted within each area-level income group. The association between income and premature mortality was significant for both measures, including after full adjustment. Area-level socioeconomic status is an inappropriate proxy for missing individual-level data. The low agreement between area- and individual-level income measures and differences in demographic profile indicate that the two socioeconomic status measures do not capture the same population groups. However, our findings demonstrate that both individual- and area-level income measures are associated with premature mortality, and describe unique socioeconomic inequities.

Project description:BackgroundStudies of agreement examine the distance between readings made by different devices or observers measuring the same quantity. If the values generated by each device are close together most of the time then we conclude that the devices agree. Several different agreement methods have been described in the literature, in the linear mixed modelling framework, for use when there are time-matched repeated measurements within subjects.MethodsWe provide a tutorial to help guide practitioners when choosing among different methods of assessing agreement based on a linear mixed model assumption. We illustrate the use of five methods in a head-to-head comparison using real data from a study involving Chronic Obstructive Pulmonary Disease (COPD) patients and matched repeated respiratory rate observations. The methods used were the concordance correlation coefficient, limits of agreement, total deviation index, coverage probability, and coefficient of individual agreement.ResultsThe five methods generated similar conclusions about the agreement between devices in the COPD example; however, some methods emphasized different aspects of the between-device comparison, and the interpretation was clearer for some methods compared to others.ConclusionsFive different methods used to assess agreement have been compared in the same setting to facilitate understanding and encourage the use of multiple agreement methods in practice. Although there are similarities between the methods, each method has its own strengths and weaknesses which are important for researchers to be aware of. We suggest that researchers consider using the coverage probability method alongside a graphical display of the raw data in method comparison studies. In the case of disagreement between devices, it is important to look beyond the overall summary agreement indices and consider the underlying causes. Summarising the data graphically and examining model parameters can both help with this.

Dataset Information

New agreement measures based on survival processes.

Publications

New agreement measures based on survival processes.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure