Dermatology Question Quality Review

Executive Summary

This review covers a candidate sample of 100 generic and 40 risky validated non-gold questions drawn from a Dermatology pool of 3,237 items. The benchmark and recent PYQ sets serve as the quality bar.

The most striking finding in this sample is a severe Bloom's level compression toward the bottom of the taxonomy. Of the 100 candidate generic questions, 42 are Bloom's Level 1 and 38 are Level 2, leaving only 20 questions at Levels 3–5 combined. The risky set is even more compressed. This is not merely a labelling problem: the questions themselves are structurally shallow — bare-fact recall, single-association lookups, and "which of the following is NOT" negation frames with no clinical context. The benchmark questions (e.g., 6e3ccb4a, 0be6885d) demonstrate that Dermatology is a subject that rewards clinical vignette construction, mechanistic reasoning, and management decision-making. The candidate sample largely fails to reach that bar.

Six operational problem categories are present in this sample. The most urgent are: (1) factually unsafe or wrong-key items that must be disabled immediately; (2) broken delivery items that are undeliverable without a missing image; and (3) a large volume of low-value, trivia-heavy recall questions that dilute the pool. A smaller but important set of worthwhile concepts is written so weakly that the item should be rewritten rather than discarded.

Headline numbers from the reviewed set:

Category	Approximate count in sample
Wrong Key or Factually Unsafe	6
Wrong Subject / Wrong Topic Placement	4
Broken Delivery (image-dependent, no image)	5
Low-Value But Correct	~38
Repetitive or Duplicative Coverage	~8
Worthwhile Concept, Weak Execution	~14

What Good Looks Like

The benchmark and PYQ sets establish a clear quality bar. Three features define a strong Dermatology item for Indian PG exams:

1. Clinical vignette with discriminating detail. The best items (6e3ccb4a, 0be6885d, 24dfb4ea, f06b4253) embed the testable fact inside a patient scenario that requires the candidate to synthesise age, sex, symptom duration, morphology, investigation result, and context before arriving at the answer. The vignette is not decorative — it is the mechanism by which distractors become plausible. A question like "What is the drug of choice for dermatitis herpetiformis?" (05e01e45) tests nothing that a vignette version would not test better.

2. Distractors that represent genuine clinical alternatives. In 0be6885d (lichenoid drug eruption), the four options represent four real management decisions a clinician might consider, each wrong for a specific reason. In 24dfb4ea (bullous pemphigoid), the distractors are the three most common differential diagnoses for subepidermal blistering. Contrast this with items like 32f8c329 ("Lupus vulgaris is a form of which of the following?") where three distractors — pulmonary TB, lymph-node TB, kidney TB — are not plausible answers to anyone who has opened a textbook.

3. Appropriate Bloom's level for the concept being tested. Even Bloom's Level 1 items can be well-constructed when the association is genuinely confusable (e.g., 001cc842, the Wood's lamp matching question; 5c7d6580, the scale-type matching question). The problem is not Bloom's 1 per se — it is Bloom's 1 items that test associations so unambiguous that they function as free marks.

4. Topic placement integrity. Dermatology questions should test dermatological reasoning. Items that are primarily about endocrinology, immunology, or microbiology and merely mention a skin finding as a hook belong in those subjects or require careful dual-tagging.

Main Issue Categories

1. Wrong Key or Factually Unsafe

Why this pattern is bad

A wrong key is the most serious quality failure. It actively teaches incorrect information to candidates preparing for high-stakes exams. Even a single wrong-key item in a daily plan or mock test damages platform credibility and can cause measurable harm to exam performance.

How it shows up

In this sample, wrong-key problems cluster around three sub-patterns: (a) oversimplified "most common" or "drug of choice" questions where the stated correct answer is contested or context-dependent; (b) "except" questions where the keyed answer is actually a feature of the condition; and (c) factual errors embedded in the correct option text.

