Every time you prescribe an antibiotic, you face two competing risks:

1. Under-treatment: If you don't cover the likely pathogen, the patient may deteriorate or die (especially in sepsis, neutropenic fever, meningitis).
2. Over-treatment: If you prescribe unnecessarily or use an overly broad agent, you drive antimicrobial resistance (AMR), expose the patient to adverse effects, increase cost, and disrupt normal flora (→ C. difficile, fungal superinfection).

Empirical therapy = choosing an antibiotic BEFORE culture results are available, based on the clinical syndrome, likely pathogens, local resistance patterns, and patient factors.

The art is in balancing coverage with stewardship. This lecture teaches you the framework for that balance.

The slide's three double-headed arrows are the organising visual for the lecture:

• Patient ↔ bacteria: virulence and inoculum interact with host defence.
• Bacteria ↔ antibiotic: susceptibility and resistance determine whether the drug is active.
• Patient ↔ antibiotic: allergy, toxicity, PK/PD, organ function, interactions, pregnancy and tissue penetration determine whether an active drug can be used safely and reach the site.

Thus, “susceptible” is not the same as “appropriate,” and patient vulnerability should trigger closer assessment—not antibiotics for a clearly viral syndrome.

• Culture and sensitivity results take 48–72 hours (at minimum).
• Many infections (e.g., community-acquired pneumonia, cellulitis) are treated entirely empirically — you may never identify the organism.
• The decision to start empirical antibiotics must be made at the bedside, immediately. You cannot wait.
• Once culture results return, you should de-escalate (narrow the spectrum) or escalate (broaden if resistant organism found). This is called "streamlining".

The clinical syndrome tells you the site of infection, which in turn predicts the likely pathogens.

These modify your antibiotic choice significantly:

Ancillary legacy question point (not a numbered question in the 37-slide deck): nitrofurantoin can cause haemolytic anaemia in G6PD deficiency.

• The lecture directs students to the IMPACT (Interhospital Multi-disciplinary Programme on Antimicrobial ChemoTherapy) data for Hong Kong resistance patterns ^[1].
• You must know which organisms are commonly resistant in your local setting. For example:
  • MRSA prevalence in HK hospitals → affects choice for skin/soft tissue infection, line infections
  • ESBL-producing E. coli/Klebsiella → amoxicillin-clavulanate may fail; may need carbapenem
  • Fluoroquinolone-resistant E. coli → no longer first-line for UTI in many settings

• IV for severe infections, sepsis, inability to take orally, poor oral bioavailability
• Oral for mild-moderate infections, step-down from IV when improving
• IV-to-oral switch: Important stewardship intervention. Switch when:
  • Afebrile for 24–48 hours
  • Clinically improving
  • Able to tolerate oral intake
  • An oral option with adequate bioavailability, penetration and spectrum exists. Do not treat “endocarditis, meningitis, osteomyelitis” as permanent IV-only categories; route decisions are diagnosis-, organism-, regimen- and phase-specific.

The pyramid is explicitly marked “an illustration only.” It places cefuroxime/amoxicillin-clavulanate/fluoroquinolones at the base, piperacillin-tazobactam above, and carbapenems at the apex, with arrows for both escalation and de-escalation. It is not a fixed “high risk = carbapenem” protocol. Assess syndrome, severity, source, prior cultures and antibiotic exposure; obtain cultures without delaying urgent therapy; then stop, narrow, target, switch route, or broaden at 48–72 hours as the evidence dictates. Current Hong Kong IMPACT guidance supersedes the brand-name diagram (IMPACT empirical therapy).

The respiratory anatomy sequence localises uncomplicated URTI to the nose/nasopharynx/larynx, tracheobronchitis to the conducting airways, and pneumonia to the alveoli. The pneumonia slide also shows a pulse oximeter: oxygen saturation belongs with temperature, heart rate, respiratory rate and focal chest findings when deciding whether this is uncomplicated bronchitis or possible pneumonia. In otherwise healthy adults, pneumonia is uncommon without abnormal vital signs or focal consolidation findings (CDC adult outpatient guidance). Pneumonia can be bacterial, viral, fungal, or mixed; give antibiotics when bacterial pneumonia is suspected, not for the word “pneumonia” automatically.

The exact Hong Kong IMPACT slide for non-hospitalised CAP recommends oral amoxicillin-clavulanate or high-dose amoxicillin, with doxycycline when atypical coverage is indicated; levofloxacin is an alternative for penicillin allergy. It specifically warns not to use a macrolide/azalide, tetracycline, or co-trimoxazole alone for empirical CAP in Hong Kong. Follow the live regimen and patient-specific assessment (IMPACT CAP guidance).

Mycoplasma pneumoniae

Ancillary legacy data (not graphed in the exact slide deck): a cited local study reported peak M. pneumoniae incidence in school-age children. Treat those percentages as study-specific historical data, not a current individual probability.

This table is an orientation to pathogen–drug matching, not a complete empirical prescribing protocol. Age, severity, recent microbiology and exposure, local resistance, contraindications, dose and care setting can change the regimen.

The slide reproduces the CDC 2021 summary:

• uncomplicated gonorrhoea: ceftriaxone 500 mg IM once if body weight is < 150 kg, or 1 g if ≥150 kg;
• chlamydia: doxycycline 100 mg orally twice daily for 7 days; and
• when treating gonorrhoea, add chlamydia treatment only if chlamydia has not been excluded—not automatically after a negative result.

Use current local sexual-health guidance for the individual patient and site of infection (CDC STI treatment guideline).

