Microbiology Question Quality Review
Executive Summary
This review covers 200 validated non-gold candidate questions sampled from the Microbiology pool of 11,104 items. The sample was analyzed across eight shards of 25 questions each. The findings are consistent and unambiguous: the Microbiology candidate pool has a severe structural quality deficit concentrated in a small number of recurring problem types, with a secondary but serious layer of factual accuracy errors that create active harm risk.
The Bloom's distribution tells the story plainly. Of the 200 candidate questions, 89 are Bloom's Level 1 and 87 are Bloom's Level 2 — together accounting for 88% of the sample. Only 19 are Level 3 and 5 are Level 4. The benchmark set, by contrast, is entirely Bloom's Level 3–4, built around clinical vignettes requiring multi-step reasoning. The gap is not marginal; it is structural.
The dominant problem is not that individual questions are wrong on isolated facts (though some are). The dominant problem is that the overwhelming majority of questions in this pool are formatted as single-sentence recall prompts — "Which organism causes X?", "What is the generation time of Y?", "Which is NOT true about Z?" — with no clinical context, no reasoning demand, and no discriminatory value at PG entrance level. These questions would be appropriate for a first-year undergraduate quiz. They are not appropriate for INICET or NEET-PG preparation.
Layered on top of this structural problem are five operationally distinct issue categories, each requiring a different remediation path. Approximately 30–40 questions in the sample carry confirmed or highly probable factual errors in their answer keys. A discrete cluster of questions is non-functional because they reference images that are absent from the text. A recurring pattern of "except/NOT/all of the above" questions uses negative framing as a substitute for clinical reasoning rather than as a complement to it. Topic misclassification is scattered but consistent enough to affect test blueprint accuracy. And a small but valuable subset of questions has genuine clinical framing that falls just short of benchmark quality and is worth upgrading rather than discarding.
Summary disposition across the 200-question sample (estimated from shard evidence):
- Keep as-is: ~25–30 questions (12–15%)
- Fix (rewrite stem, correct key, upgrade vignette, fix tag): ~55–65 questions (27–32%)
- Disable: ~110–120 questions (55–60%)
The disable rate is high but justified. The benchmark standard is clear, the PYQ gold set confirms what the exam actually tests, and speculative rewrites of trivial recall items are not a productive use of content operations time when the underlying concept is already covered by stronger items elsewhere in the pool.
What Good Looks Like
The benchmark and recent PYQ sets establish a clear and consistent quality bar. Every benchmark question shares the following structural features:
1. A clinical vignette that does real work. The patient's age, occupation, geography, exposure history, symptoms, examination findings, and laboratory results are not decorative — they are the reasoning inputs. In 32ca97a3 (ESBL-producing Klebsiella), the candidate must integrate post-surgical timing, Gram stain morphology, EMB agar appearance, cephalosporin resistance pattern, and clavulanic acid inhibition to reach the correct answer. No single clue is sufficient. In 38af2b72 (Entamoeba histolytica), the combination of rural Assam, farmer occupation, hepatomegaly, cavitary lung lesion, and liver hypoechoic lesion is required to distinguish the correct answer from Paragonimus, Echinococcus, and TB — all of which are plausible in isolation.
2. Distractors that are genuinely plausible. In 6e7f5315 (mecA gene, MRSA), the four options represent four distinct and real resistance mechanisms — altered PBP, beta-lactamase, efflux pump, and porin mutation. A candidate who does not understand the specific mechanism of mecA-mediated resistance cannot eliminate the distractors by common sense. In 5927c4ff (biofilm in prosthetic valve endocarditis), the distractors include horizontal gene transfer, inadequate penetration due to calcification, and intracellular reservoir — all of which are real phenomena that a partially-prepared candidate might select.
3. A single unambiguous correct answer. The benchmark questions do not have double-true problems, compound answers, or contestable keys. The correct answer is defensible against any standard reference.
4. Bloom's Level 3 minimum. Even the "easier" benchmark questions (bbc7b25a on Streptococcus viridans endocarditis, 77a68f11 on Blastomyces dimorphism) require the candidate to apply knowledge to a new situation, not merely retrieve a stored fact.
5. Clinical and epidemiological specificity. The benchmark questions use specific geographic anchors (Mississippi Valley, Kerala, rural Assam), specific occupational exposures (farmer, dental extraction), and specific timelines (3 weeks post-procedure, day 60 post-BMT) that are not decorative — they are part of the reasoning chain.
The PYQ set (1bda6eae, b4be7639, 19a57714) confirms that even Bloom's 1–2 items that appear in actual exams are anchored in a clinical or laboratory scenario. The question b4be7639 (Cyclospora oocyst size on ZN stain) is Bloom's 1 in isolation but is presented as a diagnostic identification task. The question 1bda6eae (Neisseria meningitidis in CSF) requires the candidate to select a multi-property biochemical profile, not a single fact.
