An in depth study on patients admitted to hospital and later deceased with the coronavirus (COVID-19) infection revealed that the majority of cases showed one or more comorbidities. About 67.8 percent of reported deceased COVID-19 patients suffered from three or more pre-existing health conditions, and 17.9 percent from two conditions. Only in 2.9 percent of COVID-19 deaths no prior health conditions were recorded. More statistics and facts about the virus in Italy are available here. For a global overview visit Statista's webpage exclusively dedicated to coronavirus, its development, and its impact.

Total number of deaths, patients alive and percentage of total (%) and case fatality rate (%) for women and men in different ethnic groups, type of healthcare provision, presence of comorbidities and travel history.

This paper used Our World data for coronavirus disease-2019 (COVID-19) death count, test data, stringency, and transmission count and prepared a path model for COVID-19 deaths. We augmented the model with age structure-related variables and comorbidity via non-communicable diseases for 117 countries of the world for September 23, 2021, on a cross-section basis. A broad-based global quantitative study incorporating these two prominent channels with regional variation was unavailable in the existing literature. Old age and comorbidity were identified as two prime determinants of COVID-19 mortality. The path model showed that after controlling for these factors, one SD increase in the proportion of persons above 65, above 70, or of median age raised COVID-19 mortality by more than 0.12 SDs for 117 countries. The regional intensity of death is alarmingly high in South America, Europe, and North America compared with Oceania. After controlling for regions, the figure was raised to 0.213, which was even higher. For old age, the incremental coefficient was the highest for South America (0.564), and Europe (0.314), which were substantially higher than in Oceania. The comorbidity channel via non-communicable diseases illustrated that one SD increase in non-communicable disease intensity increased COVID-19 mortality by 0.132 for the whole sample. The regional figure for the non-communicable disease was 0.594 for South America and 0.358 for Europe compared with the benchmark region Oceania. The results were statistically significant at a 10% level of significance or above. This suggested that we should prioritize vaccinations for the elderly and people with comorbidity via non-communicable diseases like heart disease, cancer, chronic respiratory disease, and diabetes. Further attention should be given to South America and Europe, which are the worst affected regions of the world.

Rates of deaths involving the coronavirus (COVID-19) where individuals have specific comorbidities and sex in England.

Outcomes of male vs female adults with COVID-19.

The objective of this study was to compare the effect of diabetes and pathologies potentially related to diabetes on the risk of infection and death from COVID-19 among people from Highly-Developed-Country (HDC), including Italians, and immigrants from the High-Migratory-Pressure-Countries (HMPC). Among the population with diabetes, whose prevalence is known to be higher among immigrants, we compared the effect of body mass index among HDC and HMPC populations. A population-based cohort study was conducted, using population registries and routinely collected surveillance data. The population was stratified into HDC and HMPC, according to the place of birth; moreover, a focus was set on the South Asiatic population. Analyses restricted to the population with type-2 diabetes were performed. We reported incidence (IRR) and mortality rate ratios (MRR) and hazard ratios (HR) with 95% confidence interval (CI) to estimate the effect of diabetes on SARS-CoV-2 infection and COVID-19 mortality. Overall, IRR of infection and MRR from COVID-19 comparing HMPC with HDC group were 0.84 (95% CI 0.82–0.87) and 0.67 (95% CI 0.46–0.99), respectively. The effect of diabetes on the risk of infection and death from COVID-19 was slightly higher in the HMPC population than in the HDC population (HRs for infection: 1.37 95% CI 1.22–1.53 vs. 1.20 95% CI 1.14–1.25; HRs for mortality: 3.96 95% CI 1.82–8.60 vs. 1.71 95% CI 1.50–1.95, respectively). No substantial difference in the strength of the association was observed between obesity or other comorbidities and SARS-CoV-2 infection. Similarly for COVID-19 mortality, HRs for obesity (HRs: 18.92 95% CI 4.48–79.87 vs. 3.91 95% CI 2.69–5.69) were larger in HMPC than in the HDC population, but differences could be due to chance. Among the population with diabetes, the HMPC group showed similar incidence (IRR: 0.99 95% CI: 0.88–1.12) and mortality (MRR: 0.89 95% CI: 0.49–1.61) to that of HDC individuals. The effect of obesity on incidence was similar in both HDC and HMPC populations (HRs: 1.73 95% CI 1.41–2.11 among HDC vs. 1.41 95% CI 0.63–3.17 among HMPC), although the estimates were very imprecise. Despite a higher prevalence of diabetes and a stronger effect of diabetes on COVID-19 mortality in HMPC than in the HDC population, our cohort did not show an overall excess risk of COVID-19 mortality in immigrants.

