Skip to contents

Compute entry to Intensive Care Unit

icu_entry.Rd

Determines whether an encounter has entered Intensive Care Unit (ICU) during hospital stay using CIHI Discharge Abstract Database (DAD) fields.

Usage

icu_entry(
  cohort,
  ipscu,
  as_outcome = FALSE,
  exclude_cutoff = 0,
  entry_since_cutoff = c(24, 48, 72)
)

Arguments

cohort

(data.table, data.frame)
Cohort table with all relevant encounters of interest, where each row corresponds to a single encounter. Must contain the following columns:

  • genc_id (integer): GEMINI encounter ID

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

ipscu

(data.table, data.frame)
Table equivalent to the ipscu table defined in the GEMINI Data Repository Dictionary. Table must contain fields: GEMINI Encounter ID (genc_id), SCU admission time (scu_admit_date_time), and SCU number (scu_unit_number).

as_outcome

(logical)
Whether ICU admission is as a clinical outcome or not. Default to FALSE. When set to TRUE, records with direct ICU admission before admitted to inpatient care (IP admission) (i.e. scu_admit_date_time <= admission_date_time) are excluded.

exclude_cutoff

(integer)
The number of hours to add to the IP admission time to establish a cutoff time for excluding records from being identified as ICU admissions. This parameter is only relevant when as_outcome=TRUE. For example, when exclude_cutoff = 12, records with ICU entry time scu_admit_date_time <= admission_date_time + 12 hours are excluded from being identified as ICU admissions. Default is exclude_cutoff = 0, where any ICU entries before IP admission time are excluded.

entry_since_cutoff

(integer, vector)
Time window of ICU entry since IP admission (or since x hours post IP admission when user specifies exclude_cutoff), in hours. This parameter, together with exclude_cutoff specifies the time interval during which ICU admissions are identified. By default, exclude_cutoff = 0 and entry_since_cutoff = c(24, 48, 72), function calculates ICU entry within 24, 48 and 72 hours since IP admission. Users can specify different values for the two parameters to customize the time interval of interest. For example, when exclude_cutoff = 24 and entry_since_cutoff = 48, function determines whether a patient was admitted to ICU between 24 to 72 (24+48) hours post IP admission (i.e. time interval (24, 72] hours)).

Value

(data.table)
By default, for each encounter in input "cohort" returns the corresponding derived boolean (TRUE/FALSE) fields "icu_entry_derived", "icu_entry_in_24hr_derived", "icu_entry_in_48hr_derived" and "icu_entry_in_72hr_derived". If user specified time window x hour is used, field "icu_entry_in_xhr_derived" is computed in addition to "icu_entry_derived".

Details

This function uses DAD fields Admission date-time (Group 04 Fields 01/02), and SCU Admit date-time (Group 13, Field 03/04) to derive boolean fields indicating ICU entries at any time during hospital stay, and within specified time window since hospital admission.

By definition in DAD (Group 13, Field 02), SCUs include ICUs and Step-Down Units. Step-Down Units are not considered as ICUs.

Therefore, this function excludes below CIHI defined Step-Down Units numbers from calculation:

  • 90 : Step-Down Medical Unit

  • 93 : Combined Medical/Surgical Step-Down Unit

  • 95 : Step-Down Surgical Unit

  • 99 : No SCU

Please refer to the CIHI DAD abstracting manual for more details.

Note

: By design, this function will not return any NA values for icu_entry, unless the flag is time sensitive (i.e., if as_outcome is TRUE, or when deriving ICU admission within a certain time window/after a cut-off). For time sensitive flags, any genc_ids with an ICU entry but no valid ICU admission date-time will be returned as NA. If a genc_id has at least one valid ICU admission date-time, all valid entries will be included in the calculation. Note that a lot of invalid ICU admission date-times contain date information only. Users may choose to impute missing timestamps prior to running this function.

When one tries to left-join the output of this function to another table, make sure the list of encounters aligns in both tables. As there are 2 dependent parameters (exclude_cutoff, entry_since_cutoff) determining the time interval during which ICU admissions will be identified, please read function documentation carefully when specifying a non-default value to these parameters to ensure the identification time interval is as intented. See examples below for common use cases.

Examples

## ICU admission within the first 24 hours since IP admission (i.e. you are interested in knowing % of encounters admitted to ICU):
if (FALSE) { # \dontrun{
 icu_entry (cohort, ipscu, as_outcome=FALSE, entry_since_cutoff=24)
} # }

## ICU admission within the first 24 hours since IP admission, as a clinical outcome
## excluding records with ICU entries prior to IP admission:
if (FALSE) { # \dontrun{
 icu_entry (cohort, ipscu, as_outcome=TRUE, entry_since_cutoff=24)
} # }

## ICU admission within the first 72 hours since IP admission, as a clinical outcome
## excluding records with ICU entries prior to the first 24 hours of IP admission (i.e. you are interested in knowing
## patients who were admitted to ICU between the interval of (24, 72] hours since IP admission):
if (FALSE) { # \dontrun{
 icu_entry (cohort, ipscu, as_outcome=TRUE, exclude_cutoff=24, entry_since_cutoff=48) # Note: entry_since_cutoff=48 (instead of 72) because 24+48=72
} # }