R_coding_injury_tbi.Rmd

---
title: "R Coding for State Injury Report"
author: "Mamadou Ndiaye - Washington State Department of Health"
date: "August 30, 2016"
output: html_document
subtitle: 2014 TBI Death in Washington State
abstract: 'Steps to provide count of traumatic brain injury (TBI) deaths using R:
  (1)Define the injury with regular expressions, (2)Specify the columns with the icd-10
  codes of interest, (3) Use a predefined function to match the regular expression
  patterns with the icd10 codes , (4) Create a new variable assigning ''1'' to observation
  matching the defined regular expression in any of the selected field, assigning
  ''0'' otherwise.(5) Make the tables of counts after the new variable for the defined
  injury is created.'
---  



```{r, echo=F,message=FALSE, warning=FALSE}  

options(scipen = 3)
knitr::opts_chunk$set(
  eval = FALSE,
  comment = NA,
  error   = FALSE,
  tidy    = FALSE,
  cache   = F
  )



```
## Introduction  

### R libraries and custom functions  

  The R libraries **dplyr**, **tidyr** and **Hmisc** were used in this analysis. They can be loaded with the following codes (assuming they are already installed with the function *"install.packages"*):
  
```{r}  

library(dplyr)
library(tidyr)
library(Hmisc)   

```  

The functions *"create.diag"* and *"create_cond_diag"* can be loaded from my github page with:

```{r}  

source("https://raw.githubusercontent.com/epinotes/InjuryEpi/master/create_diag")
source("https://raw.githubusercontent.com/epinotes/InjuryEpi/master/create_cond_diag")
source("https://raw.githubusercontent.com/epinotes/InjuryEpi/master/onclip")

```  
### The Death dataset

  The variables used in the death dataset (**dea2014f**) were:
  
  - The diagnosis code columns: *underly*, *mltcse1* to *mltcse20*   
  
  - The age group column (11 age groups): *agegrp11*  
  
  - The sex column: *sex*  


## Injury Definitions with Regular Expressions  


### All injuries    

  Definition from CDC: The ICD-10 codes __V01–Y36, Y85–Y87, Y89, *U01–*U03__ in the underlying cause of death field. They are re-written as regular expressions in the following that save them in the R object **.cdc_inj**.

 
```{r}

.cdc_inj <- "^V[0-9][1-9]|^[W-X][0-9][0-9]|^Y[0-2][0-9]|^Y3[0-6]|^Y8[5-7|9]|U0[1-3]"

```  

  Select the column of interest (the underlying cause of death column):

```{r}

colx <- grep("^underly$", names(dea2014f))

```
  Create the variable for any injury, *cdc_inj*:  
  
  
```{r}

dea2014f$cdc_inj <- new.diag_m(data=dea2014f, expr=.cdc_inj, colvec= colx)

```   
 Check the results with the function *describe()* (get information on the functiion by running *?Hmisc::describe*)

```{r}

describe(dea2014f$cdc_inj)

```

  Show the requested CDC table of count of injury deaths by the 11 age groups (assigned to the object *dea2014f_inj*) :
  
```{r}  

dea2014f_inj <- dea2014f %>% filter(cdc_inj == 1) %>% group_by(agegrp11) %>% summarise(deaths = n()) %>% na.omit()

print(dea2014f_inj)

```
  we can copy the table in a clipboard then paste it on the CDC injury table templates:  
  
  
```{r}

onclip(dea2014f_inj)

```

### TBI Injuries    

#### 1. All TBI Injuries


  Once a case is defined as injury (as previously done); it is TBI if any field of the multiple cause of death file has one of these ICD-10 code:

S01.0–S01.9, S02.0, S02.1, S02.3, S02.7–S02.9, S04.0, S06.0–S06.9, S07.0, S07.1, S07.8, S07.9, S09.7–S09.9, T01.0*, T02.0*, T04.0*, T06.0*, T90.1, T90.2, T90.4, T90.5, T90.8, T90.9

  or re-written as a regular expression:
  
  
```{r}
.cdc_tbi <- "S01[0-9]|S02[0|1|3|7|8|9]|S040|S06[0-9]|S07[0|1|8|9]|S09[7-9]|T0[1|2|4|6]0|T90[1|2|4|5|8|9]"

```

  Then we select the underlying cause column ("underly") and the 20 multiple cause columns ("mltcse1", "mltcse20"):
  
  
```{r}

  col_m <- grep("underly$|mltcse", names(dea2014f), value = F)

```
  Create the TBI variable
  
```{r}

dea2014f <- dea2014f %>% mutate(cdc_tbi = create_cond_diag(., expr=.cdc_tbi, colvec= col_m, cond.var = cdc_inj))

```
  Check the results:  
  
  
```{r}  

describe(dea2014f$cdc_tbi)

```

  Create a table for count of TBI deaths by the 11 age groups.

```{r}

dea2014f_tbi11 <- dea2014f %>% filter(!is.na(agegrp11) & cdc_tbi == 1) %>% mutate(agegrp11 = as.numeric(agegrp11)) %>% group_by(agegrp11) %>% summarise(deaths = n()) %>% right_join(data.frame(agegrp11 = 1:11)) %>% complete(agegrp11 = full_seq(agegrp11, period = 1),  fill = list(deaths = 0))


```   

  Print the table or copy it on the clipboard 

```{r}  

print(dea2014f_tbi11)

onclip(dea2014f_tbi11[,2])

``` 

  Create similar tables restrited to male cases:

```{r}  

dea2014f_tbi11_m <- dea2014f %>% filter(!is.na(agegrp11) & cdc_tbi == 1 & sex == "M") %>% mutate(agegrp11 = as.numeric(agegrp11)) %>% group_by(agegrp11) %>% summarise(deaths = n()) %>% right_join(data.frame(agegrp11 = 1:11)) %>% complete(agegrp11 = full_seq(agegrp11, period = 1),  fill = list(deaths = 0)) 


```


  Then restricted to female cases
  
  
```{r}  

dea2014f_tbi11_f <- dea2014f %>% filter(!is.na(agegrp11) & cdc_tbi == 1 & sex == "F") %>% mutate(agegrp11 = as.numeric(agegrp11)) %>% group_by(agegrp11) %>% summarise(deaths = n()) %>% right_join(data.frame(agegrp11 = 1:11)) %>% complete(agegrp11 = full_seq(agegrp11, period = 1),  fill = list(deaths = 0))

```

  Create a subset (**dea2014_tbi**) of the death file with all the TBI deaths. It will be used to build the other tables of specific injuries with TBI:  
  
```{r}  

sel_inj <- c(1, 3, 51, 52, grep("underly$|ecode|mltcse", names(dea2014f ), value = F), 169:180)

dea2014_tbi <- dea2014f[sel_inj] %>% filter(cdc_tbi == 1)


```  


#### 2. TBI Injuries and Motor Vehicle Traffic Crashes