Risk Factors for Admission into COVID-19 General Wards, Sub-Intensive and Intensive Care Units among SARS-CoV-2 Positive Subjects in the Municipality of Bologna, Italy

This is a retrospective cohort study aimed at identifying the risk factors for the hospitalization of patients with COVID-19 in the municipality of Bologna. A total of 32500 patients that tested positive for COVID-19 from February 28/2020 to October 13/2021 in the municipality of Bologna were included. The Kaplan-Meier method was used to estimate changes during time of ICU hospitalization for all patients as well as stratifying subjects by sex. A multi-state Cox’s proportional hazard model was fitted to investigate predictors of ICU and non-ICU hospitalization. Age, sex, calendar period of diagnosis, comorbidities and vaccination status of patients at the time of diagnosis were considered as candidate predictors. In general, male sex and advanced age resulted to be poor prognostic factors of COVID-19 outcomes. An exception was found for the over 80 age group which showed a decrease in the risk of ICU hospitalization compared to 70-79 (HR 0.57 95% CI 0.36 - 0.90 for DIAG →ICU; HR 0.40 95% CI 0.28 - 0.58 for HOSP→ICU). Having contracted the disease during the first wave was associated with a significant greater risk of hospitalization than during the second wave, while no difference in the risk of ICU admission was found between the second and third waves. Fully immunized patients showed a significant decrease in the risk of ICU and non-ICU hospitalization compared to the unvaccinated patients (HR 0.23 95% CI 0.16 - 0.31 for DIAG→HOSP; HR 0.10 95% CI 0.01 - 0.73 for DIAG→ICU). Chronic kidney failure and asthma were risk factors for non-ICU hospitalization. Diabetes and embolism were risk factors for both direct ICU and non-ICU hospitalization. The study of factors associated with a negative course of the COVID-19 disease allows to prevent fatal outcomes, establish priorities in the treatment of the disease and improve the management of hospital resources and the pandemic itself.

Since the beginning of 2020, the COVID-19 pandemic has become a threat to global 1 health. The SARS-CoV-2 virus that causes the COVID-19 disease was first found in 2 July 13, 2023 1/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. China in late 2019 [1]. Since that time it has spread fast all over the globe and Italy has 3 become one of the European most affected countries [2]. 4 The rapid diffusion of this new virus and the unpredictability of the disease course 5 has led to the development of numerous studies aimed at investigating prognostic 6 factors of COVID-19 outcomes in order to predict the severity of the disease. These 7 descriptive studies allow researchers and doctors to early identify patients most at risk 8 of disease aggravation and policy makers to better understand the evolution of this new 9 disease and implement adequate and efficient management and prevention measures. In 10 fact, the progression of COVID-19 disease can vary between asymptomatic disease, mild 11 or moderate flu-like illness and pneumonia with severe respiratory failure, that often 12 requires hospitalization in specialized wards and in severe cases in sub-intensive or 13 intensive care unit [3]. The course of the disease varies considerably between individuals 14 based on their demographic characteristics and previous health conditions, so it is 15 difficult to define the prognosis for each individual. Age and sex are confirmed and well 16 described prognostic factors, with a higher probability of developing serious illness and a 17 higher mortality rate in older subjects and males [4,5]. Several studies also argue that 18 the presence of pre-existing comorbidities, such as diabetes and hypertension, is a 19 determinant of poor prognosis [6,7]. Furthermore, the epidemiological dynamics of 20 COVID-19 has changed radically over the months due to, for example, variable climatic 21 conditions [8], the presence of more or less stringent restrictions imposed by 22 governments and the effect of vaccination [9]. From the beginning of the pandemic at 23 the end of February 2020 until fall 2021, Italy, for example, has recorded three major 24 waves of COVID-19 infections, interspersed with periods of less spread of the virus [10]. 25 In light of the foregoing, it is essential to gain a better understanding of key 26 prognostic factors of COVID-19 disease and quantify the strength of their association 27 with the patient’s likelihood of experiencing a critical event to identify patients at high 28 risk of clinical deterioration. Furthermore, in order to predict the evolution of the 29 pandemic and improve its management, an analysis of the temporal trends of the 30 disease and of the effect of the vaccination campaign on the severity of the latter is also 31 necessary. 32 With this purpose, our study considers 32500 cases of SARS-CoV-2 infection 33 diagnosed in the municipality of Bologna from February 28/2020 to October 13/2021 34 and through statistical tools analyzes the risk factors for hospitalization in general 35 COVID-19 wards and in COVID-19 intensive or sub-intensive care units. Of particular 36 interest were the age, sex and comorbidities of the patients, the vaccination status at 37 the time of diagnosis and the calendar period of diagnosis. 38

39 Study design and participants 40 We performed a retrospective cohort study on 32500 laboratory-confirmed cases of 41 COVID-19 in the municipality of Bologna from February 28/2020 to October 13/2021. 42 The analysis, performed between December 2021 and January 2023, describes the 43 progression of patients from the diagnosis of the disease to their admission to the 44 intensive or sub-intensive care unit (both referred to as ICU in the following), passing 45 through the eventual admission to COVID-19 general wards. An individual was 46 considered censored after ending the disease without experiencing hospitalization. 47 July 13, 2023 2/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint Data collection, setting and data organization 48 The data set analysed in this study was obtained with permission from the Local Health 49 Unit (AUSL, Azienda Unit` a Sanitaria Locale) of the municipality of Bologna. It 50 comprises various information about dates of diagnosis, hospital and ICU admission, 51 recovery and vaccination, socio demographic characteristics (age, sex, neighborhood of 52 residence,...) of the participants, presence of symptoms during the period of illness and 53 comorbidities. No missing data were present. We do not have reliable and certain data 54 on deaths. In this regard, our study stops at hospitalization in intensive care considered 55 as the only absorbing state. In particular, in the data set provided by the AUSL of 56 Bologna are listed all the entrances and exits from all hospital wards, COVID-19 57 dedicated and not, in the studied period of time, relating to the major hospitals in the 58 municipality of Bologna, namely Sant’Orsola Hospital, Maggiore Hospital, Bellaria 59 Hospital, Bentivoglio Hospital, Budrio Hospital, San Giovanni in Persiceto Hospital, 60 Bazzano Hospital, Porretta Hospital, Loiano Hospital and Vergato Hospital. 