We studied hypertensive patients with type 2 diabetes who had been recruited to the UK prospective diabetes study.22,23
General practitioners were asked to refer patients aged 25-65 with newly diagnosed diabetes to 23 participating centres. A total of 5102 were recruited as they met the study’s entry criterion (fasting plasma glucose concentration >6
mmol/l on two mornings), were willing to join, and did not meet the exclusion criteria for the study. Exclusion criteria were ketonuria >3
mmol/l; a history of myocardial infarction in the previous year; current angina or heart failure; more than one major vascular episode; serum creatinine concentration >175
μmol/l; retinopathy requiring laser treatment; malignant hypertension; an uncorrected endocrine abnormality; an occupation which would preclude insulin treatment (such as heavy goods vehicle driver); a severe concurrent illness likely to limit life or require extensive systemic treatment; or inadequate understanding or unwillingness to enter the study.22,23
The patients were treated by diet alone for 3 months.24
Patients who remained hyperglycaemic (fasting plasma glucose 6.1-15.0
mmol/l) without diabetic symptoms were randomly allocated conventional blood glucose control, primarily by diet, or intensive control (aiming for a fasting plasma glucose concentration <6.0
mmol/l) with additional sulphonylurea, insulin, or metformin treatment. Details of the protocol are published.22,23
Of the 4297 patients recruited to the 20 centres participating in the hypertension in diabetes study, 243 had either died or were lost to follow up before the start of the hypertension study in 1987 (fig ). Of the remaining 4054 patients, 1544 (38%) had hypertension, defined in 727 patients as a systolic blood pressure
Hg and/or a diastolic blood pressure
mm Hg or in 421 patients receiving antihypertensive treatment as a systolic pressure of
Hg and/or a diastolic pressure
Hg (fig ). Patients were enrolled on the basis of the mean of three blood pressure measurements taken at consecutive clinic visits. The exclusion criteria were a clinical requirement for strict blood pressure control (previous stroke, accelerated hypertension, cardiac failure, or renal failure) or β blockade (myocardial infarction in the previous year or current angina); severe vascular disease (more than one major vascular episode); a severe concurrent illness or contraindications to β blockers (asthma, intermittent claudication, foot ulcers, or amputations); pregnancy; or unwillingness to join the study. Of the 1544 hypertensive patients, 252 were excluded and 144 patients did not enter the study. A total of 1148 patients (637 men (55%)) with a mean age of 56.4 (SD 8.1) years entered the hypertension in diabetes study between 1987 and 1991.21
Table shows their characteristics at randomisation to blood pressure control policy.
Figure 1 Selection and random allocation of patients to treatment in hypertension in diabetes study
Characteristics of patients allocated to tight and less tight control of blood pressure. Values are numbers (percentages) of patients unless stated otherwise
Randomisation stratified for those with or without previous treatment for hypertension was performed by the coordinating centre. In all 758 patients were allocated tight control of blood pressure, aiming for a blood pressure <150/85
Hg (400 patients were given an angiotensin converting enzyme inhibitor (captopril) and 358 a β blocker (atenolol) as the main treatment); 390 patients were allocated a less tight control of blood pressure, aiming for a blood pressure <180/105
Hg but avoiding treatment with angiotensin converting enzyme inhibitors or β blockers (fig ). Sealed opaque envelopes were used and checked as described for the UK prospective diabetes study.23
The original blood pressure target of 200/105
Hg in the group assigned to less tight control was reduced in 1992 by the steering committee of the hypertension in diabetes study after publication of the results of studies in elderly, non-diabetic subjects during 1991-2.16,25,26
Randomisation produced balanced numbers of patients allocated to the various glucose and blood pressure treatment combinations for the UK prospective diabetes study and hypertension in diabetes study.
Captopril was usually started at a dose of 25
mg twice daily, increasing to 50
mg twice daily, and atenolol at a daily dose of 50 mg, increasing to 100
mg if required. Other agents were added if the control criteria were not met in the group assigned to tight control despite maximum allocated treatment or in the group assigned to less tight control without drug treatment. The suggested sequence was frusemide 20
mg daily (maximum 40
mg twice daily), slow release nifedipine 10 mg (maximum 40 mg) twice daily, methyldopa 250
mg (maximum 500 mg) twice daily, and prazosin 1
mg (maximum 5 mg) thrice daily.
Patients visited study clinics every 3-4 months. At each visit plasma glucose concentration, blood pressure, and body weight were measured, and treatments to control blood pressure and blood glucose concentration were noted and adjusted if target values were not met. If treatments and target blood pressures were not in accord with the protocol, the coordinating centre sent letters about affected patients to the clinical centres requesting appropriate action. A central record of all apparent protocol deviations was maintained. Symptoms including any drug side effects and clinical events were noted. Physicians recorded hypoglycaemic episodes as minor if the patient was able to treat the symptoms unaided and as major if third party or medical intervention was necessary.
Blood pressure measurements
Blood pressure (diastolic phase 5) while the patient was sitting and had rested for at least five minutes was measured by a trained nurse with a Copal UA-251 or a Takeda UA-751 electronic auscultatory blood pressure reading machine (Andrew Stephens, Brighouse, West Yorkshire) or with a Hawksley random zero sphygmomanometer (Hawksley, Lancing, Sussex) in patients with atrial fibrillation. The first reading was discarded and the mean of the next three consecutive readings with a coefficient of variation below 15% was used in the study, with additional readings if required. Monthly quality assurance measurements have shown the mean difference between Takeda and Hawksley machines to be 1 (4) mm Hg or less.
