Artigos

Introduction
The presence of risk factors prior to infection, such as advanced age, chronic metabolic disease (mainly diabetes) 1, and being sedentary 2, increases the risk of stronger COVID-19 symptoms due to the inefficiency of the immune system to fight the virus in this population. It is well known that lifestyle, such as poor sleep quality 3, sedentary behavior or overtraining 4, diet (e.g., energy deficit or low-fiber diet) 5, and alcohol consumption 6 negatively affect the effectiveness of the immune system to fight infections.
Vaccination is the best way to contain the pandemic and reduce the risk of severe cases when people are infected. In Brazil, several vaccines (e.g., CoronaVac from Butantã; BNT162b2 mRNA from Pfizer/BioNTech; ChAdOx1 and Ad26.COV2.S from Oxford/AstraZeneca and Janssen, respectively) with different manufacturing technologies were available to immunize the population against COVID-19. Janssen vaccine was based on viral vector technology, CoronaVac was based on the inactivated virus, while Pfizer and AstraZeneca were based on mRNA and DNA technology, respectively. All vaccines have been reported to elicit immune response and coffer efficacy against COVID-19 7-9. However, several types of adverse effects occur 10-12. Some important questions from clinical practice remain open. For instance, what vaccine has less adverse effects, and what specific adverse effects are more pronounced in one type of vaccine than others? To respond this question a machine learning approach to detect the adverse effects variation after the COVID-19 immunization can be easily implemented. The machine learning approach has already been used to automatically detect the best pathway for decision-making in COVID-19 public policies 13-15.
Despite the safety and the incalculable benefits promoted by vaccines, there is a usual perception of possible adverse effects, including nausea, vomiting, fever, general myalgia, fatigue, headache, and pain at the injection site for a few days after vaccination 16, 17. The magnitude of the adverse effects of COVID-19 vaccine in some populations was close to the effects of SARS-COV-2 infection itself. For instance, it is estimated that 93% of patients with COVID-19 infection are asymptomatic or have mild infection symptoms 18, while 48 to 90% report adverse effects after the COVID-19 vaccination 10-12. In this sense, knowing related factors that can mitigate (and explain) the adverse effects of the COVID-19 vaccine may be advantageous for public policy decision-making to increase the population's engagement in vaccination campaigns.
The literature is limited to describes the adverse effects experienced due to vaccination against COVID-19. Is know that type of vaccine (mRNA had more adverse effects (>) than DNA or viral vectors technology), age (older individual > young counterparts), sex (women > men), and previous infection before vaccination 11. Thus, is not know why age or sex-related factors is associated with adverse effect. For this, in this study, we also used a machine learning approach to explain the adverse effects variation after the COVID-19 immunization. No studies have verified the relationship between the vaccine's adverse effects and related lifestyle factors. Our hypothesis is that the adverse effects of vaccines are related to lifestyle habits and the health profile of the individuals. Therefore, the aim of this study was to investigate the relationship between lifestyle and health profile-related factors to the reported adverse effects of COVID-19 vaccines.
Method
Overall design
To achieve the aims of this study we conduct a descriptive, exploratory, explanatory, and cross-sectional study carried out with three hundred and one (301) volunteers which were vaccinated for the 'Coronavirus disease 2019' (COVID-19). As a previous study identified the association between age and COVID-19 vaccine adverse effects 11, we used G*Power (v.3.1.9.7) to calculate the sample size required to identify such an association. Thus, a sample size of 291 individuals vaccinated with COVID-19 vaccines is necessary to obtain the following statistical power in the Chi-square test: effect size= 0.3, alpha= 0.01, power= 0.90, and Df= 9. To determine the sample size from the Chi-square test, G*power requires the degrees of freedom (Df); thus, we determined that it would be 9, assuming that the hypothesis test can be done by quartiles of age and adverse effects. Thus, in a 4x4 contingency table, we have Df = (4-1) (4-1) = 9. This study was approved by the ethical committee of the Federal University of São Paulo (nº 0686P/2021).
Due to the current need for social distancing in Brazil, data was collected exclusively in the virtual environment through an online form management platform (Google Forms®). Google Forms® questionnaires were accessible through any device with access to the internet. To reach participants during the COVID-19 lockdown, we used the "snowball sampling" method to invite individuals through social media (Instagram, Facebook, and Twitter) and e-mails to participate in this research. Data collection occurred between 06/2021 to 09/2021. Currently, the application of this method is favored by social networks, is considered helpful for exploratory purposes, and is suitable for studying populations that are difficult to reach (e.g., during the lockdown condition imposed by COVID-19 pandemic) or that the number of subjects of the vaccinated population is unknown 19. Specifically, the researchers promoted the survey on social media and e-mail. The respondents and researchers encouraged others to share the survey with their networks (social media and e-mail) until the required sample size (291 individuals vaccinated with the COVID-19 vaccine) was reached.
The questionnaire formulated (provide as supplementary material) by the researchers included a set of personal questions, such as gender, age, weight, height, chronic diseases, chronic medication use (and the name of the drug), and information related to the vaccination period (at morning or afternoon), sleep quality [Pittsburg Sleep Quality Index (PSQI)], physical activity, and diet. We also collected the type of vaccine received and if they had COVID-19 (all data is described in Table 1). Regarding adverse effects, we asked the volunteers if they felt any post-vaccination symptoms. If so, which symptoms? We report the adverse effects as the number of symptoms (NS) throughout the results and discussion session. The researchers estimated the average response time for the entire questionnaire at 7 to 8 minutes.
Individuals over 18 years old, of both sexes and vaccinated against COVID-19 with at least one dose of the CoronaVac/Butantã, Pfizer/ BioNTech, or AstraZeneca/Oxford (available in Brazil at the time of this study), were included in the study. Participants who were not Brazilian residents and under 18 years old were excluded from the sample.

