The antibiotic decision in an exacerbation runs on three cardinal symptoms — increased dyspnea, increased sputum volume and increased sputum purulence — and he has one. The Winnipeg (Anthonisen) criteria call for an antibiotic when two of the three are present, or when the patient needs ventilatory support. Neither applies. Everything he does need is already running: controlled oxygen to 88–92% pending the gas, short-acting bronchodilators, and prednisone 40 mg daily for five days with no taper.

A is the reflex prescription, and the drug itself is not wrong — amoxicillin-clavulanate is first line for a high-risk patient — but the question is whether he needs an antibiotic at all, and the criteria answer it. D merges two different indications: azithromycin 250 mg daily is prophylaxis for a patient who keeps exacerbating despite optimal inhaled therapy, it requires a baseline ECG and audiogram for QT prolongation and hearing loss, and it is not treatment for the exacerbation in front of you. B reasons from how unwell he looks rather than from the criteria — hypoxemia and admission are not the trigger, ventilatory support is, and a respiratory fluoroquinolone carries the harms set out in the Canadian monograph.

Takeaway: Count the three cardinal symptoms before writing an antibiotic in COPD.

Aspiration pneumonia is treated as community-acquired pneumonia. ATS/IDSA 2019 recommends against routinely adding anaerobic coverage unless empyema or lung abscess is present, and the 2023 CMAJ review says the same: anaerobes are not the primary pathogens. She has none of the history that broadens cover — no recent hospitalization with intravenous antibiotics, no structural lung disease, no prior MRSA.

B is the answer most learners give, and it is precisely the teaching those two sources correct: metronidazole or clindamycin is reserved for empyema, abscess or necrotizing disease, not for aspiration as a category. C broadens by mechanism rather than by history — piperacillin-tazobactam belongs to advanced structural lung disease or prior Pseudomonas, neither of which she has. D names a real entity and applies it to the wrong patient: aspiration pneumonitis is a chemical injury that begins within minutes to hours and settles in 24–48 hours, whereas she has three days of fever with a new consolidation.

What the aspiration does change is the prevention plan. Ask speech-language pathology for a bedside swallowing assessment before oral intake resumes, and add positioning, texture modification and oral care. A failed assessment is not an automatic route to a feeding tube — tube feeding does not reduce aspiration or pneumonia rates.

Takeaway: Aspiration changes the prevention plan, not the antibiotic.

The pulmonary function report does not support the label she is carrying. A preserved FEV1/FVC ratio with a reduced FVC and total lung capacity is restriction, not obstruction, and a DLCO of 42% with desaturation on walking and a normal resting saturation is the physiology of interstitial disease. Fine Velcro crackles that do not clear with diuresis, and clubbing, complete it. HRCT is the test, and it goes without contrast — contrast adds nothing to the interstitium and obscures subtle ground-glass, while the protocol’s expiratory and prone sequences are what separate early basal fibrosis from dependent atelectasis.

A is the right scan for a different question, and it is a contrast study that would read the interstitium poorly; nothing here suggests an embolus. C repeats a trial that has already failed — crackles that persist through diuresis are the finding that argues against heart failure, not for more of it. D is the most tempting answer because it continues the existing plan, and it is the error this section names: a patient labelled COPD who is restrictive on formal testing has been mislabelled, and a third inhaler does not fix a diagnosis. A normal chest radiograph does not exclude ILD.

Takeaway: Read the ratio before accepting the label — restriction with a low DLCO is not COPD.

The systemic corticosteroid is the drug that alters the course of an exacerbation, and it belongs in the first hour — she is 55 minutes in without one. Prednisone 40–50 mg PO daily for 5–7 days, no taper.

A is the right drug at the wrong moment: magnesium 2 g IV over 20 minutes is for a severe attack not responding to initial treatment, and she is responding to the bronchodilators she has been given while the treatment that matters has not started. B is the commonest upgrade error — oral prednisone is as effective as intravenous, and the intravenous route is reserved for a patient who cannot swallow or absorb, which she can. C is the sequence this question exists to interrupt: the steroid is not what you reach for when the bronchodilators have failed, it is what you give alongside them so that they do not.

