Decomposition of false negatives within SIMD and urban/rural groups.

<p>Plots consider the proportion of each admission type amongst all admissions in a group (<i>A</i>) and the proportion of each admission type amongst admissions in the group with SPARRA score < 10% (<i>B</i>) and plots show (<i>A</i>) against (<i>B</i>) − (<i>A</i>). Points above the line <i>y</i> = 0 correspond to admissions which are disproportionately poorly identified by SPARRA score (strictly the criterion )., Blue vertical lines show pointwise 95% confidence intervals. Upper plots show the proportion of (unpredicted) admissions due to each cause; lower plots show the proportion of deaths due to each cause. Distinctive points are labelled.</p> <p>(PDF)</p>