Psychiatry Question Quality Review
Executive Summary
The 200-question candidate sample for Psychiatry was reviewed across eight shards of 25 questions each. The sample draws from a pool of 4,716 questions and excludes the 6 benchmark and 12 recent PYQ items used as the quality bar.
The Blooms distribution is heavily skewed toward recall: 56 questions at Blooms-1 and 119 at Blooms-2 account for 87.5% of the candidate sample, with only 25 questions at Blooms-3 or above. This is not inherently disqualifying — some Blooms-1 and Blooms-2 items in the benchmark set are excellent — but in this sample the low Blooms level correlates strongly with structural and factual problems that make a large proportion of the questions unsuitable for use without remediation.
The review identified six distinct issue categories, each with a different remediation path. The most operationally urgent are: (1) a cluster of questions with outright wrong or outdated answer keys, which will actively misinform candidates; (2) a pervasive pattern of Blooms inflation, where questions labeled Blooms-2 through Blooms-4 function as Blooms-1 recall; and (3) a large body of trivial recall and eponym-slang items that fall below the quality floor for PG entrance examination use. Secondary but still significant issues include structural defects in option construction, systematic topic misclassification, and near-duplicate content clustering around a small number of high-frequency topics.
Across the 200-question sample, the estimated disposition breakdown is approximately: Keep as-is ~25–30%, Fix before use ~30–35%, Disable ~35–40%. The disable rate is high relative to other subjects and reflects both the volume of trivial recall items and the number of questions with factually incorrect answer keys.
What Good Looks Like
The benchmark and recent PYQ items establish a clear quality bar. The best items in those sets share four properties that the candidate sample frequently lacks.
Clinical scenario with genuine differential demand. Benchmark item a050f160 (autistic thinking — "Lord Hanuman was celibate, I am celibate too, so I am Lord Hanuman") presents a real clinical utterance and requires the candidate to distinguish autistic thinking from loosening of association, neologism, and verbigeration. PYQ item beb15713 (14-year-old with sudden blindness and la belle indifférence) requires the candidate to identify conversion disorder and then apply a specific epidemiological fact about sex distribution in children. Neither question resolves on a single surface keyword.
Distractors that are clinically plausible and educationally meaningful. PYQ item 30f92771 (OCD with contamination compulsions — best treatment) offers ERP, systematic desensitization, SSRI, and DBT as options. All four are real psychological or pharmacological interventions; the question tests whether the candidate knows which is first-line for OCD specifically, not just whether they recognize the disorder. Benchmark item b93efd5a (substance use disorders — all except) uses flumazenil for long-term alcohol dependence as the keyed exception, which is a plausible-sounding but incorrect option that requires genuine knowledge to reject.
Correct answer keys that reflect current classification. Benchmark item 1585b4f2 tests ICD-10 depression criteria and correctly identifies "persistent worry about everyday matters" (a GAD criterion) as the exception. PYQ item 11efae62 tests rapid cycling bipolar and correctly identifies male sex as the exception (rapid cycling is more common in women). Both items are anchored to a named, current classification system.
Appropriate Blooms calibration. PYQ item 93cfb360 (supportive therapy most challenging in which patient) is labeled Blooms-4 and genuinely requires evaluation: the candidate must reason about what makes therapy implementation difficult, not just recall a definition. Benchmark item 7daf29e4 (ASD core criterion per DSM-5) is labeled Blooms-3 and requires the candidate to distinguish core criteria from associated features and from criteria in older classification systems.
The candidate sample contains a minority of items that meet this standard — approximately 50–60 of the 200 reviewed questions — and these are identified in the Keep calls below. The remainder fall into one or more of the issue categories described next.
Main Issue Categories
1. Wrong or Outdated Answer Keys
Why this pattern is bad. Questions with incorrect or outdated correct answers are the highest-priority problem in any question bank. A candidate who studies from these items learns the wrong fact. Unlike a poorly worded question that a candidate can reason around, a wrong answer key actively corrupts knowledge. Several of the errors in this sample arise from DSM-IV-TR or ICD-9-era content that has been superseded by DSM-5 and ICD-11 without the answer key being updated; others appear to be straightforward factual mistakes.
