Targets for dementia diagnoses will lead to overdiagnosis

Imagine the scene: you are about to review Mrs Jones in the memory clinic. When you saw her six months before you diagnosed mild cognitive impairment (MCI). But that was before your locality appeared on the wrong list on the UK government’s “name and shame” Dementia Challenge website,1 which publishes rates of dementia diagnosis for every clinical commissioning group (CCG) in the country. And it was before you viewed the accompanying video of Jeremy Hunt lambasting the “laggards” for their poor performance. And it was before your clinical commissioning group made raising diagnosis rates a priority, with financial strings attached, and started leaning heavily on you to make diagnoses—any diagnosis, it would seem; they just need to get the numbers up.

Would it really harm Mrs Jones to give her diagnosis an upgrade, to confirm her fears and call it dementia? Aren’t we supposed to diagnose it earlier these days? She’ll probably progress towards it anyway. As you call her in from the waiting room you wish that you could forget the inconvenient research that MCI often does not progress to dementia and can even improve.2

Medicine depends on a fundamental, unspoken agreement between patients and doctors. Patients make themselves profoundly vulnerable as they reveal their stories and subject themselves to examinations and investigations. Doctors promise to use all their skills to try to find an explanation for patients’ symptoms, by forming a diagnosis. It is implicit in this diagnostic process that the only factor influencing the decision to diagnose should be the best interests of patients. The idea that doctors should be motivated by self interest, such as personal or corporate gain, is abhorrent and undermines the basis of the relationship.

The making of a diagnosis is a key moment in a patient’s journey. It can bring great benefit by opening the door to effective treatments and support as well as giving a much needed explanation for worrying symptoms. But it can also bring great harm if incorrect.

Screening for dementia is not recommended in the UK3 and has been the subject of much debate in the last year.4 However, the setting of target rates for the diagnosis of any condition is a novel and unstudied concept that has now arisen in dementia care, and one which has largely gone unchallenged despite its potential to lead to substantial harm.

In its rhetoric on diagnosis rates, NHS England has been careful to use words such as “ambition” and “goal,”5 perhaps finding “target” too inflammatory, but it is the degree of pressure applied to achieving these goals that matters. CCGs listed among the 10 worst in the country will be desperate to remove themselves from such scrutiny, and at least one CCG has made the goal into a financial target. In response to a freedom of information inquiry, Herefordshire CCG has confirmed to me the incorporation of a target for dementia diagnosis rates as one of its local measures within the quality premium payment. The CCG has written to local general practitioners, stating, “The CCG needs to increase prevalence to 40% by April 2014 and 50% by April 2015—and will lose significant income if we miss these targets.”6 Accompanying the letter was a graph detailing the diagnosis rates of each practice in the CCG, with the implication that those with lower rates needed to work hardest to help the CCG earn this income.

The potential harms inherent in setting targets for diagnoses have not been analysed, despite the obvious danger that working towards targets can lead to perverse behaviours and outcomes, because the needs of patients are made secondary to the requirement to fulfil quotas. What is more, the figures themselves are highly questionable. The Delphi analysis used to estimate diagnosis rates7 is an extrapolation of data that are more than 20 years old, and the latest evidence indicates that there could be at least 200 000 fewer cases than previously thought.8 Estimates have inherent statistical error but the figures are never quoted with confidence intervals. At population level these intervals might be quite small, but when data intended for a population of 60 million is applied to a practice of 6000, the error bars must be magnified to a degree as to make the numbers meaningless—but they are quoted with an absurd degree of precision. My own practice apparently has a diagnosis rate of 126.7%,5 which can only be explained by overdiagnosis, highly erroneous figures, or both. If our practice figures are wrong by at least 30 percentage points, then this could also apply to every so called “laggardly” practice in the country.

Dementia is big business, and there are many vested interests that stand to benefit from a rise in the number of diagnoses.9 Currently the use of diagnosis target rates is proving to be a highly successful strategy for politicians and industry. If we fail to question their validity, the ethical concerns that underpin the making of a diagnosis and the danger of overdiagnosis, then we risk replication of this strategy in other areas of medicine. The diagnosis must always belong only to the patient, and never to government, policy makers, or commissioners. An urgent debate is needed before this gets out of hand.