Level of physical activity and sitting time
International Physical Activity Questionnaire – IPAQ: The International Physical Activity Questionnaire (IPAQ) 20 was used to assess the level of physical activity. 1). This is an instrument developed to estimate the usual level of physical activity in populations from different countries and sociocultural contexts. The IPAQ used was the Portuguese short version. This version consists of eight open questions, and its information allows estimating the time spent per week in different dimensions of physical activity (walking and physical efforts of moderate and vigorous intensities) and physical inactivity (sitting time). Based on the IPAQ responses, we classified the level of physical activity as low (L), moderate (M), and high (L) 21. We also calculated the metabolic equivalent for tasks (METs), and daily sitting time for the weekdays and for the weekends. The questions are attached as supplementary material.

Sleep variables
Pittsburgh Sleep Quality Index (PSQI): To assess sleep duration and quality, PSQI-BR was used, an instrument translated and validated for Portuguese 21. The questionnaire consists of 19 self-reported questions that measure, in the previous month, 7 components, namely: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disorders, use of sleeping medication, and daytime dysfunction due to drowsiness. The component scores are added, resulting in the global score (PSQI-score, ranging from 0 to 21 points; being a higher global score, the worse sleep quality). From PSQI, we also assess sleep efficiency (in percentage) and sleep quality (bad or good) as described previously 22. We also asked if the volunteers had sleep disturbance one day before, on the day, and one day after vaccination. The questions are attached as supplementary material.

Other lifestyle variables
We assessed alcohol drinking frequency and if they drank alcohol the day before, on the day, and one day after vaccination, if they followed a restrictive diet, their daily water intake, and if they believed that COVID-19 vaccine could cause adverse effects. The questions are attached as supplementary material.

