Skip to contents

Simulate ipadmdad data

dummy_ipadmdad.Rd

This function creates a dummy dataset with a subset of variables that are contained in the GEMINI "ipadmdad" table (see details in GEMINI Data Repository Dictionary).

The simulated encounter-level variables that are returned by this function are currently: Admission date-time, discharge date-time, age, gender, discharge disposition, transfer to an alternate level of care (ALC), and ALC days. The distribution of these simulated variables roughly mimics the real distribution of each variable observed in the GIM cohort from 2015-2022. Admission date-time is simulated in conjunction with discharge date-time to mimic realistic length of stay. All other variables are simulated independently of each other, i.e., there is no correlation between age, gender, discharge disposition etc. that may exist in the real data. One exception to this is number_of_alc_days, which is only > 0 for entries where alc_service_transfer_flag == TRUE and the length of ALC is capped at the total length of stay.

The function simulates patient populations that differ across hospitals. That is, patient characteristics are simulated separately for each hospital, with a different, randomly drawn distribution mean (i.e., random intercepts). However, the degree of hospital-level variation simulated by this function is arbitrary and does not reflect true differences between hospitals in the real GEMINI dataset.

Usage

dummy_ipadmdad(n = 1000, n_hospitals = 10, time_period = c(2015, 2023))

Arguments

n

(integer)
Total number of encounters (genc_ids) to be simulated.

n_hospitals

(integer)
Number of hospitals to be simulated. Total number of genc_ids will be split up pseudo-randomly between hospitals to ensure roughly equal sample size at each hospital.

time_period

(numeric)
A numeric vector containing the time period, specified as fiscal years (starting in April each year). For example, c(2015, 2019) generates data from 2015-04-01 to 2020-03-31.

Value

(data.frame)
A data.frame object similar to the "ipadmdad" table containing the following fields:

  • genc_id (integer): GEMINI encounter ID

  • hospital_num (integer): Hospital ID

  • admission_date_time (character): Date-time of admission in YYYY-MM-DD HH:MM format

  • discharge_date_time (character): Date-time of discharge in YYYY-MM-DD HH:MM format

  • age (integer): Patient age

  • gender (character): Patient gender (F/M/O for Female/Male/Other)

  • discharge_disposition (integer): All valid categories according to DAD abstracting manual 2022-2023

    • 4: Home with Support/Referral

    • 5: Private Home

    • 8: Cadaveric Donor (does not exist in GEMINI data)

    • 9: Stillbirth (does not exist in GEMINI data)

    • 10: Transfer to Inpatient Care

    • 20: Transfer to ED and Ambulatory Care

    • 30: Transfer to Residential Care

    • 40: Transfer to Group/Supportive Living

    • 90: Transfer to Correctional Facility

    • 61: Absent Without Leave (AWOL)

    • 62: Left Against Medical Advice (LAMA)

    • 65: Did not Return from Pass/Leave

    • 66: Died While on Pass/Leave

    • 67: Suicide out of Facility (does not exist in GEMINI data)

    • 72: Died in Facility

    • 73: Medical Assistance in Dying (MAID)

    • 74: Suicide in Facility

  • alc_service_transfer_flag (character): Variable indicating whether patient was transferred to an alternate level of care (ALC) during their hospital stay. Coding is messy and varies across sites. Possible values are:

    • Missing: NA, ""

    • True: "TRUE"/"true"/"T", "y"/"Y", "1"/"99", "ALC"

    • False: "FALSE"/"false", "N", "0", "non-ALC" Some entries with missing alc_service_transfer_flag can be inferred based on value of number_of_alc_days (see below)

  • number_of_alc_days (integer): Number of days spent in ALC (rounded to nearest integer). If number_of_alc_days = 0, no ALC occurred; if number_of_alc_days > 0, ALC occurred. Note that days spent in ALC should usually be < length of stay. However, due to the fact that ALC days are rounded up, it's possible for number_of_alc_days to be larger than los_days_derived.

Examples

# Simulate 10,000 encounters from 10 hospitals for fiscal years 2018-2020.
ipadmdad <- dummy_ipadmdad(n = 10000, n_hospitals = 10, time_period = c(2018, 2020))