Indirect Impact of COVID-19 Vaccination on Weekly Trends in Morbidity and Mortality Indicators in Brazil, 2020 to 2021


 Background: With the outbreak of the COVID-19 pandemic, safe and efficacious vaccines were quickly developed and made available to reduce its morbidity and mortality. This study aimed to describe the trends and evaluate the association of COVID-19 vaccination with indicators of morbidity and mortality due to the disease in Brazil.Methods: This was an ecological study whose unit of analysis was the time elapsed between February 16th, 2020, from the introduction of COVID-19 in Brazil and July 17th, 2021, six months after the start of vaccination in the country. From secondary databases, indicators of morbidity and mortality (incidence of hospitalization, proportion of severe cases, lethality and mortality) and vaccine coverage (doses 1, 2 or single dose) per epidemiological week (EW) and age groups were calculated (18 to 59 years and 60 years or older). Indicators of morbimortality were compared in the pre-vaccination (EW 8/2020 to EW 2/2021) and vaccination (EW 3/2021 to EW 28/2021) periods, analyzing the trends estimated using the JointPoint model and their associations using the Poisson model.Results: For individuals aged 60 years and over, during the period of COVID-19 vaccination there was a weekly trend towards a reduction in the incidence of hospitalizations (PMSV: -4.7%), severity (PMSV: -0.3%, lethality (PMSV: -2.0%) and mortality (PMSV: -2.8%); the increase in COVID-19 vaccination coverage was inversely associated with the incidence of hospitalizations (IRR: 0.969), lethality (IRR: 0.993), and mortality (IRR: 0.981). For individuals aged 18 to 59 years of age, during the period of COVID-19 vaccination there was a weekly trend towards a reduction in the incidence of hospitalizations (PMSV: -2.6%) and lethality (PMSV: -2.0%), while there was an increase in mortality (PMSV: 3.8%); the increase in COVID-19 vaccination coverage was inversely associated with the incidence of hospitalizations (IRR: 0.975) and lethality (IRR: 0.939). Conversely, it was directly associated with mortality (IRR: 1.054).Conclusions: it was demonstrated that COVID-19 vaccination had an indirect impact in reducing the morbidity and mortality trend from the disease in Brazil, especially in the elderly population, in which vaccination was prioritized and achieved greater coverage, when compared to individuals aged 18 to 59 years.

Initially, considering the limited amount of doses of vaccines and supplies worldwide, WHO published a guide with recommendations from the Strategic Advisory Group of Experts on Immunization (SAGE) to prioritize groups for the use of COVID-19 vaccines, including, rst, health care workers and elderly individuals. (4) In Brazil, COVID-19 vaccination started on January 18th, 2021, prioritizing health workers, elderly individuals living in long-term care facilities, individuals aged 18 years and over with institutionalized disabilities, and the indigenous population; other groups were included in the national plan to operationalize the COVID-19 vaccination, as the availability and acquisition of vaccines gradually increased. (5) Preliminarily, according to the COVID-19 National Vaccination Plan, four vaccines developed on three different platforms were introduced in Brazil: i) non-replicating viral vector (AstraZeneca/Oxford University and Janssen Pharmaceutical), ii) inactivated virus (Sinopharm), and iii) messenger RNA-based (P zer/BioNTech) platforms. (5) The e cacy of these vaccines to prevent SARS-CoV-2 infection ranged from 54% to 95%, and to prevent moderate to critical COVID-19 cases ranged from 67% to 76%. (5)(6)(7)(8)(9) The success of COVID-19 vaccination was observed in several countries that quickly managed to vaccinate a large portion of their population, such as Israel, the United States of America (USA), the United Kingdom, Chile, among others, which were successful in decreasing the number of reported cases and deaths. (10) However, with the emergence of new SARS-CoV-2 variants that invariably had greater virulence or transmission capacity, and could even escape the immunity already acquired (via vaccine or natural infection), which are called variants of concern (VOCs), a new increase in the number of con rmed cases of the disease was observed in these countries. The VOCS detected up to July 2021 are known as Alpha (formerly B.1.1.7, identi ed in the UK), Beta (formerly B.1.351, identi ed in South Africa), Gamma (formerly P.1, identi ed in Brazil), and Delta (formerly B.1.617.2, identi ed in India) variants. (11,12) Considering the possible impact of the above factors on morbidity and mortality from COVID-19 in Brazil, after six months of vaccination against the disease in the country, from February 17th to July 17th, 2021, the objective of this study was to describe the time trend of epidemiological indicators related to morbidity and mortality from COVID-19 and to analyze their association with COVID-19 vaccine coverage, in individuals aged 18 to 59 years and in those aged 60 years and over.

