In spsanderson/LICHospitalR: Hospital Data Analysis Workflow Tools

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  warning = FALSE,
  message = FALSE,
  fig.width = 8,
  fig.height = 4.5,
  fig.align = 'center',
  out.width = '95%',
  dpi = 100
)

LICHospitalR: A toolkit for working with hospital data in R

Working with Coded Consults

Getting the data

In this vignette I will discuss how to obtain data on coded consults from DSS. The coded consults come from the actual consults coded by the HIM department.

First to get started, load in the LICHospitalR library package into your R session:

library(LICHospitalR)

Get data by automated query

Once the package is loaded in, you can then get started getting data from DSS and analyzing it. The first piece of code that you will use in order to get data from DSS is the function coded_consults_query(). This function relies on a connection to DSS using the db_connect() function that makes and odbc() call out to DSS. This requires that the end user be able to make such a connection from their local desktop machine. This is typically done by having a connection file on your machine. You would normally find a connection file in the sample path here: C:\Users\bh_number_here\Documents\My Data Sources. If you are having difficulties please file a help-desk ticket with Information Services.

Now lets see an example of getting the data, we will then glimpse() the data in order to see what fields are returned and what their types are. This will require us to load the dplyr library as well.

library(dplyr)

coded_consults_query(.resp_pty = "013789") %>%
  glimpse()

You will notice that the .resp_pty was filled in with a providers ID Number. You can obtain a providers id number by using the function get_provider_id_num(), by either providing the .name parameter, or by letting all results return in alpha order. This is how you would get data using the automated query coded_consults_query().

Get data by import

If you do not use that function then you must import data into your R session with the data provided. You will ultimately need to get this from DSS.

Seasonal Time Series Diagnostics

Many times we want to see the variation by time for a great deal of things, consults are no different. That is why I built the coded_consults_seasonal_diagnostics() function. This function takes a few arguments to it. Namely .data, .date_col, and .value_col. This function was designed to work directly with the coded_consults_query() function via the pipe operator together or by take a data table from import.

Lets take a look at an example:

library(timetk)
library(dplyr)

coded_consults_query(.resp_pty = "013789") %>%
  coded_consults_seasonal_diagnositcs_plt(
    .date_col    = dsch_date
    , .value_col = record_flag
  )

Top Attending Providers

Getting Data

Now sometimes we want to see who has consulted someone the most. We can accomplish this in a single line of code using the coded_consults_top_providers_tbl() function. This function takes in four arguments and is meant to work with the pipe operator so that the .data function is not necessary to fill out. The necessary arguments after data is piped in are .top_n, .attending_col, and .consultant_col. All of these columns are explicitly provided from the coded_consults_query() function therefore helping to make a workflow efficient and tidy.

The function returns a tibble that can be analyzed by other means or used in conjuntion with other coded functions. Lets take a look.

library(dplyr)

coded_consults_query(.resp_pty = "013789") %>%
  coded_consults_top_providers_tbl(
    .top_n = 10
    , .attending_col = attending_md
    , .consultant_col = consultant
  )

Plotting Data

As we can see from the resulting code, the top 10 Attending providers ranked in order by how many times they have consulted the provider in question is returned, where n is the number of times the consultant was coded as having been consulted by the attending.

We can extend this workflow further by plotting this same information out. The only parameter this function takes is .data which can be piped in as we will see.

library(dplyr)
library(ggplot2)

coded_consults_query(.resp_pty = "013789") %>%
  coded_consults_top_providers_tbl(
    .top_n = 10
    , .attending_col = attending_md
    , .consultant_col = consultant
  ) %>%
  coded_consults_top_plt()