Total amount: € 0,00
Indexed/Abstracted in: EMBASE, PubMed/MEDLINE, Science Citation Index Expanded (SciSearch), Scopus
Impact Factor 0,536
Online ISSN 1827-1758
XVI CONGRESS DELLA SOCIETÀ ITALIANA DI NEFROLOGIA SEZIONE PIEMONTE-VALLE D’AOSTA
Castello di Rivoli, Rivoli (TO) October 10, 1998
Pacitti A., Barbieri S., Hollò S. *
Università degli Studi - Torino Cattedra di Nefrologia AOS S. Giovanni Battista, Torino UOA, Nefrologia, Dialisi e Trapianto
Background. The paradox of the increased mortality in the patients with acute renal failure (ARF) although submitted to better cares and newer renal replacement therapies (RRT) has recently prompted to the use of quantitative individual severity scores (ISS) calculating for each patient an individual death probability (DP) in correlation with the risk covariates found before the start of RRT; beside the clinical use, the ISS allow an evaluation of the effect of strategies and modalities of treatments as quantitative additive factors eventually added or subtracted to the base-line individual background of risk. The ideal index should be chosen on the basis of its precocity (origin just at the start of therapy), sensitivity (true positive against false positive results), universality (independence from the development set) and discriminative power (the capability to discern patients potentially treatable from those with an unchangeable prognosis). Indexes already validated in their development set should be used and studied into a different set (“evaluation set”).
Methods. The aim of this study has been: to evaluate a literature index (ATN-ISS, Liaño, developed prospectively in a remote set) in the local (A) environment on 340 patients with ARF successively treated with dialysis (mostly hemofiltration) studied retrospectively along a 4 year period in our regional hospital and compare its performances with a local index (PDTOR) developed by logistic analysis in the same pool; the fitness of both tests to the real outcome has been evaluated by the Limeeshow test and by ROC curves; to compare both indexes in a remote environment (B) of a dialytic pool of 345 patients extracted by a group of 25721 patients treated by 25 Italian ICU (Archidia Study group). The responses of the two indexes have been compared even with the index (SAPSII) prospectively generated at the admittance in the ICU by the Archidia Group.
Results. In the local set (A) TOR-ISS fits well with the outcomes (Limeshow test C2=N.S.) as expected being evaluated in its own “development” set, while ATN-ISS significantly underestimates deaths, perhaps working on a retrospectively built data-base, that could contain fewer risk elements than necessary. (B) In the remote set, ATN-ISS fits very well, while TOR-ISS significantly overestimates expected deaths, for its retrospective origin or for a real lower death incidence compared to that of its development set. SAPSII shows no correlation at all with the outcome because its calculation is often well before (10 days on average) than the actual start of dialytic treatment.
Conclusions. In conclusion ATN-ISS, an index built prospectively on a large cohort of patients, fits correctly in a remote prospectively built evaluation set. Retrospective built indexes or data-base don’t allow a correct ISS evaluation while ICU indexes (SAPSII, APACHE), generated at the admittance in the ICU should not be used for ARF patients submitted to dialysis.