Type of study
This was a time series study (13) whose unit of analysis was time, elapsed in epidemiological weeks (EW), in the period from February 16th, 2020 (EW 8/2020) to July 17th, 2021 (EW 28/2021). The choice of this period considered the introduction of the SARS-CoV-2 virus in Brazil in 2020 and the rst six-month period after the start of COVID-19 vaccination, counted from January 18th, 2021 (EW 3/2021).

Selection criteria
Hospitalized COVID-19 cases, con rmed by laboratory criteria, reported to the In uenza Epidemiological Surveillance Information System (SIVEP-Gripe), and who had the onset of symptoms or progression to death between February 16, 2020 and July 17, 2021, were selected and divided into the following age groups: i) 18 to 59 years of age, and ii) 60 years or older.

Variable and data sources
The following variable and data sources were consulted and analyzed: SIVEP-Gripe: o cial system for recording hospitalized cases and deaths from Severe Acute Respiratory Syndrome (SRAG) in Brazil, whose database is available without case identi cation (anonymized data) in the link: https://opendatasus.saude.gov.br/organization/ministerio-da-saude. The databases used in the study (SRAG-2020 and SRAG-2021) were obtained on August 8th, 2021. Data such as date of symptom onset, age (18 to 59 years / 60 years or more), admission to the Intensive Care Unit (ICU), use of invasive ventilatory support, nal classi cation (COVID-19), termination criteria (laboratory), progression (death), and date of progression were used for analysis.
National COVID-19 Vaccination Campaign: COVID-19 vaccination records in Brazil were made available by the National Health Data Network (RNDS) which integrates different databases such as the Information System of the National Program for Immunizations (SI-PNI), e-SUS APS (Primary Health Care), and state and municipal systems. The database was obtained for analysis on August 8, 2021, using the link: https://localizasus.saude.gov.br/. Data such as date of vaccination, age (18 to 59 years / 60 years or older) and dose (D1, D2 or single dose) were used for analysis. Population estimate: population by age (18 to 59 years / 60 years or more) calculated by the Ministry of Health of Brazil based on census data from the Brazilian Institute of Geography and Statistics (IBGE) for the year 2020.

Indicators
Incidence of hospitalizations from COVID-19 per 1,000,000 population: the numerator used was the number of cases hospitalized per EW of symptom onset and the denominator used was the number of individuals residing in the country according to the age group.
Proportion of severity: the numerator used was the number of cases admitted to the ICU and/or who received invasive ventilatory support and the denominator applied was the number of hospitalized cases, disaggregated by age group and EW of symptom onset.
Lethality: the numerator used was the number of deaths and the denominator used was the number of hospitalized cases, separated by age group and EW of symptom onset.
Mortality from COVID-19 per 1,000,000 population: the numerator used was the number of deaths per EW of the date of progression and the denominator used was the number of individuals residing in the country according to the age group.
Vaccination coverage: the numerator used was the number of individuals per age group with a complete vaccination schedule (2 doses or single dose) for any COVID-19 vaccine established in the National COVID-19 Vaccination Plan and the denominator used was the number of individuals residing in the country by the age group, calculated by EW of the vaccination date.

