A survival analysis of COVID-19 in the Mexican population

The pandemic caused by the novel coronavirus, SARS-CoV-2, has become one of the biggest health challenges worldwide. There are nearly 9 million confirmed cases and more than 465 thousand deaths as of June 22nd, 2020. This makes the COVID-19 pandemic one of the public health problems with a meaningful impact in the history of humanity.

By the third week of June, the Americas were reporting the largest number of cases worldwide. The United States was the country with the highest number of total confirmed COVID-19 cases (2,241,178), followed by Brazil (1,067,579 cases), Perú (251,338 cases), Chile (242,355 cases), and Mexico (175,202 cases). However, Mexico was third in the death count (20,781 total deaths) [1].

COVID-19 is commonly asymptomatic [2]. Asymptomatic COVID-19 patients frequently improve with time without specialized medical care [3]. But a significant proportion of COVID-19 cases develop pneumonia and acute severe respiratory failure [4]. These cases commonly require hospitalization, intensive care unit admission, and intubation [5].

Older age, the presence of comorbidities, particularly hypertension, diabetes, obesity, and smoking are factors that increase the risk of severe disease [6, 7]. The characteristics of COVID-19 may vary depending on the demographic and epidemiological profiles of each country. In the specific case of Mexico, 27% of the population is over 50 years old [8] and there is a high rate of chronic diseases [9, 10] which may increase the risk of fatal complications.

Assessing the instantaneous rate of death at any time during follow-up for the specific risk factors is crucial to determine the appropriateness of the mitigation strategies and to set up priorities to control the COVID-19 epidemic. This is especially so for countries that are in the present moment focusing their efforts on fighting the pandemic; this is the case of Mexico. The present analysis aimed to investigate the risk factors associated with COVID-19 deaths in the Mexican population using survival analysis.

This analysis includes 16,752 registers of confirmed cases of COVID-19. The mean age of the individuals was 46.55 ± 15.55 years, 58.02% (n = 9719) were men, and 9.37% (n = 1569) died. The general mortality rate was 4.94 (95% CI 4.70–5.19) per 1000 person-year, for men 5.82 (95% CI 5.48–6.18) per 1000 person-year, and for women 3.74 (95% CI 3.43–4.08) per 1000 person-year (Table 1).

When comparing individuals who died versus those who survived, the deceased were older (59.42 ± 14.29) and mostly men (67.94%), (p < 0.001). The prevalence of CKD in people who died (6.86%) was larger in comparison to those who did not die (p < 0.001). In individuals who survived, the proportion without comorbidities was higher (p < 0.001). The period from the onset of symptoms to admission in hospitalization was similar between both groups (4.27 ± 3.43, p = 0.7302). The period from the onset of symptoms to death, and the period from admission to death were 10.15 ± 5.75 and 5.86 ± 5.12 days, respectively. Among the people who died 75.33% (n = 1182) developed pneumonia, 90.25% (1416) was hospitalized, 18.61% (n = 292) was admitted to the ICU, and 23.96% (n = 376) needed intubation. These proportions were higher in comparison to individuals who did not die (p < 0.001). In the group of those who died, 42.65% received health care from IMSS, while in the surviving group, 47.94% received health care from SS services (Table 1).

The survival analysis comprises a total of 16,734 registers, 1558 of whom died. Our analysis observed a cumulated total risk time of 315,773 days. The Kaplan-Meier survival plots for the prognostic factors that resulted statistically significant are presented in Fig. 1. As can be inferred from the plot, risk is directly proportional to age group. The proportional hazards assumption is satisfied since survival risk curves do not cross during the analyzed period. Individuals who developed pneumonia, those with a previous diagnosis of CKD, and those who were admitted to hospitalization, have approximately 30% lower probability of survival after 20 days of hospitalization than individuals who did not present these characteristics. People who were admitted to ICU, have about 40% less probability of survival after 20 days of hospitalization than individuals who did not enter this unit. Finally, individuals who needed intubation have just about 50% probability of survival after 20 days of hospitalization compared to individuals who were not intubated.

The Kaplan-Meier plots demonstrated that the risk of mortality was different between sexes. Thus, Kaplan-Meier curves were plotted for men and women (Supplementary material 1 and 2), and multivariable Cox proportional hazards regression models were run for each sex. The hazard ratios for both sexes can be seen in Table 2. For women, the multivariable model containing the significant covariates was statistically significant (x² = 22.22, df = 10, p < 0.01). At any time during follow-up, age increases the risk of dying, and that risk was significantly high in older women, who had 4.41 times the risk of dying compared with the youngest age group. Women with CKD were 1.90 times likely to die compared with women without this disease. Twice as many women who developed pneumonia had died compared to women who did not develop this complication. At any time during the follow-up, women admitted to hospitalization were 6.57 times as likely to die compared with those who were not admitted to a hospital. Women who needed intubation were 2.83 times as likely to die compared with those who did not require intubation. Women receiving health care from IMSS services had 4.34 times the risk of dying compared to women who received health care in private facilities.

The multivariable Cox proportional hazards regression model for men was statistically significant (χ² = 31.83, df = 12, p < 0.001) and is presented in Table 2. The risk of dying was notably higher for men over 75 years old. They had 15.46 times the risk of dying compared with the youngest age group. Men with CKD were 1.79 times more likely to die compared with men without this disease. Pneumonia and intubation multiplied by 2.05 and 3.12 respectively the risk of dying for men. Men admitted to hospitalization and ICU were 5.42 and 1.39 times as likely to die, respectively, compared with those who did not enter these units. At any time during follow-up, men who needed intubation were 3.12 times as likely to die compared with those who did not require intubation. Men who received health care from IMSS had 6.35 times the risk of dying compared to those who received health care in private facilities.

