key: cord-0972743-8gl9d79i
authors: Sousa, T. C. M. d.; Moreira, N. d. P.; Krieger, J.; Rosa, I. S. C.; Zamudio, M.; Veras, M. A. S. M.; Kemp, B.; Barberia, L.
title: Socioeconomic Vulnerabilities and the Intensity of RT-PCR SARS-CoV-2 Testing Efforts in the Public Health System in Sao Paulo State
date: 2020-11-03
journal: nan
DOI: 10.1101/2020.10.29.20221960
sha: 8ffc31a312f3cda4ed06c1c53cf4001f92950269
doc_id: 972743
cord_uid: 8gl9d79i

Background The testing of infected persons with SARS-CoV-2 is one of the cornerstones to deploy pandemic control strategies. The public diagnostic effort is particularly important among the most vulnerable socioeconomic districts where the state is the sole health provider, such as Sao Paulo state, the Brazilian epicenter of the COVID-19 pandemic. Methods We developed an RT-PCR testing intensity effort index (RT-PCR TIEI) composed of seven indicators to assess the intensity testing efforts in the state of Sao Paulo. Each Regional Health Department (RHD) was scored using anonymized public data. We used dynamic time-series cross-sectional models to analyze the association between the RT-PCR TIEI in Sao Paulo state and its 17 RHDs from epidemiological weeks 10 to 35, and the proportion of the population living under a high level of socioeconomic vulnerability, dependent on public health service (SUS), per capita income, and population density. The regression models included an intercept and the lag of the RT-PCR TIEI, and standard errors were clustered by RHD. Findings On average, the RT-PCR TIEI score was 23.50. The maximum (47.06) was reached in week 11 and declined in subsequent weeks. The lowest score (17.65) was reached in week 25. In the long-run, socioeconomic vulnerability is negatively associated with RT-PCR TIEI (p-value=0.000, 95% CI -0.896, -0.816), with a higher proportion of the population dependent on SUS (p-value= 0.000, 95% CI -0.877, 0.808) and with population density (p-value=0.000, 95% CI -0.857; -0.806). Conclusion There was a decline in the state's testing intensity as the pandemic advanced, and the most socioeconomic vulnerable RHDs showed the lowest values where local public laboratory presence is a predictor of a higher RT-PCR TIEI score. Thus, the low RT-PCR TIEI and local laboratory capacity inequality may affect surveillance capability, especially for the most socioeconomic vulnerable population.

Governments across the world have demonstrated that more intensive efforts to identify infected individuals are essential to an effective containment of COVID-19 along with other nonpharmaceutical interventions (1) (2) . Molecular tests, commonly referred to as RT-PCR (Reverse Transcription-Polymerase Chain Reaction) tests, are considered the gold standard as they detect the virus's RNA in infected individuals and who are potentially transmitting the virus to other individuals. In turn, serological tests, such as ELISA (Enzyme-Linked Immunosorbent Assay), chemiluminescence, and rapid tests, identify the presence of antibodies produced following a SARS-CoV-2 infection (3) . Due to these differences, RT-PCR tests are considered the first-best test for controlling the transmission of SARS-CoV-2 as an essential part of a surveillance strategy that should also include identifying and isolating infected individuals and ensuring the quarantine of people who were in contact with those who tested positive (4) . In this study, we developed a new index to measure and evaluate the intensity effort of RT-PCR testing within the public health system based on the best-recommended guidelines in testing effort, especially those issued recommendations issued by the World Health Organization (WHO) (4) (5) (6) 

We then analyzed the proposed index to assess the RT-PCR testing intensity efforts undertaken by São Paulo and its 17 Regional Health Departments (RHD), considering the laboratories maintained by the public health system. São Paulo state, one of the most populous states in Brazil (8), has been an important epicenter of the pandemic in the country, and has 62% of its population depending exclusively on the public health system. There has been variation in the RT-PCR testing efforts across the state's regions over the past six months, and since April 2020, the labs in the public health system are centrally managed by São Paulo's state government (9) .

Initially, in late February and March 2020, the earliest cases were registered in the capital and few other cities outside the Metropolitan Region (RHD I), but the cases were largely concentrated in the São Paulo metropolis (10). As of August 31, 2020, the capital accounted for 32% of the cases, and the rest of the state accounted for 68% of the total cases. Since June 2020, the state's interior has surpassed the capital in the number of COVID-19 registered cases (11) , and the RT-PCR testing capacity outside the metropolis has also simultaneously expanded.

