Previous versions of the NSW Public Hospitals Comparisons Data Book have included the cost per casemix weighted separation for acute hospitals based on data received through the Unaudited Annual Return (UAR) and the Inpatient Statistics Collection (ISC). The figure is calculated by dividing the total dollars reported for the relevant programs in the UAR for the particular hospital, and dividing these by the number of acute separations (casemix weighted) from the ISC for the corresponding period.

The problem with this approach is that the activity and costs reported by hospitals can only be linked up indirectly. Therefore, cases from the ISC that may not have been in the scope of the relevant UAR Programs could potentially be included when deriving the cost per casemix weighted separation and vice versa.

The NSW Hospital Cost Data Collection (HCDC) reports activity and cost data, and therefore, it can ensure that only the relevant cases are included when deriving the cost per casemix weighted separation.

The main obstacle in using data from the HCDC to derive the cost averages for hospitals has been the lag time in receiving and processing the data relevant to a particular year’s publication. For 1997/98, the receipt and processing of the HCDC has aligned with the production of the present publication, and therefore, these costs are able to be included.

The steps calculating the cost per casemix weighted separation for the hospitals participating in the NSW HCDC are described below.

1. Data submitted for Hospital Cost Data Collection (HCDC)

Scope of the HCDC

The selection of hospitals required to submit data to the NSW HCDC is based on the NSW Peer Group classification. The groups within the scope of the Collection are:

• A1 – Principal referral
• A2 – Paediatric specialist
• A3 – Ungrouped acute
• B1 – Major metropolitan
• B2 – Major non metropolitan
• C1 – District group 1
• C2 – District group 2

The selection is based on the peer group of the hospital the year prior to the cost collection year. Therefore, for the 1997/98 collection, the 1996/97 peer group assignments were used rather than the ones reported in the present publication.

The services within the scope of the NSW HCDC are acute inpatient separations as defined by the NSW Program Structure. The following Programs are included:

• Program 3.1 Emergency department services (inpatients only);
• Program 3.2 Same day services; and
• Program 3.3 Acute overnight services.

The classification system used to assign and report costs for the 1997/98 HCDC was the Australian National Diagnosis Related Groups (AN-DRGs) classification, version 3.1.

Activity data

A requirement of the NSW HCDC is that a file of the activity being costed is submitted to NSW Health. The inpatient data differ from data provided to NSW Health through the ISC in two important ways:

1. The data provide information about the cases that were costed, at the time of costing. This may mean some differences with the ISC in the numbers of cases by AN-DRG (ie due to some records not being coded at the time of costing), and differences in DRG allocation (ie sometimes due to uncoded records, and sometimes due to differences in DRG groupers used at the hospital/Area and Department).
2. The data incorporate local knowledge about the patients being costed as far as possible. For example, some cases that are flagged as non-acute in the ISC may in fact have been acute cases in the scope of the Programs being costed. In these instances local system may automatically assign a non-acute category based on the ward the patient was admitted to, or on the specialty of the attending medical officer. If these patients genuinely required acute care, costing officers were able to override their assignment to the non-acute category in consultation with clinicians, and include their costs in the HCDC.

The activity files submitted by hospitals included the variables on which the costing was based (ie service category on admission, program, payment status, same day status and AN-DRG), and sufficient information to enable merging with ISC (ie hospital code, medical record number, date/time of admission and separation).

The data from these files were used to supplement data in the ISC, and as a quality check on the numbers and characteristics of cases reported through the ISC.

Cost results

Hospitals are required to submit costs by AN-DRG, separately for same day and overnight separations. The costs of unqualified babies are also removed from the costs of the mother and reported under a separate group (DRG 747). Cost files submitted show the number of patients being costed by AN-DRG and cost group, separately for same day and overnight patients. To assist hospitals in separating costs associated with same day patients, a ‘thousand’ series DRG set was created. Therefore, the DRG for a same day patient in AN-DRG 185 is 1185.

The cost groups that hospitals were required to identify for the 1997/98 HCDC were:

• Clinical departments
• Operating theatre
• Invasive procedure room
• Pathology
• Imaging
• Wards
• Emergency
• Intensive care
• Pharmacy/drugs
• Allied health
• Prostheses
• Depreciation
• Superannuation and workers compensation

Hospitals participating in the NSW HCDC used one of two costing approaches: patient costing or cost modelling. Patient costing sites used Trendstar (CSC Australia) software, whilst cost modelling sites used COSMOS software (NSW Health).

2. Merge of HCDC activity with ISC activity

The activity data submitted by hospitals were merged with ISC data on a case by case basis. The merge was based on hospital code, medical record number, and the dates of admission and discharge.

As mentioned above, the objectives of the merge process were twofold: to update fields for individual records reported in the ISC, and as a quality check to ensure that the numbers reported through the HCDC are similar to those reported through the ISC.

The merge was reasonably successful overall. However, there were hospitals for which the merge did not work as well. The reasons for mismatching (ie additional or insufficient records being reported through the HCDC) are currently being investigated.

3. Subset of acute activity

The data from the HCDC were used to update the service category field in the ISC records in the first instance. Where a HCDC record existed, the HCDC service category was used to overwrite the ISC service category, except for records where the ISC service category was unqualified neonate, which was retained.

This flag and the HCDC Program flag were then used to subset the records within the scope of the analysis.

The following records were considered to be within the scope of the costs reported through the HCDC, and therefore, included in the analysis:

• Unqualified babies (based on service category on admission reported in the ISC);
• Cases with a HCDC Acute Program flag (ie Program 3);
• Cases with a blank Program flag (ie no HCDC record or Program not reported in the HCDC), but with an acute service category on admission and with no days in a designated psychiatric unit).

