The Visit Information Visualization for the Emergency Department (VIVED) tool is a HTML dashboard implemented in R with the Shiny package to examine patient flow through an emergency department (ED) or urgent care center (UCC). Users can filter data to their specifications including by date, day of the week, time of the day, patient disposition, and provider. Users can then plot metrics such as number of visits per day, fraction of patients admitted per day, length of stay, boarding time, and ED census.
- Install R version 3.3.1 or greater
- Install required R packages
- Open an R terminal and install the following packages with the command install.packages("package_name")
- shiny
- ggplot2
- reshape
- RColorBrewer
- DT
- Open an R terminal and install the following packages with the command install.packages("package_name")
- Download and uncompress VIVED package
- Modify the disp_admit.txt file in the VIVED package.
- The disp_admit.txt file contains a list of dispositions values that result in admission to the relevant inpatient facility.
- Users should modify this file so that each unique disposition type resulting in admission is listed for each row. For example
- "Admit"
- "Telemetry"
- "ICU"
- If no dispositions result in admission or dispositions are unknown, leave this file empty.
- After installation, open an R terminal.
- Set the working directory to the decompressed VIVED package directory.
- setwd("/path/to/files/vived/")
- Type the following two commands:
- library(shiny)
- runApp("Vived.r")
- The app should open in your default web browser.
The input file containing ED or UCC visit information must be in comma-separated values (CSV) format with one header row and each remaining row representing a unique ED visit. Columns, along with their header labels, are described in the following table:
| Header label | Required | Date type | Description |
|---|---|---|---|
| PatientID | Yes | Integer | Patient identifier |
| TimeIn | Yes | Timestamp | Time at which the patient enters the ED or UCC |
| TimeOut | Yes | Timestamp | Time at which the patient exists the ED or UCC |
| Station | No | Integer | ED or UCC identifier. Used in combination with the PatientID to identify unique patients. |
| TimeTriage | No | Timestamp | Time at which the patient is assigned an acuity score |
| TimeSeen | No | Timestamp | Time at which the patient is first seen by a physician |
| TimeDisposition | No | Timestamp | Time at which the patient is assigned a disposition |
| Acuity | No | Integer | Acuity score (e.g. Emergency Severity Index) |
| Disposition | No | String | Disposition of the patient |
| MdAssigned | No | String | Name or identifier of the physician assigned to the patient during the visit |
| AgeAtVisit | No | Integer | Age of the patient on the day of the visit in years |
All timestamps must be in the form "YYYY-MM-DD HH:MM:SS". Columns do not have to be in order and additional columns may be included, but the column headers must be exact, including capitalization. Columns that are not required do not have to be included in the input file. However, visualizations involving non-essential data elements will not be available if the respective columns are not included.
Users can select an input file (format described above) on their local hard drive using the "Browse" button.
This dropdown allows users to select a start and end date for filtering the dataset. Users may type in the dates (format is "YYYY-MM-DD") or click on the boxes and select dates in a dropdown calendar. The default range is set to the dates of of the time of arrival of the first and last visit in the dataset.
This dropdown menu allows users to select a metric to display. Currently implemented metrics are as follows:
| Metric | Unit | Description |
|---|---|---|
| Number of visits per day | Day | Calculates the number of visits per day based on patient arrival time |
| Number per day boarding > 4 hours | Day | Number of patients per day who boarded (defined as the time between the disposition time and the patient departure from the emergency department) for more than 4 hours. Patient visits are assigned to each day by the patient arrival time. |
| Number per day returned < 72 hours | Day | Number of visits per day that correspond to a patient who visited the same ED again within 6 and 72 hours. A minimum threshold of 6 hours is included to reduce data input errors. Any filters applied to this metric only affect the initial (index) visit, not the return visit. |
| Number per day returned-admitted < 72 hours | Day | Number of visits per day that correspond to a patient who visited the same ED again within 6 and 72 hours and was admitted on the second visit. A minimum threshold of 6 hours is included to reduce data input errors. Any filters applied to this metric only affect the initial (index) visit, not the return visit. |
| Percent of patients per day | Day percent | Calculates the fraction of visits per day that correspond to the user-specified filters for disposition, provider, hour of the day, or day of the week. The visit is associated with each day by the patient arrival time. If default filter options are selected, this metric will always return 100%. The denominator corresponds to all valid visits. |
| Percent per day boarding > 4 hours | Day percent | Percentage of patients per day who boarded (defined as the time between the disposition time and the patient departure from the emergency department) for more than 4 hours. Patient visits are assigned to each day by the patient arrival time. The denominator corresponds only to those visits as specified by the user filters. |
| Percent per day returned < 72 hours | Day percent | Percentage of visits per day that correspond to a patient who visited the same ED again within 6 and 72 hours. A minimum threshold of 6 hours is included to reduce data input errors. Any filters applied to this metric only affect the initial (index) visit, not the return visit, and the denominator corresponds only to those visits as specified by the user filters. |
| Percent per day returned-admitted < 72 hours | Day percent | Percentage of visits per day that correspond to a patient who visited the same ED within 6 and 72 hours previously and was admitted on the second visit. A minimum threshold of 6 hours is included to reduce data input errors. Any filters applied to this metric only affect the initial (index) visit, not the return visit, and the denominator corresponds only to those visits as specified by the user filters. |
| Length of stay (min) | Visit | Calculates the length of stay of a patient in minutes as the difference between the time out and time in |
| Door to triage (min) | Visit | The time in minutes between the patient arrival time and the first time the acuity score is set for the patient during the visit. |
| Door to doc (min) | Visit | The time in minutes between the patient arrival time and the first time the patient is assigned to a physician. |
| Door to disp (min) | Visit | The time in minutes between the patient arrival time and the first time the disposition is set for the patient visit. |
| Boarding time: disposition-out (min) | Visit | Calculates the boarding time as the difference between the disposition time and time out. |
