---
title: "Vorhofkatheter-Statistik"
author: "Daniel Kraus"
date: '2018-12-29'
output:
  slidy_presentation: default
  ioslides_presentation: default
  beamer_presentation: default
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE, warning = FALSE)
library(tidyverse)

raw_data = read_csv('vhk.csv') %>%
  mutate(Year = lubridate::year(Date))

cath_by_year = raw_data %>% count(Year)
first_year = min(raw_data$Year)
last_year = max(raw_data$Year)
max_y_break = ((max(cath_by_year$n) %/% 10) + 1) * 10

```

## Katheterimplantationen pro Jahr
```{r cath_by_year }
cath_by_year %>%
  ggplot(aes(x = Year, y = n)) +
  geom_col() +
  scale_y_continuous(breaks = seq(from = 0, to = max_y_break, by = 10)) +
  scale_x_continuous(breaks = seq(from = first_year, to = last_year, by = 1)) +
  labs(x = NULL, y = "Anzahl Katheter")
```

<!--
## Katheterimplantationen im Jahresverlauf
```{r cath_by_month}
raw_data %>% mutate(Month = lubridate::month(Date)) %>%
  group_by(Year) %>%
  count(Month) %>%
  ggplot(aes(x = Month, y = n, group = Year, alpha = Year)) +
  geom_point() +
  geom_line()
  
```
-->

## Katheterexplantationen pro Jahr
```{r expl_by_year}
raw_data %>% mutate(ExplYear = lubridate::year(RemovalDate)) %>%
  # group_by(InsertionSite, Side) %>%
  count(ExplYear) %>%
  ggplot(aes(x = ExplYear, y = n)) +
  geom_col() +
  # facet_grid(rows = vars(InsertionSite), cols = vars(Side)) +
  scale_x_continuous(breaks = seq(from = first_year, to = last_year, by = 1)) +
  labs(x = NULL, y = "Explantationen")
```

## Explantationen pro Implantation pro Jahr
```{r expl_by_cath_by_year}
raw_data %>% mutate(ImplYear = lubridate::year(Date), ExplYear = lubridate::year(RemovalDate)) %>%
  group_by(ImplYear) %>%
  summarise(ExplByImpl = sum(!is.na(ExplYear)) / n()) %>%
  ggplot(aes(x = ImplYear, y = ExplByImpl)) +
  geom_col() +
  scale_x_continuous(breaks = seq(from = first_year, to = last_year, by = 1)) +
  labs(x = NULL, y = "Explantationen pro Implantation")
```

## Verweildauern der Katheter
```{r durations, message=FALSE}
raw_data %>% mutate(Year = lubridate::year(Date), Duration = RemovalDate - Date) %>%
  group_by(Year) %>%
  summarize(MedianDuration = median(Duration, na.rm = TRUE)) %>%
  ggplot(aes(x = Year, y = MedianDuration)) +
  geom_col() +
  scale_x_continuous(breaks = seq(from = first_year, to = last_year, by = 1)) +
  labs(x = NULL, y = "Mediane Katheter-Verweildauer [Tage]")
```

## Gründe der Katheterexplantation
### Variante A: Absolute Zahlen
```{r removal_reasons, message=FALSE}
raw_data %>% filter(!is.na(RemovalDate), !is.na(RemovalReason)) %>%
  mutate(ExplYear = lubridate::year(RemovalDate) %% 100) %>%
  group_by(ExplYear) %>%
  count(RemovalReason) %>%
  ggplot(aes(x = ExplYear, y = n)) +
  geom_point() + geom_line() +
  scale_x_continuous(breaks = scales::pretty_breaks()) +
  scale_y_continuous(breaks = scales::pretty_breaks()) +
  facet_wrap(vars(RemovalReason)) +
  labs(x = NULL, y = "Anzahl entfernter Katheter")
```

### Variante B: auf die Zahl der in dem Jahr gelegten Katheter bezogen
```{r removal_reasons_normalized, message=FALSE}
# Zur Berechnung dieses Index muß man zunächst für jeden explantierten Katheter
# berechnen, wie viele Katheter im *ex*plantationsjahr *im*plantiert wurden.
impl_per_year = raw_data %>% mutate(ImplYear = lubridate::year(Date)) %>% count(ImplYear)

