Closed acortizmd closed 1 month ago
1) we need to discuss this further, but I have wanted to keep AMA as its own category, but we should talk as a group about that. I know that for 90% of projects they will get lumped into "Home" (that's where i think it should be lumped into), but there may be times when doing studies on less severe things that would warrant being able to identify those patients. 2) agreed... should be Home 3) I think this speaks to need an admission category as well. not sure we have talked about that yet. I would argue that having the admission category be "Deceased Organ Donation" and Dispo be "Dead" would be appropriate. It would also help us filter out these admission because MOST of the time we won't want these admission at all 4) good question.... I think this brings up another issue.... technically the entire admission should be considered in these databases and the "hospitalization_id" (previously "encounter_id") should represent the whole hospitalization from start to finish. Technically no one should have a disposition of observation... it should technically be the dispo of the observation admission should be the dispo for both of those encounters (I'm assuming they are separate encounters based on the data). @08wparker does that make sense
@ingra107 I agree with 4. I am assume these are scenarios where the patient goes from the ED to an observation unit. ideally we would join the ED visit and an observation record into one hospitalization_id
if the patient never left the hospital
@ingra107 @08wparker As Nick notes in #3 having an admission category will be helpful. ('Hospital Transfer', 'ED Admit', 'Direct Admission', 'Elective Surgery'). I think this will be important; for example for any process 'early in the course of critical illness' we may want to exclude transfer patients. Certainly that would probably be the case for prone positioning and early ventilator settings...?
so @chochbe1 the idea is to make admission_type_name
and admission_type_category
and add it to the hospitalization
table?
@08wparker ... yes. I think that will be best. you can see how I do dispo and admit type in my logic below
if(!exists("admit_source_trib")){
admit_source_trib <- tribble(
~enc_admit_source, ~enc_admit_source_cat,
"Emergency Department", 1,
"Home/Work", 1,
"Skilled Nursing, Intermediate Care, or Assisted Living Facility", 1,
"Admission From Home", 1,
"External ED to Hospital", 1,
"Non-Healthcare Facility", 1,
"Admission From Skilled Nursing Facility, Intermediate Care, or Assisted Living Facility", 1,
"Non-HealthEast Emergency Department", 1,
"Court/Law Enforcement", 1,
"Admission From Psych, Substance Abuse, or Behavioral Rehab Facility", 1,
"Normal Delivery", 1,
"Hospice", 1,
"Transfer from Critical Access Hospital", 1,
"Premature Delivery", 1,
"Extramural Birth", 1,
"Born Inside this Hospital", 1,
"Trans from Hospice", 1,
"Born Outside this Hospital", 1,
"Transfer from Another Home Health Agency", 1,
"Transfer from a Designated Disaster Alternative Care Site (ACS)", 1,
"Admission From Clinic", 2,
"Physician or Clinic Office", 2,
"Prescheduled Admission/Surgery", 2,
"Transfer From Hosp Inpt Within Same Facility", 2,
"Discharge/Readmit (Same Hospital)", 2,
"Ambulatory Surgery Center", 2,
"Hospital to Hospital", 3,
"Hospital (Acute Care Facility)", 3,
"Other Type of Health Care Facility", 3,
"Other Health Care Facility To Hospital", 3,
NA, 4,
"Information Not Available", 4,
"HMO Referral", 4
)
# factor admission source
admit_source_trib <- admit_source_trib |>
mutate(enc_admit_source_cat = factor(
enc_admit_source_cat,
levels = 1:4,
labels = c("ED" ,
"Direct" ,
"OSH" ,
"Other")
))
}
### Getting discharge categories set
if(!exists("discharge_disp_trib")){
discharge_disp_trib <- tribble(
~enc_disch_disp, ~enc_discharge_disp_cat,
"Home or Self Care", 1,
"Home-Health Care Svc", 1,
"Home IV Drug Therapy", 1,
"Home Care - Other", 1,
"Home Care - HealthEast", 1,
"CORE Clinic", 1,
"Home with Planned Hospital IP Readmission", 1,
"Planned Acute Care IP Readmission - Home or Self Care", 1,
"Planned Acute Care IP Readmission - Home Care - Other", 1,
"Home-Health Care Svc with Planned Hospital IP Readmission", 1,
"IRTS - Intensive Residential Treatment Program", 1,
"Adoptive Parent / Foster Care", 1,
"IRTS", 1,
"Assisted Living", 2,
"Group Home", 3,
"Jail/Police Custody", 3,
"MN Correctional Facility", 3,
"Shelter", 3,
"Halfway House", 3,
"Court/Law Enforcement", 3,
"Jail/Police Custody with Planned Hospital IP Readmission", 3,
"Acute Rehab Facility", 4,
"Rehab Facility - Inpatient", 4,
"Acute Rehab Facility with Planned Hospital IP Readmission", 4,
"Planned Acute Care IP Readmission - Rehab Facility - Inpatient", 4,
