eduKate Learning Manual
Science | Veterinary World
Confirm the Tests Asked the Same Question → Check Specimen and Timing → Compare Methods → Revisit Pre-Test Probability → Identify What Each Test Can and Cannot See → Gather Separating Evidence → Update the Diagnostic Model

Veterinary Discordant Diagnostic Results

Why Two Good Tests Can Disagree Without Either One Being Useless

Wait, What? Two Tests Can Both Work Properly and Still Give Different Answers

A dog can have one infectious-disease test that is positive and another that is negative. A cat can have cytology that suggests one liver process while histology later reveals another. Two crossmatch methods can disagree on whether a donor-recipient pairing is compatible. One airway sample can show inflammation while another collected from a different airway region shows a different cellular pattern.

The instinct is to ask, “Which test is wrong?” Sometimes that is the correct question. Often it is too simple.

disagreement can come from different questions, different samples, different times, different thresholds or different parts of the same disease.

The Scientific Job

This manual owns one Veterinary World job:

How do veterinary teams reconcile apparently conflicting diagnostic results without automatically discarding one test or averaging them into false certainty?

Veterinary Diagnostic Tests owns general sensitivity, specificity and predictive value. Culture and Susceptibility, Serology, Cytology, Histology, Imaging and other manuals own their individual test domains. This page owns cross-test reconciliation when evidence does not line up cleanly.

Quick Answer

When diagnostic results disagree, a strong veterinary approach asks:

Did the tests examine the same biological target?
Were they performed on the same specimen type or anatomical site?
Were the samples collected at the same stage of disease?
Could treatment have changed one result before the other was obtained?
Do the methods have different sensitivity, specificity or detection limits?
Could preanalytical handling have affected either specimen?
Could one method detect exposure while another detects current organism or tissue change?
Does the animal’s clinical picture make one result more plausible?
What new evidence would best separate the competing explanations?

Primary Entry — Tests Do Not All Observe the Same Layer of Reality

Serology may detect an immune response. PCR may detect genetic material. Culture may detect viable organisms under conditions that allow growth. Cytology examines cells. Histology examines tissue architecture. Imaging shows structure. A functional test may show physiological consequence.

These methods can disagree because they are looking at different layers of the same biological process.

Part 1 — The First Question Is Whether the Tests Asked the Same Question

A positive antibody test and a negative PCR result do not necessarily contradict one another. Antibodies may persist after exposure while organism DNA is absent from the sampled blood at that moment. Likewise, a normal image does not necessarily contradict an abnormal biochemical marker if structural change has not yet developed or lies below the modality’s detection threshold.

Before choosing a “winner”, define the question each test actually answered.

Part 2 — Different Anatomical Sites Can Produce Different Truths

Disease is not always distributed evenly. Inflammatory cells can differ between central and smaller airways. A fine-needle aspirate can sample one part of a liver while infiltrative disease occupies another. A tumour can contain necrotic, inflammatory and malignant regions.

Two samples can therefore disagree because both are truthful descriptions of different places.

Part 3 — Time Changes What a Test Can See

Early infection may produce detectable organism before a mature antibody response. Later disease may reverse that pattern. Treatment can reduce microbial load while antibodies remain. Tissue injury may produce biomarkers before imaging changes become obvious.

A diagnostic result belongs to a time point. Comparing two tests without comparing their timing can manufacture a contradiction that biology does not actually contain.

Part 4 — Test Performance Is Probabilistic, Not Absolute

No diagnostic test is perfectly sensitive and perfectly specific. False positives and false negatives remain possible even when the test is used correctly.

Merck Veterinary Manual emphasises that predictive values depend on disease prevalence. A positive result for a rare disease in a low-risk patient can carry a different meaning from the same result in an animal with strongly compatible signs and exposure history.

Secondary Deepening — Discordance Is Information About the Model

When two good tests disagree, the diagnostic model should become more precise. The disagreement may reveal that the original hypothesis was too broad, that disease is heterogeneous, that timing matters, that one sample missed the lesion or that the tests measure different biological stages.

discordance is not merely a nuisance; it can expose which assumption was hidden.

Part 5 — Method Comparison Is Not the Same as Patient Diagnosis

Two laboratory methods can have only moderate agreement even when both are clinically useful. Agreement studies examine whether methods classify the same specimens similarly. Patient diagnosis asks whether the animal has the biological condition of interest.

A method disagreement should therefore not be mistaken for proof that one method is clinically worthless.

Part 6 — Preanalytical Error Must Be Reopened

If results conflict, specimen quality deserves another look. Was the sample collected from the intended site? Was it contaminated? Was there enough material? Was transport delayed? Did cells degrade? Did treatment begin before the second sample?

eduKate Veterinary World — Veterinary Preanalytical Error

Part 7 — Pre-Test Probability Helps Judge Which Result Is Surprising

If an animal has highly characteristic signs, exposure and examination findings, a negative result from a test with imperfect sensitivity may not collapse the diagnosis. If the disease is extremely unlikely before testing, a weak positive deserves careful confirmation.

This does not mean ignoring inconvenient results. It means interpreting them in the probability landscape that existed before the result arrived.

Part 8 — Repeat Testing Should Target the Source of Disagreement

Simply repeating every test can create more numbers without resolving the question. A better next step asks which uncertainty matters most.

If sampling error is plausible, obtain a better-targeted specimen.
If timing is the problem, consider whether later evidence should change.
If methods measure different targets, use a test that addresses the unresolved layer.
If the prior probability is unclear, gather history or examination evidence that narrows it.

The point is not “more testing”. It is more discriminating evidence.

