I want to revise a drug per day. So, let's start with antibiotics. As, I am General practitioner, I want to revise and get updates on each drugs . What do u say?

Asking for Preferences

Which antibiotic class or specific drug would you like to start with today? · How detailed should the revision be?

Great choice to start with Penicillins - the backbone of antibiotic therapy. Let me pull in textbook content and recent literature simultaneously.

Searching sources...

Reading File

Finding Sources

Searching PubMed

penicillin amoxicillin clinical use resistance

Searching the Web

penicillin antibiotic 2025 2026 clinical updates resistance GP prescribing

Now let me pull the actual textbook content from Goodman & Gilman and Katzung - the two most relevant pharmacology references.

Reading File

Finding Sources

Reading File

I now have comprehensive information. Let me compile the full deep-dive revision.

Searching Images

penicillin beta-lactam mechanism of action bacterial cell wall

A medical pathophysiology diagram illustrating the mechanism of action of Catharanthus roseus root ethanolic extract (EECRR) in sensitizing multiple antibiotic-resistant Staphylococcus aureus (MAR-SA) to ampicillin. The flowchart transitions from left to right, starting with a bacterial colony of MAR-SA and a test tube representing the extract. It branches into two treatment pathways: 'Ampicillin + 1/2 MIC EECRR', which results in decreased cell viability, and 'Ampicillin + MIC EECRR', which leads to cell death (indicated by a red cross). The right side of the diagram provides a molecular level illustration of a MAR-SA cell. It demonstrates that the EECRR treatment inhibits penicillin-binding protein 2a (PBP2a), which normally blocks ampicillin. By suppressing PBP2a expression and causing plasmid DNA damage, the extract allows ampicillin to bind to PBP2, disrupting the bacterial cell wall integrity. Key labels include PBP2a, PBP2, ampicillin, plasmid, and DNA damage. This visual summarizes an antimicrobial research study on overcoming antibiotic resistance using botanical extracts.

A diagnostic biochemical image displaying a SDS-PAGE gel fluorescence scan from a Bocillin competition assay. The visual demonstrates the inhibition of Pseudomonas aeruginosa Penicillin-Binding Protein 3 (PBP3) by the cyclic boronate inhibitor, vaborbactam. The gel shows a series of horizontal protein bands representing PBP3 labeled with a fluorescent Bocillin reporter. From right to left, the lanes correspond to increasing concentrations of vaborbactam (0, 0.05, 0.1, 0.25, 0.5, 1.0, 1.25, 1.5, and 1.75 mM). A clear, concentration-dependent decrease in band intensity is visible; the highest fluorescence intensity is at 0 mM (absence of inhibitor), while the lowest intensity is observed at 1.75 mM. This trend illustrates competitive binding between vaborbactam and the fluorescent reporter for the PBP3 active site. The educational focus is on the biochemical validation of non-beta-lactam antibiotic inhibitors against bacterial cell wall synthesis targets.

This composite educational graphic presents a microbiology analysis of bacterial cell wall synthesis using autoradiography and grain distribution profiling. It consists of three panels (a, b, and c), each featuring an electron micrograph of Escherichia coli alongside a corresponding quantitative graph. Panels (a) and (b) demonstrate filamentous bacteria exhibiting abnormal cell morphology with multiple visible constrictions. The associated line graphs, which plot 'Grains / µm cell length' against 'Normalized cell length', show two distinct peaks labeled 'PIPS' (Penicillin-Insensitive Peptidoglycan Synthesis). These peaks correlate spatially with the cellular constrictions, indicating localized peptidoglycan incorporation sites when Penicillin-Binding Protein 3 (PBP3) is inhibited or non-functional. Panel (c) displays normally dividing bacterial cells with shorter, rod-shaped morphology. The corresponding graph shows a single, sharp central peak at a normalized cell length of 0.5, representing focused peptidoglycan synthesis at the standard mid-cell division site. This comparison highlights the topography of cell envelope growth and the distinction between normal septal synthesis and penicillin-insensitive mechanisms in cell division mutants.

Here is your full GP-level deep dive on Penicillins - Day 1 of your antibiotic revision series.

💊 Day 1 - Penicillins: The Complete GP Revision

Sources: Goodman & Gilman's Pharmacological Basis of Therapeutics | Katzung's Basic & Clinical Pharmacology 16e | CDC Antibiotic Stewardship Report 2025 | IDSA AMR Guidance 2026

1. CLASSIFICATION AT A GLANCE

Class	Key Drugs	Spectrum
Natural penicillins	Penicillin G (IV), Penicillin V (oral), Benzathine Pen G (IM)	Narrow - Gram+, spirochetes
Antistaphylococcal	Oxacillin, Nafcillin, Dicloxacillin	Narrow - MSSA only
Aminopenicillins	Ampicillin, Amoxicillin	Broader - some Gram- added
+ β-lactamase inhibitor	Amoxicillin-Clavulanate (Co-amoxiclav), Ampicillin-Sulbactam	Broad - overcomes resistance
Antipseudomonal	Piperacillin-Tazobactam (Pip-Taz)	Extended Gram- + Pseudomonas

2. MECHANISM OF ACTION

Penicillins are β-lactam antibiotics - they are bactericidal and work by:

Irreversibly binding to Penicillin-Binding Proteins (PBPs) on the bacterial cell membrane
PBPs are transpeptidase enzymes responsible for the final cross-linking step of peptidoglycan (the structural backbone of the bacterial cell wall)
Inhibiting PBPs leads to accumulation of defective cell wall precursors → the bacteria cannot maintain osmotic integrity → cell lysis and death

Key point: They are time-dependent killers. Efficacy depends on maintaining drug concentration above the MIC for >40-50% of the dosing interval (not on peak concentration).

