One Patient, Many Samples: What Is the Denominator in a Diagnostic Study?
October 19, 2026 · 5 min read · Aqrab editorial team
A diagnostic study reports results from 300 samples. The patients who supplied them number 100. The headline may be accurate about samples, yet misleading if readers apply it to a decision made once per patient. Before interpreting the percentage, find out which unit was counted and which clinical decision the test is meant to support.
Three Denominators, Three Questions
Suppose 100 patients each contribute three specimens in a hypothetical study. The numbers below describe the design, not any real test performance.
| Unit | Count | Question it can answer |
|---|---|---|
| Patients | 100 | How often is a patient classified correctly under a specified rule? |
| Specimens | 300 | How often is an individual specimen classified correctly? |
| Testing episodes | Depends on protocol | How often does the actual sequence of testing classify an episode correctly? |
A patient-level result requires a prespecified rule: first specimen only, any positive among three, a majority result, or another clinically intended algorithm. These rules can produce different results from the same specimens. Do not silently relabel a specimen-level proportion as patient-level sensitivity or specificity.
The practical takeaway
Ask for patients, specimens per patient, and the decision rule before trusting a diagnostic accuracy headline. If the clinical claim is patient-level, request a patient-level estimate and uncertainty interval.
Why More Rows Are Not More Independent Patients
Specimens from one patient can share biology, collection conditions, and reference-status errors. Treating every specimen as an independent person can make an uncertainty interval too narrow. The amount of distortion depends on the within-patient dependence and study design; 300 specimens from 100 patients do not simply equal 300 independent patients, nor is the effective sample size automatically exactly 100.
Patients who supply more specimens can also receive more weight in a pooled specimen-level percentage. If sicker patients are sampled more often, that percentage may describe the sampling process as much as the intended-use population. A valid specimen-level question is possible, but it must be named and analyzed as such. For patient-level use, aggregate according to the intended testing algorithm and account for clustering when estimating uncertainty or comparing tests.
A Four-Question Review
- Unit: How many unique patients, testing episodes, and specimens contributed to each accuracy calculation?
- Rule: If several specimens are available, what result triggers the clinical action? Was that rule specified before examining outcomes?
- Reference: Is condition status defined for the patient, the episode, or each specimen, and is the index test kept out of that definition?
- Uncertainty: Do intervals and test comparisons respect repeated observations within a patient?
Some assays genuinely target a specimen-level property. The point is not to ban specimen-level analyses; it is to keep their conclusions aligned with the unit actually studied and the decision that will be made.
Where Aqrab Fits
When using Aqrab to critique a diagnostic paper, include the participant flow, specimen counts, reference-standard definition, and testing algorithm. A methods review can flag a unit-of-analysis mismatch and missing clustering details. It cannot infer patient-level accuracy from an unlabeled specimen table.
Source Anchors
The FDA diagnostic-test reporting guidance explicitly notes that its standard calculations assume no multiple samples from one patient. In FDA guidance for one respiratory diagnostic device category, sample-size planning is based on subjects rather than specimens and a combined patient-level rule is specified. That device-specific rule is an example, not a universal rule for all tests. STARD 2015 provides the broader diagnostic accuracy reporting framework.
Keep reading
Don't stop at one method.
Good methods judgment comes from contrast. Read the neighboring guides, see where the assumptions diverge, and avoid treating every observational problem like it needs the same hammer.
Only the Disagreements Were Checked: The Discrepant-Resolution Trap
When a third test checks only disagreements, agreement is not proof that both tests were right. Use this three-question audit before accepting a revised diagnostic-accuracy table.
The Result Arrived After the Decision: Audit Diagnostic Turnaround
A practical timeline shows why diagnostic accuracy alone cannot establish that a result can reach the clinical decision it is meant to support.
One Pooled Sensitivity, Two Hospitals: Check Diagnostic Accuracy by Site
A worked two-site example shows how a pooled sensitivity can hide a weaker result in one setting, and what to inspect before transporting a diagnostic claim.