Example question IDs and explanations

204ea4d5 ("What is the treatment for alopecia areata?" — keyed: Minoxidil): This is factually unsafe. Minoxidil is used for androgenetic alopecia and is at best an adjunct in alopecia areata. The first-line treatments for alopecia areata are intralesional or topical corticosteroids. Minoxidil as the sole correct answer is misleading and will confuse candidates who have studied current guidelines. Disable.
0c9448de ("Morphea most commonly occurs in which location?" — keyed: Forehead): This is factually incorrect. Morphea (localised scleroderma) most commonly affects the trunk, not the forehead. Linear morphea of the forehead is the "en coup de sabre" subtype, which is a specific variant, not the most common location overall. The keyed answer conflates a subtype with the general condition. Disable.
186abc7b ("Follman's balanitis is caused by?" — keyed: None of the above): The correct answer is intended to be "mixed anaerobic organisms / non-specific" but the option "None of the above" is a structurally poor key. More critically, the question as written implies Haemophilus ducreyi (chancroid) is a distractor, which is correct, but the "None of the above" key is unverifiable without knowing what the author intended as the true cause. This is both a wrong-key risk and a broken-stem problem. Disable.
f203ecd2 ("What is a characteristic finding in borderline leprosy?" — keyed: "Inveed Saucer appearance"): The option text contains a typographical corruption ("Inveed" instead of "Inverted"). Beyond the typo, the inverted saucer/punched-out appearance is a feature of borderline tuberculoid (BT) lesions specifically, not borderline leprosy as a whole spectrum. The stem is too broad for the answer to be unambiguous. Fix or disable — if fixed, the stem must specify BT leprosy and the typo must be corrected.
16132f51 ("Hyperpigmentation is seen in all except?" — keyed: Cushing's syndrome): This is factually unsafe. Cushing's syndrome can cause hyperpigmentation, particularly in the ectopic ACTH variant where ACTH levels are very high. The intended teaching point is that primary Cushing's (adrenal adenoma) causes hypopigmentation relative to Addison's, but the blanket statement that Cushing's does not cause hyperpigmentation is incorrect and will mislead candidates. Disable.
d55c3caa ("Cutaneous lesions may be produced by the following mycobacteria, except?" — keyed: M. intracellulare): This is factually unsafe. M. intracellulare (part of the MAC complex) does produce cutaneous lesions, particularly in immunocompromised patients. The intended answer may have been M. bovis or another species, but as written the key is wrong. Disable.

Recommended disposition: All six items above — disable immediately pending expert review.

2. Wrong Subject or Wrong Topic Placement

Why this pattern is bad

Misplaced questions contaminate subject-specific practice sets and daily plans. A candidate studying Dermatology who encounters a question that is primarily testing Endocrinology or Immunology receives a distorted signal about their Dermatology preparation. Topic misplacement also inflates apparent coverage of a topic while leaving genuine dermatological depth untested.

How it shows up

In this sample, misplacement occurs in two forms: (a) questions placed in Dermatology that are primarily testing another subject's core content, and (b) questions placed under a specific dermatology topic tag that belong under a different dermatology topic.

Example question IDs and explanations

6a4dae12 (topicName: Autoimmune Skin Diseases — "In which condition is mucocutaneous candidiasis an important clinical feature?" — keyed: Autoimmune hypoparathyroidism): This question is primarily testing Endocrinology / Immunology (APS-1 / APECED syndrome). The skin finding is a hook, not the subject of the question. The reasoning required is endocrinological. This belongs in Endocrinology or Immunology, not Autoimmune Skin Diseases. Disable from Dermatology pool; transfer to Endocrinology.
fdf43fd4 (topicName: Phototherapy and Photobiology — "A 26-year-old female with a history of extensive exposure to sun... What is the most likely diagnosis?" — keyed: Dermatomyositis): Dermatomyositis is an autoimmune connective tissue disease. The question is image-dependent (see Broken Delivery section) and the topic placement under "Phototherapy and Photobiology" is incorrect — dermatomyositis is not a phototherapy topic. It belongs under Autoimmune Skin Diseases. Fix topic tag; address image dependency separately.
71520746 (topicName: Fungal Skin Infections — "Posterior Midline Atrophic Candidiasis is best described as... Median rhomboid glossitis"): This question is primarily an Oral Medicine / Dental question. Median rhomboid glossitis as a manifestation of oral candidiasis is a topic that belongs in Oral Pathology or ENT, not Fungal Skin Infections in a Dermatology pool. Disable from Dermatology pool.
5aa2470b (topicName: Psoriasis — "Which of the following conditions is characterized by sausage digits?" — keyed: Psoriatic arthritis): Dactylitis / sausage digits is a rheumatological finding. While psoriatic arthritis is a complication of psoriasis, this question is testing Rheumatology recall, not Dermatology reasoning. It would be better placed in Rheumatology or as part of a psoriasis complications vignette. As a standalone bare-fact question in the Psoriasis topic, it adds no dermatological value. Disable or transfer.

