index.Rmd

---
title: "Visualize the output dataset for COVerAGE-DB"
---


```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE, warning = FALSE, message=FALSE, fig.width=8, fig.height=6)
```
 
```{r, include = FALSE}
invisible(lapply(list.files(here::here("R"), full.names = TRUE), source))
suppressPackageStartupMessages({
  library("data.table")
  library("ggplot2")
  library("readr")
  library("cowplot")
  library("osfr")
  library("countrycode")
})
source_1 <- "COVerAGE-DB: A database of COVID-19 cases and deaths by age (https://osf.io/mpwjq/)"
WorldRobinson <- readRDS(here::here("data/NEWorldRobinson.rds"))
geo_df <- readRDS(here::here("data/WorldRobinson_df.rds"))

#
dtJHU <- get.JHU.daily()
dtJHU2 <- process_dtJHU(dtJHU)

# download input.DB as zip file and read in 
# `load_fresh` will save a local copy
dt5_ori <- load_fresh(backup = TRUE)
dt5_ori <- fread(here::here("data_backup/dt5_ori.csv"))
dt5 <- clean_outputDB(dt5_ori) # dt5 has only latest day 
countries_w_new_update <- dt5[Date>"2020-10-01", unique(Country)]
all_countries <- get_cnames(dt5) # all countries in the datasets
# range of `max_date`
dt_date <- unique(dt5[,.(Country, max_date)])
  
# make dt5_country (MPIDR by country and age and sex) for Tableau 
dt5_country <- get_dt5_pool_sex(dt5) # dt5_country has all the countries, pooled sex into total and remove NA value
dt5_country[, New:=ifelse(Country%in%countries_w_new_update, 1, 0)]
dt5_country <- join_region(dt5_country) # joined to region 

dt5_country <- join_wpp_pop_dth(dt5_country)
# dt5_country[, update_date:= format.Date(dt5_ori$update_date[1], "%d %B %Y")]
dt5_country[, update_date:= as.Date(dt5_ori$update_date[1], format = "%d %B %Y")]

dt5_country_colorder <- readRDS(here::here("data/dt5_country_colorder.rds"))
setcolorder(dt5_country, dt5_country_colorder)
dt5_country[,.N]
# 2/11: 8383
# unique(dt5_country[,.(Measure, Sex2, WBRegion4, nc_MPIDR_WB)]) # Number of MPIDR countries by WB group 

# agg country and age get total
# dt5_total <- get_sum_country_age(dt5_country)
dt5_summary_simple <- get_summary_table(dt5_country)
dt5_summary_simple[, Pcent_0_9:= (`Age 0-4` + `Age 5-9`) / Sum0_19 * 100] # 0-10 % in 0-20
dt5_summary_simple[, Pcent_10_19:= (`Age 10-14` + `Age 15-19`) / Sum0_19 * 100]

# agg by WB region
dt5_summary_region <- get_summary_table_region(dt5_country, dtJHU2)
dt5_summary_region[1:6,]
# 
# aggregate at country level with JHU country data 
dt5_summary <- get_summary_table_region2(dt5_country, dtJHU2)
dt5_summary[,.N]
# 744

dt5_summary_simple

```


```{r}
# plots ----
# dt_ava shows if there is only case, or only deaths or both for each country by sex
dt_ava <- copy(dt5_country)[, Measure := paste(sort(unique(Measure)), collapse = "_"), by = .(Sex, Country)]
dt_ava[Sex!="Both", Sex:= "Sex-specific"]
dt_ava2 <- unique(dt_ava[,.(Country, Sex, Measure)]) # used for map, `plot_ava_map2`
# dt_ava2[Sex=="Both"&grepl("Cases", Measure) ,.N]


# latest date of updating country
dt_date <- unique(dt5[, .(Country, Sex, Date)])
dt_date[, Month := format(Date, "%Y-%m")]
# There are `r NC_New` countries updated after October 2020:
NC_New <- paste0(dt_date[Month >= "2020-10", uniqueN(Country)], " (", 
                 round(dt_date[Month >= "2020-10", uniqueN(Country)]/dt_date[,uniqueN(Country)]*100), "%)")

# agg country, get total by age
dt5_total_age <- get_sum_country(dt5_country)

# then get regional summary
dt5_total_region <- get_sum_country_region(dt5_country) # get regional (both total and sex-specific )

# plot barchart by age groups using dataset `dt5_total_age` 
bar_total_age1 <- make_barchart2(dt5_total_age[Sex=="Both"], Measure0 = "Cases")
bar_total_age2 <- make_barchart2(dt5_total_age[Sex=="Both"], Measure0 = "Deaths")
bar_total_age <- cowplot::plot_grid(bar_total_age1, bar_total_age2)

# plot barchart by WB region
bar_total_age_region <- make_barchart2(dt5_total_region[Sex=="Both"], label = FALSE)


# age pyramid for case and death 
dt5_total_age$Sex <- factor(dt5_total_age$Sex, levels = c("Both", "Male", "Female"))
plots_by_measure1 <- plot.measure(data = dt5_total_age[Sex!="Both"], Measure0 = "Cases", big_plot = TRUE, hide_legend = TRUE)
plots_by_measure2 <- plot.measure(data = dt5_total_age[Sex!="Both"], Measure0 = "Deaths", big_plot = TRUE, hide_legend = FALSE)
pyramid_age_sex <- cowplot::plot_grid(plots_by_measure1, plots_by_measure2) 

# age pyramid for case and death --- by WB region
rcs <- c("Low income",
           "Lower middle income",
           "Upper middle income",
           "High income")
grid0 <- expand.grid(rcs, c("Cases", "Deaths"))

plot.all <- function(region0, measure0){
  plot.measure(dt5_total_region[Sex!="Both"], measure0, region0, big_plot = TRUE, hide_legend = TRUE, hide_text = TRUE)
}
plots_by_region_sex <- Map(plot.all, region0 = grid0$Var1, measure0 = grid0$Var2)
pyramid_age_sex_region <- cowplot::plot_grid(plotlist = plots_by_region_sex, ncol = 2)    

# 
# one map each 
# In the `dt5_country` dataset, countries with single/both sex always have total
c_case <- dt5_country[Measure=="Cases" & Sex == "Both", unique(Country)]
c_death <- dt5_country[Measure=="Deaths" & Sex == "Both", unique(Country)]
c_case_sex<- dt5_country[Measure=="Cases" & Sex != "Both", unique(Country)]
c_death_sex <- dt5_country[Measure=="Deaths" & Sex != "Both", unique(Country)]

map_total <- plot_ava_map2(dt_ava2[Sex == "Both"])
map_sex_specific <- plot_ava_map2(dt_ava2[Sex == "Sex-specific"])

# aggregated plots for all countries ---- 
all_countries <- get_cnames(dt5)
# `get_dt_for_total` runs a cleaning function, check if CFR could be obtained
data_total1 <- rbindlist(lapply(all_countries, get_dt_for_total,
                                data = dt5,
                                target_interval = seq(0, 90, by = 5),
                                get_f_m = FALSE))
# for the 0-19 group total
CFR_0_19 <- round(calculate.CFR(data_total1)[Measure=="Deaths", Value]/calculate.CFR(data_total1)[Measure=="Cases", Value] * 100, 3)

data_total2 <- rbindlist(lapply(all_countries, get_dt_for_total, 
                                data = dt5,
                                target_interval = seq(0, 90, by = 5),
                                get_f_m = TRUE))

# CFR plot
data_total1 <- revise.data.total(data_total1) # calculate CFR
data_total2 <- revise.data.total(data_total2)
CFR_total <- plot.measure(data_total1, Measure0 = "Case fatality rate", 
                   big_plot = TRUE, hide_legend = TRUE) # total
CFR_sex <- plot.measure(data_total2, Measure0 = "Case fatality rate", 
                   big_plot = TRUE, hide_legend = FALSE) # sex-specific
bar_CFR <- cowplot::plot_grid(CFR_total, CFR_sex) 
```

**This page uses the harmonized dataset with 5-year age interval. **

COVerAGE-DB is an open-access database including cumulative counts of confirmed COVID-19 cases, deaths, and tests by age and sex. Original data and sources are provided alongside data and measures in standardized and age-harmonized formats. The data were extracted from reports published by official governmental institutions in a variety of formats.
The [project page](https://github.com/timriffe/covid_age) and the [medRxiv paper](https://www.medrxiv.org/content/10.1101/2020.09.18.20197228v2) contain the technical details. 

The database is still in development. On `r Sys.Date()` it includes `r dt5_ori[, uniqueN(Country)]` countries and `r dt5_ori[Region!="All", uniqueN(Region)]` subnational areas.

**Data Availability**

The `r length(c_case)` countries with national data on counts of COVID-19 cases and the `r length(c_death)` countries with national data on counts of COVID-19 deaths.
The `r length(c_case_sex)` countries with national data on sex-specific counts of COVID-19 cases and the `r length(c_death_sex)` countries with national data on sex-specific counts of COVID-19 deaths.

```{r, fig.width=8, fig.height=4}
map_total
map_sex_specific
```


**Age distribution of global COVID-19 cases and deaths**
```{r, fig.width=10, fig.height=4}

bar_total_age
```

**Age distribution of global COVID-19 cases and deaths by sex**
```{r, fig.width=10, fig.height=4}
pyramid_age_sex
```

**Case fatality rate (CFR), with 5-year age interval. **
CFR is calculated by dividing death counts by case counts for each age group. In calculating CFR, only countries with both case and death data are counted.

```{r, fig.width=10, fig.height=4}
bar_CFR
```

**Age distribution of COVID-19 cases and deaths by income group**
```{r, fig.width=10, fig.height=10}
bar_total_age_region
```

**Age distribution of global COVID-19 cases and deaths by sex and income group**
```{r, fig.width=10, fig.height=14}
pyramid_age_sex_region
```
 

## Reference  

Riffe, T., Acosta, E., and The COVerAGE-DB team (2020). COVerAGE-DB: A database of COVID-19 cases and deaths by age. medRxiv https://doi.org/10.1101/2020.09.18.20197228

Dong E, Du H, Gardner L. An interactive web-based dashboard to track COVID-19 in real time. Lancet Inf Dis. 20(5):533-534. https://doi.org/10.1016/S1473-3099(20)30120-1