Statistical analysis
Data are presented as mean (or frequency) and standard deviation (and minimum maximum, which will be indicated). The comparison between vaccines types according to sampling characteristics were verified with one-way ANOVA (or Kruskal-Wallis or Chi-Square test). The Bonferroni test was used as a post-hoc for the ANOVA test, and the Dunnett test was used as a post-hoc for the Kruskal-Wallis’s test. An exploratory decision tree regression (DTR) model was performed (using the CHAID method spontaneous growth) to identify the influential variables which could determine the number of symptoms (adverse effect) from COVID-19 vaccines. The parents’ nodes (leaves) of the DTR were pruned to compare at least 100 samples in each node, and child nodes 50 samples. We chose 50 samples in child node because lower values turn the DTR redundant, and higher values (>50 samples) do not generate sufficient nodes to draw conclusions. The contextual variables, which were identified as significantly correlated with the number of symptoms, were used as covariables in the DTR models. We validate all DTR test with K-fold cross-validation (10-fold). Significance level was set to P 0.05. Type 2 error risk probability is reported as ?. SPSS v.26 was used to carry out all analyses.
Results
Table 1 describes the characteristics of the three hundred and one (301) volunteers who responded to the survey. Three participants were excluded from this study (only three who received the Janssen-Cilag/Johnson vaccine). The age varies of participants from young adults to seniors, and a wide distribution in weight can also be observed. Also, a wide distribution occurs in BMI, hours of sitting time, physical activity level, sleep quality, water, and alcohol drinking. A wide variety of these variables might be of interest to be investigated regarding their relationship with the NS (or adverse effects) of the COVID-19 vaccines.

INSERT TABLE 1 NEAR HERE

In table 2, a comparison of the volunteer’s characteristics was carried out using vaccine type as a discrimination variable. Thus, we identified that AstraZeneca had more NS reported when compared to CoronaVac and Pfizer. CoronaVac receivers reported fewer NS than AstraZeneca and Pfizer (Table 2). Also, in Table 1S, we identified the most frequent symptoms reported by the volunteers. Then, in Table 2S, we identified the specific symptoms associated with the vaccines. In summary, in Tables 1S and 2S, we can identify that CoronaVac was associated with fewer specific symptoms. AstraZeneca was associated with a higher frequency of symptoms of general myalgia, headache, fatigue, chills, drowsiness, and joint pain when compared to CoronaVac. Also, AstraZeneca presented more symptoms of headache, chills, drowsiness, joint pain, nausea, or vomiting than Pfizer.

INSERT TABLE 2 NEAR HERE

Decision tree regression

Due to a large variation in the participant's variables such as age, BMI, hours of sitting time, physical activity level, sleep quality, water, alcohol drinking, and sleep variables, we carried out a decision tree regression to achieve the aims of the study (investigate whether the variables related to lifestyle or healthy profile are associated to the NS reported by the participants).
Thus, in Figure 1A, the dependent variable used in the decision tree regression was the “number of symptoms”, and all variables described in Tables 1 and 2 were used as independent variables. In Figure 1A, we see that the NS (which ranged from zero to twelve) was related to the type of vaccine (also confirmed in Table 2) and age. Our DTR model indicates that vaccine and age can explain 35% of NS variation. The decision tree created three nodes for each type of vaccine (CoronaVac, Pfizer, and AstraZeneca). Node 3 (AstraZeneca) branches into two other nodules identifying different amounts of NS with advancing age, demonstrating that individuals with advancing age (>32 years) have less adverse effect when compared to young individuals (?32 years). As the vaccine type behaves as an influencing variable in the regression tree, we use it as a covariate in Figure 1B. In Figure 1B, “belief in adverse effects” was related to the different adverse events reported by the volunteers. For those who believe in adverse events, the decision tree grew, indicating that age and sleep quality influences the reported adverse effects.

