C. Singh1, M. Salvitti1, H. Roberts1, D. Lysov1, K. Bressler1, R. Kiernan1, D. Karev2, S. Dirusso1,2 1New York Institute of Technology College of Osteopathic Medicine, Old Westbury, NY, USA 2St. Barnabus Hospital, Department Of Surgery, Bronx, NY, USA
Introduction:
Data has shown that all death mortality has risen during and after the COVID pandemic. This study uses a locally generated highly predictive mortality model to help analyze the effect of the COVID pandemic on a NYC inner city trauma center performance pre, during, and post COVID pandemic timelines.
Methods:
Data: Trauma Registry data from a Bronx Level 2 Trauma Center (2016-2022: n=6305). A forward entry binomial logistic regression model (LR) was derived on 2016-2018 data using demographic, physiologic, comorbidity, and deposition variables; the LR was then used to generate predicted mortalities of patients admitted in 2019 (n=846), 2020 (n=940), 2021 (n=1036), and 2022 (n=913). The sum of the predicted probabilities for each year (2019,2020,2021,2022) was calculated and compared to the number of observed deaths.
Results:
The LR had excellent discrimination (AuROC=0.974) and calibration (LH-C Statistic = 6.57). Descriptive Analysis of key demographics: percentage of predictor comorbid conditions in patients significantly changed (p<0.001); while use of public insurance declined by 11.5%, patients with no/unknown insurance increased by 10.6% (p<0.001); mean age at admission declined (p=0.008); mean NISS decreased (p<0.001); mean Total GCS score, diastolic blood pressure, Revised Trauma Score (RTS), and Abbreviated Injury Scale (AIS) score remained steady (p≥0.051). In 2019, the LR model predicted 38 (4.5%) deaths, but only 33 (3.9%) were observed (p>0.439). From 2020-2022, the LR predicted 33 (3.5%), 32 (3.1%), and 26 (2.9%) deaths, while 40 (4.3%), 43 (4.2%), and 40 (4.4%) were observed, respectively. These differences were significant (p<0.05).
Conclusion:
The relatively steady patterns in patient populations between 2016-2019 were disrupted during 2020-2022, revealing COVID19’s subtle influence on the healthcare landscape. Regardless, our LR remained sensitive and specific, enabling meaningful surveillance of trauma service performance. The difference in predicted and observed mortality in 2019 suggests maintenance of high-quality care. Despite the declining acuity of admitted patients, the presence of 32 unexpected deaths from 2020-2022 suggests barriers to the delivery of quality care. This finding reflects an elevated excess mortality stemming from direct/indirect impacts of COVID19 on healthcare at a local level. The persistence of increased unexpected deaths in 2022 reveals a sluggish, hindered return to normalcy despite laxing public concern and policies. Further investigation involves utilization of the LR to extract and evaluate these unexpected deaths to identify specific hurdles to delivering quality care that may have been precipitated by the pandemic.
