The Italian COVID-19 epidemiological surveillance dataset analyzed here contained information on over 4 million persons molecularly diagnosed with a SARS-CoV-2 infection until July 25, 2021. The dataset included information about age, sex, date of diagnosis, presence vs. absence of symptoms, date of hospitalization (if pertinent), date of death (if pertinent), and a few other data. Survival analyses on the whole series and on subsets of non-hospitalized and hospitalized patients strongly confirmed the pivotal role of age in the probability of survival of COVID-19 patients. The analysis by age category, adjusted for sex and pandemic wave, showed that age groups older than 65 had mortality risks that were hundreds of times greater than that of the 15- to 44-year-old reference class. The 0-14 years age group had a mortality risk that was about 10 times less than that of the reference class. Male sex was also confirmed to be a poor prognostic factor, but with a much smaller effect. Additionally, our analysis demonstrated that being diagnosed during the first pandemic wave (until June 2020) was associated with an approximately 3-fold higher mortality risk than being diagnosed later.
In non-hospitalized patients, the mortality risk associated with age was greater than that for the whole series. This difference might be explained by the observation that most deceased non-hospitalized patients were very old, with a median age of 86 years. As a possible interpretation, we suppose that some elderly persons deteriorated rapidly and died before they could be hospitalized.
In hospitalized patients, old age was associated with an excess risk of death, as in the whole series, although the statistical estimates were lower. For example, for the age group ³85 years old, the HR was 58.7 for hospitalized patients and 687 for the whole series. The difference may be explained by the fact that hospitalized patients were much older than subjects of the whole series, i.e., median age 70 years versus 46 years, and that, therefore, hospitalization by itself incorporates an excess risk of death, as age is a known risk factor for hospitalization [15,16], including in our series (not shown).
Our finding of age being a risk factor for COVID-19 mortality is in agreement with that of a meta-analysis by Shi et al. [8] on 27 studies (including 24 from China, two from the United States, and one from Italy) and a meta-analysis by Booth et al. [17] on 66 studies with >17 million patients from 14 countries. Both meta-analyses found an association between old age and excess risk of mortality from COVID-19, although the quantitative risk estimates differ. Of note, these meta-analyses did not report HRs associated with survival, since no Cox analyses were done. To the best of our knowledge, only one other nation-wide study, conducted in France by Semenzato et al. [18], used Cox models to analyze the effects of age on the risk of mortality in a large number of hospitalized COVID-19 patients. Although the age groups differ between the two studies, the risk estimates are similar, with HRs >50 in elderly patients in both studies.
Several immunological mechanisms responsible for the increased risk of death from COVID-19 in the elderly can be hypothesized. One study demonstrated that pre-existing T-cell immunity induced by circulating human alpha- and beta-coronaviruses is present in young adults but virtually absent in older adults [19]. Consequently, older adults had a minimal baseline frequency of cross-reactive T cells directed toward the novel SARS-CoV-2; for this reason, they may be at higher risk of severe COVID-19 disease and death. Moreover, the phenomenon of immunosenescence, which involves age-related changes in innate and adaptive immunity, has been imputed as being associated with the increased mortality of older adults infected with SARS-CoV-2 [20]. The elderly exhibit a deficient immunologic response to SARS-CoV-2 infection, which may be another reason for their increased risk of severe disease and death [21].
In the Italian nationwide COVID-19 series, male sex was an unfavorable prognostic factor for survival, with a risk that was 2-fold higher than for females in the whole series, and >80% and ~50% higher in non-hospitalized and hospitalized patients, respectively. This result is in agreement with those of several other studies [8,17], although the quantitative risk estimates differ. The HRs for male sex calculated in this study, which range from 1.46 to 2.05, are similar to those reported by Semenzato et al. [18].
The mechanism by which sex is an unfavorable prognostic factor for COVID-19 is not yet known. Most likely, several sex-related factors contribute to the higher risk of males for poorer COVID-19 outcomes. A study of 1,683 Italian patients who underwent chest computed tomography at admission showed that men had a higher prevalence of cardiovascular comorbidities, more coronary calcifications, and a higher coronary calcium score than females [22]. Notably, the higher coronary calcific burden of men appeared to be associated with higher mortality. A study of about 3,000 COVID-19 patients in a single center in China observed that the level of inflammatory cytokines in peripheral blood was higher in males than in females [23]. Also, the percentages of CD19+ B cells and CD4+ T cells were generally higher in female patients during the course of the disease. Overall, males had greater inflammation, lower lymphocyte counts, and lower and delayed antibody responses during SARS-CoV-2 infection and recovery than females. Finally, from the perspective of an immunological mechanism, it has been hypothesized that chronic, subclinical, systemic inflammation, characteristic of aging, and immunosenescence contribute to the excess risk of COVID-19 mortality in elderly men [24].
