Practice worksheet
Reading an Acupressure Study: Outcomes, Adverse Events and Follow-Up
For each result, identify the measure, comparison, unit, time point and participants actually observed or analysed. Then check how adverse events were collected and how long people were followed. An improvement score, a missing safety section and a report of zero events answer different questions; none supplies personal clearance to use a treatment.
Name the outcome, unit and comparison first
“The study had good results” leaves out the result that matters. A pilot may recruit successfully while participants struggle to keep using the intervention. A symptom score may improve while the comparison group improves too. Read one outcome at one time rather than treating every favorable statement as the same finding.
| Field | Write down | Why the label matters |
|---|---|---|
| Measure | Exact name, instrument and whether it is primary, secondary or exploratory. | Recruitment, adherence and symptom benefit are different outcomes. |
| Unit and direction | Score units or participants meeting a defined criterion; whether higher or lower is better. | A falling score can mean improvement on one scale and worsening on another. |
| Comparison | Which groups and which reported analysis. | Change within one group does not show superiority over another. |
| Time point | Baseline, end of intervention and later follow-up, separately. | An end-of-treatment result cannot establish benefit years later. |
| Analysis population | Assigned, measured and analysed participants, by group and time. | Enrollment is not the denominator for every reported result. |
| Uncertainty and meaning | Effect estimate, its appropriate uncertainty and relevance to patients. | A small p-value does not by itself establish a useful improvement in daily life. |
Check the registered outcome and the final report’s description. If they use different labels, record that difference and consult the dated protocol or record history before drawing a conclusion about changes. A current registry entry alone cannot show what was specified in its earliest version.
Also identify the actual intervention: manual pressure, an auricular device, needles and electrical stimulation are different methods. Preserve the population and additional care. A fatigue trial in cancer survivors does not directly establish treatment for ordinary stress or every symptom promoted on a point page.
The sham-control guide addresses the separate task of understanding comparison procedures and masking. Here the task is reading the resulting measurements, safety report and follow-up.
Real example: keep the pilot’s purpose and counts visible
The Knoerl and colleagues pilot report studies self-administered relaxing acupressure in adolescent and young adult cancer survivors with fatigue. It enrolled 42 people, assigned 30 to relaxing acupressure and 12 to sham, and examined feasibility with preliminary symptom outcomes. The matched NCT06442891 results page lists recruitment rate and adherence as primary outcomes and fatigue change as secondary in the checked record.
The paper’s Table 4 gives these sample sizes for observed outcome data. They are the numbers to retain when reading the measurement times, alongside the analysis method.
| Measurement time | Relaxing acupressure | Sham comparison |
|---|---|---|
| Baseline | 30 participants | 12 participants |
| 6 weeks | 20 participants | 11 participants |
| 10 weeks | 20 participants | 9 participants |
Thus 31 people supplied six-week outcome data and 29 supplied ten-week data. Missing observations differ between groups. The Methods describe an intent-to-treat approach with a mixed-effects model using full information maximum likelihood. That model does not mean every enrolled participant supplied every questionnaire. Read its assumptions and any sensitivity analysis; do not silently treat absent scores as measured answers.
The paper reports fatigue reductions within both groups and no statistically significant between-group difference at six or ten weeks. It also reports that adherence fell below its prespecified feasibility benchmark. Recruitment, outcome completion, adherence and symptom advantage are separate findings. This pilot is not a definitive equivalence trial.
The registry’s participant-flow periods add another label check: “Intervention” lists 30 and 12 completed, while “Follow Up” lists 20 and 11 completed. Do not substitute the first completion count for the paper’s questionnaire count. Preserve the period names and measurement definitions; different reporting labels need clarification rather than an invented explanation.
Fictional calculation: percentage points and relative change
All numbers in this section are invented for arithmetic practice. They are not results from the pilot or any other acupressure study. Suppose two fictional packages are compared, and the report defines a criterion before analysis. Eighty participants in each group have a recorded result; 32 in A and 24 in B meet that criterion.
| Fictional group | Participants meeting the stated criterion | Proportion |
|---|---|---|
| Package A | 32 of 80 | 40% |
| Comparison B | 24 of 80 | 30% |
| Absolute contrast | 40% − 30% | 10 percentage points |
| Relative contrast | (40% − 30%) ÷ 30% | About 33.3% more meeting the criterion |
The absolute contrast is ten additional people meeting the criterion per 100 people, as a description of these fictional proportions. The relative contrast uses B as the reference: ten divided by thirty is about one third. A headline saying “33% more” hides the baseline unless it also gives 30% and 40%.
