Pooled evidence precision of clinical trials on hydroxychloroquine for Covid-19 treatment was stabilized eight months after the outbreak. (preprint)

OBJECTIVE At the beginning of 2020, hydroxychloroquine showed promising in vitro activity for Covid-19 and several studies were oriented to assess its safety and efficacy. However, after a few months, hydroxychloroquine has proved ineffective. The randomized controlled trials (RCTs) developed quickly and in different settings represent the scientific community capacity to assess drug repositioning effectiveness during a sanitary crisis. Therefore, a critical evaluation of the evidence generated can guide future efforts in analogous situations. We aimed to analyze the RCTs assessing the efficacy of hydroxychloroquine in treating Covid-19, describe their internal validity and power, and evaluate their contribution to the precision of the combined evidence for assessing the mortality outcome. STUDY DESIGN AND SETTINGS This meta-research included RCTs assessing hydroxychloroquine to treat patients diagnosed with Covid-19. It was part of an umbrella systematic review of methods/meta-research (PROSPERO: CRD42022360331) that included a comprehensive search in MEDLINE, EMBASE, Cochrane Library, and the Latin America Database - Lilacs. We retrieved studies published until January 10th, 2022. The risk of bias was assessed using Cochrane Risk of Bias (RoB) 2.0. We analyzed methodology of the studies, precision and random error change through time from pooled evidence, study comparators, patient important outcome, power in different magnitude of effects proxy. RESULTS A total of 22 RCT were included, from that 17 (77%) assessed hospitalized patients and five (23%) outpatients setting. Mortality was related as primary endpoint in only four studies, however half of the studies included composite endpoints including mortality as a component. The internal validity analysis using RoB2 found that eight studies (36%) had a high risk of bias. Only one study had sufficient power to evaluate a moderate magnitude of effect (RR = 0,7 on mortality). The standard error to evaluate efficacy on mortality did not change appreciably after October 2020. From Oct 2020 to Dec 2021, 18 additional studies were published with 2,429 patients recruited. CONCLUSION This meta-research highlights the impact that collaborative, and network scientific research have on informing clinical decision-making. Duplicate efforts create research waste as precision analysis shows that after October 2020, there was not appreciably changes in the precision of the pooled RCT evidence to estimate the hydroxychloroquine effect on mortality.

Integrating epidemiological and clinical predictors of SARS-CoV2 infection in students and school staff in the state of Sao Paulo (preprint)

Background: There is a great deal of uncertainty concerning which contexts would be safe for returning to school and about individual criteria that would reduce contact between the infected and susceptible people in the school setting. Therefore, the purpose of this study was to estimate the prevalence of infection by SARS-CoV-2 in students and school staff; and to identify predictors of infection, including both municipal epidemiological indicators and individual variables reported by the participants. Methods: This was a virological survey carried out among students (over 14 years old) and school staff in Sao Paulo state, between epidemiological weeks 43 to 49 of the year 2020. A self-administrated questionnaire including sociodemographic and clinical information was applied. Moreover, a nasopharynx swab was performed for virological testing (RT-PCR). We evaluate the relationship of COVID-19 epidemiological indicators of the residence municipality with the odds of SARS-CoV-2 infection. For this, a composite index relating recent mortality and previous incidence (RM/PI) was proposed based on the ratio of deaths recorded in the second and third week counted back to the sum of cases during the previous seven weeks (weeks 4 to 10 counted back). We obtained a multiple model using random-effects logit regression integrating epidemiological indicators and individual variables. Results: In total, 3436 participants were included, residents of 72 municipalities. The overall prevalence of infection was 1.7% (95%CI: 1.3%-2.2%). SARS-CoV-2 infection was independently associated with loss of smell, a history of pulmonary disease, and a recent trip outside the municipality. Moreover, the RM/PI index consistently predicted the SARS-CoV-2 infection (adjusted OR: 1.45; 95%CI 1.02-2.04). Based on these associations, we proposed a classification in four groups with different SARS-Cov-2 infection prevalence (0.54%, 1.27%, 3.8%, and 4.13%). Conclusion: Epidemiological and individual variables allowed classifying groups according to the infection probability in a school population of the state of Sao Paulo. This classification could help guide the return to classes in situations in which epidemiological control is evident, maintaining basic protection measures and increasing vaccination coverage.