At entry to the UK prospective diabetes study and subsequently every three years all patients had a clinical examination which included retinal colour photography, ophthalmoscopy, measurement of visual acuity, assessment of peripheral and autonomic neuropathy, chest radiography, electrocardiography, and measurement of brachial and posterior tibial blood pressure using Doppler techniques. Annual direct ophthalmoscopy was also carried out. Every year a fasting blood sample was taken to measure glycated haemoglobin (haemoglobin A1c), plasma creatinine concentration, and concentrations of urea, immunoreactive insulin, and insulin antibodies; random urine samples were taken for measurement of albumin concentration.
Visual acuity was measured with Snellen charts until 1989, after which ETDRS (early treatment of diabetic retinopathy study) charts22
were used to assess best corrected vision, with current refraction or through a pinhole. Retinal colour photographs of four standard 30° fields per eye (nasal, disc, macula, and temporal to macular fields) were taken plus stereophotographs of the macula. Repeat photography was arranged if the quality of the photograph was unsatisfactory. Retinal photographs were assessed at a central grading centre by two independent assessors for the presence or absence of diabetic retinopathy. Any fields with retinopathy were graded by two further senior independent assessors using a modified ETDRS final scale.22
Neuropathy was assessed clinically by knee and ankle reflexes, and by biothesiometer (Biomedical Instruments, Newbury, Ohio) readings taken from the lateral malleoli and the end of the big toe.22
A 12 lead electrocardiogram was recorded and given a Minnesota code,22
and a chest x
ray film was taken for measurement of cardiac diameter.
Biochemical methods have been reported previously.23,27
Urinary albumin concentration was measured by an immunoturbidimetric method with a normal reference range of 1.4
mg/l to 36.5 mg/l.27
Microalbuminuria has been defined as a urinary albumin concentration of
and clinical grade proteinuria as a urinary albumin concentration of
Clinical end points
Twenty one clinical end points were predefined in the study protocol.22
All available clinical information was gathered for possible end points—for example, copies of admission notes, operation records, death certificates, and necropsy reports. Copies of these, without reference to the patient’s allocated or actual treatment, were formally presented to two independent physicians who allocated an appropriate code from the ninth revision of the international classification of diseases (ICD-9) if the criteria for any particular clinical end point had been met. Any disagreement between the two assessors was discussed and the evidence reviewed. If agreement was not possible the information was submitted to a panel of two further independent assessors for final arbitration. The closing date for the study was 30 September 1997.
End points were aggregated for the main analyses. The three predefined primary outcome analyses were the time to the occurrence of (a) a first clinical end point related to diabetes (sudden death, death from hyperglycaemia or hypoglycaemia, fatal or non-fatal myocardial infarction, angina, heart failure, stroke, renal failure, amputation (of at least one digit), vitreous haemorrhage, retinal photocoagulation, blindness in one eye or cataract extraction); (b) death related to diabetes (death due to myocardial infarction, sudden death, stroke, peripheral vascular disease, renal disease, hyperglycaemia or hypoglycaemia); (c) death from all causes.
Secondary outcome analyses of four additional aggregates of clinical end points were used to assess the effect of treatments on different types of vascular disease. These were myocardial infarction (fatal or non-fatal myocardial infarction or sudden death), stroke (fatal or non-fatal stroke), amputation or death from peripheral vascular disease, and microvascular complications (retinopathy requiring photocoagulation, vitreous haemorrhage, and fatal or non-fatal renal failure).
Since a patient could in sequence have different end points, he or she could be included in more than one end point category.
Surrogate end points—
Details of subclinical, surrogate variables have been published.23
Analysis was on an intention to treat basis, comparing patients allocated to tight and less tight blood pressure control. Patients allocated to tight control with angiotensin converting enzyme inhibitors or β blockers were pooled in this paper for analysis. They are compared in the accompanying paper.29
Life table analyses were performed with log rank tests, and hazard ratios were obtained from Cox’s proportional hazards models and used to estimate relative risks. Survival function estimates were calculated using the product limit (Kaplan-Meier) method. In the text relative risks are quoted as risk reductions and significance tests were two sided. For aggregate end points 95% confidence intervals are quoted, whereas for single end points 99% confidence intervals are quoted to allow for potential type 1 errors. Similarly, 99% confidence intervals were used to assess surrogate end points that were measured at triennial visits. Mean (SD), geometric mean (1 SD interval), or median (interquartile range) values are quoted for the biometric and biochemical variables, with values from Wilcoxon, t
, or χ2
tests for comparisons. Risk reductions for surrogate end points were derived from frequency tables. The overall values for blood pressure during a period were assessed for each patient as the mean during that period and for each allocation as the mean of patients with data in the allocation. Control of blood pressure was assessed in patients allocated to the two groups who had data at nine years of follow up.
Hypoglycaemia was determined from the number of patients allocated to a treatment and continuing with it who had one or more minor or major hypoglycaemic episodes each year. Urinary albumin concentration was measured in mg/l. Change in diabetic retinopathy was defined as a change of two steps (one step in both eyes or two or more steps in one eye) with a scale from the worse eye to the better eye that included retinal photocoagulation or vitreous haemorrhage as the most serious grade. Visual loss was defined as the best vision in either eye, deteriorating by three lines on an ETDRS chart.
Both the UK prospective diabetes study and hypertension in diabetes study received ethical approval from the appropriate committee in each centre and conformed with the guidelines of the Declarations of Helsinki (1975 and 1983). All patients gave informed consent.
Data monitoring and ethics committee
The data monitoring and ethics committee examined the end points every six months to consider halting or modifying the study according to predetermined guidelines. These included a difference of three or more standard deviations by log rank test in the rate of deaths related to diabetes or deaths related to diabetes and major illness between the group assigned to tight control and that assigned to less tight control or between the group given captopril and that given atenolol.22
One of the stopping criteria was attained immediately before the scheduled end of the study.