Can NICE prevent diabetes?
- Correspondence to Professor Edwin A M Gale, Department of Diabetic Medicine, University of Bristol, Learning and Research, Southmead Hospital, Bristol BS10 5NB, UK;
- Received 23 October 2012
- Revised 23 October 2012
- Accepted 30 October 2012
- Published Online First 12 January 2013
The rising tide
Diabetes (diagnosed or undiagnosed) currently affects 7.4% of the UK population, and is projected to reach 10% by 2030.1 Conventional wisdom (with just a hint of moral censure) attributes the rising prevalence of diabetes to obesity and physical inactivity. It would however be just as true to lay the blame on increasing longevity, for glucose tolerance deteriorates with age, and we live much longer than we did. By their ninth decade, 55% of European males and 74% of females will have diabetes, undiagnosed diabetes or some form of glucose intolerance.2 Diabetes is also on the increase because of simpler diagnostic tests, lower diagnostic thresholds and health policies that reward primary care physicians for finding new cases. Last but not least, those with diabetes now live longer following diagnosis, whether due to better care or lead-time bias,3 thus adding to the prevalence of the condition (figure 1). The diabetes epidemic thus reflects increased longevity, better diagnosis of diabetes and longer survival, all of which might be considered desirable outcomes. Diabetes itself is highly undesirable, however, and elements of a national strategy are now in place to reduce its impact upon our population.
This strategy is outlined in recent National Institute for Clinical Excellence (NICE) guidance entitled Preventing type 2 diabetes: risk identification and interventions for individuals at high risk. This, as the title indicates, advocates identification of those at increased risk rather than population screening,1 and outlines a two-step strategy. Stage 1 uses objective measures and questionnaires to assess the risk of diabetes in three target groups: those at high risk by virtue of ethnicity, family or personal history; all those aged 40–74; and all those aged 75 or over. Individuals with low scores are offered general lifestyle advice and reassessed at 5-yearly intervals, whereas those at higher risk progress to blood tests for fasting plasma glucose or HbA1c in Stage 2, followed by interventions for those who test positive (box 1).
What is the guidance about?
The guidance sets out to identify people aged 18 and over at high risk of type 2 diabetes, and to offer them effective lifestyle-change programmes to prevent or delay the condition. It emphasises that there should be no upper age limit, ‘as everyone can reduce their risk, including people aged 75 years and over’.
Stage one of the policy involves use of risk assessment tools and questionnaires.
Those with a high risk score progress to Stage 2, which involves measurement of fasting plasma glucose (FPG) or HbA1c.
Those with FPG ≥7.0 or HbA1c ≥48 mmol/l (6.5%) are deemed to have diabetes if symptomatic, or following confirmation by repeat testing if asymptomatic.
Those with FPG 5.5–6.9 mmol/l or HbA1c 42–47 mmol/l (6.0–6.4%) are told they are at high risk and offered participation in an intensive lifestyle change programme together with annual blood checks.
This emerging national strategy has a number of important health benefits. It will increase awareness of diabetes and associated risk factors for heart disease, and it will promote healthier behaviours. It will alert many people to their personal risk of diabetes, will diagnose them earlier and will reduce the number of those with undiagnosed diabetes. It substitutes a risk-based approach for indiscriminate population screening which, as NICE acknowledges, is not justified4—although some might think that checking anyone past the age of 40 who comes your way sounds suspiciously like it.
There is however more than a suspicion that health promotion and advocacy of healthier lifestyles has overtaken the evidence. An evidence-based strategy should be based upon well-validated markers of risk, mechanistically linked to unwanted epidemiological outcomes and reinforced by clear evidence of benefit from early intervention. The potentially harmful consequences of such a policy (labelling, adverse effects of intervention, not to mention cost) will have been weighed carefully and discussed in relation to each risk category. In this instance, such expectations are not always fulfilled.
The risks of hyperglycaemia
Hyperglycaemia is a risk factor for both arterial and small vessel disease, but the risk threshold differs between them. A formal diagnosis of diabetes implies that both thresholds have been crossed, and was initially established by relating the risk of diabetic retinopathy to glucose levels following an oral glucose tolerance test (OGTT) in prospective studies. These studies demonstrated a clear threshold effect for retinopathy, but a continuously distributed risk for arterial disease; there is no obvious lower threshold to distinguish it from the non-diabetic population. A line had nonetheless to be drawn, and the resulting shadow zone between health and diabetes came to be known as impaired glucose tolerance (IGT).
