Why the NHS must wake up to the medical technology gap

If the UK is to match other advanced countries, major investment in technology is essential. But predicting real demand can be difficult. Joe Rafferty and colleagues may have the answer

Advances in medical technology create opportunities to improve care, outcomes and patient perspectives. But variations in regulation, decision-making and funding can cause uncertainty about cost and benefits, creating dilemmas for policy-makers and care planners.

To make informed investment decisions, it is crucial to try and unbundle some of these issues.

With increases in NHS funding dwindling after 2008, investment decisions over the next couple of years are critical.

The Wanless review characterised the NHS as a system with a history of successful cost containment and late adoption of technology. It called for major investment in high-end medical technologies - around 3 per cent of the annual health budget - in order to close 'both the unacceptable gaps in performance within the UK, and the considerable gaps in performance between the UK and other developed countries'.

The difficulty the NHS often faces is how to translate sensible aspirations into practical questions of how much and in what technology the investment should be placed.

The use of MRI and CT imaging technologies illustrates these dilemmas well. This article looks at:

• the appropriateness of current and planned MRI and CT rates using international comparators;
• an approach, using MRI and CT as examples, to calculating appropriate intervention rates through applying clinically indicated interventions to population-based diagnostic data.

This approach could be adapted to help inform the commissioning of a wider range of services, which, in turn, would help the NHS assess healthcare needs on the basis of

an objective and transparent set of clinical criteria.

Benchmarking within or across countries and sharing information can provide an evidence base that policy-makers can adapt to national circumstances.

In the NHS, it is normal to see comparative data on the performance of acute and primary care trusts or strategic health authorities. It is also routine to see trends in activity levels over time similar to those shown in figure 1, which depicts the number of MRI and CT scans in England between 1995-96 and 2004-05, projected to 2008-09 and beyond.

International comparison

This suggests a strong growth in the use of these scans, with both a doubling in the decade ending 2004-05. If the trend continues, a simple projection would suggest further substantial growth during the rest of this decade.

In contrast, Organisation for Economic Co-operation and Development data for the number of MRI and CT machines per million population for a number of advanced countries suggests the rate of diffusion has been much higher elsewhere. As figure 2 demonstrates, the number of MRI and CT scanners per million population in England is only 75 and 50 per cent respectively of the median value for this group of countries. Other research suggests MRI and CT scanners are used more intensively in some other countries than in England.

In other words, we have fewer machines per head of population and the machines we have are used for fewer hours per week.

To put this in context, if England continues to ride the trend for MRI as indicated in figure 1 and achieves an MRI rate of around 38-40 per 1,000 by 2010, it will be 20 per cent below the minimum intervention rate for any state in the US in the year 2000.

All the countries in the OECD sample had fewer than five MRI scanners per million population in 1990. But by 2003 only three countries, including the UK, were still in this position. At the same time five countries had more than 10 MRI scanners per million population - over twice the UK average.

In practice, many OECD countries may have achieved English 2003 intervention rates in the mid-1990s - but have continued to invest in these technologies.

Lagging behind

Although relatively crude, the analysis suggests that MRI and CT diffusion in the UK is 10-15 years behind OECD comparators. This conclusion could not be drawn from a traditional NHS benchmarking exercise, which would only show the wide variation around the national average in England, proving the usefulness of cross-country comparisons in an NHS seeking to compare itself with the best.

This OECD data is consistent with the broader Wanless analysis, and suggests that a significant investment in high-tech equipment, in this case MRI and CT machines, is needed. If this equipment is to be used efficiently, it in turn suggests the need for a significant increase in imaging rates above the current trend.

The scale of the gap between the MRI and CT rates in England and most other OECD countries, and the fact that the gap appears to be growing, could indicate growing differences in the way health services are provided elsewhere in the world. A range of other factors such as government policy, payment systems and clinical practice have an impact on imaging rates, and the investment may be linked to a drive to break the connection between outpatient and diagnostic assessment and the provision of inpatient care.

This suggests not only the need for a change in the way patients are imaged in England, but a more radical review of how we provide healthcare.