61 For the purpose of this study, the patients were divided into eight age groups (0-19, 62 20-29, 30-39, 40-49, 50-59, 60-69, 70-79 and over 80) based on the ISS ( Istituto Superiore 63 di Sanit` a) guide lines [11]. 64 Furthermore, the participants were divided into five groups according to their 65 calendar period of diagnosis [12]: 66 1. Wave 1 (W1): from February 28/2020 to May 13/2020; 67 2. Out-wave 1 (OW1): from May 14/2020 to November 06/2020; 68 3. Wave 2 (W2): from November 07/2020 to February 01/2021; 69 4. Wave 3 (W3): from February 02/2021 to April 28/2021; 70 5. Out-wave 2 (OW2): from April 29/04 to October 31/2021. 71 These periods were identified by observing the number of daily hospitalizations in 72 COVID-19 wards. In particular, the periods in which this number exceeds 500 are called 73 waves, while those in which is less than 500 are named out-waves (S1 Fig). 74 Finally, the patients were divided into three groups according to the vaccination 75 status at the time of diagnosis. There is a well-known delay between inoculation and 76 artificial immunization [13], and in particular it was found that in case of SARS-CoV-2 77 vaccines this is about 14 days after the first dose and 7 days after the second dose for 78 mRNA vaccines (Pfizer-BioNTech, Moderna) [14] and 14 days after the single dose of 79 Johnson&Johnson vaccine [15]. For this reason, we considered individuals as not 80 vaccinated (NV) up to 14 days after the first dose of mRNA vaccine or up to 14 days 81 after the single dose of the J&J vaccine, partially vaccinated (PV) as of 14 days after 82 their first dose of mRNA vaccine and totally vaccinated (TV) as of 7 days after their 83 second dose of mRNA vaccine or 14 days after being vaccinated with the single dose of 84 J&J [16,17]. People partially vaccinated who tested positive for COVID-19 after more 85 than 150 days from the administration of the first dose (after which immunity is 86 indeterminate) were excluded in order to take into account potentially late second doses 87 but, at the same time, not excessively overestimate the effectiveness of the partial 88 immunization [16]. Furthermore, people vaccinated with Astrazeneca S.P.A. vaccine 89 were not considered since the use of this vaccine had a troubled history in Italy in the 90 period under analysis [18]. In particular, several people who received the first dose of 91 Astrazeneca S.P.A vaccine were then given a second dose of an mRNA vaccine. 92 Therefore, we felt that it was difficult to estimate the effectiveness of the vaccine in the 93 case of individuals who received at least one dose of Astrazeneca S.P.A vaccine and we 94 exclude them from the analysis. Therefore the final number of patients considered is 95 July 13, 2023 3/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint equal to 32500, i.e. the total number of COVID cases in the municipality of Bologna in 96 the period considered (34016) minus the number of patients who received at least one 97 dose of the Astrazeneca S.P.A vaccine (1516). 98 Statistical analysis 99 First of all, the data have been reworked in order to make them suitable for statistical 100 analysis. In particular, all the independent variables (sex, age group, calendar period of 101 diagnosis, comorbidities and vaccination status) were considered as categorical variables 102 and dummy coded, i.e. 1 if the individual presents a certain characteristics and 0 103 otherwise. Such pre-processing was performed using the package pandas v1.4.1 [19] in 104 Python v3.8.8. 105 Hence, the Kaplan-Meier method was used to estimate the probability of avoiding 106 hospitalization in ICU (considered as the event) at any time after the diagnosis of 107 COVID-19 and in particular the difference in this probability was observed between 108 males and females. Log rank test was used to test any difference in survival probability 109 among the two sexes. This analysis was performed using the packages lifelines 110 v0.26.4 [20] and matplotlib v3.5.1 for the visualization [21] in Python. 111 Subsequently, we built a multi-state model (Fig 1) to investigate the determinants of 112 risk for three possible transitions that patients can experience during the disease, that is 113 from COVID-19 diagnosis (DIAG) to hospitalization in non-ICU COVID-19 specialised 114 wards (HOSP); from DIAG to admission to ICU (ICU) and from HOSP to ICU. A 115 multivariate Cox’s proportional hazard model, was fit to each transition using age 116 groups, calendar period of diagnosis, vaccination status and pathologies as predictor 117 factors. 118 The Cox’s model is expressed by the hazard function denoted by h(t) (Eq (1)) which 119 can be interpreted as the instantaneous risk of experiencing the event of interest at time 120 t: 121 h(t) = h0 × e(b1x1+...+bpxp) (1) where the ( x1,...,xp) is the set of predictor factors (covariates), the coefficients ( b1,...,bp) 122 measure the impact of covariates on the event realization and h0 is the baseline hazard 123 function, that corresponds to the value of the hazard when all the xi are equal to 0. A 124 value of bi greater than 0, or equivalently an hazard ratio HR = ebi greater than 1, 125 indicates that as the value of the i-th covariate increases, the event hazard increases as 126 well. On the other hand, a value of bi smaller than 0 (HR < 1), indicates that the i-th 127 covariate is negatively associated with the event probability. 128 Therefore, through a multivariate Cox’s model we studied the effect of sex, age 129 group, calendar period of diagnosis, vaccination status at the time of diagnosis and 130 pathologies on the probability of experiencing any of the three transitions. We have 131 chosen W2 as reference period in the analyses in order to better observe the differences 132 between the beginning of the pandemic that caught the whole world unprepared (W1 133 and OW1) and a more advanced phase of the same characterized by a better knowledge 134 of the virus and the launch of the vaccination campaign (W3 and OW2). We checked 135 the Cox proportional hazard assumptions by Schoenfeld residuals test and plot. This 136 statistical analysis was conducted in R v4.1.1 using the packages survminer v0.4.9 [22] 137 and survival v3.5-5 [23]. Specifically, the multivariate Cox’s model was fit by means of 138 the function ”coxph” of the survival package with the default parameters. 139 Given the limited number of performed hypothesis tests p-values were not adjusted 140 for multiple testing. Furthermore, p-values less than 0.05 were considered statistically 141 significant. In particular, we assign one star ( *) if p < 0.05, two stars ( **) if p < 0.01 142 and three stars ( ***) if p < 0.001. 143 July 13, 2023 4/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint Fig 1. States and tran5itions for the multi-state model.

