3. Inpatient Classification and Cost Weights

3.1 AN-DRG Version 3

AN-DRG Version 3 was introduced on 1 July 1997 and will continue in 1998 - 99. The grouping software used is AN-DRG Version 3.1.

The Department engaged Hospital Services Research Group to conduct the 1997 - 98 Victorian Cost Weights Study of 1996 - 97 inpatient activity. A review of all weights was undertaken and the proposed areas of change were considered both through the Cost Weights Study itself and through formal Departmental consultations.

After reviewing submissions from clinicians and hospitals relating to clinical issues impacting on the cost weights, a small number of DRGs were identified as requiring further investigation. Almost universally issues relating to these DRGs were found to be due to DRG classification issues requiring resolution through the Australian Casemix Clinical Committee or poor cost data.

A full list of weights is given in Section C: Supplementary Information. WIES6 amendments and a full explanation of WIES6 are given in Section C: Calculation of WIES.

3.2 ICD-10-AM and AR-DRG Version 4

AN-DRG Version 3.1 will continue for 1998 - 99, although hospitals will move to code inpatient admissions according to ICD-10-AM from 1 July 1998.

The implementation of ICD-10-AM will provide substantially improved data quality. Coded data which reflects current clinical practice is essential for casemix-based funding and development, service planning and coordination, and epidemiology and other clinical research. However, it will take some time before the DRG grouper software has been redeveloped to maximise the value of the additional ICD-10-AM precision.

Data collected in the Dual Coding Study suggest that the DRG allocation for some patients differs depending on whether clinical details are coded using ICD - 9-CM directly or if clinical details were first coded in ICD-10-AM and then mapped back to ICD - 9-CM for grouping to DRGs.

Preliminary data from the Dual Coding Study indicated that reductions for some groups of DRGs are greater than for others. As a result, some hospitals may be more likely to be affected and provision has been made to modify the funding policy to ensure that they are fairly treated. The Dual Coding Study was, however, relatively small (10,000 cases) and targetted to problematic DRGs. The mapping tables used were preliminary and have since been refined. Its results therefore do not predict mapping effects in the total Victorian data of almost 800,000 cases.

The National Centre for Classification in Health (NCCH) has provided a number of mappings between ICD - 9-CM and ICD-10-AM. Forward mapping files provide ICD-10-AM mapped codes for each ICD - 9-CM code and backward mapping files provide ICD - 9-CM mapped codes for given ICD-10-AM codes. Each set of mapping files provides two types of maps: logical maps which give the most appropriate DRG allocation, and historical maps which give the most appropriate map on clinical meaning. In most cases the two are identical.

In addition, to lessen the impact of moving to ICD-10-AM the Department has developed an algorithm to improve the mapping of obstetrics codes based upon the presence of additional ICD-10-AM codes of Z37.* and Z39.0*, within the reported secondary diagnosis list. The mapping process will be incorporated into the construction of VIC-DRGs in 1998 - 99.

The effect of the move to ICD-10-AM on the ability of each hospital will be assessed and hospital specific code mapping adjustment factors used when comparing each hospital's coded throughput against target. Factors will be developed through comparing WIES6 allocation on 1997 - 98 VIMD data before and after simulated ICD-10-AM to ICD-9-CM mapping. Where necessary, adjustment factors will be re-assessed progressively through 1998 - 99.

Version 4 of the classification (AR-DRG 4) is expected to be available for implementation from 1 July 1999. A decision on its introduction must await further analyses. It will not be introduced before 1 July 1999. A collection of ICD-10-AM data is essential to the analyses and testing of the classification.

3.3 Coding Standards and Adjustments

The success and fairness of casemix funding is based on accurate and honest reporting of diagnostic information. Two coding audits have been conducted using 1993 - 94 and 1995 - 96 data. The 1993 - 94 data showed 86.5 per cent of the audited episodes were coded into the same AN-DRGs; 5.9 per cent were "overcodes"; and 7.6 per cent were "undercodes". In 1995 - 96, same coding was 88.3 per cent, a slight improvement of 1.8 per cent; 5.0 per cent were "overcodes"; and 6.5 per cent "undercodes". There are however differences across hospitals emerging that need closer examination.

