HIV Articles  
Long-Term Follow-up after Tight Control of Blood Pressure in Type 2 Diabetes
  NEJM Oct 9 2008
Rury R. Holman, F.R.C.P., Sanjoy K. Paul, Ph.D., M. Angelyn Bethel, M.D., H. Andrew W. Neil, F.R.C.P., and David R. Matthews, F.R.C.P.
From the Diabetes Trials Unit (R.R.H., S.K.P., M.A.B.) and the Division of Public Health and Primary Health Care (H.A.W.N.), and the National Institute of Health Research (NIHR) School for Primary Care Research (H.A.W.N.), Oxford Centre for Diabetes, Endocrinology, and Metabolism (R.R.H., S.K.P., M.A.B., H.A.W.N., D.R.M.); and the NIHR Oxford Biomedical Research Centre (R.R.H., H.A.W.N., D.R.M.) - both in Oxford, United Kingdom.
"Optimal blood-pressure control is of major importance in reducing the risks of microvascular and macrovascular disease in patients with type 2 diabetes but must be maintained if these benefits are to be sustained."

Background- Post-trial monitoring of patients in the United Kingdom Prospective Diabetes Study (UKPDS) examined whether risk reductions for microvascular and macrovascular disease, achieved with the use of improved blood-pressure control during the trial, would be sustained.
Methods- Among 5102 UKPDS patients with newly diagnosed type 2 diabetes mellitus, we randomly assigned, over a 4-year period beginning in 1987, 1148 patients with hypertension to tight or less-tight blood-pressure control regimens. The 884 patients who underwent post-trial monitoring were asked to attend annual UKPDS clinics for the first 5 years, but no attempt was made to maintain their previously assigned therapies. Annual questionnaires completed by patients and general practitioners were used to follow patients who were unable to attend the clinic in years 1 through 5, and questionnaires were used for all patients in years 6 to 10. Seven prespecified aggregate clinical end points were examined on an intention-to-treat basis, according to the previous randomization categories.
Results- Differences in blood pressure between the two groups during the trial disappeared within 2 years after termination of the trial. Significant relative risk reductions found during the trial for any diabetes-related end point, diabetes-related death, microvascular disease, and stroke in the group receiving tight, as compared with less tight, blood-pressure control were not sustained during the post-trial follow-up. No risk reductions were seen during or after the trial for myocardial infarction or death from any cause, but a risk reduction for peripheral vascular disease associated with tight blood-pressure control became significant (P=0.02).
Conclusions- The benefits of previously improved blood-pressure control were not sustained when between-group differences in blood pressure were lost. Early improvement in blood-pressure control in patients with both type 2 diabetes and hypertension was associated with a reduced risk of complications, but it appears that good blood-pressure control must be continued if the benefits are to be maintained.
The United Kingdom Prospective Diabetes Study (UKPDS) was a randomized, prospective, multicenter trial that indicated that improved glucose control in patients with newly diagnosed type 2 diabetes mellitus reduces the risk of clinically evident microvascular complications, with a relative risk reduction for myocardial infarction of 16% (P=0.052).1 From 1987 to 1991, UKPDS patients with hypertension were, in addition, randomly assigned in a factorial manner to a tight blood-pressure control regimen involving an angiotensin-converting-enzyme (ACE) inhibitor or a beta-blocker or to a less-tight blood-pressure control strategy that excluded these agents.2 For tight as compared with the less-tight control of blood pressure, there were relative risk reductions of 24% for any diabetes-related end point, 32% for diabetes-related death, 44% for stroke, and 37% for microvascular disease.3
A postinterventional benefit with regard to both microvascular and macrovascular complications of diabetes has been reported in the Steno-2 Study.4 Over a period of 5.5 years after multifactorial risk intervention, persistent reductions in the rates of death from any cause, death from cardiovascular causes, and progression of microvascular disease were reported, despite diminutions in the within-trial differences in glucose, blood-pressure, and lipid control, suggesting that there may be a persistent effect of earlier improved management of risk factors.4
We report here the results of a 10-year, postinterventional follow-up of the survivor cohort of the UKPDS blood-pressure study that examined whether a continued benefit of earlier improved blood-pressure control was evident and, if so, the degree to which it persisted. In a companion article in this issue of the Journal, we report the results of the post-trial monitoring in the glycemic intervention groups of the UKPDS patients.5