BackgroundCoronavirus disease 2019 (COVID-19) causes life-threatening with the high-fatality rates and spreads with high-infectious disease worldwide. We aimed to systematically review the comorbidities and complications of COVID-19 that are associated with various disease severity, progression, and mortality in China, to provide contemporary and reliable estimates in settings with centralized isolation and hospitalization.MethodsIn this systematic review and meta-analysis, we searched four main English language databases, and four main Chinese language databases for observational studies published from inception to January 2022, to identify all the related comorbidities and complications of COVID-19, in the China region with centralized isolation and hospitalization, with disease severity, progression, and mortality. Literature search, data extraction, and quality assessment were independently conducted by two reviewers. We used the generalized linear mixed model to estimate pooled effect sizes for any comorbidities and complications, and subgroup in gender ratio was done to further address the potential heterogeneity.ResultsOverall, 187 studies describing 77,013 patients, namely, 54 different comorbidities and 46 various complications of COVID-19, were identified who met our inclusion criteria. The most prevalent comorbidities were hypertension [20.37% 95% CI (15.28–26.63), 19.29% (16.17–22.85), 34.72% (31.48–38.10), and 43.94% (38.94–49.06)] and diabetes [7.84% (5.78–10.54), 8.59% (7.25–10.16), 17.99% (16.29–19.84), and 22.68% (19.93–25.69)] in mild, moderate, severe, and critical cases. The most prevalent complications were liver injury [10.00% (1.39–46.72), 23.04% (14.20–35.13), and 43.48% (39.88–47.15)] in mild, moderate, and severe cases, and acute respiratory distress syndrome [ARDS; 94.17% (20.78–99.90)] and respiratory failure [90.69% (28.08–99.59)] in critical cases. Renal insufficiency [odds ratio (OR) 17.43 (6.69–45.43)] in comorbidities and respiratory failure [OR 105.12 (49.48–223.33)] in complications were strongly associated in severe/critical than in mild/moderate cases. The highest estimated risk in intensive care unit (ICU) admission, progression, and mortality was an autoimmune disease, nervous system disease, and stroke in comorbidities, shock, and ARDS in complications.ConclusionComorbidities and complications in inpatients with COVID-19 were positively associated with increased risk in severe and critical cases, ICU admission, exacerbation, and death during centralized isolation and hospitalization. Prompt identification of comorbidities and complications in inpatients with COVID-19 can enhance the prevention of disease progression and death and improve the precision of risk predictions.

The Public Health Research Database (PHRD) is a linked asset which currently includes Census 2011 data; Mortality Data; Hospital Episode Statistics (HES); GP Extraction Service (GPES) Data for Pandemic Planning and Research data. Researchers may apply for these datasets individually or any combination of the current 4 datasets.

The purpose of this dataset is to enable analysis of deaths involving COVID-19 by multiple factors such as ethnicity, religion, disability and known comorbidities as well as age, sex, socioeconomic and marital status at subnational levels. 2011 Census data for usual residents of England and Wales, who were not known to have died by 1 January 2020, linked to death registrations for deaths registered between 1 January 2020 and 8 March 2021 on NHS number. The data exclude individuals who entered the UK in the year before the Census took place (due to their high propensity to have left the UK prior to the study period), and those over 100 years of age at the time of the Census, even if their death was not linked. The dataset contains all individuals who died (any cause) during the study period, and a 5% simple random sample of those still alive at the end of the study period. For usual residents of England, the dataset also contains comorbidity flags derived from linked Hospital Episode Statistics data from April 2017 to December 2019 and GP Extraction Service Data from 2015-2019.

