This post is one in a series on Payment by Results in mental health, written for both professional and personal reasons. The full series is introduced and linked to here, including a post with all references in it.
England is at one of the most advanced stages for introducing Payment by Results in mental health anywhere in the world. There is limited and no long-term experience from anywhere else to inform policy and practice (CHE, 2009:iii). The most relevant learning can be found in Australia, Canada, New Zealand, the Netherlands and the United States, which have all considered or introduced a form of Payment by Results for mental health (CHE, 2009:6).
Different approaches to PbR in mental health were to be found in different places. For example, in the classification of mental health problems – the equivalent to clustering in England – there was some variation:
- Australia produced 42 groups as part of its classification system: 19 for community episodes and 23 for inpatient episodes. These include diagnosis (using ICD-10), severity (using HoNOS) and other factors (CHE, 2009:16)
- The Canadian system (which focuses on inpatient care only (CHE, 2009:vi)) groups mental health assessments into one of seven categories and 47 groups. Two further categories covered no diagnoses and records for short stay episodes (CHE, 2009:21)
- In New Zealand, there were 42 classes identified: 20 inpatient and 22 community (CHE, 2009:21)
- In the United States, patients were covered by one of 468 Diagnostic Related Groups (DRGs) (CHE, 2009:40)
Different places also used a variety of different factors for determining what is and isn’t included directly within classifications, or instead adjusted for on top of the classification. For example:
- New Zealand used regression analysis that helped to incorporate patient characteristics rather than the interventions they received (CHE, 2009:21)
- New Zealand also explicitly included ethnicity in its classification system, since it had a major impact on case complexity (CHE, 2009:22)
- Australia and New Zealand used episodes as the unit of analysis, rather than interventions or services received (CHE, 2009:49)
- In Canada, each episode is adjusted for the length of stay and location of the treatment (CHE, 2009:20)
- The US doesn’t use classifications at all as a means of payment: it uses average cost, per diem case payments which are then adjusted by a range of factors, including patient characteristics (for example, age, comorbidity, use of ECT) and provider characteristics (for example, rurality, teaching status and local wage rates) (CHE, 2009:41;45).
As CHE notes (2009:48), the English approach to PbR in mental health (21 clusters that are not based on diagnosis, though people with similar diagnoses and similar levels of symptom severity are likely to be found within the same cluster) is most similar to the Australian and New Zealand approaches.
In both Australia and New Zealand, though they had trialled these classification systems, they were not subsequently used for payment purposes (CHE, 2009:iv). The difficulties of covering both inpatient and community care contributed to this decision (CHE, 2009:16;45). Furthermore, the significant variation in provider costs identified in Australia meant that an average cost approach couldn’t easily be used (CHE, 2009:17).
More generally, the complexity of the introduction of PbR in mental health has meant that, in all of the countries where it has been tried, it has been introduced gradually, over a period of years (CHE, 2009:v).
Very little is known about the impact of PbR in mental health on the quality of mental health services. All that can be concluded from work in other countries is that:
No country has developed a single system that covers inpatient hospital care and community care (CHE, 2009:48)
Even if there were consensus on optimal treatment pathways – which is far from clear (and especially so in England) – “differences in service configuration, economies of scale and scope, resource availability and local costs may mean that [the impact of PbR] may vary in practice.” (CHE, 2009:49).