The candidate pool, by contrast, is dominated by questions that ask "Which virus causes chickenpox?" or "What is the generation time of E. coli?" These are not questions that appear in INICET or NEET-PG. They are not questions that prepare candidates for INICET or NEET-PG. They should not be in this pool.
Main Issue Categories
1. Structural Recall Overload: Single-Fact Questions With No Clinical Anchor
Why this pattern is bad
This is the primary and most pervasive quality problem in the Microbiology candidate pool. A question that asks "Chickenpox is caused by which virus?" or "What is the generation time of E. coli?" or "Which cell produces IL-1?" is not a PG entrance question. It is a flashcard. It tests whether a candidate has memorized a single association, not whether they can reason about a clinical problem. The benchmark standard is unambiguous: even straightforward concepts must be embedded in a clinical scenario that requires the candidate to apply knowledge, not retrieve it. Questions at Bloom's Level 1 with no vignette have zero discriminatory value at PG level — a candidate who has done one week of revision will answer them correctly, and a candidate who has done no revision will guess correctly 25% of the time.
The harm is not just that these questions are easy. The harm is that they crowd out higher-quality items, inflate the apparent size of the question bank without adding assessment value, and — if used in practice tests — give candidates a false sense of preparation by rewarding memorization over reasoning.
How it shows up
This pattern appears in every shard and across every sub-topic: Bacteriology, Virology, Mycology, Parasitology, Immunology, and General Microbiology. It is not a cluster — it is the dominant mode of the pool. Approximately 55–60% of the 200-question sample consists of questions that are structurally pure recall with no clinical context. The stems are typically one sentence. The options are typically four named entities or four true/false statements about a single organism. There is no patient, no exposure, no lab result, no clinical decision.
Example question IDs
3f0ad811: "Chickenpox is caused by which virus?" — definitional tautology, Bloom's 1, no distractors of any discriminatory value. The answer is in the question name.de8a5d44: "Generation time for E. coli?" — isolated numerical fact, distractors (2 minutes, 2 hours, 20 days) are implausible to any medical student.577ebf4d: "IL-2 is secreted by Helper T-cells" — single cytokine-cell association, Bloom's 1 in practice despite being tagged Bloom's 2.de7cd55e: "Influenza virus belongs to which family?" — taxonomy recall, no clinical relevance.f80d8661: "What does 'pleomorphic' mean in bacteria?" — vocabulary definition, not a microbiology application question.44ab6d59: "Pontiac fever is caused by?" — single-line, no stem, Bloom's 1.9c931fc0: "Tetanospasmin encoding genes are located on plasmid" — bare-fact genetics recall.a03c8640: "Monoclonal antibodies bind to a specific epitope" — definitional, Bloom's 1.Q97dc76b0: "Which immunoglobulin has the highest serum concentration?" — universally known at undergraduate level.Qa6e6c7e1: "DHR is mediated by T lymphocytes" — single-fact recall, no vignette.Q792afb07: "Mite-transmitted disease = scrub typhus" — single-association recall.Q1d609b5d: "Genital warts caused by HPV" — pre-clinical knowledge.400630b7: "Human papillomatosis is caused by HPV?" — tautological.e46fec1f: "An example of naturally acquired passive immunity is placental transfer" — Bloom's 1, no clinical context.12b72d13: "Which cell type produces IL-1?" — textbook flashcard.fe14d494: "AFB negative slide field count" — numerical recall.2897447d: "Aflatoxin produced by Aspergillus flavus" — pure recall, no clinical hook.54ea6dc5: "Kidney-shaped organism = Gonococci" — trivial morphology recall.3c54424a: "Louse-borne relapsing fever agent" — name recall.c03edf76: "Prokaryotes lack mitochondria" — pre-medical level, appears in GNM/ANM templates.cf2ad745: "Scarlet fever caused by Streptococci" — distractors (Proteus, Klebsiella) are implausible to any medical student.
Recommended disposition
Disable for all questions that are pure Bloom's 1 recall with no clinical context and where the underlying concept is already covered by stronger items in the pool or in the PYQ/benchmark set. This applies to the majority of examples listed above. For a small subset where the concept is genuinely high-yield and no stronger item exists (e.g., specific diagnostic criteria, organism-specific virulence factors), consider a fix by adding a 2–3 sentence clinical vignette to reach Bloom's 3 — but only if the concept warrants the investment. Do not rewrite trivial facts (generation times, family names, vocabulary definitions) — disable them.
2. Confirmed and Probable Factual Errors in Answer Keys
Why this pattern is bad
A question with a wrong answer key is not merely a low-quality question — it is an actively harmful one. If used in a practice test or study session, it teaches candidates incorrect information. In a subject like Microbiology, where precise factual knowledge is tested, a wrong key on a taxonomy, serology, or mechanism question can directly cause a candidate to answer a real exam question incorrectly. This is the highest-priority remediation category in the pool.