How it shows up. The pattern appears across multiple shards and multiple topic areas. It is not confined to one topic. The most common subtypes are: (a) classification-system obsolescence, where the correct answer was true under DSM-IV-TR but is false under DSM-5; (b) pharmacological guideline drift, where a drug designated as "treatment of choice" no longer holds that position in current guidelines; (c) outright factual errors that appear to have been present at the time of authoring.
Examples from the reviewed set.
2abb2fc6— Marks catatonia as "a type of schizophrenia." Under DSM-5 and ICD-11, catatonia is a specifier applicable across mood disorders, medical conditions, and other psychotic disorders, not a schizophrenia subtype. This will directly misinform candidates preparing for INICET, which tests DSM-5/ICD-11 content.b6c612c8— Marks "suboptimal development" as the current term for Down syndrome. This is not a recognized clinical or nosological term in any current classification. The correct modern term is intellectual disability. The answer key is simply wrong.bf347669— Marks cholecystokinin as NOT an anxiogenic agent. CCK-4 is a well-validated experimental anxiogenic used in human challenge studies and is discussed in standard psychopharmacology texts. The answer key is factually incorrect.61f2a262— Marks schizophrenia as the psychiatric disorder most commonly associated with stroke. Post-stroke depression is the most common psychiatric sequela of stroke by a wide margin in the literature. The answer key is wrong.29541b35— Marks "vivid dreams" as NOT a symptom of PTSD. Recurrent distressing dreams are a DSM-5 Criterion B symptom of PTSD. The answer key is wrong.5511b8f3— Marks lithium as the sole drug of choice for acute mania. The benchmark itemd701fb08explicitly demonstrates that valproate and haloperidol are equally guideline-supported for acute mania. Designating lithium alone as the correct answer conflicts with the benchmark.1ea770e4— Structured as an "all are true except" question about ADHD, but the keyed correct answer ("symptoms before 12 years") is the TRUE DSM-5 criterion, not the false one. The false statement is "before 7 years" (the old DSM-IV criterion). The question logic is internally contradictory.d5f3cba7— Marks reactive depression as having the highest suicidal tendency. Standard teaching associates highest suicide risk with psychotic or severe melancholic depression, not reactive depression.dc42a925— Associates compulsive sexual behavior most closely with hair pulling. This is conceptually incoherent; the behavioral analog for compulsive sexual behavior is gambling disorder or other behavioral addictions, not trichotillomania. The answer key appears corrupted or misassigned.9f5c4d02— The vignette describes a familiar person (nurse) believed to be replaced by an imposter (wife) — this is Capgras syndrome. The question keys Fregoli syndrome, which is the reverse (a stranger believed to be a known person in disguise). The answer key is wrong.8f846abb— Marks benzodiazepine as treatment of choice for GAD. Current guidelines (NICE, BAP, and Indian Psychiatric Society) designate SSRIs/SNRIs as first-line for GAD. Benzodiazepines are appropriate for short-term or adjunctive use but not as the primary treatment of choice.
Recommended disposition. All items in this category require expert review before any use. Items where the error is a simple key swap or stem correction (e.g., 1ea770e4, 9f5c4d02) should be fixed. Items where the entire premise is wrong or the question cannot be salvaged without a full rewrite (e.g., 2abb2fc6, b6c612c8, dc42a925, 61f2a262) should be disabled. No item in this category should be published in its current form.
2. Structural Defects in Option Construction
Why this pattern is bad. Structural defects — duplicate options, logically incoherent distractors, "all of the above" as the keyed answer, "none of the above" as the keyed answer — make questions unanswerable or trivially easy regardless of the candidate's knowledge. They also introduce scoring disputes and erode candidate trust in the platform. These are distinct from factual errors: the problem is in the architecture of the question, not the underlying content.
