Controlled study

A controlled study pairs a group that receives an intervention — a substance, a dose, a procedure — against a comparison group that does not, allowing the outcome in each to be measured side by side. The comparison group is what makes the design controlled: it sets a baseline for what would have happened in the absence of the intervention.

Control is a design feature, not a quality stamp. A study can include a control group and still be small, unblinded, or poorly matched. On this encyclopedia, the term usually covers any design with a formal comparison arm — a range that runs from a small open-label trial to a large blinded experiment with hundreds of participants.

What this design can establish

A controlled study can establish whether an effect — a change in score, a symptom, a physiological measure — differs between the exposed group and the comparison group. It puts a number on that difference and, when the study is large enough, can assess whether the difference is likely to reflect a real pattern rather than chance alone.

When the control group receives a placebo and participants are unaware of which condition they are in, the design can separate the pharmacological contribution of the substance from expectation effects. When the control is an active comparator, the study compares two interventions against each other — answering a different question from one with a no-treatment arm.

Controlled studies can also detect safety signals: if adverse events cluster in the treatment group relative to the control, the design provides evidence for attributing them to the intervention — something a case series or uncontrolled observation cannot do.

What it cannot

A control group alone does not establish causation. If participants were not randomly assigned to their groups, systematic differences between who received the intervention and who did not may explain the result just as well as the intervention itself. Control and randomisation are separate design features; a controlled study does not automatically have both.

A controlled study also cannot generalise beyond the population it enrolled. Findings in one group — young healthy adults, for instance — may not hold for people who are older, have comorbidities, or take concurrent medications. The most common wrong inference is treating a positive result in a controlled study as evidence that an effect applies to anyone who uses the substance.

For much of the pharmacological record covered here, controlled studies have not been published. That absence appears throughout this encyclopedia as a stated gap — not as evidence of safety or of harm, but as a fact about what the record currently contains.