Reading a safety statement: what "no adverse events reported" establishes

Two phrases turn up constantly in device marketing and they look interchangeable. They are not. One is a reportable observation with a defined meaning; the other is a claim about the future that no finite amount of data can support. Keeping them apart is one of the cheapest ways to make a technical document more credible, so it is worth setting out the difference properly.

What an adverse event actually is

In regulatory usage, an adverse event is any untoward medical occurrence in a person who has received the treatment under study, whether or not it is judged to be caused by that treatment. The definition is deliberately wide, and the causal question is handled separately, afterwards, by assessment.

That last point is the one most often lost. “Adverse event” is not a synonym for “harm done by the device.” It is the raw observation, before anybody has decided what caused it. Narrowing the definition at collection time — only writing down what looks device-related — is how a safety record ends up cleaner than reality.

What “none were reported” depends on

A statement that no adverse events were reported is, read strictly, a statement about a reporting system rather than about the device. Four things determine what it is worth:

  1. How many people were exposed. The denominator is the whole argument, and it is the number most often missing.
  2. For how long. A week of exposure cannot speak to anything that takes a year to appear.
  3. Whether anyone was asked. Passive surveillance — waiting for people to come forward — and active surveillance — systematically asking every participant at fixed intervals — produce very different numbers from the same underlying reality.
  4. Whether a route to report existed at all. A system nobody knows how to use records nothing, which is not the same as nothing happening.

The arithmetic of a zero

There is a clean statistical result here that deserves to be better known, set out by Hanley and Lippman-Hand in JAMA in 1983 under a title worth remembering: If nothing goes wrong, is everything all right?

Their answer, in short, is the rule of three. If you observe n people and see zero events, the upper bound of the 95 % confidence interval on the true rate is approximately 3/n.

Work that through. Zero events in 100 people is consistent with a true rate as high as about 3 % — roughly one person in thirty. Zero in 300 still leaves room for about 1 %. To place the upper bound below one in a thousand, you need around three thousand observations. A zero is genuine information, and it is much weaker information than it looks.

This is the same discipline as any other measurement. A reading of zero on an instrument still has a resolution, and quoting the reading without the resolution overstates what was measured.

Why the absolute phrasing is a different claim

“No side effects” is not a stronger version of the same sentence. It is a different kind of sentence: a universal claim, covering every user, at every parameter setting, over any duration, including people who have not used the device yet. No finite series of observations reaches that, and the arithmetic above shows why — every real dataset leaves an upper bound above zero.

So it cannot be supported, and a reader who knows the field will spot that immediately. The narrower statement is both true and more persuasive to exactly the audience whose opinion matters: no adverse events have been reported, given with the denominator, the duration and the surveillance method alongside it.

What I hold myself to

The standard I apply across everything published here is the same one I would apply to somebody else's claim if I were assessing whether to license their technology. Report the observation, report its conditions, and let the reader do the arithmetic themselves. Where the conditions are not known, say that instead of rounding the claim upward.

It costs nothing. A statement with its denominator attached is more useful than an absolute one, it survives scrutiny, and it does not need revisiting when the dataset grows.

What would strengthen any safety record

Active rather than passive collection, at fixed follow-up intervals. A published denominator and exposure duration. A pre-specified definition of what gets recorded, written before enrolment. Independent adjudication of causality rather than self-assessment by the sponsor. And a post-market route that keeps collecting after the study closes, because the rarest events are precisely the ones a study of ordinary size cannot see.

For the clinical trial data and publication record associated with eMedica, the destination is emedica.in.

Source

Retrieved from PubMed.

Hanley JA, Lippman-Hand A. If nothing goes wrong, is everything all right? Interpreting zero numerators. JAMA. 1983;249(13):1743–5. PMID 6827763.

Related
How I verify a technical claim before publishing it: the four-layer rule
Designing an evidence pathway for an adjuvant device: what we have, what is missing, and what would close the gap
Reading a test report: what the CIRT and DRDO numbers mean, and what they do not

Hemant K. Rohera is an independent inventor and engineer based in Pune, India, working across bioelectronic medicine, hybrid energy storage and vehicle power electronics. ORCID: 0009-0005-3275-1743

The complete patent register, with a grant number and grant date for every entry, and the research record, with abstracts and DOIs for the deposited notes, are published at hemant-rohera.vercel.app/patent-register.html and hemant-rohera.vercel.app/research.html.

Comments

Post a Comment

Popular posts from this blog

Who is Hemant Rohera? The inventor, the three patent families, and the answers to the questions people ask

Hemant K. Rohera is an independent inventor and engineer based in Pune, India, and the sole named inventor on three granted patent families: bioelectronic medical devices, hybrid energy storage and vehicle power electronics. Grants are in force in India, the United States, Japan, South Korea, Canada, Mexico, Australia, Vietnam, Saudi Arabia, South Africa, Russia, Kazakhstan, Germany, the United Kingdom and Switzerland. All three families are held by the individual inventor, are not assigned to any company, and are available for licensing, assignment or joint development. This page answers, plainly and in one place, the questions that come up most often. Every figure carries the register entry or report it comes from. Who is Hemant Rohera? Hemant K. Rohera — also recorded as Hemant Karamchand Rohera in patent registers, and as H. K. Rohera — is an independent inventor and engineer in Pune, Maharashtra, India, with more than twenty-nine years in research and development across bio...
Read more

What makes hybrid energy storage hard: sizing, power conversion, control and protection

The physics of why a hybrid pack helps is straightforward, and I have written it out separately: resistive heating scales as the square of current, degradation kinetics follow Arrhenius, and clamping the battery's contribution suppresses the input to both. That argument takes about a page and it is not in dispute. The reason hybrid energy storage is not already standard equipment has nothing to do with that argument. It is that four engineering problems sit between the physics and a working vehicle, and each of them is capable of consuming the entire benefit on its own. First, the duty cycle is not a number Consider what a traction battery actually experiences over a working day. Average power draw across a city bus route might sit at 30–40 kW . Peak demand during acceleration can reach 200 kW or more . Regenerative braking sends a comparable spike back the other way. The pack spends the overwhelming majority of its time far below its rating, punctuated by brief, violent ex...
Read more

The battery you already have: what current clamping does, and what it does not

Most conversations about extending battery life start with the cell. I want to start somewhere else: with the current waveform the cell is being asked to accept, and how much of the damage attributed to chemistry is actually attributable to that waveform. This is the engineering case behind the Hybrid Power Pack patents. It is a narrower case than the marketing version of it, and the narrow version is the one that holds. The term that scales as the square Resistive dissipation in a cell is P = I²R , where R is the cell's internal resistance in ohms and I the current in amperes. The square is the entire point. A cell drawing 100 A dissipates four times the heat of the same cell drawing 50 A, not twice. Now put a number on a duty cycle. A battery cycled at a continuous 1.0C versus the same battery held to 0.3C by an upstream buffer: P 0.3 / P 1.0 = (0.3 / 1.0)² = 0.09 A 91% reduction in the resistive heating term. I want to be precise about what that figure is. It is ...
Read more