Implementation of the automated Leicester Practice Risk Score in two diabetes prevention trials provides a high yield of people with abnormal glucose tolerance
- First Online:
- Cite this article as:
- Gray, L.J., Khunti, K., Edwardson, C. et al. Diabetologia (2012) 55: 3238. doi:10.1007/s00125-012-2725-8
- 841 Downloads
The Leicester Practice Risk Score (LPRS) is a tool for identifying those at high risk of either impaired glucose regulation (IGR), defined as impaired glucose tolerance and/or impaired fasting glucose, or type 2 diabetes from routine primary care data. The aim of this study was to determine the yield from the LPRS when applied in two diabetes prevention trials.
Let’s Prevent Diabetes (LPD) and Walking Away from Diabetes (WAD) studies used the LPRS to identify people at risk of IGR or type 2 diabetes from 54 general practices. The top 10% at risk within each practice were invited for screening using a 75 g OGTT. The response rate to the invitation and the prevalence of IGR and/or type 2 diabetes in each study were calculated.
Of those invited 19.2% (n = 3,449) in LPD and 22.1% (n = 833) in WAD attended. Of those screened for LPD 25.5% (95% CI 24.1, 27.0) had IGR and 4.5% (95% CI 3.8, 5.2) had type 2 diabetes, giving a prevalence of any abnormal glucose tolerance of 30.1% (95% CI 28.5, 31.6). Comparable rates were seen for the WAD study: IGR 26.5% (95% CI 23.5, 29.5), type 2 diabetes 3.0% (95% CI 1.8, 4.2) and IGR/type 2 diabetes 29.5% (95% CI 26.4, 32.6).
Using the LPRS identifies a high yield of people with abnormal glucose tolerance, significantly higher than those seen in a population screening programme in the same locality. The LPRS is an inexpensive and simple way of targeting screening programmes at those with the highest risk.
KeywordsImpaired glucose regulation Prevalence Primary care Screening Type 2 diabetes
Anglo–Danish–Dutch Study of Intensive Treatment in People with Screen Detected Diabetes in Primary Care
Impaired fasting glucose
Impaired glucose regulation
Impaired glucose tolerance
Let’s Prevent Diabetes
Leicester Practice Risk Score
National Health Service
National Institute for Health and Clinical Excellence
Walking Away from Diabetes
Type 2 diabetes is usually preceded by the ‘prediabetic’ state in which there is impaired glucose regulation (IGR), which includes impaired fasting glucose (IFG) and impaired glucose tolerance (IGT). Randomised controlled trials have shown that progression from IGR to type 2 diabetes can be prevented through lifestyle or pharmacological intervention . Identification of individuals at high risk of type 2 diabetes is therefore warranted, so that preventive action can be taken. Nevertheless, evidence from the UK is limited despite modelling studies suggesting that screening for IGR followed by appropriate prevention interventions is likely to be cost effective .
Risk scores offer a non-invasive way of identifying those at high risk for invitation to screening programmes. A plethora of risk scores have been developed and validated over the past 10 years . Four scores are available for use in the UK. The Cambridge Risk score was designed to identify undiagnosed diabetes  and the QDScore predicts 10-year risk of developing diabetes . The Leicester Risk Assessment Score and the Leicester Practice Risk Score (LPRS) are the first scores developed for use in the UK that detect undiagnosed IGR and type 2 diabetes in multi-ethnic populations [6, 7]. The Leicester Risk Assessment score is a questionnaire completed by members of the public without intervention from a heathcare professional or the results of medical tests. The score includes age, sex, ethnicity, BMI, waist circumference, hypertension and family history of diabetes. The LPRS is similar but is for use within primary care databases to rank those listed by risk, this score does not therefore include waist circumference as this is not routinely stored. These risk scores are now recommended by the National Institute for Health and Clinical Excellence (NICE) for risk identification in primary care in the UK .
Evidence of the use of risk scores in clinical practice are lacking. Two prevention clinical trials are currently being carried out in the UK that use a risk score for case identification. The aim of this paper is to describe the cohort of ‘at risk’ people that the score identifies and to report the yield of IGR and type 2 diabetes and compare this with a population-based method of screening.
The LPD study has been described in detail elsewhere . Briefly, this was a two-stage project screening for IGR using the risk score followed by a cluster randomised diabetes prevention trial. Forty-four practices were included in the study. The risk score was calculated for all those aged 40–75 years if white European or 25–75 years if South Asian excluding those with known diabetes and terminal illness. The top 10% of patients with the highest scores were invited for screening; in large practices the first 500 were initially invited.
