index.Rmd

---
title: "International symposium"
subtitle: "COVID-19<br/>Longitudinal analysis"
author: "Luis Anunciação, Lucas Barrozo, Anna Portugal, Fabiana Ribeiro, Anja Leist, Landeira-Fernandez"
institute: "PUC-Rio, UFRJ, Universidade de Luxemburgo"
date: "`r format(Sys.time(), '%d %B, %Y %X')`"
output:
  xaringan::moon_reader:
    lib_dir: libs
    css: [default, metropolis, metropolis-fonts]
    nature:
      #ratio: '16:9'
      highlightStyle: github
      highlightLines: true
      countIncrementalSlides: false
      slideNumberFormat: |
        <div class="progress-bar-container">
          <div class="progress-bar" style="width: calc(%current% / %total% * 100%);">
          </div>
        </div>    
---
class: center, middle, inverse

# Background & Motivation


```{r, include=FALSE }
#https://github.com/yihui/xaringan/issues/109
```

```{r global options, include = FALSE}
knitr::opts_chunk$set(echo=FALSE, include = TRUE, warning=FALSE, message=FALSE, fig.retina = 3)
options(knitr.table.format = "html")
```

```{r xaringan-tile-view, echo=FALSE}
xaringanExtra::use_tile_view()
xaringanExtra::use_logo(
  image_url = "http://www.deqm.puc-rio.br/wp-content/uploads/2020/08/logo-puc-pequena-pb.png",
  width = "50px",
  
)
```

<style>
.remark-slide-number {
  position: inherit;
}

.remark-slide-number .progress-bar-container {
  position: absolute;
  bottom: 0;
  height: 4px;
  display: block;
  left: 0;
  right: 0;
}

.remark-slide-number .progress-bar {
  height: 100%;
  background-color: gray;
}
</style>

---

# Background & Motivation

.pull-left[

1. On 11 Mar 2020, the World Health Organization (WHO) declared the outbreak a global pandemic;

1. Brazil is fighting a terrible COVID-19 waves exacerbated by multiple inadequate responses;

1. The rapidly evolving situation has drastically altered people's lives;

1. Health also includes from the neck up diseases;

]

.pull-right[
1. The uncertainties and fears associated with the outbreak, along with mass lockdowns and economic recession are predicted negatively impact the mental health;

1. increase the psychological distress in the general population, as well as in persons with pre-existing mental disorders;

1. Lack of interpersonal attachments is linked to poor physical, emotional, and mental health (Baumeister & Leary, 1995);

1. There is a lot of forecasts related to economic aspects, but just a few related to mental health; 

1. Understand the consequences of the social distance in mental health;
]


---
class: center, middle, inverse

# The current study

---

# Objectives & Methodology 

.pull-left[
1. This is an exploratory survey;

1. Web-based data collection;

1. A convenience-sampling plan was the main strategy;

1. Participants were recruited at two times (the first current moment and 6 months later;

]

.pull-right[

1. To explore the impacts of social distancing and quarantine in two waves in aspects related to mental health, especially in depression and anxiety among Brazilian participants; 

1. To check for differences in anxiety and depression trajectories including time and quarantine-related variables as predictors (if you stay at home, how many hours you stay at home, how often you leave home).


]

