Papademetriou 2006.
| Methods | Randomized, double‐blind, placebo‐controlled trial (parallel arms). Factorial design. Wash‐out period = 4 to 5 weeks. Multicenter, USA | |
| Participants | Sitting DBP 95 to 114 mmHg and SBP < 180 mmHg. Mean age 53 years. Males ˜50%. Mean body mass index > 30: 57% of patients. Baseline BP was 151/100 | |
| Interventions | ER‐metoprolol 25 mg/ (N = 89), 50 mg (N = 94), 100mg (N = 96) or 200 mg/d (N = 52), ER‐metoprolol 25 mg, 50 mg, 100 mg or 200 mg/d + hydrochlorothiazide 6.25 mg/d or 25 mg/d (all combined, N = 849), hydrochlorothiazide 6.25 mg (N = 86), 12.5 mg (N = 105), 25 mg daily (N = 48) or placebo (N = 152) Trial duration = 8 weeks (followed by a 2‐week drug taper period) | |
| Outcomes | Change from baseline in trough sitting and standing DBP and SBP; peak BP; BP response rate; ECG, heart rate, hematology, urinalysis and serum biochemistry. Subgroup analyses were performed (i.e. based on age, sex and race) | |
| Notes | A sample size calculation was provided and was based on 1485 patients to detect a difference in standard deviation of 8 mmHg (sitting DBP) at 80% power. The majority of patients in the study (57%) were overweight with a body mass index (BMI) over 30, a characteristic that is not reflected in most other trials included in this review. The study authors stated that baseline patient characteristics were well balanced across treatment groups, however the reviewers determined there to be statistically significant differences (P value < 0.05) in terms of mean age, incidence of diabetes mellitus and BP goals above JNC‐7 at baseline. Total AEs were not reported | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "This was a multicenter (in the USA), randomised, double‐blind, placebo‐controlled, parallel group, unbalanced factorial study..." (line 1 under "Study Design" p.1218) "...eligible patients were randomly allocated to one of the 17 treatment groups..." (line 4 from bottom of p.1218). "A central, computer‐generated randomisation schedule using an interactive voice response system allocated patient to treatment groups within the study centre. As certain treatment groups were of greater importance for pair wise comparisons, patients were assigned in an allocation ratio of 1 (indicating least important), 2 (more important), or 3 (most important)". (line 1 under "Randomization, Blinding and Study Treatments" p.1218) |
| Allocation concealment (selection bias) | Low risk | "A central, computer‐generated randomisation schedule using an interactive voice response system allocated patient to treatment groups within the study centre." (line 1 under "Randomization, Blinding and Study Treatments" p.1218) |
| Blinding (performance bias and detection bias) All outcomes | Low risk | The dose of ER‐metoprolol was doubled after 1 week of therapy, however the investigators stated that it was a "blinded escalation to the assigned dose". (line 1 from top of p.1218) "To blind study treatments, HCTZ in identically appearing tablets of 6.25, 12.5, or 25 mg, or matching placebo and ER‐metoprolol tablets of 25 or 100 mg or matching placebo, were blister‐card packaged according to a double‐dummy design." (line 7 under "Randomization, Blinding and Study Treatments p.1218) |
| Incomplete outcome data (attrition bias) All outcomes | Unclear risk | Primary efficacy analysis was based on an intention‐to‐treat (ITT) technique with the LOCF (last observation carried forward) Exclusions: 1260/2831 (44.5%) of patients were excluded from the study during the placebo run‐in period prior to randomization. Some of the reasons included patients failing eligibility criteria (814) and those withdrawing consent (262) Attrition: 176/1571 (11%) of patients withdrew from the study for the following reasons: 46 ‐ "adverse event", 51 ‐ "lack of therapeutic response", 45 ‐ "patient consent withdrawn", 19 ‐ "loss to follow‐up" and 15 ‐ "other reasons". Note that since the data were combined, it was impossible to ascertain from which treatment groups the withdrawn patients originated (there were 17 treatment groups in total, some with patients receiving non‐thiazide mono‐ and combo‐therapies) WDAEs: a total of 46/1571 (2.9%) patients withdrew due to adverse events (5 from non‐thiazide treatment groups), 3/72 (4.2%) patients receiving HCTZ 12.5 mg and 2/66 (3%) patients on placebo. The precise reasons were not given |
| Selective reporting (reporting bias) | Unclear risk | The study was a 4 x 5 factorial design which means there were 20 possible treatment groups to assign patients, however the authors stated that 3 of the groups (combo‐therapy: ER‐metoprolol/HCTZ: 200/6.25, 25/25 and 50/25 mg) were not investigated (see line 5 under "Study Design" p.1218). Results for serum chemistry (except for potassium and uric acid levels), hematology, lipids and urinalysis tests and ECG were measured but actual values were not reported at the study's endpoint. Standing BP data were measured, but not reported on. Peak:trough BP was measured but not reported for each separate treatment group. Subgroup analyses of sex (male versus female) age (< 65 years versus ≥ 65 years) and ethnicity (African American versus other) were documented in the results section of the study only (actual values were not shown for sex and age differences); but not included in the methods. Mortalities: none. SAEs: 5 patients had a "coronary artery disease manifestation" e.g. myocardial infarction, but it was not stated from which treatment group the patients originated. Total AEs were not reported. Only the single most commonly occurring AE (headache) was reported in the study |
| Industry sponsorship | High risk | Supported by AstraZeneca LP |