Start by preserving the observation
When someone says, “I noticed this after that,” the first claim is about their experience. Symptoms, daily functioning, sleep quality, discomfort, motivation, and other person-reported outcomes can contain information that no laboratory value fully replaces.
Respecting that observation does not require accepting the first causal explanation attached to it. Relay can hold both ideas at once: the person may be accurately describing what changed, and the available information may still be unable to show why it changed.
After does not automatically mean because of
Timing is useful. If a change begins after an exposure, that sequence makes a causal explanation possible. But several other things may change around the same time: sleep, diet, stress, illness, expectations, measurement habits, other substances, or the natural course of a symptom.
A confounder is another factor connected with both the exposure and the observed outcome. Chance, measurement variation, and a return toward a person's usual state after an unusually good or bad period can also produce convincing before-and-after stories.
The more specific the record, the more useful the signal
A detailed record can preserve when an observation began, how it changed, what was measured, what else was happening, and whether the pattern repeated. That makes the experience easier to interpret and may reveal questions worth testing.
Detail improves the description; it does not automatically remove alternative explanations. A precise timeline can show that two events occurred together without proving that one produced the other.
Objective measurements strengthen the observation—not always the cause
A scale, wearable, laboratory value, or performance measure can confirm that something changed beyond memory alone. Repeated measurement can also show whether the change was brief, sustained, or part of an existing pattern.
The measurement still does not isolate the cause if other conditions changed at the same time. Better measurement narrows uncertainty about what happened. A comparison design is what helps narrow uncertainty about why.
Several similar stories create a signal, not a rate
When multiple people independently report a similar event, the pattern may deserve attention. Safety systems use spontaneous reports in exactly this way: to identify signals that might warrant investigation.
The number of stories alone cannot show how common the event is. We may not know how many people had the exposure, how many experienced nothing, who chose to report, whether reports overlap, or how completely other explanations were recorded. Without that denominator and a suitable comparison, frequency and risk remain uncertain.
Personal experiences can improve the questions research asks
The FDA defines patient experience data broadly: symptoms, functioning, quality of life, treatment experiences, preferences, and the outcomes people consider important. These observations can shape what researchers measure and which burdens deserve attention.
An experience can therefore be scientifically valuable before it establishes causation. It may identify a missing outcome, an unusual timing pattern, a subgroup question, a possible adverse event, or a result that should be measured more carefully in future research.
Stronger individual evidence requires structure
An N-of-1 trial is a planned comparison conducted within one person, often using repeated treatment periods, predefined outcomes, and design features intended to reduce bias. It asks a different question from an anecdote: whether outcomes reliably differ across controlled conditions for that individual.
These designs are only appropriate for certain questions and settings. They may be impractical or unsafe when effects are irreversible, conditions change quickly, washout is impossible, or withholding an intervention creates risk. Even a rigorous individual result does not automatically generalize to other people.
Key terms
- Personal experience data
- Information about a person's symptoms, functioning, priorities, treatment experiences, preferences, or outcomes that matter to them.
- Temporal association
- A pattern in which one event occurs before or around the time of another; timing is compatible with causation but does not establish it alone.
- Confounder
- Another factor related to both an exposure and an outcome that can distort the apparent relationship between them.
- Reporting bias
- A distortion that occurs when some experiences are more likely to be shared, recorded, or published than others.
- Denominator
- The total number of relevant people or exposures needed to interpret how frequently an event occurred.
- Signal
- A pattern suggesting that a possible relationship deserves closer investigation, without yet establishing cause or frequency.
- N-of-1 trial
- A planned, repeated comparison within one person designed to estimate how outcomes differ under controlled conditions for that individual.
Relay diagram
From experience to a supported conclusion
What this can—and cannot—tell us
What it can tell us
- What one person noticed, measured, valued, or found concerning.
- The timing and context of an observed change.
- Which outcomes may matter enough to measure in future research.
- Whether an unexpected possibility or recurring pattern deserves investigation.
- How an experience affected daily life in ways standardized measures may miss.
- Which details, alternative explanations, and unanswered questions should be preserved.
What it cannot establish alone
- That the exposure caused the observed change.
- How much of the change was due to expectation, natural variation, another exposure, or chance.
- How frequently the same event occurs without knowing the relevant denominator.
- Whether people who did not report had different experiences.
- That the same result will occur in another person, setting, exposure, or timeframe.
- That a community pattern establishes safety, effectiveness, or comparative superiority.
Go deeperOptional · about 2 minutes
The experience and its explanation are separate claims
“I experienced nausea” is a report about an outcome. “This compound caused my nausea” is a causal interpretation. Evidence can support the first claim more directly than the second.
Separating them protects the person's observation instead of dismissing it. It also keeps the explanation open to competing possibilities until stronger evidence distinguishes among them.
Why a hundred reports still may not reveal frequency
A count becomes a rate only when the relevant total is known. One hundred reports among one thousand exposed people would mean something different from one hundred reports among one million.
Voluntary reporting adds another uncertainty: people who notice an event may be more likely to report than those who do not. Duplicate, incomplete, stimulated, or publicity-driven reports can further change the apparent pattern.
What structure adds within one person
Predefined outcomes reduce the temptation to focus only on favorable changes. Repeated baseline measurements reveal ordinary variation. Comparison periods, randomization, and blinding can reduce some expectation and time-order effects when the question permits them.
Those features can strengthen an individual causal inference. They still do not answer whether the same effect occurs across a population, and they do not remove ethical or practical limits.
The takeaway
If you only remember one thing from this guide:
A personal experience can reveal a real observation and a valuable signal. It cannot establish cause and effect by itself.
Sources and support7 sources
- Patient-Focused Drug Development Glossary U.S. Food and Drug Administration
Defines patient experience data to include symptoms, functioning, quality of life, treatment experiences, preferences, needs, and patient-prioritized outcomes.
- Patient-Focused Drug Development: Collecting Comprehensive and Representative Input U.S. Food and Drug Administration
Explains how the research question, target population, sampling strategy, and collection method shape the usefulness and representativeness of patient input.
- FDA Adverse Event Monitoring System public dashboard U.S. Food and Drug Administration
States that spontaneous reports reflect observations and opinions, do not establish causation, and cannot alone determine event incidence.
- Data Mining at FDA — White Paper U.S. Food and Drug Administration
Explains that reporting associations can identify possible safety signals but cannot be interpreted as causal or as estimates of risk.
- Analyzing and Interpreting Data — Field Epidemiology Manual Centers for Disease Control and Prevention
Explains that an observed association may reflect causation, chance, selection bias, information bias, confounding, or other error.
- Design and Implementation of N-of-1 Trials: A User's Guide Agency for Healthcare Research and Quality
Describes the design, appropriate uses, limitations, analysis, and implementation of structured within-person trials.
Provides reporting standards for prospectively planned, multiple-crossover trials conducted within individuals.