It was decided to postpone a third audit planned for 1997 - 98 data due to additional hospital resources required for the introduction of ICD-10-AM coding. It is now planned to conduct an audit on ICD-10-AM codes and resultant AN-DRGs using 1998 - 99 data, commencing early in the 1999 calendar year. As well as providing a check on the accuracy of codes, the audit will be of great educational assistance to hospital coders using the new classification system.

It is intended that the next audit contract will cover a three year period to allow both annual audits and supplementary audits with larger sample sizes where the primary audit identifies hospitals with significant coding anomalies. In these cases the cost of the supplementary audit, VIMD data correction and WIES adjustment will be borne by the hospital.

3.4 HIV/AIDS

Due to changes in admission practice, coding and funding, this cluster of DRG codes has been assessed in light of the latest cost weight data. Some weights for these DRGs have increased significantly.

Part of the problem in the earlier weights appeared to relate to the classification of same day patients. The proportions of same day patients in the various HIV/AIDS DRGs differed significantly between the 1996 - 97 and 1997 - 98 Victorian Cost Weights Studies. Time series data demonstrated that the proportion of same day separations in these DRGs was variable, potentially resulting in inappropriate variations in the cost weights.

The Commonwealth has indicated that under AN-DRG Version 4 that all same day HIV/AIDS cases are likely to be in a single DRG.

After discussions with representatives of the Ministerial Advisory Committee on HIV/AIDS and Related Diseases and service providers and detailed consideration of all available data, an addition four HIV/AIDS DRGs have been designated as same day DRGs. In addition, AN-DRG 780 Chemotherapy, was split into VIC-DRG 778 (Chemotherapy without HIV/AIDS) and VIC-DRG 779 (Chemotherapy with HIV/AIDS).

3.5 Calculation of Inlier Boundaries: Trim Points

For WIES6 new inlier boundaries were calculated from the 1997 - 98 Victorian Cost Weights Study. For most DRGs the low boundary was set at a third of the average length of stay for the DRG and the high boundary was set at three times the average length of stay for the DRG. Inlier boundaries were converted to integers by truncating the low boundary and rounding the high boundary. The average length of stay was calculated after excluding extreme cases (up to 2.5 per cent of the longest stays and up to 2.5 per cent of the shortest stays). Where more than 2.5 per cent of separations occurred in either the highest or lowest length of stay category, no cases were excluded. For example, in many DRGs more than 2.5 per cent of separations are same day so for these DRGs no separations were excluded under low trimming.

For some DRGs the low boundary was calculated using a multiplier of 2/3 rather than 1/3 and the high boundary was calculated using a multiplier of 3/2 rather than 3. These modifications were based upon clinical discussions during the development of WIES1 to WIES4 and are detailed in Section C.

Where no episodes occurred for a DRG within the Cost Weight Study data base, the WIES5 boundaries were retained.

After clinical consultations regarding an altered payment structure for DRG 003, WIES5 boundaries were also retained for DRG 003. (This modification resulted in a slightly higher high outlier boundary and fewer high outliers than if the boundaries had been calculated using the 1997 - 98 Victorian Cost Weights Study). The 1997 - 98 inlier boundary points were retained for DRG 954 as small numbers of patients with extended stay were considered likely to inappropriately increase the boundary and reduce the appropriateness of the payment rates for shorter stay patients.

For the purpose of calculating inlier boundaries, same day separations were excluded when calculating the DRG mean for those DRGs that had designated same day DRGs. In such cases including same day cases would have inappropriately lowered the high boundary point resulting in a large proportion of non-same day separations being classified as high outliers.

Exceptions to this trimming method are neonates; high cost AN-DRGs; and specific AN-DRGs where the trim points have been modified, as advised by clinical specialists, to better delineate between levels of patient severity. The trimming changes notified in 1997 - 98 have been continued for 1998 - 99.

