VA‐I 1967.

Methods	Randomized, double‐blind, placebo‐controlled trial conducted in ambulatory patients in USA
Participants	N = 143, mean age 51 years, range not reported. 53.8% patients were African‐Americans. Male (100%). Baseline mean SBP/DBP was 186/121 mmHg and pulse pressure was 65 mmHg. The inclusion criterion was DBP < 115 to 129 mmHg. Patients were followed for 1.5 years. Exclusion criteria: surgically curable hypertension, uremia, and concomitant fatal diseases such as carcinoma. Patients with hemorrhages, exudates, or papilledema in the optic fundi, history of cerebral or subarachnoid hemorrhage, dissecting aneurysm, or congestive heart failure resistant to digitalis and mercurial diuretics were excluded. Additional exclusions included patients who wished to return to the care of their private physicians, those who would be unable to attend clinic regularly, for geographical or other reasons, and patients of dubious reliability such as alcoholics, vagrants, and poorly motivated patients.
Interventions	Step 1. HCTZ 100 mg plus reserpine 0.2 mg plus hydralazine 75 mg Step 2. hydralazine 150 mg
Outcomes	Mortality, stroke, CHD, CHF, and diastolic BP
Notes	Study design published: Freis ED. In: Gross F, editor(s). Antihypertensive Therapy; Principles and Practice, an International Symposium. New York: Springer‐Verlag, 1966:345‐54.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "A group of 143 male hypertensive patients with diastolic blood pressures (at the clinic) averaging between 115 mmHg and 129 mmHg were randomly assigned to either active (hydrochlorothiazide plus reserpine plus hydralazine hydrochloride) or placebo treatment." "A table of random numbers was utilized by the statistician in determining the assignments." "Patients classified by severity scores as having mild hypertension were randomized in a separate stratification from those with moderate hypertension." "There were no significant differences with regard to age, weight, duration of known hypertension, or family history of hypertension, between the placebo and active treatment groups (Tables 1 and 2). There were more Negro and diabetic patients in the actively treated than in the placebo group, but the differences were not significant. The various indices of severity, such as hospital and clinic blood pressure, funduscopic, cardiac, central nervous system, and renal abnormalities were essentially similar in the two groups." Comment: method of randomization was adequate.
Allocation concealment (selection bias)	Low risk	"At the time of randomization, a sealed envelope was opened, which assigned the patient to one of two possible regimens—active antihypertensive medications or their placebos." Fries 1966 stated allocation was to be accomplished by opening a numbered sealed envelope containing a card assigning patient to a code number regimen. The cards were made by statistician from table of random numbers. Comment: Allocation concealment was probably done.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Quote: "The tablets of hydralazine and the Veratrum compound were made up to appear and taste the same. Chlorothiazide, reserpine, and hydralazine also were prepared to resemble their respective placebos. As in the previous study, three different code numbers were assigned to each preparation, including the placebos." "A similar appearing placebo also was manufactured. These tablets were given the same code number identifications as used in the prior study except that the letter 'C' preceded the series of digits. By substituting the 'C' series medication for the prior 'A' series, those patients taking either reserpine plus hydralazine or reserpine plus placebo of hydralazine had chlorothiazide 500 mg twice daily added to their regimens, while those patients who were not treated with active preparations had only placebo of chlorothiazide added." "The double‐blind technique was employed by utilizing a series of complex code numbers to disguise the identity of the randomized treatments and by making active drugs and placebos identical in appearance. It is realized, however, that blood pressure levels and side effects made the maintenance of such a double‐blind study difficult and imperfect." Comment: although blinding was attempted, it was probably not successful.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Blinding of the outcome assessor was not reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	The total number of dropouts was 12, or 8.4%. Nine occurred during the first two months following randomization. Seven had been randomized to placebos and five to active drugs. Thus, the dropout rate was small and was approximately equally divided between the active‐ and placebo‐treated patients. There is considerable variation in duration of observation as patients recruited from April 1964 to December 1966, and study ended in May 1967. 38% of patients followed for two years or more. Duration of study averaged 15.7 months for placebo‐treated patients and 20.7 months for active‐treated group.
Selective reporting (reporting bias)	Low risk	The study reports blood pressure data and morbid events, such as death, dissecting aortic aneurysm, ruptured aortic aneurysm, cerebral hemorrhage, fundi striate hemorrhage, high BP leading to re hospitalization, cerebrovascular accident, increased creatinine and BUN, hyperglycemia and depression as terminating events Non‐terminating events reported are MI, CHF, cerebrovascular thrombosis, and TIA. Treatment failures were also reported.
Other bias	Unclear risk	COI was not reported