Advanced Modeling of the Evolution of the Epidemiological Outbreak of SARS-CoV-2 Pandemic

Design. Retrospective observational study. The modeling will be based on the SARS-CoV-2 pandemic that started at the beginning of 2020.

Subjects of the study. Information will be collected on daily incidence data aggregated by age and sex for: tests performed, positive cases, hospital admissions and ICU admissions for SARS-CoV-2, hospital discharges and ICU discharges, recovered and mortality (in ICU, in hospital or in the community) of individuals with Coronavirus Disease of 2019 (COVID 19).

Criteria for inclusion. Of positive cases: Having a SARS-CoV-2 infection laboratory-confirmed by a positive result on the reverse transcriptase-polymerase chain reaction assay for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) or a positive antigen test from March 1, 2020 to January 9, 2022.

For hospital admissions: Hospital admissions since the start of the pandemic. Considering different episodes as a single admission when it comes to transfers from one center to another. Consider exclusively income due to the COVID19.

Exclusion criteria: Patients admitted for other reasons who have developed the disease during their hospital stay.

Variables. The data to be collected is aggregated data in the form of incidents. The population will be stratified into ten age groups (0 - 9, 10 - 19, ..., 70 - 79, 80 - 89, 90+) and by sex. Variables:

• Individuals in the study population by age.
• Number of new confirmed positive cases of COVID19 by age and day.
• Number of new hospital admissions due to COVID19 by age and day. Number of ICU admissions due to COVID19
• Number of total deaths from COVID19 by age and day.
• Number of hospital discharges (live patients) of patients who have been hospitalized for COVID19 by age and day (excluding transfers).
• Number of deaths in hospital due to COVID19 by age and day.
• Number of deaths in the ICU due to COVID19 by age and day.

The outcome variables that will be obtained from the proposed modeling are:

• Number of estimated positive COVID19 cases by age and day.
• Number of estimated COVID19 hospital admissions by age and day.
• Number of estimated total deaths due to COVID19 estimated by age and day.
• Number of estimated ICU admissions due to COVID19 estimated by age and day.

Analysis of data. The investigators will use P-splines and Negative Binomial Distribution. P-splines, or penalized splines, are a powerful tool for modeling nonlinear relationships in temporal data. By combining them with the negative binomial distribution, a model is obtained that is especially suitable for counting data with over-dispersion, as is the case with pandemic data.

Procedure:

• Data Collection: Daily data on positive cases, hospital admissions and ICU admissions will be obtained from the beginning of the pandemic until september 2022.
• Modeling: A P-splines model based on the negative binomial distribution will be fitted to the data. This model will be designed to capture temporal trends and seasonal patterns, as well as to handle the over-dispersion present in the data.
• Model with Random Effect for Day of the Week: Specifically for the prediction of hospital admissions, a random effect for the day of the week will be incorporated. This adjustment will be made because a systematic variability in income was identified depending on the day of the week. Incorporating this random effect significantly will improve the accuracy of the model for this variable.
• Prediction: Predictions will be made for two time horizons: short term (1 and 2 days) and medium term (5 days). These predictions will allow us to anticipate the evolution of the pandemic and make informed decisions.

Validation of Predictions: To validate the accuracy and robustness of the predictions, a retrospective analysis will be carried out at different times (or waves) of the pandemic. Model predictions will be compared to actual observed data, and error metrics will be calculated to evaluate model performance.

Limitations. One of the limitations of the study is the possible loss of hospitalizations due to the disease considered and death (or recovery) in individuals whose temporal sequence of testing, admission and death (or recovery) has not followed the sequence used in searches carried out.

Ethical aspects. This study uses only anonymized information to meet its objectives. There is no data available to identify a patient.

The processing, communication and transfer of personal data of all participating persons complies with the provisions of the European Data Protection Regulation (EU2016/679) regarding the protection of natural persons with regard to processing. of personal data and the free circulation of these data and Organic Law 3/2018, of December 5, on the Protection of Personal Data and guarantee of digital rights. Virtually all of the data necessary for this study is aggregated data that in no case can be associated with individuals. All information will be treated absolutely confidentially.

Regarding obtaining informed consent from the patient, this research team proposes carrying out the study without asking the patient for informed consent. The reasons why this proposal is made are based on article 58 of Law 14/2007, of July 3, on Biomedical Research (""..exceptionally, coded or identified samples may be treated for the purposes of biomedical research without the consent of the source subject, when obtaining said consent is not possible or represents an unreasonable effort. In these cases, the favorable opinion of the corresponding Research Ethics Committee will be required. ")