Efficacy is whether a treatment can work under ideal conditions, the kind a trial arranges on purpose; effectiveness is whether it does work in the conditions you actually live in, with missed doses, competing illnesses, and a clinic that runs late. The two words sound like synonyms, so most claims you read treat them as one. A treatment can have strong efficacy and only modest effectiveness, and that gap is not a scandal. It is the predictable distance between a controlled question and a messy one. When someone tells you a treatment works, the useful reflex is to ask which kind of working they mean. The specifics belong with a clinician who knows your situation.
Key points#
- Efficacy is measured under optimal, tightly controlled conditions; effectiveness is measured in routine, real-world care.
- The same treatment can produce two honest but different numbers, and neither is a lie.
- When real-world results fall short, the cause is usually delivery (missed doses, side effects, cost), not a weaker treatment.
- Reading a claim well means asking who was studied, how the treatment was delivered, and who dropped out before the result was reported.
The two questions hiding in one word#
"Does it work" is really two questions wearing the same coat. The first asks whether a treatment can produce a benefit at all when nothing is allowed to get in the way. The second asks whether that benefit survives contact with ordinary life. Researchers separate them deliberately, and the distinction is old enough to have formal criteria for sorting one kind of trial from the other, as the AHRQ framework on efficacy versus effectiveness trials lays out.
Neither question is a criticism of the other. Efficacy comes first for a reason, because isolating a treatment from everything that could muddy the signal is the only way to learn whether the treatment itself does anything. Effectiveness is what you ultimately care about, because it describes the world you occupy, and trouble starts only when a number earned by answering one question gets quoted as if it answered the other.
A test kitchen and a weeknight#
Think of a recipe developed in a test kitchen. Professional cooks measure to the gram, own sharp knives, and repeat the dish until it is reliable. The published result is real, and it tells you the dish can turn out beautifully. That is efficacy.
Now cook the same recipe on a Tuesday after work, halfway distracted, missing one ingredient, with an oven that runs hot. The result is usually decent and rarely identical to the photograph. That is effectiveness. Nobody lied to you. The two outcomes describe the same recipe meeting two very different sets of demands.
A trial built to measure efficacy behaves like the test kitchen on purpose. As reviews of pragmatic and explanatory trials describe, such studies often enroll people who are younger, more likely to take every dose, and carrying fewer other conditions than the average patient, then deliver care through experienced staff who follow up closely. Each choice is defensible. Together they describe a ceiling, not a floor.
Why the two numbers drift apart#
Everyday care removes the supports a trial supplies. Doses slip because life is busy. Other conditions complicate the picture. A side effect that a motivated volunteer accepted for a few months becomes, for someone with a full life, the reason they stop after a few weeks, and the benefit anyone actually captures depends on whether they can stay on the treatment long enough to receive it.
So when effectiveness lands below efficacy, the usual explanation is undramatic. The treatment still does what the science said, but fewer people receive its full benefit; something that markedly lowers a risk when taken daily lowers it less across a population where many people stop within a year. The medicine did not weaken. The amount of it that reached people did. Work on real-world evidence, including reviews of studies that bridge trials and routine care, frames this as a delivery gap to close rather than a reason to abandon a sound treatment. The practical target becomes the thing in the way: a side effect that needs managing, a regimen too complex to keep up, a cost that forces people to ration.
The opposite error is more flattering and just as common. A strong efficacy figure gets read as a promise of everyday benefit. Then routine care underdelivers, and people feel cheated by a number that was never a forecast. Treating a ceiling as an average manufactures disappointment the evidence never earned.
How a claim can be true and still mislead#
Promotional language likes to hold the efficacy number and the everyday feeling in the same hand: a message can display a result obtained under controlled conditions while implying, through tone and imagery, that this is what to expect on an ordinary morning. The figure is accurate. The suggestion wrapped around it is not, because the conditions that produced it go unmentioned. This is a pattern in how claims get framed, not an accusation against any one maker.
A subtler version reports results only from the people who completed the full course, after those who could not tolerate it have already left the frame. What remains is a flattering portrait of the disciplined few, presented as if it described everyone who started. You do not have to assume bad faith to protect yourself from either move. Ask what conditions produced the claimed effect, and ask who is no longer in the picture when the result is reported.
A quick checklist for reading a claim#
Start with the people. Who was studied, and who was screened out? Broad eligibility that resembles the people who will actually use the treatment leans toward effectiveness. Narrow criteria, or volunteers far healthier than typical, lean toward efficacy, a best case rather than an average one.
Next, look at delivery and staying power. Specialist centers, intensive monitoring, and near-perfect adherence point to efficacy. Ordinary clinicians, normal follow-up, and an honest count of who stopped point to effectiveness. Analyses that keep everyone who started respect exactly the messiness effectiveness is meant to capture.
Finally, read the outcome. An effect on something people can feel and live by, measured over a meaningful stretch of time, supports an effectiveness claim. An effect on a short-term measurement that lives mainly inside the study supports an efficacy claim. When a claim cannot tell you its conditions, treat it as incomplete rather than proven.
Sources and further reading
Questions and answers
Does lower effectiveness mean a treatment is bad?
Usually not. A drop from efficacy to effectiveness most often reflects how much of the treatment actually reaches people in daily life, not a flaw in the treatment itself, and it points toward fixing the obstacle, such as side effects, complexity, or cost, rather than discarding a treatment that works when taken as intended.
How can I tell which one a study measured?
Look at three things: who was enrolled, how the treatment was delivered and monitored, and how long people were followed. Tight eligibility, expert delivery, and short specialized outcomes suggest efficacy. Broad enrollment, ordinary care, honest dropout accounting, and outcomes people can feel suggest effectiveness.
Why keep both ideas in mind?
They answer different questions and need each other. Asking only whether something can work collects benefits that never reach a kitchen table. Asking only whether it works here would never explain why something fell short. Holding both makes any single claim much harder to oversell, and for anything touching your own health, the specifics belong with a clinician who knows you.