The First Randomized Controlled Trial and Why Its Design Still Matters

The trial widely accepted as the first properly randomized controlled trial was the Medical Research Council's 1948 study of streptomycin for pulmonary tuberculosis, published in the British Medical Journal and guided by the statistician Austin Bradford Hill. Its lasting contribution was the method rather than the drug. Patients were assigned to streptomycin plus bed rest or to bed rest alone using a schedule of random sampling numbers, and that schedule was concealed inside sealed envelopes so no clinician could steer who received the new treatment. Those two features, random allocation and concealment of the allocation, are still what separate trustworthy evidence from the impressive but unreliable kind.

Why streptomycin, and why 1948

Streptomycin arrived in the mid-1940s as the first antibiotic with real activity against the tuberculosis bacillus, at a moment when tuberculosis was still a leading cause of death among young adults. Post-war Britain faced a severe dollar shortage and could import only small quantities of the drug. As the James Lind Library account documents, Bradford Hill argued that precisely because supply was scarce, it would be defensible, and arguably fairer, to allocate the limited streptomycin at random rather than by physician preference or social standing. Scarcity turned a controlled comparison from an ethical problem into an ethical solution: no one could receive the drug on demand, so a lottery was a reasonable way to distribute it while learning whether it actually worked.

The design itself was disciplined. The investigation enrolled 107 patients aged 15 to 30 with acute progressive bilateral pulmonary tuberculosis, a form serious enough that a genuine effect should be visible yet variable enough that individual recoveries could mislead. Fifty-five were allocated to streptomycin plus bed rest and 52 to bed rest alone. As the retrospective account in the peer-reviewed record notes, within the first six months there were 4 deaths among the 55 streptomycin patients against 15 among the 52 given bed rest alone. The direction was clear, and it was clear because of how the comparison was built.

The problem randomization solves

Before 1948, most treatments were judged by comparing treated patients with earlier patients, or with whichever patients a doctor chose to treat. That approach hides a fatal flaw. If clinicians decide who gets a promising new drug, they tend, consciously or not, to give it to the patients most likely to do well, or sometimes to the sickest as a last resort. Either way the groups differ before treatment begins, and any difference afterward cannot be cleanly attributed to the drug. Statisticians call this selection bias, and it has flattered countless useless remedies.

Random allocation breaks that link. When a coin toss or a random number decides the assignment, the two groups become comparable on average, both on the factors we can measure, such as age and disease severity, and on the ones we cannot. Whatever difference emerges at the end is far more likely to reflect the treatment than a stacked deck. The streptomycin trial was among the first to make that logic the backbone of a clinical comparison rather than an afterthought.

Allocation concealment, the quieter innovation

Randomization gets the headlines, but the subtler achievement of the 1948 trial was allocation concealment, and the two are not the same. A trial can generate a perfectly random sequence and still be corrupted if the people enrolling patients can see or guess what comes next. A clinician who knows the next slot is the streptomycin arm might nudge a favored patient into the queue, or hold a frail one back. In the MRC trial, as Iain Chalmers has documented in the James Lind Library, the allocation schedule was kept from the physicians admitting patients. Assignments sat in sealed envelopes marked only with a hospital name and a number, opened in sequence and only after a patient had already been accepted into the study. The decision to enroll was therefore made blind to the assignment, which protects the randomization at the exact moment it is most vulnerable.

Outcome assessment was guarded too. The chest radiographs, the main measure of progress, were read independently by assessors who did not know which treatment each patient had received. That blinding matters whenever an outcome requires human judgment, because expectation quietly shapes interpretation. The trial was not placebo-controlled in the modern sense, and its authors were candid about that, but its combination of concealed random allocation with blinded assessment set a template that the field spent the next half century formalizing.

Why the design still matters

These are not antiquarian details. Decades later, Kenneth Schulz and colleagues examined a large set of controlled trials and found, as recorded in the James Lind Library, that trials with inadequate or unclear allocation concealment tended to exaggerate treatment effects compared with trials that concealed allocation well. In other words, the specific safeguard Bradford Hill built into a tuberculosis study in 1948 turns out to predict, empirically, whether a trial's result can be believed.

That gives any reader a practical checklist. When a supplement, device, longevity intervention, or new therapy is promoted on the strength of a study, the questions worth asking are the ones the streptomycin trial answered first: were participants actually randomized, was the upcoming assignment hidden from the people enrolling them, and did someone blinded to treatment assess the outcome? Testimonials, before-and-after images, and comparisons with the past fail all three tests, which is why they persuade so easily and mislead so often. The 1948 trial is remembered as a landmark less for curing patients than for teaching medicine how to know when it has. This article is educational and is not medical advice.