Evaluation of Integrated Resource Framework Test Sites

3 PRODUCING THE MAPPING INFORMATION

3.1 In all four test sites the starting point for IRF was to build an aggregate picture of health and social care activity, together with their associated costs, using NHS and Local Authority data. This mapping exercise began by looking at 2008/09 data. It was anticipated that mapping of cost and activity data across both partners would create a process which could be repeated more easily in subsequent years, thus adding the possibility of analysing trends over time.

3.2 In practice, at the outset of IRF, one test site (Lothian) elected to build a picture of resource use and costs at the level of individual patients and clients. The other three test sites took the decision to map costs and activity at the level of general practice, locality and CHP. Tayside test site decided in mid-2011 to apply a patient/client level approach to its data.

Progress with mapping NHS and Local Authority data

3.3 There has been progress in mapping health and social care data. All test sites undertook high level mapping of Local Authority and NHS cost and activity across health and adult social care. It was reported that this was the first time that mapping information across the partners had been undertaken and it was found to be helpful in building a more complete picture of activity patterns and the associated costs.

3.4 The first attempt at mapping 2008/09 cost and activity across health and social care was not straightforward and encountered a number of challenges. Some of these challenges were overcome, some were worked around and others continue to limit the accuracy, acceptance and use of the mapping data. These challenges and the steps taken to overcome them are discussed in paragraphs 3.5 to 3.25 below.

3.5 The process has been streamlined since the first iteration of the mapping and, for 2009/10, the Scottish Government and NHS National Services Scotland Information Services Division (ISD) have worked with partnerships to produce "all age", "over 65" and "over 75" health and adult social care mapped data for all partnerships in Scotland. This has reduced the workload burden for partnerships while ensuring greater consistency at the national level. The Scottish Government and ISD are currently working on the 2010/11 data for Scotland.⁹

Collating cost and activity data from different reporting systems

3.6 NHS and Local Authority systems for calculating costs and activity operate in different ways. Broadly speaking, Local Authority systems record direct costs charged for care, whereas NHS systems record activities and then allocate average costs to these. The mapping produced collated information from these different accounting systems based on a number of estimates and assumptions. This was reported to be useful in helping partners to make better sense of how resources are currently spent.

3.7 The IRF mapping has helped to collate these two cost structures into single documents and has therefore simplified analysis of partners' spending and activity. However, there remain challenges in apportioning costs across different systems. One of the most obvious differences between the systems is the way in which they allocate overheads - NHS overheads are included in the average costs applied to each activity whereas Local Authority overheads are normally accounted for separately. Resolving these differences to include costs on a like for like basis is not easy. These differences can be acknowledged and accepted when the purpose is discussing variation in spend in different geographical communities (e.g. the total spend on the 75+ population in the different local authorities with the NHS board). However, when the analysis goes a stage further, and discussions focus on potential transfer of resources between partners, then the different accounting systems become more problematic (e.g. in transferring spend from one partner to another, different ways of allocating overheads can be much more contentious.)

Different levels of mapping

3.8 Stakeholders reported that the mapping information was built on assumptions and estimates which make practical sense at a macro level but become less robust as the data is broken down to more detailed levels. Test sites have each seen value in the mapping process but reported that the main benefits were achieved by mapping health and social care information down to CHP level. At this level, assumptions about the mixed approach, of breaking down overheads and building up patient level costs and activity, were generally accepted by partners as reasonable. Moving to more detailed information started to raise questions of credibility and confidence in how community health and social care costs are produced and applied.

3.9 Currently, the mapping data can be used to examine hospital-based care and community-based care at NHS Board level, CHP level, and for some activity at the level of GP practice. Hospital based activity is built up from individual patient records and can be analysed by geographic area, GP practice or age grouping. A large proportion of hospital costs can be mapped at an individual level using centrally collected SMR and prescribing data. This enables 67 percent of NHS Board expenditure to be mapped at individual level, a proportion that increases to 78 percent for people aged 65+ years, and 80 percent for those aged 75+ years. The community based activity delivered by the NHS can be analysed at CHP level but becomes less robust when broken down below CHP level.¹⁰ Similarly, Local Authority spend is only available at an aggregate level by CHP.

3.10 When looking at the costs for a whole NHS Board area, all health costs and Local Authority spend should be included. This eases the pressure resulting from any differences in the way that costs are apportioned. Test sites reported that producing and analysing mapping data at this level was robust and useful.