Recommended disposition: 6a4dae12 and 71520746 — disable from Dermatology. fdf43fd4 — fix topic tag. 5aa2470b — disable or transfer to Rheumatology.

3. Broken Delivery (Missing Image, Malformed Options, Incomplete Stem)

Why this pattern is bad

Image-dependent questions without a functioning image are completely undeliverable. The candidate sees a stem that says "the image shown" or "the given picture depicts" and then has no image. This is not a quality problem that can be fixed by editing text — it requires the image asset to be verified and attached. Until that is confirmed, the item must be disabled. Delivering a broken image question in a live test is a serious user experience failure and can generate support tickets and trust issues.

How it shows up

Several items in the candidate sample have stems that explicitly reference an image ("shown below," "the given picture depicts," "characteristic appearance") but are in the generic/risky pool without confirmed image assets. Additionally, one item has a malformed option structure.

Example question IDs and explanations

ca1f3fe2 ("What is the most likely cause of these skin changes?" — keyed: Chronic venous stasis): The stem contains no clinical information whatsoever. The entire question depends on an image that is not present in the text. Without the image, all four options are equally plausible. This is completely undeliverable as text. Disable until image is confirmed attached and functional.
fdf43fd4 ("A 26-year-old female with a history of extensive exposure to sun comes to your clinic with presentation shown below"): The stem says "shown below" — image-dependent. The text alone does not provide enough discriminating information to distinguish dermatomyositis from the other options. Disable until image confirmed.
befaa146 ("The given picture depicts:" — keyed: Tinea imbricata): Entirely image-dependent. The stem is four words. Without the image, this is undeliverable. Note also the topic placement (Blistering Diseases) is wrong for Tinea imbricata, which is a fungal infection. Disable until image confirmed; fix topic tag.
ea76375f (topicName: Skin Tumors — "Identify the condition shown in the image" — keyed: Sebaceous cyst): Entirely image-dependent. The options include alopecia areata and tinea capitis, which are hair disorders, not skin tumors — suggesting the topic tag is also wrong. Disable until image confirmed; fix topic tag.
3eca31ca ("An HIV-positive patient presents with a characteristic appearance of the leg and a history of HHV-8 infection"): The phrase "characteristic appearance" implies an image was intended. The text alone (HIV + HHV-8) makes the answer trivially obvious (Kaposi sarcoma), so this may function without an image, but the stem is poorly constructed regardless. See also Low-Value section. Fix stem to remove image dependency or attach image.
186abc7b (Follman's balanitis — "None of the above" as correct key): Beyond the wrong-key concern noted above, the option structure using "None of the above" as the keyed answer is a malformed option pattern that provides no educational value and is not used in benchmark or PYQ items. Disable.

Recommended disposition: ca1f3fe2, fdf43fd4, befaa146, ea76375f — disable until image assets are confirmed. 3eca31ca — fix stem. 186abc7b — disable.