Simultaneously, in June, the state is loosening non-pharmaceutical interventions contingent on assessing the pandemic's evolution in each RHD (12) . As a result, the observed differences in testing intensity within the public health system at this spatial unit of analysis are critical. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. ; https://doi.org/10.1101/2020.10.29.20221960 doi: medRxiv preprint

We developed an RT-PCR testing intensity effort index (RT-PCR TIEI) and conducted an observational time-series cross-sectional study to apply the proposed index to assess the RT-PCR public testing efforts in São Paulo state and its 17 Regional Health Departments (RHD).

Given time lags in data reporting, we analyzed information reported from epidemiological week 10 to week 35 (March 1st to August 29, 2020 ).

The official data for the state of São Paulo is based on several government sources. Specimens of anonymized daily data on the RT-PCR tests conducted in all public laboratories and the number of daily new confirmed cases in the state of São Paulo, including mild and severe cases, are drawn from the Health Secretariat's Intelligent Monitoring System (SIMI) (11) . For testing, this data source provides the number of RT-PCR performed in the network of public labs for hospitalized and ambulatory patients. The SIMI database informs only one date; it is not clear if it corresponds to the collection of samples to perform the RT-PCR test or when the result was reported in the system. Data on the 2020 population and social vulnerability were provided by the Foundation for Statewide System Data Analysis (13) (14) . The 2010 Social Vulnerability Index of São Paulo State is based on five socioeconomic and four demographic indicators (14) . Social vulnerability is measured as the percentage of people living in each of the 17 RHD under high or very high social vulnerability (including urban and rural areas), capturing the regional vulnerability. For each RHD, these values represent the mean value of all municipalities in the specific region.

We also obtained the proportion of the population which was exclusively dependent on the Brazilian public health system, known as Sistema Único de Saúde (SUS) in Portuguese, in May 2020. These data are provided by the Information Health System of the Secretariat of São Paulo State (15) . The per capita income estimates are based on 2010 census data (16) ; while the population density was estimated as the ratio between the total population living in each RHD in relation to the geographic area.

To identify specific policies that should be included to develop an RT-PCR Testing Intensity Effort Index (RT-PCR TIEI), we analyzed WHO and CDC technical guidelines and daily press conferences that specifically addressed testing (3) (4) (5) (6) (7), and other academic publications and platforms (17) (18) . Although the CDC is a U.S. institution, its recommendations can be applied in Brazil considering its expertise and experience in is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. ; https://doi.org/10.1101/2020.10.29.20221960 doi: medRxiv preprint surveillance strategies and the COVID-19 pandemic similarities in Brazil and the USA, with stabilization of the number of cases and deaths at a high level (19) .

The RT-PCR TIEI is based on seven indicators, as shown in Table 1: 1. Daily testing targets for RT-PCR tests (17) 2. The RT-PCR positivity rate (4) 3. The diagnostic (e.g., RT-PCR) test turnaround time (specimen collection to test report) (5) 4. Local or regional public laboratory test processing capacity (5) 5. Repeat testing of individuals (3) 6. The percentage of cases interviewed for contacts (7) 7. The quality of RT-PCR testing data provided (18) The composite score is an additive index re-scaled to range from zero (lowest) to 100 (highest level of testing effort).

The first five indicators are based on testing surveillance policies per se; one indicator includes a measure to assess if testing is being used for contact tracing, and the last indicator aims to evaluate the quality of the information provided from official sources. The maximum score is attributable to the observance of mainly WHO and CDC recommendations, and the remaining scores are declining ordinal measures with a minimum value of zero. Table 1 .

We estimated multivariate regression models to examine the association between the RT-PCR TIEI scores (dependent variable), per capita income, population density, the proportion of the population that depends on the SUS and living under high social vulnerability across 17 RHDs in São Paulo State in epidemiological weeks 10 to 35. The regression models included an intercept, and the lag of the RT-PCR TIEI and standard errors were clustered by RHD (20) . In our analyses, the p-values less than 0.05 were considered statistically significant. All analyses were performed using STATA version 16. 