Neither number supplies a confidence interval, a p-value, a causal conclusion or a clinically meaningful threshold. You would still need the design, measurement definition, analysis and uncertainty. Do not calculate a personal chance of benefit from this exercise.
If some participants lack follow-up, write the observed numerator and denominator and examine the stated analysis. Do not automatically count every missing participant as improved or not improved. That is an assumption requiring a suitable method or sensitivity analysis, not a harmless formatting choice.
Read statistical uncertainty and practical importance separately
Look for the reported between-group estimate and its uncertainty at the relevant time. A confidence interval around group A’s mean is not the confidence interval for the A-versus-B difference. Use the actual contrast from the analysis rather than subtracting two interval endpoints.
A statistically significant difference can be too small to matter in daily life. Conversely, an imprecise estimate can leave important possibilities unresolved. NCCIH’s journal-reading guidance distinguishes statistical and clinical significance and explains why a single study rarely provides a final answer.
For a symptom scale, check which direction means improvement and whether an appropriate meaningful-change threshold is justified for that population and instrument. A threshold from another scale cannot be imported by matching the number. A mean change also does not show that every participant improved by that amount.
Record function, acceptability and burden where measured, alongside symptoms. Completing a technique accurately, enjoying an app and receiving a symptom benefit are not interchangeable. Avoid making a finding stronger just because one of the measures sounds encouraging.
Read adverse events as a separate result
An adverse-event report concerns unwanted experiences or medical occurrences in the study. Whether an event is related to the intervention is a separate assessment. Check whether the report counts people with at least one event, total events or both; one person can experience more than one event.
| Report wording | What to check | Conclusion to avoid |
|---|---|---|
| No adverse events reported | Who was monitored, how, for how long and with what completeness. | “There can never be harm.” |
| No serious adverse events | The definition and whether non-serious events were also collected. | “Nobody had any unwanted experience.” |
| Adverse events not described | Methods, results, supplements and the matched registry. | “The event count was zero.” |
| Events occurred during the study | Event type, affected people, event count and assessed relatedness. | “Every event was caused by the intervention.” |
| Only frequent events listed | The reporting threshold and separately reported serious events. | “Unlisted events did not occur.” |
“Serious” has a defined meaning, rather than simply being a stronger adjective for pain. The FDA’s serious-adverse-event guidance, in its medical-product reporting context, includes outcomes such as death, life-threatening events, hospitalization, disability and other medically important events. Read the study’s applicable definition and its non-serious reporting too. This guide does not classify a reader’s symptoms.
The pilot Methods say events were monitored throughout the study and reviewed for severity, relatedness and need for follow-up. Its Results state that no adverse events were reported. The checked registry separately shows zero affected among 30 and 12 at risk for all-cause mortality, serious events and other non-serious events. It lists a 0% threshold for reporting other events, while the separate adverse-event reporting-description field is not specified.
Those are the report’s data and limits, not proof that every possible event was detected. A small study with zero reported events cannot rule out uncommon or delayed harms, nor establish safety for pregnancy, children, different health conditions or another pressure method.
Follow-up is a measurement, not just a date
Distinguish how long the whole study was active from how long each person was observed. The checked registry’s adverse-event time frame says the study was active for 14 months and participants were assessed for an average of 40 days each. That does not give everyone 14 months of observation. An average also does not establish identical follow-up for all participants.
The paper’s symptom assessments were at baseline, six weeks and ten weeks. Safety and symptom fields must each retain their own time frame. Do not replace a safety observation window with the latest symptom visit simply because both appear in the same record.
A planned follow-up is not a completed measurement. Record the number actually observed at each visit, known reasons for missing data and the analysis used. In the registry’s named Follow Up period, non-completion reasons include lack of time and loss to follow-up. The labels do not justify assuming that everyone missing got better, got worse or withdrew because of harm.