4. Low-Value But Correct (Too Simple, Low-Yield, Trivia-Heavy, Weak Exam Relevance)

Why this pattern is bad

This is the largest single problem category in the reviewed set. Low-value questions are not wrong — they are simply not worth the slot they occupy. They test single-word associations that any candidate who has read a textbook once will answer correctly. They do not discriminate between prepared and unprepared candidates, they do not build clinical reasoning, and they crowd out higher-quality items in daily plans and mock tests. The Bloom's distribution data confirms this: 42% of the candidate generic sample is Bloom's Level 1, and the risky sample is even more bottom-heavy.

The specific sub-patterns observed are:

Bare eponym lookups with no clinical context
"Drug of choice" questions with no patient scenario, no competing options, and no reasoning required
Single-feature association questions ("X is seen in Y") where the association is unambiguous and widely known
Negation ("except") questions with three obviously correct statements and one obviously incorrect one
Numerical trivia (wavelengths, percentages, mite counts) with no clinical application

How it shows up

This pattern appears throughout the generic and risky samples. The following are representative examples — the full list is considerably longer.

Example question IDs and explanations

05e01e45 ("What is the drug of choice for dermatitis herpetiformis?" — keyed: Dapsone): Bare DOC recall. No vignette, no competing clinical scenario. Bloom's 1. The benchmark already has a richer DH question embedded in a celiac disease vignette (8faf099e). Disable — concept is covered better elsewhere.
b6a84c85 ("What is the treatment of choice for Mongolian spots?" — keyed: None of the above): Trivial. Mongolian spots are benign, self-resolving birthmarks. The "correct" answer is "none" — a non-answer. No clinical reasoning is required. Disable.
2d4d1505 ("En coup de sabre is seen in:" — keyed: Scleroderma): Single eponym-to-disease lookup. Bloom's 1. No vignette. Disable — the concept is better tested within a scleroderma vignette.
f9f6a510 ("Which sign is observed on membrane removal of a psoriatic lesion?" — keyed: Auspitz sign): Bloom's 1 eponym lookup. The benchmark set already covers psoriasis histopathology at a higher level (5a7b778f). Disable.
57323a18 ("Gottron papules are pathognomonic for which condition?" — keyed: Dermatomyositis): Single-association lookup. Bloom's 1. The benchmark has a dermatomyositis vignette (6e3ccb4a) that tests the same concept at Bloom's 4. Disable.
adadd92d ("What is the wavelength of the carbon dioxide laser?" — keyed: 10600 nm): Pure numerical trivia. No clinical application. Not tested in recent PYQs. Disable.
8ecd2d30 ("What is the average number of mites found on the body in a person suffering from regular scabies?" — keyed: 10 to 15): Numerical trivia with no clinical application. Disable.
3a06bd25 ("Which 'P' describes the characteristic shape of lichen planus lesions?" — keyed: Polygonal): Bloom's 1 mnemonic recall. The "6 Ps of lichen planus" is a standard mnemonic; this tests one letter of it. Disable.
e2cdc335 ("What is the typical duration for pityriasis rosea to resolve?" — keyed: 6–12 weeks): Numerical trivia. No clinical context. Disable.
a484e333 ("Lesions of pityriasis rosea are distributed mostly on which part of the body?" — keyed: Trunk): Bloom's 1. Trivially obvious. Disable.
6ca4a7f6 ("Which syndrome is characterized by a port wine stain?" — keyed: Sturge Weber Syndrome): Single eponym lookup. Bloom's 1. Disable.
6ff07e17 ("What is the primary function of Langerhans cells?" — keyed: Antigen presentation): Basic science recall with no dermatological application. Disable.
919cc760 ("In a primary lesion of herpes, what is typically seen?" — keyed: vesicle): Bloom's 1. Trivially obvious. Disable.
7fbb01c1 ("Toxic epidermal necrolysis involves what percentage of body surface area?" — keyed: >30%): Numerical threshold recall. While the TEN/SJS distinction is clinically important, this bare-number question tests memorisation, not reasoning. A vignette asking the candidate to classify a patient's presentation would be far superior. Disable — replace with a vignette-based item.
f85bc2a6 ("Most common pattern of onychomycosis?" — keyed: Distal and lateral subungual): Bloom's 1 "most common" recall. Disable.
c89c0809 ("A 40-year-old farmer with a history of recurrent attacks of porphyria... What is the diagnosis?" — keyed: Porphyria cutanea tarda): The stem gives away the answer in the first clause ("history of recurrent attacks of porphyria"). No reasoning is required. Fix — remove the diagnosis from the stem and build a genuine clinical scenario.
b52c356c ("The Ridley-Jopling classification of leprosy is a type of?" — keyed: Clinical, bacteriological, immunological, and histological classification): Pure definitional recall. No clinical value. Disable.
16035116 ("Leprosy affects all organs except which of the following?" — keyed: Uterus): Anatomical trivia. No clinical application. Disable.
4742a460 ("What is true about Thrush?" — keyed: Responds to nystatin): Bloom's 1. The three distractors are obviously false to any medical student. Disable.
d2dac61b ("Koenen tumor is seen in which of the following conditions?" — keyed: Tuberous sclerosis): Single eponym-to-disease lookup. Bloom's 1. Disable.
32f8c329 ("Lupus vulgaris is a form of which of the following?" — keyed: Skin tuberculosis): Bloom's 1. The distractors (pulmonary TB, lymph-node TB, kidney TB) are not plausible. Disable.
eb68c7ea ("Apple jelly nodules are seen in which condition?" — keyed: Lupus vulgaris): Bloom's 1 eponym lookup. Disable.
469728b0 ("Ritter's disease is caused by which bacterium?" — keyed: Staphylococcus aureus): Bloom's 1. While the concept (SSSS) is clinically relevant, the bare eponym-to-organism format adds no value. Disable — replace with a neonatal SSSS vignette.
06988036 ("Which peripheral nerve is most commonly affected in leprosy?" — keyed: Ulnar nerve): Bloom's 1. Widely known. Disable.
fd819d7e ("What is the treatment for lepra reaction with acute neuritis?" — keyed: Prednisolone): Bloom's 1 DOC recall. The PYQ set already has a richer lepra reaction question in pregnancy (f06b4253). Disable.

