Introduction

At the end of 2019, some cases of “pneumonia of unknown cause” were noticed by the Wuhan Municipal Health Commission in China’s Hubei Province [1]. Collected bronchoalveolar-lavage samples were consistent with an RNA virus of the Coronaviridae family [2]. Thereafter, the World Health Organization (WHO) named the novel infectious pneumonia “coronavirus disease 2019” or COVID-19 [3]. COVID-19 was quickly proven to be highly contagious, reaching approximately 300 cases in China by January 2020 [4]. Soon COVID-19 became a pandemic, with more than 44,000 infections and more than 1,000 deaths in China, with 441 cases outside China in 24 countries [5].

The first case of COVID-19 in Brazil was reported on February 27, 2020 in the city of São Paulo. Based on published events, eight of the 27 federated units of Brazil present cumulative mortality rates above 10 per 100,000 inhabitants: four in the north, two in the northeast, and two in the southeast region (including Rio de Janeiro and São Paulo) [6]. Until November 2020, Brazil officially recorded 5,468,270 cases of COVID-19 (2,602 per 100,000 inhabitants) and 158,456 COVID-19 deaths (75 per 100,000 inhabitants). The five federative units with the highest mortality counts are São Paulo (39,007 deaths), Rio de Janeiro (20,376 deaths), Ceará (9,325 deaths), Minas Gerais (8,872 deaths), and Pernambuco (8,587 deaths). The highest cumulative mortality rates above 10 per 100,000 inhabitants are found in Ceará, with 102 deaths per 100,000 inhabitants [7]. In the state of Rio Grande do Sul (RS), the southernmost state in Brazil with 11.3 million people, the first case of COVID-19 was diagnosed on February 29, 2020. As of August 6, 2020, 76,563 confirmed cases (673 per 100,000 inhabitants) and 2,163 deaths (19 per 100,000 inhabitants, 2.8% of confirmed cases) have been reported [8, 9]. As of November 2020, RS recorded 240,694 COVID-19 cases (2,116 per 100,000 inhabitants) and 5,699 deaths (50 per 100,000 inhabitants) [8].

Although a significant investment has been made worldwide to provide antiviral prophylaxis for COVID-19, to test different drugs for prevention or treatment COVID-19 cases, and to develop vaccines [10], current recommendations to reduce the spread of COVID-19 include physical distancing [11], quarantining, and large-scale lockdowns of entire populations [12, 13]. Evidence indicates that the implementation of social distancing can suppress COVID-19 transmission rates to prevent the disease from overwhelming the healthcare system. In an analysis of 49 countries, Atalan [14] showed that the COVID-19 pandemic can be suppressed by lockdown measures. In another study including data from 131 countries [15], a decrease in the transmission rate of COVID-19 was observed within 1–3 weeks following the introduction of school closures, workplace closures, public events bans, stay-at-home orders, and limits on internal movement. However, the reduction of transmission ranged from 3% to 24% approximately one month following the introduction of the recommendations, and the effect was only statistically significant for public events bans [15]. Similarly, in New Zealand (a country of 4.886 million inhabitants), the estimated COVID-19 case infection rate decreased from 8.5 to 3.2 per one million people after the implementation of a nationwide the lockdown, resulting in a low relative burden of disease [16]; until now New Zealand has accumulated only 1,973 COVID-19 cases and 25 deaths. Although social distancing and lockdown measures appear to be successful, there is “social fatigue” associated with following these recommendations, leading many societies to return to a usual lifes, increasing COVID-19 transmission [17, 18].

To investigate this, a population-based survey EPICOVID-RS was conducted in nine cities in Rio Grande do Sul from April to June of 2020 [9], including the city of Ijuí, population 83,475, located in the northwestern region of the state. The survey tested 2,500 subjects for the presence of antibodies against SARS-CoV-2 over the course of five different rounds of ~500 subjects each. The seroprevalence of antibodies against SARS-CoV-2 was undetectable (i.e. 0 positive cases) on April 12^th (round 1), and increased to 0.042% [8, 9]. Participants answered questions regarding social distancing adherence (SDA) and daily preventive routines (DPR) [9]. In parallel, the local government registered and published the official number of COVID-19 cases in Ijuí (for details, see the official government website [19]), while the mobile social distancing index (MSDI) in Ijuí was monitored by means of data collected from mobile geolocation [20].