Takeaway: The steroid goes in within the first hour, not when the bronchodilators fail.

Two decisions follow the diagnosis: which drug, and for how long. CARAVAGGIO (NEJM 2020) found apixaban non-inferior to dalteparin in cancer-associated VTE without excess major bleeding, and ADAM-VTE found significantly less major bleeding — so an oral agent is reasonable here, with the caveat about gastrointestinal and genitourinary tumours, which she does not have. Duration follows the provocation, and active cancer is a persistent provocation: anticoagulation continues while the cancer is active rather than stopping on a date.

B was the standard before CARAVAGGIO and is still a reasonable choice in the gastrointestinal and genitourinary tumours where DOAC bleeding is higher — but “required in all” is no longer true, and daily injections are not a neutral cost in palliative treatment. C gets the drug right and the duration wrong; three months is the answer for a transient provocation that has resolved, and hers has not. D is effective anticoagulation, but warfarin’s remaining place is triple-positive antiphospholipid syndrome, mechanical valves and severe renal impairment, and INR control during chemotherapy is difficult.

Takeaway: Name the provocation and it names the duration — active cancer never leaves.

A pH below 7.35 with hypercapnia is the threshold, and he is well past it — pH 7.24 with a pCO2 of 74. Non-invasive ventilation is the treatment, and its value is that it is started before exhaustion, not after. Drowsiness and a respiratory rate that is falling while the pCO2 climbs are signs of tiring, not of improvement.

A treats the number that looks most abnormal and makes the real problem worse: the target in COPD stays 88–92%, and pushing the saturation to 94–98% in a patient who is retaining CO2 worsens hypercapnia. D is not wrong about where this may end, and it is the reason critical care is already involved — but NIV is what is offered first in a hypercapnic exacerbation, and it is the intervention most likely to make the tube unnecessary. B is the modern-sounding wrong answer and the one worth being able to argue against at the bedside: high-flow nasal oxygen delivers heated humidified gas at high flow and improves oxygenation, but it does not provide the inspiratory pressure support that unloads the respiratory muscles and clears CO2. His problem is not oxygenation — he is 90% on 28% oxygen — it is ventilation. High-flow has been studied in hypercapnic exacerbations and may have a place in milder cases, but at a pH of 7.24 the guidelines and the trial evidence point to NIV.

Takeaway: A pH below 7.35 with a rising pCO2 is a call for ventilation, not another round of nebulizers.

By Light’s criteria this is an exudate, and on all three counts: the protein ratio is 35 ÷ 68 = 0.51, the LDH ratio 200 ÷ 260 = 0.77, and the fluid LDH of 200 exceeds two-thirds of the upper limit of normal serum LDH (167). Any one of the three is enough. It is still a transudate. Five days of diuresis removes water faster than protein and concentrates what is left behind, so a cardiac effusion drifts across the thresholds while its cause never changes — in Romero-Candeira’s serial taps, pleural protein rose from 23 to 33 g/L and LDH from 177 to 288 U/L over five days of treatment.

The check is a gradient, which is a subtraction rather than a ratio and is therefore far less disturbed by concentration. Serum albumin minus pleural albumin is 34 − 21 = 13 g/L, above 12, which means transudate; the serum-to-fluid protein gradient agrees at 68 − 35 = 33 g/L, above 31.

A is the cascade a pseudoexudate causes when the ratios are taken at face value — a CT, a repeat tap and a respirology consult for an effusion that was always cardiac. C picks up the label the fluid actively excludes: the pH is 7.44, the glucose normal and the Gram stain negative, and she has neither fever nor pneumonia. D is the cautious-sounding option and the least defensible of the four — repeating the tap cannot resolve the ambiguity, because the numbers will drift further in the same direction as long as she is being diuresed. The gradients are already on the chart.

Takeaway: When the ratios say exudate and the patient says heart failure, calculate the gradient.

