We used a SUEIHCDR compartmental model to project thousands of scenarios for the transmission dynamics of COVID–19 in São Paulo, Brazil through the next two years. We used advanced algorithms to model scenarios related to strategy type, SD magnitude, time window, and level of personal protection. The goal was to determine the best-case scenario to control the current peak of infections and avoid a second pandemic wave.
Controlling first peak
Currently, Brazil has the highest rate of transmission in the world with an estimated R0 of 2.816, indicating that it has yet to contain the first peak in infections and associated deaths due to the COVID–19 pandemic. Data from around the world (e.g., Asia, Europe, and North America) indicate that it is possible to mitigate the spread of COVID–19 with widespread SD and PPM measures3. However, our analysis of location data11,12 indicate that the current level of SD is at only 42% for Brazil and for 52% São Paulo. In addition, the current protection values are 60% for the state and 69% for the country. Not only does our model indicate that current SD and protection values are insufficient in controlling the pandemic, they will have dire consequences on the overall number of infections and associated deaths with an extremely large first peak. In addition, our model suggest that this will result in the need of public health resources, especially ICU, exceeding what is currently available. With the current levels of SD and protection, our model predicts that the ICU needed will surpass the available ICU in Sao Paulo at the first drop of SD levels, unless protection increases (Figure 1). The scenario may be even worse because not all ICU beds available are exclusively dedicated to covid–19 patients and because some cities may experience the health-care system failure before others.
According to our model, it is possible for the state of São Paulo to gain control over the first peak if levels of personal protection are significantly increases (at least 10%). Alternatively, an immediately increase SD to values over 75% may solve the problem of the first peak as well. However, increasing SD to values in 75% may result in a second peak if the restrictions are lifted within 40days. SD at 75% represents a complete lock down13. Thus, such a strategy should be used with caution. Further, it should only be used in critical cities and not the State as a whole. Our model suggest that if this high-level of SD is used, it is only necessary for a short time (until end of June 2020).
Note, however, that lower levels of SD will need to be in place for years to come to maintain control over the pandemic14,15. Widespread use of PPM (e.g., wearing facemasks, frequently washing hands, using hand sanitizer, maintaining physical distance between other people, and avoiding agglomerations) and high rates of testing have been emphasized to mitigating the spread of COVID–19 in addition to SD15–19. The results of our model agree; our model indicates that an increase in personal protection to a level of 70% for São Paulo in combination with SD is necessary to contain the concurrent peak in infections and associated deaths. Eikenberry et al. (2020) estimated that the efficacy of using face masks ranges between 50 to 90%, depending upon mask material and fitting20. They assumed that at least 50% protection factor would be achievable for well-made and well-fitted mask usage by the entire population20. The fitting of our model predicts that current level of protection is about 60%. An increase in protection levels would need a massive effort by public health officials to enforce and/or educate people to use facemask and to maintain a 2 meters safe distance from other people. Note, some Brazilian cities have implemented strict SD guidelines and require some level of personal protection19. For example, Belo Horizonte, Rio de Janeiro, and Salvador require facemasks in public21. São Paulo recently announced they will require facemask as well.
Avoiding a second peak
Assuming that Brazil can effectively reduce R0 below 1 with SD and personal protection, it is important to take steps to reduce the likelihood of a second peak. Our models indicate that if SD and personal protection measures are stopped too soon or reduced too much after the containment of the first peak, a second peak in infections and associated deaths will occur. Many experts agree that a secondary pandemic wave is likely if SD restrictions are lifted too quickly22,23. Therefore, determining when and how to relax restrictions has become the focus of epidemiological work around the world. It has been proposed that a responsible exit strategy should continue SD restrictions alongside widespread testing and contract tracing24. Laboratory testing to confirm COVID–19 exposure and/or diagnosis has been a major obstacle for the mitigation efforts worldwide. Unfortunately, testing efforts in Brazil have been deficient. Brazil has conducted the least COVID–19 tests per capita worldwide (https://www.worldbank.org/). In addition, reports indicate that Brazil uses substandard testing kits and only tracks hospitalized cases (https://p.dw.com/p/3cBQi). The lack of data for contact tracing in Brazil suggest efforts are insufficient as well.
Nevertheless, our models suggest it is possible for Brazil, considering São Paulo as a model state, to avoid a second peak. According to our results, the best-case exit strategy to prevent a second peak in São Paulo was a stepping-down strategy over a two-year period. A stepping-down approach would involve a gradual stair-step down. For example, the stepping-down approach we modeled multiplied SD values of 40%, 30% and 20% by 1, then 1/2, then 1/4, then 1/8 and then back to 1 and the we repeated each stair-step down. A stepping-down strategy was also the best-case exit strategy modeled for the US15. A stepping-down approach may be beneficial because it allows for periods of transmission leading to heard immunity without overwhelming public health resources14,25. Alternatively, a one-time SD may result in a catastrophic second peak (Figure 1, yellow curve), if the virus reoccurred and not enough people have immunity14,25. In addition, the stepping-down approach resulted in a 6.5% reduction in total time required to SD over the two-year period, potentially reducing the economic and social costs associated with SD.
The results of the current investigation suggest that an 80-day window between each step was the most beneficial strategy. This result is also consistent with that for the US15. Note, however, the current investigation modeled more time windows and more complex algorithms than the US model. In addition, the best-case stepping-down strategy with an 80-day windows was a magnitude of SD of at least 50%, in the highest windows, or approximately 24% average through the period of analysis. Furthermore, the results suggest that higher protective measures could account for lower SD values, which may be associated with economic and/or psychological benefits. Alternatively, higher SD values delay the onset of the second peak. Delaying the onset of the peak may allow Brazil to procure additional public resources and more time for the development of effective treatments or even a vaccine.
The minimum protection rate for this scenario was 70% for São Paulo. If protection rates are maintained at these levels the models suggest that a second peak would be avoided. However, if protection drops, a second peak would occur and would cross the threshold for available public resources with perhaps devastating consequences. Mortality rates associated with COVID–19 may rise when hospitals become overwhelmed and have fewer resources to treat patients with life-threatening symptoms (https://www.hopkinsmedicine.org/coronavirus). This result provides additional support for the notion that personal protection is critical for maintaining control over the COVID–19 pandemic15–19.