With no pharmaceutical treatment for COVID-19 available, and a vaccine still months away, interventions have focused on quarantine and social distancing. The aim of these strategies is to slow down the spread of infection and reduce the intensity of the transmission (i.e., “flatten the curve”) [21]. The initial COVID-19 outbreak in Wuhan, China indicates that critical care capacity can be exceeded if distancing measures are not implemented quickly or strongly enough [22]. An effective social distancing strategy and monitoring may reduce the risk of overwhelming the health system, allowing adequate patient care and decreasing mortality rates [22]. Using data on the local COVID-19 cases and social distancing behavior, we demonstrate that low levels of social distancing were insufficient to prevent the COVID-19 outbreak in Ijuí, Brazil.

Materials and methods

Results

The daily MSDI computed using data from mobile monitoring for the city of Ijuí is shown in Fig 1A. On the first day of mobile monitoring, the MSDI was 30.7% (Saturday February 1, 2020), after which point it increased to 39.3% (Sunday February 2, 2020) and then decreased to 27.5% in the first weekday registered (Monday February 3, 2020). These data describe the normal social behavior in Ijuí, which remained stable with only a few fluctuations until mid-March, when MSDI grew to 70.2% (Sunday March 22, 2020) and stabilized between 50% and 55% on subsequent weekdays. However, at the end of June, the MSDI decreased to 31.4% (June 19, 2020), reaching the lowest value since city’s first registered case of COVID-19 on March 18, 2020 (Fig 1B). The last registered MSDI value was 35.3% (June 30, 2020). For a fair comparison, percentages pointed above the open circles in Fig 1A correspond to MSDI on weekends, whereas the closed circles correspond to MSDI on working days. We observed an increase in distancing on the weekends, indicating that more people tended to stay home on weekends vs. weekdays. Specifically, on weekends MDSI was ~12% higher than on working days (44.87 ± 9.70 [95% CI: 41.99–47.75] vs. 36.07 ± 7.45 [95% CI: 34.66–37.48], p < 0.0001, Student’s t-test). However, a decrease in overall MSDI (weekends and weekdays) was registered between March 22, 2020 (70.2%) and June 21, 2020 (48.2%). From a macro point of view, it is evident that the population of Ijuí is loosening social distancing and increasing social interaction up to the end of June.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Mobile social distancing index and COVID-19 cases in Ijuí, Brazil.

A) Mobile Social distancing index (MSDI) B) Total and new weekly COVID-19 cases C) Distribution of COVID-19 cases and hospitalizations by age.

https://doi.org/10.1371/journal.pone.0246520.g001

The person to test positive for COVID-19 in Ijuí (Fig 1B) started to feel symptomatic on March 18, 2020 and was diagnosed shortly thereafter. Until May 17, 2020, there was a slight increase in the number of cases registered in Ijuí (Fig 1B). However, on June 7, 2020 there was a 2.1-fold increase in total COVID-19 cases. A majority of these cases occurred in women (52.4%) and people younger than 60 years of age (86%) (Fig 1C). Until the end of June, Ijuí had a 9.5% hospitalization rate for positive COVID-19 cases; 74.1% of hospitalizations occurred in people over 50 years of age, and no hospitalizations occurred for people under 30 years of age. The majority (70.4%, n = 19) of the hospitalized cases were men as opposed to women (29.6%, n = 8). Until the end of June, there was just one death caused by COVID-19 in Ijuí (an 85-year-old oncologic patient with diabetes and hypertension, male). The urban distribution of COVID-19 cases in Ijuí indicates that approximately 30% were downtown inhabitants (S1 Fig in S1 File).