and Discussion 144 Description of study cohort 145 Our study cohort was composed of 32500 people, who tested positive for COVID-19 146 between February 28/2020 and October 13/2021 in the municipality of Bologna, i.e. the 147 total number of COVID cases in the municipality of Bologna in the period considered 148 (34016) minus the number of patients who received at least one dose of the Astrazeneca 149 S.P.A vaccine (1516). 150 Overall, 16696 (51.37%) of the participants were females, while 15804 ( 48.63%) were 151 males. 152 Based on ISS ( Istituto Superiore di Sanit` a) guide lines [11], the following age groups 153 were considered for the analysis: 0-19, 20-29, 30-39, 40-49, 50-59, 60-69, 70-79 and over 154 80 years old. 155 Of all participants, 1497 (4 .61%) have contracted COVID-19 during W1, 4090 156 (12.58%) during OW1, 10017 (30 .82%) during W2, 12376 (38.08%) during W3 and the 157 remaining 4520 (13.91%) during OW2. 158 Regarding hospital admissions, 29423 individuals (90 .53%) did not require 159 hospitalization, 2475 (7.62%) were hospitalised but did not enter ICU and 602 (1.85%) 160 entered the ICU. 161 Kaplan-Meier curves of ICU hospitalization 162 Table 1 shows the percentages of patients, divided by sex, age group, calendar period of 163 diagnosis, vaccination status and comorbidities who were admitted to ICU during the 164 infection and those of censored people, that is who left the analysis without having 165 experienced such hospitalization. All patients were followed up from the first positive 166 COVID-19 test until censored or ICU entry (event). Of the 602 registered events, 586 167 (97.34%) occurred in less than 14 days following initial COVID-19 diagnosis. The 168 Kaplan-Meier estimation technique was used to calculate the probability of not being 169 admitted to ICU as days pass from the diagnosis (Fig 2 - A). As expected, the overall 170 July 13, 2023 5/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint Kaplan-Meier curve decreases rapidly during the first 14 days after COVID-19 diagnosis 171 showing that most entries in ICU occurred during this time. Separate Kaplan-Meier 172 curves stratified by sex (Fig 2 - B) show a statistically significant difference between 173 males and females in the probability of not being admitted to ICU. In particular, the 174 latter is higher in women at any time after COVID-19 diagnosis. A similar result was 175 found in several articles [5,24,25], confirming male sex as a poor prognostic factor. 176 Fig 2. Kaplan-Meier curves. Overall (A) and stratified by sex (B) estimate of the probability of not being admitted to ICU during COVID-19 infection. July 13, 2023 6/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint T able 1. Distribution of patients by sex, age group, calendar period of diagnosis, vaccination status and pathologies. Infection ICU Censored No. No. % No. % T otal 32500 602 1.85 31898 98.15 Sex Male 15804 403 2.55 15401 97.45 Female 16696 199 1.19 16497 98.81 Age group 0-19 5521 5 0.10 5516 99.90 20-29 3942 7 0.18 3935 99.82 30-39 4688 27 0.58 4661 99.42 40-49 5437 60 1.10 5377 98.90 50-59 5461 115 2.11 5346 97.89 60-69 2952 158 5.35 2794 94.65 70-79 1833 146 7.97 1687 92.03 over 80 2666 84 3.15 2582 96.85 Calendar period of diagnosis W1 1497 70 4.68 1427 95.32 OW1 4090 85 2.08 4005 97.92 W2 10017 179 1.79 9838 98.21 W3 12376 236 1.91 12140 98.09 OW2 4520 32 0.71 4488 99.29 V accination status NV 30938 596 1.93 30342 98.07 PV 417 5 1.20 412 98.80 TV 1145 1 0.10 1144 99.90 Pathology Essential hypertension 1380 71 5.14 1309 94.86 Hypertensive heart disease 456 30 6.58 426 93.42 Diabetes mellitus 1230 103 8.37 1127 91.63 Hypercholesterolemia 384 18 4.69 365 95.05 Malignant neoplasm 1177 45 3.82 1132 96.18 Multiple sclerosis 68 0 0.00 68 100.00 Hypothyroidism 620 66 10.65 554 89.35 Glaucoma 256 11 4.30 245 95.70 Toxic diffuse goiter 62 1 1.61 61 98.39 Epilepsy 78 3 3.85 75 96.15 Heart and pulmonary circulation diseases 872 59 6.77 813 93.23 Chronic kidney failure 184 14 7.61 170 92.39 Celiac disease 170 0 0.00 170 100.00 Asthma 357 8 2.24 349 97.76 Rheumatoid arthritis 108 9 8.33 99 91.67 Chronic Hepatitis 301 15 4.98 286 95.02 Embolism and thrombosis of other veins 145 18 12.41 127 87.59 Psoriasis 52 0 0.00 52 100.00 Hashimoto’s thyroiditis 320 3 0.94 317 99.06 Ulcerative colitis 106 1 0.94 105 99.06 Parkinson’s disease 30 2 6.66 28 93.34 Pituitary dwarfism 58 0 0.00 58 100.00 Endometriosis (III - IV ASRM stage) 40 0 0.00 40 100.00 July 13, 2023 7/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint Multi-state Cox’s Regression model 177 A multivariate and multi-state Cox’s proportional hazard model was fitted to identify 178 the factors associated to the risk of experiencing three potential transitions during the 179 course of the disease. The scheme of the model’s events and transitions is represented in 180 Fig 1. Three transitions were studied: from COVID-19 diagnosis to hospitalization in 181 any COVID-19 ward except ICU (DIAG →HOSP); from COVID-19 diagnosis to 182 hospitalization in ICU (DIAG →ICU); from admission to any non-ICU COVID-19 ward 183 to admission to ICU (HOSP →ICU). The percentages of patients divided by gender, age, 184 period of diagnosis, vaccination status and pathologies who have experienced each of 185 these transitions are shown in Table 2. 186 Specifically, the survival model was used to analyse the effects of age, sex, calendar 187 period of diagnosis, vaccination status at the time of diagnosis and pathologies on the 188 risk of experiencing any of those transition. The results of this model can be useful both 189 immediately and in the future for making public health decisions. For example, to 190 identify patients most at risk of hospitalization who need a priority for vaccination and 191 to improve the use of resources in hospitals, such as oxygen tanks. Furthermore, these 192