3.6 New Same Day DRGs

Changes in clinical practice over the last few year has resulted in significant declines in average length of stay for many DRGs, resulting in substantially lower boundaries and in some cases lower weights. In many cases reduced length of stay resulted mainly from a higher proportion of patients being treated on a same day basis. Where reductions in inlier weights were identified as mainly due to high growth in same day cases, the DRGs were classified as same day DRGs. This resulted in twelve additional same day DRGs plus the additional four HIV/AIDS DRGs which were classified as same day DRGs. These are listed in Section C.

3.7 Calculation of Inlier Weights

Weights were calculated from the average costs of inliers based upon the new inlier boundaries. Trimming was undertaken according to the criteria used for the 1997 - 98 Victorian Cost Weights Study. In calculating weights the following adjustments were made:

• The average costs of some DRGs were increased to adjust for prosthetic costs;
• All weights were subjected to rebasing to maintain state wide WIES equivalence between WIES versions. This was done by calculating both WIES5 and WIES6 on the same twelve months VIMD dataset and then scaling all WIES6 weights by the ratio of total WIES5 to total WIES6. Agreed target WIES5 levels were adjusted by similar hospital specific indices;
• Where there were fewer than 150 inliers in 1996 - 97 and where 1996 - 97 average cost differed by more than 20 per cent from the 1995 - 96 average cost, data were combined for 1996 - 97 and 1997 - 98. This process was undertaken to reduce the extent of statistical variation due to small numbers. The exceptions were for the average costs associated with the HIV/AIDS DRGs (801 to 805), which were calculated on 1996 - 97 data to reflect changes in actual drug costs; and
• In some cases preliminary weights were inconsistent with grouper logic. For these DRGs data for multiple DRGs were combined for the purpose of calculating inlier boundaries and weights. For example, where the preliminary weight for the DRG with CC was lower than the preliminary weight for the corresponding DRG without CC, data for both DRGs were aggregated to calculate a combined set of inlier boundaries and corresponding weight.

In addition a number of DRG specific adjustments were made:-

• The total and variable inlier weights for DRG 148 were increased by the equivalent of $20,000 to incorporate the cost of cochlear implants previously paid via specified grants;
• The average cost of DRG 003 was reduced by 20 per cent to adjust for the extension of mechanical ventilation payments to DRG 003;
• Between the 1995 - 96 Victorian Cost Weights Study and the 1997 - 98 Victorian Cost Study cost increases for same day patients in DRG 421 (knee procedures) were considered unlikely. The 1998 - 99 weight has been set based upon an average cost mid way between these points. The appropriateness of the reported costs will be examined during the coming year; and
• Rapid growth in both separations and average costs for multiday patients within DRG 942 (other factors influencing Health Status Age >79 or W CC) occurred between the 1996-97 and 1997 - 98 Victorian Cost Weights Studies. Growth was particularly pronounced for patients with principal diagnosis related to social reasons for admission. Consequently the 1998 - 99 weights have been set at last year's values while the appropriateness of casemix payment for these patients is assessed.

3.8 Prostheses Adjustments

Historically, prostheses costs have been poorly allocated to patients within hospital information systems. In many cases, costs associated with prosthetic devices are recorded under operating room costs and allocated accordingly. Consequently, under WIES4 and WIES5, adjustments were made to increase the reported average price for a number of DRGs where prosthesis costs were known to be significant. Data collected from the National Costing Study Service Weight Study and data from two Victorian Hospitals were used as a basis for making these adjustments. Costs were part recovered by reducing theatre costs across most surgical DRGs.

The adjustment for prostheses were recalculated for WIES6. This was necessary because the adjustment factors used previously, based upon data collected for patients admitted in the early 1990s, did not accurately reflect current medical practice and prosthetic prices. Further, in the 1997 - 98 Victorian Cost Weights Study eight hospitals reported prostheses costs separately. Consequently, a significant proportion of prosthetic costs were allocated by the hospitals, lessening the need for adjustment.