In each round of the population-based survey, >400 adults were surveyed and tested for the presence of SARS-CoV-2 antibodies, for a total of 2,222 study participants. Characteristics of the study population are described in Tables 1 and 2. The majority of the subjects interviewed were women (~60%) and white (~80%), with a roughly equal distribution across age and education categories (Table 1). Between 32% and 39% of participants reported having hypertension and ~13% and ~10% had diabetes and asthma, respectively (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Sociodemographic characteristics of Ijuí subjects by date of population-based survey.

https://doi.org/10.1371/journal.pone.0246520.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Comorbidities characteristics of Ijuí subjects by date of population-based survey.

https://doi.org/10.1371/journal.pone.0246520.t002

The social distancing population-based survey (initiated on April 12, 2020) revealed that the majority of the subjects reported high SDA (Fig 2A). However, we observed a decrease in the proportion of subjects that declared high adherence in parallel to an increase in the proportion that reported partial adherence (Fig 2A). Similarly, the proportion of the subjects that reported staying at home all the time or only going out for essential needs has decreased, while there was an increase in the number of people that declared leaving their homes daily for work (Fig 2B). This is consistent with the decrease in MSDI just 15 days after the first survey wave, which continued to decrease until the end of May (blue lines in Fig 2C). In contrast, there was a significant increase in the number of COVID-19 cases in Ijuí during the period of May to June (red line, Fig 2C), which may evocate a recovered of SDA and DPR approximately one month later (June) (blue lines in Fig 2C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The association of social distancing adherence (SDA), daily preventive routine (DPR), and mobile social isolation index (MSDI) with COVID-19 cases in Ijuí, Brazil.

A) Social distancing adherence, as indicated in the population-based survey study. B) Daily preventive routine, as indicated in the population-based survey study C) Association between total COVID-19 cases and measures of social distancing.

https://doi.org/10.1371/journal.pone.0246520.g002

We have also analyzed two scenarios to estimate the number of cases in Ijuí after May 17, 2020 (Fig 3A): First, a sustained controlled community infection scenario represented by a linear trend (green line in Fig 3A, estimated by the equation y = 1.416667x - 1.416667, R² = 0.848, P = 0.0004) and second, an uncontrolled outbreak community infection in Ijuí, represented by an exponential progression in the number of cases (blue line in Fig 3B, estimated by the equation y = 1.351649 - (-0.2424262/-0.3155474)×(1—^{e^(+0.3155474x)}, R² = 0.955, P = 0.00019). Worryingly, the number of actual COVID-19 cases registered through July 7, 2020 (277, red line in Fig 3A) was higher than what was estimated by linear and exponential prediction (23 and 120, respectively). The exponential estimation (blue line) predicted that Ijuí could reach 1,000 cases of COVID-19 on August 23, 2020, indicating that Ijuí’s public health system could be overwhelmed.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Predictions for COVID-19 cases, hospitalizations, and deaths in Ijuí, Brazil.

A) Difference between actual and predicted cases of COVID-19 using linear regression (green line) and exponential equation (blue line). B) Predictions without any new intervention after the 100^th case (R₀ = 2.79). C) Predictions after implementing strategies to reduce the transmission rate by 50% 30 days after the 100^th case (R_t = 1.40). D) Predictions after implementing strategies to reduce the transmission rate by 50% 15 days after the 100^th case (R_t = 1.40). E) Predictions without any new intervention after the 100^th case (R₀ = 1.44). F) Predictions after implementing strategies to reduce the transmission rate by 50% 30 days after the 100^th case (R_t = 0.79). G) Predictions after implementing strategies to reduce the transmission rate by 50% 15 days after the 100^th case (R_t = 0.79). All predictions were performed in the free epidemic simulator app from São Paulo University. Available at: https://ciis.fmrp.usp.br/covid19/epcalc/public/index.html.

