Skip to main content
Figure 5 | BMC Medical Research Methodology

Figure 5

From: A review of statistical estimators for risk-adjusted length of stay: analysis of the Australian and new Zealand intensive care adult patient data-base, 2008–2009

Figure 5

Upper panel: Violin plots of ICU length of stay (raw-scale, length of stay < 16 days) and predicted length of stay (raw-scale, < 16 days; determination and validation data sets) for estimators: GLM Poisson-log, GLM negative binomial-log, GLM gamma-log, GLM inverse Gaussian-log. Lower panel: Hosmer-Lemeshow (F) test of residuals across deciles of linear-predictor, for estimators: GLM Poisson-log, GLM negative binomial-log, GLM gamma-log, GLM inverse Gaussian-log. X-axis, linear predictor; Y-axis, parameter estimates. determin., determination. Validat., validation.

Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us