Intended for healthcare professionals

Education And Debate

Glycated haemoglobin and metabolic control of diabetes mellitus: external versus locally established clinical targets for primary care

BMJ 1995; 310 doi: (Published 25 March 1995) Cite this as: BMJ 1995;310:784
  1. Chris Butler, lecturera,
  2. John Peters, consultant physicianb,
  3. Nigel Stott, professora
  1. a Department of General Practice, University of Wales College of Medicine, Llanedeyrn Health Centre, Cardiff CF3 7PN
  2. b Department of Medicine, University Hospital of Wales, Cardiff
  1. Correspondence to: Dr Butler.
  • Accepted 12 January 1995


Objectives: To examine current targets for glycated haemoglobin as a marker for metabolic control in diabetes mellitus in relation to datasets from several areas, and to consider whether target setting could be improved.

Design: Data collected from enhanced care records of general practices for a representative community based sample of people with diabetes.

Setting and subjects: 3022 people with diabetes on the lists of 37 general practices (total list size 222 550) in South Glamorgan in 1992; samples of glycated haemoglobin had been processed at two laboratories with different methodologies and reference ranges.

Main outcome measures: Last glycated haemoglobin level measured in subjects for 1992 and published data from other studies considered in relation to existing goals and standards for the metabolic control of diabetes.

Results: An ascertainment rate for people with diabetes of 1.36% was obtained. The rate of data capture for haemoglobin A1 was 75.7%, and the mean level for study samples was 10.5% at one laboratory and 10.0% at the other (similar values to those of comparable studies). These mean levels of haemoglobin A1 in representative populations of people with diabetes are poor or very poor according to published standards, including those of the British Diabetic Association. These findings are set in the context of the psychology of goal setting and performance in complex clinical situations.

Conclusion: Targets for clinical care that are set in the absence of normative data and local feasibility assessments should be treated with caution. Targets are more likely to enhance health care if target setters recognise the importance of psychological aspects of goal setting and motivation.

Key messages

  • Key messages

  • Prevailing “top down” approaches to clinical target setting ignore principles of patient centred care and proper consideration of goals as motivational tools

  • Effective targets are like to be attainable in the short term, believed in, negotiated, and flexible

  • Those who are expected to achieve the targets should undergo the necessary skills development, believe in their capabilities, be properly supported, and receive feedback


A plethora of charters stating explicit performance targets for a wide variety of services has appeared recently. This preoccupation is reflected in health care by the precise targets set by the Health of the Nation1 and the growing number of guidelines for managing diseases that contain specific goals and targets.2 3 Few have challenged the assumption that setting targets for health care workers will result in long term benefit for patients, and the uncertainties associated with clinical standards apply equally to clinical targets.4 5 Despite the present climate of multidisciplinary cooperation, few target setters have drawn on the experience of other disciplines. Theory and experience from the field of psychology, for example, suggest that the relation between goal setting and performance in complex settings is not a simple one. Indeed, setting goals may have a negative effect on performance for tasks that are intrinsically interesting and require complex strategies for achieving success.6 7 8 9 Moreover, some target setters may be operating counter to the spirit of their own disciplines. Patient centred principles regarding clinicians motivating their patients are often ignored in the relationship between target setter and clinician; the first some clinicians hear about clinical targets is when they receive uninvited, printed instructions through the post telling them what to do.

Two large, well publicised trials have shown that fewer microvascular complications occur with better metabolic control in people with insulin dependent diabetes.10 11 While poor metabolic control remains one of several important risk factors for the complications of diabetes,12 13 these trials have strengthened the conviction that the burden of morbidity and mortality can be lightened by improved care. The outcome of a large trial assessing the relative contribution of metabolic control to complications in people with noninsulin dependent diabetes is awaited.14 It seems reasonable, however, to assume that achieving better glycaemic control is also worth while in this group as the mechanism of microvascular damage is thought to be the same as in people with insulin dependent diabetes.15 The question of optimal metabolic control is therefore now in sharp focus.15 16

Measurement of glycated haemoglobin is the most widely used method of evaluating long term glycaemic control.17 Doctors' awareness of this measure has been shown to be associated with better glycaemic control for their patients,18 and patients are invited to evaluate their own achievements in controlling their diabetes in terms of this measure.19

