Background/Introduction
-
Many studies require information on the number of days spent in hospitals during the year. The hospital discharge abstracts data that we receive from Manitoba Health is separation based - a record is only created when an individual is discharged from the hospital. Separation based records may start before the beginning of the year, and/or continue beyond the end of the year. Using the LOS reported on the separations abstract may include days that fall outside of the year. Near the end of the last year of information, days may be lost because people have not yet been discharged from hospital. In order to account for these situations, we can calculate the
In-year Length of Stay (LOS).
For information related to this topic, see the concepts on Hospital Days - How to Measure Length of Stay (LOS), Hospital Overnight Census Counts and Expected Length of Stay (ELOS) in a Personal Care Home (PCH).
Basic Definition
-
To calculate in-year LOS and the corresponding number of patient days, several years of data should be used to capture individuals who were separated after the end of the year or where records were not processed by the end of the year.
Test for the following situations and assign the length of stay accordingly.
A. Admission and Separation fall within the year of interest.
Use the whole stay (DATESEP-DATEADM, or LOS) to calculate in-year days.
B. Hospital stay starts before the beginning of the year and ends during the year
.
Count only those days that fall after the beginning of the year (DATESEP - start of year).
C. Admission to hospital is within the year, but separation falls after the end of the year.
Count only those days from admission to the end of the year (end of year - DATEADM).
D. Admission and Separation fall outside the start and end of the year.
Count only those days that fall within the year (end of year - start of year).
Notes
-
A. Counting All Days
When using separation based data, very long stays may be missed if enough years of data are not used. At least one year beyond the end of the study year should be used to capture late filed records and long stays. The difference in days counted may be significant.
e.g. Counting days for 1995/96 fiscal year all MB hospitals.
B. Duplicate records
Year of Data Number of Days Number of Separations % Change - Days % Change - Separations
95/96 1274565 159665 . .
+96/97 1468354 165230 13.2 3.4
+97/98 1501709 165441 2.2 0.1
When combining multiple years of hospital data duplicate records should be removed. Test for duplicates after combining data sets, but before doing any selections. Use at least PHIN, DATEADM, DATESEP, and HOSP to identify duplicate records. Keep the most recent record if duplicates occur.
See the associated concept on Duplicate Records - Hospital Discharge Abstracts for further information.