Data analysis
The following analyzes were performed: 1. Temporal trends were estimated for all indicators using the JointPoint model. To identify the EW (independent variable) in which there were statistically signi cant changes in the trend of epidemiological indicators and vaccine coverage (dependent variables), up to ve in ection points in each time series were used. The weekly percentage of variation (PSV) was calculated for each segment in which there was an in ection in the time series and the average weekly percentage of variation (PMSV), for the pre-vaccination (EW 8/2020 to EW 2/2021) and vaccination (EW 3/2021 to 28/2021) periods, including their respective 95% con dence intervals (95% CI). The signi cance test used to select the best model was the traditional Bayesian Information Criteria (BIC3) method, considering a statistical signi cance level < 0.05.
2. Association of epidemiological indicators with vaccination coverage: analyzed by simple Poisson models, whose dependent variables were the epidemiological indicators (incidence of hospitalizations, severity, lethality and mortality from COVID-19), analyzed separately, and the explanatory variable in all cases was the vaccination coverage with a complete vaccination schedule (D2 or single dose).
Incidence-rate ratios (IRR) and their respective 95% CI were calculated, considering a signi cance level < 0.05. Model t was assessed using the pseudo R 2 statistical method, which measures the reduction in deviation due to the explanatory variable.
The statistical programs used were JointPoint Trend Analysis 4.9 and STATA v. 13. The databases obtained for analysis are publicly accessible and do not have personal identi cation variables, so the study was not submitted for appraisal by an ethics committee for research with human beings, according to the Resolution of the National Health Council number 510/ 2016 which provides rules applicable to methodological procedures involving the use of secondary data in Brazil.

Results
There were 2,083,217 cases of SARS reported during the study period, of which 1,371,187 (65.8%) were con rmed for COVID-19. Of these, 1,215,314 (88.6%) cases were con rmed by laboratory criteria, of which 621,703 (51.2%) were between 18 and 59 years of age and 573,812 (47.2%) were aged 60 years or older. There were 263,408 deaths from COVID-19 in individuals aged 60 years and over, and for 692 (0.3%) there was no record of the date of death. There were also 120,271 deaths from COVID-19 in individuals between 18 and 59 years of age, and for 376 (0.3%) there was no record of the date of death.
Six months after the introduction of COVID-19 vaccination, rst and second doses vaccination coverage reached, respectively, 47.7% and 35.0% in individuals aged 60 years and over, and 45.9% and 9.2% in individuals between 18 and 59 years of age.
Considering that individuals aged 60 years and over, for the most part, were prioritized to vaccination in Brazil, the results for this age group will be shown before the ndings for individuals aged 18 to 59 years, which will be shown in sequence.
Progression of COVID-19 and vaccination in individuals aged 60 years and over Figure 1 shows the distribution of observed and estimated weekly trends of epidemiological indicators of morbidity and mortality from COVID-19 and rst and second doses vaccination coverage, in individuals aged 60 years and over. Table 1 shows the PSV that was calculated for each change in the time trend of indicators of morbidity and mortality and vaccine coverage, whereas Table 2 shows the PMSV of these indicators for the COVID-19 pre-vaccination and vaccination periods. Finally, Table 3 shows the result of the association between indicators of morbidity and mortality and COVID-19 vaccination coverage with a complete scheme (2nd dose or single dose).
The data show that the progression of vaccination coverage in individuals aged 60 years and over had three in ection points, and four trend segments for the rst dose have been estimated: 1st from EW 3/2021 to EW 6/2021 with weekly increase (PSV: 112.2%), 2nd from EW 6/2021 to EW 13/2021 with new weekly increase (PSV: 37.8%), 3rd from EW 13/2021 to EW 18/2021 with the lowest weekly increase (PSV: 11.3%) and 4th from EW 18/2021 to EW 28/2021 with the greatest reduction in the weekly increase (PSV: 0.3%). As for the vaccination coverage of the second or single doses, there were also three in ection points, and four trend segments were estimated: 1st from EW 3/2021 to EW 5/2021 with weekly increase (PSV: 1st from EW 3/2021 to EW 5/2021 with weekly increase (PSV: 72.0%), 2nd from EW 5/2021 to EW 8/2021 with the highest weekly increase (PSV: 234.0%), 3rd from EW 8/2021 to EW 17/2021 with the lowest weekly increase (PSV: 42.4%) and 4th from EW 17/2021 to EW 28/2021 with the greatest decrease in the weekly increase (PSV: 4.0%). In the rst six months of COVID-19 vaccination, there was a weekly increase trend in vaccination coverage, in individuals aged 60 years and over, for both the rst dose (PMSV: 22.5%) and for second or single dose (PMSV: 39.5%).
On the other hand, the proportion of individuals aged 60 years or over who manifested a more severe clinical condition showed three in ection points with four time trend segments: 1st from EW 8/2020 to 10/2020 with weekly increase (PSV: 19.8%), 2nd from EW 10/2020 to EW 17/2020 with weekly reduction (PSV: -7.0%), 3rd from EW 17/2020 to EW 34/2020 with new weekly increase (PSV: 1.0%), and 4th from EW 34/2020 to EW 28/2021 with a new weekly reduction (PSV: -0.3%). Severity showed a decrease trend both in the pre-vaccination period (PMSV: -0.1% and in the vaccination period (PMSV: -0.3%). Furthermore, COVID-19 vaccination coverage was not statistically associated with the proportion of severity in individuals aged 60 years and over (p-value: 0.762).
COVID-19 vaccination coverage was inversely associated with the weekly incidence of hospitalizations for COVID-19 in individuals aged 60 years and over (IRR: 0.98).   Progression of COVID-19 and vaccination in individuals aged 18 to 59 years Figure 2 shows the distribution of observed and estimated weekly trends of epidemiological indicators of morbidity and mortality and vaccination coverage, for the rst and second doses, for COVID-19 in individuals aged 18 to 59 years. Table 4 shows the PSV that was calculated for each change in the time trend of indicators of morbidity and mortality and vaccine coverage, whereas Table 5 shows the PMSV of these indicators for the COVID-19 pre-vaccination and vaccination periods. Finally, Table 6  On the other hand, lethality in individuals hospitalized from COVID-19 who were between 18 and 59 years old had ve in ection points and six time trend segments: 1st from EW 8/2020 to EW 10/2020 with weekly decrease (PSV-28.0%), 2nd from EW 10/2020 to EW 15/2020 with weekly increase (PSV: 11.3%), 3rd from EW 15/2020 to EW 44/2020 with new weekly reduction (PSV: -1.7%), 4th from EW 44/2020 to EW 12/2021 with new weekly increase (PSV: 3.6%), 5th from EW 12/2021 to EW 24/2021 again with weekly reduction (PSV: -2.4%), and 6th) from EW 24/2021 to 28/2021 with the greatest weekly reduction (PSV: -19.4%). Lethality in this population showed a decrease trend in the pre-vaccination period (PMSV: -0.5%) and towards a greater reduction in the vaccination period (PMSV: -3.3%). COVID-19 vaccination coverage was inversely associated with lethality in individuals hospitalized for the disease who were between 18 and 59 years of age (IRR: 0.94).