This 10-year follow-up study of the survivor cohort from the UKPDS blood-pressure-intervention trial shows that the benefits seen in patients assigned to a strategy of tight blood-pressure control3 were not maintained once the differences in blood pressure seen during the trial itself were lost. Participating patients were a hypertensive cohort recruited from among people with newly diagnosed diabetes and have been followed for over 15,000 person-years since their randomization to tight or less-tight blood-pressure control. The lower blood-pressure levels attained early in the course of diabetes did not appear to confer a legacy effect, as seen with the more sustained benefits of earlier improved blood-glucose control.5,9
The UKPDS blood-pressure trial showed substantial reductions in the relative risk of any diabetes-related end point, diabetes-related death, stroke, and microvascular disease associated with a strategy of tight blood-pressure control, as compared with less-tight control, that maintained systolic pressure that was 10 mm Hg lower, and diastolic pressure that was 5 mm Hg lower, over a median of 8.4 years.3,7 After the trial, blood-pressure levels fell in the less-tight-control group and rose in the tight-control group, with no significant between-group differences after 2 years. In line with this equalization of blood pressures, the risk reductions observed were substantially smaller by year 10 after the trial, and none remained significant. No significant risk reductions were seen, during or after the trial, for myocardial infarction or death from any cause. The risk-reduction point estimate of 49%, which was not significant (P=0.17), for the small number of patients with peripheral vascular disease during the trial3 remained unchanged, but with the accumulation of additional events during the post-trial monitoring period, the risk reduction was significant at year 10 (50%, P=0.02). No significant differences were found during or after the trial in the comparison of ACE-inhibitor therapy and beta-blocker therapy with regard to any aggregate end point, apart from a nominally significant increase in the risk of death from any cause with ACE-inhibitor use at the end of the post-trial monitoring period.
Since the publication of the UKPDS blood-pressure results, other randomized, controlled trials have confirmed the importance of treating hypertension in patients with type 2 diabetes10,11,12,13,14 and have examined blood-pressure targets lower than the UKPDS target of 150/85 mm Hg.3 None of those subsequent studies have reported post-trial follow-up data, so the longevity of benefits observed during the trial is unknown, and we are unaware of any other blood-pressure trials that have specifically examined post-trial legacy effects. A 10-year post-trial follow-up of the Cooperative North Scandinavian Enalapril Survival Study (CONSENSUS),15 which evaluated the usefulness of enalapril for severe heart failure rather than for hypertension (and which was discontinued after 6 months), showed a significant improvement in survival for patients previously assigned to enalapril (P=0.008).16 The Steno-2 Study, in which patients with type 2 diabetes were randomly assigned, in a nonfactorial manner, to conventional or intensive multiple risk factor interventions, showed additive benefits of improved glycemic control, the use of aspirin, and antihypertensive and lipid-lowering therapies after a median of 7.8 years.17 Post-trial follow-up for a median of 5.5 years showed continued benefits in the intensive-therapy group as compared with the conventional group, with an overall 20% absolute reduction in the risk of death from any cause (P=0.02).4 However, these sustained risk reductions cannot simply be regarded as a legacy effect, since favorable differences in systolic blood pressure (a difference of 6 mm Hg), low-density lipoprotein and HDL cholesterol levels, and blood glucose levels persisted beyond the trial period.4
Hypertension is a major risk factor for cardiovascular disease in patients with type 2 diabetes,18,19 conferring an age-adjusted 82% increase in the risk of diabetes-related death.20 Hypertension is more common among those with diabetes than in the general population, as reported in the Multiple Risk Factor Intervention Trial (MRFIT), which showed that the absolute risk of death from cardiovascular causes among men with diabetes was three times that among men without diabetes.21 The UKPDS blood-pressure trial was initiated in 1987 in recognition of the probable need to control both glycemia and blood pressure.22 The trial showed not only the benefits of improved blood-pressure control associated with a stepwise addition of blood-pressure-lowering therapies in a treat-to-target approach but also the necessity of increasing the intensity of these therapies over time.3,23 A more recent combined analysis of the blood-glucose and blood-pressure data from the UKPDS has confirmed the additive effects of blood glucose and blood pressure over time on the risk of complications in patients with type 2 diabetes and has shown that those assigned to strategies of both intensive glucose control and tight blood-pressure control had fewer events than those assigned to either strategy alone or to neither one.24