Adjustments were made by assuming that hospitals that were unable to allocate prosthesis costs have the same average prosthetic costs as the hospitals that allocated prosthetic costs to individual patients. For the purpose of calculating WIES6 weights, the average cost of inliers was adjusted for prostheses in 41 DRGs where the prosthesis adjustment increased the average cost by at least $20 and by at least 1 per cent of average costs.

For nineteen DRGs, the average cost allocated to patients in 1996 - 97 was significantly lower than the previous adjustment factor. In these cases the previous adjustment factor was used as the basis for adjustment rather than the average from the eight hospitals.

For three DRGs this process appeared to give anomalous results. The WIES4 adjustment for lenses (DRGs 98 & 99) was significantly higher than the published price in 1996 - 97. The average cost reported by hospitals, however, was significantly lower than the published price. After consultation, the prosthesis adjustment was set at the highest price of prostheses actually used plus an allowance for ocular fluid.

The other DRG treated differently was DRG 241 where the changed method of adjusting for prostheses resulted in a reduction in prosthesis costs of almost $10,000. Because of the large reduction, advice was obtained about preferred prosthesis for this DRG. Based upon this advice a prosthesis cost of approximately $8,000 higher than the reported average cost in the eight hospitals was used in developing the cost weights. This increase related to recent development of a superior prosthesis. Actual costs will be reviewed in 1998 - 99.

3.9 High Outliers

High outlier weights have been adjusted to ensure that, when using Network payment rates, variable payments for high outlier days are at least $125 per day (equivalent to the nursing home rate) and no more than $496 per day. As for WIES4 and WIES5 high outlier weights are adjusted by the specific high outlier adjustment factor. Surgical DRGs were allocated a high outlier factor of 0.7, medical DRGs were allocated a high outlier factor of 0.8 and some specialist DRGs were allocated a high outlier factor of 1.0.

Implicit in the allocation of the high outlier factor is the assumption that high outlier days on average cost less than inlier care. This assumption was questioned by some hospitals during 1997 - 98. Data provided by the Royal Children's Hospital suggested that, for a few DRGs, while the intensity of care given to high outliers reduced over time the cost of care past the high boundary was still significantly higher than the average daily cost for inlier care. However, the significance of the results were difficult to interpret due to small sample size. Average daily costs for inliers and high outliers were compared based upon the Victorian Costing Study data for the DRGs identified as potentially having high outlier per diems by the Royal Children's Hospital. These data supported the Royal Children's Hospital data for two DRGs (DRG 718 and DRG 725). High outlier weights have been set at 1.2 and 1.3 for these DRGs based upon average daily costs from the cost weight study.

As the costs associated with prostheses and theatre are usually incurred early in a patient's stay these costs are excluded when calculating high outlier WIES for DRGs with significant theatre and prosthesis costs.

With the complexity for trimming and outlier weights increasing as more precise costing data become available, a review of trimming and outlier funding policy is planned for 1998 - 99 in consultation with AR-DRG Version 4 evaluations.

3.10 Mechanical Ventilation Co-Payment

3.10.1 Extension to DRG 003

During 1997 - 98, analysis of the available data for DRG 003 (Tracheostomy except for Mouth Larynx or Pharynx, Age > 15) demonstrated that a considerable proportion of the variation in costs for patients allocated to DRG 003 related to the time spent in ICU rather than the time spent in hospital. Time spent in ICU, and days of mechanical ventilation, varies considerably between hospitals for inliers and it was therefore decided to provide a separate funding factor for mechanical ventilation in this DRG.

Under WIES6, DRG 003 will be eligible for mechanical ventilation co-payments after a patient has spent four days on mechanical ventilation (i.e. day 5 is the first day for which co-payments are made). The setting of a threshold for mechanical ventilation co-payments means that the costs associated with the first 4 days of ICU are fully included in the inlier weight for DRG 003. This is consistent with the grouper assignment of patients with more than 94 hours of continuous ventilation to this DRG and with the observation that many non-ventilated patients in this DRG spend about four days in intensive care.