Overview

The COVID-19 Patient Recovery Dataset is a synthetic collection of anonymized records for around 70,000 COVID-19 patients. It aims to assist with classification tasks in machine learning and epidemiological research. The dataset includes detailed clinical and demographic information, such as symptoms, existing health issues, vaccination status, COVID-19 variants, treatment details, and outcomes related to recovery or mortality. This dataset is great for predicting patient recovery (recovered), mortality (death), disease severity (severity), or the need for intensive care (icu_admission) using algorithms like Logistic Regression, Random Forest, XGBoost, or Neural Networks. It also allows for exploratory data analysis (EDA), statistical modeling, and time-series studies to find patterns in COVID-19 outcomes.
The data is synthetic and reflects realistic trends found in public health data, based on sources like WHO reports. It ensures privacy and follows ethical guidelines. Dates are provided in Excel serial format, meaning 44447 corresponds to September 8, 2021, and can be converted to standard dates using Python’s datetime or Excel. With 70,000 records and 28 columns, this dataset serves as a valuable resource for data scientists, researchers, and students interested in health-related machine learning or pandemic trends.

Data Source and Collection

Source: Synthetic data based on public health patterns from sources like the World Health Organization (WHO). It includes placeholder URLs.
Collection Period: Simulated from early 2020 to mid-2022, covering the Alpha, Delta, and Omicron waves.
Number of Records: 70,000.
File Format: CSV, which works with Pandas, R, Excel, and more.
Data Quality Notes:

About 5% of the values are missing in fields like symptoms_2, symptoms_3, treatment_given_2, and date.
There are rare inconsistencies, such as between recovery/death flags and dates, which may need some preprocessing.
Unique, anonymized patient IDs.

Key Column Details

patient_id: Unique identifier (e.g., P000001).
country: Reporting country (e.g., India, USA, Brazil, Germany, China, Pakistan, South Africa, UK).
region/state: Sub-national region (e.g., Sindh, California, São Paulo, Beijing).
date_reported, date_of_recovery, date_of_death: Excel serial dates (convert using datetime(1899,12,30) + timedelta(days=value)).
age: Patient age (1–100 years).
gender: Male or Female.
comorbidities: Pre-existing conditions (e.g., Diabetes, Hypertension, Cancer, Heart Disease, Asthma, None).
symptoms_1, symptoms_2, symptoms_3: Reported symptoms (e.g., Cough, Fever, Fatigue, Loss of Smell, Sore Throat, or empty).
severity: Case severity (Mild, Moderate, Severe, Critical).
hospitalized, icu_admission, ventilator_support: Binary (1 = Yes, 0 = No).
vaccination_status: None, Partial, Full, or Booster.
variant: COVID-19 variant (Omicron, Delta, Alpha).
treatment_given_1, treatment_given_2: Treatments administered (e.g., Antibiotics, Remdesivir, Oxygen, Steroids, Paracetamol, or empty).
days_to_recovery: Days from report to recovery (5–30, or empty if not recovered).
recovered, death: Binary outcomes (1 = Yes, 0 = No; generally mutually exclusive).
tests_conducted: Number of tests (1–5).
test_type: PCR or Antigen.
hospital_name: Fictional hospital (e.g., Aga Khan, Mayo Clinic, NHS Trust).
doctor_assigned: Fictional doctor name (e.g., Dr. Smith, Dr. Müller).
source_url: Placeholder.

Summary Statistics

Total Patients: 70,000.
Age: Mean ~50 years, Min 1, Max 100, evenly distributed.
Gender: ~50% Male, ~50% Female.
Top Countries: USA (20%), India (18%), Brazil (15%), China (12%), Germany (10%).
Comorbidities: Diabetes (25%), Hypertension (20%), Cancer (15%), Heart Disease (15%), Asthma (10%), None (15%).
Severity: Mild (60%), Moderate (25%), Severe (10%), Critical (5%).
Recovery Rate: ~60% recovered (recovered=1), ~30% deceased (death=1), ~10% unresolved (both 0).
Vaccination: None (40%), Full (30%), Partial (15%), Booster (15%).
Variants: Omicron (50%), Delt...