The errors observed in this sample are not borderline or reference-dependent. Several are clear-cut factual mistakes: marking HAV as belonging to Hepadnaviridae (it belongs to Picornaviridae), marking HIV as not being an RNA virus, marking lymphocytes as an exception to MHC Class I expression (red blood cells are the classic exception), and marking Gram staining as the test that differentiates Streptococcus from Pneumococcus (both are Gram-positive cocci; the differentiating tests are bile solubility and optochin sensitivity). These errors suggest that some questions in this pool were generated or edited without adequate subject-matter expert review.
How it shows up
Factual errors appear across Bacteriology, Virology, and Immunology sub-topics. They are not confined to a single topic or shard. In this sample, at least 8–10 questions carry confirmed or highly probable answer key errors. Given that this is a random sample of 200 from a pool of over 11,000, the absolute number of erroneous questions in the full pool is likely substantial.
Example question IDs
dac5b285: Marks "Hepadnaviridae" as the family of Hepatitis A virus. HAV belongs to Picornaviridae (genus Hepatovirus). Hepadnaviridae is HBV. This is an unambiguous factual error.e3f4c9f5: Marks "Gram staining" as the test differentiating Streptococcus from Pneumococcus. Both are Gram-positive cocci; Gram staining cannot differentiate them. The correct differentiating tests are bile solubility, optochin sensitivity, and inulin fermentation.4807126e: Marks "It is an RNA virus" as incorrect for HIV. HIV is definitively an RNA retrovirus. This is a factual error that would actively misinform candidates.c5621bf6: States lymphocytes are an exception to MHC Class I expression. This is factually incorrect — lymphocytes express MHC Class I. Red blood cells are the classic exception.8e26e4dd: Marks IgG as the "most avidly complement-fixing antibody." Standard immunology teaching (Janeway, Roitt) is that IgM is the most potent classical pathway activator; a single IgM molecule can activate complement versus two IgG molecules required. If the intended answer is IgG, the question must specify a different qualifier.1f268261: Marks "Belongs to Filoviridae family" as the false statement about Ebola. Ebola does belong to Filoviridae — this is the defining feature of the virus family. The answer key is inverted.6c84280e: States S. pneumoniae does not require IgA protease. S. pneumoniae does produce IgA1 protease as a virulence factor. This is a content error.700a939c: Stem says "in vitro test for pneumococcus virulence" but the correct answer is intraperitoneal mouse inoculation — an in vivo animal model. The stem directly contradicts the answer.8dd52a3b: Marks "Sunlight" as the correct method for sputum disinfection. Sunlight is not a standard or recommended method in any major microbiology text; autoclaving or burning are standard.ebdbf455: Marks Streptococcus pyogenes as the most common organism causing upper respiratory tract infections in adults. Rhinoviruses are the actual most common cause; among bacterial causes, the framing is contested.a200e2da: States dengue is "NOT transmitted by blood transfusion." Dengue transmission via transfusion is documented in the literature; the absolute framing is factually contestable.853e12c0: Uses "Deuteromycetes" as a valid taxonomic classification. Deuteromycetes is no longer a valid taxonomic group in current mycology nomenclature.
Recommended disposition
Disable immediately for all questions with confirmed factual errors in the answer key (dac5b285, e3f4c9f5, 4807126e, c5621bf6, 1f268261, 700a939c). These cannot be used in any form until corrected by a subject-matter expert. Fix for questions where the error is in the stem framing or qualifier rather than the core fact (8e26e4dd, a200e2da, ebdbf455) — these require stem revision and expert review before reuse. Expert review for 6c84280e, 8dd52a3b, 853e12c0 before any disposition decision.
3. Broken Image-Dependent Questions
Why this pattern is bad
A question that references an image but does not include the image is not a question — it is an incomplete text fragment. In the best case, it is unanswerable. In the worst case, it is answerable by guessing from the remaining text clues, which means it tests nothing. Several questions in this sample were clearly designed around microscopy images, clinical photographs, or labeled diagrams, and the image has been lost or was never properly embedded. These questions cannot be fixed by editing the text; they require either image restoration or complete rewrite as text-based questions.
This is a discrete data integrity problem, not a content quality problem. The remediation path is different from other categories: it requires either locating and attaching the original image, or rewriting the question from scratch as a text-based item that does not depend on visual information.
How it shows up
In this sample, at least four questions are confirmed broken image-dependent items. The pattern is identifiable by stems that say "shown in the image below," "as seen in the figure," "the smear of pus below," or "the stool examination reveals a specific finding" with no image present in the question text. The correct answer cannot be derived from any information provided in the text.
Example question IDs
32e6fefd: Stem asks "What is the likely cause of this manifestation?" with no image and no clinical description. The correct answer (Pseudomonas) cannot be derived from any information provided. Completely unintelligible as a text question.c0ba6205: "Bacteria shown in the smear of pus below" — image absent. This is a PYQ-tagged item (NEET-PG 2017) with Bloom's 4 reasoning potential; it is worth restoring if the image can be located.24c5b336: "Stool examination reveals a specific finding" — image absent. Unanswerable as text-only.1757246b: "HIV antigen marked X in figure" — figure absent. The question cannot function without the structural diagram.Q11c35be7: Image-based LGV question (PYQ-tagged, NEET-PG 2017, Bloom's 4). High-quality item if image is available; non-functional without it.