The WAD study screened for those at high risk of diabetes followed by a cluster randomised trial to improve metabolic health and increase physical activity . Ten general practices were included in this study. The risk score was run on the practice database and the top 10% aged 18 years and over (excluding those with known diabetes and those taking steroids) were invited to screening.
Measurements and outcomes
Both the LPD and the WAD study used the same standard operating procedures for data collection at the screening visit and the assessment of type 2 diabetes status. All participants received a 75 g OGTT. Participants were categorised according to WHO 1999 criteria . Anyone who had an OGTT result in the diabetes range was recalled for a confirmatory test. In these studies IGR was defined as IFG and/or IGT. IFG was defined as a fasting blood glucose concentration of between 6.1 and 6.9 mmol/l inclusive and IGT as a 2 h blood glucose concentration of between 7.8 and 11 mmol/l inclusive. Those with type 2 diabetes were referred back to their general practitioner (GP) for standard care.
At the screening visit the following data were also collected: HbA1c, cholesterol, BMI, body weight, waist and hip circumference and blood pressure, as well as data on a number of self-reported questionnaires that have been described previously [10, 11].
The primary outcome of the screening phase of both studies was the proportion of people detected with IGR or type 2 diabetes. Although HbA1c was not used for diagnosis in these studies, we also report data on the following categories using HbA1c: 6.0–6.4% (42–47 mmol/mol), ≥6.5% (48 mmol/mol) and ≥6.0% (42 mmol/mol), and those suggested by the ADA: 5.7–6.4% (38–47 mmol/mol) and ≥5.7% (38 mmol/mol) . In addition, we also assessed the yield when using a fasting blood glucose concentration of 5.5–6.9 mmol/l to define IFG, as recommended by NICE . We also assessed the response rate to the invitation to screening.
The response rates and prevalences were compared with those found in the Anglo–Danish–Dutch Study of Intensive Treatment in People with Screen Detected Diabetes in Primary Care (ADDITION)-Leicester study . ADDITION-Leicester used a population-based screening approach in the same locality as LPD and WAD. Random samples of 40- to 75-year-olds (25–75 years for South Asian ethnicity) from 20 general practices were invited for an OGTT .
Means (SD) and counts (%) were used to summarise the data. Differences between the studies and screening approaches were assessed using the t test for continuous variables and the χ2 test for categorical. The prevalence (positive predictive value) of each glucose disorder was calculated for each study. All statistical tests are two-sided and p < 0.05 reflects statistical significance throughout.
Characteristics of those screened compared with those screened as part of the ADDITION-Leicester study
p value (LPD vs WAD)
p value (Total vs ADDITION-Leicester)
Age (years), mean (SD)
Weight (kg), mean (SD)
BMI (kg/m2), mean (SD)
Waist circumference (cm), mean (SD)
Systolic BP (mmHg), mean (SD)
Diastolic BP (mmHg), mean (SD)
Total cholesterol (mmol/l), mean (SD)
LDL-cholesterol (mmol/l), mean (SD)
HDL-cholesterol (mmol/l), mean (SD)
Fasting glucose (mmol/l), mean (SD)
Two-hour glucose (mmol/l), mean (SD)
HbA1c (%/mmol/mol), mean (SD)
Compared with the participants in the ADDITION-Leicester study, those screened using the risk-score approach were significantly older, more likely to be male, heavier, had higher blood pressure, glucose levels and HbA1c and more likely to be taking lipid-lowering and antihypertensive medications. Higher levels of total cholesterol were observed in the ADDITION-Leicester study, which is probably a reflection of the lower treatment levels (Table 1).
