patient_data.qmd

---
title: "GECKO data dashboard"
format: 
  html:
    embed-resources: true
theme: cerulean
editor: source
execute:
  echo: false
  warning: false
editor_options: 
  chunk_output_type: console
callout-appearance: simple
---

```{r}
library(tidyverse)
library(scales)
library(bslib)
library(bsicons)
library(shiny)
library(consort)


# GECKO green is #74a73e

source("patient_data_pull.R")
#save(patient_data_orig, file = "patient_data_orig.rda")
#load("patient_data_orig.rda")

source("registrations_pull.R")
#save(registrations_orig, file = "registrations_data_orig.rda")


#patient_data_orig$period %>% unique() %>% factor() %>% levels()

n_records = nrow(patient_data_orig)

# patient_data_orig %>% 
#   count(internal_check)

patient_data = patient_data_orig %>% 
  # REDCap sends the data fields including sub-var names in {}, removing here:
  mutate(op_performed = str_extract(op_performed, "^.*(?=(\\{))")) %>% 
  mutate(period = fct_explicit_na(period, "Missing")) %>% 
  mutate(wb = fct_relevel(wb, "High income", "Upper middle income", "Lower middle income"))

# patient_data %>% 
#   count(inclusion_exclusion, internal_check)

patient_data_cleaned = patient_data %>% 
  filter(internal_check == "Pass - included in analysis")

patient_data = patient_data %>% 
  mutate(inclusion_exclusion = na_if(inclusion_exclusion, "Include - this is a valid record for a patient who fulfils inclusion criteria")) %>% 
  mutate(internal_check = na_if(internal_check, "Pass - included in analysis"))

# patient_data %>% 
#   count(inclusion_exclusion, internal_check)


n_records_cleaned = nrow(patient_data_cleaned)

registrations = registrations_orig

source("world_map.R")


```

```{r}
# Functions

theme_set(theme_classic())
theme_update(axis.text  = element_text(size = 16, colour = "black"),
             axis.title = element_text(size = 16, colour = "black"),
             plot.title = element_text(size = 20, colour = "black"),
             strip.text = element_text(size = 16, colour = "black"))

barplot_count = function(df, var, title = NA, high_n = 50, infreq = TRUE){
  if (infreq){
    df = df %>% 
      mutate(plotvar = fct_infreq({{var}}))
  } else{
    df = df %>% 
      mutate(plotvar = {{var}})
  }
  df %>% 
    drop_na({{var}}) %>% 
    ggplot(aes(y = fct_rev(plotvar) %>% fct_relevel("Missing"))) +
    geom_bar(colour = "#2166ac", fill = "#d1e5f0") +
    geom_text(stat='count', aes(label=comma(after_stat(count)),
                                hjust = if_else(after_stat(count) > high_n, 1, 0)),
              size  = 10,
              label.padding=unit(0.3, "lines")) +
    scale_y_discrete(labels = label_wrap(15)) +
    scale_x_continuous(expand = expansion(mult = c(0, 0), add = c(0, 0))) +
    ggtitle(str_wrap(title, 30)) +
    ylab("") +
    xlab("Number of patients")
  
}

histogram = function(df, var, title = NA, xlab = NA, ylab = NA){
  df %>% 
    ggplot(aes({{var}})) +
    geom_histogram(binwidth = 10) +
    ggtitle(str_wrap(title, 30)) +
    xlab(xlab) +
    ylab(ylab) +
    scale_x_continuous(expand = expansion(mult = c(0, 0), add = c(0, 0))) +
    scale_y_continuous(expand = expansion(mult = c(0, 0.1), add = c(0, 0)))
}

```

::: {layout-ncol="1"}

Records that pass internal quality check:

```{r}
bslib::value_box(
  title = "GECKO patient data:",
  value = paste(prettyNum(n_records_cleaned, big.mark = ","), "records from"),
  showcase = bs_icon("clipboard2-pulse-fill"),
  h3(paste(n_distinct(patient_data_cleaned$redcap_data_access_group) %>% 
             prettyNum(big.mark = ","), "hospitals in")),
  h3(paste(n_distinct(patient_data_cleaned$country), "countries")),
  br(),
  p("Updated:"),
  p(format(Sys.time(), "%d-%b %Y")),
  theme_color = "success"
)
```

```{r}
#| include: false
consort_plot(patient_data,
             order = list(record_id    = "Records started on REDCap",
                          inclusion_exclusion   = "Excluded with reasons:",
                          record_id = "Records considered",
                          internal_check  = "Internal DQ check",
                          #record_id    = "Included in analysis",
                          wb      = "Included in analysis",
                          record_id    = " "),
             side_box = c("inclusion_exclusion", "internal_check"),
             allocation = "wb",
             cex = 0.8,
             text_width = 30)