could be used to prevent overcrowding in intensive care units, which has been 193 one of the major problems in managing the pandemic in Italy [26], by identifying the 194 factors that increase the risk of such recovery especially among patients already 195 hospitalized in order to be able to treat them adequately before their conditions worsen. 196 Overall, our results show that the Cox’s model fits the data well. Table 3 shows in 197 fact that, considering a 5-folds cross-validation, a concordance index of 0.82 is obtained 198 on the test set for the DIAG →HOSP transition, a concordance index of 0.80 for the 199 DIAG→ICU transition, and a concordance index of 0.64 for the HOSP →ICU transition. 200 For what concerns the partial effects of the covariates in the model, a common result 201 of all three transitions analyzed is an increased risk of experiencing them associated 202 with males compared to females (HR 1.65 95% CI 1.53 - 1.79 for DIAG →HOSP 203 transition; HR 1.81 95% CI 1.39 - 2.37 for DIAG →ICU transition; HR 1.92 95% CI 1.50 204 - 2.45 for HOSP →ICU transition). This finding is in line with many studies in which a 205 significantly higher risk of mortality, hospitalisation and ICU admission was found to be 206 linked with male sex [24,25]. 207 July 13, 2023 8/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint T able 2. Distribution of patients by sex, age group, calendar period of diagnosis, vaccination status, pathologies and transition. DIAG DIAG→HOSP DIAG→ICU HOSP→ICU No. No. % of DIAG No. % of DIAG No. % of HOSP T otal 32500 2831 8.71 246 0.76 356 12.58 Sex Male 15804 1572 9.95 152 0.96 251 15.97 Female 16696 1259 7.54 94 0.56 105 8.34 Age group 0-19 5521 11 0.20 4 0.07 1 9.09 20-29 3942 48 1.22 4 0.10 3 6.25 30-39 4688 133 2.84 12 0.26 15 11.28 40-49 5437 352 6.47 31 0.57 29 8.24 50-59 5461 512 9.38 47 0.86 68 13.28 60-69 2952 528 17.89 63 2.13 95 17.99 70-79 1833 517 28.21 51 2.78 95 18.38 over 80 2666 730 27.38 34 1.28 50 6.85 Calendar period of diagnosis W1 1497 436 29.12 28 1.87 42 9.63 OW1 4090 323 7.90 44 1.08 41 12.69 W2 10017 700 6.99 72 0.72 107 15.29 W3 12376 1137 9.19 89 0.72 147 12.93 OW2 4520 235 5.20 13 0.29 19 8.09 V accination status NV 30938 2746 8.88 242 0.78 354 12.89 PV 417 38 9.11 3 0.72 2 5.26 TV 1145 47 4.10 1 0.10 0 0.00 Pathology Essential hypertension 1380 331 23.99 21 1.52 50 15.11 Hypertensive heart disease 456 97 21.27 13 2.85 17 17.53 Diabetes mellitus 1230 326 26.50 41 3.33 62 19.02 Hypercholesterolemia 384 84 21.88 4 1.04 15 17.86 Malignant neoplasm 1177 219 18.61 16 1.36 29 13.24 Multiple sclerosis 68 11 16.18 0 0.00 0 0.00 Hypothyroidism 620 66 10.65 3 0.48 12 18.18 Glaucoma 256 57 22.27 3 1.17 8 14.04 Toxic diffuse goiter 62 3 4.84 1 1.61 0 0.00 Epilepsy 78 6 7.69 1 1.28 2 33.33 Heart and pulmonary circulation diseases 872 200 22.94 24 2.75 35 17.50 Chronic kidney failure 184 74 40.22 5 2.72 9 12.16 Celiac disease 170 5 2.94 0 0.00 0 0.00 Asthma 357 41 11.48 4 1.12 4 9.76 Rheumatoid arthritis 108 22 20.37 4 3.70 5 22.73 Chronic Hepatitis 301 49 16.28 9 2.99 6 12.24 Embolism and thrombosis of other veins 145 42 28.97 10 6.90 8 19.05 Psoriasis 52 8 15.38 0 0.00 0 0.00 Hashimoto’s thyroiditis 320 19 5.94 1 0.31 2 10.53 Ulcerative colitis 106 14 13.21 0 0.00 1 7.14 Parkinson’s disease 30 10 33.33 0 0.00 2 20.00 Pituitary dwarfism 58 1 1.72 0 0.00 0 0.00 Endometriosis (III - IV ASRM stage) 40 2 5.00 0 0.00 0 0.00 July 13, 2023 9/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint T able 3. Mean concordance indices and standard errors assigned on the training set and test set in a 5-fold cross validation. T ransition T raining set T est set DIAG→HOSP 0.820 (0.001) 0.817 (0.004) DIAG→ICU 0.827 (0.003) 0.798 (0.011) HOSP→ICU 0.687 (0.002) 0.642 (0.011) T ransition DIAG→HOSP 208 Table 4 shows the Cox’s model results for the DIAG→HOSP transition. 209 The hazard ratio generally increases with age. However, no significant increase is 210 observed in the risk of being admitted to any COVID-19 ward for those over 80 211 compared to the reference age group 70-79 (HR 1.12 95% CI 0.99 - 1.27). A similar 212

was obtained in a Spanish study examining the association between age and the 213 risk of COVID-19 diagnosis, hospitalization and death which shows that the risk of 214 hospitalization after a diagnosis of COVID-19 peaked, for both sexes, around the age of 215 75 and then decreased slightly for older ages [27]. 216 Taking the not vaccinated status (NV) as baseline, it results that being partially 217 immunized does not imply a significant change in the risk of experiencing 218 DIAG→HOSP transition (HR 0.75 95% CI 0.54 - 1.05). On the other hand, the totally 219 vaccinated status (TV) is associated with a significant reduction in the hazard of 220 hospitalization (HR 0.23 95% CI 0.16 - 0.31). 221 As regards the effect of the calendar period of diagnosis, having contracted the 222 disease during W1 or W3 implied a significant greater risk of being hospitalized in a 223 non-ICU COVID-19 ward than during W2 (3.51 95% CI 3.09 - 4.00 for W1; 1.15 95% 224 CI 1.35 - 1.63 for W3). Despite some differences in the definition of the three waves, our 225

are comparable with those obtained from an analysis of the COVID-19 outcomes 226 of 4 million inhabitants of Northwest Italy during the first three waves of the 227 pandemic [28]. In particular, this study confirmed a greater risk of hospitalization in 228 both non-ICU and ICU wards of an infected person in the first wave compared to the 229 second wave. This may have been due to the improvements in COVID-19 treatment 230 following the first wave of the epidemic, such as the use of more effective 231 pharmacological strategies, and the widening of access to test for COVID-19 that was 232 initially restricted to the most severely ill patients. However, as regards what obtained 233 for W3, it is difficult to find an explanation to this result since no significant changes in 234 the management and treatment of the pandemic are present between W2 and W3 235 periods. Probably, this finding is due to factors beyond our knowledge such as, for 236 example, the availability of beds in hospitals dedicated to infected people during the 237 two waves which can affect the number of hospitalized patients. 238 Among all the pathologies taken into consideration, diabetes (HR 1.55 95% CI 1.36 - 239 1.76), hypercholesterolemia (HR 1.31 95% CI 1.05 - 1.62), multiple sclerosis (HR 1.89 240 95% CI 1.95 - 3.39), chronic kidney failure (HR 1.92 95% CI 1.49 - 2.48), asthma (HR 241 1.69 95% CI 1.23 - 2.30) and embolism (HR 1.58 95% CI 1.15 - 2.17) were significantly 242 associated with a increased risk of hospitalization. These results are in line with several 243 studies that have observed a significantly higher risk of non-ICU hospitalization among 244 patients with diabetes, chronic kidney failure and asthma [24,29,30]. 