Recommended disposition
Disable immediately for all confirmed broken image-dependent questions. For PYQ-tagged items with high intrinsic quality (c0ba6205, Q11c35be7), flag for image restoration — if the original image can be located and properly embedded, these items are worth keeping. For non-PYQ items (32e6fefd, 24c5b336, 1757246b), disable and rewrite as text-based questions if the concept is worth preserving.
4. Negative-Stem and "All/None of the Above" Format Defects
Why this pattern is bad
The "except/NOT true/false statement" format is not inherently bad — several PYQ gold-standard items use it effectively (e.g., 8cab8858, e32dbe37). The problem is how it is used in this pool. In the candidate sample, negative-stem questions are overwhelmingly applied to bare factual lists with no clinical context, producing items where the cognitive task is "identify the one wrong fact from a list of facts" rather than "reason about a clinical problem." This is a lower-order task than it appears, because the distractors are typically so obviously true that the correct answer (the false statement) is identifiable by elimination without understanding the underlying concept.
The "all of the above" and "none of the above" answer options are a separate but related defect. When "all of the above" is the correct answer, a candidate who knows any two of the three true statements can select it by elimination — the question tests partial knowledge rather than complete understanding. When "none of the above" is the correct answer, the question provides no positive educational content and gives no feedback about what the correct answer actually is.
How it shows up
Negative-stem questions appear in every shard. The "all of the above" / "none of the above" defect appears in at least 6–8 questions in the sample. The pattern is cross-topic, appearing in Bacteriology, Virology, Immunology, and Parasitology.
Example question IDs
0af97e6d: "All transmitted by blood except EBV" — bare factual list, no clinical context, trivially easy because EBV is the obvious outlier.0abbc19e: "Incorrect about tetanus" — "caused by endotoxin" is an obvious distractor that makes the question trivially easy; tetanus toxin is an exotoxin, universally known.f17a6a9d: "False about macrophages" — bare factual list, no clinical scenario.6df11b4f: "PJP FALSE statement" — marks "frequently associated with CMV" as FALSE, but co-infection with CMV in AIDS patients is well-documented; the word "frequently" makes this contestable.b90f6e15: "All of the above" is the correct answer for passive-active immunity — structural defect allowing shortcut by elimination.6fc8ca50: "All of the above" for chromosomal transfer mechanisms — structural defect plus imprecise option framing.388a7503: "Which type of microscope is commonly used in microbiology?" — "All of the above" as correct answer, trivially easy, no discriminatory value.19c76144: "None of the above" as the correct answer for occult filariasis — provides no positive educational content.509f0f60: "All of the above" as correct answer for false positive VDRL causes.Q9c5bb62d: "NOT physical sterilization" — decontextualized, classification ambiguity around filtration.Qb22dde40: "Ig NOT used for" — bare factual list, no clinical scenario.
Recommended disposition
Fix for negative-stem questions where the underlying concept is high-yield and the question is salvageable by adding a clinical vignette or replacing the problematic distractor. For example, 6df11b4f can be fixed by replacing the contestable CMV distractor with a clearly false statement. Disable for questions where "all of the above" or "none of the above" is the correct answer and the question has no other redeeming features (388a7503, 19c76144, 6fc8ca50). Apply a standing policy: "all of the above" and "none of the above" should not be used as the keyed correct answer in this question bank.
5. Near-Duplicate and Over-Concentrated Topic Coverage
Why this pattern is bad
When two or more questions in the pool test the same specific fact with only superficial variation in the stem (e.g., "endoscope sterilization" vs. "bronchoscope sterilization," both answered by glutaraldehyde), they add no incremental assessment value. They inflate the apparent coverage of a topic without actually testing different aspects of it. In a large pool like this one (11,104 questions), near-duplicates are particularly wasteful because they consume slots that could be used for higher-quality items on under-covered topics.
The sterilization and disinfection sub-topic is the clearest example in this sample, but the pattern also appears in Neisseria differentiation (maltose fermentation appears as the key fact in at least three separate questions across shards), syphilis serology (VDRL, prozone, false positives appear as a cluster), and opportunistic infections in HIV (Cryptococcus, PJP, CMV, Candida each appear multiple times at Bloom's 1–2).
How it shows up
Near-duplicates are identifiable when two questions share the same correct answer, the same key discriminating fact, and differ only in the surface framing of the stem. Over-concentration is identifiable when a single sub-topic (e.g., glutaraldehyde sterilization, Neisseria maltose fermentation, VDRL false positives) accounts for multiple questions in the sample without any of them reaching Bloom's 3.