Discussion
The present study described and analyzed the progression of COVID-19 vaccine coverage in relation to indicators of morbidity and mortality trends from the disease, both in individuals aged 18 to 59 years and in individuals aged 60 years and over, in the rst six months of vaccination, in Brazil.
As far as is known, this was the rst study published that analyzed the indicators of morbidity and mortality time trends from COVID-19, comparing the pre-vaccination and the vaccination periods, in Brazil. In this sense, this study adds up-to-date ndings of interest not only to the national community, but also to the international community, as the control of the pandemic will depend on the efforts that each country has made/is making to face COVID-19, especially countries large and populous like Brazil, which have an intense migratory movement of individuals and strong international trade and economic relations.
The methods of this study have also been used to assess the indirect impact of vaccination on morbidity and mortality trends of other vaccine-preventable diseases, such as pneumococcal disease and chickenpox, both of which are infectious and transmitted mainly from person to person by respiratory route, as well as COVID-19. (14,15) The main ndings of this study show that there were important changes in indicators of morbidity and mortality trends from COVID-19 over time in both populations evaluated, especially after the start of vaccination against the disease, with the exception of the proportion of hospitalized severe cases that remained. practically stable in the period evaluated. It was also shown that the indirect impact of vaccination was more intense in reducing indicators of morbidity and mortality trends for individuals aged 60 years and over compared to those aged 18 to 59 years, and for the latter group, vaccination coverage was lower in the period evaluated, as COVID-19 vaccination for this group started in immunocompromised individuals and practically expanded in EW 20 to other population groups.
The interpretation of these results must take into account, at least, the limitation imposed by the use of administrative data, which may have underestimated the indicators analyzed, especially vaccine coverage, which depends on timely records of cases and vaccinated individuals in their respective information systems.
It must also be taken into account that the observed statistical association does not necessarily re ect a causal association between the increase in vaccination and the decrease in indicators of morbidity and mortality from COVID-19, as the reduction in cases and deaths that occurred during the vaccination period may have been an event that coincided with the end of an epidemic peak, such as that which occurred in 2020 when there were no COVID-19 vaccines yet. However, it is noteworthy that during vaccination in 2021, the reduction of cases and deaths from COVID-19 was more accelerated and intense than that which occurred in 2020, especially in the population aged 60 years and over, which was prioritized for vaccination and also achieved greater vaccine coverage compared to individuals aged 18 to 59 years. In addition, the reduction in hospitalizations in the older-age group had 54% of its occurrence explained solely by the increase in vaccination coverage (Table 3).
Another important consideration is the possible heterogeneity of vaccine coverage in subgroups within the population studied, which may have contributed to a reduction in morbidity and mortality, even with the apparent low vaccine coverage. For example, within the group of individuals aged 60 and over, there were individuals who were 80 years old or over, among others. According to P.C. Victoria et al (2021), four months after the start of COVID-19 vaccination, approximately 95% of Brazilians aged 80 years or older had already received at least the rst dose of the vaccine.(16) It is important to remember that the risk of morbidity and mortality from COVID-19 is directly proportional to increasing age. (17) Therefore, by decreasing the number of susceptible individuals in the age group that also has the highest risk of morbidity and mortality from the disease, those authors observed a reduction in the percentage of deaths from this underlying cause in individuals aged 80 years or more, which went from 25% -30% at the start of vaccination to 13% after four months of COVID-19 vaccination in Brazil. (16) On the other hand, the severity trend in hospitalized cases was practically constant throughout the period evaluated, with a slight weekly reduction observed in the vaccination period, especially among individuals aged 60 years and over. Therefore, there was no statistically signi cant association between COVID-19 vaccination and the severity trend in both groups analyzed. The study population, consisting of patients hospitalized for COVID-19, may have in uenced this result, as hospitalized individuals are more likely to progress to a severe clinical condition due to the disease.(18) In addition, there is a possible confounding factor that was not controlled in this study and may have in uenced the result of this study; it is the presence of comorbidities, which in addition to increasing the risk for more severe clinical conditions of COVID-19, it was also adopted as an inclusion criterion for vaccination of priority groups, in uencing vaccine coverage in different age groups, according to the National Plan of Vaccination.