INSERT FIGURE 1 NEAR HERE

As we cannot identify whether the “believe in the adverse effects” have a nocebo effect, or if it is a direct effect of the adverse effects suffered by the volunteers, we removed this variable from the independent variables and performed a new decision tree regression (Figure 2). Therefore, when using the vaccine type as a covariate in the DTR model (Figure 2A), we observed that age is related to different NS reports (Nodes 1, 2, and 3). Node 3 branches into the other two nodes (4 and 5), identifying that good sleep quality decrease the NS in individuals over 32 years of age. However, age and sleep quality can explain only 7% of NS variation. To identify which age-related factors could explain the NS, we sought to identify which variables (lifestyle or health profile, described in Table 2) are age-related in our volunteers (Figure 2B). Thus, the DTR identified that disease incidence (nodules 1 and 2; Figure 2B) is related to age. Therefore, individuals in the mean age group of 33.2 years (nodule 1) were associated with not having disease, while individuals in the mean age of 43.9 years (nodule 2) were associated with diseases. Next, in Figure 2C, we sought to identify factors that may be related to diseases, and the variable medication (taking or not) was significantly associated (nodules 1 and 2, respectively). Thus, the main factor related to NS (beyond vaccine type) is age, and age is significantly associated with disease and medication use.

INSERT FIGURE 2 NEAR HERE

Besides vaccine type, because we identified that age is the most important variable to explain the NS, we then used age as a co-variable in Figure 3A. Thus, we identified that a change in sleep efficiency (SE) is related to the NS (nodes 1, 2, and 3). Node 2 branches in nodules 4 and 5, suggesting that medication use could decrease the NS. However, sleep efficiency and medication can explain only 3% of NS variation.
In node 3 (Figure 3A), we identified that high sleep efficiency was associated with NS reported (nodes 1 to 3) and medication (in nodule 2) seems to interact with sleep efficiency. Then, in Figure 3B, we sought to identify which factors could be associated with sleep efficiency. Interestingly, the DTR identified that using or not using medication (node 1 and 2, respectively) was associated with sleep efficiency, indicating that individuals with higher sleep efficiency did not use medication.