3.10.2 Mechanical Ventilation Co-payment Rate

During 1997 - 98, the accuracy of the co-payment rate (0.7729) for mechanical ventilation was checked against the 1997 - 98 Victorian Cost Weights Study data. Given that the co-payment represents an adjustment for treating the most severely ill patients in ICU, the total ICU cost, less the total indirect ICU costs for eligible DRGs, was divided by the estimated number of mechanical ventilation days. This estimate of the per diem cost of mechanical ventilation was very similar to the current payment rate (0.7729 * Full WIES price).

The WIES5 mechanical ventilation co-payment rates will be retained for WIES6:

• 0.7729 WIES6 per eligible day on mechanical ventilation; and
• 3.132 WIES6 per eligible neonate episode on mechanical ventilation.

To be eligible for the copayment the patient must:-

• Have been ventilated for at least six hours (or 102 hours for DRG 003);
• Be admitted to a hospital with a recognised intensive care unit; and
• Be allocated to a DRG which is eligible for the co-payment.

3.11 Thalessaemia

Thalessaemia cases were demonstrated by costing data to require more resources than other patients within relevant DRGs. For 1998 - 99, each thalessaemia case in DRGs 760 and 761 will continue to receive a co-payment of 0.2648 WIES. These WIES will be part of the hospital's WIES target and general funding arrangements

3.12 Rehabilitation

Rehabilitation is an area that requires special attention as the nature of its activities do not readily fit into one AN-DRG category. In recent years considerable work has been undertaken to assess the feasibility and development of a new classification system, and in 1998 - 99 a proposed funding model that will more accurately reflect the diverse nature and cost of services required by rehabilitation patients will be circulated for discussion and possible implementation on 1 July 1999. This funding model is based on the Casemix Rehabilitation and Funding Tree (CRAFT) classification.

Analyses have been carried out on a full year's rehabilitation data from the Victorian Inpatient Minimum Dataset (VIMD). The preliminary model developed through clinical and statistical analyses has been further modified on clinical advice.

In brief, rehabilitation will be classified according to sixteen sub-groups: two groups are for stroke-neurological patients; seven groups for orthopaedic patients depending on type of procedure and functional status; one for cardiac/pulmonary patients; and a separate category each for spinal injury, burns, head injury and patients with amputations. There are a further two groups for general "other" rehabilitation.

Cost weights for payment for the sixteen categories have been provided from rehabilitation cost data which has been collected from a small number of acute hospital rehabilitation units as part of the 1997 - 98 Victorian Cost Weights Study.

A paper detailing the new funding model will be circulated to the field for comment in July 1998. The paper also proposes that funding for spinal, amputation, head injury and burns patients should not be based on cost weights but, due to the nature of these patients and the type of rehabilitation required, be provided by a specified grant.

Designation status would also be retained with the introduction of the model. Over 1998 - 99, issues relating to the introduction of the model, such as outlier payments, and admission and discharge rules will be addressed, and working groups will be formed to provide input and recommendations. Seminars will be also be held during the year to refine the proposed model. Shadow budgets outlining the impact of the introduction of the model will be provided to agencies during 1998 - 99 and refinements incorporated into the model.

Rehabilitation grants for 1998 - 99 will continue on a similar basis as previous years. Bed day payments will continue in 1998 - 99 at rates comparable to those established for Aged Care Services Output Group 113, with increments for Consumer Price Index (CPI) factors.

Funding will be provided at the following rates:

Level 1: $345 per bedday

Level 2: $287 per bedday

A capped number of beddays will be allocated to designated agencies for each level of service.

3.13 Specified Grants

In 1998 - 99 specified grants will continue to be paid to compensate hospitals for services which do not fall neatly into inpatient or outpatient service arrangements, and for classes of hospital care which DRGs do not measure well. The following specified grants will be retained with some modifications in 1998 - 99:

• Heart and Liver Transplants;
• Neonatal Intensive Care Unit (NICU) Cots;
• Spinal Injuries;
• Neonatal Cardiac Surgery;
• Paediatric Cardiac Investigations; and
• Paediatric Weights.