Keep the conclusion within the measured interval. If longer-term benefit, delayed adverse events or outcomes after stopping were not measured or reported, write that uncertainty. A six-week program with a ten-week visit does not demonstrate permanent benefit.
Six outcome and safety reading checks
Try the original checks below before opening their explanations. They test reporting distinctions and the labelled fictional arithmetic, not clinical decisions.
Case 1
In the fictional table, 40% of A and 30% of B meet the criterion. Is the absolute difference “33.3 percentage points”?
Show answer and explanation for case 1
No. The absolute difference is 10 percentage points.
About 33.3% is the relative increase compared with B’s 30%. Keep the base proportion visible; neither calculation establishes statistical uncertainty or clinical importance.
Case 2
A paper enrolled 42 participants, but only 31 supplied a six-week questionnaire. It used an intent-to-treat model. Were 42 actual six-week responses collected?
Show answer and explanation for case 2
No. A model and an observed response count are different things.
Record assigned and observed participants separately, then read how the analysis handled missing data. A method label does not create the absent questionnaires.
Case 3
A registry says the study was active for 14 months and participants were assessed for an average of 40 days. Does that give every participant 14 months of safety follow-up?
Show answer and explanation for case 3
No. Calendar duration and each person’s observation period are different.
Preserve both labels. Do not extend an average observation time into identical follow-up for everyone, or assume the safety and symptom windows match.
Case 4
The article contains no adverse-event section. Can your summary say “zero adverse events”?
Show answer and explanation for case 4
No. Write that adverse-event reporting was not found in the material checked.
Check the methods, supplements and matched registry. Missing information is not a measured zero, and it does not establish that nobody was monitored.
Case 5
A small trial reports no serious adverse events. Does this alone show that no participant had soreness or another non-serious event?
Show answer and explanation for case 5
No. Read the non-serious-event reporting separately.
“Serious” concerns defined outcomes such as hospitalization or life-threatening events; it is not a synonym for every unwanted symptom or a complete measure of symptom intensity.
Case 6
A symptom estimate is shown at six weeks. Can you describe its benefit as permanent if no later measurements are reported?
Show answer and explanation for case 6
No. Limit the claim to the measured time and analysis.
A later visit must actually supply relevant results. The presence of a scheduled follow-up does not mean everyone attended or that a durable effect was demonstrated.
Save one conclusion with its limits
Download the editable outcome, harms and follow-up reading record. It keeps assigned, observed and analysed participants separate and leaves room for missing reporting. The form is an original aid without a validated score.
Population and actual comparison: Outcome / instrument / unit / direction: Time point and observed participants: Between-group estimate and uncertainty: Adverse events: collection, counts, relatedness, observation window: Missing data and longer-term questions: The narrow conclusion this report supports:
For this pilot, a bounded summary retains the cancer-survivor population, feasibility purpose, variable adherence, incomplete outcome observations, reported lack of a statistically significant fatigue advantage over sham and limited observation windows. It also attributes the zero adverse-event report rather than converting it into personal safety clearance.
If the task is finding a record or interpreting its recruitment status, use the registered-trial guide. If a result might affect your own care, discuss the actual evidence with a qualified clinician. This article supplies no point locations, pressure schedules or instructions to change treatment.
Sources and further reading
- Knoerl and colleagues: Determining the feasibility of self-administered relaxing acupressure for fatigue among adolescent and young adult cancer survivorsRead abstract, Introduction, all Methods, Results including outcome-table notes, Discussion/limitations and Conclusions. Pilot purpose, observed samples, missing-data method, adherence, event monitoring and reported zero adverse events; no independent efficacy reanalysis.
- ClinicalTrials.gov: NCT06442891 official record and result data (JSON)Checked primary/secondary labels, named participant-flow periods, event counts, reporting threshold and safety observation time frame against the displayed Results Posted page. Machine-readable primary record; public reading page is linked in the example.
- FDA: What is a Serious Adverse Event?Complete medical-product reporting guidance read. Serious outcome categories are distinct from a general unwanted-symptom count; not a classification of an individual reader’s symptoms.
- NCCIH: How To Make Sense of a Scientific Journal Article — ResultsComplete results-reading guidance read for statistical versus clinical significance, replication and conflicts. No claim that a p-value is a personal probability of benefit.