How it shows up. The most common subtypes observed in this sample are: (a) duplicate option text within a single question; (b) "all of the above" as the correct answer, which eliminates all discrimination; (c) logically self-contradictory option sets (e.g., "all of the above statements are true" as a distractor in a FALSE-statement question); (d) implausible distractors that make the correct answer obvious without any knowledge of the topic.
Examples from the reviewed set.
1fc42046— Options A and C are identical ("Echolalia") in a catatonia "except" question. The question is unanswerable as presented because two options are the same text, one of which is marked correct and one incorrect.a4f0e594— Uses "All of the above statements are true" as a distractor in a question asking which statement is FALSE. This is logically self-contradictory and cannot be a valid distractor in a false-statement format.326284e3— Two options have identical text in a hysteria vs. hypochondria differentiation question. One must be replaced with a genuinely distinct differentiating feature before the question is usable.905aeb01— Uses "None of the above" as the correct answer in an early-onset Alzheimer's "except" question. This option provides no discriminatory information and teaches nothing; the candidate learns only that none of the listed options is correct, not what the correct fact is.cb282137— "Non-memory cognitive symptoms in Alzheimer's disease" with "All of the above" (aphasia, apraxia, agnosia) as the correct answer. Zero discriminatory value; any candidate who recognizes any one of the three options as correct will select "all of the above."df9b8197— "Common treatment for unipolar depression" with "All of the above" (fluoxetine, sertraline, citalopram) as the correct answer. Three individually correct options collapsed into a format that tests nothing.44fc1149— "Which of the following causes withdrawal syndrome?" with "All of the above" (alcohol, methadone, morphine) as the correct answer. Trivially easy; tests only whether the candidate knows that all three cause withdrawal, which is basic knowledge.3cc64266— Options include "All of the options" alongside "Intellectualization" and "Depersonalization" in a question about cognitive disorders. Intellectualization is a defense mechanism, not a cognitive disorder, making the option set internally incoherent.272e2469— Uses "None of the options" as a distractor when a clearly correct answer exists among the other options, creating unnecessary confusion.
Recommended disposition. Items with duplicate options (1fc42046, 326284e3) must be fixed before any use — they are currently unanswerable. Items with "all of the above" as the correct answer (cb282137, df9b8197, 44fc1149) should be disabled; the format is a known quality failure and the underlying concepts are better tested with a properly constructed item. Items with logically incoherent option sets (a4f0e594, 3cc64266) should be fixed if the underlying concept is worth testing, or disabled if the concept is already covered by a better item.
3. Trivial Recall and Eponym-Slang Items Below the PG Quality Floor
Why this pattern is bad. A question that any candidate can answer correctly by recognizing a single word or recalling a single isolated fact adds no discriminatory value to a PG entrance examination. These items inflate the easy/Blooms-1 count, consume question slots that could be occupied by higher-quality items, and — when they test obsolete terminology or street slang — actively waste study time. The benchmark set uses eponyms and named syndromes only when they carry diagnostic or management implications (e.g., Korsakoff's syndrome, Capgras syndrome in a differential context). The candidate sample contains a substantial cluster of items that do not meet this standard.
How it shows up. Three subtypes are observed: (a) pure vocabulary/definition recall with no clinical context; (b) historical or obsolete eponyms with no current clinical utility; (c) street slang and colloquial drug terminology with no diagnostic relevance.
Examples from the reviewed set.
