Intensive care units (ICUs) are increasingly interested in assessing and improving their performance. ICU Length of Stay (LoS) could be seen as an indicator for efficiency of care. However, little... Show moreIntensive care units (ICUs) are increasingly interested in assessing and improving their performance. ICU Length of Stay (LoS) could be seen as an indicator for efficiency of care. However, little consensus exists on which prognostic method should be used to adjust ICU LoS for case-mix factors. This study compared the performance of different regression models when predicting ICU LoS. We included data from 32,667 unplanned ICU admissions to ICUs participating in the Dutch National Intensive Care Evaluation (NICE) in the year 2011. We predicted ICU LoS using eight regression models: ordinary least squares regression on untransformed ICU LoS,LoS truncated at 30 days and log-transformed LoS; a generalized linear model with a Gaussian distribution and a logarithmic link function; Poisson regression; negative binomial regression; Gamma regression with a logarithmic link function; and the original and recalibrated APACHE IV model, for all patients together and for survivors and non-survivors separately. We assessed the predictive performance of the models using bootstrapping and the squared Pearson correlation coefficient (R2), root mean squared prediction error (RMSPE), mean absolute prediction error (MAPE) and bias. The distribution of ICU LoS was skewed to the right with a median of 1.7 days (interquartile range 0.8 to 4.0) and a mean of 4.2 days (standard deviation 7.9). The predictive performance of the models was between 0.09 and 0.20 for R2, between 7.28 and 8.74 days for RMSPE, between 3.00 and 4.42 days for MAPE and between -2.99 and 1.64 days for bias. The predictive performance was slightly better for survivors than for non-survivors. We were disappointed in the predictive performance of the regression models and conclude that it is difficult to predict LoS of unplanned ICU admissions using patient characteristics at admission time only. Show less
OBJECTIVES\nErrors in the registration or extraction of patient outcome data, such as in-hospital mortality, may lower the reliability of the quality indicator that uses this (partly) incorrect... Show moreOBJECTIVES\nErrors in the registration or extraction of patient outcome data, such as in-hospital mortality, may lower the reliability of the quality indicator that uses this (partly) incorrect data. Our aim was to measure the reliability of in-hospital mortality registration in the Dutch National Intensive Care Evaluation (NICE) registry.\nMETHODS\nWe linked data of the NICE registry with an insurance claims database, resulting in a list of discrepancies in in-hospital mortality. Eleven Intensive Care Units (ICUs) were visited where local data sources were investigated to find the true in-hospital mortality status of the discrepancies and to identify the causes of the data errors in the NICE registry. Original and corrected Standardized Mortality Ratios (SMRs) were calculated to determine if conclusions about quality of care changed compared to the national benchmark.\nRESULTS\nIn eleven ICUs, 23,855 records with 460 discrepancies were identified of which 255 discrepancies (1.1% of all linked records) were due to incorrect in-hospital mortality registration in the NICE registry. Two programming errors in computer software of six ICUs caused 78% of errors, the remainder was caused by manual transcription errors and failure to record patient outcomes. For one ICU the performance became concordant with the national benchmark after correction, instead of being better.\nCONCLUSIONS\nThe reliability of in-hospital mortality registration in the NICE registry was good. This was reflected by the low number of data errors and by the fact that conclusions about the quality of care were only affected for one ICU due to systematic data errors. We recommend that registries frequently verify the software used in the registration process, and compare mortality data with an external source to assure consistent quality of data. Show less