INSERT FIGURE 3 NEAR HERE

Discussion

In this study, we used the decision tree regression (DTR) method to identify the most important variables to explain the NS effects from the COVID-19 vaccine in a sample of Brazilians. The approach of machine learning in public health has been effective in discovering relationship between variables that are not detected by the orthodox statistic that uses linear regressions or comparison between group that are arbitrarily created (for example, statistic presented in Table 2). Also, the machine learning approach automatically detects the best pathway to be taken in public policies 13, 15, a fact that it would be difficult to be reached using orthodox statistics. Here, using CHAID spontaneous growth model, we use sleep variables, several other variables related to lifestyle and factors related to the health profile (described in Table 1) as independent variables to detect if has association with NS of COVID-19 vaccines. Thus, if the variables that have significant association (even in a subset data from the variable) the DTR could detect their potential to explain the dependent variable. According to our DTR model, there is a negative association (Chi-square: P= 0.001; ?= 0.9) between age and the amount of reported adverse effects. Such negative association between age and NS is probably due to the presence of diseases and the chronic use of medications. The literature suggests that older individuals are less likely to report adverse effects due to COVID-19 vaccines 11. It was speculated that fewer adverse events might be because older people produce less INF-I than younger individuals 23. As our data analysis also identified a negative association between age and NS due to COVID-19 vaccines (see Figure 2A). Therefore, we sought to explore whether health and lifestyle factors were associated with the age of our sample. In this sense, the DRT model (see Figure 2B) suggested that older individuals were associated with a greater likelihood of having a disease. Then, having a disease was associated with chronic medication use (see Figure 2C). The use of these medications seems to mitigate the NS of vaccines (see Figure 3A). This was confirmed, when we compared individuals who take or not medication, with statistical differences in the NS of vaccines (Mann-Whitney test: P=0.042). Also, there is a significant association of age and medication users (Chi-square: <33 vs. >32 age years, P= 0.01). Therefore, older individuals reporting fewer adverse effects might be related to medication use.
It is well known that sleep quality significantly affects the effectiveness of the immune system 3, therefore, our hypothesis was that the NS might be related with participants' sleep profile. In Figure 3A, sleep efficiency was related to the NS (lower sleep efficiency was associated with a higher NS of COVID-19 vaccines). Also, nodes 4 and 5 from Figure 3A indicates that this relationship is also mediated by the use of medication (see also Figure 3B). It has been previously identified that the medications used to manage chronic medical conditions contribute significantly to sleep worsens 24, which could induce a low NS (from COVID-19 vaccines) in individuals with low sleep efficiency. In our sample, the use of medication is associated with low sleep efficiency. Also, individuals with poor sleep efficiency (for instance, ?84.21% versus ?85.68%) in our sample have a significantly higher NS due do COVID-19 vaccines. It is also suggested by the literature that individuals with poor sleep quality (insomnia/sleep deprivation) are at greater risk of developing medical disorders (mainly cardiovascular diseases) 25. This might explain the association identified here between sleep efficiency and medical use (and disease and medical use). Therefore, our data suggest that sleep efficiency may play a significant role on the NS experienced due to COVID-19 vaccines, but medication used also plays a significant role on sleep efficiency. Studies have suggested a relationship between sleep efficiency and the immune system response to vaccines 26, 27. For instance, it has been identified that short sleep duration negatively affects the antibody response to novel antigens 27. Also, poor sleep efficiency and short sleep duration were associated with a higher likelihood of developing a cold 28. Thus, future studies should investigate the role of poor sleep efficiency on adaptive immune system from COVID-119 vaccines. In summary, our results suggest that a low sleep efficiency can increase the adverse effects of the COVID-19 vaccine. On the other hand, individuals with low sleep efficiency who use chronic drugs can mask these adverse effects.
The class of drugs used by the volunteers is shown in Table S3. Several types of drugs (and their combination) were reported by the volunteers of this work, which makes it challenging to identify the possible origin (which drug) could mitigate the adverse effects of COVID-19 vaccines. Also, it isn't easy to establish whether there is a causal relationship between taking medication and sleep efficiency (Figure 3A and 3B). We suggest that randomized controlled studies (using individuals who take or not medication with high and low sleep efficiency) examining the NS events from COVID-19 vaccines may help to clarify the role of sleep efficiency and drug class on the adverse effects of vaccines.
Also, we identified that the three different types of vaccines (CoronaVac, Pfizer, and AstraZeneca) differ from each other in the amounts and specificity of reported adverse effects. The literature has reported incidences of adverse effects from the first dose of CoronaVac, Pfizer, and AstraZeneca vaccines of 48.1% 10, 65% to 82% 10, 11, 65% to 84% 10, 12, respectively. Our data indicate an incidence of 48.6%, 85.5%, and 89.9% for CoronaVac, Pfizer, and AstraZeneca, respectively. Furthermore, the NS in our sample population is statistically different between the three vaccines (see Table 2). Therefore, these results indicate that in fact, AstraZeneca and Pfizer vaccines produce a greater NS than CoronaVac.
