Peer Groups

Peer grouping is a process by which a cohort of facilities is divided into mutually exclusive and exhaustive subsets. Peer grouping is performed in a logical manner designed to meet a set of clearly defined ‘principles’.

In NSW, peer groups are used as the basis for:

• Activity and cost comparisons presented in the NSW Public Hospitals Comparison Data Book;
• Cost benchmarking. Cost benchmarks that individual hospitals are expected to achieve are the peer hospital group averages;
• Selecting hospitals to participate in the NSW Hospital Cost Data Collection;
• Planning services. For example, patient activity is analysed based on the clinical conditions of the patient, as well as the level of service sought indicated by the type of hospital at which the patient was treated; and
• Other benchmarking activities at a hospital or clinical level. For example, comparisons of length of stay of patients receiving a hip replacement.

Principles for grouping peer hospitals

Peer groups should be categorised using evidence on hospital activity such that each peer group has the following characteristics:

  a. Sufficient number of hospitals in each group

  b. Groupings should be based on relatively strong evidence

  c. Reliability and robustness

  d. Relative casemix homogeneity

  e. Relative resource homogeneity

  f. Relative similar hospital structure

In 1998, a Working Group consisting of AHS and NSW Health representatives was established to consider approaches to peer grouping. The group determined that peer groups should be based on role for facilities with specific functions (for example, psychiatric hospitals and nursing homes), on size for general acute hospitals (measured by total acute casemix weighted separations), and on the level of acuity for small community hospitals. Areas were requested to provide comments, which resulted in the re-allocation of some hospitals.

Verification of the methodology

One of the main strengths of peer grouping methodology is that it was developed using the valuable input of the AHSs through the Peer Hospitals Group Working Party. The methodology developed is relatively simple and relies upon either special roles of the facilities, size and/or measure of acuity to determine the peer groups.

Aisbett (1998) has developed a Clustering algorithm that uses iterated correlation binary-clustering to develop peer groups. In reporting his results Aisbett (1998) has compared his results with the 1996/97 NSW peer groups. He concludes that ‘clusters’ results ‘have a high level of agreement with the input defined Peer Groups devised by the NSW Health Department’.

Whilst the methodology developed for the 1996/97 edition is not perfect, (indeed a number of small changes were deemed necessary this year), this research by Aisbett (1998) confirms that the results of the methodology are at least equal to those derived using state of the art statistical methods.

Enhancing the 1996/97 methodology

NSW Health has made a commitment to an annual review of the peer groups. Rather than make further radical changes, the 1997/98 review aimed to:

1. Provide an overview of the existing methodology
   
2. Review the consistency of the existing methodology
   
3. Re-allocate hospitals which had had a role change since the previous year
   
4. Re-allocate hospitals to reflect changes in activity during 1997/98
   

• Extended adjustments to the data. Acute separations occurring in designated psych units and separations in rehabilitation DRGs where they were coded as acute were not included in the calculation of the acute casemix weighted separations. This was done for the purpose of consistency with the NSW Hospital Cost Data Collection and the Casemix Standards for NSW 1998/99.
  
• Clearer classification of the ungrouped acute (E) peer group, now referred to as A3 to reflect the referral role performed by the majority of the hospitals in the group.
  
• Internal reviews of the results to confirm that all facilities have been appropriately grouped.
  

Summary of the 1997/98 methodology

The methodology use for the purposes of classification in the peer grouping process does not differ greatly from that developed by the Working Group for the 1996/97 publication. It has been necessary to redefine the methodology to better describe the stages and decisions made. The methodology, depicted in Figure 1, follows six steps:

Step 1:    Adjustments to data
Step 2:    Group facilities with a specialist function (peer groups B2, E, F1, F2, F3, F4, F5, F6, F7 and F8)
Step 3:    Allocate hospitals to peer groups A1, B1 and C1 by acute casemix weighted separations
Step 4:    Allocate hospitals to peer group C2 by acute casemix weighted separations and separations
Step 5:    Allocate hospitals to peer groups D1 and D2 by acuity
Step 6:    Verify classification (hospitals close to boundaries and changes).

Figure 1:    A logical explanation of the peer grouping methodology (excluding step six)

The review resulted in a relatively small number of changes in the classification of peer groups. Peer group E was reclassified A3 to reflect the major referral role of all but one of the hospitals. Another thirteen facilities were reviewed and reclassified due to changes in activity. They are as follows:

Facility	AHS	Peer 1996/97	Peer 1997/98
A202 Canterbury District Hosp	CSAHS	E - Ungrouped acute	B1 - Major metropolitan
B206 Wyong Hosp	CCAHS	C1 - District group 1	B1 - Major metropolitan
J214 Prince Albert Memorial, Tenterfield	NEAHS	D1 - Community acute	D2 - Community non acute
K206 Coonabarabran District Hosp	MAHS	D2 - Community non acute	D1 - Community acute
K215 Gulgong District Hosp	MAHS	D2 - Community non acute	D1 - Community acute
M204 Berrigan War Memorial Hosp	GMAHS	D1 - Community acute	D2 - Community non acute
M206 Corowa Hosp and NH	GMAHS	D2 - Community non acute	D1 - Community acute
M209 Henty District Hosp	GMAHS	D1 - Community acute	D2 - Community non acute
M212 Mercy Care Centre, Albury	GMAHS	E - Ungrouped acute	F8 - Ungrouped non acute
N204 Boorowa District Hosp	SAHS	D1 - Community acute	D2 - Community non acute
N209 Goulburn Base Hosp	SAHS	C2 - District group 2	C1 - District group 1
Q208 Merriwa District Hosp	HAHS	D1 - Community acute	D2 - Community non acute
R211 Leeton District Hosp	GMAHS	D2 - Community non acute	D1 - Community acute

In addition a number of other hospitals were reviewed for possible changes in peer group. This was because using the above methodology they were highlighted as needing regrouping or were borderline cases. However, at this stage it was decided that it would be more beneficial not to change the peer grouping of these facilities.

