Dataset Information

A dynamic approach for reconstructing missing longitudinal data using the linear increments model.

ABSTRACT: Missing observations are commonplace in longitudinal data. We discuss how to model and analyze such data in a dynamic framework, that is, taking into consideration the time structure of the process and the influence of the past on the present and future responses. An autoregressive model is used as a special case of the linear increments model defined by Farewell (2006. Linear models for censored data, [PhD Thesis]. Lancaster University) and Diggle and others (2007. Analysis of longitudinal data with drop-out: objectives, assumptions and a proposal. Journal of the Royal Statistical Society, Series C (Applied Statistics, 56, 499-550). We wish to reconstruct responses for missing data and discuss the required assumptions needed for both monotone and nonmonotone missingness. The computational procedures suggested are very simple and easily applicable. They can also be used to estimate causal effects in the presence of time-dependent confounding. There are also connections to methods from survival analysis: The Aalen-Johansen estimator for the transition matrix of a Markov chain turns out to be a special case. Analysis of quality of life data from a cancer clinical trial is analyzed and presented. Some simulations are given in the supplementary material available at Biostatistics online.

SUBMITTER: Aalen OO

PROVIDER: S-EPMC3293429 | biostudies-literature | 2010 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A dynamic approach for reconstructing missing longitudinal data using the linear increments model.

Aalen Odd O OO Gunnes Nina N

Biostatistics (Oxford, England) 20100413 3

Missing observations are commonplace in longitudinal data. We discuss how to model and analyze such data in a dynamic framework, that is, taking into consideration the time structure of the process and the influence of the past on the present and future responses. An autoregressive model is used as a special case of the linear increments model defined by Farewell (2006. Linear models for censored data, [PhD Thesis]. Lancaster University) and Diggle and others (2007. Analysis of longitudinal data ...[more]

PMID: 20388914

Similar Datasets

Project description:BackgroundMultiple sources of influenza surveillance data are becoming more available; however integration of these data streams for situational awareness of influenza activity is less explored.Methods and resultsWe applied multivariate time-series methods to sentinel outpatient and school absenteeism surveillance data in Hong Kong during 2004-2009. School absenteeism data and outpatient surveillance data experienced interruptions due to school holidays and changes in public health guidelines during the pandemic, including school closures and the establishment of special designated flu clinics, which in turn provided 'drop-in' fever counts surveillance data. A multivariate dynamic linear model was used to monitor influenza activity throughout epidemics based on all available data. The inferred level followed influenza activity closely at different times, while the inferred trend was less competent with low influenza activity. Correlations between inferred level and trend from the multivariate model and reference influenza activity, measured by the product of weekly laboratory influenza detection rates and weekly general practitioner influenza-like illness consultation rates, were calculated and compared with those from univariate models. Over the whole study period, there was a significantly higher correlation (ρ = 0.82, p≤0.02) for the inferred trend based on the multivariate model compared to other univariate models, while the inferred trend from the multivariate model performed as well as the best univariate model in the pre-pandemic and the pandemic period. The inferred trend and level from the multivariate model was able to match, if not outperform, the best univariate model albeit with missing data plus drop-in and drop-out of different surveillance data streams. An overall influenza index combining level and trend was constructed to demonstrate another potential use of the method.ConclusionsOur results demonstrate the potential use of multiple streams of influenza surveillance data to promote situational awareness about the level and trend of seasonal and pandemic influenza activity.

Dataset Information

A dynamic approach for reconstructing missing longitudinal data using the linear increments model.

Publications

A dynamic approach for reconstructing missing longitudinal data using the linear increments model.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets