COVID Symptom Study Sweden Open Dataset.

covidsymptom R Package

The covidsymptom R package provides an easy way to import open data from the COVID Symptom Study-Sweden. The package includes five datasets:

• national_estimates - daily estimated incidence of symptomatic COVID-19 in Sweden
• county_estimates - daily estimated incidence of symptomatic COVID-19 in Swedish counties
• postcode_estimates - daily estimated incidence of symptomatic COVID-19 in smaller Swedish regions (2-digit postcodes)
• csss_tests - daily proportion of positive COVID-19 tests reported by CSSS users
• symptoms - daily prevalences of symptoms reported by CSSS users

End of data collection

On 11 July 2022, we have completed the data collection phase of CSSS. We would like to thank our participants for their contribution to the research and our study since the start of the covid-19 pandemic. Without their data, it would not have been possible for us to follow the spread of infection in Sweden. We will continue to publish our results in scientific articles and at scientific meetings and we will share these results on the study’s website as soon as they have been reviewed and published. You can read more about the end of data collection here (Swedish). We are, therefore, working on the final version of this package for CRAN.

Table of Contents

• Installation
• Update Data
• Usage
  • National Estimates
  • Counties’ Estimates
  • 2-Digit Postcodes’ Estimates
  • COVID-19 Tests Results
  • Symptoms Prevalences
• Dashboard
• About COVID Symptom Study - Sweden

Installation

Install the CRAN version:

install_packages("covidsymptom")

Development version of the package can be installed from Github with:

install.packages("remotes")
remotes::install_github("csss-resultat/covidsymptom")

Update data

In order to respect CRAN best use practices, we will only push a new version of the package with most recent data every one month. However, data from COVID Symptom Study - Sweden is updated on a daily basis (see end of data collection), thus the dev version of the package is also updated daily. The function update_csss_data() (based on a similar function from the coronavirus package), checks updates to the dataset and re-install the package with the most recent data (dev version).

library(covidsymptom)
update_csss_data()

If you want to avoid updating the package to have the most recent data, you can use the function get_latest_data() to import the latest version available. Notice however that this won’t re-write the package datasets.

library(covidsymptom)
national_estimates <- get_latest_data(data_level = "national")

head(national_estimates)
#> [1] "Your data is up-to-date"

Usage

• National estimates

library(ggplot2)
library(dplyr)

covidsymptom::national_estimates %>%
  ggplot(aes(x = Datum, y = Uppskattning)) +
  geom_line(col = "#a60f61") +
  geom_point(col = "#a60f61", size = 0.5) +
  labs(x = "Date", y = "Predicted number of cases as % of the population",
       title = "Predicted Number of COVID-19 Cases by Date",
       subtitle = "")  +
  scale_y_continuous(limits = c(0, 1.3),
                     breaks = seq(0, 1, 0.2),
                     labels = paste0(format(seq(0, 1, 0.2), decimal.mark = ","), " %"), # add % sign to the labels
                     expand = c(0,0)) +
  scale_x_date(date_breaks = "30 days", date_labels = "%d %b, %Y") + # can be "1 day", "2 days", etc.
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1), panel.grid.minor.x = element_blank()) +
  geom_ribbon(aes(ymin = Low_CI, ymax = High_CI), fill = "#a60f61", alpha = 0.09)

• Counties’ estimates

library(ggplot2)
library(dplyr)
library(lubridate)

counties_selection <- c("Skane", "Stockholm", "Vastra Gotaland", "Uppsala")

covidsymptom::county_estimates %>%
  filter(Lan %in% counties_selection) %>%
  ggplot(aes(x = Datum, y = Uppskattning, color = Lan)) +
      geom_line() +
      #geom_point(size = 0.5) +
      labs(x = "Datum", y = "Uppskattad förekomst", title = "% Uppskattad förekomst av symtomatisk covid-19", subtitle = "") +
      scale_x_date(date_breaks = "50 days") +
      theme_minimal() + 
      theme(axis.text.x = element_text(angle = 45, hjust = 1), panel.grid.minor.x = element_blank(),
            legend.position = "none", plot.title = element_text(hjust = 0.5)) + 
      geom_ribbon(aes(ymin = Low_CI, ymax = High_CI), color = "transparent", fill = "#a60f61", alpha = 0.09) +
      facet_wrap(. ~ Lan) 

• 2-digit postcodes’ estimates

COVID Symptom Study - Sweden provides also predictions at 2-digit postcode levels.

library(ggplot2)
library(dplyr)
library(lubridate)
library(gt)

stockholm_codes <- c('11', '12', '13', '14', '15', '16', '17', '18', '19', '76')

filtered_data <- covidsymptom::postcode_estimates %>%
  filter(Postnummer %in% stockholm_codes & Datum == as.Date("2021-01-31") & !is.na(Uppskattning))

min_pred <- min(filtered_data$Uppskattning)
max_pred <- max(filtered_data$Uppskattning)
pred_palette <- scales::col_numeric(c("#f9dee2", "#5E0B21"), domain = c(min_pred, max_pred), alpha = 0.75)


filtered_data %>%
  arrange(desc(Uppskattning)) %>%
  gt(.) %>%
  tab_header(
    title = md("**Predicted number of cases as % of the population**"),
    subtitle = "Stockholm's 2-digit regions"
  ) %>% 
  cols_width(starts_with("Datum") ~ px(95)) %>%
  tab_style(
    locations = cells_column_labels(columns = everything()),
    style     = list(
      cell_borders(sides = "bottom", weight = px(3)),
      cell_text(weight = "bold")
    )) %>% 
  cols_align("center") %>% 
  data_color(columns = vars(Uppskattning),
             colors = pred_palette) %>%
  tab_source_note(source_note = "Data: COVID Symptom Study Sweden") %>%
  tab_options(
    column_labels.border.top.width = px(3),
    data_row.padding = px(3),
    source_notes.font.size = 12,
    table.font.size = 12,
    heading.align = "center",
    row_group.background.color = "grey")

stockholm_codes <- c('11', '12', '13', '14', '15', '16', '17', '18', '19')

covidsymptom::postcode_estimates %>%
  filter(Postnummer %in% stockholm_codes) %>%
  ggplot(aes(x = Datum, y = Uppskattning, color = Postnummer)) +
  geom_line() +
  #geom_point(size = 0.5) +
  labs(x = "Datum", y = "Uppskattad förekomst", title = "% Uppskattad förekomst av symtomatisk covid-19", subtitle = "") +
  scale_x_date(date_breaks = "90 days") +
  theme_minimal() + 
  theme(axis.text.x = element_text(angle = 45, hjust = 1), panel.grid.minor.x = element_blank(),
        legend.position = "none", plot.title = element_text(hjust = 0.5)) + 
  geom_ribbon(aes(ymin = Low_CI, ymax = High_CI), color = "transparent", fill = "#a60f61", alpha = 0.09) +
  facet_wrap(. ~ Postnummer, scales= "free")   

• covid-19 tests results

library(gganimate)
library(ggdark)

tests_df <- covidsymptom::csss_tests

a_plot <- ggplot(tests_df, aes(x=date, y=total_tests)) +
  geom_line() +
  geom_segment(aes(xend=max(date), yend = total_tests), linetype=2) +
  geom_point(size = 3) + 
  geom_text(aes(x = max(date), label = sprintf("%2.0f", total_tests)), hjust=0, color = "grey") +
  transition_reveal(date) + 
  view_follow(fixed_y = TRUE)+
  coord_cartesian(clip = 'off') + 
  labs(title = 'Reported antigen tests in the Covid Symptom Study Sweden', y = 'N# Antigen tests', x = "Date") +
  enter_drift(x_mod = -1) + exit_drift(x_mod = 1) +
  dark_theme_gray() +
  theme(plot.margin = margin(5.5, 40, 5.5, 5.5))

animate(a_plot, fps = 5)

• Symptoms prevalences

library(ggplot2)
library(dplyr)
library(ggrepel)

symptoms_df <- covidsymptom::symptoms %>%
  mutate(Vikt = factor(Vikt, levels = c("Positiv", "Negativ")))
symptoms_df[symptoms_df$Symptom == "Förlorat eller förändrat lukt-/smaksinne",]$Symptom <- "Förlorat eller förändrat\nlukt-/smaksinne"
data_ends <- symptoms_df %>% 
  filter(Datum == max(Datum)) 

symptoms_df %>%
  ggplot(aes(x = Datum, y = Andel, color = Symptom)) +
  geom_line() +
  theme_light(base_size = 11) +
  labs(x = "Datum", y = "Andel deltagare med symptom", title = "Covid19 symptom", subtitle = "", color = "") +
  scale_x_date(date_breaks = "60 days", date_labels = "%d %b, %Y", limits = as.Date(c("2020-04-015","2022-07-01"))) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1), panel.grid.minor.x = element_blank(),
        plot.title = element_text(hjust = 0.5, face="bold"), legend.position = "none",
        strip.text = element_text(face="bold")) + 
 geom_label_repel(
   aes(label = Symptom, fill = Symptom), data = data_ends,
   fontface ="plain", color = "black", size = 3,
   xlim = as.Date(c("2022-03-20", "2022-07-01"))) +
  facet_wrap(~Vikt, scales = "free_y", ncol =1) 

Dashboard

A supporting dashboard is available here

About COVID Symptom Study - Sweden

The COVID Symptom Study was a non-commercial project that used a free smartphone app to facilitate real-time data collection of symptoms, exposures, and risk factors related to COVID-19. The app was developed by researchers at King’s College and Guys and St Thomas’ Hospitals in London in partnership with health science company Zoe Global Ltd. Baseline data and recurring daily questions are described in Drew et al (Science, 2020). The app was launched in the UK and US March 2020. In Sweden, the study is based at Lund University and, as per a collaboration agreement on 28 July 2020, Uppsala University. More about the Swedish part of the study is described in Kennedy & Fitipaldi, et al. (Nat Comms, 2022) The app was launched in Sweden on April 29, 2020 as part of a national research initiative on COVID-19. >4.5 million participants in the three countries used the app, ~220,000 of whom live in Sweden. On July 11, 2022, with over 500 million data entries, in which Swedish participants contributed ~20 million of these, the project officially ended the data collection phase.