Recommended disposition: All items listed above — disable. The concepts are either covered by better gold-standard items or are too trivial to merit a slot in the pool.

5. Repetitive or Duplicative Coverage

Why this pattern is bad

When multiple items in the pool test the same narrow fact at the same cognitive level, they crowd each other out in daily plans and mock tests, creating the illusion of broad coverage while actually drilling the same point repeatedly. Repetition is particularly wasteful when the repeated item is low-value (Bloom's 1) and a higher-quality version of the same concept already exists in the gold-standard set.

How it shows up

In this sample, repetition is observed across three concept clusters: pityriasis rosea distribution/duration, leprosy classification/nerve involvement, and blistering disease differentiation.

Example question IDs and explanations

Pityriasis rosea cluster: e2cdc335 (duration), a484e333 (distribution), b115b347 (clinical description), 3583c564 (clinical description with collarette scales). All four test overlapping facts about pityriasis rosea at Bloom's 1–2. The PYQ set already has a pityriasis rosea item embedded in a clinical scenario. Three of these four should be disabled; at most one clinical-vignette version should be retained.
Leprosy bare-fact cluster: 06988036 (most common nerve), 16035116 (organs spared), b52c356c (Ridley-Jopling classification), c07dd03e (innumerable lesions in lepromatous leprosy), 32f8c329 (lupus vulgaris = skin TB — different disease but same low-value pattern). These five items collectively test leprosy at Bloom's 1 with no clinical context. The PYQ set has a high-quality leprosy management question (f06b4253). The bare-fact items should be disabled and replaced with vignette-based items covering leprosy classification, reactions, and management.
Blistering disease differentiation cluster: eba3e9eb (bullous pemphigoid vignette — reasonable), 2ae1fd63 (pruritus in BP — bare fact), 4638b78a (PV hypersensitivity type — bare fact), 6e9c24bf (immunologically mediated blistering diseases except — bare fact). The PYQ set already has two strong blistering disease items (ee9f8201, 24dfb4ea). The bare-fact items 2ae1fd63, 4638b78a, and 6e9c24bf are redundant and low-value. Disable the three bare-fact items; retain eba3e9eb as a reasonable vignette.