https://doi.org/10.1371/journal.pone.0246520.g003

Next, we analyzed predictions for the months after the 100th case of COVID-19 in Ijuí. In the worst scenario (R₀ = 2.79), the public and private healthcare systems would be overwhelmed with >4,000 hospitalizations and >1,500 deaths in 120 days (Fig 3B). In contrast, if the transmission rate of COVID-19 is reduced by 50% 30 days after the 100^th case (R_t = 1.40), a moderate decrease in the impact on the healthcare system is expected (Fig 3C). We further analyzed the impact on the transmission rate of COVID-19 if the transmission rate fell by 50% (R_t = 1.40) 15 days after the 100^th case (Fig 3D). This scenario reduced the peak infection to 1,314 cases, expected to occur in 66 days. Consequently, the peak of hospitalization (1,582) would occur later, 85 days after the 100^th case, and result in 862 deaths in 120 days. If this RS transmission rate (R₀ = 1.44) persisted without any alteration after the 100^th case, it would result in a higher number of infections, hospitalizations, and deaths peaking after two months (Fig 3E). However, a 50% decrease in the transmission rate 30 days after the 100^th case (R_t = 0.79), would considerably minimize the impact of COVID-19 on the healthcare system, resulting in a peak of 156 daily cases 34 days later, a peak of 120 daily hospitalizations 60 days later, and a peak of 62 daily deaths 120 days later (Fig 3F). Finally, if the transmission rate was reduced by 50% (R_t = 0.79) 15 days after the 100^th case (Fig 3G), the infection peak would reach 75 daily cases and 52 daily hospitalizations, resulting in 27 deaths. The proportion of the whole population that needs recovery decreased from approximately 82.0% in the worst prediction to 1.37% in the best one. Specifically, in the most optimistic prediction (a 50% reduction in the transmission rate 15 days after the 100^th case of COVID-19) approximately 98.7% of the population of Ijuí would remain without infection, and approximately 1.37% would be recovered (1,162 subjects).

Discussion

Our study describes the SDA, DPR, and MSDI among the population of Ijuí, Brazil, and predicts the progression of COVID-19 under different scenarios. To the best of our knowledge, this is the first study to describe the social distancing behavior of a community based on a population-survey procedure and mobile monitoring data that preceded the COVID-19 outbreak. Furthermore, this is the first report to estimate the progression of COVID-19 in Ijuí.

Although there are currently many studies about social distancing behavior and COVID-19 cases in Brazil, it is imperative to study local data separately. Brazil is a vast country comprised of 26 federative states and the Federal District, and there are many cultural, economic, educational, and geographic differences between states and between different cities in the same state. Given these differences, trends in the number of cases and deaths differs between states and cities, and municipalities have the autonomy to determine which measures to adopt in order to best mitigate COVID-19 according to their respective scenarios [33]. All of Brazil’s states implemented distancing measures, mostly after March 15, 2020. Partial economic lockdown was implemented before the tenth confirmed case of COVID-19 by 18 (67%) states and before the first death from COVID-19 by 24 (89%) of the states [33]. In April 2020, of nine major cities in Rio Grande do Sul, the biggest cities (Porto Alegre and Santa Maria) exhibited the highest degree of social distancing, while Ijuí had a less favorable pattern, with a strikingly higher percentage reported being “out of the house all day” [25].

In February 2020, the city of Ijuí, as well as its neighboring cities, had not registered any cases of COVID-19. The first case registered in the state of RS occurred on March 10, 2020 in the city of Campo Bom– 402 km away from Ijuí, and very close to the capital of the state, Porto Alegre. The first case of COVID-19 registered in Ijuí dates back to March 18, 2020. Ijuí joined the official counting of cases after eight days; most probably, the absence of cases before this date explains the non-change in the percentage of social distancing (respecting its natural peaks on Sundays).