JC Deepening — Discordance Can Be Modelled as Competing Explanations

Suppose Test A is positive and Test B is negative. Several models may fit:

the disease is present and Test B is falsely negative;
the disease is absent and Test A is falsely positive;
the disease was present earlier but not at the second time point;
the tests detect different biological states;
the sampled sites differ;
preanalytical error affected one specimen;
the disease is heterogeneous and both results are locally correct.

The best next observation is the one that separates these models most efficiently.

Part 9 — Agreement Statistics Do Not Replace Clinical Reasoning

Studies may report percent agreement, kappa, sensitivity, specificity and predictive values. These help us understand test behaviour in populations. They do not automatically tell us what happened in one individual animal.

The individual case still requires history, examination, sampling context and consequences of being wrong.

Part 10 — A Good Handoff Preserves the Disagreement

When a case moves to a specialist or reference laboratory, it is tempting to summarise aggressively: “Test A positive, so disease suspected.” But the disagreement itself may be diagnostically valuable.

A strong handoff includes the conflicting results, specimen types, timing, treatments and the question that remains unresolved.

How Do We Know?

Veterinary studies repeatedly demonstrate imperfect agreement between diagnostic methods. Airway brush cytology and bronchoalveolar lavage can identify different inflammatory patterns. Point-of-care and laboratory crossmatch methods can disagree in some donor-recipient pairings. Fine-needle cytology may miss infiltrative hepatic disease later identified histologically. Serology and PCR can produce different patterns because exposure, immune response and detectable organism are not identical biological targets.

Observation vs Inference

Observation: serology is positive and PCR is negative.
Inference: exposure with low or absent detectable circulating organism is possible; neither active disease nor absence of disease is established by the pair alone.
Observation: cytology suggests hepatic lipidosis but histology shows infiltrative disease.
Inference: the aspirate may have sampled a non-representative region; cytology was not necessarily analytically defective.
Observation: two crossmatch methods classify a pairing differently.
Inference: method-specific detection differences exist; the safest transfusion interpretation requires the responsible veterinary team.

Evidence Boundaries

discordant results ≠ one test definitely wrong.
positive result ≠ disease definitely present.
negative result ≠ disease definitely absent.
higher sensitivity ≠ better for every clinical question.
method agreement ≠ diagnostic truth.
repeat testing ≠ useful unless it targets uncertainty.
clinical fit ≠ permission to ignore contradictory evidence.
educational reconciliation ≠ individual diagnostic advice.

Common Misconceptions

Misconception	Better model
Two tests disagree, so one laboratory made a mistake.	Different targets, sites, timing, thresholds and test characteristics can create legitimate discordance.
The more advanced test automatically wins.	Usefulness depends on the biological question and validated performance.
Repeat both tests until they agree.	Choose the next evidence that best separates the competing explanations.
Clinical signs can override test results.	Clinical context changes probability but does not erase contradictory evidence.

Unfamiliar Transfer

Animal A has a positive antibody test but negative organism PCR. Animal B has an abnormal biomarker but normal imaging. Animal C has cytology and histology that disagree. Animal D has two test methods with moderate population agreement and conflicting individual results.

A strong learner asks what each method actually detected, when and where the sample came from, how likely disease was before testing, and what observation would most efficiently separate the remaining models.

Checkpoint Questions

Why can serology and PCR legitimately disagree?
How can anatomical sampling create discordant results?
Why does timing matter?
What does pre-test probability contribute?
Why is method agreement not the same as patient diagnosis?
When should preanalytical error be reconsidered?
Why can indiscriminate repeat testing be inefficient?
What information should accompany a specialist handoff?

Answer key

They may measure immune response and detectable organism rather than the same target.
Disease can be patchy, so samples from different regions can show different states.
Disease stage, immune response and treatment can change detectability over time.
It changes how surprising a positive or negative result should be.
Method agreement compares tests; patient diagnosis asks about the animal’s actual state.
Whenever sample handling, contamination, site selection or delay could plausibly explain the conflict.
More tests can create more noise unless the next test discriminates between competing explanations.
The conflicting results, specimen types, timing, treatments and unresolved question.

Edge Science — Can Multi-Modal Diagnostic Systems Learn From Disagreement?

Future veterinary decision systems may combine imaging, laboratory measurements, pathology, history and longitudinal data. The difficult cases will not be those where every signal agrees. They will be the cases where evidence conflicts.

A robust system should not hide disagreement by averaging it away. It should preserve which source said what, estimate why they may differ and ask for the next observation that best reduces uncertainty.

Veterinary World Direction Graph

Veterinary diagnostic discordance → define each test target → compare sample site → compare time → check preanalytical quality → review test performance → revisit pre-test probability → generate competing explanations → choose discriminating evidence → specialist handoff if needed → update the animal-level model.

Diagnostic Tests owns general test performance. Individual test manuals own their methods. This page owns reconciliation when credible evidence conflicts.

Research Sources and Further Reading

Educational safety boundary: Conflicting veterinary test results can have many explanations and may involve high-consequence decisions. This manual teaches diagnostic reasoning only; interpretation for an individual animal belongs to the responsible veterinarian and relevant specialist laboratory or clinician.

Teaching Guide for Parents, Tutors and Teachers

For the people who teach because somebody depends on them.

Use two observers looking through different windows at the same playground. One sees rain on the far side; the other sees dry pavement under shelter. Ask: “Do they disagree, or are they observing different parts of the same scene?”

define the question → define what each test sees → check place and time → preserve disagreement → seek separating evidence → return to the animal.

The mastery target is a learner who does not panic when evidence conflicts. Instead, the learner treats disagreement as a clue about hidden assumptions, sampling, timing and the structure of the diagnostic problem.

eduKate Singapore

eduKate Learning Manual: Veterinary Discordant Diagnostic Results | Why Two Good Tests Can Disagree Without Either One Being Useless