The facilities that were considered for review were as follows:

It was also noted that peer group F8 contained a number of sub-groups. These included facilities that may be considered for the most part as hospices or rehabilitation units. No decision was made on these units, however. The 1998/99 peer grouping process will be informed by the collection of provisional SNAP data, which will assist in better classifying these facilities.

The results of the 1997/98 peer grouping can be summarised using the number of acute inpatients or total inpatients as follows:

* Excludes error DRG, same day dialysis and chemo DRG, and rehab DRG seps, unqualified babies and seps in designated psych units
** Excludes error DRG and same day dialysis and chemo DRG seps, and unqualified babies

Table 1: Peer Group Definitions
Peer Group	Definition
A1 Principal Referral	Acute hospitals, treating 25,000 or more acute casemix weighted separations* per annum.
A2 Paediatric Specialist	Establishments where the primary role is to provide specialist acute care services for children.
A3 Ungrouped Acute	Establishments whose primary role is the provision of acute services of a specialised nature for which there is insufficient peers to form additional peer groups. Limited comparisons can be made with other hospitals in either A1 or A2.
B1 Major Metropolitan	Acute hospitals, treating 10,000 or more, but less than 25,000 acute casemix weighted separations* per annum.
B2 Major Non-Metropolitan	Establishments located in rural areas providing acute specialist and referral services for a catchment population from a large geographical area.
C1 District Group 1	Acute hospitals, treating 5,000 or more, but less than 10,000 acute casemix weighted separations* per annum.
C2 District Group 2	Acute hospitals, treating 2,000 or more, but less than 5,000 acute casemix weighted separations* per annum, plus acute hospitals treating less than 2,000 acute casemix weighted separations per annum but with more than 2,000 separations* per annum.
D1 Community Acute	Acute hospitals, treating less than 2,000 acute casemix weighted separations* per annum, and less than 2,000 acute separations* per annum, and with less than 40% non-acute and outlier bed days* of total bed days.
D2 Community Non-Acute	Hospitals, treating less than 2,000 acute casemix weighted separations* per annum, and less than 2,000 acute separations* per annum, and with more than 40% (or more) non-acute and outlier bed days* of total bed days.
F1 Psychiatric	Establishments devoted primarily to the treatment and care of inpatients with psychiatric, mental or behavioural disorders. Centres of non-acute treatment of drug dependence, developmental and intellectual disability are not included here. This group also excludes institutions mainly providing living quarters or day care.
F2 Nursing Homes	Establishments which provide long-term care involving regular base nursing care to chronically ill, frail, disabled or convalescent persons or senile inpatients. They must be approved by the Commonwealth Department of Health and Family Services and / or licensed by the State, or controlled by government departments.
Multi-Purpose Services F3 Current F4 Future	Multi-Purpose Services (MPSs) which provide integrated acute health, nursing home, hostel, community health and aged care services under one organisational structure, as agreed between the Commonwealth and State Governments. MPSs provide a range of services which are negotiated with the community, the service providers and the relevant Departments. This group is further split into current and future MPSs.
F5 Hospices	Establishments with a specific function of providing palliative care to terminally ill patients
F6 Rehabilitation	Establishments with a primary role in providing services to persons with an impairment, disability or handicap where the primary goal is improvement in functional status.
F7 Mothercraft	Establishments where the primary role is to help mothers acquire mothercraft skills in an inpatient setting.
F8 Ungrouped Non-Acute	Establishments whose primary role is the provision of non-acute services, but for which there are insufficient peers to form an addition peer group. Limited comparisons can be made within this peer group and with other non-acute facilities.
* - Please refer to Figure 1 for inclusions and exclusions in calculating the figures used to determine the peer groups.

References

Aisbett, C., (1998) ‘Formal casemix comparison of hospitals’, The Tenth Casemix Conference in Australia: Conference Proceedings, Melbourne.

Australian Institute of Health and Welfare, (1998) National Health Data Dictionary Version 7.0. AIHW: Canberra.

Australian Institute of Health and Welfare, (1997) National Health Data Dictionary Version 6.0. AIHW: Canberra.

Commonwealth Department of Human Services and Health, (1997) Day Only Procedure Manual. Commonwealth of Australia: Canberra.

NSW Health Department, (1997) NSW Public Hospitals Comparison Data Book 1995/96. Information and Data Services Branch, NSW Health Department: Sydney.

NSW Health Department, (1998a) Technical Paper: Casemix Standards for NSW 1997/98. Structural and Funding Policy Branch, NSW Health Department: Sydney, January 1998.

NSW Health Department, (1998b) Casemix Standards for NSW 1998/99, Structural and Funding Policy Branch, NSW Health Department: Sydney, November 1998.

Page Owner: IMCS         Last Updated : Monday, 21-Jun-99 09:53:20
URL: internal.health.nsw.gov.au/iasd/imcs/