01_table1_demographics.Rmd

---
title: "Demographics for table 1"
author: "Andrew McDavid"
date: "`r Sys.Date()`"
output: 
  github_document: default
  html_document: default
---

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

## Load data

```{r}
subject = read_csv('data/subject_covariates.csv') %>% mutate(ispreterm = preterm_weeks >= 0)

hospital_humilk = read_csv('data/milk_hospital.csv') %>% left_join(subject)
nabx_polish = read_csv('data/antibiotic_exposure.csv')

# calculate summary statistics by term
calc_stat = function(tab, var, value){
  tab = tab %>% group_by(ispreterm)
  varq = var
  # continuous, mean SD
  if(missing(value)) res = tab %>% dplyr::summarize(mean = mean(!!varq), sd = sd(!!varq),  min = min(!!varq), max = max(!!varq), iqr = IQR(!!varq))
  # binary number / proportion
  else res = tab %>% dplyr::summarize(n = sum(!!varq == value), prop = mean(!!varq==value))
  res
}
```

## Any Milk (in hospital)

```{r}
calc_stat(hospital_humilk, quo(`Any Human Milk Perinatal`), TRUE)

```

## Days to discharge

```{r}
days_discharge = read_csv('data/subject_timeline.csv') %>% group_by(Subject) %>% filter(`Sequence Num` %in% c(1, 7)) %>% arrange(cga) %>% summarize(days_to_discharge = DOL[2] - DOL[1])
days_discharge %>% left_join(subject) %>% calc_stat(quo(days_to_discharge))
```


## Milk after discharge

```{r}
milk = read_csv('data/milk_survey.csv')
milk = milk %>% rename(milk_months = `For how many months did baby receive breast milk for more than half of feedings`) 
milk = milk %>% mutate(milk_months = case_when(is.na(milk_months)~ '0', milk_months == 'Less than 1' ~ '0.5', TRUE ~ milk_months), milk_months = as.numeric(milk_months))

milk_sum = milk %>% group_by(Subject) %>% dplyr::summarize(months_surveyed = n()*3, milk_months = mean(milk_months/3, na.rm = TRUE)*12, any_milk_quarter = mean(`Did baby receive breast milk since last visit`=='Yes', na.rm = TRUE)*4)

write_csv(milk_sum, path = 'intermediates/milk_subject.csv')
```

These values were reported to the study by the infants' caregiver in a quarterly survey

### Months of > 50% human milk

```{r}
milk_sub = left_join(milk_sum, subject)
calc_stat(milk_sub, quo(milk_months))
```

Months that the child recieved > 50% human milk (normalized to one year of follow up)

### Quarters of milk

```{r}
calc_stat(milk_sub, quo(any_milk_quarter))
```
Number of 90-day periods in which any milk was recieved.

```{r}
#latex(summary( ispreterm ~ milk_months + any_milk_quarter, data = milk_sum, method = 'reverse', test = TRUE, prmsd = TRUE, continuous = 3))

ggplot(milk_sub, aes(x = preterm_weeks, y = milk_months)) + geom_point() + geom_smooth()

ggplot(milk_sub, aes(x = preterm_weeks, y = any_milk_quarter)) + geom_point() + geom_smooth()
```

Dose-response between milk received and gestational age.

## Antibiotics

```{r}
nabx_polish = left_join(nabx_polish, subject)
calc_stat(filter(nabx_polish, !discharge), quo(`Number of systemic antibiotic`)) 
```

Days of antibiotics (IV, in patient)

```{r}
calc_stat(filter(nabx_polish, discharge), quo(`Number of systemic antibiotic`)) 
```

Courses of antibiotics reported by care giver (likely oral).