It is not surprising that the beneficial effects seen in study populations randomly assigned to blood-pressure-lowering therapies are lost once the blood-pressure differences are lost. Most trials involving lowering of blood pressure, such as the 4.5-year Systolic Hypertension in the Elderly Program (SHEP) trial demonstrating a significant reduction in the risk of stroke,25 show an early separation of the cumulative risk of macrovascular end points between study groups. It might be expected that the relatively rapid "on-effects" of improved blood-pressure lowering on cardiovascular risk would be reflected in similarly rapid "off-effects" if blood-pressure control were to be relaxed. Alternatively, post-trial improvements in blood-pressure therapy in groups treated less aggressively during the trial would be likely to bring their cardiovascular risks into line with those treated more aggressively during the trial. In either case, no long-term differences in cardiovascular risk would be seen, as was found in our trial. This is in direct contrast to the legacy effects of earlier intensive glucose control seen in patients with type 2 or type 1 diabetes during follow-up studies of the UKPDS5 and the Epidemiology of Diabetes Interventions and Complications (EDIC) study,9 respectively, both of which showed extended and improved treatment benefits despite early post-trial loss of between-group glycemic differences.
Our study has certain important limitations. Questionnaires may not have captured all nonfatal end points. Biochemical and clinical measurements were not collected in years 6 to 10, although by year 3 after the trial, it was already evident that blood-pressure differences had been lost. The glucose and blood-pressure data from the trial were analyzed separately, given the factorial design, but there may have been some post-trial confounding, since at baseline the glycated hemoglobin levels were higher in patients previously assigned to tight blood-pressure control than in those assigned to less-tight control, presumably because of greater beta-blocker use and possibly the use of thiazide diuretics. The lack of detailed information on risk factors in years 6 through 10 also precludes the use of proportional-hazards analyses to assess effects of other time-dependent covariates such as microalbuminuria.
In conclusion, these 10-year post-trial follow-up data from the UKPDS indicate that the relative risk reductions seen in the tight blood-pressure-control group did not persist when blood-pressure differences were no longer maintained. Optimal blood-pressure control is of major importance in reducing the risks of microvascular and macrovascular disease in patients with type 2 diabetes but must be maintained if these benefits are to be sustained.
Characteristics of the Patients

Baseline characteristics for the 884 patients undergoing post-trial monitoring among the 1148 patients randomly assigned to tight or less-tight blood-pressure control (Figure 1), and for the 592 patients in the tight-control group who were randomly assigned to ACE-inhibitor or beta-blocker therapy, are shown in Table 1. As compared with patients who had been assigned to less-tight blood-pressure control, those who had been previously assigned to tight blood-pressure control during the study had a lower mean systolic blood pressure (by 9 mm Hg, P<0.001) and a lower diastolic blood pressure (by 3 mm Hg, P<0.001) but a higher median glycated hemoglobin value (by 0.8 percentage points, P=0.001). In the group undergoing tight blood-pressure control, 61% of patients were taking two or more antihypertensive agents, as compared with 36% in the group undergoing less-tight blood-pressure control. Overall, less than 2% of patients were taking lipid-lowering drugs. The two groups did not differ significantly with respect to age; sex; race or ethnic group; body weight; levels of fasting plasma glucose, lipids, or plasma creatinine; or the albumin-to-creatinine ratio (Table 1). Baseline characteristics for patients previously assigned to an ACE inhibitor or beta-blocker also did not differ significantly between the two groups, except that the beta-blocker group had a greater mean body weight, by 4 kg (P=0.01), a lower total cholesterol level, by 7 mg per deciliter (0.18 mmol per liter) (P=0.04), and a lower high-density lipoprotein (HDL) cholesterol level, by 3 mg per deciliter (0.08 mmol per liter) (P=0.009). No significant differences were found at baseline for any variable between patients with and those without available final-year data (Table 1), except that those with data were less often white (P<0.001).
