Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 May 9;20(6):100.
doi: 10.3892/br.2024.1788. eCollection 2024 Jun.

Time‑dependent ROC curve analysis to determine the predictive capacity of seven clinical scales for mortality in patients with COVID‑19: Study of a hospital cohort with very high mortality

Affiliations

Time‑dependent ROC curve analysis to determine the predictive capacity of seven clinical scales for mortality in patients with COVID‑19: Study of a hospital cohort with very high mortality

Martha A Mendoza-Hernandez et al. Biomed Rep. .

Abstract

Clinical data from hospital admissions are typically utilized to determine the prognostic capacity of Coronavirus disease 2019 (COVID-19) indices. However, as disease status and severity markers evolve over time, time-dependent receiver operating characteristic (ROC) curve analysis becomes more appropriate. The present analysis assessed predictive power for death at various time points throughout patient hospitalization. In a cohort study involving 515 hospitalized patients (General Hospital Number 1 of Mexican Social Security Institute, Colima, Mexico from February 2021 to December 2022) with COVID-19, seven severity indices [Pneumonia Severity Index (PSI) PaO2/FiO2 arterial oxygen pressure/fraction of inspired oxygen (Kirby index), the Critical Illness Risk Score (COVID-GRAM), the National Early Warning Score 2 (NEWS-2), the quick Sequential Organ Failure Assessment score (qSOFA), the Fibrosis-4 index (FIB-4) and the Viral Pneumonia Mortality Score (MuLBSTA were evaluated using time-dependent ROC curves. Clinical data were collected at admission and at 2, 4, 6 and 8 days into hospitalization. The study calculated the area under the curve (AUC), sensitivity, specificity, and predictive values for each index at these time points. Mortality was 43.9%. Throughout all time points, NEWS-2 demonstrated the highest predictive power for mortality, as indicated by its AUC values. PSI and COVID-GRAM followed, with predictive power increasing as hospitalization duration progressed. Additionally, NEWS-2 exhibited the highest sensitivity (>96% in all periods) but showed low specificity, which increased from 22.9% at admission to 58.1% by day 8. PSI displayed good predictive capacity from admission to day 6 and excellent predictive power at day 8 and its sensitivity remained >80% throughout all periods, with moderate specificity (70.6-77.3%). COVID-GRAM demonstrated good predictive capacity across all periods, with high sensitivity (84.2-87.3%) but low-to-moderate specificity (61.5-67.6%). The qSOFA index initially had poor predictive power upon admission but improved after 4 days. FIB-4 had a statistically significant predictive capacity in all periods (P=0.001), but with limited clinical value (AUC, 0.639-0.698), and with low sensitivity and specificity. MuLBSTA and IKIRBY exhibited low predictive power at admission and no power after 6 days. In conclusion, in COVID-19 patients with high mortality rates, NEWS-2 and PSI consistently exhibited predictive power for death during hospital stay, with PSI demonstrating the best balance between sensitivity and specificity.

Keywords: COVID-19; ROC curve; hospital mortality; sensitivity; severity of illness index; specificity.

PubMed Disclaimer

Conflict of interest statement

The authors declare that they have no competing interests.

Figures

Figure 1
Figure 1
Flowchart of recruitment of hospitalized patients with COVID-19. COVID-19, Coronavirus Disease 2019.
Figure 2
Figure 2
Clinical parameters and treatment of patients with COVID-19 over the first 8 days of hospitalization. (A) Proportion of patients who died or survived. Compared with baseline data, the proportion of patients who died increased significantly on days 7 (P=0.037) and 8 (P=0.007). (B) Proportion of patients in critical condition significantly increased on days 6 (P=0.024), 7 (P=0.002) and 8 (P<0.001). (C) Proportion of patients requiring mechanical significantly increased ventilation on days 3 (P=0.034) and 4-8 (all P<0.001). (D) Proportion of patients with elevated serum D-dimer significantly increased on days 6-8 (all P<0.001). (E) Proportion of patients with elevated serum lactate dehydrogenase significantly increased on days 5 (P=0.016), 6 (P=0.036), 7 (P=0.029) and 8 (P=0.039). (F) Proportion of patients with elevated serum ferritin significantly increased on days 2-8 (all P<0.001). (G) Proportion of patients with elevated blood urea nitrogen significantly increased on days 6 (P=0.011), 7 (P=0.001) and 8 (P=0.006). (H) Proportion of patients requiring antibiotics significantly on days 3 (P=0.014), 4 (P=0.002) and 5-8 (all P<0.001). (I) Proportion of patients requiring amine therapy significantly increased on days 4 (P=0.009), 5 (P=0.002) and 6-8 (all P<0.001). All comparisons were conducted using mixed-effects multinomial logistic regression analysis. *P<0.05 vs. baseline. (J) National early warning score 2, (K) pneumonia severity index and (L) COVID-GRAM remained relatively constant over time, although their values differ depending on the reason for discharge from the hospital (improvement or death) *P<0.001. COVID-19, Coronavirus Disease 2019; COVID-GRAM, Critical Illness Risk Score.
Figure 3
Figure 3
Changes in predictive power (AUC, with 95% confidence intervals) during hospitalization for (A) National Early Warning Score, (B) Pneumonia Severity Index, (C) COVID-GRAM, (D) MuLBSTA, (E) Quick Sequential Organ Failure Assessment Score, (F) Kirby and (G) fibrosis-4 index. AUC, area under the curve; COVID-GRAM, Critical Illness Risk Score; MuLBSTA, the Viral Pneumonia Mortality Score.

Similar articles

Cited by

References

    1. Barbero MG. ¿Como ha afectado la COVID- 19 al sistema sanitario y la formación de los médicos y que hemos aprendido? Educación Médica. 2021;22:S1–S2.
    1. Pan American Health Organization: Epidemiological update: SARS-CoV-2 and other respiratory viruses in the Americas Region-8 January 2024., 2024.
    1. Delgado-Enciso I, Paz-Garcia J, Barajas-Saucedo CE, Mokay-Ramírez KA, Meza-Robles C, Lopez-Flores R, Delgado-Machuca M, Murillo-Zamora E, Toscano-Velazquez JA, Delgado-Enciso J, et al. Safety and efficacy of a COVID-19 treatment with nebulized and/or intravenous neutral electrolyzed saline combined with usual medical care vs. usual medical care alone: A randomized, open-label, controlled trial. Exp Ther Med. 2021;22(915) doi: 10.3892/etm.2021.10347. - DOI - PMC - PubMed
    1. Mascellino MT, Di Timoteo F, De Angelis M, Oliva A. Overview of the main anti-SARS-CoV-2 vaccines: Mechanism of action, efficacy and safety. Infect Drug Resist. 2021;14:3459–3476. doi: 10.2147/IDR.S315727. - DOI - PMC - PubMed
    1. Havers FP, Pham H, Taylor CA, Whitaker M, Patel K, Anglin O, Kambhampati AK, Milucky J, Zell E, Moline HL, et al. COVID-19-associated hospitalizations among vaccinated and unvaccinated adults 18 years or older in 13 US States, January 2021 to April 2022. JAMA Intern Med. 2022;182:1071–1081. doi: 10.1001/jamainternmed.2022.4299. - DOI - PMC - PubMed