How to improve variation in inappropriate admission performance

In the wake of the Health Bill’s royal assent lie the structural reconfigurations around strategic health authority and primary care trust clusters that have already commenced as part of the 2012 Operating Framework plans and the creation of commissioning support services.

However, the NHS is now in danger of suffering from temporary paralysis. During a transition period in which PCT management has been aggregated upwards and become more detached from previously-localisedresponsibility, there has been no handover to clinical commissioning groups as they remain in shadow form. 

Additionally, the infrastructure for monitoring progress of the Nicholson challenge, which used to span from SHAs to general practices, has been dismantled, leaving managers at all levels in the new hierarchy flying blind for the foreseeable future at a time when direct line of sight on performance transparency is most needed. 

Once the commissioning support units/organisations come online, some form of performance visibility will emerge again. But that is likely to take several months and will only start once CSUs are fully established and won’t address the full NHS hierarchy, as they will only be concerned with their CCGs.

What is urgently needed is a way to get to grips with the unwarranted variations that persist but go unnoticed because they are masked by the way data is aggregated.

Inappropriate admissions

A prime example can be found in management of inappropriate hospital admissions related to long-term conditions, diabetes in particular. These are admissions for which the primary diagnosis, i.e. the condition for which the patient was hospitalised, would not normally justify an admission. In some of these instances, it may turn out that the admission was in fact appropriate, but experience shows that the vast majority are cases that should have been treated outside an acute hospital setting.

The NHS as a whole should be concerned because these represent a significant share of the overall inappropriate admissions annually and have already cost the NHS in excess of £15 million over the first three quarters of the current financial year alone as illustrated in table 1.

The challenges in tackling this come from the current difficulty in accessing the existing data which can reveal where problems arise, together with the lack of dedicated staff who can work with stakeholders to take action to resolve the problem.  

On the question of access to the data, as SHAs and PCTs have begun to cluster, the problem of integrating reporting systems to provide relevant views has not been a priority. This means that although the data exists, not enough effort has gone into converting it into meaningful information.

The rate of inappropriate inpatient admissions associated with Diabetes illustrates the difficulties. Overall, the figures show reasonable stability across the board through time and, in particular, for the NHS North of England, as table 2 shows.

Regional variations

As table 3 (attached right) shows, the apparent stability in the North covers substantial variation between the different PCTs across the old SHAs. The aggregate level, suggesting steady figures at around 44 per cent, disguises wide differences in performance between the component organisations.

The figures contain both good and bad news.  There have been significant gains in several clusters, but NHS South Yorkshire’s position has deteriorated dramatically from 39.7 per cent to 48.4 per cent.  NHS Humber, conversely, shows what can be achieved - that a target level of 32 per cent is attainable.

The next stage is to drill down further into the data. Table 4 shows values for individual PCTs. Where, again, more granular data shows more variation locally. In particular, several PCTs show significant deterioration over a two-year period.

This is not just a northern phenomenon.  The same situation can be found across PCTs in other regions as table 5 demonstrates for the top 25 PCTs with highest adverse variation compared with last year.

Management has been moved upwards into clustered organisations but, as the examples above show, it is not enough to view data at that level, because it tends to average out serious levels of variability at local level, disguising variations from managers.

Addressing variations

To address the problem of unwarranted variation, PCTs and their successors need access to information on which they can base effective action and they need staff with the skills and the confidence to use it to drive discussions with providers and GP practices around changing their behaviours. 

Consolidation of PCTs into clusters, the delay in legislation around establishing CCGs and the current uncertainty around future commissioning support has left a void in local engagement with providers and GP practices. This is having a direct impact on achieving the potential savings required by the QIPP programme.

Returning to the example of inappropriate admissions associated with Diabetes and looking at sample PCTs from each of the four SHA clusters, the extent to which these patients are triggering unwarranted escalation of care and cost is immediately clear.

Figure 1 (see attached right) demonstrates the degree to which simple Diabetic patients with no complications across a PCT patch are admitted into hospital only to be discharged without having any procedure.  This may indicate an unwarranted escalation from primary care that deserves further investigation.

Figure 2 (right) highlights turnstile treatment in secondary care in the same four SHAs: a significant portion of the cases being admitted stay one day or less, which might suggest that the escalation to hospital care could have been avoided.  Avoiding hospitalisationmight have been significantly less costly.

To achieve this shift in care, PCTs and CCGs need to not only challenge acute providers to justify admissions but also examine why their practices are failing to prevent escalation of care. This requires even more detailed data showing performance within individual practices, underpinned by individual patient records, to facilitate discussions around care plans.

PCTs and their CCG successors must focus their effort on the practices where intervention will generate the highest pay off.  Segmenting practices by volume and value metrics is a useful way to determine where to concentrate their initiatives.  Figures 3 and 4 illustrates how simply this can be done showing that the top 10 in each category tend to be the same practices and usually account for the highest significant share of the patients and concentrated expenditure.

Undertaking this work requires good analytics to support local commissioning management teams and CCGs who can harness the information and act on it. 

At present there is a disconnection between what is occurring at the coal face locally and the aggregated PCT and SHA clusters currently overseeing it. This is aggravated by a lack of structured reporting and coordination of resources to act on it.

If not addressed quickly, this will result in a growing backlog of deferred savings having to be delivered by CCGs, even if the proposed commissioning support services are able to quickly provide them with the information they need when they are launched.