How Trials Report Harms, and Why the Safety Half Is Often Thin

Randomized trials usually report benefits in more detail than harms, so the safety half of a paper is often thin. The CONSORT harms guidance sets out how adverse events should be defined, collected, analyzed, and reported so that safety is built into the trial report. Before trusting a low adverse-event count, check how harms were collected, since active structured surveillance finds far more events than passive spontaneous reporting.

The lopsided ledger

Most trials are designed to detect whether a treatment works, and their machinery is tuned to that question. Harms get far less attention by default. The result is a lopsided ledger in which the benefit column is detailed and the harm column is vague, even though patients and clinicians need both to make a decision.

Reporting guidance for harms exists because this imbalance is the norm, not the exception. Knowing that pattern is the first step to reading the safety section with the right level of skepticism.

What CONSORT says about harms

The main CONSORT statement, the standard checklist for reporting randomized trials, contains only one item that specifically addresses harms. To close that gap, the CONSORT group published a harms extension, later updated as CONSORT Harms, spelling out how adverse events should be defined, collected, analyzed, and presented.

The update modified a set of the main checklist items and added new ones, so that harms reporting is woven into the trial report rather than bolted on at the end. When a paper follows it, the safety section stops being an afterthought.

Active versus passive collection

How harms are gathered changes the numbers enormously. Passive or spontaneous collection, where patients mention problems unprompted, captures far fewer events than active surveillance with structured questionnaires at set visits.

So a trial reporting few adverse events may have a genuinely safe treatment, or it may simply have asked poorly. Good reporting states the method of ascertainment, which lets you tell the difference between a quiet drug and a quiet questionnaire.

Denominators, definitions, and severity

A harm count means little without its denominator and its definition. You need to know how many people were at risk, over what period, how an event was defined, and how its severity was graded.

Withdrawals due to adverse events are a particularly honest signal, because they combine frequency and importance from the patient's own point of view. Vague phrases such as well tolerated are claims, not data, and should be read as an invitation to go find the actual numbers.

Why harms hide

Several forces push harms into the shadows. Trials are usually too small and too short to detect rare or delayed harms with confidence. Composite or selective reporting can bury specific events inside broad categories.

There is also a subtle asymmetry. An underpowered efficacy result is rarely treated as proof of no benefit, yet an underpowered safety result is often treated as reassurance. Absence of reported harm is not the same as evidence of safety, and the two are easy to confuse.

What to check before you trust the safety half

Ask whether harms were prespecified as outcomes or only noted incidentally. Look for the method of collection, clear definitions, denominators, and severity grading. Check whether the analysis included everyone randomized, and whether serious events and dropouts due to side effects are reported separately by arm.

When a paper devotes pages to efficacy and a single sentence to safety, read that ratio as information about the reporting, not about the drug. A thin safety section tells you to withhold confidence until better data arrive.