How we read a study

You do not need a science degree to tell a strong study from a weak one. You need a handful of questions, and the willingness to ask them before getting excited. This is the checklist behind every issue of PD Brief, written out so you can apply it to anything you read.

1. How many people were in it?

Size is the single most useful signal available to a non-specialist. A study of 20 people can suggest an idea. It cannot establish one. Small studies produce unstable results: chance alone will hand a few of them a striking finding that vanishes when someone tries again with 200 participants.

A rough guide for treatment trials: under 50 people is preliminary, 50 to 200 is worth attention, and several hundred over multiple years is the kind of evidence that changes practice.

2. Was there a comparison group, chosen at random?

If everyone in a study received the treatment, and they improved, that improvement could be the treatment — or the natural fluctuation of the disease, or the attention that comes with being in a study, or the placebo effect.

Parkinson's has an unusually strong placebo response. Expecting a benefit can trigger real dopamine release and produce measurable improvement in movement. Any Parkinson's trial without a placebo group should be read with that firmly in mind.

The strongest design is the randomised controlled trial: assign people to treatment or placebo purely by chance, so the groups are otherwise comparable.

3. Did anyone know who was getting what?

If participants know they are on the real drug, they tend to report feeling better. If the doctor scoring their symptoms knows, they tend to score more generously. Neither requires dishonesty; it is how human judgement works.

Double-blind means neither side knows. It is the standard for good reason.

4. Was it in people, or in mice?

A great deal of exciting Parkinson's coverage describes work in mice, rats, or cells in a dish. This research is essential — it is where ideas come from — but the failure rate on the way to humans is brutal. Many treatments have cured Parkinson's-like conditions in rodents. None of them has cured Parkinson's.

When you read a hopeful headline, check what the subjects were. It is often the whole story.

5. Did it measure what actually matters?

Some studies measure how a person feels and functions. Others measure a number in a scan or a blood sample. The second kind is faster and cheaper, but a drug can move a laboratory measurement without helping anyone.

Ask what changed: a person's ability to walk, dress, and sleep — or a marker that is merely believed to correspond to those things.

6. Did it hit the target it named in advance?

Researchers declare a primary endpoint before a trial starts: the one main question it will answer. This exists to prevent a trial that failed from being rewritten around whichever measurement happened to look good.

If a report emphasises a benefit in a subgroup — "worked well in women under 60" — while the main result was negative, that is a hypothesis for a future study, not a finding.

7. Is the effect big enough to feel?

Statistically significant does not mean large. With enough participants, a difference far too small to notice will still clear the significance threshold.

So ask the second question: how big was it, and would a person living with the disease be able to tell? The two questions are independent, and the second is often the more important one.

8. Who paid for it, and who ran it?

Funding does not invalidate research, and most drug trials are necessarily funded by the companies that make the drug. But it is worth knowing. Independently funded studies of a treatment tend to report smaller effects than company-funded ones.

Good papers state their funding and conflicts of interest openly. The information is usually near the end.

9. Has anyone repeated it?

A single study is a data point. Science advances when independent groups reproduce a finding. The first report of anything, however striking, is provisional by definition.

This is the reason PD Brief keeps an archive rather than only publishing the newest thing: watching a claim get confirmed, qualified, or overturned across several years is where the real understanding lives.