| Census per minute | Minute | Calculates the number of patients in the ED for each minute of the day as assessed by the number of visit intervals overlapping that minute. |
| Census per minute by acuity | Minute | Calculates the number of patients in the ED for each minute of the day as assessed by the number of visit intervals overlapping that minute and subsets the ED census by patient acuity. |
| Census per minute by age | Minute | Calculates the number of patients in the ED for each minute of the day as assessed by the number of visit intervals overlapping that minute and subsets the ED census by the decade of the patient age at the time of the visit. |
| Census per minute by disposition | Minute | Calculates the number of patients in the ED for each minute of the day as assessed by the number of visit intervals overlapping that minute and subsets the ED census by patient disposition. |
| Census per minute by progression | Minute | Calculates the number of patients in the ED for each minute of the day as assessed by the number of visit intervals overlapping that minute and subsets the ED census by patient progression. ??? Progression indicates the stage of the patient visits. The first stage is between the visit time in and the time of triage. The second state is the time between triage and being seen by the doctor. The third stage is between being seen by the doctor and having the disposition set. The fourth stage is between having a disposition set and leaving the ED. |
| Census per minute by status | Minute | Calculates the number of patients in the ED for each minute of the day as assessed by the number of visit intervals overlapping that minute and subsets the ED census by patient status. ??? Patient status is split into 4 categories: patients who will be home, patients who will be admitted but their disposition has not yet been set, and patients who will be admitted and their disposition has been set who have been boarding for at most 4 hours and more than 4 hours. |
This dropdown allows users to select an aggregate view to display the metrics described above. The available aggregation views differ by selected metric and will dynamically update depending on the metric selected by the user. The views are described below:
| View | Description |
|---|---|
| Histogram | Plots metric distribution as a block histogram |
| Year | Groups the metrics by year associated with each visit time in and displays the distribution associated with each year as a boxplot for relevant years. |
| Month | Groups the metrics by year and month of the visit time in. The distribution of each group is shown as a boxplot for relevant year-month. |
| Month of the year | Groups the metrics by month of the year of the visit time in. Metrics in different years but the same month will be placed in the same group. The distribution for each group is shown as a boxplot for relevant months. |
| Week | Groups metrics by week of the visit time in where weeks begin on Monday. |
| Day | Groups metrics by the date of the visit time in. If only one metric per date is available (e.g. number of visits per day), the plot is a line plot with points indicating each unique date. If more than one metric per date is available, each unique date's distribution is displayed as a boxplot for relevant dates. |
| Day of the week | Groups metrics by the day of the week of the visit time in. Metrics in different weeks but the same day of the week will be placed in the same group. The distribution for each group is shown as a boxplot for relevant days of the week. |
| Hour of the day | Groups metrics by the hour of the day of the visit time in. Metrics from different days but occurring in the same hour of the day will be placed in the same group. The distribution for each group is shown as a boxplot for relevant hours of the day. |
| Minute by day | Groups metrics by minute of the day and day of the week of the visit time in. Metrics are displayed as a bar plot showing the metric's value for that minute on a particular day. Multiple bar plots are created if more than one day of the week is specified. If the data represents more than one week, then data is averaged for overlapping minutes. |
The views available for each metric by unit type is delineated below:
| Metric unit / View | Histogram | Year | Month | Month of the year | Week | Day | Day of the week | Hour of the day | Minute by day | Acuity | Disposition | Provider |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Day | X | X | X | X | X | X | X | X | ||||
| Day percent | X | X | X | X | X | X | X | X | ||||
| Visit | X | X | X | X | X | X | X | X | X | X | X | |
| Census | X | X | X | X | X | X | X | X | X |
This dropdown allows the user to filter patient visits by the patient disposition assigned during the visit. This dropdown updates dynamically to display all unique dispositions in the full dataset. The default is "All".
This dropdown allows the user to filter patient visits by the assigned acuity (e.g. Emergency Severity Index). This dropdown updates dynamically to display all unique acuity labels in the full dataset. The default is "All".
This dropdown allows the user to filter patient visits by the provider assigned to the patient during the visit. This dropdown updates dynamically to display all unique providers in the full dataset. The default is "All".
These four dropdowns allow the user to filter patient visits within a particular time interval of day. The start time and end time are inclusive of the selected minute. For census metrics, the census is set to missing during these periods and so is not displayed or included in any visualization. For other metrics, patient visits are excluded if the time in associated with the visit lies outside the requested time interval. The default is to include all hours (i.e. start time: "00:00" and end time "23:59").
This group of checkboxes allows the user to select any combination of days of the week to filter patient visits. Any patient visit's time in which lies in a checked day of the week will be retained for analysis and visualization. The default is to include all days of the week.
When checked, this option allows users to display visualizations and tables calculated without visits occurring on the ten annual federal holidays. A list of federal holidays is available on the U.S. Office of Personnel Management website.
Having selected all appropriate filters and options, the user can click the "Run" button to produce the respective visualization and table. Depending on the size of the dataset and the metric selected, the visualization may take as much as several seconds to produce. ED census plots are particularly time consuming when calculated over long ranges of time, so starting with a smaller time interval (e.g. one week) is recommended.
Having produced a plot, the user can click the "Download plot" button to save a PDF of the visualization. A file explorer box will appear asking the user to select a folder in which to save the plot.
Having produced a plot and table, the user can click the "Download table" button to save a CSV file of the respective table. A file explorer box will appear asking the user to select a folder in which to save the table.