4. Calculation of cost weighted separations

In the first instance, all records in the ISC database were assigned a cost weight based on the cost weights from the 1997/98 HCDC (see NSW Acute Care Costs 1997/98, NSW Health, May 1999). Cost weights for unqualified babies are also included in this set.

The DRG for cost weight allocation was sourced from the ISC.

5. Preparation of HCDC cost result files and merge with activity database

As described above, each hospital submitted cost results by AN-DRG separately for same day and overnight cases, broken down into the 13 cost groups. The cost groups included a group for depreciation, which was subsequently used for deriving the cost per casemix weighted separation less depreciation.

File formats for sites using COSMOS and sites using Trendstar varied slightly. Trendstar sites also included a variable identifying the number of separations for each DRG. However for COSMOS sites, data needed to be derived from a separate file (PDAT file) to identify the number of separations for each costed DRG.

For COSMOS sites unqualified babies were assigned to a separate DRG (747 for overnight separations or 1747 for same day separations). However for Trendstar sites the results were submitted with the costs of unqualified babies assigned to the original DRG to which these babies were assigned. Data for unqualified babies for Trendstar sites were extracted from the original DRG (using statistics from the activity file), and assigned to either DRG 747 or 1747.

Results data for each hospital were aggregated into a single file. From this file the total reported acute care costs could be derived, and compared with data reported through the UAR. There are legitimate reasons why the total acute inpatient costs reported through the UAR vary from costs reported through the HCDC. The reconciliation return provides a basis for identifying these variations.

One source of systematic variation between Trendstar and COSMOS sites relates to the costs of patients who are in hospital at the beginning and end of the year. For patient costing sites (Trendstar), costs incurred during the year are allocated to these patients. As patients who were in hospital at the beginning of the year incurred costs prior to the start of the year, their costs will be under-estimated. As patients who were in hospital at the end of the year will incur costs in the following year, their costs will also be underestimated. To avoid these under-estimates, only patients that were admitted and separated within the year were costed. As a consequence the total costs reported in the results for the Trendstar sites will be less than the costs reported in the UAR.

The situation is different for cost modelling (COSMOS) sites. These sites only allocated costs to patients who have separated within the year. This includes patients who were admitted prior to the beginning of the year, but excluded patients who were still in hospital at the end of the year. On balance this approach does not result in an under-estimate of costs. This conclusion was verified by analysing data related to these situations. In the case of COSMOS sites the total costs reported in the costing results will be the same as the costs reported in the UAR.

Taking these factors into account, the costing results were compared with the acute inpatient costs reported in the UAR. Several hospitals were identified where the costing results were significantly below the total costs reported in the UAR. For these hospitals the UAR costs rather than the HCDC costs were used to calculate the average cost per casemix weighted separation (ie as for previous years).

The next step was to incorporate the HCDC costs by AN-DRG and same day or overnight status reported by individual hospitals into the activity database created at step 2.

6. Subset of cases and costs used to derive the cost per casemix weighted separation

The following cases and associated costs were removed to ensure that only reliable cost estimates were included in calculating the cost per casemix weighted separation:

• Observations from the results file where no DRG had been assigned (ie ungrouped cases);
• Error DRGs: 951, 952, 955 and 956; l Rehabilitation DRGs: 940 and 941, where the Casemix Standards do not assign a cost weight;
• Psychiatric and Drug/Alcohol DRGs: 841-863. These were excluded as separations with a psych flag were not included in the merged ISC/Activity file, and consequently the average cost per DRG may not be representative of costs for the remaining separations; and
• Tertiary neonate DRGs: 705-710, only for sites without the required level of facilities to provide these services.

Observations with missing DRGs and error DRGs were checked to ensure costs excluded through this process were not excessive.

7. Calculation of Cost Per Casemix Weighted Separation

For non-excluded hospitals, total costs and total depreciation costs for each DRG were calculated by multiplying the number of separations for the DRG (as identified in the ISC) by the average cost reported for the DRG and the average depreciation cost reported for the DRG.

Total costs and depreciation costs were summed for the hospital, as were the total casemix weighted separations.

The average cost per casemix weighted separation was derived by dividing the total cost for the hospital by the total casemix weighted separations. The average cost per casemix weighted separation less depreciation was calculated by taking the total cost for the hospital, deducting total depreciation costs, and dividing the remainder by the total casemix weighted separations for the hospital.

8. Analysis of results

The average costs by hospital were compared to the averages derived using the UAR data as a quality check. Hospitals with known problems (for example, those reporting a large proportion of implausible costs by DRG compared to their peer group average, or those with large proportions of mismatched data between the ISC and HCDC) also generally had large differences in costs when comparing the two approaches. The final step was to make decisions about whether the averages using the UAR approach were to be reported rather than those derived using the HCDC. For a few hospitals, it was decided that the UAR average would be reported. These hospitals were:

A202 Canterbury District

B202 Gosford

B206 Wyong

J208 Inverell District

M207 Deniliquin

M212 Mercy Hospital, Albury

N201 Batemans Bay District

P212 Illawarra Regional

Q209 Muswellbrook District

R205 Griffith Base

R219 Wagga Wagga Base

Details of the calculations in each of these steps will be made available to hospitals and Areas, and will be used as a basis for discussion with these sites in order to improve the process in future years.

Page Owner: IMCS         Last Updated : Tuesday, 20-Jul-99 09:53:20
URL: internal.health.nsw.gov.au/iasd/imcs/