IntroductionWhile it is established that vaccination reduces risk of hospitalization, there is conflicting data on whether it improves outcome among hospitalized COVID-19 patients. This study evaluated clinical outcomes and antibody (Ab) responses to severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) infection/vaccines in patients with acute respiratory failure (ARF) and various comorbidities.MethodsIn this single-center study, 152 adult patients were admitted to Ohio State University hospital with ARF (05/2020 – 11/2022) including 112 COVID-19-positive and 40 COVID-19-negative patients. Of the COVID-19 positive patients, 23 were vaccinated for SARS-CoV-2 (Vax), and 89 were not (NVax). Of the NVax COVID-19 patients, 46 were admitted before and 43 after SARS-CoV-2 vaccines were approved. SARS-CoV-2 Ab levels were measured/analyzed based on various demographic and clinical parameters of COVID-19 patients. Additionally, total IgG4 Ab concentrations were compared between the Vax and NVax patients.ResultsWhile mortality rates were 36% (n=25) and 27% (n=15) for non-COVID-19 NVax and Vax patients, respectively, in COVID-19 patients mortality rates were 37% (NVax, n=89) and 70% (Vax, n=23). Among COVID-19 patients, mortality rate was significantly higher among Vax vs. NVax patients (p=0.002). The Charlson’s Comorbidity Index score (CCI) was also significantly higher among Vax vs. NVax COVID-19 patients. However, the mortality risk remained significantly higher (p=0.02) when we compared COVID-19 Vax vs. NVax patients with similar CCI score, suggesting that additional factors may increase risk of mortality. Higher levels of SARS-CoV-2 Abs were noted among survivors, suggestive of their protective role. We observed a trend for increased total IgG4 Ab, which promotes immune tolerance, in the Vax vs. NVax patients in week 3.ConclusionAlthough our cohort size is small, our results suggest that vaccination status of hospital-admitted COVID-19 patients may not be instructive in determining mortality risk. This may reflect that within the general population, those individuals at highest risk for COVID-19 mortality/immune failure are likely to be vaccinated. Importantly, the value of vaccination may be in preventing hospitalization as opposed to stratifying outcome among hospitalized patients, although our data do not address this possibility. Additional research to identify factors predictive of aberrant immunogenic responses to vaccination is warranted.

Background and aims There is poor knowledge on characteristics, comorbidities and laboratory measures associated with risk for adverse outcomes and in-hospital mortality in European Countries. We aimed at identifying baseline characteristics predisposing COVID-19 patients to in-hospital death. Methods and results Retrospective observational study on 3894 patients with SARS-CoV-2 infection hospitalized from February 19th to May 23rd, 2020 and recruited in 30 clinical centres distributed throughout Italy. Machine learning (random forest)-based and Cox survival analysis. 61.7% of participants were men (median age 67 years), followed up for a median of 13 days. In-hospital mortality exhibited a geographical gradient, Northern Italian regions featuring more than twofold higher death rates as compared to Central/Southern areas (15.6% vs 6.4%, respectively). Machine learning analysis revealed that the most important features in death classification were impaired renal function, elevated C reactive protein and advanced age. These findings were confirmed by multivariable Cox survival analysis (hazard ratio (HR): 8.2; 95% confidence interval (CI) 4.6-14.7 for age ?85 vs 18-44 y); HR = 4.7; 2.9-7.7 for estimated glomerular filtration rate levels <15 vs ? 90 mL/min/1.73 m2; HR = 2.3; 1.5-3.6 for C-reactive protein levels ?10 vs ? 3 mg/L). No relation was found with obesity, tobacco use, cardiovascular disease and related-comorbidities. The associations between these variables and mortality were substantially homogenous across all sub-groups analyses. Conclusions Impaired renal function, elevated C-reactive protein and advanced age were major predictors of in-hospital death in a large cohort of unselected patients with COVID-19, admitted to 30 different clinical centres all over Italy.