On March 19, 2020 the city administration issued its most restrictive decree in the analyzed period, The decree restricted public transport, closed stores, suspended classes at schools and universities, and established special protocols for restaurants and other services for one week [19]. Accordingly, on March 22, 2020 the MSDI reached its highest value: 70.2%. Other subsequent decrees were issued in Ijuí, but with more relaxed restrictions. In parallel, the Federal Government of Brazil has been minimizing this pandemic and, in most cases, encouraging people to keep their regular routines [34]. Thus, we observe a continuous reduction of the social distancing, to the point that at the end of June, MSDI fell to the same levels seen at the beginning of the analyzed period, when there were no cases of COVID-19 in Ijuí or even in Brazil as a whole [7, 35]. Our results were similar to those reported by Oliveira and colleagues [36], presenting a mean isolation index from February 1, 2020 to April 10, 2020 of 40.2%, ranging from 18.5% to 69.4%. Specifically, this study analyzed mean isolation index in the states of São Paulo (13.5% to 67.9%) and Rio de Janeiro (16.6% to 69.4%) and found that social isolation indexes of 46.7% have the highest accuracy (93.9%) to predict R(t) <1 [36], which means that the epidemic is slowing. These data reinforce the validity and reliability of MSDI as a behavioral indicator of social distancing.

In the final days of June 2020, we observed a slight increase in social distancing, most likely as a consequence of a high number of new cases reported in the second half of June, which encouraged the city administration to review its protocols and implement stricter recommendations [19]. The mean MSDI over the total period of mobile monitoring was 38.5%. Taking into account that “staying at home” was defined as a mobile not moving outside a 450 m radius [20], it is possible that the actual social distancing level in Ijuí was below 30% in regular times, since a 450 m distance still allows neighborhood social interactions, including small markets.

Our results about SDA and DPR may be related to the official recommendations by local and state governments. Since May 10, 2020 state of RS has implemented the “Controlled Social Distancing Model” (CSDM). To minimize the spread of COVID-19, this model outlines the safety of various economic activities in a color-coded manner, ranging from yellow (representing a low risk of infection and promoting high health care capacity) to black (representing a high risk of transmission and promoting high occupation rates of hospitalization and intensive care unit visits). Since May 10, 2020, every city in RS is classified according to this model to represent the progression of the COVID-19 outbreak [8]. Ijuí was categorized as yellow from May 10 to June 8, when it became orange (one level above of risk) because of the rapid increase in the number of cases (95 new cases) and hospitalization rates in the intensive care units (5 subjects) in one week and the subsequent decrease in the ratio of recovered-to-active ratio cases. As a local initiative to counteract the COVID-19 outbreak, Ijuí Municipal Decree number 7.107 (issued June 16, 2020) improved the recommendations to decrease the risk of community spread of the coronavirus [19]. In detail, the decree allows the maintenance of regular service in restaurants (between 7 AM and 11 PM) with reduced capacity, while snack bars and coffee shops were allowed to work only in delivery or drive-thru settings. The non-essential commercial businesses were allowed to operate, but only at a ratio of one customer per employee, respecting the limits established in the occupational and operational protocols. Sports clubs were allowed to operate exclusively for the physical conditioning of the respective contracted professional athletes, observing the minimum distance of two meters between them. Physical contact or agglomerations were prohibited in all public settings [19]. Thus, the levels social distancing we measured in Ijuí may have been in accordance with these official recommendations, but were nevertheless insufficient to prevent a rapid increase in the number of COVID-19 cases. (For Ijuí urban geographic details, please see S1 Fig in S1 File, with geographic localization of COVID-19 cases in June and August).