Measuring glycated haemoglobin, the “gold standard” method for assessing glycaemic control, is therefore fixed firmly in the sights of the clinical target setters. In its advice to general practitioners, the British Diabetic Association offers targets for haemoglobin A1 and A1c with the proviso that they may have to be adjusted according to the variation in reference ranges of different assays (see table I).3 The Royal College of General Practitioners states that a level of haemoglobin A1 of less than 9% should be the goal for those with insulin dependent diabetes and those with non-insulin dependent diabetes who are aged under 65.20 The St Vincent Declaration's guidelines for children and adolescents with diabetes state that glycated haemoglobin should be “around the upper normal range, not lower than the mean of normal range (indicating hypoglycaemia) and not more than 1-2% above normal range.”21 The American Diabetic Association has recently abandoned its precise categories of control (“good,” “acceptable,” and “poor”) in favour of this more general approach (see table II).19


Target percentages for ranges of haemoglobin A1 and A1c recommended by British Diabetic Association3 in assessment of metabolic control in people with diabetes

View this table:

Target percentages for levels of haemoglobin Alc recommended by American Diabetic Association19 in assessment of metabolic control in people with diabetes

View this table:

Some target setters try to avoid the pitfalls of absolute values and categorise control according to standard deviations of haemoglobin A1 values around the non-diabetic mean. A value less than two standard deviations above the non-diabetic mean has been classified as “good control,” less than four “acceptable control,” and more than four “poor control.”22 “Good control” has recently been redefined, without explanation, to include “three standard deviations above the upper limit of mean value for the reference range.”13

We evaluated these published targets by comparing them with real data derived from a large group of patients with diabetes and with other published data-sets. In the light of this and the experiences of our programme of audit enhanced care for people with diabetes in South Glamorgan, we considered possible improvements that could be made to the process of setting clinical targets.


The development of our district-wide database for the audited care of people with diabetes since 1988 has been described elsewhere.23 Important aspects of the database are that it (a) is district-wide and represents one of the largest British datasets of community based care of people with diabetes, (b) incorporates a cycle of ongoing annual audit in parallel with continuing education, and (c) has shown a year on year improvement in the recording of process and outcome measures to levels higher than in most comparable studies.

The dataset presented here included all people with diabetes on the lists of the 37 participating general practices in 1992 irrespective of whether their care was largely hospital based or general practice based. One haemoglobin A1 level per person with diabetes (the specimen taken nearest to the annual review) is included in the data.

Assays were performed at two hospitals. Laboratory 1 used agar gel electrophoresis (Corning, normal reference range 5.7%-8%) and laboratory 2 used cation exchange chromatography (Helena, 5.5%-7.7%). We compared our data in South Glamorgan and datasets from other areas to published targets for glycated haemoglobin.


By the end of 1992, 51 of a possible 74 (69%) practices in South Glamorgan were participating in the programme, and 37 of these practices completed an annual audit. This represented a combined list size of 222550 patients, which included 749 people with insulin dependent diabetes and 2273 people with non-insulin dependent diabetes. An ascertainment rate for diabetes of 1.36% was therefore obtained.

In the 1992 annual audit, levels of haemoglobin A1 were recorded in 446 people with insulin dependent diabetes and 1789 people with non-insulin dependent diabetes. When calculating data capture, we excluded eighty subjects in the paediatric age group (all with insulin dependent diabetes) as their metabolic control was managed almost exclusively in hospital; the capture rates were therefore 65.7% for people with insulin dependent diabetes and 78.7% for those with non-insulin dependent diabetes (75.7% overall).

Table III gives the overall mean level of haemoglobin A1 and the mean levels for each type of diabetes obtained at each of the two laboratories in South Glamorgan. Figure 1 illustrates the skewed distribution of haemoglobin A1 levels obtained at each laboratory. Figure 2 gives the mean level of haemoglobin A1 for these two South Glamorgan datasets and for four other British studies24 25 26 27 together with the normal reference range for the British Diabetic Association's targets and the normal reference ranges quoted for each assay used. South Glamorgan's laboratory 2 was the only laboratory to use cation exchange chromatography—the rest used agar gel electrophoresis.