In years 1 through 5, the majority of patients (92 to 97%) attended a UKPDS clinic. The median duration of follow-up after randomization with regard to the comparison of the tight-control and the less-tight-control groups was 14.5 years (15,583 person-years), including a median of 8.0 years of post-trial monitoring. The median duration of follow-up after randomization with regard to the comparison of patients receiving an ACE inhibitor and those receiving a beta-blocker was 14.6 years (10,358 person-years), including a median of 7.6 years after the trial.
The mortality rate was 51%, with the leading causes of death being cardiovascular events (53%) and cancer (21%). The overall rate of loss to follow-up was 2%. Baseline differences between the less-tight-control group and the tight-control group in the number of antihypertensive agents taken were no longer evident by year 5 after the trial, with 73% and 75% of patients, respectively, taking two or more agents. By year 5, among all patients, 24% were receiving lipid-lowering therapy and 44% were receiving aspirin, with no significant differences between the two groups. Baseline differences between the groups in mean systolic and diastolic blood pressures were lost by year 1 and year 2, respectively (Figure 2). There were no significant changes in body weight (Figure 2). There were no significant differences in glycated hemoglobin, fasting plasma glucose, or plasma creatinine levels or albumin-to-creatinine ratio, at any time point, between the tight and less-tight-control groups, or between the ACE inhibitor and beta-blocker groups, with the exception of fasting plasma glucose at year 3 and glycated hemoglobin at year 4 (P=0.04 for both end points) (see the Supplementary Appendix, available with the full text of this article at
Clinical Outcomes
Among patients undergoing tight blood-pressure control as compared with those undergoing less-tight control, significant relative risk reductions seen during the trial - for any diabetes-related end point (relative risk reduction of 24%, P=0.005), diabetes-related death (32%, P=0.02), stroke (44%, P=0.01), and microvascular disease (37%, P=0.009) - were not sustained during the post-trial follow-up period. These differences diminished over time (Figure 3), with risk reductions that were not significant at year 10 of 7% for any diabetes-related end point (P=0.31), 16% for diabetes-related death (P=0.12), 23% for stroke (P=0.12), and 16% for microvascular disease (P=0.17) (Table 2 and Figure 4).
The nonsignificant relative risk reductions observed during the trial for myocardial infarction (21%, P=0.13) and death from any cause (18%, P=0.17) were also diminished during the follow-up period (Figure 3). Risk reductions in the tight-control group as compared with the less-tight-control group 10 years after the trial ended were 10% for myocardial infarction (P=0.35) and 11% for death from any cause (P=0.18) (Table 2 and Figure 4). Overall, few patients had peripheral vascular disease (8 in each group), with maintenance of the risk reduction seen during the trial (49%, P=0.17) (Figure 4); however, the number of patients with this end point increased to 21 in each group during the post-trial monitoring period, and there was a significant risk reduction in the tight-control group at year 10 (50%, P=0.02) (Table 2 and Figure 4). Between the patients receiving an ACE inhibitor or beta-blocker, no significant differences in aggregate end points were seen during the trial or during post-trial monitoring, apart from an emerging, nominally significant increase in the risk of death from any cause in the ACE-inhibitor group (P=0.047) (Table 2 and Figure 3 and Figure 4).

Details on recruitment of patients for the UKPDS and the protocols, methods,2,6 and post-trial monitoring procedures5 have been reported previously. The embedded Hypertension in Diabetes Study (HDS)2 involved randomization of 1148 of 1544 UKPDS patients with hypertension (blood pressure, 160/90 mm Hg or 150/85 mm Hg if the patient was receiving antihypertensive treatment) in a factorial manner, over a 4-year period beginning in 1987, to a tight blood-pressure control regimen (with randomized assignment to up to 50 mg of captopril [Capoten, Bristol-Myers Squibb] twice daily or up to 100 mg of atenolol [Tenormin, Hoechst] once daily) or to a less-tight blood-pressure control regimen (without the use of ACE inhibitors or beta-blockers). A total of 144 patients declined to participate in the blood-pressure study. A total of 252 other patients were excluded because they already required strict blood-pressure control (owing to previous stroke, accelerated hypertension, cardiac failure, or renal failure) or beta-blockade (owing to myocardial infarction in the previous year or current angina), had severe vascular disease or a severe concurrent illness, had contraindications to the use of beta-blockers, or were pregnant. The 1148 patients had a mean (}SD) age of 56.4}8.1 years and 56% were male; 87% reported that they were white, 4% Asian Indian, and 7% Afro-Caribbean.
