Table 1 Identifying additional data for E linkage.

Findings

Events for people on leave from permanent RAC (either for hospital leave or for social leave) have both start and end dates recorded.

E strategy adjustment

Matches to the admission, hospital leave and social leave events were carried out separately, and start and end dates for hospital and social leave events were used when matching to hospital discharges.

Question 2

Is there any information on the hospital data which indicates how the hospital event started or from which sector the patient came?

Findings

Information on usual residence (private dwelling, RAC etc.) is not reliably available on the hospital dataset. A variable on hospital 'mode of admission' shows whether or not the hospital event started with a within-sector transfer or as an admission into hospital.

E strategy adjustment

Hospital mode of admission data influenced the type of date comparisons made when matching to RAC leave events.

Question 3

Is there any information on the hospital data concerning where the patient went after discharge?

Findings

'Mode of discharge' is reported on the hospital data. Categories include whether the patient transferred within the sector, died in hospital, returned to their usual residence or went to RAC. Death in hospital is well-reported but return to RAC as the usual residence and admission to RAC are not well-distinguished.

E strategy adjustment

Hospital episodes ending with a within-sector transfer were excluded from the linkage process (also done for the basic E linkage). Hospital discharges were divided into three groups defined by reported destination on discharge: 'to usual residence', 'died' and 'other' (predominately people reported as discharged to RAC). The three groups of RAC events (admissions, hospital leave and social leave) were then matched separately to each of these three groups, with the exception that RAC admissions were not matched to hospital discharges due to death. This reduced the incidence of within-dataset duplicates with respect to match variables and so allowed greater flexibility in choice of match variables.

Question 4

Is there any information on the RAC data concerning where a resident was just prior to admission?

Findings

The RAC dataset does not explicitly contain information about a client's location just prior to admission. However, data on place of assessment for aged care (including whether in hospital) and date of the assessment are available.

E strategy adjustment

Place and date of aged care assessment were used to assist linkage when matching using hospital region (see 6. below)

Question 5

Is it possible for more than one event to be reported for the same person on the same day – either for hospital discharge events or RAC entry events?

Findings

Hospital episodes may last less than a day (same-day episodes), so a person may have two hospital episodes starting or ending on the same day. In addition, people may go to a different RAC facility at the end of their hospital leave. Such a change in RAC facility is recorded both as a return from hospital leave and as an admission into RAC, resulting in two entry events reported for the same day relating to the same event (i.e. collision events).

E strategy adjustment

Because same-day hospital episodes are highly unlikely to be related to a RAC entry event, same-day hospital episodes were excluded from the linkage process (also done for the basic E linkage). To select between any duplicate matches arising from multiple events on the same day for the same person or multiple comparisons between the groups of RAC events and hospital discharge groups, likely matches from the various comparisons were given a priority ranking. Priority ranks were based on reported destination on discharge from hospital, in addition to a preference for matches to a RAC hospital leave record over a RAC admission record (as a link to the former indicates that the person was already living in RAC), with matches to social leave being least preferred.

Question 6

Is there any additional address-type information on either dataset that could assist in identifying events for people who change their usual residence on leaving hospital?

Findings

Other available address-type information includes postcode of the hospital and postcode of the receiving RAC facility.

E strategy adjustment

Usual residence was grouped into regions of various size to allow for movement within a specified neighbourhood. Hospital region and RAC facility region as well as reported region of the person's usual residence were considered for matching.