674729eb— "Acid head" as slang for a heavy LSD user. Pure street-slang trivia with zero clinical or diagnostic relevance. No PYQ precedent at this level of trivia.c0e68e0a— Schizophrenia associated with "asthenic" personality per Kretschmer's constitutional typology. This is an archaic, discredited framework with no place in modern PG examination.1725371d— DSM-IV-TR Axis III = medical disorders. DSM-IV-TR has been superseded by DSM-5, which abolished the multiaxial system. Testing a deprecated classification at this level of trivia is not appropriate.b2f35d15— "Othello syndrome = jealous killer." The framing is imprecise (Othello syndrome is morbid jealousy, not specifically homicide) and the question is Blooms-1 eponym recall. Compare with0859671fin the same sample, which tests Othello syndrome through a genuine clinical vignette with differential reasoning — that is the correct approach.9e90b74d— "Dementia praecox coined by whom?" → Kraepelin. Historical trivia, Blooms-1, no clinical relevance.c740d1ad— "Who introduced psychoanalysis?" → Freud. Same pattern.ed0355f5— "Elderly gentleman develops severe depression — what is this called?" → Involutional melancholia. The term is not in DSM-5 or ICD-11; the question tests obsolete nomenclature.f9f4ba00— Thanatophobia = fear of death. Pure vocabulary recall, no clinical utility at PG level.72e76fc5— Onanism = a sexual behaviour term. Trivial terminology, wrong topic tag (filed under Substance Use Disorders), no discriminatory value.17e1ef86— "Derailment is NOT a defense mechanism." Tests only whether the candidate recognizes a word; any candidate who knows either thought disorder terminology or defense mechanisms answers instantly.2826a727— "Delusion of persecution = belief of being persecuted." The question and correct answer are literally the same phrase. Zero discriminatory value.07fd4afb— "Hallucination = false perception without external stimulus." Textbook definition reproduced verbatim.db0ab214— "Empathy = therapist sharing client's emotions." Definitional recall with no clinical context.e5c711a0— "Obsessions best defined as recurrent intrusive thoughts." Blooms-1 definition, no clinical scenario.baad0713— "ECT pre-medication → muscle relaxant." Does not distinguish succinylcholine from other agents; reduces a clinically important pharmacology point to a generic label.750b0b7c— "Moclobemide = RIMA." Blooms-1, no clinical context about tyramine interaction or dietary restriction comparison.213ab5d8— "Biological rhythm disruption → depression." Trivially easy, no plausible distractors.
Recommended disposition. Disable across the board. These items do not meet the quality floor for PG entrance examination use. The underlying concepts — where they are worth testing — should be covered by vignette-based or application-level items. Several of these concepts (Othello syndrome, defense mechanisms, ECT pharmacology, RIMA mechanism) already have better-quality items elsewhere in the benchmark or PYQ sets, making these redundant as well as weak.
4. Blooms Inflation — Recall Items Mislabeled as Application or Analysis
Why this pattern is bad. Blooms-level labels drive question selection for tests, daily plans, and difficulty calibration. When a recall item is labeled Blooms-3 or Blooms-4, it gets selected for slots intended for application and analysis questions, displacing genuinely higher-order items and making tests easier than intended. Conversely, when a genuine clinical reasoning item is labeled Blooms-1, it gets underused. The inflation in this sample is predominantly in the upward direction: recall items carrying Blooms-2, Blooms-3, or Blooms-4 labels.
How it shows up. The pattern is pervasive and cross-cutting — it appears in every shard reviewed. The most common mechanism is a thin clinical wrapper: a one-sentence scenario is added to a recall question, and the Blooms label is upgraded to reflect the vignette format rather than the actual cognitive demand. In practice, the scenario adds no reasoning requirement because the answer is immediately obvious from a single keyword in the stem.
Examples from the reviewed set.
efab522a— "What is dysthymia?" labeled Blooms-1 (correctly), but the question is a bare definition with no clinical context. The label is accurate but the item is below the quality floor.06512d5c— "Core feature of manic episode → elevated mood" tagged NEET-PG-2013 and Blooms-2. No application required; the answer is the defining feature of the condition by name.f1629371— "Projection = blaming others" labeled Blooms-2. Pure definition recall.ea38f6aa— Labile affect vignette labeled Blooms-3. The vignette adds no reasoning requirement; the answer is immediately obvious from the single clinical descriptor.5511b8f3— "Drug of choice for acute mania" labeled Blooms-4 (analysis/evaluation). This is a straightforward recall item regardless of the Blooms label.29541b35— PTSD symptoms question labeled Blooms-4. Single-fact retrieval about whether vivid dreams are a PTSD symptom.9f27db4f— "Carbamazepine for temporal lobe epilepsy" labeled Blooms-2. Single-fact recall.957a34c3— "Postoperative delirium" labeled Blooms-2. Bare-fact recall with no clinical reasoning.6e0cb603— "11-month-old boy restless in class → ADHD" labeled Blooms-3. The answer is immediately obvious from the scenario; no differential reasoning is required (and the age is almost certainly a typographical error — see Category 5).35b8864d— "Lady can't speak after fight → Conversion Disorder" labeled Blooms-3. The answer is immediately obvious from the single surface feature (functional neurological symptom after emotional trigger).