```


Records started on REDCap:

```{r}
bslib::value_box(
  title = "GECKO patient data:",
  value = paste(prettyNum(n_records, big.mark = ","), "records from"),
  showcase = bs_icon("clipboard2-pulse-fill"),
  h3(paste(n_distinct(patient_data$redcap_data_access_group) %>% 
             prettyNum(big.mark = ","), "hospitals in")),
  h3(paste(n_distinct(patient_data$country), "countries")),
  br(),
  p("Updated:"),
  p(format(Sys.time(), "%d-%b %Y")),
  theme_color = "success"
)
```



```{r}
bslib::value_box(
  title = "GECKO registrations:",
  value = paste(nrow(registrations)%>% 
             prettyNum(big.mark = ","), "hospitals from"),
  showcase = bs_icon("clipboard2-pulse-fill"),
  h3(paste(n_distinct(registrations$country), "countries")),
  br(),
  p("Updated:"),
  p(format(Sys.time(), "%d-%b %Y")),
  theme_color = "warning"
)
```

```{r}
map_patients
```

```{r}
map_registrations
```
:::

::: {layout-ncol="2"}

::: {.callout-tip icon=false collapse=true}

# Patient numbers Top 10

```{r}
patient_data %>% 
  count(country, sort = TRUE, name = "n_patients") %>% 
  slice(1:10) %>% 
  knitr::kable()
```

## NIHR Global Surgery Unit hub countries

```{r}
hub_countries = c("Mexico", "Nigeria", "India", "Ghana", "Rwanda", "Benin", "South Africa")
patient_data %>% 
  filter(country %in% hub_countries) %>% 
  mutate(country = fct_expand(country, hub_countries)) %>% 
  count(country, sort = TRUE, name = "n_patients", .drop = FALSE) %>% 
  knitr::kable()

```

:::

::: {.callout-warning icon=false collapse=true}

# Registrations Top 10

```{r}
patient_data %>% 
  count(redcap_data_access_group, sort = TRUE, name = "n_patients") %>% 
  slice(1:10) %>% 
  knitr::kable()
```

#### Registrations NIHR Global Surgery Unit hub countries

```{r}
hub_countries = c("Mexico", "Nigeria", "India", "Ghana", "Rwanda", "Benin", "South Africa")
registrations %>% 
  filter(country %in% hub_countries) %>% 
  mutate(country = fct_expand(country, hub_countries)) %>% 
  count(country, sort = TRUE, name = "n_hospitals", .drop = FALSE) %>% 
  knitr::kable()

```

:::

:::

# Operative approach - Minimally invasive surgery (MIS) vs Open
::: panel-tabset

## Map

```{r}
map_op_approach
```

## Plot

Interactive dataset explorer: https://argoshare.is.ed.ac.uk/gecko_shinyviz/

```{r, fig.width = 12, fig.height = 12}
barplot_count(patient_data_cleaned, op_approach,
              "Operative approach by World Bank income group",
              high_n = 1500) +
  scale_x_continuous(expand = expansion(mult = c(0, 0.01), add = c(0, 0)), labels = scales::label_comma()) +
  facet_wrap(~wb, ncol = 1)
```


## Table

```{r}
op_approach %>%
  gt::gt() %>% 
  gt::opt_interactive(use_search = TRUE,
                      use_pagination = FALSE)
```



## Histogram

```{r}
op_approach %>% 
  ggplot(aes(prop_laprob)) +
  geom_histogram() +
  xlab("Proportion of Lap/Rob/Conv") +
  ylab("n countries")
```


:::



::: {layout-ncol="2"}
```{r, fig.height=10}
barplot_count(patient_data_cleaned, period, "Data collection period", high_n = 2000, infreq = FALSE)
```

```{r}
barplot_count(patient_data_cleaned, wb, "World bank income group", high_n = 2000, infreq = FALSE)
```

```{r}
barplot_count(patient_data_cleaned, op_performed, "Operation performed", high_n = 4000) +
  scale_x_continuous(expand = expansion(mult = c(0, 0.01), add = c(0, 0)))
```

:::