A comparative study of models for the different covariance structure analysis of repeated measurement data

In repeated measurement data, the variables are not independent, and a certain auto- correlation typically exists between different levels of repeated measurement factors. The random error is composed of at least two parts, i.e. the individual random effect and the intra-individual multi-repeated measurement effect. Traditional statistical analysis methods （such as the t-test and the one-way analysis of variance） are not applicable. The linear mixed model has been widely applied for the analysis and design of repeated measurement data. This paper focuses on medical examples and describes the selection of a covariance structure for the linear mixed model of repeated measurement in the modeling of different variance-eovariance structures. By selecting different covariance structures, we can perform the parameter estimation and statistical test for the fixed effect of repeated measurement data, the parameters of random effects, and the covari- ance matrix. The results are analyzed and compared to provide a reference for applying the linear mixed model of repeated measurement to medical research.

Evaluation of neuro-intensive care unit performance in China： predicting outcomes of Simplified Acute Physiology Score II or Glasgow Coma Scale

Background Severity scoring systems are useful tools for measuring the severity of the disease and its outcome. This pilot study was to verify and compare the prognostic performance of the Simplified Acute Physiology Score II （SAPS II） and Glasgow Coma Scale （GCS） in neuro-intensive care unit （N-ICU） patients. Methods A total of 1684 patients consecutively admitted to the N-ICU at Xuanwu Hospital between January 1, 2005 and December 31, 2011 were enrolled in this study. The data-base included admission data, at 24-, 48-, and 72-hour SAPS II and GCS. Repeated measure data analysis of variance, Logistic regression analysis, the Hosmer-Lemeshow goodness-of-fit statistic, and the area under the receiver operating characteristic were used to evaluate the performance. Results There was a significant difference between the SAPS II or GCS score at four time points （F=16.110, P=0.000 or F=8.108, P=0.000）. The SAPS II scores or GCS score at four time points interacted with the outcomes with significant difference （F=116.771, P=0.000 or F=65.316, P=0.000）. Calibration of the SAPS II or GCS score at each time point on all patients was good. The percentage of a risk estimate prediction corresponding to observed mortality was also good. The 72-hour score have the greatest consistency. Discriminations of the SAPS II or GCS score at each time were all satisfactory. The 72-hour score had the greatest discriminative power. The cut-off value was 33 （sensitivity of 85.2% and specificity of 74.3%） and 6 （sensitivity of 70.6% and specificity of 65.0%）. The SAPS II at each time point on all patients showed better calibration, consistency and discrimination than GCS. The binary Logistic regression analysis identified physiological variables, GCS, age, and disease category as significant independent risk factors of death. After the two variables including underlying disease and type of admission were excluded, we built the simplified SAPS II model. A correlation was suggested between the simplified SAPS II score at each time point and outcome, regardless of the diagnosis. Conclusions The GCS scoring system tends to be a little weaker in the predictive power than the SAPS II scoring system in this Chinese cohort of N-ICU patients. The advantage of SAPS II scoring system still exists that it dose not need to take into account the diagnosis or diseases categories, even in the special N-ICU. The simplified SAPS II scoring system is considered a new idea for the estimation of effectiveness.