Social distancing measures appear effective, mainly when implemented in conjunction with the isolation of people who test positive for COVID-19 and quarantining of anyone who has been in contact with them [31, 37]. Our data indicates that that preventive behavior among the population of Ijuí, related to the SDA recommendation and DPR, did not reach 70% participation in social distancing in any of the five waves of the survey. We mentioned the threshold of 70% since it was proposed early that maintaining social distancing at a maximum of 76% could prevent 90,000 COVID-19-related deaths and keep intensive care units in São Paulo from being overwhelmed [38]. As of November 23, 2020, São Paulo has registered more than 394,000 COVID-19 cases and 14,000 deaths [7], suggesting that maintaining and strengthening current social distancing measures, isolating COVID cases, and quarantining people who have been in contact with others who have tested positive, is absolutely vital to avoid serious stress to Brazil’s healthcare system. Furthermore, it has been suggested that more restrictive recommendations can be more effective in reducing the number of infected subjects [1, 38–41], and it is necessary to apply such recommendations immediately and rigorously, especially to control the spread of COVID-19 in schools, since children and teenagers may have a disproportional contribution to an increase in the transmission rates [18]. However, as reported by WHO, many countries have reported an increase in “pandemic fatigue” among the population, characterized by lack of motivation to follow the recommended social distancing behaviors to protect themselves and others from the virus (WHO) [42]. On March 13, 2020 the United Stated issued a national proclamation that almost immediately resulted in a large number of people sheltering at home and reducing their daily movements, in line with the MSDI trends we observed in Ijuí in March. From early April to mid-April, the MSDI reached an upper limit followed by a plateau, indicating “social distancing inertia” in Ijuí. After that, a reduction in social distancing measures occurred even in the states that maintained the recommendation of mobility restriction, an example of “quarantine fatigue” [17].

Alongside this, considering data from Middle East respiratory syndrome (MERS) in Saudi Arabia in 2014 and SARS caused by SARS-CoV-1 in China in 2003, the psychological effects of a new pandemic tend to be more pronounced, widespread, and longer-lasting than the pure somatic effects of the infection, and the “epidemic of the fear” may be worse than the disease itself. It has been estimated anxiety about the possibility of infection ranges from 24% to 83% at the beginning of an epidemic, while the long-term epidemic period may trigger or exacerbate stress-related mental disorders such as mood disorders, anxiety disorders, and post-traumatic stress disorder [18].

An intrinsic limitation of our study is the variability of self-perception about COVID-19 risk and what it means to socially distance. People see and act in different ways, depending on whether those things are perceived as psychologically relatable [43]. Thus, people respond to social distancing recommendations according to their empirical constructs. The construal level theory (CLT) of psychological distance has been recognized as a way to discuss the judgment and decision-making related to distance perception, which comprehends a mutual meaning of distance dimensions: temporal distance, social distance, spatial distance, and hypotheticality (i.e., distance from actuality) [44]. As a consequence, individual constructs about distance may influence evaluation, prediction, and behavior. In the same way, individuals may have different conceptions about risk, even if we consider risk directly as the chance of injury, damage, or loss [45]. As for the concept of temporal distance, Li and colleagues [46] suggested that people answering the survey could be influenced by the immediate pandemic-related context and details. Furthermore, because COVID-19 is primarily transmitted through close contact, people are more sensitive to implement social distancing with strangers and tend to believe that their behavior is in accordance with municipal social distancing regulations [46].

In this scenario, perceptions of risk play a key role in a process called “social amplification of risk.” Social amplification of risk is triggered by the occurrence of an adverse event (whether major or minor) and reflects the fact that the adverse impacts of such an event sometimes extend far beyond the direct damages to victims and may result in massive indirect impacts [45]. Also, extensive media coverage of an event can contribute to heightened perceptions of risk and may have influenced the answers in our survey. Recent studies have shown that factors such as gender, race, political views, affiliations, emotional affect, and trust are strongly correlated with risk perception. Equally important is that these factors can influence the judgments of experts as well as laypeople [18, 45]. Our study is limited in its ability to investigate psychological influences during COVID-19 pandemic period, and further studies regarding the “feelings of subjects” about social distancing are recommended to better understand the phenomenon of “fatigue quarantine behavior” worldwide.

At the end of June 2020, Brazil reached more than 1.4 million COVID-19 cases [7, 35]. On June 30, RS reported 26,941 cases, or 344 cases per 100,000 inhabitants [8]. From June to July 2020, the hospitalization rate in RS decreased from 13% to 11%, while the proportion of COVID-19 deaths among hospitalized patients rose from 2.3 to 2.5% (currently 8.5 per 100,000 inhabitants). If Ijuí had the same rates of cases, hospitalization, and deaths as RS in June, it would have 287 cases, 31 hospitalizations, and seven deaths. In reality Ijuí had 283 cases, 27 hospitalizations, and one death. Ijuí ended the month of June with 80.8% of intensive care unit beds occupied and three COVID-19 patients [8, 19].