3.11 The process of producing cost and activity data across Local Authority areas within an NHS Board was susceptible to anomalies wherever health costs could be skewed by location of Board-wide facilities located within a council area. For example, Ayrshire and Arran found that the costs of Allied Health Professionals (AHPs) were allocated to hospitals rather than to the communities in which the AHPs were active. This weighted costs more heavily in council areas where the hospitals were located. In another example, Dundee stakeholders reported that many residents do not live in the same geographic locality as their general practice.¹¹ Test sites reported that undertaking the mapping process had proven helpful in creating a better understanding of how to produce aggregated NHS Board and Local Authority data to take account of the different geographical boundaries.

Data sharing and confidentiality

3.12 Respondents in the e-mail surveys noted frustration at the lack of data sharing between the key stakeholder organisations. Stakeholders cited lack of willingness to share data and lack of consistency in available data as the main challenges for data sharing. One example of work to address this is in NHS Tayside where the Chief Executive is now taking forward work to establish a generic data-sharing agreement.

3.13 Test sites reported that information on health community based activity and costs are underdeveloped across Scotland. GP community health activity data, such as district nurse contact hours, is not yet included in the IRF dataset.¹² Test sites reported that some GPs are reluctant to release this data in some cases because of concerns that resource use profiles will enable individual patients to be identified. They require detailed information about the use to which such data will be put and the steps which will be taken to anonymise the data, in order to agree to its release. The Scottish Government and ISD recognise that this is an important element in shifting the balance of care and are working to improve the information available. Options to do this are under review, including using sample data from a subset of general practices.

3.14 It was reported that data protection presented significant challenges, both in terms of Local Authorities passing data to the NHS Board, and the NHS Board returning mapped data to Local Authorities. For example, the need to define individual agreements between NHS Lothian and each of its four Council partners created considerable delays during the mapping process. It was reported that detailed national guidance in this area would be desirable to expedite data sharing agreements in the context of resource integration in future.

3.15 As described in 3.5 above, the Scottish Government is supporting NHS Boards and Local Authority partners as they work to map health and social care cost and activity at as granular a level as possible. ISD and Scottish Government Health Analytical Services Division (ASD) are supporting this work with the aim of adding impetus to the mapping process and making a stronger case for consistency within and between partners. All NHS Boards have agreed to participate in developing a standard patient level costing methodology, overseen by a costing advisory group convened for this purpose. NHS Boards and their Council partners are being encouraged to establish local costing groups to ensure consistency within partnerships.

Different approaches to mapping in the test sites

3.16 The mapping process was not routinely undertaken by partners so additional resources were required to manipulate data from different local systems. This meant that test sites had to make decisions about the point at which further mapping detail no longer justified the additional effort required. Three related factors affected this: (i) the resources available to produce the mapping data; (ii) the credibility of the mapping data (as described in 3.9); and (iii) the potential for the mapping data to affect resource realignment.

3.17 Analysing information in greater detail - to a patient level for all health and social care - was reported to be very time intensive. Therefore, Ayrshire & Arran, Highland and (in the first instance) Tayside undertook this level of "drill down" mapping only for specified groups of patients associated with the IRF pilots.¹³

3.18 In contrast, Lothian's decision to map at the patient level for all services took longer and required a significant commitment from staff involved. Local partners maintain that this approach holds considerable potential for a much more detailed understanding of cost and activity. Partners in Lothian are now working to show how this detail can be translated into better decision making, more efficient and effective service provision, and improved outcomes for patients. Another challenge that Lothian has worked to overcome is how to take the large volume of data created by this system and present it in a format that is simple enough to be useful. While this work remains ongoing, service planners responded positively when presented with the level of detail about costed care pathways which this work has produced.

Local flexibility and national coordination of mapping data

3.19 Test sites reported that flexibility to develop local systems for working out costs was useful in helping test site partners to assume ownership of the mapping process and of the data produced. As noted elsewhere in this report, local stakeholders consistently indicated that acceptance of the validity of the mapping data was very important. Flexibility to build cost data in a way that made sense at a Local Authority and NHS Board level was therefore reported as a positive element of the IRF approach.