---
# Demographic aspects 

```{r, include=FALSE }
#base
load("C:/Users/luisf/Dropbox/Puc-Rio/Projeto - COVID longitudinal/R base - covid 19 longitudinal.RData")
#packages
pacman::p_load(tidyverse, arsenal, janitor, dfSummary, DT, knitr, kableExtra, lme4, lmerTest, ggalluvial, patchwork)
```


```{r}
theme_set(theme_bw())
```


```{r }
ds_original_t1 %>% 
  select(sexo, 
         idade, escolaridade, estado_civil, rio_sp, quantos_filhos, plano_saude,
         como_vive, pessoas_na_casa, pessoas_em_risco_casa,
         renda_familiar_mensal, orientacao_politica,
         fica_em_casa_t1, tem_saido_de_casa, falou_familia_pessoalmente,testou_coronavirus, 
         conhece_alguem_coronavirus, segue_recomendacoes, concordo_covid_preocupacao_saude, 
         concordo_quarentena, concordo_covid_preocupacao_saude, 
         trabalho_antes_quarentena, trabalho_agora_quarentena, area_profissao,
         preocupacao_pagar_conta, pensa_em_adiar_pagamento, starts_with("preocupacao_"), 
         quanto_tempo_voce_acha_que_que_a_vida_voltara_ao_normal, 
         antes_da_quarentena_coronavirus_voce_fumava, agora_durante_a_quarentena_coronavirus_voce_fuma, 
         antes_da_quarentena_coronavirus_voce_bebia, agora_durante_a_quarentena_coronavirus_voce_bebe,
         antes_da_quarentena_coronavirus_voce_fazia_exercicios_fisicos, 
         agora_durante_a_quarentena_coronavirus_voce_faz_exercicios_fisicos,
         agora_durante_a_quarentena_coronavirus_sua_alimentacao,
         agora_durante_a_quarentena_coronavirus_seu_peso) %>% 
  tableby(~.,.) %>% 
  summary(., text = TRUE) %>% 
  data.frame() -> tab_demo
```



```{r}
#rmarkdown::paged_table(tab_demo)
#summary(tab_demo, text=TRUE) %>% 
#  as.data.frame(width = 20) %>% 
#  knitr::kable(format = "markdown", longtable = TRUE)

#kable(tab_demo, "markdown", booktabs = TRUE, longtable = TRUE, caption = "Test") %>%
#  kable_styling(latex_options = c("hold_position", "repeat_header"))

datatable(tab_demo,          
          extensions = c('FixedColumns',"FixedHeader"),
          options = list(scrollX = TRUE, 
                         dom = 'tp', #no search, but pagination
                         fixedHeader=TRUE))
```

---
class: center, middle, inverse

# Descriptives

---
class: middle, center

# First wave
## Contextual variables 

```{r, eval = FALSE }
ds_original_t1 %>% select(fica_em_casa_t1:falou_familia_pessoalmente) %>% 
  DataExplorer::plot_bar()
```

```{r,  }
col_names <- ds_original_t1 %>% select(fica_em_casa_t1:tem_saido_de_casa) %>% names
plot_list <- list()
for (i in col_names){
  plot <- ds_original_t1 %>% filter(!is.na(.data[[i]])) %>% ggplot(., aes_string(x=1, y=i)) +
    geom_bar(stat = "identity") 
    #geom_text(aes(y = ((..count..)/sum(..count..)), 
    #              label = scales::percent((..count..)/sum(..count..))), 
    #          stat = "count", vjust = -0.25) 
  #coord_polar("y", start=0, direction = -1) +
  plot_list[[i]] <- plot
}
plot_grob <- gridExtra::arrangeGrob(grobs=plot_list)
plot(plot_grob)
```

```{r out.width = "60%", eval = FALSE }
pieplotter <- function(col) {
  tibble(var = col) %>% 
    count(var) %>% 
    mutate(
      p = n/sum(n),
      y_mid = lag(cumsum(p), default = 0) + (p/2) 
    ) %>% 
    ggplot() +
    geom_col(
      aes(x = "", y = p, fill = var)
    ) +
    coord_polar(theta = "y") +
    #geom_text(
    #  aes(x = "", y = y_mid, label = scales::percent(p))
    #) +
    theme(
      axis.text.x = element_blank()
    )
}
map(ds_original_t1 %>% select(fica_em_casa_t1:tem_saido_de_casa) , pieplotter) %>% wrap_plots()& theme_minimal()
```


---
class: middle, center

# First wave
## Contextual variables 


```{r }
col_names <- ds_original_t1 %>% select(area_profissao,preocupacao_pagar_conta,pensa_em_adiar_pagamento) %>% names
plot_list <- list()
for (i in col_names){
  plot <- ds_original_t1 %>% filter(!is.na(.data[[i]])) %>% ggplot(., aes_string(x=1, y=i)) +
    geom_bar(stat = "identity") 
    #geom_text(aes(y = ((..count..)/sum(..count..)), 
    #              label = scales::percent((..count..)/sum(..count..))), 
    #          stat = "count", vjust = -0.25) 
  #coord_polar("y", start=0, direction = -1) +
  plot_list[[i]] <- plot
}
plot_grob <- gridExtra::arrangeGrob(grobs=plot_list)
plot(plot_grob)
```

---
class: center, middle, inverse

# Results


---
class: middle, center

# First wave

## I'm worried about (...)

```{r out.width = "65%"}
library(likert)
ds_original_t1 %>% select(starts_with("preocupacao_"),-preocupacao_pagar_conta) %>%
  rename_all(., ~paste(str_remove_all(.,"preocupacao_"))) %>% 
  rename_all(., ~paste(str_replace_all(.,"_", " "))) %>% 
  mutate_all(.,factor, levels =  c("Nenhuma","Pouca","Bastante","Extrema")) %>% 
  data.frame() %>% 
  likert() %>% plot(., wrap = 10,ordered = T,
                    low.color='darkblue', high.color='maroon') + 
  theme(legend.position = "right")