245 July 13, 2023 10/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint T able 4. Cox’s model results for DIAG →HOSP transition. Covariates HR HR 95% CI p-value Age group a 0-19 0.01 0.004 - 0.01 < 0.001*** 20-29 0.04 0.03 - 0.06 < 0.001*** 30-39 0.10 0.08 - 0.12 < 0.001*** 40-49 0.23 0.20 - 0.27 < 0.001*** 50-59 0.33 0.29 - 0.38 < 0.001*** 60-69 0.63 0.56 - 0.71 < 0.001*** over 80 1.12 0.99 - 1.27 0.067 Sexb Male 1.65 1.53 - 1.79 < 0.001*** V accination statusc TV 0.23 0.16 - 0.31 < 0.001*** PV 0.75 0.54 - 1.05 0.091 Calendar period of diagnosis d W1 3.51 3.09 - 4.00 < 0.001*** W3 1.48 1.35 - 1.63 < 0.001*** OW1 1.30 1.14 - 1.48 < 0.001*** OW2 1.99 1.68 - 2.35 < 0.001*** Pathology Essential hypertension 1.02 0.90 - 1.15 0.741 Hypertensive heart disease 0.97 0.78 - 1.20 0.767 Diabetes mellitus 1.55 1.36 - 1.76 < 0.001*** Hypercholesterolemia 1.31 1.05 - 1.62 0.017* Malignant neoplasm 1.12 0.98 - 1.29 0.103 Multiple sclerosis 1.89 1.95 - 3.39 0.033* Hypothyroidism 1.00 0.78 - 1.27 0.986 Glaucoma 1.04 0.79 - 1.37 0.776 Toxic diffuse goiter 0.74 0.24 - 2.22 0.587 Epilepsy 1.51 0.67 - 3.40 0.325 Heart and pulmonary circulation diseases 1.15 0.99 - 1.34 0.067 Chronic kidney failure 1.92 1.49 - 2.48 < 0.001*** Celiac disease 0.66 0.29 - 1.49 0.317 Asthma 1.69 1.23 - 2.30 0.001** Rheumatoid arthritis 1.46 0.96 - 2.23 0.080 Chronic Hepatitis 1.20 0.90 - 1.59 0.216 Embolism and thrombosis of other veins 1.58 1.15 - 2.17 0.004** Psoriasis 1.04 0.51 - 2.12 0.911 Hashimoto’s thyroiditis 0.85 0.55 - 1.32 0.474 Ulcerative colitis 1.27 0.74 - 2.18 0.393 Parkinson’s disease 1.69 0.85 - 3.37 0.135 Pituitary dwarfism 0.80 0.12 - 5.29 0.815 Endometriosis (III - IV ASRM stage) 1.09 0.27 - 4.37 0.899 a70-79 reference age group, bFemale reference sex, cNV reference vaccination status, dW2 reference calendar period of diagnosis. July 13, 2023 11/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint T ransition DIAG→ICU 246 Table 5 reports the hazard ratios resulted from the Cox’s model for the DIAG →ICU 247 transition. 248 The hazard increases with age but only up to 60-69. In fact, the 60-69 group does 249 not show a significantly different hazard compared to 70-79 (HR 0.83 95% CI 0.56 - 250 1.22). For what concerns the over 80 group, instead, a slightly significant lower hazard 251 is observed with respect to the reference (HR 0.57 95% CI 0.36 - 0.89). These findings 252 are in line with what obtained in a Swedish study, that is a decrease of the risk in ICU 253 hospitalization for ages older than 60-69 [29]. This could be due to the fact that several 254 elderly people may have died even before being admitted to intensive care or to the fact 255 that in some critical periods of the pandemic, priority was given to younger people who 256 needed admission to intensive care [31]. Furthermore, the fact that the elderly were the 257 most protected and least exposed category of people during the pandemic as the Italian 258 government recommended them to limit contacts as much as possible [32], could have 259 influenced the decrease in the risk of admission to intensive care for these people. 260 Concerning the effect of the vaccination status, no significant difference in the 261 hazard of being directly hospitalized in ICU after diagnosis was observed among 262 partially (PV) and non vaccinated (NV) patients (HR 1.15 95% CI 0.36 - 3.68). As for 263 those who have contracted the disease after completing the vaccination cycle (TV), 264 instead, a slightly significant decrease in risk was found (HR 0.10 95% CI 0.01 - 0.73). 265 As regards the influence of the calendar period of diagnosis, W1 is associated with 266 an increased hazard to experience the DIAG →ICU transition compared to W2 (HR 2.30 267 95% CI 1.47 - 3.60). On the other hand, no significant difference in the risk of 268 hospitalization in ICU between patients infected by SARS-CoV-2 during W3 and those 269 tested positive for COVID-19 during W2 was observed (HR 1.06 95% CI 0.77 - 1.44). 270 Although the definitions of the different waves do not match perfectly as they are based 271 on arbitrary criteria for each study, our results are fairly in line with other studies in 272 which changes in ICU admission and mortality have been analyzed over time [28,33]. In 273 particular, in each of these studies, as in ours, a reduction in the risk of serious illness 274 was observed in the second wave compared to the first one. As already discussed, this 275

is the effect of the development and improvement of the clinical management of 276 the new disease after the first few months. Furthermore, the absence of difference in the 277 risk of ICU admission found between W2 and W3 may be justified, as said before, by 278 the fact that there were no significant changes in the management of the pandemic and 279 treatment of the disease between the two periods. 280 Being affected by diabetes (HR 2.43 95% CI 1.65 - 3.59), heart or pulmonary 281 circulation disease (HR 1.65 95% CI 1.06 - 2.56), rheumatoid arthritis (HR 3.90 95% CI 282 1.43 - 10.63), chronic hepatitis (HR 2.39 95% CI 1.23 - 4.65) or embolism (HR 4.94 95% 283 CI 2.57 - 9.51), was found to be associated with a significantly higher risk of being 284 admitted to ICU. The same result regarding diabetes was obtained in a nationwide 285 Swedish study [29]. 286 July 13, 2023 12/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint T able 5. Cox’s model results for DIAG →ICU transition. Covariates HR HR 95% CI p-value Age group a 0-19 0.03 0.01 - 0.09 < 0.001*** 20-29 0.04 0.02 - 0.01 < 0.001*** 30-39 0.11 0.06 - 0.21 < 0.001*** 40-49 0.24 0.15 - 0.39 < 0.001*** 50-59 0.34 0.22 - 0.53 < 0.001*** 60-69 0.83 0.56 - 1.22 0.342 over 80 0.57 0.36 - 0.89 0.013* Sexb Male 1.81 1.39 - 2.37 < 0.001*** V accination statusc TV 0.10 0.01 - 0.73 0.023* PV 1.15 0.36 - 3.68 0.820 Calendar period of diagnosis d W1 2.30 1.47 - 3.60 < 0.001*** W3 1.06 0.77 - 1.44 0.729 OW1 1.64 1.13 - 2.39 0.010* OW2 0.99 0.53 - 1.87 0.981 Pathology Essential hypertension 0.79 0.49 - 1.28 0.338 Hypertensive heart disease 1.50 0.84 - 2.68 0.174 Diabetes mellitus 2.43 1.65 - 3.59 < 0.001*** Hypercholesterolemia 0.56 0.21 - 1.47 0.237 Malignant neoplasm 0.91 0.54 - 1.53 0.707 Multiple sclerosis - - - Hypothyroidism 0.46 0.15 - 1.48 0.194 Glaucoma 0.65 0.20 - 2.09 0.470 Toxic diffuse goiter 3.05 0.51 - 18.36 0.224 Epilepsy 2.39 0.33 - 17.49 0.390 Heart and pulmonary circulation diseases 1.65 1.06 - 2.56 0.028* Chronic kidney failure 1.44 0.55 - 3.73 0.457 Celiac disease - - - Asthma 1.75 0.64 - 4.77 0.276 Rheumatoid arthritis 3.90 1.43 - 10.63 0.008** Chronic Hepatitis 2.39 1.23 - 4.65 0.010* Embolism and thrombosis of other veins 4.94 2.57 - 9.51 < 0.001*** Psoriasis - - - Hashimoto’s thyroiditis 0.55 0.08 - 3.97 0.553 Ulcerative colitis - - - Parkinson’s disease - - - Pituitary dwarfism - - - Endometriosis (III - IV ASRM stage) - - - a70-79 reference age group, bFemale reference sex, cNV reference vaccination status, dW2 reference calendar period of diagnosis. July 13, 2023 13/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint T ransition HOSP→ICU 287 Table 6 shows the results of the Cox’s model for the HOSP →ICU transition. 