The answer is in the pharmacy record. Two puffs twice daily of a 120-dose inhaler is roughly one device a month, so three in a year is about three months of controller therapy — and seven reliever inhalers meets the CTS definition of SABA overuse at more than two a year, a risk factor both for severe exacerbation and for asthma-related death. Her asthma has not failed medium-dose ICS-formoterol, because she has not taken it.

That is the distinction between difficult asthma and severe asthma, and it has to be settled before the label is applied: adherence, inhaler technique, comorbidities such as allergic rhinitis, chronic rhinosinusitis with nasal polyps, reflux, obesity and obstructive sleep apnea, inducible laryngeal obstruction, and continuing exposure at home or at work.

A is the referral this question exists to interrupt, and the eosinophil count is what makes it tempting — but biologics are specialist-initiated and phenotype-directed, and 0.42 does not convert uncontrolled asthma into severe asthma. B is the standard next rung of the ladder and would be right in a patient taking her controller; here it increases a dose that is not being delivered. C adds a fourth device to a regimen already too complicated to be followed. Her body mass index of 31 is worth pursuing in its own right, as one of the comorbidities that makes asthma difficult.

Takeaway: Count the inhalers in the pharmacy record before adding a fourth drug.

Fever, diarrhea, hyponatremia, transaminitis and confusion after freshwater exposure is the classic cluster — suggestive, though not discriminating on its own. Note first how sick he is: confusion, a respiratory rate of 30 and multilobar consolidation are three ATS/IDSA minor criteria, which makes this severe pneumonia. That matters twice over. It is the severity band in which Legionella is most over-represented, and it is why he should have had both urine antigens and a respiratory culture on admission — and never a beta-lactam alone.

What settles the management is the blind spot in the test: the urine antigen detects Legionella pneumophila serogroup 1 only, roughly 86% sensitive for that serogroup and 74–79% across all, so a negative result does not exclude the diagnosis. And a beta-lactam does not cover Legionella at all — the likeliest reason he has not improved. Start levofloxacin 750 mg daily, which is the usual choice in severe or immunocompromised disease, and send sputum culture or PCR before the first dose. Legionnaires’ disease is reportable to public health.

A is the error this question targets, and it is a reasonable-sounding one: a named microbiological test came back negative. It only ever answered a narrower question than the one being asked. C repeats a test whose sensitivity does not improve with time, and delays therapy in a patient who is deteriorating. D broadens the spectrum without covering the organism actually suspected — piperacillin-tazobactam and vancomycin add nothing against Legionella, and his pneumonia was community-onset, present on admission.

Takeaway: A negative Legionella urine antigen rules out one serogroup, not the disease.

Pregnancy invalidates the usual shortcuts, which is why it has its own algorithm. The pregnancy-adapted YEARS pathway asks three questions — clinical signs of deep vein thrombosis, hemoptysis, and whether pulmonary embolism is the most likely diagnosis — and she has none of them. With none of the three, the D-dimer threshold is 1000 µg/L, and hers is 1600. That does not diagnose an embolus; it means she cannot be excluded without imaging. Both CTPA and ventilation-perfusion scanning deliver fetal doses well below the threshold of concern — V/Q gives less breast radiation, CTPA less to the fetus.

B is the most instructive answer here, because the strategy is correct and the patient is wrong. Compression ultrasound is the first test in the pregnant patient who has clinical signs of deep vein thrombosis: a proximal DVT is treated exactly as PE is, so finding one confirms venous thromboembolism and spares her the chest imaging entirely. She has no leg swelling and no calf tenderness, so that branch does not open — and most patients with PE have no clinical DVT at all, which is why a negative scan here would only cost time. A starts from a true statement and lands in the wrong place: the D-dimer does rise through pregnancy, which is why the algorithm supplies a pregnancy-specific threshold rather than discarding the test. D treats a diagnosis she may not have, committing her to months of injections, an altered delivery plan and a label on her chart that no one can later disprove. Radiation fear is the commonest reason this diagnosis is missed in pregnancy.

Takeaway: In pregnancy, symptoms in the leg decide whether the chest gets scanned.