It is important to mention, that the most common adverse effect reported by our volunteers was pain at the application site, general myalgia, headache, fever, fatigue, chills, sore throat, and drowsiness, among other symptoms of common colds (Table 1S and 2S). It is also important to note that believing in adverse effects was associated with a high NS in our study, indicating here that the high adverse effect caused by vaccines may have led the volunteers in our study (~80%) to believe the adverse effects of vaccines. In fact, from our data, we cannot identify if this result is a nocebo effect or a direct effect of the NS of vaccines. However, the negative expectations from adverse effects (a nocebo effect) 29 and the NS caused by previous vaccines 30 reduce the adherence to vaccination campaigns. Therefore, future research (and public policies) should adopt strategies to reduce the COVID-19 vaccine’s adverse effects. Our studies suggest that the use of medications (probably with analgesic components) and high-quality sleep can mitigate the adverse effects of vaccines.
It has been suggested that the magnitude of these symptoms are related (as a by-product) to an exuberant type I interferon (IFN-I) response 23. The mRNA vaccines are known to be powerful inducers of IFN-I 31, and it is well known that COVID-19 vaccines have different development techniques; for example, CoronaVac is based on the inactivated virus, while Pfizer and AstraZeneca are based on mRNA and DNA technology, respectively. The mRNA vaccine presents higher values of NS when compared to the vaccine with DNA or viral vector technologies 11; these data are in accord with the adverse effects identified in our volunteer sample (see Table 2S). Thus, the symptoms of cold (e.g. fever, chills, headache, muscle pain and nausea) experienced during COVID-19 vaccination (a foreign body infusion) are caused by the release of several amounts of cytokine from neutrophils and macrophages when the vaccine molecule is detected 32. The IFN-I production is required for optimal immune response, thus suggesting that the magnitude of vaccine adverse effects reported here might be related to the IFN-I response 23.The literature has reported efficacy against COVID-19 symptomatic infection in the Brazilian population with different percentages of efficacy [for example, CoronaVac, Pfizer, and AstraZeneca vaccines has been 51% 7, 87.7% 8, and 64.2% 9, respectively]. However, studies verifying whether, in fact, the magnitude of symptoms (like the ones felt on common colds) after COVID-19 vaccination is actually associated with the IFN-I response remains an open question. Also, no studies are identifying an association between cytokines or NS and vaccine efficacy.
This study has some important limitations. The data are online self-reported habits of sleep quality, physical activity level, and dietary habits, which might increase report errors and the statistical type 1 error 33. Also, our sample is of convenience, which violates the assumption of statistical analysis tests in which data are from a random sample. Another potential limitation is the “snowball sampling” methodology with online data collection adopted in this study, which imposes several limitations in this study, namely: (1) this methodology may not capture the representativeness of the vaccinated COVID-19 population and might limit itself to a social “bubble”; (2) community bias: initial participants disproportionately influence the sample, leading to potential misrepresentation of the target population; (3) non-randomness: snowball sampling violates principles of randomness and representativeness, which is important for generalizable research findings; (4) the online data collection can increase the anchoring effect, which might impact the data reliability and accuracy. Therefore, future studies that carry out direct prospective design with random sampling are necessary to confirm our findings. To mitigate such limitations and to decrease type 1 and 2 statistical errors to identify the association between age and NS due to COVID-19, we collect sufficient data, i.e., 301 individuals who have previously been vaccinated against COVID-19. Also, our machine learning analysis used a K-fold cross-validation test (10-fold). Therefore, considering that our sample was a convenience sample, we adopted a conservative approach during statistical analysis, which provided greater confidence in our data report.
We chose snowball sampling methodology due to the conditions imposed by the COVID-19 pandemia (mainly due to isolation actions, such as quarantine). This methodology provides several strengths that help achieve the aims of this study 19. This methodology allows access to a specific population (i.e., vaccinated patients that would be difficult to locate using traditional sampling methods due to their geographic dispersion, anonymity, or hesitation to expose themselves); efficiency in sample formation (once initial participants identify other potential participants, recruitment becomes more agile); participant trust and adherence (the nominations are made by people known, there is greater trust and acceptance to participate in the research, resulting in greater engagement and reduced refusal rates); reduced cost (compared to methods that require extensive sampling lists or mass recruitment, the snowball technique tends to be less costly and more practical, as the referral process minimizes the need for large investments in advertising and screening) 19.
Conclusion
The use of medications associated with diseases usually associated with older individuals mitigates the NS from COVID-19 vaccines. Sleep efficiency also plays a significant role in mitigating adverse effects from COVID-19 vaccines. Specifically, individuals with lower sleep efficiency report higher NS from COVID-19 vaccines, and the use of medication also appears to mitigate it. Future research should investigate the impact of the interaction between chronic medication use and the efficacy of COVID-19 vaccines. 80% of our sample believes in the adverse effects of the vaccines. In this sense, future research and public policy should investigate and pay attention, respectively, to the impact of this variable on adherence to subsequent doses of the COVID-19 vaccines. Here, we demonstrate that the DTR is a valuable model for investigating the NS related to COVID-19 vaccines and their potential to assist decision-making in public policies. For example, the use of the machine learning technique is a formidable tool to investigate issues that might related to large numbers of factors. Thus, with this technique, it is possible to discriminate important factors to explain the problem and discard factors that do not explain anything.

