There are up to three different types of hospital identifiers included in the HCUP databases.

• The data source's own number scheme for identifying hospitals and facilities (DSHOSPID),
• The hospital identifier used by the American Hospital Association (AHAID and IDNUMBER), and
• A unique HCUP hospital identifier (HOSPID).

The hospital entity as defined by the data source may differ from the hospital entity as defined by the AHA. For example, the data source treats two separate facilities as two hospitals, while the AHA Annual Survey treats the two facilities as a single hospital, or vice versa. For consistency across states, HCUP defines hospitals in accordance with the American Hospital Association Annual Survey of Hospitals. During HCUP data processing, the data source's identification of the hospital is reconciled with the identification of the hospital in the AHA Annual Survey of Hospitals. For detailed information about this linking process, see the special report on HCUP Hospital Identifiers.

DSHOSPID is missing if the data source that contributed discharge data to the NIS prohibits the release of hospital identifiers.

Prior to 1998, the variable DSHOSPID is length 9 with the first digit indicating the level of care, the next two digits for state "06", and then a 6-digit hospital identifier that included the county code.

Beginning in 1998, DSHOSPID is length 6 and only contains the unique hospital identifier. The level of care indicator is retained in the HCUP variable LEVELCARE.

Regardless of whether the information on the level of care is stored in the first digit of DSHOSPID or variable LEVELCARE, the values are defined as follows:

The reliability of this indicator for the type of care depends on how it was assigned.

Prior to 1995. The type of care was assigned by California based on the hospital's licensed units and the proportion of records in a batch of submitted records that fall into each Major Diagnostic Category (MDC). Hospitals were permitted to submit discharge records in one of two ways: submit separate batches of records for each type of care OR bundle records for all types of care into a single submission. How a hospital submitted its records to California determined the accuracy of the type of care indicated in the first digit of DSHOSPID. Consider a hospital which is licensed for more than one type of care:

• If the hospital submitted one batch of records per type of care, then the distribution of each batch of discharges into MDCs would clearly indicate the type of care (acute, psychiatric, etc.). The data source could then accurately assign the first digit of DSHOSPID.
• If the same hospital submitted all of its records in one batch, then the distribution of discharges into MDCs would be a mixture of acute and other types of care. The first digit of DSHOSPID would be set to "general acute care" (value = 1) on all records and would not distinguish the types of care.

Prior to 1995, most hospitals submitted only one batch of records to California which meant that the type of care indicated in the first digit of DSHOSPID did not distinguish among types of care.

Beginning in 1995. Hospitals were required to assign type of care codes to individual records for certain discharges. These discharges included:

• general acute care (value = 1),
• skilled nursing and intermediate care (value = 3), and
• rehabilitation care (value = 6).

For discharges from facilities licensed as psychiatric care (value = 4) or alcohol/chemical dependency recovery treatment (value = 5), California continued to assign the type of care code to all discharges from the facility.

In 2000, some values of DSHOSPID have leading blanks. These DSHOSPIDs need to be left justified to be consistent with the coding of the same DSHOSPID in 2000 and other years.

In 2000, some values of DSHOSPID have leading blanks. These DSHOSPIDs need to be left justified to be consistent with discharges from the same DSHOSPID in 2000 and other years.

Beginning with 1995 data, Oregon changed the format of the state-specific hospital identification numbers stored in DSHOSPID. The new format is incompatible with the format used in previous years.

The coding of DSHOSPID varies by data year.

• Prior to 1995, the hospital identifier supplied by Pennsylvania contained a three character prefix "PAF".
• From 1995-1997, this prefix was not included in the supplied data. For consistency with previous years of HCUP data, the prefix "PAF" was added to the beginning of the Pennsylvania hospital identifier (DSHOSPID) during HCUP processing.
• Beginning in 1998, the prefix "PAF" is not included in the DSHOSPID for Pennsylvania.

Three discharges which can be identified by:

DSHOSPID < '010'

should be deleted from the 1996 South Carolina SID. These are partial records that were included in the raw source file.

A small number of discharges (<50) which can be identified by:

DSHOSPID - '999'

should be deleted from the 2005 South Carolina SID. These are out of state records which were not rejected during HCUP processing.

Included with the records of general acute care stays from community hospitals are records from alcohol dependency units, bone marrow transplant units, extended care units, psychiatric units, rehabilitation units, group health units, and swing bed units. Records for these different types of care can be identified by the fourth digit of the supplied hospital identifier (DSHOSPID) on each patient record:

Washington assigns this value to DSHOSPID based upon the type of unit discharging the patient.