This systematic review aimed to provide an overview of the clinical profile and outcome of COVID-19 infection in patients with hemoglobinopathy. The rate of COVID-19 mortality and its predictors were also identified. A systematic search was conducted in accordance with PRISMA guidelines in five electronic databases (PubMed, Scopus, Web of Science, Embase, WHO COVID-19 database) for articles published between 1st December 2019 to 31st October 2020. All articles with laboratory-confirmed COVID-19 cases with underlying hemoglobinopathy were included. Methodological quality was assessed using the Joanna Briggs Institute (JBI) critical appraisal checklists. Thirty-one articles with data on 246 patients with hemoglobinopathy were included in this review. In general, clinical manifestations of COVID-19 infection among patients with hemoglobinopathy were similar to the general population. Vaso-occlusive crisis occurred in 55.6% of sickle cell disease patients with COVID-19 infection. Mortality from COVID-19 infection among patients with hemoglobinopathy was 6.9%. After adjusting for age, gender, types of hemoglobinopathy and oxygen supplementation, respiratory (adj OR = 89.63, 95% CI 2.514–3195.537, p = 0.014) and cardiovascular (adj OR = 35.20, 95% CI 1.291–959.526, p = 0.035) comorbidities were significant predictors of mortality. Patients with hemoglobinopathy had a higher mortality rate from COVID-19 infection compared to the general population. Those with coexisting cardiovascular or respiratory comorbidities require closer monitoring during the course of illness. More data are needed to allow a better understanding on the clinical impact of COVID-19 infections among patients with hemoglobinopathy.Clinical Trial Registration:https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42020218200.

Frailty is a syndrome of increased vulnerability to incomplete resolution of homeostasis (healing or return to baseline function) following a stressor event (such as an infection or fall) and it is associated with poor outcomes including increased mortality and reduced quality of life. The pathophysiology of frailty is poorly understood. Age and frailty have been proven to be independently predictive of outcomes in patients admitted with an acute illness. In COVID-19, routine frailty identification informed decision making about treatment.

This dataset from 01-03-2020 to 01-04-2022 of 327,346 patients admitted during all waves of the COVID pandemic both with and without COVID-19. The dataset includes granular demographics, frailty scores, physiology and vital signs, all care contacts and investigations (including imaging), all medications including dose and routes, care outcomes, length of stay and outcomes including readmission and mortality.

Geography: The West Midlands (WM) has a population of 6 million & includes a diverse ethnic & socio-economic mix. UHB is one of the largest NHS Trusts in England, providing direct acute services & specialist care across four hospital sites, with 2.2 million patient episodes per year, 2750 beds & > 120 ITU bed capacity. UHB runs a fully electronic healthcare record (EHR) (PICS; Birmingham Systems), a shared primary & secondary care record (Your Care Connected) & a patient portal “My Health”. 

Data set availability:  Data access is available via the PIONEER Hub for projects which will benefit the public or patients.  This can be by developing a new understanding of disease, by providing insights into how to improve care, or by developing new models, tools, treatments, or care processes.  Data access can be provided to NHS, academic, commercial, policy and third sector organisations.  Applications from SMEs are welcome.  There is a single data access process, with public oversight provided by our public review committee, the Data Trust Committee.  Contact pioneer@uhb.nhs.uk or visit www.pioneerdatahub.co.uk for more details. 

Available supplementary data: Matched controls; ambulance and community data. Unstructured data (images).  We can provide the dataset in OMOP and other common data models and can build synthetic data to meet bespoke requirements. 

Available supplementary support: Analytics, model build, validation & refinement; A.I. support.  Data partner support for ETL (extract, transform & load) processes.  Bespoke and “off the shelf” Trusted Research Environment (TRE) build and run.  Consultancy with clinical, patient & end-user and purchaser access/ support.  Support for regulatory requirements.  Cohort discovery. Data-driven trials and “fast screen” services to assess population size.

In Italy, approximately ******* deaths were registered in 2022. According to the data, ischemic heart diseases were the most common cause of death in the country, with ****** cases registered, closely followed by cerebrovascular diseases. COVID-19 was the third illness causing the largest number of deaths in Italy. COVID-19 death comorbidities Most patients admitted to the hospital and later deceased with the coronavirus (COVID-19) infection showed one or more comorbidities. Hypertension was the most common pre-existing health condition, detected in **** percent of patients who died after contracting the virus. Type 2-diabetes, ischemic heart disease, and atrial fibrillation were also among the most common comorbidities in COVID-19 patients who lost their lives. Cancer deaths The number of people who died from a tumor in Italy decreased constantly between 2006 and 2021. Indeed, the rate of deaths due to cancer among Italians dropped from **** deaths per 10,000 inhabitants in 2006 to **** in 2021. The Italian region with the highest cancer mortality rate was Campania, followed by Sardinia, and Sicily.

Summary of demographics and characteristics of male vs female adults with COVID-19.

