Abstract
Intro We develop a straightforward ICU acuity score (Q3) that is calculated every 3 hours throughout the first 10 days of the ICU stay. Q3 uses components of the Oxford Acute Severity of Illness Score (OASIS) and incorporates a new component score for vasopressor use. In well-behaved models of ICU mortality, the marginal effects of Q3 are significant across the first 10 days of the ICU stay. In separate models, Q3 has significant effects on ICU remaining length of stay. The score has implications for work that seeks to explain modifiable mechanisms of changing acuity during the ICU stay.
Methods From the MIMIC-III database, select ICU stays from 5 adult ICUs were partitioned into consecutive 3-hour segments. For each segment, the number of vasopressors administered and all 10 OASIS component scores were computed. Models of ICU mortality were estimated. OASIS component effects were examined, and vasopressor count bins were weighted. Q3 was defined as the sum of 8 retained OASIS components and a new weighted vasopressor score. Models of ICU mortality quadratic in Q3 were estimated for each of the first 10 ICU days and were subjected to segment-level, location-specific tests of discrimination and calibration on newer ICU stays. Marginal effects of Q3 were computed at different levels of Q3 by ICU day, and average marginal effects of Q3 were computed at each location by ICU day. ICU remaining length of stay (LOS) models were also estimated and the effects of Q3 were similarly examined.
Results Daily ICU mortality models using Q3 show no evidence of misspecification (Pearson-Windmeijer p>0.05, Stukel p>0.05), discriminate well in all ICUs over the first 10 days (AUROC ∼ 0.72 – 0.85), and are generally well calibrated (Hosemer-Lemeshow p>0.05, Spiegelhalter’s z p>0.05). A one-unit increase in Q3 from typical levels (Q3=15) affects the odds of ICU mortality by a factor of 1.14 to 1.20, depending on ICU day (p<0.001), and the ICU remaining LOS by 5.8 to 9.6% (p<0.001). On average, a one-unit increase in Q3 increases the probability of ICU mortality by 1 to 2 percentage points depending on location and ICU day, and ICU remaining LOS by 5 to 10 hours depending on location and ICU day.
Conclusion Q3 significantly affects ICU mortality and ICU remaining LOS in different ICUs across the first 10 days of the ICU stay. Depending on location and ICU day, a one-unit increase in Q3 increases the probability of ICU mortality by or 1-2 percentage points and ICU remaining LOS by 5 to 10 hours. Unlike static acuity scores or those updated infrequently, Q3 could be used in explanatory models to help elucidate mechanisms of changing ICU acuity.
Competing Interest Statement
All authors have a financial interest in Indicator Sciences, LLC, which funded this work.
Funding Statement
This study was funded by Indicator Sciences, LLC
Author Declarations
I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained.
Yes
The details of the IRB/oversight body that provided approval or exemption for the research described are given below:
All analysis was performed on MIMIC III database. All investigators were certified to use the MIMIC-III database. All data were previously de-identified, and the public use of MIMIC has been approved by the IRB of its hosting organization. No additional IRB approval was required. Project code is available from the authors.
I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals.
Yes
I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance).
Yes
I have followed all appropriate research reporting guidelines and uploaded the relevant EQUATOR Network research reporting checklist(s) and other pertinent material as supplementary files, if applicable.
Yes
Data Availability
Project code is available from the authors.