analysis.Rmd

---
title: "Analysis"
author: "Rachel Gold"
date: '2022-07-12'
output: github_document
editor_options: 
  markdown: 
    wrap: 72
---

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

## R Markdown

The purpose of this markdown is to analyze CMS Hospital Survey Deficiencies that we've already filtered down to EMTALA violations involving pregnant patients.

```{r}
library(readxl)
library(dplyr)
library(stringr)
library(writexl)
library(readr)
```


## Data

Reading in violations from 2011-Q3 2022 that we automatically filtered by features of the inspection_text field (0_etl, 1_etl, and 2_etl). These were then manually reviewed by Rachel Gold and Karen Rodriguez for whether they, in fact, pertained to pregnant patients.

```{r}
emtala_pregnant <- read_xlsx("data/manual/confirmed_pregnant.xlsx")
emtala_all <- read_xlsx("data/processed/0_etl_all_emtala_deficiencies.xlsx")

```

## Hospital counts, features

Web: “Our investigation found 389 hospitals spanning 44 states have violated EMTALA statutes while attending to pregnant patients, racking up at least 683 federal violations since 2011.” 
 
Script: “OUR SCRIPPS NEWS INVESTIGATION FOUND IT IS AMONG 389 HOSPITALS (FC) AROUND THE COUNTRY REPONSIBLE FOR NEARLY 700 VIOLATIONS OF THE EMERGENCY MEDICAL TREATMENT AND LABOR ACT – OR EMTALA"

## Violation count

Web: “racking up at least 683 federal violations since 2011.” 
```{r}
facility_count<-n_distinct(emtala_pregnant$facility_id)

violation_count<-n_distinct(emtala_pregnant$key_identifier)

state_count<-n_distinct(emtala_pregnant$state)

print(paste("facility_count:",facility_count,"violation_count:",violation_count,"state_count:",state_count))

emtala_pregnant%>%
  count(state,sort=TRUE)

```

## Date range
Script:“...ALL INVOLVING PREGNANCY EMERGENCIES BETWEEN 2011 AND 2022.” 
 
```{r}
emtala_pregnant%>%
  count(year)
```

## Percentage of EMTALA overall

Web: “Cases involving pregnant women made up about 15% of all EMTALA investigations.” 

```{r}

#number of EMTALA pregnant patient investigations since 2011
count_preg<-n_distinct(emtala_pregnant$EVENT_ID)

#number of EMTALA investigations overall since 2011
count_invest<-n_distinct(emtala_all$EVENT_ID)

#percent of EMTALA investigations involving a pregnant patient since 2011 
print(paste(count_preg,"investigations of violations against pregnant patients, out of",count_invest,"investigations of overall. (",count_preg/count_invest*100,"%)"))

```

## EMTALA violation ranking
Web: “We found the most common EMTALA violation was “failure to provide medical screening examinations.” (medical screening exam = 2406)

```{r}
emtala_pregnant %>%
  count(deficiency_tag,sort=TRUE)

# filter emtala_pregnant for deficiency tag == 2406
count_all_preg <- n_distinct(emtala_pregnant$key_identifier)
count_2406 <- n_distinct((emtala_pregnant %>% filter(deficiency_tag == 2406))$key_identifier)

print(paste(count_2406,"medical screening violations against pregnant patients, out of",count_all_preg,"violations against pregnant patients of overall. (",count_2406/count_all_preg*100,"%)"))
```

## Turnaway count

Rachel Gold and Karen Rodriguez also manually reviewed the EMTALA violations for whether the pregnant patient was actually turned away by the hospital.

Web: "Our investigation found of the cases that were investigated by CMS, at least 241 involved pregnant women going to the ER for care and being turned away."

```{r}
# I added "turnaway" as a column in confirmed_pregnant.xlsx
turnaway <- emtala_pregnant %>% filter(turnaway == TRUE)
n_distinct(turnaway$EVENT_ID)

n_distinct(turnaway$facility_id)
```

## Rural vs. urban

Web: "EMTALA violations involving pregnancy [...] included small hospitals and large ones, in both rural and urban areas."

```{r}
emtala_facilities <- read.csv("data/processed/3_etl_facility_summary.csv")
emtala_facilities %>%
  count(Rural.Status)
```