Recommended disposition: Disable the redundant low-value items in each cluster. Retain at most one item per concept cluster, preferring the highest Bloom's level version.

6. Worthwhile Concept, Weak Execution (Keep the Concept, Fix the Stem/Options/Vignette)

Why this pattern is bad

Some items in this sample test genuinely high-yield concepts that appear in PYQs and benchmark tests, but the execution is weak enough that the item underperforms its potential. The concept should not be abandoned — it should be rewritten. Weak execution manifests as: a vignette that gives away the answer in the stem, distractors that are not plausible alternatives, a stem that is ambiguous about what is being asked, or a clinical scenario that is internally inconsistent.

How it shows up

This pattern is observed in approximately 14 items in the reviewed set. The following are the most actionable examples.

Example question IDs and explanations

6db3fdb5 ("Benzoyl peroxide acts in acne vulgaris by:" — keyed: Decreasing bacterial count): The concept (mechanism of BPO) is high-yield and appears in PYQs. However, the correct answer is incomplete — BPO works by releasing free oxygen radicals that kill P. acnes, and it also has mild keratolytic and comedolytic effects. The distractor "Acts as a keratolytic agent" is partially true, making this item ambiguous. The stem should be rewritten to ask specifically about the primary mechanism, and the distractors should be made mutually exclusive. Fix.
821bff99 ("A young person presents with comedones and papulo-pustular acne affecting the face, trunk, and back. What is the recommended management approach?" — keyed: Oral doxycycline plus topical retinoic acid): The concept (moderate-to-severe acne management) is high-yield. The vignette is reasonable but thin — no severity grading, no prior treatment history, no contraindications mentioned. The correct answer is defensible but the stem should specify severity (moderate-to-severe) and the distractors should include isotretinoin as a plausible option for severe disease. Fix — add severity context and improve distractor set.
bb2f369a (Hidradenitis suppurativa vignette — keyed: Hidradenitis suppurativa): The clinical scenario is well-constructed (obese, smoker, axillary subcutaneous nodules with scarring). The concept is high-yield. However, the distractors are weak — lipodystrophy and xeroderma pigmentosum are not plausible differentials for this presentation. Replace with pyoderma gangrenosum (already present), infected epidermoid cyst, and nodular acne inversa. Fix distractors.
4c69002f (Epidermolysis bullosa — gene mutation — keyed: Collagen type 7): The concept is high-yield and appears in PYQs. The stem is a bare-fact question with no clinical context. This should be rewritten as a vignette: a neonate with blistering at trauma sites, subepidermal split on histology, asking for the gene mutation. The current format tests memorisation; a vignette would test application. Fix — add clinical vignette.
1cf5eb6a (Fixed drug eruption on penis — keyed: Aceclofenac): The concept (most common drug causing FDE on genitalia) is high-yield. However, the correct answer is debatable — NSAIDs as a class (particularly cotrimoxazole, tetracyclines, and NSAIDs) are all common causes of FDE. Aceclofenac specifically as the most common cause of genital FDE is not universally agreed upon in standard references. The stem should be rewritten to present a more specific clinical scenario (e.g., recurrent erythematous patch at the same site after taking a specific drug) and the key should be verified against current Indian dermatology textbooks. Fix with expert verification.
5f8bcb56 (Photoallergic contact dermatitis from chromium in a construction worker — keyed: Photoallergic contact dermatitis from chromium): The concept is genuinely high-yield and the clinical scenario is well-constructed. The weakness is in the distractors — "Berloque dermatitis" is a phototoxic reaction from bergapten in perfumes, not a plausible differential for a construction worker handling cement. Replace with a more clinically relevant distractor (e.g., irritant contact dermatitis from cement alkali, or allergic contact dermatitis without photosensitisation). Fix distractors.
86085361 (Occupational hand dermatitis prevention — Bloom's 5 synthesis question — keyed: Implement A, B, and C): This is the only Bloom's 5 item in the entire candidate sample and the concept is genuinely important. However, the question is written as a policy synthesis exercise that reads more like a public health question than a clinical dermatology question. The stem is also somewhat leading — option D ("reducing glove use frequency") is obviously wrong from an infection control standpoint, making the synthesis less demanding than it appears. Fix — reframe as a clinical scenario (e.g., a dermatology registrar advising an occupational health team) and make option D less obviously wrong.
9325ec63 ("Nail deformities commonly seen in lichen planus include all of the following EXCEPT:" — keyed: Beau's lines): The concept is high-yield. However, the "all except" format is weak. Beau's lines are transverse grooves caused by any systemic illness that temporarily arrests nail growth — they are not specifically associated with lichen planus, making the key defensible. But the question would be stronger as a positive vignette: "A patient with lichen planus develops a specific nail change — which of the following is most characteristic?" Fix — convert from negative to positive format.
c2394d5a (Tinea corporis vignette — keyed: Tinea corporis): The vignette is reasonable but the KOH positivity is stated explicitly, making the answer obvious. A stronger version would present the clinical picture and ask for the most appropriate investigation, or present the KOH result and ask for the causative organism. Fix — increase cognitive demand.
3e657964 (Reverse Koebner's phenomenon in granuloma annulare — keyed: Reverse Koebner's phenomenon): The concept is genuinely high-yield and rarely tested. The vignette (diabetic patient, annular orange lesions disappearing after biopsy) is well-constructed. The main weakness is that "annular orange skin lesions" is an unusual description for granuloma annulare (which is typically flesh-coloured to erythematous). The stem should be reviewed for accuracy of the clinical description. Fix — verify clinical description.