The reverse error also appears, though less frequently: 75ec5ae5 (haloperidol → uprolling of eyes → acute dystonia) is labeled Blooms-1 but involves genuine clinical reasoning from a drug-induced EPS scenario.
Recommended disposition. Blooms inflation is a calibration problem, not a content problem. Items that are otherwise factually correct and structurally sound should be fixed by correcting the Blooms label and, where feasible, upgrading the item to match the labeled level (e.g., by adding a clinical scenario or a more demanding distractor set). Items that are both Blooms-inflated and trivially easy should be disabled. A systematic re-audit of all Blooms-3 and Blooms-4 labels in the Psychiatry pool is warranted given the pervasiveness of this pattern.
5. Vignette Integrity Failures — Demographic Errors, Ambiguous Stems, and Thin Clinical Context
Why this pattern is bad. Vignette-based questions are the primary vehicle for testing clinical reasoning at PG entrance level. When the vignette contains a factual impossibility (wrong age, impossible clinical scenario), an ambiguous qualifier that changes the correct answer, or so little clinical detail that the scenario adds no reasoning requirement, the question fails its core purpose. These failures are distinct from Blooms inflation (Category 4): the issue here is not just that the question is too easy, but that the clinical scenario itself is broken or misleading.
How it shows up. Four subtypes appear in this sample: (a) demographic errors that create clinical impossibilities; (b) ambiguous qualifiers in the stem that make the correct answer depend on an unstated assumption; (c) vignettes so thin that they function as recall prompts rather than reasoning tasks; (d) stems that conflate two different clinical constructs without justification.
Examples from the reviewed set.
6e0cb603— "An eleven-month-old boy is restless in class → ADHD." ADHD cannot be diagnosed at 11 months; the child is not yet walking, let alone attending class. This is almost certainly a typographical error for "eleven years," but as written it is a clinical impossibility. The question cannot be used until corrected.702ea0e3— Smoker contemplating quitting: the correct answer conflates Prochaska's Contemplation stage with the Health Belief Model's "perceived susceptibility." These are constructs from two different theoretical frameworks; the question does not specify which model is being tested, making the correct answer a category error.483fdc3c— "Commonest phobia in clinical practice" without specifying whether epidemiological prevalence or clinical referral frequency is intended. Specific phobia is more prevalent epidemiologically; agoraphobia is most common in clinical referral settings. The question is ambiguous without this qualifier.46822cbb— "OCD not responding to sertraline 50 mg → next step." The therapeutic dose range for OCD is 100–200 mg; 50 mg is a sub-therapeutic starting dose. The question does not specify whether an adequate dose was tried, making "increase dose" vs. "switch class" genuinely ambiguous. A prepared candidate who knows OCD dosing guidelines will dispute the answer.439c71a1— Illusion vs. hallucination: the stem states the patient "can see this face every time she walks past that tree" — the constancy and specificity of the perception makes it a borderline case between illusion and a formed visual hallucination. The stem needs clarification (e.g., explicitly state she sees it only in poor lighting or at a distance) to unambiguously anchor the correct answer.f7fb21b5— La belle indifférence keyed to "dissociative disorders" broadly. La belle indifférence is classically associated with conversion disorder (functional neurological symptom disorder) specifically, not dissociative disorders as a category. The correct answer is imprecise in a way that could mislead learners about the distinction between conversion and dissociation.906ef8c6— Diazepam loading dose for alcohol withdrawal given as 60–150 mg. This range is unusually wide and not standard across major references; most guidelines describe symptom-triggered protocols rather than fixed ranges. The question will generate disputes among candidates who have studied from different references.baae6a8f— "Novel neurotransmitter with effect in pathophysiology of mood disorders" — the stem is vague enough that schizophrenia and OCD (offered as distractors) are also plausible associations for nitric oxide. The question needs a more precise stem or a supporting literature reference to justify a single correct answer.