These indicators are designed to accompany the SHMI publication. As well as information on the main condition the patient is in hospital for (the primary diagnosis), the SHMI data contain up to 19 secondary diagnosis codes for other conditions the patient is suffering from. This information is used to calculate the expected number of deaths. 'Depth of coding' is defined as the number of secondary diagnosis codes for each record in the data. A higher mean depth of coding may indicate a higher proportion of patients with multiple conditions and/or comorbidities, but may also be due to differences in coding practices between trusts. Contextual indicators on the mean depth of coding for elective and non-elective admissions are produced to support the interpretation of the SHMI. Notes: 1. As of the July 2020 publication, COVID-19 activity has been excluded from the SHMI. The SHMI is not designed for this type of pandemic activity and the statistical modelling used to calculate the SHMI may not be as robust if such activity were included. Activity that is being coded as COVID-19, and therefore excluded, is monitored in the contextual indicator 'Percentage of provider spells with COVID-19 coding' which is part of this publication. 2. Please note that there was a fall in the overall number of spells from March 2020 due to COVID-19 impacting on activity for England and the number has not returned to pre-pandemic levels. Further information at Trust level is available in the contextual indicator ‘Provider spells compared to the pre-pandemic period’ which is part of this publication. 3. There is a shortfall in the number of records for The Princess Alexandra Hospital NHS Trust (trust code RQW). Values for this trust are based on incomplete data and should therefore be interpreted with caution. 4. Frimley Health NHS Foundation Trust (trust code RDU) stopped submitting data to the Secondary Uses Service (SUS) during June 2022 and did not start submitting data again until April 2023 due to an issue with their patient records system. This is causing a large shortfall in records and values for this trust should be viewed in the context of this issue. 5. There is a high percentage of invalid diagnosis codes for Chesterfield Royal Hospital NHS Foundation Trust (trust code RFS), Milton Keynes University Hospital NHS Foundation Trust (trust code RD8), and West Suffolk NHS Foundation Trust (trust code RGR). Values for these trusts should therefore be interpreted with caution. 6. A number of trusts are now submitting Same Day Emergency Care (SDEC) data to the Emergency Care Data Set (ECDS) rather than the Admitted Patient Care (APC) dataset. The SHMI is calculated using APC data. Removal of SDEC activity from the APC data may impact a trust’s SHMI value and may increase it. More information about this is available in the Background Quality Report. 7. East Kent Hospitals University NHS Foundation Trust (trust code RVV) has a submission issue which is causing many of their patient spells to be duplicated in the HES Admitted Patient Care data. This means that the number of spells for this trust in this dataset are overstated by approximately 60,000, and the trust’s SHMI value will be lower as a result. Values for this trust should therefore be interpreted with caution. 8. Further information on data quality can be found in the SHMI background quality report, which can be downloaded from the 'Resources' section of this page.

BackgroundThe new coronavirus SARS-CoV-2 pandemic has been relatively less lethal in children; however, poor prognosis and mortality has been associated with factors such as access to health services. Mexico remained on the list of the ten countries with the highest case fatality rate (CFR) in adults. It is of interest to know the behavior of COVID-19 in the pediatric population. The aim of this study was to identify clinical and sociodemographic variables associated with mortality due to COVID-19 in pediatric patients.ObjectiveUsing National open data and information from the Ministry of Health, Mexico, this cohort study aimed to identify clinical and sociodemographic variables associated with COVID-19 mortality in pediatric patients.MethodA cohort study was designed based on National open data from the Ministry of Health, Mexico, for the period April 2020 to January 2022, and included patients under 18 years of age with confirmed SARS-CoV-2 infection. Variables analyzed were age, health services used, and comorbidities (obesity, diabetes, asthma, cardiovascular disease, immunosuppression, high blood pressure, and chronic kidney disease). Follow-up duration was 60 days, and primary outcomes were death, hospitalization, and requirement of intensive care. Statistical analysis included survival analysis, prediction models created using the Cox proportional hazards model, and Kaplan-Meier estimation curves.ResultsThe cohort included 261,099 cases with a mean age of 11.2 ± 4 years, and of these, 11,569 (4.43%) were hospitalized and 1,028 (0.39%) died. Variables associated with risk of mortality were age under 12 months, the presence of comorbidities, health sector where they were treated, and first wave of infection.ConclusionBased on data in the National database, we show that the pediatric fatality rate due to SARS-CoV-2 is similar to that seen in other countries. Access to health services and distribution of mortality were heterogeneous. Vulnerable groups were patients younger than 12 months and those with comorbidities.

