Four questions answer most of what an abstract can support: who was studied, what they were compared against, what was measured, and for how long. Answer those before reading anyone’s summary of the result.
Population: who was actually enrolled
Populations are narrow on purpose. A 2007 randomized trial of tesamorelin enrolled 412 patients with HIV and abdominal fat accumulation, and a 2021 trial of semaglutide enrolled 1,961 adults with overweight or obesity without diabetes alongside a lifestyle intervention. Those are different questions with different answers. [Falutz et al., 2007][Wilding et al., 2021 (STEP 1)]
Comparator: compared with what
Without a comparator, change over time can reflect the natural course, measurement drift or co-interventions. Placebo-controlled randomized designs , such as the 48-week phase 2 retatrutide trial , exist to separate the intervention from everything else happening to participants. [Jastreboff et al., 2023]
Endpoint: what was measured
Endpoints can be specific and narrower than the headline suggests. A visceral-fat measure is not general weight loss; a laboratory marker is not a symptom; and an indication in a regulated label is written for a defined population, as in United States bremelanotide labelling limited to acquired, generalized hypoactive sexual desire disorder in premenopausal women, with stated exclusions. [Falutz et al., 2007][US bremelanotide prescribing information]
Follow-up: for how long
Durations in these reports range from roughly six months to over a year. A 26-week or 48-week result says nothing certain about year three, and a short study cannot characterise effects that emerge slowly or resolve quickly. [Falutz et al., 2007][Jastreboff et al., 2023]
Reading safety results: what “none reported” can and cannot tell you
Read the safety table, not just the reassurance
A study summary may say a treatment was “well tolerated.” Before accepting that description, look for the numbers and the method used to collect them. Missing information and a measured zero tell different stories.
An adverse event is an unwanted medical occurrence during treatment; the term alone does not establish that treatment caused it. The task is to understand what was observed, in whom, and under which conditions. [Phillips et al., 2019]
Three statements that should stay separate
| Statement | What it tells you |
|---|---|
| Not reported | You cannot find the result in the material reviewed. Record the gap; do not enter zero. |
| No events observed | A result for a defined event, group and observation period. Keep those details attached. |
| No statistically significant difference | The analysis did not establish a difference at its chosen threshold. That does not establish equal risk. [Phillips et al., 2019] |
Even a well-run trial may be too small or too short to reveal uncommon or delayed problems. Its eligibility rules also limit whose experience it describes. “No serious adverse events” is therefore a narrower statement than “safe.” [CONSORT Harms, 2023]
Five questions to ask of the report
1. How were events collected? Look for scheduled questions, examinations or tests, and the observation period. If collection methods are unclear, mark them as unclear. [CONSORT Harms, 2023]
2. Who is in the denominator? Check the safety-analysis population rather than assuming it equals everyone randomized. Keep the number assessed beside each result. [Phillips et al., 2019]
3. What is being counted? People with an event and total events are different measures. Repeated events can disappear in a count of people affected. [Phillips et al., 2019]
4. What did the table leave out? Check thresholds, such as showing only common events, and consult the supplement. A restricted table is not a complete event inventory. [Pitrou et al., 2009][CONSORT Harms, 2023]
5. What happened to participants who stopped? Look for withdrawals due to adverse events, with reasons and results by group. Do not infer their experience from those who completed follow-up. [Pitrou et al., 2009]
This is a reading aid, not a validated score. Use the paper, supplement and protocol together. The BLOOM guide to following Canadian trial records can help trace those documents.
A blank cell is not a zero
Imagine a hypothetical table that includes only events affecting at least 5% of participants. A particular event does not appear. You cannot conclude that nobody experienced it: the table’s inclusion rule could have kept it out. A useful note is: “Not listed in this table; check the full event results.” This example describes no actual trial or BLOOM product. [CONSORT Harms, 2023]
Why this check is worth doing
In a 2009 audit of 133 randomized-trial reports, Pitrou and colleagues found that 47.4% supplied no data on withdrawals due to adverse events. About one-third restricted which harms they reported. These were reports from a historical publication window, not a survey of today’s peptide literature. [Pitrou et al., 2009]
A separate review published in 2019 examined 184 drug-trial reports from four major medical journals. Only 29% described the specific prompts used to collect adverse-event information. This audit examined reporting practices, not whether particular drugs or research products were safe. [Phillips et al., 2019]
These findings justify checking the documentation. They do not tell us that every omitted item concealed a problem, or that the same percentages apply to current research.
What Canadian adverse-reaction records can tell you
Canada Vigilance contains reports of suspected reactions to health products. It can help identify possible safety signals, but a report does not by itself establish causation. Health Canada cautions that reporting is incomplete and the number of people exposed is unknown. [Canada Vigilance guidance, 2022]
That means you cannot turn the number of reports into an incidence rate, or rank two products by comparing their totals. Likewise, an empty search is not evidence of safety. Keep a database search separate from a trial result with a defined population and follow-up period. [Canada Vigilance guidance, 2022]
A study is not a product authorization
Health Canada’s 9 April 2026 advisory warns against unauthorized injectable peptide drugs and states that research-use labelling does not exempt products from regulatory requirements. A study citation or certificate of analysis does not establish authorization for a particular product. See the BLOOM guide to research materials and authorized medicines for the separate product-status check. [4]
Leave a note another reader can check
Record the event definition, group sizes, observation period, results, withdrawals and anything missing. Cite the table or supplement you used. A specific statement about what the study measured is more useful than a broad reassurance.
One-sentence discipline
- Write: in [population], over [duration], [intervention] versus [comparator] showed [endpoint result].
- Add the reported harms in the same sentence or the next one.
- Name the publication year and study stage exactly as reported.
- Do not substitute a product you can buy for the product that was studied.