Mean (SD) percentages for haemoglobin A1 and number of samples by laboratory and type of diabetes

View this table:

Distribution of levels of haemoglobin A1 obtained at two laboratories in South Glamorgan. Laboratory 1, with samples from 1326 patients, had mean of 10.5% (upper limit of laboratory's reference range 8%); laboratory 2, with samples from 909 patients, had mean of 10.0% (7.7%)

The mean level of haemoglobin A1 quoted for the Ipswich study is for a cross section of people with diabetes three years after a changeover to an integrated system of diabetes care based in a centre for diabetes.27 The data from Oxford are for people with diabetes aged 60 and over,26 and the data from Trowbridge were obtained from a community-wide screening programme.25 The approach in Royal Tunbridge Wells was cross sectional and general practice based.24

In the Ipswich study the normal reference range was obtained by calculating two standard deviations around a mean for 150 subjects; it is not stated how these subjects were selected. The normal range used for South Glamorgan's laboratory 1 is based on the frequency distribution of values obtained from 61 laboratory workers who were not known to have diabetes. Since values did not follow a normal distribution, the use of standard deviations about the mean to calculate this normal range was thought inappropriate. (Targets similarly based on standard deviations about this mean may be open to question on grounds of statistical validity. Elsewhere, the use of laboratory normal reference ranges as quoted for the assays used are generally still derived by calculating two standard deviations about a non-diabetic mean.)

Figure 2 illustrates the well established methodological and statistical difficulties inherent in making accurate and meaningful comparisons of data on haemoglobin A1.28 29 In each study the mean level of haemoglobin A1 was well above its local normal range, however this was calculated, and a long way from “acceptable” according to published targets. It is apparent that, even if haemoglobin A1 levels could be manipulated somehow to take into account different methodologies and reference ranges, at least half of all patients in these large British audits of diabetic care would still have metabolic control that was classified as poor or very poor according to the criteria of the British Diabetic Association and others. Roughly the same proportion would still be rated as having poor metabolic control if four standard deviations above the nondiabetic mean was taken as the cut off point for having acceptable control.


Mean level of haemoglobin A1 for six datasets and normal ranges for each dataset compared with reference range for British Diabetic Association's targets

The highly selected and motivated group of young diabetic volunteers in the tightly controlled arm of the diabetes control and complications trial, supported by extensive education programmes and monitoring in a research milieu, achieved a mean level of haemoglobin A1c of around 7%.10 This value lies on the border of the British Diabetic Association's categories of acceptable and poor metabolic control (the non-diabetic mean for the diabetes control and complications trial assay and the British Diabetic Association's reference range are almost exactly the same). Fewer than 5% of patients in this sophisticated trial maintained an average level of glycated haemoglobin that the British Diabetic Association would classify as good.

In no large scale clinical study (community based or experimental) did investigators come close to achieving metabolic control which would be considered adequate according to published targets.


The findings of this study suggest that existing targets for the metabolic control of people with diabetes mellitus, when viewed in the light of normative data, are impractical for those involved in the day to day management of patients. Is there a better way of setting targets than the prevailing “top down” approach, in which national expert bodies may set targets that ignore the realities and skills of clinicians in many different and complex settings? Are locally established targets perhaps more valuable than national targets?


Goal setting is essentially a strategy for enhancing performance, and several considerations about goals and human motivation are relevant to the process of setting clinical targets. Firstly, at the best of times unrealistic goals will produce repeated failure that can eventually weaken motivation,6 and setting very difficult goals for people who are already under great pressure will simply worsen discomfort rather than improve performance.7 It is an established principle in the field of individual behaviour change that goals are more likely to be achieved if they are short range and attainable and if the easiest ones are attempted first. Research into “self efficacy” suggests that the inclination to engage in challenging activities is affected by an individual's beliefs about what he or she can do.6 Seemingly impossible goals may cause people to perceive the situation as uncontrollable, and their performance suffers. This takes its toll on what they believe they are capable of. Thus, those wishing to motivate others need to pay attention to raising people's beliefs in their competence, and they should try to structure situations in ways that are likely to bring success.

Secondly, expectancy theory suggests that people are more likely to do things if they believe in their value and that a given amount of effort will achieve them.30 Thirdly, neither goals alone nor feedback alone motivates performance optimally; both ingredients are necessary.31

Behaviourists, often interested in short term results within narrow parameters, are inclined to a “black box” view of human motivation and see activity as an outcome of conditioning; setting targets is an attempt to condition behaviour. External goals can be counterproductive, however, when applied to people doing intrinsically interesting activities,6 7 8 9 who may focus their energy on achieving goals instead of using initiative to focus on the aspects of excellence that are most appropriate to each unique, complex situation. There is a risk of making people dependent on external inducements, and the perceived “locus of causality” may change from internal to external.32 Deci and Ryan believe that the “self” is not simply the outcome of social evaluations and external pressures.9 Human beings, by nature, seek out optimal challenges and try to master and integrate new experiences. “Competence,” “autonomy,” and “relatedness” are innate psychological needs, and addressing these needs will promote a person's identification with the underlying value of the activity and acceptance of it as his or her own. This kind of behaviour change is called integrated regulation and will strike a chord with the clinicians who are interested in the self empowerment of their patients as a step towards lasting improvements in behaviour across a wide range of situations in life.

Indeed, many well researched principles governing the interaction between a clinician and a patient are relevant also to the relationship between goal setters and clinicians. Exploring the patient's perspective lies at the heart of the discipline of general practice.33 Studies of addictions also suggest that telling people what they should achieve may entrench resistance and undermine motivation, especially when the advice is uninvited as it may cause resentment and damage rapport.34 Negotiation about therapeutic goals35 and real valuing of patients' views36 seems to improve outcomes.


While it is too soon to be able to evaluate the South Glamorgan programme of audit enhanced care for people with diabetes in relation to improved outcomes for patients, the programme continues to grow and has shown an annual improvement in the recording of measures of the processes of care and the active participation of primary care teams.23 We believe that part of this success lies in attempts to operate within the spirit of appropriate goal setting and human motivation. In the programme, any explicit and implied targets for care are based on past performance and the group performance of peers. The possibility of achieving targets is therefore firmly established, and until recently the group had little awareness of the external targets promulgated by the St Vincent Declaration and its antecedents.

Clinicians from each general practice receive a confidential report enabling them to compare their year to year progress in caring for people with diabetes; they then attend a meeting where their achievements are fed back in a way that facilitates confidential comparison of the performance of local peers. Continuing medical and nursing education is integral, so that skills are enhanced and people feel more able to do what is expected of them.

The programme germinated from within general practice and involves hospital specialists, academics, and all members of the primary care team in a cooperative style of interaction. The process of audit, feedback, and continuing education tries to be responsive to the changing needs and new ideas of all the participants. We are aware of a developing sense of ownership of the data and any targets that are generated.

In the light of these data, the considerations about goals in human motivation, and the impressions gained from our practical experiences, we suggest that the following assertions are important for potential target setters:

  1. People are more likely to attain targets if they believe in the benefits of achieving them and perceive themselves as competent to achieve them.

  2. Just as clinicians should avoid dictating to patients, so should target setters negotiate locally with those who are expected to achieve the targets.

  3. Targets that are attainable are likely to be based on normative data derived from comparable care settings.

  4. The appropriate use of targets incorporates important elements of personal development and feedback.

  5. Since locally established targets are most likely to be congruent with these principles, they will probably be more effective motivating tools than those generated externally.

Funding in the early years of the South Glamorgan audit enhanced programme of care for people with diabetes mellitus was by the Primary Health Care Development Fund, the Welsh Scheme for Health and Social Research, and Novo Nordisk. The South Glamorgan Medical Audit Advisory Group and the region's family health services authority have funded the project since 1992, and the former continues to recruit practices to the scheme and organise opportunities for practical continuing education for each audit meeting on diabetes. The participating doctors, nurses, dieticians, chiropodists, and others with a specialist interest in diabetes determine the direction of the programme and keep it flourishing through their enthusiasm and their administrative, educational, and clinical efforts. The practice nurses of South Glamorgan deserve special mention for their dedicated care and follow up of patients.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.