However, despite the severity trend having been practically constant, the lethality trend in cases hospitalized for COVID-19 reduced considerably, during the period of vaccination, in both groups analyzed. It can be argued that, over time, with greater accumulated knowledge on the clinical management of cases, the reduction in lethality may have occurred due to an improvement in the quality of care, or even due to a reduction in the number of hospitalized patients, consequently alleviating the burden on health care services and workers, ensuring greater attention and intensive care to hospitalized patients, and therefore avoiding the worsening of cases and the consequent progression to death. (19) But as shown above, the proportion of severity in the cases was constant over time, therefore refuting these hypotheses and reinforcing the impact of vaccination in reducing mortality from COVID-19. Despite the increase in vaccination coverage, in both groups analyzed, did little to explain the decrease in lethality (6.3% for individuals aged 60 years and over, and 6.5% for individuals aged 18 to 59 years), each increase in vaccination coverage reduced by 0.7% and by 6.1%, respectively, the weekly lethality from COVID-19 in the two populations studied.
The increase in mortality associated with the increase in vaccination coverage for individuals aged 18 to 59 years, despite the statistically signi cant association, re ected much more the in uence of outlier values that in uenced the average and increased trend variability in the period analyzed, characterizing more a type I error, than a time association that weighs against vaccination. Furthermore, with the advance of vaccination in the elderly group and the reduction in morbidity and mortality from COVID-19 in this age group, the perception of risk has changed for younger individuals (20). With the easing of containment measures, they also began to be more frequently exposed, resulting in a change in the morbidity and mortality pro le of the disease, which began to affect this age group with greater intensity, as vaccine coverage among individuals aged 18 to 59 years was still incipient.
The COVID-19 vaccination trend showed a great difference between vaccination coverage for the rst, second or single dose, in both groups evaluated. Initially, due to the worldwide unavailability of doses and the need to accelerate the national COVID-19 vaccination campaign, the application of the rst dose to the largest possible number of individuals was prioritized for the most vulnerable groups, as the rst dose was relatively effective, and the time until the application of the second dose could be long enough for the countries to receive more doses, and then complete the vaccine schedules with the second dose. (21) However, the low vaccination coverage for the second dose in Brazil may also have occurred due to a false sense of security that was perceived by the population as a result of the decrease in morbidity and mortality that occurred during this period, therefore decreasing the adherence to vaccination to complete the vaccination schedule, (22) in addition to the possibility of underreporting of vaccine doses in the aforementioned information systems.
Vaccine hesitancy, which is the act of refusal or delay in vaccination, may lead to the accumulation of susceptible individuals, whether as unvaccinated individuals, incomplete vaccination schedules, or natural loss of acquired immunity, which allied to the maintenance of viral circulation, including VOCs of greater transmissibility, such as the Delta variant, may cause new epidemic waves that, with the increase in morbidity and mortality from COVID-19, could jeopardize the population's trust in vaccines and in vaccination as a whole. (23,24) In the study conducted by Yi-Tui Chen (2021), the effect of COVID-19 vaccination was evaluated in eight countries that had reached, by May 2nd, 2021, at least 50% of their population vaccinated (Israel, United Arab Emirates, Chile, United Kingdom, United States of America [USA], Hungary, Qatar, and Serbia), in addition, he estimated the vaccination coverage needed to reach an in exion point in the disease trend. In six of these countries (except the USA and the UK), a peak of cases was also observed after the start of vaccination and before obtaining mass immunity. When the reduction in disease cases began, vaccination coverage in these countries ranged from 1.46 to 50.91%, and the R2 of the models ranged from 57.2% to 89.9%, i.e., the models also had a high explanatory power. For the USA and UK, Pearson coe cients were calculated, which showed a reduction in COVID-19 rates from the start of vaccination. (25) However, in many of these countries, including Israel, the United Kingdom and the USA, there was an increase in COVID-19 cases even after reaching high vaccination coverage. (10,26,27) Several explanations have been put forward to explain this phenomenon, including the early relaxation of non-pharmacological measures (such as social distancing and the use of protective masks), the role of VOCs in greater transmissibility, the effectiveness of the vaccines used and the selection of priority groups for vaccination.