raw_data %>% 
  select(Date, RemovalDate, RemovalReason) %>%
  mutate(ImplYear = lubridate::year(Date) %% 100, ExplYear = lubridate::year(RemovalDate)) %>%
  left_join(impl_per_year, by = c("ExplYear" = "ImplYear")) %>% # creates column "n"
  filter(!is.na(RemovalDate), !is.na(RemovalReason)) %>%
  group_by(ExplYear) %>%
  add_count(RemovalReason) %>% # creates column "nn"
  ungroup() %>%
  select(ExplYear, RemovalReason, n, nn) %>%
  mutate(i = nn/n) %>%
  group_by(ExplYear, RemovalReason) %>%
  # summarize(i = sum(i)) %>%
  distinct() %>%
  ggplot(aes(x = ExplYear, y = i)) +
  geom_point() + geom_line() +
  scale_x_continuous(breaks = scales::pretty_breaks()) +
  scale_y_continuous(breaks = scales::pretty_breaks()) +
  facet_wrap(vars(RemovalReason)) +
  labs(x = NULL, y = "Anzahl entfernter Katheter / gelegter Katheter")
```


## Alter der Patienten
```{r patient_age}
raw_data %>%
  ggplot(aes(group = Year, x = Year, y = Age)) +
  geom_boxplot() +
  coord_cartesian(ylim = c(20, 100)) +
  scale_x_continuous(breaks = seq(from = first_year, to = last_year, by = 1)) +
  scale_y_continuous(breaks = seq(from = 20, to = 100, by = 10))  +
  labs(x = NULL, y = "Jahre")
```

## Geschlecht der Patienten
```{r patient_sex}
raw_data %>% group_by(Year) %>% summarise(PercentFemale = sum(Sex == "weiblich") / n()) %>%
  ggplot(aes(x = Year, y = PercentFemale)) +
  geom_col() +
  scale_x_continuous(breaks = seq(from = first_year, to = last_year, by = 1)) +
  coord_cartesian(ylim = c(0, 1)) +
  scale_y_continuous(labels = scales::percent_format(accuracy = 1)) +
  labs(x = NULL, y = "Anteil Frauen")
```

## Katheterlokalisation
Ist da ein Trend hin zu immer mehr Kathetern von links?!

```{r insertion_site}
raw_data %>% mutate(Side = factor(Side, levels = c("rechts", "links"))) %>%
  ggplot(aes(x = Year)) +
  facet_grid(InsertionSite ~ Side) +
  geom_bar() +
  labs(x = NULL, y = "Anzahl Katheter")
```

## Anteil der Arztrollen
Um 2014 herum haben einige die Facharztprüfung abgelegt, ist das der Grund für die Auffälligkeit 2015/2016?

```{r percent_residents}
raw_data %>% group_by(Year) %>% 
  summarize(Assistenzarzt = sum(SurgeonRole == "Assistenzarzt") / n(),
            Facharzt = sum(SurgeonRole == "Facharzt") / n(),
            Oberarzt = sum(SurgeonRole == "Oberarzt") / n()) %>%
  gather(key = Role, value = Percent, Assistenzarzt, Facharzt, Oberarzt) %>%
  ggplot(aes(x = Year, y = Percent)) +
  scale_x_continuous(breaks = seq(from = first_year, to = last_year, by = 1)) +
  scale_y_continuous(labels = scales::percent_format(accuracy = 1)) +
  facet_grid(Role ~ .) +
  geom_col() +
  labs(x = NULL, y = "Anteil in den gelegten Kathetern")
```

## Hitparade der Durchleuchtungsdauern
```{r greatest_fluoroscopy}
raw_data %>%
  group_by(Surgeon) %>%
  summarize(FluoroscopyIndex = median(InsertionFluoroscopyDuration, na.rm = TRUE)) %>%
  arrange(desc(FluoroscopyIndex)) %>%
  top_n(-10, FluoroscopyIndex) %>%
  mutate(Surgeon = factor(Surgeon, levels = Surgeon)) %>%
  ggplot(aes(x = Surgeon, y = FluoroscopyIndex)) +
  geom_col() +
  coord_flip() +
  labs(x = NULL, y = "Median der Durchleuchtungsdauer [s]")
```

## Individuelle Durchleuchtungsdauern
Nur Operateure der letzten 4 Jahre

```{r individual_fluoroscopy, message=FALSE}
raw_data %>% filter(Year > lubridate::year(lubridate::today()) - 4, !is.na(InsertionFluoroscopyDuration)) %>%
  mutate(Year = Year %% 100) %>%
  group_by(Surgeon, Year) %>%
  summarize(FluoroscopyIndex = median(InsertionFluoroscopyDuration, na.rm = TRUE)) %>%
  ungroup() %>%
  # mutate(Surgeon = factor(Surgeon, levels = Surgeon)) %>%
  ggplot(aes(x = Year, y = FluoroscopyIndex)) +
  geom_point() +
  geom_line() +
  facet_wrap(vars(Surgeon)) +
  labs(x = NULL, y = "Median der Durchleuchtungsdauer [s]")
```

## Hitparade der Implanteure
```{r greatest_surgeons}
raw_data %>% count(Surgeon) %>% arrange(n) %>% top_n(10, n) %>% mutate(Surgeon = factor(Surgeon, levels = Surgeon)) %>%
  ggplot(aes(x = Surgeon, y = n)) +
  geom_col() +
  coord_flip() +
  labs(x = NULL, y = "Gesamtzahl Katheter")
```

## Hitparade der Assistenten
Einsame Spitze... Romana Ziegler!

```{r greatest_assistants}
raw_data %>% count(Assistant) %>% arrange(n) %>% top_n(10, n) %>% mutate(Assistant = factor(Assistant, levels = Assistant)) %>%
  ggplot(aes(x = Assistant, y = n)) +
  geom_col() +
  coord_flip() +
  labs(x = NULL, y = "Gesamtzahl Katheter")
```