Comparative work we have done suggests - despite major increases in MRI and CT image rates over the last few years and projections suggesting they would continue to increase - that the NHS could face a major shortfall in appropriate, timely and accessible imaging capacity by the end of the decade.

However, although it is endorsed as helpful by bodies such as the OECD, a comparative benchmarking approach is always subject to the challenge that it is comparing the outputs from different systems and, as a result, the findings may not be robust enough, on their own, to determine policy.

The holy grail is to develop an approach that would independently verify the conclusions of the cross-country analysis. This led us to develop an approach we have

called the clinically indicated intervention rate analysis.

The aim was to develop an estimate of the clinically appropriate level of provision of MRI and CT scans nationally, if traditional capacity constraints relating to financial and cultural hindrances were removed. This would give a view of the underlying demand for imaging, based on patient need.

The objective was to develop a tool sensitive to individual populations, reflecting the morbidity in each population in the calculation of imaging requirements on a local basis, rather than applying a national average rate to a local population.

The major problem in developing a clinically indicated approach that could be applied nationally or to any SHA, PCT or practice population was the lack of a consistent database as to who was being imaged and why. This is because the analysis and coding is symptom-based, so the clinical/diagnostic problem presented is recorded (for instance, chronic headache) but the ultimate diagnosis is not recorded and aggregated nationally.

Using primary diagnosis data

This problem was overcome using hospital episode statistics (HES) data. Although it lacks information about the symptoms which led to patients presenting for imaging, the HES data contains the primary diagnosis of every person admitted on an inpatient or day-case basis.

An exercise was carried out, initially by local radiologists, subsequently extended in collaboration with the Royal College of Radiologists, to determine by primary diagnosis: whether either CT or MRI was clinically indicated in determining that diagnosis;

and for every patient actually diagnosed with the condition for which MRI and CT was deemed appropriate, how many other patients would be imaged only

for the potential diagnosis to be ruled out (the imaging ratio)?

The details of the approach are being written up separately; however, in simple terms this information can be married to the HES data for an SHA, PCT or practice to identify the number of patients who would have required an MRI or CT, based on their recorded diagnosis.

This figure is then factored up by the imaging ratio to arrive at the number of patients who estimates predicted would have needed an MRI or CT, based on the clinical data available.

The estimates produced using this method have been fine-tuned to reflect the impact of an ageing population, changing technology and policies to produce ranges based on the lower end of the clinical indications.

Current MRI and CT intervention rates per 1,000 population are 19 and 44 respectively. The study suggests that, based on clinical indications, MRI and CT imaging rates should be significantly higher than rates currently found in the NHS. It proposes that MRI rates should be within the range of 70 to 92 per 1,000 of population and CT rates should be within the range of 100 to 144 (giving central estimates of 81 and 122 per 1,000 respectively) by the end of 2008.

While these proposed increases may be considered radical, they would still see the population of England enjoying MRI and CT imaging rates well below many other developed countries.

Completely independently, in April 2005, the Canadian expert panel on MRI and CT published a report showing that the recommendation from the Ontario Association of Radiologists was for an MRI ratio of 60-69 per 1,000 population and for a CT ratio of 110-130 per 1,000 population - consistent with the results of the clinically indicated intervention analysis, and far higher than currently planned in England.

The key to success

If the NHS continues to follow trends, it will remain a late adopter - potentially lagging even further behind other countries. If it is to follow much of the rest of the world, and put into practice its policy of developing alternatives to hospital provision, it needs to do much more than produce a step change in access to imaging.

The key to success lies in an increase in appropriate imaging, but also in improved access to imaging and other diagnostics in primary and community settings - the right diagnostics in the right setting and sequence.

The wider importance is that it could be applied to many other areas of diagnostics or healthcare, forming the core of a population-based health needs assessment process that reflects local morbidity and engages clinicians in determining appropriate responses.

Local health service needs should be based on verifiable information and local clinical and patient views as to the appropriate response to those needs.

Joe Rafferty is director of commissioning and performance at North West strategic health authority, Paul White is lead consultant at Health Horizons and Gerard Marchand is managing director of GPT Health Strategies