3.20 Flexibility to design local approaches to mapping enabled this to best fit the expectations and uses of each test site. However, it also makes it more difficult to build a consistent picture of health community care and Local Authority social care costs across the country. This may complicate analysis of activity and costs that cross council or NHS Board boundaries and would undermine attempts at national benchmarking. It also requires more resources for each area to build local systems for the mapping if they do not have a template to follow.

3.21 When the mapping was carried out as a one-off exercise for the first phase of the IRF, the local approach was reported to be appropriate and the mix of methods justified. However, the mapping is now being repeated on a more regular basis, thereby strengthening the argument for standardisation. By making use of the learning from the mapping approaches used in each test site, the Scottish Government is working with ISD to simplify and prioritise the information gathered. This holds the potential to increase the efficiency with which mapping data can be produced, and the ease with which it can be compared, within and across sites.

3.22 Whether the mapping is carried out using local systems or national templates, the information is currently limited to data already captured by existing data systems. It was reported that, if an integrated resource framework was being designed from scratch, then it would be reasonable to design integrated systems focused on gathering information relevant to resource planning in a similar format across all local partners. However, test sites considered the cost and complexity of developing a new system specifically for this purpose, and transferring from individual accounting systems into a new information framework, to be prohibitively expensive and the process unnecessarily disruptive.

Potential to make the mapping more comprehensive

3.23 Examining the resources committed by health and social care partners represents a significant shift towards more integrated service planning and delivery. However, the cost and activity data included in the IRF mapping did not include public health activity which was reported to have a substantial effect on demand into the system. The IRF mapping focuses on making best use of resources within the health and community care system, but remains susceptible to demand pressures (obesity, alcohol problems, drugs misuse etc.) which are not currently measured within the mapping information. Stakeholders reported that there is scope to extend the analysis to look at, for example, demographic pressures, and future cost implications based on assumptions about cost of illness. For example the Scottish Government Alcohol and Drug Partnership (ADP) team has commissioned a study of whether the IRF can be used to produce information on the activity and cost of services used by substance misusers in Dundee City ADP. NHS Board and Local Authority social work activity datasets will be used to map partnership expenditure to individual and aggregate population levels.

3.24 IRF test sites also identified some additional partners whose work has an impact on health and social care (e.g. education and police departments). Test sites were aware of the potential value of including a broader range of partners whose work can significantly affect the level and location of demand into health and community care systems. The significant challenges posed by attempting to improve integration across primary care, secondary care and Local Authority community care meant that this was the focus of initial pilot work across the test sites. However, partners repeatedly acknowledged that future work on integration needed to more proactively involve a wider range of partners. The work in NW Perthshire to develop consumption budgets for localities seeks to address this challenge (see Chapter 6).

3.25 The IRF mapping focussed on making best use of existing information in order to limit the extra work required. However, this limited the mapping to the quality of existing data collection methods. This potentially skews analysis on to what has traditionally been measured and may miss significant factors which are less easily quantified. One of the main concerns reported by stakeholders throughout the evaluation was that the mapping information does not include quality and outcome measures alongside the cost and activity data. The IRF pilots recognised the need to build jointly agreed quality and outcome indicators into the aggregate descriptive activity and cost data, but had not achieved this by the time of the final evaluation.

"There is an obsession with beds as measure of capacity...but we don't understand capacity in the community well. The information systems are less well developed in the community, and talk about patient outcomes doesn't stand up to the same level of scrutiny."

Key findings

• There has been progress in mapping health and social care data. All test sites mapped cost and activity across health and social care. It was reported that this was the first time that mapping information across the partners had been undertaken and that this was helpful in building a more complete picture of activity patterns and the associated costs.
• Flexibility to design local approaches to mapping enabled this to best fit the expectations and uses of each test site. However, it also made broader comparison across NHS Boards and Local Authorities more difficult.
• A large proportion of hospital costs and prescribing costs can be mapped at an individual level using centrally collected SMR and prescribing data. However, detailed information on community health and social care costs is not yet standardised or centrally collected raising questions of credibility and confidence in how data are produced and applied.
• Mapping at a patient level has raised significant data protection challenges. In order to share address this Local Authorities, NHS Boards and GP practices require clarity on the use to which such data will be put, and the steps which will be taken to ensure anonymity and confidentiality.
• The IRF cost and activity data did not include public health activity which was reported to have a substantial effect on demand into the system.