This study has been described according to the CONSORT guidelines for the presentation of clinical trials.11
This randomised, double blind, double dummy study was performed in 37 centres (12 in Australia, nine in Denmark, four in Finland, and 12 in Israel) in patients with type 2 diabetes who were aged between 30 and 75 years old and had previously diagnosed hypertension and microalbuminuria.
We included patients if the urinary albumin:creatinine ratio was 2.5-25 mg/mmol and the diastolic blood pressure was 90-110 mm Hg after two and four weeks of placebo treatment, respectively. Exclusion criteria were: body mass index
, systolic blood pressure >200 mm Hg, non-diabetic cause of secondary hypertension, cardiovascular event in the past six months, serum creatinine concentration
130 ×6d mol/l in women and
150 ×6d mol/l in men, serum potassium concentration >5.5 mmol/l, glycated haemoglobin concentration (HbA1c
) >10%, pregnancy or potential pregnancy, and breast feeding.
After four weeks of placebo treatment, eligible patients were randomised to four treatment groups. Figure gives details of the randomisastion and treatment. Consequently, half the patients received candesartan and half received lisinopril for the first 12 weeks. From 12 to 24 weeks, one third of the patients received candesartan alone, one third lisinopril alone, and one third the combination, unless patients had diastolic blood pressure below 80 mm Hg at 12 weeks.
Distribution of participants in study. Doses were: candesartan 16 mg once daily, lisinopril 20 mg once daily, or their combination
The patients attended the clinic for a total of nine study visits: at four and two weeks before randomisation, at randomisation (week 0), and at 1, 6, 12, 13, 18, and 24 weeks after randomisation. At each visit blood pressure was measured in the morning after five minutes of rest, about 24 hours after the previous drug administration, with an automatic device (Omron HEM-705 CP, Omron Electronics, Tokyo, Japan). Sitting blood pressure was measured three times with an interval of about two minutes, and the mean was calculated. The standing blood pressure was measured once after one minute of standing.
Microalbuminuria was determined two weeks before randomisation and at weeks 0, 12, and 24 by calculation of the urinary albumin:creatinine ratio.3
For each determination the patients brought early morning voided urine samples from two consecutive days. Albumin concentration was measured by immunoturbidimetry, and creatinine concentration was measured by autoanalyser. Creatinine clearance was calculated with the Cockroft-Gault formula ((140−age)×body weight (kg)×K/serum creatinine (μmol/l). K (constant) was 1.25 for men and 1.03 for women. Haemoglobin A1c
was measured by high performance liquid chromatography at weeks 0, 12, and 24. Clinical chemistry, haematology, and urinalysis were performed at weeks 0, 12, and 24 with standard methods. Serum creatinine and potassium concentrations were also measured at weeks 1 and 13. The ACE genotype was determined as previously described.12
Tolerability was assessed by using spontaneously reported adverse events, recorded in response to an open question or observed by the investigator at each visit.
The study was performed in accordance with the principles stated in the Declaration of Helsinki and approval was obtained from each institution's ethics committee. All patients gave their informed consent before being included in the study.
The assumed standard deviation for the change in urinary albumin excretion was 1.1 on a logged scale. This would allow estimation of the ratio of the expected medians with a relative error of at most 33% with a probability of 95%. As we used the mean of two early morning measurements we predicted that the variability between indiviuals would be reduced. In consequence, the observed relative error could be expected to be smaller than assumed. We therefore calculated that we needed about 220 patients.
For all treatments we analysed the changes from baseline (randomisation) to 12 and 24 weeks in blood pressure, urinary albumin:creatinine ratio, and creatinine clearence with a linear model for analysis of covariance with factors for treatment, centre, and interaction between them and baseline value as a covariate. For urinary albumin:creatinine ratio the changes in diastolic blood pressure and body weight were also used as covariates. The urinary albumin: creatinine ratio was analysed after logarithmic transformation. Differences between treatments were estimated from the fitted model (analysis of covariance). The results for urinary albumin:creatinine ratio are presented as estimates of the true treatment geometric means and as estimates of the ratios of the true treatment geometric means, with their 95% confidence intervals and corresponding P values. All analyses were based on intention to treat (defined as all patients who took at least one dose and had efficacy data available after randomisation), with the last value carried forward for missing values.