Recommended disposition. Items with demographic impossibilities (6e0cb603) must be corrected before any use. Items with ambiguous qualifiers (483fdc3c, 702ea0e3, 46822cbb) should be fixed by adding the missing qualifier to the stem. Items with imprecise correct answers (f7fb21b5) should be fixed by tightening the answer text. Items where the ambiguity cannot be resolved without a full rewrite should be disabled.
6. Topic Misclassification — Systematic Misfiling Across Multiple Topics
Why this pattern is bad. Topic tags drive question selection for topic-specific quizzes, daily plans, and analytics. When questions are systematically misfiled, topic-level performance data is corrupted, candidates studying a specific topic receive irrelevant questions, and the content team cannot accurately assess coverage gaps. The misclassification pattern in this sample is not random — it reflects a small number of topics being used as catch-all bins, particularly Personality Disorders and Psychiatric Assessment and Diagnosis.
How it shows up. The pattern appears across multiple shards. The most common misclassification directions are: (a) OCD-related content filed under Personality Disorders; (b) sexual behaviour and paraphilia content filed under Psychiatric Assessment and Diagnosis or Substance Use Disorders; (c) organic/neuropsychiatric content filed under Psychotic Disorders or Psychiatric Assessment and Diagnosis; (d) psychotherapy content filed under Anxiety Disorders or other clinical topic areas; (e) forensic/legislative content filed under Psychiatric Assessment and Diagnosis without a dedicated forensic tag.
Examples from the reviewed set.
f9f4ba00(Thanatophobia) and8d0d6811(OCD treatment — ERP) — both filed under Personality Disorders. OCD-related content belongs under Obsessive-Compulsive and Related Disorders.72e76fc5(Onanism) — filed under Substance Use Disorders. Sexual behaviour terminology belongs under Personality Disorders or Sexual Disorders.78c574a3(Korsakoff's psychosis) — filed under Neurobiology and Genetics rather than Substance Use Disorders.34b1d5a4(anorexia nervosa features) — filed under Personality Disorders. Anorexia belongs under Eating Disorders.501236d2(paraphilia classification) — filed under Psychiatric Assessment and Diagnosis rather than Personality/Sexual Disorders.a742ea9e(National Trust Act) — filed under Psychiatric Assessment and Diagnosis. Forensic and legislative psychiatry items need a dedicated tag to distinguish them from clinical assessment content.12f4bca3(vascular dementia) — filed under Psychotic Disorders. Vascular dementia belongs under Geriatric Psychiatry or Organic Psychiatry.0e9df778(delirium definition) — filed under Psychiatric Assessment and Diagnosis. Delirium belongs under Organic/Geriatric Psychiatry.6018b8c3(frontal lobe tumor features) — filed under Psychiatric Assessment and Diagnosis. This is a neuropsychiatry/organic psychiatry item.0b3ff6da(Alcoholics Anonymous) — filed under Anxiety Disorders. Belongs under Substance Use Disorders.a9f615b8(aversion therapy) — filed under Anxiety Disorders. Belongs under Psychotherapies.
Recommended disposition. Topic misclassification is a fix-only issue — the underlying question content may be acceptable, but the tag must be corrected. A batch reclassification pass is more efficient than item-by-item correction. The content team should audit the Personality Disorders and Psychiatric Assessment and Diagnosis topic bins specifically, as these appear to be the primary catch-all destinations for misfiled content. A dedicated Forensic Psychiatry tag and an Organic/Geriatric Psychiatry tag should be verified as existing and correctly applied.
Prioritization
The six issue categories are not equally urgent. The following prioritization reflects both the severity of the problem and the operational effort required to address it.
Immediate action required — do not publish in current form.