BackgroundThis study aims to develop a machine learning model to predict the 30-day mortality risk of hospitalized COVID-19 patients while leveraging federated learning to enhance data privacy and expand the model's applicability. Additionally, SHapley Additive exPlanations (SHAP) values were utilized to assess the impact of comorbidities on mortality.MethodsA retrospective analysis was conducted on 6,321 clinical records of hospitalized COVID-19 patients between January 2021 and October 2022. After excluding cases involving patients under 18 years of age and non-Omicron infections, a total of 4,081 records were analyzed. Key features included three demographic data, six vital signs at admission, and 79 underlying comorbidities. Four machine learning models were compared, including Lasso, Random Forest, XGBoost, and TabNet, with XGBoost demonstrating superior performance. Federated learning was implemented to enable collaborative model training across multiple medical institutions while maintaining data security. SHAP values were applied to interpret the contribution of each comorbidity to the model's predictions.ResultsA subset of 2,156 records from the Taipei branch was used to evaluate model performance. XGBoost achieved the highest AUC of 0.96 and a sensitivity of 0.94. Two versions of the XGBoost model were trained: one incorporating vital signs, suitable for emergency room applications where patients come in with unstable vital signs, and another excluding vital signs, optimized for outpatient settings where we encounter patients with multiple comorbidities. After implementing federated learning, the AUC of the Taipei cohort decreased to 0.90, while the performance of other cohorts improved to meet the required standards. SHAP analysis identified comorbidities including diabetes mellitus, cerebrovascular disease, and chronic lung disease to have a neutral or even protective association with 30-day mortality.ConclusionXGBoost outperformed other models making it a viable tool for both emergency and outpatient settings. The study underscores the importance of chronic disease assessment in predicting COVID-19 mortality, revealing some comorbidities such as diabetes mellitus, cerebrovascular disease and chronic lung disease to have protective association with 30-day mortality. These findings suggest potential refinements in current treatment guidelines, particularly concerning high-risk conditions. The integration of federated learning further enhances the model's clinical applicability while preserving patient privacy.

Participants A total of 203 patients with type 2 diabetes mellitus who attended nine primary care areas in four Cuban provinces belonging to different regions of the country (Pinar del Río, La Habana, Ciego de Ávila and Santiago de Cuba) participated in the study. Participants were selected by non-probabilistic sampling based on the following inclusion criteria: 1. have a diagnosis of type 2 DM according to the World Health Organization criteria, 2. be older than 18 years old, 3. be patients of the health care areas mentioned above, and 4. be willing to participate in the study and to sign the informed consent form.

Patients with mental illness, cognitive deficit (dementia, psychosis or mental disability) or other apparent condition that prevents understanding and completion of the questionnaire were excluded. Although retrospective data on infection rates in diabetic patients suggest that people with type 1 DM are at higher risk for infectious diseases in general, and death rates are similar to those of people with type 2 DM,this study focused on the latterfortwo main reasons. First, patients with type 1 DM are mostly children and young people and the prevalence of this type of diabetes is lower compared to type 2 DM, which leads to a lower number of patients seen in consultation and primary health care. Second, the study was conducted in the context of the COVID-19 pandemic and patients with type 2 DM were the most accessible population to be surveyed by the research team in primary care areas.

The minimum sample size was calculated with the Soper software package for a multiple regression study, according to the desired probability level (α=0.05), the number of predictors in the model (18 predictors), the anticipated effect size (f2=0.15) and the desired statistical power level (1- β=0.80). The software suggested a minimum number of 118 participants; however, the final number was higher than the minimum required.

Instruments Socio-demographic and health information A questionnaire was developed specifically for this study, where participants were asked to provide information about their sex, age, educational level, type of work, cohabitation, marital status, presence of chronic complications, presence of comorbidities, family or friends infected with COVID-19, and time since diagnosis with DM.