The glycaemic thresholds for diabetes and IGT are qualitatively different. The risk of retinopathy is unifactorially linked to glucose levels, responds to glucose-lowering therapies, is sharply inflected above the diagnostic threshold, and is reproducible in widely differing populations. In contrast, the risk of macrovascular disease is multifactorial, responds poorly (if at all) to glucose lowering therapies, shows a smooth increase which begins well within the non-diabetic glucose range, and varies markedly from one age group or ethnic group to another. No surprise that the diagnostic threshold for diabetes has never been challenged, whereas the threshold for macrovascular intervention has given rise to endless controversy.
A further concern scarcely mentioned in the NICE guidance is that glucose homeostasis deteriorates quite markedly with age, whatever the measure used. In one study, 85% of elderly people without known diabetes who were found to have an HbA1c ≥48 mmol/l (≥6.5%)—and who would therefore be classed as diabetic by NICE criteria—did not have diabetes by glucose criteria; 34% were actually normoglycaemic.5 False positives apart, it is scarcely a triumph to ‘predict’ a condition that will affect the majority of octogenarians (by definition, among the healthiest members of our population), or to offer interventions whose benefit is untested in this age group.
Problems with measurement
It would help if we had better screening measures (box 2). The OGTT is a clumsy test, poorly reproducible, and heavily influenced by age; IGT (which can only be diagnosed by the OGTT) was never considered a satisfactory diagnostic entity. Measurement of fasting plasma glucose was introduced as a substitute for the OGTT in 1997, and impaired fasting glucose (IFG) was proposed as an alternative for IGT. In practise, both measures have been used interchangeably since then. The problem is that IFG correlates poorly with IGT. One analysis found, for example, that IGT was present in 751 people in the study population, as against 520 people with IFG; but only 219 (17%) tested possible on both measures.6 Integrated measures of glycaemic exposure appeared to offer a way forward, and HbA1c was next introduced as a diagnostic test for diabetes. As we have seen, this did not solve the problem, for up to 50% of those found to have diabetes by the OGTT are non-diabetic by HbA1c and vice versa.7 HbA1c is affected by age and ethnicity, and is on average 4 mmol/l (0.4%) higher in South Asians, Afro-Caribbean people and those over the age of 70.7 All this means that non-diabetic hyperglycaemia is now identified by three measures which associate weakly with one another and with cardiovascular risk. It may be convenient to lump all three measures together under the label of impaired glucose regulation as NICE has done, but this merely widens the net without increasing the value of the catch.
Diagnosis of Hyperglycaemia
Glycaemic status can be evaluated by measurement of fasting glucose levels, glucose levels 2 h after oral glucose or glycated haemoglobin (HbA1c).
Diabetes is indicated by fasting plasma glucose ≥7.0 mmol/l, a 2 h value ≥11.1 mmol/l or HbA1c ≥48 mmol/l (≥6.5%).
Many different terms are used to denote the range between normality (however defined) and diabetes. Impaired fasting glucose signifies the fasting range 5.5–6.9 mmol/l. Impaired glucose tolerance is defined by a 2-h glucose 7–11.0 mmol/l, and the equivalent HbA1c range (according to some authorities) is 42–47 mmol (6.0–6.4%). Values from 39–47 mmol/l (5.7–6.4%) are used in the USA to indicate increased risk of diabetes.
NICE uses the blanket term impaired glucose regulation to cover all the above abnormalities. Other terms include dysglycaemia, high risk of diabetes, non-diabetic hyperglycaemia or (especially in the USA) prediabetes.
Screening for early abnormalities is based upon the rationale that detection will allow us to prevent progression to frank diabetes. The evidence derives from randomised trials which show that glucose-lowering strategies such as lifestyle change and/or medication can delay progression to a diabetic OGTT,1 although sceptics might wonder whether earlier treatment of hyperglycaemia is really the same thing as diabetes prevention. Such interventions, although effective, are labour-intensive and costly, and have mainly been restricted to individuals with a mean age from 35–55 years.