Funding
The author(s) declare financial support was received for the research, authorship, and/or publication of this article. This work is supported by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP—Grant number: 2021/03601-1).

Data Availability Statement
The databases used in the article, including the extraction codes, analyses, and results, are available in the repository: de França , Elias; Marques, Camila G.; Leite, Geovana S. Fogaça; Magalhães, Ana Carolina O.; Caperuto, Erico C.; Thomatieli-Santos, Ronaldo Vagner, 2025, "Data for: A practical machine learning approach to identifying adverse effects due to COVID-19 vaccines", https://doi.org/10.48331/SCIELODATA.PIDUOS, SciELO Data, V1, UNF:6:jVnNNYszr4XeDipk3Wou6A== [fileUNF]

References
1. Rod J, Oviedo-Trespalacios O, Cortes-Ramirez J. A brief-review of the risk factors for covid-19 severity. Rev saude publica. 2020; 54.
2. Sallis R, Young DR, Tartof SY, Sallis JF, Sall J, Li Q, et al. Physical inactivity is associated with a higher risk for severe COVID-19 outcomes: a study in 48 440 adult patients. Br J Sports Med. 2021; 55:1099-105.
3. Moldofsky H. Sleep and the immune system. Int J Immunopharmacol. 1995; 17:649-54.
4. Walsh NP, Gleeson M, Shephard RJ, Gleeson M, Woods JA, Bishop NC, et al. Position statement. Part one: Immune function and exercise. Exerc Immunol Rev. 2011; 17:6-63.
5. Bermon S, Castell LM, Calder PC, Bishop NC, Blomstrand E, Mooren FC, et al. Consensus Statement Immunonutrition and Exercise. Exerc Immunol Rev. 2017; 23:8-50.
6. Díaz LE, Montero A, González-Gross M, Vallejo AI, Romeo J, Marcos A. Influence of alcohol consumption on immunological status: a review. Eur J Clin Nutr. 2002; 56:S50-S3.
7. Jin L, Li Z, Zhang X, Li J, Zhu F. CoronaVac: A review of efficacy, safety, and immunogenicity of the inactivated vaccine against SARS-CoV-2. Hum Vacc Immunother. 2022:2096970.
8. Polack FP, Thomas SJ, Kitchin N, Absalon J, Gurtman A, Lockhart S, et al. Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine. N Engl J Med. 2020; 383:2603-15.
9. Voysey M, Clemens SAC, Madhi SA, Weckx LY, Folegatti PM, Aley PK, et al. Safety and efficacy of the ChAdOx1 nCoV-19 vaccine (AZD1222) against SARS-CoV-2: an interim analysis of four randomised controlled trials in Brazil, South Africa, and the UK. Lancet. 2021; 397:99-111.
10. Lai FTT, Leung MTY, Chan EWW, Huang L, Lau LKW, Peng K, et al. Self-reported reactogenicity of CoronaVac (Sinovac) compared with Comirnaty (Pfizer-BioNTech): A prospective cohort study with intensive monitoring. Vaccine. 2022; 40:1390-6.
11. Beatty AL, Peyser ND, Butcher XE, Cocohoba JM, Lin F, Olgin JE, et al. Analysis of COVID-19 Vaccine Type and Adverse Effects Following Vaccination. JAMA Netw Open. 2021; 4:e2140364-e.
12. Chekol Abebe E, Mengie Ayele T, Tilahun Muche Z, Behaile T/Mariam A, Dagnaw Baye N, Mekonnen Agidew M, et al. Evaluation and comparison of post-vaccination adverse effects among Janssen and Oxford-AstraZeneca vaccinated adult individuals in Debre Tabor Town: A cross-sectional survey in Northwest Ethiopia. Hum Vacc Immunother. 2022:2104059.
13. Magazzino C, Mele M, Coccia M. A machine learning algorithm to analyse the effects of vaccination on COVID-19 mortality. Epidemiol Infect. 2022; 150:e168.
