Second Edition Miranda Wolpert, Rita Harris, Sally Hodges, Peter Fuggle
Yüklə 1,66 Mb. Pdf görüntüsü
|
- Bu səhifədəki naviqasiya:
- Table 1
- Informal confidence range for predicted resource use**
- Table 2
15 1. To be potentially allocated to the getting advice needs-based grouping, children, young people and families had to have at maximum one problem rated as moderate, no problems rated as severe and no problems rated as potentially significant and enduring (such as psychosis or eating disorders) on the current view at outset. On this basis, 28% of the episodes of care were considered potentially appropriate to include in this grouping. 2. To be potentially allocated to the getting help needs-based grouping, children, young people and families had to have a signature problem rated as moderate or above, or one problem rated as severe. On this basis 60% of the episodes of care were considered potentially appropriate to include in this grouping. Of these about half (30% of all episodes of care) are estimated to be allocated to potentially benefiting from intervention guided by one of the ten NICE guidelines subsumed under “getting help”, while the other half belong to the three “co-occurring problem” groups (30% of all episodes of care). 3. To be potentially allocated to the getting more help needs-based grouping, children, young people and families had to have a difficulty that indicated likelihood of need for substantive resource use, such as eating disorders, psychotic symptoms, or multiple severe problems. On this basis, 10% of the episodes of care were considered potentially appropriate to include in this grouping. Of these, around a quarter are allocated by the algorithm to potentially benefit from help guided by one of the three NICE guidelines subsumed under “getting more help”, while the other three-quarters belong to the non-NICE specified “difficulties of severe impact” (8% of all episodes of care). An important finding from the payment system work was that algorithm assignment did not fit neatly with actual resource use. This is consistent with findings in the development and analysis of other algorithm-based classifications. There was significant variability in actual resource use for children and young people and families potentially allocated to the groupings as outlined in Table 1 below.
Predicted resource use for needs-based groupings, from payment systems project analysis Needs-based groupings Predicted % in grouping based on application of the algorithm 95% confidence interval of group percentage Predicted average no. of sessions 95% confidence interval of estimated average appointments Predicted % resource use for a typical service* Informal confidence range for predicted resource use** Getting advice 28% 27%-29% 6.2
4.6-8.4 24%
20%-29% Getting help 61% 60%-62%
6.9 5.1-9.5
59% 53%-65%
Getting more help
11% 11%-12%
10.4 7.5-14.5
16% 13%-22%
Total 100%
-- 7.2
6.6-7.8 100%
-- Note: The estimation of “% in grouping” is based on closed and open cases from 11 CAMH services (n=11,353). The estimation of “average number of sessions” is based on the sample of closed cases whose points of contact began between 1 September 2012 and 28 February 2013 (n=757). The latter sample was constructed in an attempt to minimise bias towards shorter periods of contact, which arises because data collection ended on 30 June 2014 (giving an overall data collection period of 22 months). Nonetheless, by definition no child in the data set attended NHS outpatient CAMHS for longer than 22 months. We therefore think that the predicted averages of numbers of sessions given in the table (as well as their confidence intervals) are underestimates. *Data only included face-to-face work as data quality for indirect work was too poor, so number of sessions is taken as proxy for resource use. No data was known about more or less expensive staff so each contact is treated as of equal resource use. **The confidence range of estimated percentage of appointments takes into account the uncertainty about the estimated percentage of service users in each grouping, as well as the uncertainty about the average number of appointments within each grouping. This is not a precise confidence interval. 16 Table 2 below sets out an entirely hypothetical allocation to groupings and allied resource use which draws on the analysis above but assumes resource use that follows tighter allocation to clusters and includes hypothesised use by groupings not addressed in the payment systems work but core to THRIVE: thriving and risk support (see elaboration sections p.17 and p.23 below).
Getting advice i 30%
3 10%
8% Getting help 60% 10
56% Getting more help 5% 30
14% Getting risk support 5% 15
7% Thriving
n/a n/a
n/a 15%
Total 100%
9.2 100%
100% Note: The predicted average number of sessions here was set to 9.2, which is similar to the average number of sessions observed in data collected by CORC. This is higher than the 7.2 observed in Payment Systems data (reported in Table 1), since Payment Systems data are biased toward shorter periods of contact. It is crucial to note that Table 2 is entirely hypothetical. This framework must be tested and we do not want to make extravagant claims of cost savings without evidence. We hypothesised that targeting help may result in overall savings that would then free resources for community support, but this assumption is something to be tested as part of implementation trials. One of the key tasks of THRIVE is to make more explicit how resource usage links to need and for this to be examined, considered and refined as part of ongoing implementation and framework development. We now turn to a detailed discussion of each of the proposed needs-based groupings that make up the THRIVE framework. 9 This includes neuro-psychological assessment thought to be relevant in around 3% of cases and assumed to be happening in addition to other elements. Yüklə 1,66 Mb. Dostları ilə paylaş:
|