The predominant rationale for preventing (ie, delaying) diabetes is that this will prevent cardiovascular disease. The excess cardiovascular risk conferred by IFG is however modest at around 20%, with an excess of around 10% for those who have both IFG and IGT.8 Furthermore, it is well demonstrated that glucose-lowering therapy has little or no effect upon cardiovascular risk in established diabetes,9 so there is little reason to suppose that it might prove useful at lesser degrees of hyperglycaemia. It is thus hard to see how the interventions advocated by NICE could be evaluated, let alone judged effective. A secondary reason for early detection of diabetes is that this might permit earlier detection and treatment of microvascular complications, but recent work suggests that the yield of treatable retinopathy in newly diagnosed patients within the existing system is already too low to justify more aggressive screening.10
In summary, the indications are that current practise is reasonably effective in detecting diabetes and/or cardiovascular risk. We can and should do better, but the excess cardiovascular mortality in diabetes is strongly associated with socioeconomic deprivation,11 and this gaping hole in healthcare provision will not be plugged by policy documents which will in practise, if not in intent, result in more intensive surveillance of those already within the system. Guidelines on glucose monitoring or glucose control in diabetes also show a clear bias towards more aggressive implementation than the evidence would appear to justify.12 ,13 Turkeys do not vote for Christmas, and diabetes specialists do not vote for a reduced emphasis upon glucose. Less cynically, the guidance is directed towards improved health outcomes, and seems prepared to cut an evidential corner or two if this promotes greater health awareness, healthier lifestyles and earlier detection of disease. And why not simplify the message if this enables the management of large numbers of people to be streamlined more effectively?
Such are the siren songs which tempt health policy makers into dangerous waters. The policy outlined by NICE, or so it seems to me, falls victim to evidence lag, good intentions and the desire for ‘one size fits all’ solutions. The policy is appropriate for young people at increased risk of diabetes; these have much to gain and little to lose. It probably does more good than harm in 40–74 year olds—but what about the elderly?
Old people seem to have been included in the programme almost as an afterthought, perhaps because healthy behaviours are beneficial at any age. Nonetheless, this is the point at which advocacy overtakes the evidence (box 3). This strategy confers disease status upon two-thirds of the extreme elderly, and to what benefit? In this age group, risk assessment tools are poorly validated, there little or no evidence to support screening and intervention for diabetes and there is no evidence at all for intervening in age-related hyperglycaemia. The mortality of those who progress to diabetes over the age of 70 is only marginally elevated,14 and there is no good reason to believe that aggressive implementation of glucose lowering strategies will affect this other than adversely. Greater health awareness is a good thing, and selective implementation of this policy will benefit some members of an increasingly spry elderly population, but what about the down side?
Early intervention in Hyperglycaemia: how useful?
The three proposed screening measures are poorly correlated.
The false positive rate is high, whatever the screening strategy.
The risk of diabetes and cardiovascular disease is asymmetric within the population, and ‘one-size fits all’ policies may be inappropriate.
The value of screening for diabetes, let alone prediabetes, has not been established.
Glucose lowering strategies have little impact upon cardiovascular disease in established type 2 diabetes, and have not been shown to be effective at lesser degrees of hyperglycaemia.
Age-related changes in glucose metabolism affect the majority of the old people. Treating this as disease exposes many members of the elderly population to unproved and potentially harmful interventions.
Opportunistic screening of those who come to medical attention is another example of the inverse care law: it will do little to alleviate the high impact of diabetes and cardiovascular disease upon the socially deprived.
Younger people can and do buffer excesses of therapeutic enthusiasm by the simple expedient of non-compliance: the elderly are more likely to have pills pushed down their throats by anxious carers. Any health information provided to older people which includes the word diabetes will almost certainly entail anxiety, increased supervision, the denial of simple pleasures and the infliction of additional medication. Side effects include nausea, diarrhoea and feeling dreadful (metformin), hypoglycaemia, not uncommonly fatal in this age group (sulfonylureas or insulin), weight gain, fluid retention and fractures (pioglitazone), pancreatitis (GLP-1 based therapies) and the oily faeces which have prompted a less elegant synonym for orlistat. For whose benefit?
In summary, this guidance has much to commend it. It sets out to mobilise our dwindling health resources towards a nasty disease with nastier consequences. The focus on prevention is welcome. The policy is well intended and thoughtfully constructed. On the utilitarian principle of the greater good for the greater number, it scores well. All it lacks, in sum, is a certain amount of common sense and quite a lot of the evidence. As David Sackett has said, ‘there are simply too many examples of the disastrous inadequacy of lesser evidence as a basis for individual interventions among the well’.15
Competing interests None.
Provenance and peer review Commissioned; externally peer reviewed.