"Skilled Nursing Facility", 5,
"Medicaid Only Certified Nursing Facility", 5,
"Nursing Facility with Planned Hospital IP Readmission", 5,
"Skilled Nursing Facility with Planned IP Hospital Readmission", 5,
"Planned Acute Care IP Readmission - Skilled Nursing Facility", 5,
"Left Without Being Seen", 6,
"Left Against Medical Advice", 6,
"Psych Facility - HealthEast", 7,
"Psychiatric Hospital", 7,
"Psych Facility - Other", 7,
"Substance Abuse Treatment Program - Inpatient/Not Part of Acute Care Facility", 7,
"CD Facility - HealthEast", 7,
"Psychiatric Hospital with Planned Hospital IP Readmission", 7,
"CD Facility - Other", 7,
"Planned Acute Care IP Readmission - Psych Facility - Other", 7,
"Mental Health Jud Commit Anoka", 7,
"Crisis Residence", 7,
"Short Term Hospital", 8,
"Another Health Care Institution Not Defined", 8,
"Acute Hospital - Non-HealthEast", 8,
"Intermediate Care Facility", 8,
"Cancer Center or Children's Hospital", 8,
"Acute Hospital - HealthEast", 8,
"Another Health Care Institution with Planned Hospital IP Readmission", 8,
"Short Term Hospital with Planned Hospital IP Readmission", 8,
"Swing Bed", 8,
"Disaster Alternate Care", 8,
"Disch/Trans to Designated Disaster Alternative Site", 8,
"Planned Acute Care IP Readmission - Acute Hospital", 8,
"Critical Access Hospital", 8,
"Critical Access Hospital with Planned Hospital IP Readmission", 8,
"Intermediate Care Facility with Planned Hospital IP Readmission", 8,
"Federal Hospital", 8,
"Intermediate Care Facility - State Designated", 8,
"Federal Hospital with Planned Hospital IP Readmission", 8,
"Planned Acute Care IP Readmission - Federal Hospital", 8,
"Cancer Center or Children's Hosp with Planned IP Hospital Readmission", 8,
"Swing Bed with Planned Hospital IP Readmission", 8,
"Planned Acute Care IP Readmission - LTACH", 9,
"Long Term Acute Care", 9,
"LTACH - Bethesda", 9,
"LTACH - Non-HealthEast", 9,
"Long Term Acute Care with Planned Hospital IP Readmission", 9,
"Hospice/Home", 10,
"Hospice - Home Care", 10,
"Hospice - Medical Facility", 10,
"Hospice/Medical Facility", 10,
"Expired in Hospice", 11,
"Expired", 11,
"Expired in Medical Facility", 11,
"Expired - Place Unknown", 11,
"Expired at Home", 11,
"ED Dismiss - Never Arrived", 12,
"FV Cutover", 12,
"ED Dismiss - Diverted Elsewhere", 12,
"Level III Nursery", 12,
"ED Dismiss - Code 9", 12,
NA, 12,
"Admitted as an Inpatient", 13,
"Still a Patient", 13
)
# factor discharge destination
discharge_disp_trib <- discharge_disp_trib |>
mutate(enc_discharge_disp_cat = factor(
enc_discharge_disp_cat,
levels = 1:13,
labels = c(
"Home",
"Assist Living",
"Group Home/Shelter/Jail" ,
"Acute Inpt Rehab" ,
"SNF" ,
"Against Med Advice (AMA)" ,
"ChemDep/Psych" ,
"Another Hosp" ,
"LTACH" ,
"Hospice" ,
"Expired" ,
"missing",
"Still Admitted"
)
))
}
looks like your admission_type_category
proposal is ("ED" , "Direct" , "OSH" , "Other") which seems like a good start to me.
@ingra107 looks like this enc_discharge_disp_cat
field has several more levels than the CLIF discharge_category
which has levels
"Home"
"Hospice"
"Acute care hospital"
"Still Admitted"
"Inpatient Rehab Facility"
"Skilled Nursing Facility"
"Long Term Care Hospital"
"Psychiatric Hospital"
"Dead"
"Other"
Do you want to propose a modification to our discharge_category
CDE as the point of contact for the hospitalization
table? Remember we will still have discharge_name
.
Also I would recommend storing as string VARCHAR data types (instead of that specialty R factor data type with levels/labels)
re VARCHAR: totally agree... you can tell this was the first thing i did when i translated my stata to R code lol.... very newbie mistake to factor something so quickly
re discharge_category
: These are the ones I use for studies, I wasn't sure if ppl had put a lot of thought in to the discharge_category
before and thus didn't want to step on toes. I think finding that sweet spot of keeping some granularity without getting too into the weeds is ideal and I think these categories suffice. I think adding "Against_medical_advice" would be appropriate. The only other I have thought about is whether we have a Chemical_dependecy_detox category as well.
I think this makes the most sense to me
1 Home
2 SNF
3 Expired
4 Acute Inpt Rehab
5 Hospice
6 LTACH
7 Another Hosp
8 Group Home/Shelter/Jail
9 ChemDep/Psych
10 Against Med Advice (AMA)
11 Assist Living
12 Still Admitted
13 missing/other
Notes:
discharge_category == Hospice
@kaveriC - add UMMN mapping to CLIF repository & data dictionary
Good afternoon! I am currently working on disposition_category for the Penn data transformation. I have a few questions perhaps somebody can help clear up.