As of August 31, 2020, the state of São Paulo had the highest number of RT-PCR tests performed 3,049,073 tests, with 6,640 tests performed per 100,000 people (11) . All RHD received the highest score for T1 (=3) in the first weeks analyzed (March). During that period, there were at least 30 RT-PCR tests for every positive case. After this period, all RHDs received . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. ; https://doi.org/10.1101/2020.10.29.20221960 doi: medRxiv preprint the lowest possible score for T1 (=0) for the majority of the weeks analyzed. In the RHDs, the ratio between RT-PCR tests and COVID-19 confirmed cases was between 0 and 9, which implies that for each registered COVID-19 case, there were less than 10 RT-PCR tests performed in a week. Only four RHDs received a score of 2 for T1, but these scores were never maintained for more than two weeks in any RDH. There were at least ten RT-PCR tests conducted for each positive case in a given week (generally in weeks 14 and 15). Finally, every RHD with at least one local public lab received a score of one in the lab indicator (T4). In our sample, seventy percent of RHDs (12 RHDs) had at least one local public laboratory. Over the course of 25 weeks, the highest score of 2 was observed in 40.27% of the sample, an intermediate score of 1 was observed in 30.3%, and the lowest score in 29.41% of the sample. Among the RHDs with a local public laboratory, only RHD XIII did not achieve higher overall RT-PCR TIEI scores since this RHD could not score highly on T2 and T4. There was significant heterogeneity across the public labs in the processing capacity of RT-PCR tests per day. On September 17th, the range of variation of RT-PCR tests per day was between 0 (RHDs II (Araçatuba), VII (Campinas), and XI (Presidente Prudente)) and 4,500 (RHD I -Grande São Paulo) (11).

The last proposed indicator, the quality of RT-PCR testing data provided, was similar for all RHDs since there is only one database for all RHDs of the São Paulo State and this source reports the number of collected specimens. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. Additionally, we analyzed how population density, income per capita, the proportion of the population dependent on SUS, and social vulnerability are related to RT-PCR testing. Figure 3 reports Table 2 shows the results of the dynamic time-series cross-sectional models. We use four models to analyze how the RT-PCR TIEI is associated with each explanatory variable in the seventeen RHD between the epidemiological week 10 to 35 . In the bottom panel of the table, we report the long-term predicted effects on the RT-PCR TIEI, including considering the lag dependent variable. In the long-run, socioeconomic vulnerability is negatively associated with RT-PCR TIEI (p-value=0.000, 95% CI -0.896, -0.816), with a higher proportion of the population dependent on SUS (p-value= 0.000, 95% CI -0.877, 0.808) and with population density (p-value=0.000, 95% CI -0.857; -0.806). No evidence of first or second-order autocorrelation was found. Table 2 .

The RT-PCR TIEI, which comprises seven indicators, is a useful tool to analyze the intensity of RT-PCR testing effort to control the SARS-CoV-2 pandemic and provide a metric to assess the public health testing efforts in association with other surveillance measures. Using this approach, our data show that between epidemiological week 10 to 35, the RT-PCR testing intensity effort was highly heterogeneous across the 17 RHD in the state and trending down. The presence of a local public laboratory was the indicator with the highest impact on the final score.

This finding is consistent with the lack of data for three indicators that received zero scores:

diagnostic RT-PCR test turnaround time (T3); repeat testing of individuals (T5); and, percentage of cases interviewed for contact elicitation (T6). All RHDs presented a reduction in the RT-PCR is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. ; https://doi.org/10.1101/2020.10.29.20221960 doi: medRxiv preprint TIEI over the 26-weeks analyzed. In addition, we found a positive association between the lowest scores of RT-PCR TIEI and the RHDs with the largest proportion of people living under high levels of social vulnerability, the population dependent on the public health system and with higher population densities. In the course of the pandemic, the state of São Paulo advanced the intensity of its testing efforts in the public health system using RT-PCR tests for SARS-CoV-2 (11) . However, the response was not homogenous and remained below WHO and CDC recommendations and the best practices adopted in other regions ( Table 1 ). The positivity rates were above 5% and increased or remained at this high level over time. Positivity rates greater than 5% are considered insufficient for WHO (4), as they suggest that most RHDs are largely concentrating RT-PCR testing efforts directed at symptomatic cases. RT-PCR testing as a pandemic control strategy remained limited in the state and was never an effective tool to mitigate the virus's spread. In consonance with our finding, the positivity rate averaged in all state of São Paulo was 30% on 18 October 2020, according to official data (11) , which reveals that the state continues to rank below international counterparts, such as New York State (USA) (21).

