Anatomy Question Quality Review

Status Note

This is a pilot narrative report, not a full re-review of the newer 100-question randomized subject sample now shown on the site. The qualitative conclusions here came from an earlier smaller manual packet review and should be read as an issue taxonomy prototype for Anatomy, not as the final word on the refreshed full-scope packet.

Scope

• Course slice: Indian Medical PG -> Anatomy
• Gold reference used for calibration: INI-CET April 2026 Benchmark Test plus recent 5-year PYQ packet
• Candidate pool reviewed: validated non-PYQ, non-benchmark questions only
• Pilot sample reviewed in this report: 8 benchmark, 12 PYQ, 16 generic, 16 risky

What good looks like in this subject

The better Anatomy questions in this packet are not necessarily difficult, but they do one useful thing well: they anchor an anatomical concept to a concrete clinical localization problem. Representative examples include 02a9b480, a5c9a1c7, 1e6ce134, e800b52c, and cf41a988. These are still solvable from core anatomy, but they at least ask the learner to use anatomy rather than simply recite it.

Core problem

The weak side of the Anatomy bank is dominated by low-yield factual recall and topic drift. Many questions are factually correct, but they feel like isolated viva facts, comparative-anatomy trivia, or image-spotter leftovers rather than questions worth preserving in a serious PG exam bank. The bank also contains multiple questions where a short pseudo-clinical wrapper has been added around a fundamentally direct recall task.

Main issue categories

1. Comparative-anatomy and evolutionary trivia that should not occupy exam slots

This is the clearest pathology in the Anatomy sample. Questions such as d5d1ae0b ("human appendix is homologous to which structure in herbivorous mammals?") and c82c3ac6 ("which embryonic structure in humans is homologous to the cloaca...") are factually valid, but they are poor fits for the target exam standard. They do not reflect the way good Anatomy questions create value for the learner.

Why this is bad:

• They test curiosity-level facts rather than clinically useful anatomy.
• They consume bank space without improving conceptual coverage.
• They make the subject feel random and low-yield.

Suggested disposition:

• disable unless there is a very explicit reason to preserve an evolutionary-anatomy micro-topic.

2. One-step direct recall with no reasoning burden

Questions like 5a1b21a2 ("Which artery supplies the deep cerebellar nuclei?"), c9f25e39 ("Which paranasal sinuses are present at birth?"), 2abe4520 ("Inversion and eversion of the foot occur at which joint(s)?"), and 8d2b2edc ("Which one of the following structures is a part of the diencephalon?") are too simple for the intended benchmark.

Why this is bad:

• The candidate either remembers the fact or does not.
• Distractor elimination does most of the work.
• These items do not discriminate well between prepared and truly strong students.

Suggested disposition:

• disable for the simplest examples.
• fix only when the underlying concept is important and can be converted into a better localization or injury-based question.

3. Exclusion-format anatomy questions that reward partial recall

Questions like 72512aa1 ("All the following structures are found in the lateral nasal wall except"), 34e1df46 ("In ulnar nerve injury in the arm, all of the following are seen except"), and 4f3fcd5a ("Which structure is NOT found in the deep perineal pouch in males?") rely on negative framing rather than positive reasoning.

Why this is bad:

• These are often solved by spotting the odd option rather than understanding the anatomical setup.
• They feel exam-like without actually being high-quality.
• They become especially weak when the distractors are uneven.

Suggested disposition:

• fix if the concept matters.
• disable when the same concept already exists in a better positive-format question.

4. Subject bleed and misclassified content

The packet contains questions that do not feel like real Anatomy bank priorities. b525bedd asks for the normal AP length of the eyeball, which feels more like ophthalmology spotter material than durable Anatomy value. fb3fbb97 asks about fructose transport by GLUT, which is not even an Anatomy question in substance. This kind of bleed is one reason the bank feels noisy.

Why this is bad:

• It weakens subject identity.
• It makes clean curation harder because the content owner is unclear.
• Learners experience these items as arbitrary.

Suggested disposition:

• disable or reassign only if there is a strong subject owner and use case.

5. Fake-clinical wrapping over a direct textbook fact

Some benchmark examples show the style the generic bank is trying to imitate: a short trauma or surgical vignette followed by a structure-identification problem. That format can work, but low-quality items often only paste a vignette onto a fact that remains one-step. The better benchmark examples are acceptable because the localization is still clinically coherent; the weak bank copies the surface form without the reasoning depth.

Why this matters:

• The bank risks producing questions that look premium but remain shallow.
• This is not as bad as the trivia bucket above, but it is still below the quality bar if the vignette adds no reasoning.

Suggested disposition:

• fix when the clinical frame can be made to genuinely matter.

Patterns worth telling the content team

• The single biggest Anatomy issue is not "wrong facts." It is low-yield concept selection.
• The second biggest issue is recall-only items disguised as a complete question bank.
• Comparative-anatomy trivia and exclusion-format anatomy are the clearest candidates for aggressive cleanup.

Example keep / fix / disable calls

• keep: 02a9b480, a5c9a1c7, 1e6ce134, cf41a988
• fix: 34e1df46, 72512aa1
• disable: d5d1ae0b, c82c3ac6, 5a1b21a2, c9f25e39, fb3fbb97, b525bedd

Bottom line

Anatomy looks weak not because the facts are wrong, but because too many questions are asking the wrong kind of thing. The bank needs fewer comparative and spotter-style leftovers, fewer exclusion stems, and many fewer one-step fact tests. The quality bar should move toward clinically anchored structural reasoning and away from isolated anatomy trivia.