```

---
class: middle, center

# First wave

## Depression

```{r}
ds_original_t1 %>% 
  filter(!is.na(sexo)) %>% 
  ggplot(., aes(x=sexo, y = total_ces_t1)) + 
         geom_boxplot() +
  ggpubr::stat_compare_means(label.y = 65, label.x = 1.5,label = "p.signif") +
  geom_hline(yintercept = 16, size = 1, color = "red", linetype = "dashed")

```
.footnote[[*] Overall mean = `r round(mean(ds_original_t1$total_ces_t1, na.rm=T),2)` (SD: `r round(sd(ds_original_t1$total_ces_t1, na.rm=T),2)`).]

---
class: middle, center

# First wave

## Anxiety

```{r}
ds_original_t1 %>% 
  filter(!is.na(sexo)) %>% 
  ggplot(., aes(x=sexo, y = total_gad_t1)) + 
         geom_boxplot() +
  ggpubr::stat_compare_means(label.y = 25, label.x = 1.5,label = "p.signif") +
  geom_hline(yintercept = 10, size = 1, color = "red", linetype = "dashed")
```
.footnote[[*] Overall mean = `r round(mean(ds_original_t1$total_gad_t1, na.rm=T),2)` (SD: `r round(sd(ds_original_t1$total_gad_t1, na.rm=T),2)`).]

---
class: middle, center

# First wave

## Depression & Anxiety

```{r}
ggplot(ds_original_t1, aes(x = total_ces_t1, y = total_gad_t1)) +
  geom_jitter() +
  geom_smooth(method = "lm") +  
  ggpubr::stat_cor()
```



---

class: center, middle, inverse

# Follow up

---
class: middle, center

# Longitudinal analysis  



```{r}
d<-ds_t1_t2 %>% 
  select(unique_id, fica_em_casa_t1, fica_em_casa_t2) %>% 
  pivot_longer(-unique_id, names_to = "survey", values_to = "response") %>% 
  rename(subject = unique_id) %>%  #daqui para baixo é detalhe
  mutate(survey = str_replace_all(survey, "fica_em_casa_t1", "First wave")) %>% 
  mutate(survey = str_replace_all(survey, "fica_em_casa_t2", "Second wave")) %>% 
  mutate(response = factor(str_remove_all(response, "Entre*")))  %>% 
  mutate(response = factor(str_remove_all(response, "ou o dia todo*")))  
```


```{r}
#create a long dataset
d<-d %>% 
  group_by(survey,subject, response) %>% 
  count() %>% 
  mutate(pct = n / sum(n))
```


```{r  out.width = "65%"}
#change levels order
d <- transform(d,response = factor(response, rev(levels(response))))
#plot
ggplot(d,
       aes(x = survey, stratum = response, alluvium = subject,
           y = n,
           fill = response)) + #data
  scale_x_discrete(expand = c(.1, .1)) + #
  geom_flow() + #shade area conecting the rectangles
  geom_stratum(alpha = .7) + 
  geom_text(aes(label = paste0(..stratum.., "\n", n, "\n", 
                             scales::percent(..prop..) )), stat = "stratum", size = 3) + #text inside each rectangle
  scale_fill_brewer(type = "qual", palette = "Set2") +
  theme_void() +
  theme(legend.position = "none") #hide legends
```



---
class: inverse, middle, center

# Hypothesis
## Mental health (Time evolution and group condition)

--
Depression (CES)

---
# Table & Graph


.pull-left[

```{r, results = 'asis'}
ds_t1_t2 %>% 
  tableby(group ~ total_ces_t1 + total_ces_t2, total = FALSE, test = FALSE,data = .) %>% 
  summary()
```
]
.pull-right[


```{r }
ds_t1_t2 %>% 
  select(total_ces_t1,total_ces_t2, unique_id,group) %>% 
  pivot_longer(-c(group, unique_id), names_to = "time") %>% 
  ggplot(., aes(x = time, y = value, group = group, color = group)) +
  stat_summary(geom = "line", fun = "mean") +
  stat_summary(fun.data = mean_se, geom = "errorbar", width = 0.2) +
  theme_bw()

```
]