The results of our analysis, with the Kaplan-Meier survival method and multivariable Cox proportional hazards regression model, demonstrated that men had a higher risk of dying due to COVID-19. In both sexes, older age, CKD, development of pneumonia, hospitalization, intubation, and health care in public health services are independent risk factors increasing the risk of death due to COVID-19. ICU admission was a significant risk factor only in men.

The systematic review and meta-analysis from Galbadage et al. [12] showed that men are more likely to die from COVID-19 than women. Although a clear explanation of this association has not been established yet, it has been suggested that some biological and immunological pathways can play an essential role in the differential behavior of COVID-19 [13]. Some socio-behavioral and cultural aspects can also explain the more unfavorable scenario for men [14, 15]. It has been reported that being over 50 years old increases the risk of fatal complications for this disease [16, 17]. Both sex and old age play an essential role in the risk of the Mexican population. It is especially striking that men in the oldest age group have about 16 times the risk of dying from COVID-19.

ICU admission and intubation are indicators of high-level severity of the disease, both represent an increased risk of death [18]. In our analysis, intubation represents a notable risk factor since it more than doubles the risk of death in both men and women.

The mean time from the onset symptoms to death (Table 1) was different from that reported in other studies suggesting that death occurs between 14 and 21 days after the onset of symptoms [19]. We found that the time between the onset of symptoms until seeking care was similar between people who died and those who survived (p = 0.7302). It can be hypothesized that people with a high risk of severe COVID-19, e.g. with comorbidities, arrived in a critical state to the health services.

A high proportion (58%) of people who survived COVID-19 had no comorbidities. The high prevalence of chronic diseases, in both individuals who died and those who did not, reflects the epidemiological context of Mexico [9, 10] but also the impact of chronic disease in the COVID-19 presentation and severity. These data allows us to hypothesize that chronic diseases may not only increase the risk of complicated disease but also the risk of acquiring the novel coronavirus. A possible explanation is that chronic inflammation, adverse effects on immunomodulation, and metabolic stress that characterize systemic diseases, decrease the ability to react against external agents; in this case, SARS-CoV-2 [20‐22].

In our analysis, CKD a common comorbidity of diabetes [23] and hypertension [24], increased almost two times the risk of death in comparison to people without this disease. Cheng et al. [25] reported that 13.1% of the people admitted to hospitalization in China had kidney failure. This prevalence is significantly larger than the 2.34% we found in the Mexican population diagnosed with COVID-19. Besides, in the Chinese study, the mortality rate due to CKD (33%) was more extensive than in our population. While in China, kidney disease was diagnosed at the hospital admission by laboratory tests, in Mexico, the diagnosis was registered on a self-reported basis. Thus, the prevalence and mortality due to this cause may be underestimated in Mexico. The Kaplan-Meier curves in the Chinese study also show that individuals with kidney disease had a significantly higher risk of in-hospital death. A similar pattern can be observed in the present study. It has been suggested that SARS-CoV-2 may bind to renal epithelial cells, injure these cells, and subsequently disrupt the whole body fluid, acid-base, and electrolyte homeostasis, thus increasing the failure in those with preexisting CKD therefore accelerating their deaths [26, 27].

A relevant finding of our analysis is that the crude lethality rate of 9.4% is above the 5.3% initially reported in Hubei, China [28], and 6.05% in the United States. However, it is below the 11.76% for Spain, 13.98% for Italy, 14.37% for the United Kingdom and 19.35% for France [29]. Such discrepancies can be explained by the way the COVID-19 cases are confirmed and deaths registered. In most countries, only in-hospital deaths are recorded, and in countries like Mexico, in the absence of symptoms, tests are not used in large quantities [30].

Finally, according to our results, different health service providers may have an essential role in lethality. Public providers reported a larger risk of death in comparison with private ones. IMSS presents the highest risk of death (hazard ratio 4.34 in women and 6.35 in men), followed by SS services (hazard ratio 3.22 and 3.52 in women and men, respectively). It is especially important to note that SS and IMSS are the institutions that have received the most cases of COVID-19 in this epidemic, both of uninsured and insured population, hence, having a high probability of over saturation.

The analysis of these results must consider some limitations. First, the dataset does not include clinical variables related to the evolution of the COVID-19, which could have been useful in adjusting the model. Second, since the COVID-19 pandemic is ongoing, not all outcomes of the pandemic have been observed. Finally, The database contains information on all deaths that were confirmed as positive cases of COVID-19. Nevertheless, there is a high percentage of deaths classified as atypical pneumonia which were not confirmed as cases of COVID-19. Those cases may lead to underestimate the case fatality rate for this disease.

Nonetheless, the present paper contributes to understand the dynamics of the pandemic in Mexico. We highlight risk factors for death that have been little studied, especially the presence of CKD and data about the health system that reflects the social inequality in Mexico.

This analysis adds to previous evidence [31] on the higher risk of death for men and older people, both in Mexico and elsewhere. Further research is necessary to understand the origin of this differential risk. One of the specific goals of the basic and clinical research should be to identify differences in response to vaccination to ensure the efficacy of the novel vaccines.

We consider that other risk factors will become relevant over time, so epidemiological research will continue to be paramount.

Given the persistence of SARS-Cov-2, it is vital to focus mitigation campaigns on the population at higher risk of fatal outcomes. We also need stronger public health campaigns aimed at reducing the prevalence of obesity, diabetes, hypertension, and their comorbidities. These conditions may increase the susceptibility of the Mexican population to COVID-19 and cause a rapid progression to severe states of the disease and death.