Retrospective reimbursement

At present, NHS mental health services are primarily funded through block contracts agreed between commissioners and providers of care, or on the basis of levels of existing ‘inputs’ such as the number of beds (Reference Mason, Goddard and MyersMason 2011a). This method of financing, often termed retrospective reimbursement, offers little incentive for providers to deliver an efficient level of care (Reference Mason, Goddard and MyersMason 2011a) because it does not encourage them to control costs or increase output (activity levels). This is because an agreed fixed sum is paid regardless of the number of patients treated. Budgets are typically notionally fixed and renegotiated on the basis of past expenditure.

Retrospective payment typically allows the provider to receive funding for each day of care, each diagnostic test, or each procedure which is conducted (e.g. fee for service) and the full cost of each patient is therefore shifted to the payer (e.g. the insurer or the state). There is a clear link for the provider between their revenue and the use of resources per case. Under retrospective reimbursement, the payer refunds the actual costs incurred by each provider, which shifts risk to the payer.

Currency

The currency is the nationally agreed unit of activity attracting payment. It can take a number of forms covering different time periods, from an out-patient attendance or a stay in hospital (e.g. an HRG) to a year of care for a long-term condition, or in the case of mental health, a care cluster. The HRG currency in acute care is based on procedure codes (OPCS-4) and diagnosis codes (ICD-10). Clinical coders translate the patient’s notes into codes, which are then translated into HRGs.

Reference costs

The reference costs are a schedule each provider submits to the Department of Health detailing how much it costs them to provide each unit of the currency (Box 1). Reference costs for mental health services have been collected by cluster since 2010–2011. Costs are collected for three types of activity: admitted (in-patient) and non-admitted (out-patient) care, and initial assessments.

BOX 1

Hypothetical example of a national schedule of reference costs for mental health clusters

	Cluster 1 costs, £	Cluster 2 costs, £	etc.
Trust A	800	8000
Trust B	700	8500
Trust C	900	9000
Trust D	1000	7500
etc.
National average = tariff	850	8000

It costs Trust A £800 to provide Cluster 1 services and £8000 to provide Cluster 2, which is a more costly activity. It costs Trust B £700 and £8500 respectively to provide these same services. These are provider-specific costs that take into account both clinical costs (e.g. costs of diagnostics and monitoring interventions) and non-clinical costs (e.g. capital charges, food, cleaning and maintenance). All provider trusts submit these costs to the Department of Health, which calculates the national average cost for each cluster across all provider trusts. This is turned from a retrospective cost into a prospective fixed price – the tariff – the payment that all trusts will subsequently receive from commissioners. So Trust A will make a surplus on Cluster 1 because the tariff (£850) is higher than their cost to provide it, and they will break even on Cluster 2, whereas Trust C will make a loss on both clusters as their unit costs exceed the national tariff for both activities.

Reference costs are collected from NHS providers only. Private providers do not submit such costs, even though PbR applies to them when they provide care for NHS patients (Reference Mason and GoddardMason 2009a). A large proportion of mental health activity is carried out by private providers and their cost structures may look very different from those of the NHS (e.g. contributions to employees’ pension schemes).

The tariff is based on the national average of all NHS provider reference costs for a given currency, which is then turned into the prospective national fixed price. Whether the public and private sectors should face the same tariffs is a matter of some debate (Reference Mason, Street and MiraldoMason 2009b), but in England a uniform tariff is set for provision for NHS patients.

Table 2 shows the maximum review period for each of the 21 mental health clusters (determining when reassessments should occur), the indicative cluster unit costs and cluster costs per review period, and the number of patients in each cluster. As expected, the unit costs per cluster increase as illness severity and patient need increase for particular conditions. Commissioners will pay providers monthly for each patient in each cluster.

TABLE 2

Mental health clusters, cluster review periods, indicative costs and patient numbers for 2011–2012^a

Although mental health services do not yet have a fixed tariff, these costs provide an indication of the data that will be used and the method that will be applied to generate the tariff.

Tariff

The tariff, which has been delayed, will be calculated annually as a weighted average of admitted and non-admitted care and initial assessments. The clusters are designed to be independent of setting, thus providing an incentive to treat people in the least restrictive but also the most cost-effective setting. This strengthens incentives to shape care pathways and to keep patients out of hospital, since providers will only be able to make a surplus if they minimise the more expensive inpatient costs relative to treating more patients in an out-patient setting.

