How to Read a Nutrition Study Abstract
To read an abstract, pull out six facts (who was studied, what they did, what it was compared with, what was measured, how many people and for how long) before you look at the result.
On this page
- The six facts to pull out
- Where each fact sits in an abstract
- Worked example 1: a diet trial (Shai 2008)
- Worked example 2: a multivitamin trial (Gaziano 2012)
- Reading the numbers
- A one-minute checklist
- Related
The six facts to pull out
Researchers often shorten the first four to "PICO." Size and duration are the two that headlines drop most often.
| Fact | Question | Why it matters to you |
|---|---|---|
| Population | Who was in the study? | Results in older men may not apply to a woman of 45 |
| Intervention | What did they do or take? | "Low-carb" can mean 20 g a day or 150 g a day |
| Comparison | What was it tested against? | Beating "no change" is easier than beating another diet |
| Outcome | What did they measure? | Weight and blood markers are not the same as heart attacks |
| Size | How many people? | Small studies swing more by chance |
| Duration | How long did it run? | Many diet differences shrink after 6 to 12 months |
Where each fact sits in an abstract
Most medical journals use a structured abstract with labeled parts. Each fact has a usual home.
- Background or Objective. The question the study asked.
- Methods or Design. Population, intervention, comparison, size and duration. Look for "randomized," which means people were assigned by chance.
- Results. The outcome numbers, usually with a confidence interval or p-value.
- Conclusions. The authors' own reading. Check it against the results; a conclusion can say more than the numbers do.
Abstracts are free on PubMed. For search tips, see how to search PubMed for diet studies.
KetoDick on this: he points viewers to Google Scholar and PubMed and says a study has to state its protocol, meaning what was actually done (video 1JbK1_ClDmY). He also says he lists his sources so viewers can "look at the quality of the research" (from the video) and take it to their doctor (video UygKGoP_BtA).
Worked example 1: a diet trial (Shai 2008)
Shai et al., 2008, the DIRECT trial, is one of the most-cited low-carb trials. Here are the six facts from the evidence page.
| Fact | DIRECT |
|---|---|
| Population | 322 adults |
| Intervention | A low-carb diet |
| Comparison | A low-fat diet and a Mediterranean diet |
| Outcome | Weight loss and blood lipids, including the total-cholesterol-to-HDL ratio |
| Size | 322 |
| Duration | 2 years |
| Design | Randomized controlled trial |
Result: mean weight loss was 2.9 kg on low-fat, 4.4 kg on Mediterranean and 4.7 kg on low-carb. Low-carb gave the largest improvement in the total-cholesterol-to-HDL ratio.
How to read it:
- These are averages. Some people lost more, some less, and some may have gained.
- The gap between low-carb and Mediterranean was 0.3 kg on average. The gap between low-carb and low-fat was 1.8 kg.
- Two years is long for a diet trial, which makes the result more useful than a 12-week study.
- The cholesterol result is a ratio, a lab marker. It is not a count of heart attacks.
Compare it with Foster et al., 2003: 63 adults, where low-carb lost more at 3 and 6 months but the difference was not significant at 12 months, and many people dropped out of both groups. Shorter and smaller studies can tell a different story. See keto vs low-fat.
Worked example 2: a multivitamin trial (Gaziano 2012)
Gaziano et al., 2012 reports the cancer results of the Physicians' Health Study II (PHS II).
| Fact | PHS II (cancer) |
|---|---|
| Population | Male physicians aged 50 and over |
| Intervention | A daily multivitamin |
| Comparison | Placebo |
| Outcome | Total cancer |
| Size | 14,641 |
| Duration | Median 11.2 years |
| Design | Randomized controlled trial |
Result: a modest reduction in total cancer, with a hazard ratio (HR) of 0.92.
How to read it:
- The population is narrow. Every participant was a male doctor aged 50 or over. The result may not apply to women, and the trial did not test women.
- An HR of 0.92 means the multivitamin group had about 8% lower relative risk over the study. It does not tell you how many cancers that is per 100 people. See relative vs absolute risk.
- The same trial also measured heart outcomes. Sesso et al., 2012 found no reduction in major cardiovascular events (HR 1.01). One trial, two outcomes, two different answers.
For the wider multivitamin picture, see vitamins on keto and do multivitamins work?.
KetoDick on this: in his beginner basics he suggests a common multivitamin with magnesium and calcium, naming no brand (video UygKGoP_BtA). PHS II tested a daily multivitamin in men and found a modest cancer reduction and no heart benefit; it did not test a keto diet.
Reading the numbers
| Term | Plain meaning |
|---|---|
| Mean | The average across the group |
| Hazard ratio (HR) | Risk in one group compared with the other over time; 1.0 means no difference |
| 95% confidence interval (CI) | The range the true effect probably falls in; if it crosses 1.0 (for a ratio) or 0 (for a difference), the result may be chance |
| p-value | How likely a result this large would be by chance alone; below 0.05 is the usual cut-off for "significant" |
| "Not significant" | The study could not tell the groups apart, which is not the same as proving they are equal |
A small study can show a big effect that disappears in a larger one. Hall et al., 2016 had 17 men in a metabolic ward, which gives tight control but very few people.
A one-minute checklist
- Find "randomized" or "cohort" in the methods.
- Write down who, how many and how long.
- Name the comparison group.
- Check whether the outcome is a real event or a lab marker.
- Read the result numbers before the conclusion sentence.
- Ask whether the people studied look like you.
If you take diabetes or blood pressure medication, have kidney disease or are pregnant, a single study is a reason to ask your doctor a question, not to change your dose.
Related
Guide: Chapter 12: Reading nutrition news and studies · Observational studies vs randomized trials · Diet headlines · Red flags in diet advice · Keto vs low-fat
Blog: How to search PubMed · Relative vs absolute risk · Do multivitamins work? · Keto vs low-carb
Videos: Men's Health hack job on keto · Keto basics for beginners
Last verified: 2026-10-31