14. Magazzino C, Mele M, Schneider N. Assessing a fossil fuels externality with a new neural networks and image optimisation algorithm: the case of atmospheric pollutants as confounders to COVID-19 lethality. Epidemiol Infect. 2021; 150:e1.
15. Mele M, Magazzino C, Schneider N, Strezov V. NO2 levels as a contributing factor to COVID-19 deaths: The first empirical estimate of threshold values. Environ Res. 2021; 194:110663.
16. Meo S, Bukhari I, Akram J, Meo A, Klonoff DC. COVID-19 vaccines: comparison of biological, pharmacological characteristics and adverse effects of Pfizer/BioNTech and Moderna Vaccines. Eur Rev Med Pharmacol Sci. 2021:1663-9.
17. Riad A, Pokorná A, Attia S, Klugarová J, Koš?ík M, Klugar M. Prevalence of COVID-19 vaccine side effects among healthcare workers in the Czech Republic. J Clin Med. 2021; 10:1428.
18. Barnes E, Goodyear, C. S., Willicombe, M., Gaskell, C., Siebert, S., I de Silva, T., ... & McInnes, I. B. . SARS-CoV-2-specific immune responses and clinical outcomes after COVID-19 vaccination in patients with immune-suppressive disease. Nat Med. 2023; 29:1760-74.
19. Vinuto J. A amostragem em bola de neve na pesquisa qualitativa: um debate em aberto. Temáticas. 2014; 22:203-20.
20. Matsudo S, Araújo T, Marsudo V, Andrade D, Andrade E, Braggion G. Questinário internacional de atividade f1sica (IPAQ): estudo de validade e reprodutibilidade no Brasil. Rev bras ativ fís saúde. 2001:05-18.
21. Bertolazi AN, Fagondes SC, Hoff LS, Dartora EG, da Silva Miozzo IC, de Barba MEF, et al. Validation of the Brazilian Portuguese version of the Pittsburgh sleep quality index. Sleep Med. 2011; 12:70-5.
22. Buysse DJ, Reynolds CF, Monk TH, Berman SR, Kupfer DJ. The Pittsburgh sleep quality index: A new instrument for psychiatric practice and research. Psychiatry Res. 1989; 28:193-213.
23. Sprent J, King C. COVID-19 vaccine side effects: The positives about feeling bad. Sci Immunolo. 2021; 6:eabj9256.
24. Kumar S, Wong PS, Hasan SS, Kairuz T. The relationship between sleep quality, inappropriate medication use and frailty among older adults in aged care homes in Malaysia. PLoS One. 2019; 14:e0224122.
25. Krystal AD. Treating the Health, Quality of Life, and Functional Impairments in Insomnia. J Clin Sleep Med. 2007; 03:63-72.
26. Robertson BD, Collen JC. Sleep touches everything. J Clin Sleep Med. 2020; 16:1997-8.
27. Prather AA, Hall M, Fury JM, Ross DC, Muldoon MF, Cohen S, et al. Sleep and Antibody Response to Hepatitis B Vaccination. Sleep. 2012; 35:1063-9.
28. Cohen S, Doyle WJ, Alper CM, Janicki-Deverts D, Turner RB. Sleep Habits and Susceptibility to the Common Cold. Arch Intern Med. 2009; 169:62-7.
29. Colloca L, Miller FG. The nocebo effect and its relevance for clinical practice. Psychosom med. 2011; 73:598.
30. Monami M, Gori D, Guaraldi F, Montalti M, Nreu B, Burioni R, et al. COVID-19 vaccine hesitancy and early adverse events reported in a cohort of 7,881 Italian physicians. Ann Ig. 2021; 10:1-13.
31. Cagigi A, Loré K. Immune Responses Induced by mRNA Vaccination in Mice, Monkeys and Humans. Vaccines-Basel. 2021; 9.
32. Vogel WH. Infusion reactions: diagnosis, assessment, and management. Clin J Oncol Nurs. 2010; 14:E10-21.
33. de Reuver M, Bouwman H. Dealing with self-report bias in mobile Internet acceptance and usage studies. Inf Manag. 2015; 52:287-94.