The blood-pressure target for the tight-control group was less than 150/85 mm Hg, and the target for the less-tight-control group was less than 180/105 mm Hg, with therapies added as necessary in the following recommended sequence: 20 mg of furosemide daily (maximum, 40 mg twice daily), 10 mg of slow-release nifedipine daily (maximum, 40 mg), 250 mg of methyldopa daily (maximum, 500 mg), and 1 mg of prazosin thrice daily (maximum, 5 mg). The median duration of follow-up for the comparison of tight control with less-tight control was 8.4 years,2 and within the tight-control group, the duration was 8.6 years for the comparison of ACE inhibitors with beta-blockers.
Post-Trial Monitoring
When the interventional trial closed on September 30, 1997, the survivor cohort entered a 10-year post-trial monitoring program planned to coincide with a projected 50% mortality rate. In September 1998,3,7 after publication of the UKPDS results, patients and clinicians were advised to aim for the lowest feasible blood-glucose and blood-pressure levels. Patients returned to their usual physicians, with no attempt to maintain previously randomized therapies, and were seen annually for 5 years in UKPDS clinics. Standardized collection of end-point data was continued, as were measurements of blood pressure, levels of fasting plasma glucose, glycated hemoglobin, and plasma creatinine; calculation of the albumin-to-creatinine ratio; and clinical examinations every 3 years. Questionnaires - the European Quality of Life-5 Dimensions (EQ-5D) instrument8 and a questionnaire on health-resource use - were administered annually. Patients who were unable to attend clinics were mailed the EQ-5D and health-resource questionnaires, and additional questionnaires were sent to their general practitioners to capture possible end points. During years 6 to 10, because of funding constraints, these questionnaires were used to follow all patients remotely. After the censoring date of post-trial monitoring (September 30, 2007), final questionnaires were sent to all remaining patients.
Clinical Outcomes
The study administrator obtained full documentation from hospitals and general practitioners for all putative end points, whether reported at a clinic visit or on a questionnaire. The vital status of patients was obtained from the U.K. Office of National Statistics. End points were adjudicated exactly as they had been during the trial and by the same end-point committee, the members of which were unaware of the treatment assignments. The seven prespecified UKPDS aggregate clinical end points2 were any diabetes-related end point (sudden death, death from hyperglycemia or hypoglycemia, fatal or nonfatal myocardial infarction, angina, heart failure, fatal or nonfatal stroke, renal failure, amputation, vitreous hemorrhage, retinal photocoagulation, blindness in one eye, or cataract extraction), diabetes-related death (death from myocardial infarction, stroke, peripheral vascular disease, renal disease, hyperglycemia or hypoglycemia, or sudden death), death from any cause, myocardial infarction (sudden death, fatal or nonfatal myocardial infarction), fatal or nonfatal stroke, peripheral vascular disease (amputation of at least one digit or death from peripheral vascular disease), and microvascular disease (vitreous hemorrhage, retinal photocoagulation, or renal failure).
Statistical Analysis
Statistical analyses were performed on the basis of the intention-to-treat principle with the use of descriptive statistics presented as the number of patients (percentage) or appropriate measures of central tendency and dispersion. Continuous and categorical study variables were compared between tight and less-tight blood-pressure-control therapies by means of nonparametric tests. Kaplan-Meier time-to-event analyses were used for aggregate clinical end points, with log-rank tests used to test for differences between previous treatment assignments. Since recruitment was performed during a 4-year period, patients could have participated in the interventional trial for 6 to 10 years. Since there was no common time from randomization to the commencement of post-trial monitoring, we used serial hazard-ratio plots with confidence intervals, after checking that proportional-hazards assumptions were not violated, to estimate post-trial relative risks, with P values calculated only at the end of the follow-up period. Absolute risk rates are expressed as the number of events per 1000 person-years.
Post-trial monitoring was initiated by the investigators and was sponsored for 5 years by the Medical Research Council and then by the University of Oxford. The investigators designed and conducted the study, analyzed the data, and prepared the manuscript, independently of any funding bodies. The investigators vouch for the completeness and accuracy of the data.
  iconpaperstack View older Articles   Back to Top