One of the factors affecting the disparity in testing intensity across RHDs is the lack of a local laboratory, reflecting low past public health investments in infrastructure before the onset of the pandemic. The majority of RHDs have a local laboratory (12 of 17 RHDs). Laboratories are expensive and difficult to install during a public health emergency (22) . It is expected that tests can be processed more quickly in regions with local laboratories that can provide timely results, given they have the capacity to accommodate the testing demand, and that supplies necessary to process tests (e.g., reagents) are readily available (5) . Some regions were able to further increase their test processing capacity by partnering with universities and research institutes. However, the universities, laboratories, and research institutes collaborating in providing public access to RT-PCR are concentrated in only five RHDs, while others have public laboratories in their territories. There is no public laboratory in RHD XII (23), and its municipalities are among those with the lowest levels of social and economic development (14) .

Indicator seven (the quality of RT-PCR testing data provided) permits examination of the ability to access information, which is fundamental to evaluate and monitor pandemic control measures (24) . Since the beginning of the COVID-19 pandemic, the importance of sharing data has been underscored by the necessity to build and integrate computational infrastructure that can provide large datasets and facilitate data access by public servants, researchers, and the general population (24) (25) . Three of the seven indicators were not available for any RHD. These indicators would provide information that could help identify if those tested receive diagnoses in is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. ; https://doi.org/10.1101/2020.10.29.20221960 doi: medRxiv preprint adequate time to ensure treatment and reduce the viral spread; this is an important area that deserves further transparency. Although some RHDs claimed to have implemented contact tracing programs, such as RHD I (26), III, and VI (27) , an evaluation of contact-tracing efforts across all RHDs was not possible with currently available data.

Our results confirmed that socioeconomic vulnerability is a determining factor affecting access to RT-PCR testing and its subsequent control strategies in São Paulo State. This finding is in line with Souza et al. (26) that found an association between confirmed COVID-19 diagnosis and per capita income in São Paulo, the capital of São Paulo State. Given that demand for testing is not necessarily associated with regional testing supply, and further considering each region's population density, our study's findings highlight structural inequities contributing to the severity of the COVID-19 pandemic. The WHO has emphasized both the importance and the distribution of laboratories that should also be monitored to reduce inequities (5) . The findings provide evidence that a key component of surveillance capacity, laboratory testing, was underperformed during the pandemic and might have contributed to the emergency's poor handling even in Brazil's richest and most well-endowed state.

Differences in access to COVID-19 testing were also found in New York City by race and socioeconomic characteristics based on tested individuals' ZIP codes (28) . The city's most vulnerable communities did not have proportional access to testing, and this disparity affected New York's public health response. The same inequality related to COVID-19 testing access has been found in Brazil, where people living in areas with higher income per capita are more likely to have been tested (29) , despite the difference between the health system in Brazil, which have a public health system, and the USA. The State of Sao Paulo does not disclose individuals' ZIP codes, which could be used to assess if the inequalities across RHDs also exist within a municipality, such as the metropolis. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. ; https://doi.org/10.1101/2020.10. 29.20221960 doi: medRxiv preprint In the third scenario, which includes most of the RHDs in Sao Paulo, the number of tests performed was close to the number of cases recorded for most of the analyzed period. Most likely, RT-PCR testing was used in these RHDs to test only symptomatic hospitalized patients. Therefore, testing was not used to identify new cases and people who have been in contact with infected people.

There are limitations to this study. The Health Secretariat's Intelligent Monitoring System does not report anonymized individual RT-PCR testing data conducted in private labs (11) . Given these limitations, we were not able to estimate the diagnostic RT-PCR test turnaround time.

When observing the evolution of the pandemic in the state of Sao Paulo and the RT-PCR TIEI in the RHDs, there is evidence suggesting that there have been significant public resources directed at RT-PCR testing since March. Nevertheless, the results presented here indicate that the RT-PCR testing effort's intensity needs to increase to improve pandemic control measures.

Unfortunately, there were no data available in processing capacity by laboratory over time.

Given our goal of assessing RT-PCR testing efforts in public laboratories, our study also highlights that the Brazilian Health System (SUS) can play an instrumental role in reducing inequalities. Its decentralized structure allows for targeted policies, considering the distinct epidemiological profiles and vulnerabilities (30) . During pandemics, control measures must be guided by the characteristics of the pandemic settings and the health system's capacity in different locations. In this regard, the SUS allows municipalities and RHDs to adopt distinct strategies to test, identify, and isolate infected people, contact tracking, and quarantine infected people. Finally, our data suggest that in addition to the improvement in testing capacity, the autonomy of RHDs to modify their surveillance and control measures requires that federal and state authorities adopt policies that are aligned with these efforts, which was not the case through the course of the pandemic.