Example question IDs
8bda6999andd64bc470: Endoscope sterilization vs. bronchoscope sterilization — both answered by glutaraldehyde, functionally identical questions with a different instrument name.8afcff95,8bda6999,d64bc470,817c57cd,3082168e,e6258424: Six sterilization/disinfection questions in the sample, all at Bloom's 1–2, several testing the same glutaraldehyde fact.- Neisseria maltose fermentation: Appears as the key discriminating fact in
5a3e13fc(shard 002),c7993be4(shard 003),Qbc5ab010(shard 005), andf7505e67-adjacent items — the same fact tested repeatedly without clinical context. - VDRL/syphilis serology cluster:
baa7e2b4,509f0f60,605991c8,35f9f5fa— multiple questions on syphilis serology at Bloom's 1–2 without clinical vignette framing.
Recommended disposition
For near-duplicate pairs, disable the weaker item and keep or fix the stronger one. For over-concentrated topic clusters, retain at most one or two items per specific fact, preferably the one with the best clinical framing. For the sterilization/disinfection cluster, retain one well-constructed question on the principle (e.g., heat-labile equipment requiring chemical sterilization) and disable the rest. For Neisseria differentiation, retain the best-framed item (the one closest to a clinical scenario) and disable the pure recall versions.
6. Topic Misclassification Affecting Blueprint Accuracy
Why this pattern is bad
When a question is filed under the wrong topic, it affects test blueprint accuracy, retrieval for targeted practice, and the ability to audit coverage gaps. A question about Candida oral lesions in HIV filed under "Virology" will not appear in a Mycology coverage audit. A question about VDRL flocculation filed under "Immunology" will not appear in a Bacteriology/Syphilis coverage audit. At scale, systematic misclassification creates the illusion of coverage in some topics and genuine gaps in others.
This is an operational problem with a clear remediation path: topic reassignment. It does not require content rewriting, but it does require a systematic audit pass, not case-by-case correction.
How it shows up
Topic misclassification appears in at least 6–8 questions in this sample. The most common patterns are: (a) opportunistic infection questions filed under Virology when the causative organism is a fungus or parasite; (b) syphilis serology questions filed under Immunology when they belong under Bacteriology or Diagnostic Microbiology; (c) Rickettsia questions filed under Bacteriology rather than Zoonotic Diseases; (d) culture medium questions filed under Microbiome rather than Diagnostic Microbiology.
Example question IDs
5664999c: Oral lesions in HIV, correct answer is Candida (a fungus), filed under Virology. Should be Mycology or Diagnostic Microbiology.baa7e2b4: VDRL flocculation test, filed under Immunology. Should be Bacteriology (Syphilis) or Diagnostic Microbiology.6c27edee: Streptococcal DNase B antigenicity, filed under Immunology. Should be Bacteriology.c7c2d6aa: Epidemic typhus/Rickettsia, filed under Bacteriology. Should be Zoonotic Diseases or Rickettsia.f7505e67: Thayer-Martin medium for gonorrhea, filed under Microbiome. Should be Diagnostic Microbiology or Bacteriology.407faca4: Skin-penetrating parasite question, topic tag inconsistent with content.
Recommended disposition
Fix topic tag for all confirmed misclassified questions. This is a low-effort, high-impact remediation that does not require content rewriting. Recommend a systematic topic-reassignment audit for the following high-risk categories: opportunistic infections (Virology vs. Mycology vs. Parasitology), syphilis serology (Immunology vs. Bacteriology), Rickettsia (Bacteriology vs. Zoonotic Diseases), and culture media (Microbiome vs. Diagnostic Microbiology).
7. Thin Clinical Vignettes: Clinical Framing That Does Not Drive Reasoning
Why this pattern is bad
This category is distinct from pure recall questions (Category 1) and distinct from benchmark-quality vignettes. It describes a specific failure mode: questions that have the surface structure of a clinical vignette — a patient age, a symptom, a lab result — but where the clinical context is decorative rather than functional. The answer is immediately deducible from a single keyword in the stem without integrating the other clinical information. The vignette does not add reasoning demand; it adds length.
This matters because these questions are the most fixable items in the pool. They have the right structure; they just need the clinical details to do real work. A question about Cryptosporidium in an HIV patient where the only discriminating clue is "4–6 µm oocyst size" is not a clinical reasoning question — it is a size-recall question with a patient attached. A question about H. pylori where the vignette says "34-year-old man diagnosed with H. pylori after biopsy" and then asks where H. pylori lives is not using the vignette at all.
How it shows up
This pattern appears in approximately 15–20 questions in the sample. It is most common in Parasitology (oocyst size questions dressed as HIV vignettes), Bacteriology (organism location questions with non-functional patient framing), and Mycology (dimorphism questions with a patient attached but no reasoning required). The benchmark question 38af2b72 (Entamoeba histolytica) demonstrates what a functional vignette looks like: every clinical detail — occupation, geography, symptoms, imaging — is necessary to distinguish the correct answer from the distractors.