Concern about COVID-19 contagion We used the COVID-19 contagion concern scale (PRECOVID-19) originally developed for the general population, which assesses worry about becoming infected with COVID-19 and its impact on people’s mood and ability to perform daily activities. In this study we used the version validated for Cuban patients with diabetes, which consists of 5 items. All items have 4 Likert-type response options, ranging from 1=never or rarely to 4=almost all the time. The PRE-COVID-19 has a unidimensional structure, where the total score is calculated by adding the scores of each of the 5 items. Higher scores indicate greater concern about becoming infected with COVID19. The reliability of the PRE-COVID-19 for this study was very good (ω=0.91).

Blood glucose level Fasting blood glucose values were obtained from the patients’ clinical histories and from blood tests performed in the last three months in laboratories equipped for this purpose. Based on this, poor glycemic control was determined as fasting blood glucose greaterthan or equal to 7 mmol/L (126 mg/dl) in the last three months and good control as figures below this value. The criterion based on glycosylated hemoglobin (HbA1c) could not be used because it is not a test regularly available in the primary health care system where the survey was applied. Other control criteria using continuous glucose monitoring systems were not possible either, as they are not generally available for patients with DM living in Cuba.

Procedure The questionnaire was applied by properly trained researchers, who complied with strict COVID-19 prevention health protocols, between the months of January and April 2021. The questionnaire was administered during patients’ visits to primary care centers or in their homes. During this period of time, the fight against COVID-19 in Cuba suffered some setbacks, characterized by an increase in the number of infected people, even higherthan that observed during the first stage of the disease, in 2020. Thus, during those dates, more than 64,414 positive diagnoses and 384 deaths were reported in the country. Participation was voluntary and without any financial compensation. Participants signed the informed consent form and were informed that they could withdraw from the study at any time. Similarly, the reliability of the data was guaranteed. The study protocol was approved by the Ethics Committee of theUniversidad Privada delNorte in Peru (registration number: 20213002).

Data Analysis The frequencies and percentages of the categorical variables included in the model were examined. In the case of the outcome variable (concern about COVID-19 contagion), the mean±standard deviation (SD) was calculated for the total sample. These values were then also calculated for each category of each variable. For inferential purposes, bivariate associations were examined with a series of analyses of variance (ANOVA). The assumption of homoscedasticity was reasonably well met in most cases; however, a possible noncompliance with the assumption of normality of the residuals was observed. Therefore, we repeated the analyses after a power transformation of the outcome variable. Since the results were practically identical with both procedures, only those obtained with the variable in its original form are reported.

Variables that reached statistical significance (p<.05) in the ANOVAs were selected as potential predictors in a linear regression. Crude (simple) regressions were run, which replicated the ANOVAs but also allowed for a more detailed examination of between-group differences. Finally, a fitted (multiple) regression was run with all predictors simultaneously. Statistical significance was judged from the 95% CIs, which provide a set of possible values of the coefficient in the population. A CI that does not include zero is equivalent to a p<.05.

BackgroundThe effective treatment of coronavirus disease 2019 (COVID-19) remains unclear. We reported successful use of high-dose intravenous immunoglobulin (IVIg) in cases of severe COVID-19, but evidence from larger case series is still lacking.MethodsA multi-center retrospective study was conducted to evaluate the effectiveness of IVIg administered within two weeks of disease onset at a total dose of 2 g/kg body weight, in addition to standard care. The primary endpoint was 28-day mortality. Efficacy of high-dose IVIg was assessed by using the Cox proportional hazards regression model and the Kaplan-Meier curve adjusted by inverse probability of treatment weighting (IPTW) analysis, and IPTW after multiple imputation (MI) analysis.ResultsOverall, 26 patients who received high-dose IVIg with standard therapy and 89 patients who received standard therapy only were enrolled in this study. The IVIg group was associated with a lower 28-day mortality rate and less time to normalization of inflammatory markers including IL-6, IL-10, and ferritin compared with the control. The adjusted HR of 28-day mortality in high-dose IVIg group was 0.24 (95% CI 0.06–0.99, p<0.001) in IPTW model, and 0.27 (95% CI 0.10–0.57, p=0.031) in IPTW-MI model. In subgroup analysis, patients with no comorbidities or treated in the first week of disease were associated with more benefit from high-dose IVIg.ConclusionsHigh-dose IVIg administered in severe COVID-19 patients within 14 days of onset was linked to reduced 28-day mortality, more prominent with those having no comorbidities or treated at earlier stage.