Recommended disposition: All items in this category — fix. Do not disable; the underlying concepts are worth testing. Prioritise 6db3fdb5, 821bff99, 4c69002f, and 1cf5eb6a as the highest-yield fixes.

Prioritization

The following table summarises recommended actions by urgency.

Priority	Action	Items	Rationale
P1 — Immediate	Disable (wrong key / factually unsafe)	204ea4d5, 0c9448de, d55c3caa, 16132f51, 186abc7b, 0c9448de	Active misinformation risk
P1 — Immediate	Disable (broken delivery, no image)	ca1f3fe2, fdf43fd4, befaa146, ea76375f	Undeliverable in live tests
P2 — This sprint	Disable (wrong subject / misplaced)	6a4dae12, 71520746, 5aa2470b	Subject contamination
P2 — This sprint	Disable (low-value, concept covered by gold)	05e01e45, b6a84c85, 2d4d1505, f9f6a510, 57323a18, adadd92d, 8ecd2d30, 3a06bd25, e2cdc335, a484e333, 6ca4a7f6, 6ff07e17, 919cc760, 7fbb01c1, f85bc2a6, b52c356c, 16035116, 4742a460, d2dac61b, 32f8c329, eb68c7ea, 469728b0, 06988036, fd819d7e, 2ae1fd63, 4638b78a, 6e9c24bf	Pool dilution
P3 — Next sprint	Fix (worthwhile concept, weak execution)	6db3fdb5, 821bff99, bb2f369a, 4c69002f, 1cf5eb6a, 5f8bcb56, 86085361, 9325ec63, c2394d5a, 3e657964	Salvageable with rewrite
P3 — Next sprint	Fix topic tag	fdf43fd4 (→ Autoimmune), befaa146 (→ Fungal), ea76375f (→ Hair/Nail or Pediatric)	Metadata integrity
P4 — Backlog	Review for duplication	Pityriasis rosea cluster, leprosy bare-fact cluster	Pool hygiene

Example Keep / Fix / Disable Calls

The following are concrete, item-level disposition calls with brief justifications.