288 Only few patients in the age ranges 0-19, 20-29, or 30-39 were present in the initial 289 state (i.e. few patients of these age groups were hospitalized in non-ICU wards; see 290 Table 2 for details), making the hazard ratios related to those groups not robust. For 291 this reason, we focused our discussion only on older ages. A significantly lower risk was 292 observed for patients aged 40-49 and over 80 compared to 70-79 (HR 0.45 95% CI 0.29 - 293 0.70 for 40-49; HR 0.40 95% CI 0.28 - 0.58 for over 80), while no significant difference in 294 transition risk was noticed for ages 50-59 and 60-69 (HR 0.77 95% CI 0.55 - 1.08 for 295 50-59; HR 1.09 95% CI 0.81 - 1.48 for 60-69). Likewise, in a study from Kuwait [34], it 296 was found that the probability of ICU admission among hospitalized patients was 297 highest at around age 70 and then decreases for older ages. As already discussed, these 298 findings could be the effect, at least in part, of the rationing of healthcare resources 299 during the critical phases of the pandemic with younger patients being prioritized to 300 receive hospital care and intensive services if needed [31]. 301 For what concerns the influence of the vaccination status, as for the transitions 302 previously analyzed, being partially vaccinated (PV) does not imply any change in the 303 risk of experiencing HOSP →ICU transition (HR 0.48 95% CI 0.12 - 1.99). On the other 304 hand, no person among those totally vaccinated (TV) at the time of diagnosis was 305 transferred to ICU from another COVID-19 ward. Therefore, consistent with an official 306 ISS report dated September 2021 that studies the impact of vaccination with mRNA 307 vaccines on the risk of COVID-19 infection, hospitalization and death by analyzing data 308 from the National Vaccination Registry and the COVID-19 Integrated Surveillance 309 System [35], we observed that the completion of the vaccination cycle significantly 310 reduces the risk of COVID-19-related hospitalisation. Our results are also in line with a 311 Norwegian study in which it was observed that fully vaccinated hospitalized patients 312 had a significantly lower risk of ICU admission than unvaccinated ones [36]. 313 As regards the effect of the calendar period of diagnosis, a significantly lower risk of 314 HOSP→ICU transition was observed during W1 than during W2 (HR 0.53 95% CI 0.37 315 - 0.76), while during the other calendar periods of diagnosis there is no significant 316 difference in the same hazard compared to the reference. 317 None of the pathologies considered was associated with a significant increase in the 318 risk of moving to ICU from a non-ICU ward. This result is in contradiction with what 319 emerged in some studies in which for example patients with diabetes [37] and 320 asthma [24] are significantly more at risk of experiencing this transition. 321 July 13, 2023 14/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint T able 6. Cox’s model results for HOSP →ICU transition. Covariates HR HR 95% CI p-value Age group a 0-19 0.50 0.06 - 4.01 0.514 20-29 0.38 0.12 - 1.23 0.106 30-39 0.64 0.36 - 1.17 0.147 40-49 0.45 0.29 - 0.70 < 0.001*** 50-59 0.77 0.55 - 1.08 0.126 60-69 1.09 0.81 - 1.48 0.562 over 80 0.40 0.28 - 0.58 < 0.001*** Sexb Male 1.92 1.50 - 2.45 < 0.001*** V accination statusc TV - - - PV 0.48 0.12 - 1.99 0.315 Calendar period of diagnosis d W1 0.53 0.37 - 0.76 < 0.001*** W3 0.94 0.72 - 1.21 0.613 OW1 0.77 0.53 - 1.12 0.179 OW2 0.67 0.39 - 1.13 0.133 Pathology Essential hypertension 1.12 0.81 - 1.56 0.495 Hypertensive heart disease 1.21 0.72 - 2.95 0.478 Diabetes mellitus 1.18 0.87 - 1.59 0.286 Hypercholesterolemia 1.44 0.85 - 2.43 0.178 Malignant neoplasm 0.82 0.54 - 1.25 0.361 Multiple sclerosis - - - Hypothyroidism 1.62 0.84 - 3.15 0.152 Glaucoma 1.01 0.50 - 2.07 0.973 Toxic diffuse goiter - - - Epilepsy 1.82 0.44 - 7.51 0.408 Heart and pulmonary circulation diseases 1.13 0.79 - 1.62 0.496 Chronic kidney failure 0.77 0.38 - 1.53 0.447 Celiac disease - - - Asthma 0.79 0.28 - 2.23 0.663 Rheumatoid arthritis 2.17 0.79 - 5.93 0.132 Chronic Hepatitis 0.89 0.40 - 2.00 0.780 Embolism and thrombosis of other veins 1.70 0.83 - 3.50 0.147 Psoriasis - - - Hashimoto’s thyroiditis 1.30 0.31 - 5.46 0.724 Ulcerative colitis 0.45 0.07 - 3.10 0.420 Parkinson’s disease 0.91 0.22 - 3.74 0.898 Pituitary dwarfism - - - Endometriosis (III - IV ASRM stage) - - - a70-79 reference age group, bFemale reference sex, cNV reference vaccination status, dW2 reference calendar period of diagnosis. July 13, 2023 15/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint Study limitations 322 The results of our study should be interpreted taking into account some limitations. 323 The major constrain of this study is the lack of available data on deaths among 324 SARS-CoV-2 infected people. Therefore, it was not possible for us to analyze the 325 factors that influence mortality but only those associated with the risk of developing a 326 severe form of the disease that requires hospitalization in intensive care. This makes the 327 multi-state model incomplete as we have been forced to consider ICU as an absorbing 328 state when in reality it is possible to get out of it by dying or passing to a state other 329 than death, such as discharge, recovery or admission to another non-intensive ward. 330 Furthermore, this study only concerns the municipality of Bologna, hence the results 331 obtained cannot be generalized to other municipalities or to all of Italy or other 332 countries. We believe that an accurate comparison with other studies should also take 333 into account specific factors of each territory such as, for example, the socio-economic 334 conditions of the study cohort, the criteria of virus testing in the study setting, the 335 availability of, the availability of beds in local hospitals for COVID-19 patients and in 336 general the management of the pandemic by the competent authorities during the study 337 period. 338 Another limitation of this study is that the criterion used to define the waves is 339 arbitrary and based only on daily hospitalizations in the municipality of Bologna during 340 the study period. However, our definition is quite in line with other Italian 341 studies [28,33] and national reports [10]. 342 In addition, we did not take into account the circulation of several variants of 343 SARS-CoV-2 over the study period, whose different virulence may have affected the 344 rate of hospitalization and the mortality. 345 Finally, although studies have found that COVID-19 vaccines only start to provide 346 significant protection after one week after the second dose of mRNA vaccines or two 347 weeks after the single J&J vaccine dose [14,15], the immune response triggered by 348 vaccination is gradual and variable between individuals. Therefore, our categorization of 349 patients by vaccination status at the time of COVID-19 diagnosis, although supported 350 by several articles [16,17], may be too simplistic not taking into account this gradual 351 growth and variability of immunity. 352

