Objectives: To develop a simplified Therapeutic Intervention Scoring System (TISS) based on the TISS-28 items and to validate the new score in an independent database.
Design: Retrospective statistical analysis of a database and a prospective multicentre study.
Setting: Development in the database of the Foundation for Research on Intensive Care in Europe with external validation in 64 intensive care units (ICUs) of 11 European countries.
Measurements and results: Development of NEMS on a random sample of TISS-28 items, cross validation on another random sample of TISS-28, and external validation of NEMS in comparison with TISS-28 scored by two independent raters on the day of the visit to the ICUs participating in an international study. Multivariable regression techniques, Pearson's correlation, and paired sample t-tests were used (significance at p < 0.05 level). Intraclass correlation, rate of agreement, and kappa statistics were used for interrater reliability tests. The TISS-28 items were reduced to NEMS (9 items) in a random sample of 2000 records; the means of the two scores were no different: TISS-28 26.23 +/- 10.38, NEMS 26.19 +/- 9.12, NS. Cross-validation in a random sample of 996 records; mean TISS-28 26.13 +/- 10.38, NEMS 26.17 +/- 9.38, NS; R2 = 0.76. External validation on 369 pairs of TISS-28 and NEMS has shown that the means of the two scores were no different: TISS-28 27.56 +/- 11.03, NEMS 27.02 +/- 8.98, NS; R2 = 0.59. Reliability tests have shown an "almost perfect" interrater correlation. Similar to studies correlating TISS with Simplified Acute Physiology Score (SAPS)-I and/or Acute Physiology and Chronic Health Evaluation II scores, the value of NEMS scored on the first day accounts for 30.4% of the variation of SAPS-II score.
Conclusions: NEMS is a suitable therapeutic index to measure nursing workload at the ICU level. The use of NEMS is indicated for: (a) multicentre ICU studies; (b) management purposes in the general (macro) evaluation and comparison of workload at the ICU level; (c) the prediction of workload and planning of nursing staff allocation at the individual patient level.