KEEP

821017f7 (BCC vignette — medial canthus, rolled-out margins, peripheral palisading): Well-constructed clinical vignette with histopathological correlation. Bloom's 4. Distractors are genuine clinical alternatives. Difficulty 3. Meets benchmark standard. Keep.
eba3e9eb (Bullous pemphigoid vignette — tense blisters, negative Nikolsky, no mucosal involvement): Reasonable clinical vignette. Bloom's 3. Distractors are the correct differential diagnoses. Keep.
8faf099e (Dermatitis herpetiformis vignette — celiac disease history, grouped vesicles): Good clinical context. Bloom's 4. Distractors are plausible. Keep.
4eaa76dc (Infant scabies — itchy exudative lesions, palms/soles, siblings affected, treatment): Good clinical vignette. Bloom's 3. Correct answer (topical permethrin) is appropriate for age. Keep.
bb2f369a (Hidradenitis suppurativa vignette): Concept is high-yield; vignette is clinically sound. Keep with distractor fix (see Fix section above).
846738c6 (Actinic lichen planus — false statement question): Bloom's 4. The false statement (severe pruritus) is a genuine teaching point — actinic LP is characteristically less pruritic than classic LP. Distractors are all true statements. Keep.
06b9387b (Kerion — painful boggy scalp swelling, KOH mount as investigation): Bloom's 3. Good clinical scenario. Correct investigation identified. Keep.

FIX

f203ecd2 (Borderline leprosy — "Inveed Saucer appearance"): Concept is high-yield. Fix the typographical error ("Inverted" not "Inveed"), narrow the stem to borderline tuberculoid leprosy specifically, and verify the clinical accuracy of the other distractors. Fix.
6db3fdb5 (Benzoyl peroxide mechanism): High-yield concept. Rewrite to make the correct answer unambiguous (primary mechanism = bactericidal via free radical release) and ensure distractors are mutually exclusive. Fix.
4c69002f (Epidermolysis bullosa gene mutation): High-yield concept. Add a neonatal clinical vignette. The bare-fact format is salvageable with a two-sentence stem. Fix.
c89c0809 (Porphyria cutanea tarda): The concept is high-yield. Remove the diagnosis from the stem ("history of recurrent attacks of porphyria" gives it away). Rewrite as a farmer presenting with photosensitivity, blistering on sun-exposed areas, and hypertrichosis, asking for the diagnosis. Fix.
3e657964 (Reverse Koebner's phenomenon): High-yield, rarely tested concept. Verify the clinical description of granuloma annulare (colour, morphology) and correct if needed. Fix.

DISABLE

204ea4d5 (Alopecia areata — Minoxidil as treatment): Wrong key. Minoxidil is not first-line for alopecia areata. Disable.
0c9448de (Morphea — forehead as most common location): Wrong key. Trunk is most common. Disable.
d55c3caa (M. intracellulare does not cause cutaneous lesions): Wrong key. It does in immunocompromised patients. Disable.
16132f51 (Cushing's syndrome does not cause hyperpigmentation): Factually unsafe. Ectopic ACTH Cushing's does cause hyperpigmentation. Disable.
ca1f3fe2 ("What is the most likely cause of these skin changes?"): No stem, no image, completely undeliverable. Disable.
05e01e45 (DOC for dermatitis herpetiformis — bare fact): Concept covered by better items. Bloom's 1. No clinical context. Disable.
adadd92d (CO₂ laser wavelength — 10600 nm): Pure numerical trivia. No clinical application. Not in recent PYQs. Disable.
6a4dae12 (Mucocutaneous candidiasis in autoimmune hypoparathyroidism): Wrong subject. Primarily an Endocrinology question. Disable from Dermatology pool.
b6a84c85 (Treatment of Mongolian spots — None of the above): Trivial. Non-answer key. No educational value. Disable.
8ecd2d30 (Average mite count in scabies — 10 to 15): Numerical trivia. No clinical application. Disable.