01_create_tables.Rmd

---
title: "SARS-CoV-2 vs Influenza: unpack cleaned data, create basic tables"
author: "Patrick G. Lyons, MD, MSc"
date: "2022-08-26"
output:
  html_document: default
---

This script unpacks cleaned SARS-CoV-2 and Influenza data, consolidates data to one row per encounter, and creates basic comparative tables.

# Load libraries
```{r libraries, include = FALSE}
library(data.table)
library(tidyverse)
library(tidytable)
library(here)
library(tableone)
```

# Load cleaned data
```{r load data}
df <- fread(here("covflu_combined_deidentified.csv"))
```

# Drop variables related to multistate models
## (Code for multistate models is in 03_multistate_models.Rmd)
```{r}
df <-
  df %>%
  select.(
    -c(max_o2_12h, hours_in_hosp, hours_12)
  )

n_distinct.(df$arb_mrn) # 4785
n_distinct.(df$arb_enc) # 4785
dim(df) # 57907, 125
```

# Collapse data to one row per encounter
```{r}
df <-
  df %>%
  distinct.()

n_distinct.(df$arb_mrn) # 4785 - no patients were completely dropped
n_distinct.(df$arb_enc) # 4785
dim(df) # 4785, 125 - one row for every patient encounter
```

# Create LCA characteristics tables
```{r characteristics table}
# characteristics, stratified by virus
print(
  tableone::CreateTableOne(
    vars = c(
      "age_years", 
      "female", 
      "race_nonwhite", 
      "bmi", 
      "min_spo2",
      "min_sbp",
      "max_pulse", 
      "max_resp",
      "min_anc",
      "max_potassium",
      "min_sodium",
      "max_lactate",
      "min_albumin",
      "min_hco3",
      "elix_chf",
      "elix_arrhythmia",
      "elix_chronic_pulmonary_disease",
      "elix_dm_combined",
      "elix_renal_failure",
      "elix_metastatic_cancer",
      "elix_solid_tumor_without_metastasis",
      "elix_coagulopathy"
    ),
    factorVars = c(
      "female",
      "race_nonwhite",
      "elix_chf",
      "elix_arrhythmia",
      "elix_chronic_pulmonary_disease",
      "elix_dm_combined",
      "elix_renal_failure",
      "elix_metastatic_cancer",
      "elix_solid_tumor_without_metastasis",
      "elix_coagulopathy"
    ),
    strata = "virus",
    data = df
  ),
  nonnormal = c(
    "age_years", "bmi", "min_spo2", "max_pulse", "max_resp",
    "min_sbp", "min_sodium", "max_lactate",
    "min_albumin", "min_hco3"
  ),
  quote = TRUE, 
  noSpaces = TRUE
)
```

```{r outcomes table}
# outcomes, stratified by virus
print(
  tableone::CreateTableOne(
    vars = c(
      "los_hosp_hours", 
      "los_icu_hours", 
      "dead_or_hospice", 
      "icu_enc_01", 
      "imv_enc_01",
      "vfd_28"
    ),
    factorVars = c(
      "dead_or_hospice", 
      "icu_enc_01", 
      "imv_enc_01"
    ),
    strata = "virus",
    data = df
  ),
  nonnormal = c("los_hosp_hours", "los_icu_hours", "vfd_28")
)
```