Category 1 (Wrong or Outdated Answer Keys) is the highest priority. Questions with incorrect answer keys will actively misinform candidates. The items identified in this category should be pulled from all active test templates and daily plans immediately pending expert review. The volume observed in this sample — at least 10–12 items with clearly wrong keys across 200 questions — suggests this is not an isolated problem in the full 4,716-question pool.
Category 2 (Structural Defects in Option Construction) is equally urgent for the specific items with duplicate options or logically incoherent option sets, which are currently unanswerable. These should be pulled from active use immediately.
High priority — batch remediation pass.
Category 3 (Trivial Recall and Eponym-Slang Items) represents the largest volume of disable candidates. A batch review pass targeting questions with no exam tags, no template membership, Blooms-1 labels, and single-word or single-fact stems would efficiently identify the bulk of this category for disabling. The content team should not attempt to rewrite these items individually; the underlying concepts are either already covered by better items or are not worth testing at PG level.
Category 6 (Topic Misclassification) is a batch fix that does not require content expertise — it requires a systematic audit of the Personality Disorders and Psychiatric Assessment and Diagnosis topic bins and a reclassification pass. This should be done before the next analytics cycle to avoid corrupting topic-level coverage data.
Medium priority — item-by-item review.
Category 4 (Blooms Inflation) requires a calibration pass across all questions labeled Blooms-3 and Blooms-4 in the Psychiatry pool. Items that are factually correct and structurally sound but mislabeled should have their Blooms level corrected. Items that are both mislabeled and trivially easy should be disabled. This pass requires content expertise but can be done systematically.
Category 5 (Vignette Integrity Failures) requires item-by-item review because the fixes are stem-specific. The demographic error in 6e0cb603 is a simple correction; the ambiguous qualifier issues in 483fdc3c and 46822cbb require more careful rewriting. These items should be flagged for individual expert review rather than batch processing.
Example Keep / Fix / Disable Calls
The following calls are drawn from across the eight shards and are intended to illustrate the application of the issue categories above to specific items.
KEEP
4a8ad485 — NMS vignette with muscle rigidity, fever, elevated CPK → dantrolene. Clean clinical scenario, appropriate Blooms-3 application, correct answer, good distractors. Matches benchmark standard.
c08bd50b — Wandering patient with amnesia for travel → dissociative fugue. Well-constructed vignette, Blooms-3, plausible distractors (dementia, dissociative amnesia, schizophrenia).
0859671f — Othello syndrome vignette with plausible distractors (schizophrenia, Clerambault's, delusional parasitosis). Requires genuine differential reasoning. Blooms-4, difficulty-3. Matches benchmark standard.
dc4aee15 — Delirium tremens "except" question testing the distinction between DT and Wernicke's encephalopathy. Clinically important, PYQ-tagged, distractors are meaningful.
4df1068d — Thioridazine → pigmented retinopathy. Clinically relevant adverse-effect question with good distractor set (haloperidol, chlorpromazine, clozapine all plausible).
fd5a3b72 — Lorazepam NOT an anticraving agent for alcohol. Correct answer is factually sound; distractors (naltrexone, topiramate, acamprosate) are all genuine anticraving agents. Tests a practically important distinction.
06c82233 — 72-year-old with suicidal intent → ECT. Genuine Blooms-5 management question where age, acuity, and suicidal intent converge. Distractors are clinically meaningful.
236f0482 — Anti-NMDA receptor encephalitis → immunotherapy. High-yield, contemporary, Blooms-5 synthesis. Correctly distinguishes autoimmune psychosis from primary psychosis.
321d9be6 — Anorexia nervosa, no weight gain despite diet → observe post-meal. Genuine Blooms-3 clinical reasoning; correct answer requires understanding of the disorder's behaviour, not just recall.
53cf528b — GAD with SSRI-induced initial anxiety → add benzodiazepine temporarily. Well-constructed clinical vignette, appropriate Blooms-4, clinically defensible correct answer.
FIX
6e0cb603 — "Eleven-month-old boy restless in class → ADHD." Correct to "eleven-year-old." As written, the question contains a clinical impossibility. After correction, the vignette is usable at Blooms-2.
