What is casemix?
Casemix is a way to group like ‘cases' or 'outcomes episodes' of mental health care so outcomes can be compared. It enables a better understanding of the real difference in service user outcomes achieved between differing services.
The purpose of a casemix classification is to group episodes, or periods of care, based on factors which best predict the need for, and the cost of, care. Episodes are grouped into classes based on two criteria.
- Each class should contain episodes with similar patterns of resource consumption (the assumption being that service users who consume similar resources have similar needs).
- Each class should contain episodes that are clinically similar.
Variation in outcomes achieved by different teams, services or provider organisations, may be due to different treatment models or the mix of service users accessing mental health services. Casemix allows the adjustment of outcomes information to standardise for the mix of service users, the casemix, and to make comparisons between the variations in their outcomes more meaningful.
Controlling for the variation between service users assists in identifying and understanding the differences between service providers. By providing this means to understand the variations in casemix, providers can better focus on differences between the way services are delivered and what may contribute to improved outcomes for particular groups of service users.
Casemix in New Zealand
The main objective of the NZ Classification and Outcomes Study (CAOS) (Gaines et al., 2003) was the development of a casemix classification for New Zealand mental health services. The secondary objective was to trial the use of routine measures to assess service user outcomes.
Following the study’s completion in 2003, the Ministry of Health mandated collection of the HoNOS family of outcome measures in New Zealand. The Ministry of Health had also established a national programme, MH-SMART, to assist district health boards in implementing the collection of these measures.
With the implementation of the Programme for the Integration of Mental Health Data (PRIMHD) in July 2008, outcomes information, in the form of the HoNOS measures, now forms a mandatory part of the national mental health dataset.
To allow analysis and comparison of outcomes achieved by different mental health services for benchmarking and quality improvement purposes, outcomes need to be casemix-adjusted to take account of the unique mix of service users at each district health board. Work has recently been completed (Te Pou, 2011) to validate the 2003 classification with PRIMHD data. The Ministry of Health has since established a project to build casemix into the PRIMHD datamart. This will allow for analysis, monitoring and use of casemix adjusted outcomes from late 2012 onwards.
The New Zealand CAOS Casemix Classification
The NZ-CAOS Casemix Cassification groups service user episodes, or ‘periods of care’ into one of 42 casemix classes based on a range of nine variables.
Episodes are first split into two branches:
- episodes from an inpatient setting
- episodes from a community setting.
Each of these is then split into adult or child/youth episodes.
The nine NZ CAOS casemix variables include direct service measures, direct service user measures (such as age, ethnicity) and a blend of the two (such as focus of care). Some variables are selected from the ‘episode start’ collection and some from the ‘episode end’ collection.
The casemix variables are:
|Child and youth episodes||
Overview of the NZ-CAOS casemix classification
How casemix might be applied
Casemix can be a key information tool in planning, reviewing and improving specific team and service activities. It can also provide a source of information for nationally coordinated continuous quality improvement and benchmarking initiatives. This includes continuity of care indicators, and relationships between outcomes and other indicators.
Team / service level
Casemix adjusted change information can be useful in measuring performance at the team or service level – in any team or service there will be a mix of different service users and casemix classes. Mean change can be monitored at the team level, or by individual casemix class level within a team to identify for which service user groups improvements can be made. Access to regional or national casemix adjusted change information will allow services to identify gaps, network with those services achieving greater than expected change, and implement measures to improve service quality.
The following graph provides an example of mean change in score for community teams. This shows the actual change in total score, as well as the casemix adjusted mean improvement score, taking into account the casemix for each team.
Figure: CARMI and mean change in total score – adult community teams, DHB A
National (DHB/service provider) level
A range of national reports could be produced to demonstrate improvement (change) for services users at DHB level. Casemix adjusted outcomes can be reported by casemix branch, by setting (inpatient or community) or for specific, high cost, casemix classes.
PRIMHD outcome reports
The current PRIMHD outcome reports are not casemix adjusted but report on individual collection data rather than episode level data. With the implementation of casemix in PRIMHD, national reporting on casemix adjusted outcomes at the episode level should be included in the national set of outcome reports made available to service providers.
PRIMHD integrated reports
With the integration of activity and outcomes information and the capacity to include casemix adjusted change, analysis could be undertaken to identify the types and quantity (number and duration) of services provided. The associated change in outcomes by class, team/service, and DHB can identify what may contribute to improved outcomes for particular classes or groups of service users. This analysis would be enhanced in the future by the inclusion of treatment intervention in addition to the existing activity information in PRIMHD.
These should be included in all reporting for providers where collection compliance has not reached the target compliance rates. A minimum collection compliance rate may be recommended to avoid inaccurate reporting of data.