Example question IDs
b31cda53: Cryptosporidium in HIV patient — the 4–6 µm oocyst size is the sole discriminating clue; the clinical context adds no reasoning demand. The question is a size-recall item with a vignette wrapper.9ccd1f82: H. pylori location — "34-year-old man diagnosed with H. pylori after biopsy" framing adds nothing; the answer is a standalone anatomical fact.7aef27cd: Pneumococcus differentiation — reasonable clinical scenario but the question is answerable without reading the vignette.8c86ad48: CMV post-bone marrow transplant — clinically reasonable concept but the stem is bare; no patient demographics, no timeline, no lab context.1e231e7d: Lymphocytic CSF → herpes virus — bacterial options are eliminated by the lymphocytic profile alone; no genuine differential reasoning required.bc203b1b: Gardnerella vaginalis with clue cells — acceptable but could be upgraded with additional lab data.d178be0a: Vomiting after ice cream → Staph aureus — single-keyword deduction (short incubation + preformed toxin), no genuine differential.2ce7e7eb: Whooping cough specimen — clinically framed but does not specify phase of illness, which is the key reasoning step.bd26cc23: Dimorphic fungi at body temperature — correct concept but too simple; the benchmark77a68f11demonstrates how to test this concept properly.dbdb89c0: Farmer with fever and livestock exposure → Coxiella — vignette structure present but clinical details are sparse and the answer is immediately obvious from "livestock exposure."
Recommended disposition
Fix for all questions in this category where the underlying concept is high-yield and the question has the right structure. The fix is specific: add 2–3 clinical details that are necessary to distinguish the correct answer from the distractors. For b31cda53, add a treatment decision or epidemiological reasoning element. For 9ccd1f82, either remove the non-functional vignette and rewrite as a pure recall item (then disable per Category 1 policy) or add clinical details that make the location matter (e.g., biopsy site selection, CLO test interpretation). For 2ce7e7eb, specify the phase of illness in the stem. Do not invest in fixing questions where the concept is trivial or already covered by a stronger item.
Prioritization
The seven issue categories identified above are not equally urgent. The following prioritization reflects both the severity of harm and the operational effort required for remediation.
Tier 1 — Immediate action required (before any use in practice tests or study plans)
Category 2 (Factual Errors in Answer Keys) is the highest priority. Questions with wrong answer keys cause direct harm to candidates. The confirmed errors in this sample — HAV family, HIV RNA nature, MHC Class I on lymphocytes, Gram staining for Streptococcus/Pneumococcus differentiation, Ebola/Filoviridae, in vitro vs. in vivo pneumococcus test — must be disabled immediately. No question with a confirmed factual error should appear in any user-facing context until it has been reviewed and corrected by a subject-matter expert.
Category 3 (Broken Image-Dependent Questions) is equally urgent from a functional standpoint. These questions are non-functional as text items and will confuse or mislead candidates. Disable all confirmed broken image-dependent questions immediately. Flag PYQ-tagged items for image restoration.
Tier 2 — High priority, address in the next content operations cycle
Category 1 (Structural Recall Overload) is the largest category by volume and the primary driver of the pool's low Bloom's distribution. The recommended action is bulk disable for the most egregious items (pure Bloom's 1 with no clinical context, trivial facts, vocabulary definitions) and selective fix for items where the concept is high-yield and a vignette upgrade is feasible. Given the volume, a triage approach is recommended: disable all items where the stem is a single sentence with no clinical context and the correct answer is a single named entity. This will remove the majority of Category 1 items efficiently.
Category 4 (Negative-Stem and All/None Format Defects) requires a policy decision as well as item-level remediation. Recommend establishing a standing rule: "all of the above" and "none of the above" are not acceptable as keyed correct answers. Apply this rule systematically across the full pool, not just the sampled items.
Tier 3 — Address in the next quality audit cycle
Category 5 (Near-Duplicates and Over-Concentration) requires a topic-level coverage audit rather than item-level review. Recommend a deduplication pass for the sterilization/disinfection, Neisseria differentiation, and syphilis serology sub-topics across the full pool of 11,104 questions.
Category 6 (Topic Misclassification) requires a systematic tag-reassignment audit for the high-risk topic pairs identified above. This is low-effort per item but requires a systematic pass rather than case-by-case correction.
Category 7 (Thin Clinical Vignettes) is the highest-value upgrade opportunity in the pool. These questions have the right structure and cover high-yield concepts; they just need clinical details that do real work. Recommend prioritizing these for vignette upgrade over writing new questions from scratch.
Example Keep / Fix / Disable Calls
The following table summarizes representative disposition calls drawn from the shard evidence. These are illustrative of the patterns described above, not an exhaustive list.