1fc42046 — Catatonia "except" question with echolalia listed as both option A and option C. Replace one duplicate with a distinct distractor (e.g., waxy flexibility or posturing) before use.
46822cbb — OCD not responding to sertraline 50 mg → next step. Revise stem to specify that 50 mg is below the therapeutic dose range for OCD (100–200 mg), or specify duration of treatment, to resolve the ambiguity between dose escalation and class switch.
905aeb01 — Early-onset Alzheimer's "except" with "None of the above" as the correct answer. Replace with a specific factually incorrect statement (e.g., "presents after age 65" or "exclusively associated with ApoE4") so the item tests actual knowledge.
78c574a3 — Korsakoff's psychosis question correctly constructed but misfiled under Neurobiology and Genetics. Reclassify to Substance Use Disorders. Also check for near-duplicate status against PYQ items 51d0d375 and 19fd06f6.
8d0d6811 — OCD treatment (ERP) question with correct answer but misfiled under Personality Disorders. Reclassify to Obsessive-Compulsive and Related Disorders only.
f7fb21b5 — La belle indifférence keyed to "dissociative disorders" broadly. Tighten the correct answer to "conversion disorder" specifically to avoid misleading learners about the dissociation/conversion distinction.
75ec5ae5 — Haloperidol → uprolling of eyes → acute dystonia. Good clinical vignette mislabeled Blooms-1. Correct Blooms tag to 3.
96403a19 — Best drug for thin-built 21-year-old with paranoid schizophrenia keyed to olanzapine. Olanzapine's metabolic profile makes it a poor first choice in this demographic. Either remove the body-habitus qualifier from the stem or change the correct answer to risperidone with an explanatory rationale.
b373076f — Long-term alcohol management with "all of the above" (acamprosate, disulfiram, naltrexone) as the correct answer. Rewrite as: "Which of the following is used for relapse prevention in alcohol dependence?" with one clearly incorrect option substituted to restore discriminatory value.
DISABLE
2abb2fc6 — Catatonia marked as "a type of schizophrenia." Factually wrong under DSM-5 and ICD-11. Will actively misinform candidates. Disable.
b6c612c8 — Down syndrome "now commonly referred to as suboptimal development." Not a recognized clinical term. Answer key is simply wrong. Disable.
674729eb — "Acid head" as slang for heavy LSD user. Pure street-slang trivia with zero clinical relevance. Disable.
c0e68e0a — Schizophrenia associated with "asthenic" personality per Kretschmer's typology. Archaic, discredited framework. Disable.
1725371d — DSM-IV-TR Axis III = medical disorders. Tests a deprecated classification system at trivial recall level. Disable.
9e90b74d — "Dementia praecox coined by whom?" Historical trivia, Blooms-1, no clinical relevance. Disable.
c740d1ad — "Who introduced psychoanalysis?" Same pattern. Disable.
cb282137 — Non-memory cognitive symptoms in Alzheimer's with "All of the above" as correct answer. Zero discriminatory value. Disable.
df9b8197 — Common treatment for unipolar depression with "All of the above" as correct answer. Disable.
8f318862 — Main symptom of Korsakoff's = confabulation. Redundant with PYQ items 19fd06f6 and 51d0d375 at lower quality. Disable.
dc42a925 — Compulsive sexual behavior most closely associated with hair pulling. Answer key appears corrupted or misassigned. Disable pending full review.
2826a727 — "Delusion of persecution = belief of being persecuted." Question and correct answer are the same phrase. Disable.
07fd4afb — "Hallucination = false perception without external stimulus." Textbook definition verbatim. Disable.
adfb3824 — Kleine-Levin syndrome → hypersomnia. Single-feature identification with implausible distractors. Disable.
274400a1 — PANDAS → Streptococcus. The acronym contains the answer. Disable; replace with a clinical vignette requiring application of PANDAS criteria.
ed0355f5 — "Elderly gentleman develops severe depression → Involutional melancholia." Obsolete nomenclature not in DSM-5 or ICD-11. Disable.