From related GC slide content (GC 071 "It is red and painful") ^[4]:

From GC 098 (Antibiotic Prophylaxis) and CFB WCS27 (Surgical Infection) ^[5][6]:

• Surgical prophylaxis: most agents are started within 60 minutes before incision; agents requiring a longer infusion (for example vancomycin) are started earlier, not squeezed into the final 30 minutes
• Choice depends on the site: Cefazolin for most clean procedures; cefuroxime + metronidazole for colorectal/abdominal
• Duration: Usually single dose; re-dose if surgery > 2 half-lives of antibiotic or significant blood loss; stop within 24 hours post-op
• Prophylaxis ≠ treatment: If infection is already present, you are treating, not prophylaxing

1. Delayed/watchful prescribing: Give the patient a prescription but advise them to fill it only if symptoms worsen or don't improve in 48–72 hours. Works well for otitis media, mild sinusitis.

2. Use clinical prediction rules: The Centor/McIsaac score for pharyngitis. CURB-65 for pneumonia severity (guides whether outpatient oral or hospital IV antibiotics are needed).

3. Patient education: Explain that most URTIs are viral, self-limiting, and antibiotics won't help. Reassure about expected duration (common cold: 7–10 days; cough can last 2–3 weeks). Use phrases like "This is a viral infection. Antibiotics won't make you better faster and could cause side effects."

4. Point-of-care testing: Rapid strep test, CRP testing, procalcitonin — help distinguish viral from bacterial and guide antibiotic decisions.

The CDC materials referenced in the lecture emphasize evidence-based guidelines for when antibiotics are appropriate for: acute rhinosinusitis, acute uncomplicated bronchitis, common cold, and pharyngitis — in both adults and children. ^[1]

The otoscopy sequence supplies the distinction missing from a text-only summary:

Erythema alone is nonspecific. AOM is supported particularly by moderate/severe TM bulging or new otorrhoea not caused by otitis externa. Selected children with mild AOM and reliable follow-up can be observed for 2–3 days while pain is treated; CDC notes that about two-thirds of mild cases improve without antibiotics (CDC watchful waiting).

Slide 35's 70% (2 weeks), 40% (1 month), 20% (2 months), and 10% (3 months) residual-effusion figures come from a 1980 cohort. They illustrate that MEE after AOM is common and not an automatic treatment failure, but they are historical population data. Likewise, the slide's “80–90% resolve” statement overstates what its cited Cochrane review found: about 60% were better at 24 hours regardless of antibiotics, with no antibiotic pain benefit at that time and only a modest later benefit.

• Drug allergies — type of reaction (rash vs. anaphylaxis); timing
• Previous antibiotic use — within last 3 months → risk of resistance
• Travel history — MDR organisms, tropical infections
• Recent hospitalization — hospital-acquired pathogens
• Immunosuppression — HIV, chemotherapy, biologics, transplant
• G6PD status — especially in HK (high prevalence in Southern Chinese)
• Pregnancy/breastfeeding
• Baseline renal and hepatic function
• Indwelling devices — catheters, prosthetic valves, joints → biofilm infections

1. Blood cultures — when bacteraemia/sepsis or another guideline-defined indication is suspected; fever alone is not enough
2. Urine culture — for pyelonephritis, complicated/recurrent infection, pregnancy, treatment failure, or another local indication; not mandatory for every classic uncomplicated cystitis
3. Respiratory specimen — for selected pneumonia according to severity and guideline, not simply every productive cough
4. Pus/tissue or aspirate — prefer a representative specimen over a superficial wound swab when drainage/source control is performed
5. CSF — when meningitis is suspected; imaging before lumbar puncture is only for defined indications
6. Other site-specific — ascitic fluid (SBP), joint aspirate (septic arthritis), peritoneal fluid

• Definition: Neutrophils < 0.5 × 10⁹/L + single temperature ≥ 38.3°C or sustained ≥ 38.0°C for 1 hour
• Medical emergency: Start empirical broad-spectrum antibiotics within 1 hour
• First-line: Piperacillin-tazobactam or meropenem (must cover Pseudomonas)
• Add vancomycin if: catheter-related, skin/soft tissue, MRSA risk, hemodynamic instability
• Add antifungal if: persistent fever > 4–7 days despite antibiotics ^[10]

Mycoplasma pneumoniae peaks in school-age children (2–11 years) — this is a local HK finding. ^[1]

• Acute otitis media: Observation option in children > 2 years with mild unilateral disease; if treating → amoxicillin first-line
• Neonatal sepsis: Ampicillin + gentamicin (covers GBS, E. coli, Listeria)
• Dose calculation: Always weight-based (mg/kg) in children

1. "A 25-year-old with sore throat, cough, runny nose, fever 38°C — calculate the Centor score and advise on management." → Score = 1 → No antibiotics.

2. "A 60-year-old with COPD presents with productive cough, fever, CXR shows consolidation. Name the most likely organism and empirical antibiotic." → S. pneumoniae (consider H. influenzae); amoxicillin-clavulanate or ceftriaxone + macrolide.

3. "A young woman with uncomplicated cystitis and G6PD deficiency. Which commonly used UTI antibiotic is contraindicated?" → Nitrofurantoin.

4. "Can empirical PPI–amoxicillin–clarithromycin triple therapy be assumed for H. pylori?" → No. Use a current susceptibility-aware local regimen and confirm eradication.

5. "A 7-year-old with community-acquired pneumonia. What organism has the highest incidence in this age group in Hong Kong?" → Mycoplasma pneumoniae.

6. "When are antibiotics indicated for acute rhinosinusitis?" → Persistent > 10 days, double-sickening, severe (high fever + purulent discharge ≥ 3 days).

7. "Name 3 characteristic features of Scarlet fever." → Strawberry tongue, sandpaper rash, Pastia's lines.

8. "What are the indications for IE prophylaxis before dental procedures?" → Prosthetic valve, previous IE, cyanotic CHD, repaired CHD with prosthetic material.