KEEP
| Question ID | Topic | Reason |
|---|---|---|
d0a4a503 |
Parasitology | Well-constructed Bloom's 3 vignette (Acanthamoeba keratitis), specific morphological clue, plausible distractors, meets benchmark standard |
8309e5fb |
Zoonotic Diseases | Good clinical scenario (cat scratch, regional lymphadenopathy), requires Bartonella species differentiation |
30aa35e8 |
Mycology | Clinical context (elderly diabetic, panophthalmitis), specific morphological descriptor, correct answer requires species-level knowledge |
aff4875b |
Virology | Clean Bloom's 3 vignette (HIV, CD4 139, India ink positive → Cryptococcus), appropriate distractors |
5af22a43 |
Virology | PYQ-tagged, Bloom's 3, Tzanck smear with multinucleated giant cells, plausible herpesviridae distractors |
e32dbe37 |
Mycology | PYQ-tagged, Bloom's 4, tests conceptual understanding of dimorphic fungus classification |
afc68c4e |
Bacteriology | Classic food poisoning scenario (boil on neck, unrefrigerated food, 2-hour onset), genuinely discriminating distractors |
ab09534b |
Bacteriology | Bloom's 3 vignette (Listeria, refrigerated vegetables), plausible distractors |
Q03e9efa3 |
Bacteriology | Clinical vignette (gas gangrene, shrapnel injury), PYQ-tagged, Bloom's 3 |
e850420e |
Zoonotic Diseases | Genuine clinical scenario (cat-scratch disease), organism-to-source reasoning required |
7bd92984 |
Virology | Clinically grounded (West Nile encephalitis, Southern Minnesota, CSF profile), Bloom's 3 |
6732fdcc |
Bacteriology | Bloom's 3 clinical scenario (submandibular mass, branched gram-positive rods, actinomycosis), next-step management framing |
3507ee8d |
Immunology | Specific high-yield fact (CD55/DAF vs CD59), plausible distractors, factually correct |
6ec63cf8 |
Diagnostic Microbiology | Practical diagnostic fact (buffered glycerol saline for Shigella), correct, reasonable distractors |
61fc55df |
Immunology | PYQ-tagged, Bloom's 3, tests MHC restriction concept with plausible distractors |
FIX
| Question ID | Issue | Required Action |
|---|---|---|
e3f4c9f5 |
Wrong answer key (Gram staining cannot differentiate Streptococcus from Pneumococcus) | Correct answer key to bile solubility/optochin sensitivity; revise stem to specify clinical/lab context |
8e26e4dd |
Probable factual error (IgG marked as most potent complement fixer; IgM is standard teaching) | Expert review; either correct answer to IgM or qualify stem to specify secondary response/abundance |
bd0dcfec |
Ambiguous answer key (only stool antigen assay marked non-invasive, excluding urease breath test) | Revise stem to ask for "gold standard non-invasive test" or correct key to acknowledge both |
35f9f5fa |
Compound answer option (two distinct facts combined in one option) | Split into separate options or restructure to test one concept at a time |
b31cda53 |
Thin vignette (size recall dressed as clinical scenario) | Add treatment decision or epidemiological reasoning element to reach Bloom's 3 |
9ccd1f82 |
Non-functional vignette (clinical framing adds nothing to H. pylori location question) | Either remove vignette and disable per Category 1 policy, or add clinical details that make location matter |
2ce7e7eb |
Missing key clinical detail (phase of illness not specified for whooping cough specimen) | Add "early catarrhal phase" to stem to make reasoning explicit and defensible |
6df11b4f |
Contestable distractor (PJP/CMV co-infection "frequently" is debatable) | Replace with clearly false statement (e.g., "responds to fluconazole") |
5664999c |
Topic misclassification (Candida oral lesions filed under Virology) + ambiguous stem | Reassign to Mycology; add clinical descriptors to distinguish from oral hairy leukoplakia |
407faca4 |
Factual incompleteness (Strongyloides excluded from skin-penetrating parasites) | Reframe to specify hookworm specifically, or revise distractors so Strongyloides is not a competing correct answer |
8c86ad48 |
Thin vignette (CMV post-BMT, no demographics or timeline) | Add patient demographics, day post-BMT, and bilateral infiltrates on CXR to reach Bloom's 3 |
a200e2da |
Contestable absolute framing (dengue "cannot" be transmitted by transfusion) | Reframe as "least likely to be transmitted by blood transfusion" |
ac418ed0 |
Poorly sourced factoid (diphtheria toxin resembles snake venom) | Verify against standard references; if unsupported, replace with question on ADP-ribosylation of EF-2 |
949c4141 |
Double-true distractor (Vibrio cholerae vaccine distractor is also arguably false) | Replace vaccine distractor with unambiguously true statement |
b7642c83 |
Unanchored stem ("Which one of the following statements is true?" with no topic) | Rewrite stem with explicit topic anchor; verify all distractors are unambiguously false |
Q25b75983 |