Percentage of abnormal glucose tolerance in those screened (positive predictive values) using OGTT
p value (LPD vs WAD)
p value (Total vs ADDITION-Leicester)
Fasting glucose 5.5–6.9 mmol/l 
36.4 (34.8, 38.0)
35.9 (32.6, 39.1)
36.3 (34.9, 37.8)
22.2 (21.2, 23.2)
7.8 (6.9, 8.7)
9.6 (7.6, 11.6)
8.2 (7.4, 9.0)
5.0 (4.5, 5.6)
22.4 (21.0, 23.8)
21.4 (18.6, 24.2)
22.2 (21.0, 23.5)
13.3 (12.4, 14.1)
25.5 (24.1, 27.0)
26.5 (23.5, 29.5)
25.7 (24.4, 27.0)
16.1 (15.2, 17.0)
Type 2 diabetes
4.5 (3.8, 5.2)
3.0 (1.8, 4.2)
4.2 (3.6, 4.8)
3.2 (2.8, 3.6)
Any abnormal glucose tolerance
30.1 (28.5, 31.6)
29.5 (26.4, 32.6)
30.0 (28.6, 31.3)
19.3 (18.3, 20.2)
Percentage of abnormal glucose tolerance in those screened (positive predictive values) using HbA1c
p value (LPD vs WAD)
p value (Total vs ADDITION-Leicester)
HbA1c 6.0–6.4% (42–47 mmol/mol)
35.0 (33.4, 36.6)
31.9 (28.6, 35.1)
34.4 (33.0, 35.8)
17.5 (16.6, 18.4)
HbA1c ≥6.5% (48 mmol/mol)
10.0 (8.9, 11.0)
8.9 (6.9, 10.8)
9.7 (8.8, 10.6)
5.2 (4.7, 5.8)
HbA1c ≥6.0% (42 mmol/mol)
45.0 (43.3, 46.6)
40.7 (37.3, 44.1)
44.1 (42.6, 45.6)
22.8 (21.8, 23.8)
Cut-points suggested by ADA 
HbA1c 5.7–6.4% (38–47 mmol/mol)
65.0 (63.4, 66.6)
63.5 (60.2, 66.8)
64.7 (63.3, 66.1)
55.3 (54.1, 56.5)
HbA1c ≥5.7% (38 mmol/mol)
75.0 (73.5, 76.4)
72.3 (69.2, 75.4)
74.5 (73.1, 75.8)
60.6 (59.4, 61.7)
Finding effective methods to prevent type 2 diabetes is a public health priority and given the unequivocal success of prevention programmes there is an increasing need to develop tools with which to identify high-risk individuals who may benefit from these programmes. This study has shown a higher ‘hit rate’ when using a two-step screening approach compared with population screening; this is in line with previous studies . This approach has also been shown to be cost effective in modelling studies .
Interestingly, although the yield was increased, the uptake to the screening invitation was low and comparable with that found in a population-based screening programme carried out in the same area . This finding suggests that informing people that they are at high risk of diabetes does not increase attendance to screening compared with a generic invitation. This is in contrast to other studies which have shown that risk stratification increases attendance . There are many possible explanations for this. First, the low response rate may be due to people not wanting to take part in the clinical trial, rather than the screening. The National Health Service (NHS) Health Checks Programme, which is not part of a research project, has seen uptake rates of around twice the level we have reported here . Second, everyone received an OGTT. The OGTT is costly, time consuming and inconvenient . Both patients and healthcare professionals have reported that the OGTT is a barrier to attending screening . The ADDITION-Europe study used a variety of screening methods to establish a cohort with screen-detected type 2 diabetes . Centres that used a three-step approach, combining a risk score followed by a random blood glucose test followed by an OGTT, had a significantly greater uptake compared with a one-step approach using the OGTT only . Third, both studies have been carried out in Leicester, UK and the ethnic makeup of this locality may have an impact on the response to screening. The area has a high prevalence of ethnic minority groups (ranging from 32% in the city of Leicester to 8% in the county of Leicestershire [according to the 2006 census]). Lower uptake to screening in those of South Asian ethnicity has consistently been observed [14, 21]. This problem is amplified by the known increased risk of type 2 diabetes in this community . Finally, the method of risk communication used was minimal: potential participants were told that there are risk factors for diabetes, some of which are modifiable and that their medical records suggest they might be at high risk. Informing people of their absolute risk of diabetes and potential benefits of early diagnosis and intervention may have increased uptake to screening. Further investigation into this is warranted
Recently the WHO recommended that the diagnostic criteria for type 2 diabetes be revised to include those with an HbA1c of ≥6.5% (48 mmol/mol) [9, 22]. The WHO found insufficient evidence to classify IGR using HbA1c ; however, an international expert panel and ADA have suggested that ranges of 6.0–6.4% (42–46 mmol/mol) and 5.7–6.0% (38–42 mmol/mol) can be considered . Based on the ADA criteria, up to two-thirds of the population could fall into this category, which would overwhelm prevention initiatives. Nevertheless, moving away from the OGTT has the potential to increase uptake to screening.
Both studies invited the top 10% at risk within a practice for screening. NICE suggest that the top 50% be invited for further testing . Lowering the cut-point will give potentially lower positive predictive values than seen here as the prevalence of disease will be reduced in a lower risk group. The benefit of using a practice-based risk score for screening is that GP practices can decide where to set the cut-point based on the resources available. The LPRS is the first tool developed for use in GP practices that includes IGR in the conditions detected. Therefore there is limited data with which to compare these results. Alongside the development of the risk score we have also developed a piece of software that enables integration of this score into routine clinical practice in primary care. Using this tool optimises the high-quality data stored in general practices .
There are a number of limitations that must be considered. First, only those at high risk were invited for screening so we are unable to assess sensitivity, specificity and the negative predictive value of the score in this setting. Second, the response rate was low, although this is similar to other studies in similar populations . This reflects the difficulty in recruiting a multi-ethnic urban population with wide variations in social economic status into research studies, which may have affected the representativeness and generalisability of the data. Due to ethical constraints we were not able to collect data on those who were invited but did not attend screening; therefore we are unable to compare the characteristics of those who attended with those who did not. These data were collected in the ADDITION-Leicester study and overall those who attended were older and more likely to be female . Finally, the score used here uses the percentage of South Asians within the practice as a proxy for individual ethnicity. Since the conception of these studies the recording of ethnicity within general practice has improved and the updated version of the LPRS includes an individual’s ethnicity. Therefore some South Asian participants may have received a falsely reduced score if they belonged to a practice with a high white European population and vice versa.
In summary, using a risk score to identify those at high risk of IGR/type 2 diabetes identifies a high previously undiagnosed yield using either an OGTT (30%) or HbA1c (44%) for diagnosis. Previous population-based screening programmes using universal OGTT have had lower yields of 20%. The LPRS is an inexpensive and simple way of targeting screening programmes at those with the highest risk.
This report/article presents independent research commissioned by the National Institute for Health Research (NIHR) under its Programme Grants for Applied Research scheme (RP-PG-0606-1272). The views expressed in this publication are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health. The project was supported by the NIHR Collaboration for Leadership in Applied Health Research and Care–Leicestershire, Northamptonshire and Rutland and the NIHR Leicester–Loughborough Diet, Lifestyle and Physical Activity Biomedical Research Unit, which is a partnership between University Hospitals of Leicester NHS Trust, Loughborough University and the University of Leicester.
The authors thank the LPD collaborators: A. Farooqi (Department of Health Sciences, University of Leicester, Leicester UK); M. Carey (Leicester Diabetes Centre, University Hospitals of Leicester, Leicester UK); K. Abrams (Department of Health Sciences, University of Leicester, Leicester UK); T. Skinner (Rural Clinical School, University of Tasmania, Tasmania, Australia); J. Tuomilehto (Department of Public Health, University of Helsinki, Helsinki, Finland); S. Heller (Department of Human Metabolism, University of Sheffield, Sheffield, UK); N. Samani (Department of Cardiovascular Sciences, University of Leicester, Leicester UK); B. Stribling (Leicester Diabetes Centre, University Hospitals of Leicester, Leicester UK); and K. Jones (Patient Representative, Leicester, UK).
This research was funded by the NIHR under its Programme Grants for Applied Research scheme (RP-PG-0606-1272) and Health Research Collaboration for Leadership in Applied Health Research and Care – Leicestershire, Northamptonshire and Rutland.
Duality of interest
K. Khunti (Chair), M. J. Davies and T. Yates are members of the NICE Public Health Guidance on prevention of type 2 diabetes among people with prediabetes. M. J. Davies and K. Khunti are advisors to the UK Department Health for the NHS Health Checks Programme. M. J. Davies has received funds for research, honoraria for speaking at meetings and has served on Advisory Boards for Lilly, sanofi-aventis, MSD, Novo Nordisk, BMS, BI and Roche. K. Khunti has received funds for research, honoraria for speaking at meetings and or served on Advisory Boards for AstraZeneca, GSK, Lilly, Novartis, Pfizer, Servier and sanofi-aventis. All other authors have no conflicts of interest to declare.
LJG, KK, DW, TY and MJD made substantial contribution to conception and design. CE, SG, JH, DS, SW and TY made substantial contribution to acquisition of data. LJG and DHM analysed and interpreted data. LJG drafted the article and KK, CE, SG, JH, DHM, DS, DW, SW, TY and MJD revised the paper. All authors approved the version to be published.