INTRODUCTION
As GPs we make referrals to consultants and allied health professionals every day. It is a necessary part of what we do. We know that we try to take patients from square one in primary care and pass them on to more specialised colleagues if they need to go further into the healthcare system. We try to do this well, and to make careful and well judged clinical decisions about whether we, or a more specialised colleague, will provide better care to a particular patient. We are proud of our gatekeeper function and our role in making sure that the right patients reach the right consultants. We are also proud of the converse which is our key role in guiding patients who do not need specialist care safely away from the cost and potential perils of specialised medical care. We can point to Barbara Starfield’s1 work that shows how effective primary care can be when done well, in terms of allowing specialists to work more effectively and in saving lives and reducing healthcare costs.
We also know that secondary care is the expensive part of any healthcare system. Primary care is cheap, and deals well with 90% of cases presenting to the healthcare system.2 The alternative view of this fact is that secondary care deals with the most severe 10% of cases, who are the ones ill enough to actually need treatment.3
The decisions GPs make about referral are crucial both for the patient’s progress through the system, and for their effect on the costs incurred by the system. Any clinical decision to refer a patient is unavoidably also a financial decision to spend NHS resources.4 NHS resources are finite and the demands on them are increasing. The service is currently facing ‘the Nicholson Challenge’5 to make huge efficiency gains. We now see the advent of GP commissioning consortia tasked with achieving these gains. We have many circumstances coming together that will make analysis of referrals, and their costs, an important part of running the NHS in the forthcoming years.
WHAT DO WE ACTUALLY KNOW ABOUT THE REFERRAL PROCESS ITSELF?
Is it a good and necessary process? Does it get patients who need care to the right place for that care? Is it best thought of as a barricade or as a conduit? Are GPs a bit too keen on their gatekeeper role, as the paper by Vedsted and Olesen suggests?6 Do we gatekeep too well, at the price of reduced sensitivity and a risk of diagnostic delay? Would GPs be better to think of themselves more as ‘system navigators’ as the recent King’s Fund report suggests?7 Are there many inappropriate referrals as this NICE report suggests?8
Criteria by which to answer the questions
The truth is that sadly the important questions above are currently unanswerable. The criteria by which we could judge a referral good or bad, relevant or irrelevant, appropriate or inappropriate are not yet defined. It is not clear who should judge the merit of the referral. Should the opinion of the GP, the consultant, the patient, the local consortium, or the referral management centre staff count more or less? Is guideline adherence any indicator of high quality medicine? How specific should the guidelines be?
There is some evidence from reviews of referrals either by a consultant reviewing the referrals they have received,9,10 or from a GP reviewing a series of their own referrals11 about quality as seen from particular viewpoints. But there is as yet no overall criterion available for measurement.
When we move up to analysis of overall referral rates we are struck by the variation between practices within an area.12 Further analysis will often reveal significant variation in referral rates between doctors within the same practice. The extent of variation between practices is becoming more widely seen as area-wide figures of numbers referred and costs incurred are distributed as part of commissioning processes.
As far as PCTs and consortia are concerned, the low referrers are the better referrers. But this judgement is largely based on financial criteria. We do not have any knowledge of whether high or low referrers are better or worse clinically. We do not know what set points of sensitivity and specificity the referral process should operate at. We do not know whether lower referrers are very specific, or whether they are simply missing pathology. We do not know if high referrers are over sensitive, or whether they are simply better at spotting significant pathology than their lower referring colleagues.12
EFFECT OF PRIOR EXPERIENCE
We do not know whether additional training in a speciality increases or decreases a GP’s referral rate to that speciality. What evidence we do have suggests that GPs who have had additional training in a speciality do more work in it, and generate more referrals to it. Perhaps their colleagues without the additional training are not even aware that there is something of which to be aware of?
WHAT WE CAN DO WITH CURRENT MEASUREMENTS
For practical purposes we currently have to make a pragmatic assumption that those referrers in the middle of the distribution are about right, and that we should look to explain outlying referral rates, both high and low. But no one knows what the right referral rates would be for an area, and there is no agreement about the criteria that could be used to define what ‘right’ would be.
We do know that there is much unmet need for both diagnosis and treatment of many conditions which, ‘GPs are ideally placed to spot’ but that they signally manage to miss. The King’s Fund sees this phenomenon as a significant failing in current primary care and an area for future development.7
WHAT CAN WE DO TO ALTER CLINICAL BEHAVIOUR AROUND REFERRALS?
It is far from clear how amenable to change an individual doctor’s behaviour around referrals is. The issues involved are complex and include questions about clinical knowledge, recognition of pathology, expected outcomes of the referral, responsiveness to felt or expressed patient pressure, and the doctor’s own tolerance for uncertainty. This is a complex mixture in which, besides relatively straightforward clinical considerations, the worries of the patient and the worries of the doctor may become equal and additive, and so reach threshold for a referral.
This year in UK general practice we are experiencing new Quality and Outcomes Framework indicators that encourage GP practices to get together and review their referral data collectively in an educational way, which hopefully will lead to improvements in care pathways.
There is some observational evidence to support such an approach from Evans et al13 and their work on this process in Torfaen, South Wales. They showed how, by using an educational approach, they could achieve some reduction of high referral rates, and some increase in lower referral rates.
The idea of using an educational approach seems more likely to be successful and engaging than blunt blocking instruments such as referral management centres in use in some areas.14
WHY OUR LACK OF KNOWLEDGE ABOUT REFERRAL DECISIONS MATTERS
The analysis of referral rates at levels of clinical appropriateness, clinician behaviour, and financial implications is still in its infancy. It is at the place where prescribing analysis was about 25 years ago. There seems to be a dearth of recent literature on how we should analyse referrals.
This lack is acute in today’s NHS. It matters because doctors have no yardstick by which to measure the value of their referrals. It matters because patients have no information on which doctors are better or worse at referrals. It matters because there is potential for errors of commission (excessive referring and subsequent interventions, incidentalomas, and so on) and errors of omission (failure to diagnose, failure to treat). It matters because different doctors cost vastly different amounts of NHS resources to run.4 It matters because if referral rates are not brought into some sort of alignment then it will not be possible to maintain a financially viable NHS.5
In short the answer to ‘What do we actually know about the referral process?’ is actually, ‘Not very much’. It is a historically hallowed process15 that we all do, but it is not clear whether we do it well or not. It is a cultural rather than a planned process. Given that this process is responsible for generating about 90% of NHS costs this ignorance is no longer supportable and there is a significant opening for some operational research in this area.
In the meantime we will have to hope that primary care trusts and subsequent GP commissioning consortia analyse referral rates gently, and with a good appreciation of the limitations of the current data both in terms of its accuracy, and in terms of the limitations of our explanatory framework about it.16
Notes
Provenance
Freely submitted; not externally peer reviewed.
- © British Journal of General Practice 2011