Acute services had a decade of collecting and refining their costing processes before tariffs were introduced. For mental health the timescales of moving towards a national tariff have been considerably shorter and this may ultimately prove to be unwise with respect to the implementation of PbR. To date, there is evidence of very wide variation in costs at cluster level both within and across providers (Healthcare Financial Management Association 2012), suggesting that validation of the costing processes and consideration of the appropriateness of the clusters for costing purposes are required. There will inevitably be variation in data quality due to variations in provider efficiency, differences in service delivery models, the quality of services, the level of local authority input, as well as historical and procedural factors. There is also significant variation between providers in information technology (IT) systems, which adds complexity. Very few mental health providers, compared with acute providers, underpin the costing data with patient level information costing systems (PLICS) (PriceWaterhouseCoopers 2012), which can generate more detailed information on cost drivers (Reference Street and MaynardStreet 2007) and provide an important understanding of the relationship between cost and price. Setting prices on the basis of poor-quality cost data could significantly compromise PbR.

As there is a long time lag between submission of reference costs by providers and the publication of the tariff in acute services, a number of adjustments are made to the reference costs, including a general inflationary uplift (applied each year reflecting assumptions about price pressures), and a national efficiency requirement to encourage performance improvement (Reference Mannion, Marini and StreetMannion 2008) (giving a net tariff uplift which in recent years has resulted in negative real growth in tariff prices) (Table 3).

TABLE 3

Tariff uplifts and efficiency requirements

The national tariff is adjusted by a market forces factor (MFF) to account for unavoidable differences in costs across the country (e.g. regional variation in wages and land and building costs). It generally costs more to run a hospital in London than it does in the north of England. The MFF is nationally determined and unique to each provider, and adds about 8% to the value of the tariff (Department of Health 2012a). A basic formula for the income of a provider is:

$\text{income}$

Efficiency

Evidence broadly supports the argument that PbR has been associated with reductions in unit costs through reductions in length of in-patient stay and increases in the proportion of day cases (Reference Farrar, Chalkley and YiFarrar 2010, Reference Farrar, Yi, Boyle, Mays, Dixon and Jones2011; Reference O'Reilly, Busse and HakkinenO’Reilly 2012).

Quality of care

For reductions in length of stay to be interpreted as increases in efficiency, it would require that quality remained unchanged or improved over the same period. Use of prospective payment systems may lead to ‘skimping’ on the quality and intensity of treatment, which may later result in readmission or higher mortality rates after discharge.

Reference Farrar, Yi and SuttonFarrar et al (2009) used administrative data to construct indicators of quality of hospital care: rates of in-hospital mortality, 30-day post-surgical mortality; and emergency readmissions. They found no change in quality after the introduction of PbR, although other unmeasured dimensions of quality of care may have been affected by it.

Although there is no direct evidence that fixed prices lead commissioners to focus on quality, the available evidence suggests that the focus of providers on cost reductions has not been to the detriment of quality (Reference Farrar, Sussex and YiFarrar 2007, Reference Farrar, Yi and Sutton2009; Audit Commission 2008a).

Volume of activity

There is evidence to support the hypothesis that the volume of activity did increase following the introduction of PbR (Reference Farrar, Sussex and YiFarrar 2007, Reference Farrar, Yi and Sutton2009; Audit Commission 2008a). However, it is not possible to disentangle the impact of the tariff from the effect of the rise in NHS funding during the same period, and ultimately other policies, such as waiting-time targets, may have been more important in driving growth in volumes (Reference Propper, Sutton and WhitnallPropper 2007).

Administrative costs

In terms of the costs of setting up and implementing the new funding system, evidence shows increased administrative costs per organisation of around £100 000 (Audit Commission 2005; Reference Mannion, Marini and StreetMannion 2006). This is due to higher costs of negotiation, data collection, monitoring and enforcement.

Given the administrative burden of managing the system, critics argue that PbR may tend to favour larger providers and squeeze out smaller providers such as local charities and third-sector organisations (Reference Grimwood, Edmonds and McGuinessGrimwood 2013), who play a crucial role in mental health service provision.

One administrative cost which will be particularly relevant to NHS mental health services is the time cost of grouping patients using the new clustering tool. Although clinical teams already familiar with HoNOS may not find the clustering a challenge, for others the task of gaining familiarity with the currency and building it into assessments will take time and have clinical resource implications.