---

# Overall results 

```{r}
t.test(ds_t1_t2$total_ces_t1, ds_t1_t2$total_ces_t2, paired = T) %>% 
  pander::pander()
```




---

# Interaction 

```{r}
mod_ces <- ds_t1_t2 %>% 
  select(total_ces_t1,total_ces_t2, unique_id,group) %>% 
  pivot_longer(-c(group, unique_id), names_to = "time") %>% 
  lmer(value ~ factor(group) * factor(time) + (1|unique_id), data = .)

anova(mod_ces, type = 3, ddf = "Satterthwaite") %>% 
  kable(., digits = 2)
```

---
class: inverse, middle, center

# Hypothesis
## Mental health (Time evolution and group condition)

--

Depression (CES)

--

Anxiety (GAD)

---
# Overall results 

.pull-left[

```{r, results = 'asis'}
ds_t1_t2 %>% 
  tableby(group ~ total_gad_t1 + total_gad_t2, total = FALSE, test = FALSE,data = .) %>% 
  summary()
```
]
.pull-right[


```{r }
ds_t1_t2 %>% 
  select(total_gad_t1,total_gad_t2, unique_id,group) %>% 
  pivot_longer(-c(group, unique_id), names_to = "time") %>% 
  ggplot(., aes(x = time, y = value, group = group, color = group)) +
  stat_summary(geom = "line", fun = "mean") +
  stat_summary(fun.data = mean_se, geom = "errorbar", width = 0.2)

```
]


---

# Overall results 

```{r}
t.test(ds_t1_t2$total_gad_t1, ds_t1_t2$total_gad_t2, paired = T) %>% 
  pander::pander()
```

---

# Interaction 


```{r}
mod_gad <- ds_t1_t2 %>% 
  select(total_gad_t1,total_gad_t2, unique_id,group) %>% 
  pivot_longer(-c(group, unique_id), names_to = "time") %>% 
  lmer(value ~ factor(group) * factor(time) + (1|unique_id), data = .)

anova(mod_gad, type = 3, ddf = "Satterthwaite") %>% 
  kable(., digits = 2)
```

```{r}
ds_original_t2 %>% 
  select(
    como_vive, pessoas_na_casa, pessoas_em_risco_casa,
    renda_familiar_mensal, orientacao_politica,
    fica_em_casa_t2, tem_saido_de_casa, falou_familia_pessoalmente, 
    segue_recomendacoes, concordo_covid_preocupacao_saude, 
    concordo_quarentena, concordo_covid_preocupacao_saude, 
    
    trabalho_apos_quarentena, 
    
    preocupacao_pagar_conta, pensa_em_adiar_pagamento, 
    
    starts_with("preocupacao_"), 
    
    quanto_tempo_voce_acha_que_que_a_vida_voltara_ao_normal,
    atualmente_voce_fuma,
    atualmente_voce_bebe,
    atualmente_voce_faz_exercicios_fisicos,
    atualmente_sua_alimentacao,
    atualmente_seu_peso,
    plano_saude,
    como_vive
    ) %>% 
  tableby(~.,.) %>% 
  summary(., text = TRUE) %>% 
  data.frame() -> tab_demo_t2
```


```{r, eval = FALSE }
datatable(tab_demo_t2,          
          extensions = c('FixedColumns',"FixedHeader"),
          options = list(scrollX = TRUE, 
                         dom = 'tp', #no search, but pagination
                         fixedHeader=TRUE))
```


---
# Conclusions  

.pull-left[

1. The depression levels were above the recommended cutoff score at both times;

1. The anxiety levels were only close to cutoff, suggesting that these participants were not experiencing high anxiety during both times (however, subclinical symptoms are not desconsidered);

1. Compared to men, Women presented significantly higher results for depression and anxiety, as observed in the literature, in both times;

1. Social distancing, lockdowns, and quarantine are supposed to partially explain these outcomes;
]

.pull-right[
1. No significant differences were found in depression and anxiety among the groups (stay x leave)

1. Regardless the passage of time or being able to chose between stay of leave from home, they remained with high results

1. Non-significant results <b>are results</b> and we are still discussing our findings



]

.footnote[[*] Not really. See next page.]


---
# Thank you

This research was supported by PUC-Rio, UFRJ, and University of Luxembourg     
Public note: 4.125.060  (Plataforma Brasil)
emails:  luisfca@puc-rio.br, lucas.psic.rj@gmail.com, portugal.aca@gmail.com, landeira@puc-rio.br