353 As the novel coronavirus disease has become a major public health event due to its rapid 354 transmission and large-scale spread, it is necessary to prepare in advance about the risks 355 of serious illness to prevent pandemic degeneration based on local demographics, the 356 current situation of medical resources and the progress of the vaccination campaign. For 357 this reason, our study aimed to investigate the factors associated with the risk of 358 developing a severe form of the disease requiring hospitalization in both intensive and 359 non-intensive wards. Using a multi-state model, the study revealed that older age, male 360 sex, having contracted the disease during the first wave, and being unvaccinated or 361 partially vaccinated are poor prognostic factors for COVID-19. The multi-state model 362 approach made it possible to study the transition of patients between diagnosis, 363 admission to non-intensive wards, and admission to intensive care unit in a more detail 364 way than using the two-state model. In addition, we identified some diseases associated 365 with an increased risk of ICU and non-ICU hospitalization, including diabetes, asthma 366 and chronic kidney failure. In conclusion, our findings suggest that to limit the number 367 of severe cases of COVID-19 and consequently the overcrowding of hospitals, the 368 completion of the vaccination cycle and a more careful monitoring and prioritization in 369 the care of the elderly and of people with certain specific comorbidities are essential. 370 July 13, 2023 16/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint Supporting information 371 S1 Fig. Graphical representation of the number of monthly hospitalizations 372 in any COVID-19 ward during the study period in the hospitals of the 373 municipality of Bologna. (TIFF) 374 S2 Fig. Distribution of patients by transitions for each vaccination status. 375 (TIFF) 376 Ethics statement 377 Ethical approval for the study was obtained from the University of Bologna’s Ethical 378 Committee (approval number 283066, 5 October 2021). The data set was previously 379 anonymized, and the authors did not have access to any information that could have 380 been used to identify the individual patients involved in the study. In compliance with 381 Italian Data protection Authority [38], no informed consent was required for carrying 382 out observational studies for scientific research purposes. 383 Data Availability Statement 384 Data of the present study were provided from the Local Health Unit (AUSL, Azienda 385 Unit` a Sanitaria Locale) of the municipality of Bologna, Italy, and the authors do not 386 have the right to share them. In order to gain access to the data, contact Local Health 387 Unit of the municipality of Bologna, Italy ([email protected]). 388 Financial Disclosure Statement 389 The authors received no specific funding for this work. 390 Competing interests 391 The authors have declared that no competing interests exist. 392 Acknowledgments 393 We acknowledge the Bologna MODELS4COVID Study Group of the University of Bologna and the National Institute for Nuclear Physics (INFN): Armando Bazzani, Valerio Carelli, Paolo Tubertini, Luca Clissa, Stefano Diciotti, Enrico Lunedei, Michela Milano, Luca Palmerini, Daniel Remondini, Giulia Roli, Michele Scagliarini, Roberto Spighi, Vincenzo Vagnoni, Antonio Zoccoli, Lorenzo Chiari.

1. Worobey M, Levy JI, Malpica Serrano L, Crits-Christoph A, Pekar JE, Goldstein SA et al. The Huanan Seafood Wholesale Market in Wuhan was the early epicenter of the COVID-19 pandemic. Science. 2022 Aug 26;377(6609):951-959. doi: 10.1126/science.abp8715. July 13, 2023 17/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint 2. COVID-19 situation update for the EU/EEA [cited 2022 October 08]. In: European Centre for Disease Prevention and Control [Internet]. Available from: https://www.ecdc.europa.eu/en/cases-2019-ncov-eueea. 3. Clinical Spectrum of SARS-CoV-2 Infection [cited 2022 October 08]. In: National Institutes of Health [Internet]. Available from: https://www. covid19treatmentguidelines.nih.gov/overview/clinical-spectrum/. 4. Parohan M, Yaghoubi S, Seraji A, Hassan Javanbakht M, Sarraf P, Djalali M. Risk factors for mortality in patients with Coronavirus disease 2019 (COVID-19) infection: a systematic review and meta-analysis of observational studies. The Aging Male. 2020 Jun 08;23(5):1416-1424. doi: 10.1080/13685538.2020.1774748. 5. Pijls BG, Jolani S, Atherley A, Derckx RT, Dijkstra JIR, Franssen GHL et al. Demographic risk factors for COVID-19 infection, severity, ICU admission and death: a meta-analysis of 59 studies. BMJ Open. 2021;11:e044640. doi: 10.1136/bmjopen-2020-044640. 6. Guo W, Li M, Dong Y, Zhou H, Zhang Z, Tian C et al. Diabetes is a risk factor for the progression and prognosis of COVID-19. Diabetes Metab Res Rev. 2020 Mar 31;36(7):e3319. doi: 10.1002/dmrr.3319. 7. Pranata R, Lim MA, Huang I, Raharjo SB, Lukito AA. Hypertension is associated with increased mortality and severity of disease in COVID-19 pneumonia: A systematic review, meta-analysis and meta-regression. J Renin Angiotensin Aldosterone Syst. 2020 Apr-Jun;21(2):1470320320926899. doi: 10.1177/1470320320926899. 8. Liu X, Huang J, Li C, Zhao Y, Wang D, Huang Z et al. The role of seasonality in the spread of COVID-19 pandemic. Environ Res. 2021 Apr;195: 110874. doi: 10.1016/j.envres.2021.110874. 9. Liu Q, Qin C, Liu M, Liu J. Effectiveness and safety of SARS-CoV-2 vaccine in real-world studies: a systematic review and meta-analysis. Infect Dis Poverty. 2021;10: 132. doi: 10.1186/s40249-021-00915-3. 10. Coronavirus in Italia: dati, infografiche e mappe [cited 17 October 2022]. In: Sky tg24 [Internet]. Available from: https: //tg24.sky.it/cronaca/approfondimenti/coronavirus-italia-contagi. 11. Coronavirus [cited 21 September 2022]. In: EpiCentro-Istituto Superiore di Sanit` a [Internet]. Available from: https://www.epicentro.iss.it/coronavirus/. 12. Zeleke AJ, Moscato S, Miglio R, Chiari L. Length of Stay Analysis of COVID-19 Hospitalizations Using a Count Regression Model and Quantile Regression: A Study in Bologna, Italy. Int J Environ Res Public Health. 2022 Feb 16;19(4):2224. doi: 10.3390/ijerph19042224. 13. Pollard AJ, Bijker EM. A guide to vaccinology: from basic principles to new developments. Nat Rev Immunol. 2021 Feb; 21(2):83-100. doi: 10.1038/s41577-020-00479-7. 14. Polack FP, Thomas SJ, Kitchin N, Absalon J, Gurtman A, Lockhart S et al. C4591001 Clinical Trial Group. Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine. N Engl J Med. 2020 Dec 31;383(27):2603-2615. doi: 10.1056/NEJMoa2034577. July 13, 2023 18/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint 15. FDA Briefing Document. Janssen Ad26.COV2.S Vaccine for the Prevention of COVID-19. 2021 February 26 [cited 2022 October 20]. Available from: https://www.fda.gov/media/146217/download. 16. Cocchio S, Zabeo F, Facchin G, Piva N, Furlan P, Nicoletti M et al. The Effectiveness of a Diverse COVID-19 Vaccine Portfolio and Its Impact on the Persistence of Positivity and Length of Hospital Stays: The Veneto Region’s Experience. Vaccines (Basel). 2022 Jan 11; 10(1):107. doi: 10.3390/vaccines10010107. 17. Tartof SY, Slezak JM, Fischer H, Hong V, Ackerson BK, Ranasinghe ON, et al. Effectiveness of mRNA BNT162b2 COVID-19 vaccine up to 6 months in a large integrated health system in the USA: a retrospective cohort study. Lancet. 2021 Oct 16; 398(10309):1407-1416. doi: 10.1016/S0140-6736(21)02183-8. 18. Pignataro M. Grey areas and uncertainties: the AstraZeneca case in Italy. Lex-Atlas: Covid-19. 2021 Aug 10 [cited 08 October 2022]. Available from: https://lexatlas-c19.org/ grey-areas-and-uncertainties-the-astrazeneca-case-in-italy/. 19. McKinney W. Data Structures for Statistical Computing in Python. Proceedings pf 9th Python in Science Conference. 2010; 56-61. doi: 10.25080/Majora-92bf1922-00a. 20. Davidson-Pilon C. lifelines: survival analysis in Python. Journal of Open Source Software. 2019;4(40):1317. doi: 10.21105/joss.01317. 21. Hunter JD. Matplotlib: A 2D graphics environment. Computing in Science & Engineering. 2007;9(3):90-95. doi: 10.1109/MCSE.2007.55. 22. Kassambara A, Kosinski M, Biecek P. survminer: Drawing Survival Curves using ’ggplot2’. R package version 0.4.9. 2021 [cited 03 April 2023]. Available from: https://CRAN.R-project.org/package=survminer. 23. Therneau T. A Package for Survival Analysis in R. R package version 3.5-5. 2023 [cited 03 April 2023]. Available from: https://CRAN.R-project.org/package=survival. 24. Bennett KE, Mullooly M, O’Loughlin M, Fitzgerald M, O’Donnell J, O’Connor L et al. Underlying conditions and risk of hospitalisation, ICU admission and mortality among those with COVID-19 in Ireland: A national surveillance study. Lancet Reg Health Eur. 2021 Jun;5:100097. doi: 10.1016/j.lanepe.2021.100097. 25. Sis´ o-Almirall A, Kostov B, Mas-Heredia M, Vilanova-Rotllan S, Sequeira-Aymar E, Sans-Corrales M et al. Prognostic factors in Spanish COVID-19 patients: A case series from Barcelona. PLoS One. 2020 Aug 21;15(8):e0237960. doi: 10.1371/journal.pone.0237960. 26. Dettaglio terapie intensive in Italia [cited 17 October 2022]. In: Statistiche coronavirus [Internet]. Available from: https: //statistichecoronavirus.it/coronavirus-italia/terapie-intensive/. 27. Burn E, Teb´ e C, Fernandez-Bertolin S, Aragon M, Recalde M, Roel E et al. The natural history of symptomatic COVID-19 during the first wave in Catalonia. Nat Commun. 2021 Feb 3;12(1):777. doi: 10.1038/s41467-021-21100-y. July 13, 2023 19/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint 28. D’Arminio Monforte A, Tavelli A, Bai F, Tomasoni D, Falcinella C, Caramello V et al. Improvements throughout the Three Waves of COVID-19 Pandemic:

from 4 Million Inhabitants of North-West Italy. J Clin Med. 2022 Jul 25;11(15):4304. doi: 10.3390/jcm11154304. 29. Bergman J, Ballin M, Nordstr¨ om A, Nordstr¨ om P. Risk factors for COVID-19 diagnosis, hospitalization, and subsequent all-cause mortality in Sweden: a nationwide study. Eur J Epidemiol. 2021;36:287–298. doi: 10.1007/s10654-021-00732-w. 30. Ko JY, Danielson ML, Town M, Derado G, Greenlund KJ, Kirley PD et al. COVID-NET Surveillance Team. Risk Factors for Coronavirus Disease 2019 (COVID-19)-Associated Hospitalization: COVID-19-Associated Hospitalization Surveillance Network and Behavioral Risk Factor Surveillance System. Clin Infect Dis. 2021 Jun 1;72(11):e695-e703. doi: 10.1093/cid/ciaa1419. 31. Privitera G. Italian doctors on coronavirus frontline face tough calls on whom to save. POLITICO. 2020 Mar 09 [cited 2022 October 21]. Available from: https://www.politico.eu/article/ coronavirus-italy-doctors-tough-calls-survival/. 32. Covid-19, le raccomandazioni per le persone anziane [cited 2022 October 21]. In: Ministero della Salute [Internet]. Available from: https://www.salute.gov.it/ portale/nuovocoronavirus/dettaglioNotizieNuovoCoronavirus.jsp?id= 4172&lingua=italiano&menu=notizie&p=dalministero. 33. Giacomelli A, Ridolfo AL, Pezzati L, Oreni L, Carrozzo G, Beltrami M et al. Mortality rates among COVID-19 patients hospitalised during the first three waves of the epidemic in Milan, Italy: A prospective observational study. PLoS One. 2022 Apr 11;17(4):e0263548. doi: 10.1371/journal.pone.0263548. 34. Kipourou DK, Leyrat C, Alsheridah N, Almazeedi S, Al-Youha S, Jamal MH et al. Probabilities of ICU admission and hospital discharge according to patient characteristics in the designated COVID-19 hospital of Kuwait. BMC Public Health. 2021;21:799. doi: 10.1186/s12889-021-10759-z. 35. Istituto Superiore di Sanit` a. Impact of COVID-19 vaccination on the risk of SARS-CoV-2 infection and hospitalization and death in Italy. Report n. 4 of 2021 Sept 30 [cited 2022 October 17]. Available from: https://www.iss.it/ documents/20126/0/report_on_vaccine_effectiveness_Italy+%281%29. pdf/53d71dc2-c8c5-24c1-3467-705a8587a339?t=1633529045681. 36. Whittaker R, Br˚ athen Kristofferson A, Valcarcel Salamanca B, Sepp¨ al¨ a E, Golestani K, Kv˚ ale R et al. Length of hospital stay and risk of intensive care admission and in-hospital death among COVID-19 patients in Norway: a register-based cohort study comparing patients fully vaccinated with an mRNA vaccine to unvaccinated patients. Clin Microbiol Infect. 2022 Jun;28(6):871-878. doi: 10.1016/j.cmi.2022.01.033. 37. Kim L, Garg S, O’Halloran A, Whitaker M, Pham H, Anderson EJ et al. Risk Factors for Intensive Care Unit Admission and In-hospital Mortality Among Hospitalized Adults Identified through the US Coronavirus Disease 2019 (COVID-19)-Associated Hospitalization Surveillance Network (COVID-NET). Clin Infect Dis. 2021 May 4;72(9):e206-e214. doi: 10.1093/cid/ciaa1012. July 13, 2023 20/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint 38. Garante per la Protezione dei Dati Personali. General Authorisation to Process Personal Data for Scientific Research Purposes - 1 March 2012 [1884019] [cited 2023 April 20]. Available from: https://www.garanteprivacy.it/home/ docweb/-/docweb-display/docweb/1884019. July 13, 2023 21/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 13, 2023. ; https://doi.org/10.1101/2023.07.12.23292559doi: medRxiv preprint