The Riyadh Intensive Care Program mortality prediction algorithm assessed in 617 intensive care patients in Glasgow

Abstract

SummaryThe

Limitations of score-based daily outcome predictions in the individual intensive care patient

“…After Chang moved to London, the RIP ICU management software was introduced in several hospitals in the UK resulting in additional validation reports [8,10,11] (Table 3). In a Canadian multicenter trial including 3,350 ICU admissions, Rogers and Fuller [12] observed that only 336 out of 569 patients predicted to die actually died (positive predictive value = 59 %).…”

“…Although all authors applied the same algorithm, the results show a large variability. Among the independent validation studies there seem to be two clusters of results, one group with a low false prediction rate (<10%) but with insufficient sensitivity ( [8,11] and our study), and a second group with moderate sensitivity around 50% but with a high rate of false predictions [9,10,12,23]. Since the applied algorithm and the cut-off levels were the same in all investigations, S172…”

Using serial severity scores to predict death in ICU patients: a validation study and review of the literature

“…Atkinson et al [13] applied the Riyadh algorithm to predict outcomes in 3600 patients and found that it predicted 137 patients to die, 6 of whom lived at least 90 days (false prediction of death rate 4.4%). Similarly, Hope and Plenderleith [7] used the prediction algorithm in 617 general ICU patients and it predicted 119 to die, 24 of whom lived (false prediction of death rate 20.2%). Finally, Rogers and Fuller [8] applied an adaptation of the Riyadh model to 3350 consecutive ICU admissions and found that of the 569 patients that met one of the 'predicted to die' criteria, their mortality rate was only approximately 55-60%.…”

“…This is different from the purpose for which these scores were created which was to predict outcome by data obtained generally in first 24 h of admission. A group of studies in the past 3 decades have shown that a rising level of severity of illness as determined by either proprietary or commercially available scoring systems can be a very specific, although not sensitive, predictor of either ICU death or hospital death [1 ,2 , [3][4][5][6][7][8][9].…”