In our “optimistic” prediction (reducing transmission by 50%, R = 0.79, for 120 days after the 100^th COVID-19 case) the infection peak was predicted to have 75 simultaneous COVID-19 cases and 52 hospitalizations. However, at its peak (August 6, 2020, ~60 days after the 100^th COVID-19 case) Ijuí had 156 COVID-19 cases and eight hospitalized patients (the total number of hospitalized patients was 49, or 9.4% of COVID-19 cases). Our predictions also showed that 120 days after the 100^th COVID-19 case, Ijuí would have a total of 27 deaths. Until August 6, 2020 Ijuí had only five deaths, allowing us to infer that the population at least partially followed the recommendations to avoid the transmission of the coronavirus and also reflects improvements in medical knowledge about COVID-19 and better ways of treating patients. In fact, 120 days after the 100^th COVID-19 case, Ijuí reported a total of 25 deaths, closer to our prediction. The worst period of the COVID-19 pandemic in Ijuí (as of November 2020) was in September and October. Cases and deaths have continued to accumulate with a sustained number of 100–183 cases per week for seven consecutive weeks) reaching 2,153 cases and 38 deaths in the middle of November. This sustained transmission of COVID-19 indicates that our model failed to predict the real behavior of the community over 120 days even if was able to correctly predict the number of COVID-19 deaths.

Up-to-date COVID-19 worldwide data preliminary indicates an increased risk for developing a severe form of COVID-19 in people with comorbidities, advanced age, or who are male. Pre-existing comorbidities were present in about half of patients with the severe form of disease; 30% had hypertension, 19% had diabetes, 8% had coronary heart disease, and 3% had a previous pulmonary condition such as chronic obstructive lung disease [47, 48]. COVID-19 patients with these comorbidities were also among those with highest mortality rates, with an adjusted OR of 7.42 (95% CI: 6.33–8.79) for people with hypertension, 9.03 (95% CI: 7.39–11.35) for people with diabetes, 12.83 (95% CI: 10.27–15.86) for people with coronary heart disease, and 7.79 (95% CI: 5.54–10.43) for chronic obstructive lung disease [49]. Furthermore, 31% of COVID-19 cases, 45% of hospitalizations, 53% of intensive care unit admissions, and 80% of deaths occurred among subjects aged ≥65 years [50]. Similarly, the case-fatality rates among individuals aged ≥80 years are approximately 20% [51]. The discrepancy between males and females is also noteworthy: Men account for 60% of hospital admissions and 70% of hospital deaths [1]. In the Lombardy Region of Italy, these figures are even more exaggerated: 82% of the patients admitted to intensive care units were older men [50]. This risk profile (i.e. male, older, and with existing cardiovascular disease) may be related to an impaired immune-metabolism stress response. Thus, these subjects cannot resolve virus-induced inflammatory bursts physiologically and are susceptible to exacerbated forms of inflammation [52], leading to a fatal “cytokine storm” [53, 54]. Thus, in future studies it is necessary to examine adherence to protective social distancing behaviors together with the presence of comorbidities [3, 55]. Greater engagement in preventive behavior among older subjects and people with chronic diseases in the community may reduce hospitalization and mortality rates.

Conclusion

The insufficient engagement in social distancing behavior registered in this population-based study in Ijuí, Brazil may be related to the rapid increase of COVID-19 cases in this city. Our data predict an approaching outbreak of community spread of COVID-19, which could be avoided or attenuated if the levels of the social distancing among the population increase in the coming weeks.

Supporting information

Acknowledgments

We would like to thank all volunteers who participated in the population-based survey. We would also like to thank Professor Airam Sausen for her courses on COVID-19 prediction math.