Our multivariable analysis provides strong support for the hypothesis that mortality from COVID-19 was much greater during the first wave (January to June 30, 2020) than later. Indeed, taking the first wave as reference, in the subsequent periods we observed a ~3-fold reduced risk of death, both in the whole series and in non-hospitalized patients . In hospitalized patient, the excess risk of death was ~30% lower in the two semesters after the first wave. The excess risk of death associated with pandemic wave was first reported in an Italian study of hospitalized patients [12], and then confirmed by studies of Massachusetts healthcare workers [13], patients of the U.S. Veterans Affairs healthcare system [25], and UK patients [26]. The reasons for this effect could include the initial lack of preparedness of national health systems for pandemic management, the lack of knowledge about the most effective therapies for COVID-19 patients with severe disease, and the possibility that frailer people were more affected at the beginning of the pandemic than the rest of the population. Another possible explanation for a lower risk of mortality after June 30, 2020, and in particular, during the third semester of the pandemic, compared with the first semester, may be mass vaccination, which began in January 2021; indeed, vaccines are associated with a reduced risk of severe COVID-19 and mortality [27].
In the Italian COVID-19 epidemiological surveillance dataset, more than 2 million infected persons were symptomatic (50.4% of all cases). Modeling studies on the prevalence of infection in different populations suggested that the total number of SARS-CoV-2-positive individuals exceeds symptomatic cases by an order of magnitude or more [28–30]. If this holds true for the Italian population, then ~20 million people in Italy have been infected by SARS-CoV-2, i.e., 10 times the 2 million symptomatic cases. Why some infections are asymptomatic and others lead to severe COVID-19 has not yet been elucidated. Cross-reactive immunity, pre-existing in individuals who had been exposed to other coronaviruses, could be one of the mechanisms for asymptomatic and moderate courses of SARS-CoV-2 infection in many individuals [31].
A limitation of our study is the lack of data about COVID-19 patients’ comorbidities, which are important risk factors for outcome [8]. This lack of information prevented us from analyzing other risk factors for death. Moreover, the reasons why some hospitalized patients were classified as asymptomatic are not known, but their hospitalization may have been due to reasons other than COVID-19. For example, in 5,432 cases, the date of SARS-CoV-2 infection detection was after the date of hospitalization and in 13,144 patients the diagnosis was on the same day.
Overall, this study confirms that age and male sex are independent risk factors for COVID-19 mortality for both hospitalized and not-hospitalized patients. Because age was found to be the most impactful negative prognostic factor, it should be considered in pandemic management, by giving priority to strategies aimed at protecting elderly people. Additionally, this is the first country-wide study to demonstrate a high risk of mortality during the first pandemic wave than later. Similar nation-wide studies in different countries, to the best of our knowledge, have not been published. Thus, we cannot compare our study with those from other nations with different mortality rates, and we cannot exclude that such differences are due to unequal pandemic management in the first wave, considering that Italy was the first Western nation to be affected. Our study also suggests that the medical research that started with the pandemic onset and that led to the development of increasingly more effective clinical protocols contributed to improving COVID-19 patient survival. Despite the limitations of this study, principally due to the lack of some clinical data (e.g. about comorbidities), this study demonstrates the usefulness of a national database for studying a new disease such as COVID-19. Efforts should be made in Italy to create a more detailed national database like those of the United Kingdom [32] and France [33] that collect more data on demographics, symptoms, diagnostic tests and treatments. National health databases, especially when accompanied by a national biobank of blood samples, offer great possibilities for biomedical research. They allow the construction of cohorts with unparalleled statistical power and help study risk factors for common diseases, rare diseases, and new emerging diseases such as COVID-19. Their availability could impact treatment and public health. Therefore, the creation of such databases in countries that do not yet have them and the creation of European databases are desirable.