Conclusions
Page 17/22 COVID-19 vaccination in Brazil, in the rst six months of the national campaign, had an impact on reducing morbidity and mortality from the disease in both adult and older-age age groups, even in the context of low vaccination coverage in a tropical country with a great diversity in terms of population, logistics, and recording information in the vaccination system.
In this regard, every effort must be made to increase vaccination coverage and establish a communication channel with the population, in a clear and direct way, to inform about the risks of not getting vaccinated or not completing the recommended vaccination schedule. Furthermore, it is important to design a strategy to mitigate vaccine hesitancy, including monitoring performance indicators, identifying partners who are working on this issue, and implementing strategies and activities based on successful experiences. The information systems must also be fed with reliable data that will re ect the vaccination scenario as close to reality as possible. Only with quality information will the impact of vaccination be ensured to accelerate, with greater security, the end of the public health emergency brought on by the COVID-19 outbreak.

Declarations
Ethical approval and consent to participate Secondary publicly accessible data were used and therefore the need for ethical approval was waived.

Consent to publication
Not applicable Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request. Furthermore, the original databases are publicly accessible and can be found in the links mentioned in the methods.

Con ict of interest
The authors declare that they have no con icting interests.

Financing
No nancing.
Authors' contributions JP carried out the data collection, processing and analysis, bibliographical research, and drafting of the article. CMC, FF, LG contributed to the interpretation of the data. WNA contributed to the study conception and design. All authors read and approved the nal manuscript.

Figure 2
Trends in the incidence of hospitalizations, severity, lethality and mortality from COVID-19 compared to vaccination coverage in individuals aged 18 to 59 years, Brazil, epidemiological week 8/2020-28/2021