The impact of legislation on Covid-19 mortality in a Brazilian federative unit was mediated by social isolation. (preprint)

This study aimed to estimate the effect of restrictive laws on actual social isolation and COVID-19 mortality. Moreover, we evaluated how community adherence, measured with an index of social isolation, would mediate the lockdown effect on COVID-19 mortality. Methods: This ecological study assessed the legislations published until June 30, 2020, in the Brazilian state of Ceara. We performed a systematic review and classification of restrictive norms and estimated their immediate effect on social isolation, measured by an index based on mobile data, and the subsequent impact on COVID-19 mortality (three weeks later). A mediation analysis was performed to estimate the effect of rigid lockdown on mortality that was explained for effective social isolation. Results: The social isolation index showed an increase of 11.9% (95% CI: 2.9% - 21%) during the days in which a rigid isolation norm (lockdown) was implemented. Moreover, this rigid lockdown was associated with a reduction of 26% (95% CI: 21% - 31%) in the three-week-delayed mortality. We also calculated that the rigid lockdown had the indirect effect, i.e., mediated by adherence to social isolation, of reducing COVID-19 mortality by 38.24% (95% CI: 21.64% to 56.07%). Therefore, the preventive effect of this norm was fully explained by the actual population adherence, reflected in the social isolation index. On the other hand, mandatory mask use was associated with 11% reduction in COVID-19 mortality (95% CI: 8% - 13%). Conclusions: We estimated the effect of quarantine regulations on social isolation and evidenced that a rigid lockdown law led to a reduction of COVID-19 mortality in one state of Brazil. In addition, the mandatory masks norm was an additional determinant of the reduction of this outcome.

Monitoring social distancing and SARS-CoV-2 transmission in Brazil using cell phone mobility data (preprint)

Social distancing measures have emerged as the predominant intervention for containing the spread of COVID-19, but evaluating adherence and effectiveness remains a challenge. We assessed the relationship between aggregated mobility data collected from mobile phone users and the time-dependent reproduction number R(t), using severe acute respiratory illness (SARI) cases reported by Sao Paulo and Rio de Janeiro. We found that the proportion of individuals staying home all day (isolation index) had a strong inverse correlation with R(t) (rho

The impact of early social distancing at COVID-19 Outbreak in the largest Metropolitan Area of Brazil. (preprint)

We evaluated the impact of early social distancing on the COVID-19 transmission in the Sao Paulo metropolitan area. Using an age-stratified SEIR model, we determined the time-dependent reproductive number, and forecasted the ICU beds necessary to tackle this epidemic. Within 60 days, these measures might prevent 89,133 deaths.

A model to predict SARS-CoV-2 infection based on the first three-month surveillance data in Brazil. (preprint)

Background: COVID-19 diagnosis is a critical problem, mainly due to the lack or delay in the test results. We aimed to obtain a model to predict SARS-CoV-2 infection in suspected patients reported to the Brazilian surveillance system. Methods: We analyzed suspected patients reported to the National Surveillance System that corresponded to the following case definition: patients with respiratory symptoms and fever, who traveled to regions with local or community transmission or who had close contact with a suspected or confirmed case. Based on variables routinely collected, we obtained a multiple model using logistic regression. The area under the receiver operating characteristic curve (AUC) and accuracy indicators were used for validation. Results: We described 1468 COVID-19 cases (confirmed by RT-PCR) and 4271 patients with other illnesses. With a data subset, including 80% of patients from Sao Paulo (SP) and Rio Janeiro (RJ), we obtained a function which reached an AUC of 95.54% (95% CI: 94.41% - 96.67%) for the diagnosis of COVID-19 and accuracy of 90.1% (sensitivity 87.62% and specificity 92.02%). In a validation dataset including the other 20% of patients from SP and RJ, this model exhibited an AUC of 95.01% (92.51% - 97.5%) and accuracy of 89.47% (sensitivity 87.32% and specificity 91.36%). Conclusion: We obtained a model suitable for the clinical diagnosis of COVID-19 based on routinely collected surveillance data. Applications of this tool include early identification for specific treatment and isolation, rational use of laboratory tests, and input for modeling epidemiological trends.