Predictive performance of scales during CoViD-19 disease

Since the beginning of the COVID-19 pandemic, many researchers have constructed mathematical models related to the development of the epidemic. There has been a rapid interest in both deterministic and stochastic models. However, statistical studies devoted to the validation of the risk of complications in patients with SARS-CoV-2 are also important. For this purpose, various types of scales are used, which, taking into account various factors, allow for prediction of patient mortality. Documentation of these scales can be found on the MDCALC website (https://www.mdcalc.com/). Many scales were originally proposed to study other phenomena. For example, C2HEST was initially proposed to predict incident atrial fibrillation (AF) in the general population, CHA₂DS₂-VASc Score for Atrial Fibrillation Stroke Risk.

In our research, we check the predictive abilities of scales during CoViD-19 disease. The work is ongoing, but the first results are very promising and have already been published. One of the works is devoted to the study of the predictive abilities of C2HEST in clinical outcomes in elderly subjects with COVID-19-subjects [1]. All medical records were collected as part of the COronavirus in LOwer Silesia — the COLOS registry. The study group was classified into three risk subgroups: high, medium and low. The categorization was performed on the basis of the statistical log-rank test so that the differences between the hazard functions in the individual groups were statistically significant. We noticed significant differences in the in-hospital and 3-month and 6-month mortality.

Time-dependent AUC was calculated for the C2HEST score in predicting all-cause mortality in a period extending from the day of hospital admission up to 240 days after the initial diagnosis. In Figure 1 we can see the outcome of these calculations.

Figure 1. Time-dependent ROC analysis was performed. Calculations were made in R software

One of the results of our research is predicting all-cause mortality among elderly subjects with COVID-19. For this purpose, estimation of the survival curves was performed. Kaplan-Meier functions were used for this purpose. The p-value for log-rank test was <0.0001.

Figure 2. Survival function estimation for different risk strata.

We have only presented some of the results here. They are published in open-access journals (see the literature below). We encourage everyone to read the results. As we have already mentioned, the works are underway. We conduct research on various scales in various groups of patients. We are very excited about the preliminary result on the predictive ability of the VACO and GRAM scales. We will explain all details in the works that are currently being reviewed.

Finally, we would like to highlight two important things here. Statistical analyzes and reports were prepared by the Statistical Analysis Center of Wroclaw Medical University in cooperation with Wrocław University of Science and Technology. During these studies, many interviews were conducted with doctors from the University Clinical Hospital in Wrocław. We believe that this collaboration can contribute to the advancement of technology in the fields of medicine and demography and is important for the pharmaceutical industry. The second essence of our research is the benefits for patients. Determining the risk will allow for the development of better health prevention and treatment methods.

References

[1] P. Rola, A. Doroszko, M. Trocha, K. Giniewicz, K. Kujawa, M. Skarupski, J. Gawryś, T. Matys, E. Szahidewicz-Krupska, D. Gajecki, B. Adamik, K. Kaliszewski, K. Kilis-Pstrusinska, K. Letachowicz, A. Matera-Witkiewicz, M. Pomorski, M. Protasiewicz, K. Majchrzak, J. Sokołowski, E. A. Jankowska, K. Madziarska, Mortality predictive value of the C2HEST score in elderly subjects with Covid-19. a subanalysis of the COLOS study, Journal of Clinical Medicine 11 (4).  doi: 10.3390/jcm11040992

[2] D. Gajecki, A. Doroszko, M. Trocha, K. Giniewicz, K. Kujawa, M. Skarupski, J. Gawryś, T. Matys, E. Szahidewicz-Krupska, P. Rola, B. Stachowska, J. Halupczok-Zy la, B. Adamik, K. Kaliszewski, K. Kilis-Pstrusinska, K. Letachowicz, A. Matera-Witkiewicz, M. Pomorski, M. Protasiewicz, M. Madziarski, K. Konikowska, A. Remiorz, M. Or lowska, K. Proc, M. Szymala-Pedzik, J. Żórawska, K. Lindner, J. Sokołowski, E. A. Jankowska, K. Madziarska, Usefulness of the C2HEST score in predicting the clinical outcomes of Covid-19 in diabetic and non-diabetic cohorts, Journal of Clinical Medicine 11 (3). doi:10.3390/jcm11030873

By Marek Skarupski (Faculty of Pure and Applied Mathematics, Wroclaw University of Science and Technology)

%d bloggers like this: