Back grey_arrow_rt.gif
Effect of early intensive multifactorial therapy on 5-year cardiovascular outcomes in individuals with type 2 diabetes detected by screening (ADDITION-Europe): a cluster-randomised trial
  Download the PDF here

The Lancet, Early Online Publication, 25 June 2011

Dr Simon J Griffin DM a Corresponding AuthorEmail Address, Prof Knut Borch-Johnsen DMSc b c d, Prof Melanie J Davies MD e, Prof Kamlesh Khunti MD f, Guy EHM Rutten PhD g, Prof Annelli Sandbak PhD d, Stephen J Sharp MSc a, Rebecca K Simmons PhD a, Maureen van den Donk PhD g, Prof Nicholas J Wareham PhD a, Prof Torsten Lauritzen DMSc


Intensive treatment of multiple cardiovascular risk factors can halve mortality among people with established type 2 diabetes. We investigated the effect of early multifactorial treatment after diagnosis by screening.

In a pragmatic, cluster-randomised, parallel-group trial done in Denmark, the Netherlands, and the UK, 343 general practices were randomly assigned screening of registered patients aged 40-69 years without known diabetes followed by routine care of diabetes or screening followed by intensive treatment of multiple risk factors. The primary endpoint was first cardiovascular event, including cardiovascular mortality and morbidity, revascularisation, and non-traumatic amputation within 5 years. Patients and staff assessing outcomes were unaware of the practice's study group assignment. Analysis was done by intention to treat. This study is registered with, number NCT00237549.

Primary endpoint data were available for 3055 (99·9%) of 3057 screen-detected patients. The mean age was 60·3 (SD 6·9) years and the mean duration of follow-up was 5·3 (SD 1·6) years. Improvements in cardiovascular risk factors (HbA1c and cholesterol concentrations and blood pressure) were slightly but significantly better in the intensive treatment group. The incidence of first cardiovascular event was 7·2% (13·5 per 1000 person-years) in the intensive treatment group and 8·5% (15·9 per 1000 person-years) in the routine care group (hazard ratio 0·83, 95% CI 0·65-1·05), and of all-cause mortality 6·2% (11·6 per 1000 person-years) and 6·7% (12·5 per 1000 person-years; 0·91, 0·69-1·21), respectively.

An intervention to promote early intensive management of patients with type 2 diabetes was associated with a small, non-significant reduction in the incidence of cardiovascular events and death.

Type 2 diabetes mellitus is common, expensive to manage, and associated with a substantial burden of morbidity and mortality, particularly owing to cardiovascular complications.1 Risk of cardiovascular events and death can be halved among patients with longstanding diabetes and microalbuminuria by intensive multifactorial treatment.2, 3 Treatment of individual risk factors, such as blood pressure,4, 5 cholesterol,6 and glucose,7 is effective. Outcomes might be improved if this approach were used early in the course of the disease.8 The effect of starting multifactorial treatment from the time of diagnosis is unknown.

Type 2 diabetes is detectable well before it is clinically diagnosed9 and many patients already have evidence of diabetic complications and potentially modifiable cardiovascular risk factors at the time of diagnosis.10 Early detection by screening is not associated with harmful psychological effects11 and, therefore, diabetes meets many suitability criteria for screening.12 Modelling studies have indicated that screening would be an efficient use of resources,13 but there are several critical uncertainties that have prevented its routine widespread implementation.14 No evidence from trials is available to show whether early intensive multifactorial treatment improves outcomes when started between detection by screening and clinical diagnosis. We did the multicentre Anglo-Danish-Dutch Study of Intensive Treatment In People with Screen Detected Diabetes in Primary Care (ADDITION-Europe) to investigate this issue.

Of 1312 general practices invited, 379 (29%) agreed to participate in the study. 36 (9%) of these were excluded and 343 (91%) were randomised to provide routine diabetes care (n=176) or intensive multifactorial treatment (n=167; figure 1), of which 327 (routine care n=165, intensive treatment n=162) completed screening and 318 (routine care n=157, intensive treatment n=161) included eligible patients. Participating practices in the UK and the Netherlands have been described.17, 18, 20 The mean patient list size was 7378 in the routine care group and 7160 in the intensive treatment group. The median prevalence of known diabetes was 3·5%. Data for the prevalence of diabetes in the Denmark practices were not available.

Screening identified 3233 patients with type 2 diabetes, of whom 3057 agreed to take part (Denmark, n=1533; Cambridge, UK, n=867; Netherlands, n=498; and Leicester, UK, n=159). The characteristics of these patients did not differ significantly from those of the 176 patients who were eligible but chose not to participate. Baseline demographic, clinical, biochemical, and treatment characteristics of patients in the two treatment groups were well matched overall (table 3). In Denmark, however, more patients were identified in practices assigned intensive treatment than in those assigned routine care (910 vs 623), and more patients in the former group had a history of ischaemic heart disease (International Classification of Diseases [version 10; ICD-10] codes I20-25, 102 [11·2%] vs 53 [8·5%]) or other cardiac diagnosis (ICD10 codes I30-52, 76 [8·4%] vs 28 [4·5%]). As previously described, patients overall exhibited high levels of untreated cardiovascular risk factors at diagnosis.19

Screening programmes varied by centre (table 1). The median number of follow-up meetings for audit and feedback was four (range two to ten). Primary endpoint data were available for 3055 (99·9%) of 3057 participants. The mean follow-up period was 5·3 (SD 1·6) years, during which 238 first cardiovascular events occurred, with similar numbers and risk in each group (table 4). The cumulative probability plot for the primary endpoint seemed to diverge after 4 years of follow-up (figure 2). In the predefined subgroup analyses, no interactions were seen between the intervention and age or previous cardiovascular event (p>0·1). Estimated HRs, however, were 1·12 (95% CI 0·70-1·79) in patients younger than 60 years and 0·70 (95% CI 0·52-0·95) in those aged 60 years and older. HRs for individual components of the composite endpoint all favoured the intensive treatment group (table 4), although none achieved significance (figure 2). No patient underwent amputation as a first event. The correlation coefficient within clusters for the primary endpoint was 0·002 (Denmark: 0·014, UK [combined] 0·0000016, and the Netherlands 0·025), which suggests that the cluster design had little effect on study power.

196 patients died overall (60 [30·6%] cardiovascular deaths, including six patients with first events classified as myocardial infarction, two as stroke, and four as revascularisation, 97 [49·5%] cancer deaths, and 39 [19·9%] from other causes; table 4). The combined HR for death in the intensive treatment group compared with the routine care group was 0·91 (95% CI 0·69-1·21). Heterogeneity of results between countries was not significant (figure 3). In the UK, significantly fewer patients died in the intensive treatment group than in the routine care group, but in Denmark the risk in the routine care group was lower, although not significantly so. All results were unchanged after sensitivity analysis that excluded participants in the two practices for which endpoint data were obtained while aware of treatment allocation (webappendix p 2).

Of the 2859 patients still alive at 5 years, 2400 (84%) returned to a clinical research facility for follow-up health assessments. Clinical and biochemical outcomes could be obtained from general practice records for a further 328 (11·5%) participants. Compared with patients for whom follow-up data were available, the 131 with missing data were more likely to be from an ethnic minority group (10·2% vs 5·7%, p=0·04) and have higher baseline total cholesterol (5·9 mmol/L vs 5·6 mmol/L, p=0·004) and LDL cholesterol values (3·7 mmol/L vs 3·4 mmol/L, p=0·009). Changes in clinical and biochemical values and prescribed medications in the two groups are shown in table 3. By 5 years of follow-up improvements were seen for cardiovascular risk factors in both groups. Small but significant differences between groups were seen in change from baseline for glycated haemoglobin A1c (HbA1c), systolic and diastolic blood pressure and total and LDL cholesterol, in favour of the intensive treatment group. Prescription of glucose-lowering, antihypertensive, and lipid-lowering drugs increased in both groups. At follow-up more patients in the intensive treatment group were prescribed aspirin, angiotensin-converting-enzyme inhibitors or angiotensin-receptor blockers, glucose-lowering drugs, antihypertensive drugs, and lipid-lowering drugs than were those in the routine care group (table 3).

In both groups, more patients had values below target thresholds for HbA1c concentrations, blood pressure, and cholesterol concentrations at follow-up than at baseline (figure 4). The proportion of patients meeting the targets was higher in the intensive treatment group than in the routine care group. The proportions of individuals reporting hypoglycaemia, as assessed by the diabetes treatment satisfaction questionnaire, did not differ (χ2 4·44, p=0 ·62).33 All results were unchanged after sensitivity analysis was done that excluded participants with follow-up clinical data obtained from general practice records.

An intervention to promote target-driven, intensive management of patients with type 2 diabetes detected by screening was associated with slightly, but significantly, increased prescription of treatments and improvements in cardiovascular risk factors, and with a non-significant relative reduction in the incidence of cardiovascular events at 5 years. Differences between study groups for all components of the primary endpoint favoured the intensive treatment group. Differences were greatest for myocardial infarction and smallest for stroke.

We cannot rule out the possibility that these findings were due to chance. Rates of cardiovascular events seemed to diverge after 4 years of follow-up, although the risk of adverse outcomes, including death or self-reported hypoglycaemia, was not increased. The incidence of cardiovascular events in the routine care group (8·5%) was lower than expected compared with that in newly diagnosed patients in the UKPDS (12·1%).34 Mortality in this group (6·7%) was lower than that in patients with type 2 diabetes detected by screening in Hoorn35 (25% over 10 years) and newly diagnosed patients in a study in Denmark (33% over 7·4 years).36 In our study, mortality was similar to that reported for people of the same age without diabetes in the general population of Denmark in 1995-2006.37 This lack of difference is likely to be due to the quality of care delivered to patients early in the course of their disease in both groups.

This trial shows that screening for type 2 diabetes and early intensive multifactorial treatment of the detected patients are feasible in general practice. In both study groups, cardiovascular risk factors, such as blood pressure and cholesterol concentrations, improved notably after diagnosis and glycaemia and weight did not increase. Absolute values for risk factors at follow-up and changes from baseline compare favourably with those in other trials in which patients were recruited at clinical diagnosis and followed up for 6 years.34, 35 Clinically important differences in risk factors between groups at 1 year,18 however, were not maintained and were considerably smaller than those achieved in similar studies.2, 34

Adherence to treatment algorithms might have been suboptimum in this pragmatic trial. In three centres the screening programme used risk scores, including treatment for hypertension as one of the factors to predict the risk of type 2 diabetes, which could have limited the achievable differences between groups in blood pressure. Furthermore, the trial was undertaken against a background of improvements in the delivery of diabetes care in general practice, such as that associated with the introduction of the Quality and Outcomes Framework for primary care in the UK,38 and evidence-based guidelines in Denmark39 and the Netherlands,40 which might also have lowered the achievable differences in treatment between groups.

Participants were drawn from a large, representative, population-based sample in three different European countries. They were identified by a range of different screening programmes and all were diagnosed according to WHO criteria.21 The intensive treatment intervention incorporated various methods that encouraged changes in practitioner behaviour, and treatment algorithms and targets that were based on robust trial data.2, 4, 6, 26 Because the intervention was delivered mainly via family physicians and nurses, we randomised general practices rather than individuals to minimise the risk of contamination. The analyses were appropriate to the cluster design. We achieved high participant retention and independently adjudicated endpoint ascertainment in both trial groups. We assessed clinically important outcomes with standard equipment and protocols by especially trained staff members who were unaware of study group allocation. Patients with and without follow-up data differed little.

The generalisability of our findings to other settings should be considered in light of the non-random recruitment of general practices. We did, however, cover a large geographical area in each country, and the 26% of invited practices that were randomised were nationally representative for key descriptive characteristics.17, 18, 20 Assessment of hypoglycaemia was imprecise, as it was limited to one question, but we believe this factor is unlikely to be biased because the questionnaire was administered in the same way to participants in the two study groups by staff unaware of study group allocations. We feel that the size of the study means that even with a potentially imprecise instrument the power was sufficient to quantify the effects of the intervention on the frequency of hypoglycaemic events. In two centres (Denmark and the Netherlands), assays used to assess biochemical factors differed slightly between baseline and follow-up, but not between groups (webappendix p 1). By necessity cluster randomisation was done before screening and recruitment of patients, a design feature that can introduce differences between groups in relation to participant characteristics. Overall, patients in the two study groups were well matched, although the intervention could have influenced the number and type of patients recruited to the intensive treatment group. This effect might have been particularly relevant in Denmark, where staff in the intensive treatment group might have been especially alert to the potential benefits of early detection and treatment and, therefore, have been more active testing in individuals at high cardiovascular risk, which could have led to the observed differences in baseline characteristics between study groups in this centre. These differences between groups might have lessened the apparent effects of the intervention and, therefore, could explain, at least partly, differences between centres, which seemed to be present but were not necessarily significant owing to the limited power of heterogeneity test.

Methods of screening, outcome assessment, and laboratory testing were standard across patient groups within centres but did differ between centres. These differences might have contributed to heterogeneity, even though interlaboratory validation suggested that methods were consistent between centres and all potential primary endpoints were adjudicated in the same way. Other sources of heterogeneity include the characteristics of patients, which varied by centre, presumably because of differences in the screening programmes and underlying populations, the delivery of the intensive treatment intervention (table 1), and other unmeasured factors.

The lower-than-expected event rate suggests that 5 years of follow-up is insufficient, and the apparent divergence from 4 years onwards suggests that further follow-up is justified to test whether early intensive multifactorial treatment reduces cardiovascular risk in the long term, as seen in the UKPDS.8 Finally, the complex nature of the intensive treatment intervention and the pragmatic design of the trial make it difficult to say whether any components are individually associated with a reduction in cardiovascular risk.

The observed absolute and relative risks for cardiovascular events and mortality among participants with a mean HbA1c concentration of around 6·5% 5 years after diagnosis, should allay concerns about early intensive treatment of hyperglycaemia.

We saw no interaction between intensive treatment and age or history of cardiovascular events, although benefits associated with this intervention seemed to be greatest in patients aged 60 years or older at diagnosis. Multifactorial treatment of screen-detected patients in both study groups was associated with improvements in cardiovascular risk factors and lower than expected rates of cardiovascular events and mortality. The small differences in treatment between groups at 5 years were associated with a non-significant 17% reduction in the incidence of cardiovascular events, with no obvious adverse consequences. We are undertaking analyses of trial data on microvascular endpoints, quality of life, functional status, and health service costs. Although there is no evidence of harm associated with screening11 and intensive therapy,41 the extent to which the complications of diabetes can be reduced by earlier detection and treatment remains uncertain (panel).

Research in context
Systematic review

Studies have shown cardiovascular benefits with treatment to lower blood pressure and glucose and cholesterol concentrations in patients with type 2 diabetes mellitus.4-7 Intensive multifactorial treatment halved risk of cardiovascular events and death in patients with longstanding diabetes and microalbuminuria in the STENO-2 trial.2 We searched PubMed for relevant articles by use of "diabetes mellitus, type 2" and "cardiovascular disease" as MeSH headings and the term "randomised controlled trial" in any heading, in combination with the terms "multifactorial" or "screen*". We placed no restriction on language, year of publication, or study quality. We found no published trial evidence of the effects of multifactorial treatment on cardiovascular outcomes in individuals with screen-detected type 2 diabetes.


Population screening for type 2 diabetes and subsequent intensive treatment is feasible. Cardiovascular risk factors improved in the 5 years after detection by screening, and rates of first cardiovascular events and mortality were lower than expected. The small differences in prescribed treatment, and levels of cardiovascular risk factors, between groups at 5 years were associated with a non-significant 17% reduction in the incidence of cardiovascular events, with no obvious adverse consequences. The extent to which the complications of diabetes can be reduced by earlier detection and treatment remains uncertain.

When compared with routine care, an intervention to promote target-driven, intensive management of patients with type 2 diabetes detected by screening was associated with small increases in the prescription of drugs and improvements in cardiovascular risk factors, but was not associated with significant reductions in the incidence of cardiovascular events or death over 5 years.
  icon paper stack View Older Articles   Back to Top