Contestable correct answer (orchitis as most common mumps complication) | Revise to "most common extrasalivary gland complication in post-pubertal males" |
Qd1db02d1 |
Structurally broken options (declarative statements as answer choices) | Rewrite as clinical vignette; replace declarative-statement options with named entities |
Q5654967f |
Implausible distractors (bacteria as distractors in mycology question) | Replace bacterial distractors with fungal alternatives |
c7c2d6aa |
Topic misclassification (Rickettsia filed under Bacteriology) | Reassign to Zoonotic Diseases |
f7505e67 |
Topic misclassification (Thayer-Martin medium filed under Microbiome) | Reassign to Diagnostic Microbiology |
baa7e2b4 |
Topic misclassification (VDRL filed under Immunology) | Reassign to Bacteriology or Diagnostic Microbiology |
8bda6999 + d64bc470 |
Near-duplicate pair (endoscope vs. bronchoscope sterilization, same answer) | Disable one; rewrite the other to test a different aspect of sterilization |
DISABLE
| Question ID | Reason |
|---|---|
3f0ad811 |
"Chickenpox caused by VZV" — definitional tautology, Bloom's 1, zero discriminatory value |
de7cd55e |
"Influenza family" — taxonomy recall, no clinical relevance |
de8a5d44 |
"E. coli generation time" — isolated numerical fact, implausible distractors |
e6258424 |
"Hot air oven 160°C for 1 hour" — rote numerical recall, no reasoning |
577ebf4d |
"IL-2 secreted by Helper T-cells" — single cytokine-cell association, Bloom's 1 |
4c579945 |
"Paul-Bunnell test for infectious mononucleosis" — rote eponym recall, creates confusion with Monospot |
400630b7 |
"Human papillomatosis caused by HPV" — tautological |
e46fec1f |
"Naturally acquired passive immunity = placental transfer" — Bloom's 1, no clinical context |
f80d8661 |
"What does pleomorphic mean?" — vocabulary definition, not a microbiology application |
44ab6d59 |
"Pontiac fever caused by?" — single-line, no stem, Bloom's 1 |
9c931fc0 |
"Tetanospasmin genes on plasmid" — bare-fact genetics recall |
a03c8640 |
"Monoclonal antibodies bind specific epitope" — definitional, Bloom's 1 |
Q97dc76b0 |
"Highest serum Ig concentration" — universally known, no discriminatory value |
Qa6e6c7e1 |
"DHR mediated by T lymphocytes" — single-fact recall, no vignette |
Q792afb07 |
"Mite-transmitted disease = scrub typhus" — single-association recall |
Q1d609b5d |
"Genital warts caused by HPV" — pre-clinical knowledge |
Q3761395b |
"Which is a trematode?" — answerable by elimination without knowing what a trematode is |
Qd555100b |
"All of the above" correct answer, trivially obvious correct answer |
dac5b285 |
Confirmed factual error — HAV marked as Hepadnaviridae |
4807126e |
Confirmed factual error — HIV marked as not an RNA virus |
c5621bf6 |
Confirmed factual error — lymphocytes marked as exception to MHC Class I |
1f268261 |
Confirmed factual error — Ebola/Filoviridae marked as false statement |
700a939c |
Stem-answer contradiction — "in vitro" test answered by in vivo animal model |
32e6fefd |
Broken image-dependent — completely unintelligible as text question |
24c5b336 |
Broken image-dependent — unanswerable without image |
c03edf76 |
"Prokaryotes lack mitochondria" — pre-medical level, appears in GNM/ANM templates |
cf2ad745 |
"Scarlet fever caused by Streptococci" — implausible distractors, no discriminatory value |
fa09277c |
"DT 104 strain belongs to Salmonella Typhimurium" — highly specific strain nomenclature, no clinical relevance framing |
10a39258 |
"Pertussis cannot spread in laboratory" — factually contestable, better replaced with biosafety level question |
388a7503 |
"Which microscope is commonly used?" — "All of the above" answer, trivially easy, not PG-level |
2897447d |
"Aflatoxin produced by Aspergillus flavus" — pure recall, no clinical hook |
54ea6dc5 |
"Kidney-shaped organism = Gonococci" — trivial morphology recall |
d4029a04 |
"In cholangitis, organism mostly responsible?" — informal phrasing, Bloom's 1, no clinical context |
853e12c0 |
Outdated taxonomy (Deuteromycetes no longer valid) |
a875d6ef |
"Swollen belly syndrome — Strongyloides fuelleborni" — extremely obscure, no clinical framing, not aligned with any PYQ |
6fc8ca50 |
"All of the above" correct answer + imprecise option framing for chromosomal transfer |
19c76144 |
"None of the above" as correct answer for occult filariasis — no positive educational content |
54a5170c |
"AIDS primarily affects T lymphocytes" — Bloom's 1, categorically inconsistent options (immunoglobulins and opsonins are not cell types) |
fe14d494 |
"AFB negative slide field count" — numerical recall, below PG standard |
817c57cd |
"High-level disinfectant" — Bloom's 1 classification recall, no clinical context |
76bffa9d |
"Unenveloped ssRNA virus" — pure virology classification memorization |
12b72d13 |
"Which cell type produces IL-1?" — textbook flashcard |