Non-English Speaking Background Grants will continue in 1998 - 99. These grants are available to all public acute hospitals with greater than 1,000 annual admissions of patients from non-English speaking backgrounds (NESB) and are aimed at helping hospitals to develop planned and integrated approaches to service delivery for these patients. In 1998 - 99 the non-admitted component of the NESB grant, previously identified as part of the Victorian Ambulatory Classification System, will be combined with the existing NESB specified grant.

The Department is aiming to improve the future allocation of NESB grants and to develop performance indicators for hospitals in receipt of these grants, to help ensure that the grants are being used to meet identified needs in this area. Better information on the preferred language of service users is required in order to meet these aims. A trial to develop ethnicity identifiers for frontline service providers is currently being undertaken as part of a joint Commonwealth Department of Immigration and Multicultural Affairs / Australian Bureau of Statistics initiative.

All specified grants will be subject to review as a means of improving the identification of acute health outputs.

3.14 Victorian Maintenance Dialysis Program (VMDP)

Maintenance dialysis services are funded by way of a combination of a fixed capitation grant and WIES throughput payments. Updating overall dialysis costs is not a component of the annual cost weight study. A study to examine renal dialysis service provision and costs, in the context of the VMDP, was put to tender in early 1998, with the successful tenderer being ACIL Consulting. Based on the results of the study, and after consultation with the Renal Dialysis Clinical Committee, program grants for the five treatment modalities have been set for the 1998 - 99 financial year.

As per the recommendation of ACIL Consulting, hospitals participating in the VMDP will continue to receive funding in two components the fixed capitation grant and a WIES equivalent throughput (variable) payment. The variable payment for in-centre and satellite haemodialysis has been retained at approximately $17,182 per patient per annum. The variable component for peritoneal dialysis is included in the fixed capitation grant. Table 5 provides a comparison of the 1997 - 98 and new 1998 - 99 fixed capitation grant.

 

Table 5: Fixed Capitation Grants 1997-98 and 1998-99
Treatment Modality	1997-98 Fixed Capitation Grant per patient per annum *	1998-99 Fixed Capitation Grant per patient per annum
In-Centre:	$26,400	$24,526
Satellite:	$26,400	$19,517
Home Haemodialysis:	$32,200	$27,809
Continuous Ambulatory Peritoneal Dialysis:	$34,200	$34,992
Intermittent Peritoneal Dialysis:	$34,200	$25,180
* Contains capital component of approximately $5,000 per annum.

 

All but one of the program grants for the five treatment modalities were set at the price recommended by the consultants. The price for satellite treatment is higher than the recommended price.

The study revealed that the cost of in-centre dialysis is higher than satellite dialysis due to in-centre patients requiring a particular procedure not required by the type of patients receiving dialysis at satellite centres. This is reflected in the higher capitation grant for in-centre patients. The option of continuing the weighted average of the in-centre and satellite rate was considered but was not adopted as it would impact disproportionately on hospitals with significant winners and losers according to the ratio of in-centre to satellite utilisation. Although more costly, the alternative of fully reimbursing in-centre costs and paying a premium for satellite treatment of $1,500 per patient per annum has been preferred.

The study of renal dialysis service provision and costs did not specifically address capital equipment replacement, but noted that a solution to the problem of capital consumption was integral to a long term plan for the provision of services. This issue will be addressed and a capital funding solution determined by 1 October 1998. The report also recommended that the Department consider tendering out renal dialysis services in the medium term.

The study revealed that, over the period of June 1995 to June 1997, the number of total dialysis patients in Victoria grew an average 1.95 per cent per quarter, or 8.03 per cent per annum. The highest level of growth was in satellite patients, while the number of patients in other treatment modalities was stable or declining slightly. Home haemodialysis has shown the most significant change with patients declining from 200 in June 1995 to less than 150 in June 1997. The above growth rate and use of treatment modalities are expected to continue in 1998 - 99 and funding for these levels has been provided.