Upcoding

The PbR tariff for HRGs ‘with complications’ is higher than for those ‘without complications’, suggesting a potential incentive for upcoding. Early evidence of ‘HRG creep’ suggested little deliberate upcoding, since coding errors were found to be random (Audit Commission 2008b). However, a comparison of HRG codings from English and Scottish hospitals before and after PbR was introduced in England found a higher rate of growth of spells in HRGs ‘with complications’ in the English data (Reference Farrar, Yi, Boyle, Mays, Dixon and JonesFarrar 2011).

In acute care the possibility to ‘game’ the system is potentially both easier and more difficult. On the one hand, coders record the diagnosis and procedure from the discharge notes, but do not directly allocate the HRGs. This is done automatically via an electronic algorithm on submission to the national data warehouse. Given that there are around 28 000 codes, upcoding would seem difficult, and therefore unlikely, en masse. On the other hand, the complexity of the system makes detection of upcoding also very difficult, making efforts to systematically upcode potentially easier and more appealing.

In mental health services, the incentives for something akin to ‘cluster creep’ are more direct. Scoring for the MHCT is performed by members of the clinical team rather than clinical coders. With only 21 clusters, it will soon become apparent to clinical teams what the monetary value is for each (see Table 2). There is a computerised algorithm to support decision-making, but this can still be manually overridden. Small changes to the HoNOS item score profiles will map through to the MHCT. Patients could be moved between clusters to directly influence provider revenue and the moves could be disguised as variations in clinical judgement. There will therefore need to be checks in place to audit the mechanics of the clustering process and the integrity of this coding pathway. This will be very difficult to perform routinely and with neutral impact, since it raises suspicion while aiming to remove deception, and commissioners will not readily be in a position to audit upcoding at individual patient level. A continual validation of the algorithm for the MHCT will also be needed to ensure a consistent grouping of patients with similar needs.

Equity of care

The evidence shows little support of concerns that PbR would lead to certain (more expensive) patients being ‘dumped’ or deterred from accessing care, whereas other (less expensive) patients were ‘cherry-picked’. Reference Cookson, Laudicella and Li DonniCookson et al (2012) found no change in socioeconomic equity of healthcare use between 2001–2002 and 2008–2009 for elective procedures, and some signs that equity might actually have improved slightly, since inpatient admission rates rose slightly faster in low-income areas than elsewhere. These results raise some optimism that socioeconomic disparities in healthcare utilisation may be impervious to changes in the provider reimbursement system.

Cross-subsidisation

If providers are able to decide the service mix, they might choose to treat more of the profitable HRGs, and even drop some that are less profitable. A recent study by PriceWaterhouseCoopers (2012) found that, although hospitals are potentially able to alter their service mix according to more profitable HRGs and divest themselves of loss-making services, this has in practice been mitigated by cross-subsidy from other sources of income, including non-tariff services (i.e. those services currently lying outside the scope of the PbR tariff, such as mental health). The report found that commissioners took funding away from mental healthcare because they found it difficult to assess what they could get for their money.

This report also showed that there is temporal volatility in tariffs, providing little basis for providers to invest/disinvest in profit- or loss-making services. It is rational for hospital managers to focus on the overall financial viability of the organisation as a whole and not concern themselves with particular ‘service lines’ that might be profit- or loss-making. Cross-subsidisation may help hospital managers prevent withdrawal of services that are seen as valuable to patients, particularly if there are dependencies between services. However, cross-subsidisation reduces or even negates incentive effects at the specialty level.

PriceWaterhouseCoopers concluded that the flaw in the system is that cost information underlying PbR is too weak. An implicit assumption of the national tariff is that all providers face the same cost structures and the same opportunities for making cost reductions (Reference Appleby, Harrison and HawkinsAppleby 2012). This has not been the case in practice, which means that the tariff is not an accurate price signal, as it is based on flawed costing data.

The PriceWaterhouseCoopers study found that PbR is increasingly ignored at a local level because the system has lost credibility. The Audit Commission (2008a) concluded similarly, finding that ‘the credibility of the tariff is an issue for primary and secondary clinicians alike’ (p. 46). Because of this lack of credibility, providers and commissioners are increasingly negotiating prices locally, and ‘working around’ the tariff by, for example, agreeing levels of activity within specified budgets – more or less the system in place before the introduction of PbR. Playing outside the PbR rules may be common, yet there has been little attempt by the Department of Health to assess the scale of non-compliance.