|| Checking for direct PDF access through Ovid
The aim of this study was to determine the association of early treatment intensity with hospital mortality of intensive care unit (ICU) patients.We performed an observational study based on a national ICU registry. We included adult patients treated in Finnish ICUs between 2003 and 2013 with the length of ICU stay of more than 3 days. We measured treatment intensity with the Therapeutic Intervention Scoring System (TISS-76). We assessed mean and daily TISS scores. To define the change in treatment intensity during the first days in the ICU, we calculated the difference between the TISS score on day 3 and the score on day 1 (ΔTISS). We used multivariate logistic regression to adjust for baseline differences and continuous net reclassification improvement (NRI) to determine the impact of adding TISS data to the baseline prediction model on its prognostic performance.We identified 42,493 patients eligible for the study. For 71% of the patients, ΔTISS was ≤ 0 and crude hospital mortality was 18%. ΔTISS > 0 was observed for 29% of the patients, with a hospital mortality of 23%. When compared to the group ΔTISS ≤ 0, the category ΔTISS > 0 was independently associated with substantially increased mortality. Adding TISS data to the prediction model resulted in the improvement of prognostic performance particularly in the patients with the lowest initial baseline risk.Early increase in TISS scores was associated with increased risk of death, especially in patients with a lower initial severity of illness.