The proposed RT-PCR testing intensity effort index allowed us to analyze the COVID-19 diagnostic testing policies in São Paulo State, the pandemic epicenter in Brazil. Since the obtained RT-PCR scores decline as the pandemic advanced, we conclude that the state's testing intensity effort has been insufficient to control the SARS-CoV-2 transmission. The results confirm that socioeconomic vulnerable RHDs have lower RT-PCR TIEI scores, and local public laboratories are an important predictor of a higher RT-PCR TIEI score. Thus, the low RT-PCR TIEI and local laboratory capacity may affect the pandemic control, especially in the most socioeconomic vulnerable population. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. Regions should have at least one local laboratory for diagnostics for effective pandemic control (5)

Regional (state-wide or intra-state) testing facilities process the majority of the tests, such that they control the quality of the sample collection process, diagnostics and accurate testing 2 Regional (state-wide or intra-state) testing facilities do not process the majority of the tests, such that they control the quality of the sample collection process, diagnostics and accurate testing is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. ; https://doi.org/10.1101/2020.10.29.20221960 doi: medRxiv preprint T5. Repeat Testing of Individuals "Testing that is repeated at different points in time may be more likely to detect infection among close contacts of a COVID-19 case than testing done at a single point in time" (6) .

Local accurate testing availability is being undertaken to repeatedly test individuals over time 1 Local accurate testing availability is being undertaken to repeatedly test individuals over time 0

Contact tracing success may be evaluated by considering the number of contacts tested for SARS-CoV-2 (and identified) in relation to the number of contacts interviewed (7) . 

RT-PCR TIEI Index 0 = minimum score 18 = maximum score

Re-scaled to range from 0-100 Note: 1 suppression level testing "allows a state or community to quickly find and isolate new cases before they lead to a wider outbreak, with an aim of keeping new case levels at or near zero"; 2 mitigation level testing is focused "on reducing the spread of the virus through broad testing of symptomatic people, tracing and testing a recommended 10 contacts per new case and isolating positive contacts, and social distancing, mask wearing or stayat-home orders as necessary" (2) . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. ; https://doi.org/10.1101/2020.10.29.20221960 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. ; https://doi.org/10.1101/2020.10.29.20221960 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. ; https://doi.org/10.1101/2020.10.29.20221960 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. ; https://doi.org/10.1101/2020.10.29.20221960 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted November 3, 2020. ; https://doi.org/10.1101/2020.10.29.20221960 doi: medRxiv preprint

Contact Tracing during Coronavirus Disease Outbreak, South Korea, 2020. Emerging Infect Dis

SARS-CoV-2 Testing Strategy: Considerations for Non-Healthcare Workplaces

Public health criteria to adjust public health and social measures in the context of COVID-19

World Health Organization. Laboratory testing strategy recommendations for COVID-19: interim guidance

World Health Organization

Evaluating Case Investigation and Contact Tracing Success

Regulamenta, no âmbito do Estado de São Paulo, o fluxo para o diagnóstico do novo coronavírus -COVID-19

Sistema de Monitoramento Inteligente -SIMI-SP

Sistema de projeções populacionais para os municípios do estado de São Paulo

Índice Paulista de Vulnerabilidade Social

Estimativa da população SUS dependente (com base na saúde suplementar)

Brazilian Institute of Geography and Statistics (IBGE)

New Testing Targets: As COVID-19 outbreaks grow more severe, most U.S. states still fall far short on testing

The COVID Tracking Project: Counting COVID-19 Tests: How States Do It, How We Do It, and What's Changing

COVID-19 Dashboard by the Center for Systems Science and Engineering (CSSE)

Taking Time Seriously

Coordination and resource-related difficulties encountered by Quebec's public health specialists and infectious diseases/medical microbiologists in the management of A (H1N1) -a mixedmethod, exploratory survey

Open access epidemiological data from the COVID-19 outbreak. The Lancet Infectious Diseases

Data sharing and outbreaks: best practice exemplified. The Lancet

Hearing Loss in Infants with Microcephaly and Evidence of Congenital Zika Virus Infection -Brazil

Araraquara é uma das 3 cidades escolhidas para projeto piloto de rastreamento da Covid-19

Disparities in COVID-19 Testing and Positivity

Pesquisa Nacional por Amostra de Domicílios : PNAD COVID19 : August, 2020 : resultado mensal / IBGE, Coordenação de Trabalho e Rendimento

Integralidade e Autonomia no SUS: Desvinculação da aplicação de recursos federais e os desafios da coordenação