3. PHARMACOKINETICS - DEEP DIVE

Parameter	Details
Absorption	Amoxicillin - good oral bioavailability (~80%), can be taken with food. Ampicillin and Pen V - 30-55%, must be given on empty stomach (1-2 hrs before or after meals)
Distribution	Widely distributed into tissues, joint/pleural/pericardial fluid, bile. Poor penetration into: CSF (normal meninges <1% of plasma), prostate, phagocytic cells, intraocular fluid. CSF penetration improves to ~5% with meningeal inflammation
Protein binding	Variable. Nafcillin/Dicloxacillin - highly protein-bound. Amoxicillin/Ampicillin - lower binding
Metabolism	Minimal hepatic metabolism. Nafcillin is the exception - primarily biliary excretion
Elimination	Predominantly renal - glomerular filtration AND tubular secretion. Short half-life: 30-90 minutes for most penicillins
Dose adjustment	Reduce dose in significant renal impairment (except Nafcillin)
Probenecid interaction	Blocks renal tubular secretion of penicillins → prolongs half-life and raises drug levels. Still used occasionally for gonorrhoea or syphilis regimens

Goodman & Gilman's, p.1170

4. SPECTRUM OF ACTIVITY

Natural Penicillins (Pen G, Pen V, Benzathine Pen G)

Gram-positive: Most streptococci (S. pyogenes, viridans group), penicillin-susceptible S. pneumoniae, non-β-lactamase staphylococci
Gram-negative: Neisseria meningitidis (still very sensitive), some anaerobes (Clostridium spp., Actinomyces)
Spirochetes: T. pallidum (syphilis), Borrelia, Leptospira - remain highly sensitive
⚠️ >90% of S. aureus is now resistant to natural penicillins due to β-lactamase production
⚠️ Penicillin-resistant S. pneumoniae is common in paediatric populations

Aminopenicillins (Amoxicillin, Ampicillin)

All of the above, PLUS:
H. influenzae (non-β-lactamase strains), E. coli, Proteus mirabilis, Salmonella, Enterococcus faecalis
Amoxicillin is the most active oral β-lactam against S. pneumoniae with elevated MICs - use higher doses (80-90 mg/kg/day) for AOM in children
⚠️ Ampicillin-resistant H. influenzae widespread - use Co-amoxiclav instead
⚠️ No activity against Klebsiella, Enterobacter, Pseudomonas, Serratia, Citrobacter (intrinsic β-lactamases)

Amoxicillin-Clavulanate (Co-amoxiclav)

Adds coverage for β-lactamase-producing H. influenzae, Moraxella catarrhalis, E. coli, Klebsiella, S. aureus (MSSA)
Does NOT cover MRSA, Pseudomonas, Enterobacter

Piperacillin-Tazobactam

Extended Gram-negative coverage including Pseudomonas aeruginosa, most Enterobacterales (including some ESBL-producers), anaerobes
A hospital-level drug; not for GP outpatient prescribing typically
Katzung 16e, p.1244-1245

5. GP-RELEVANT CLINICAL INDICATIONS + DOSES

Amoxicillin

Indication	Adult Dose	Duration	Notes
Streptococcal pharyngitis	500 mg TDS or 1g BD	5-7 days	Still first-line; penicillin V is an option
Community-acquired pneumonia (mild, low-risk)	1g TDS	5 days	For suspected pneumococcal; use with azithromycin if atypicals suspected
Acute otitis media	500 mg TDS	5-7 days	High-dose 80-90 mg/kg/day in children
Acute sinusitis (bacterial)	500 mg TDS	5 days	Only for bacterial, not viral
H. pylori eradication	As part of triple/quadruple therapy	10-14 days	See regimens below

Amoxicillin-Clavulanate (Co-amoxiclav)

Indication	Adult Dose	Duration
Otitis media (treatment failure/severe)	625 mg TDS or 1g BD	5-7 days
Exacerbation of chronic bronchitis	625 mg TDS	5 days
Sinusitis (with β-lactamase risk)	625 mg BD/TDS	5 days
UTI (Co-amoxiclav, second-line)	625 mg TDS	7 days
Animal bites	625 mg TDS	5-7 days
Cellulitis (mild, community-acquired)	625 mg TDS	5-7 days

2025-2026 Update (Wales/UK): National guidance now targets ≥75% of amoxicillin prescriptions to be 5-day courses rather than 7 days. Evidence shows equivalent clinical outcomes with reduced antibiotic exposure and AMR risk. (GOV.Wales AMR 2025-2027)

Benzathine Penicillin G (IM)

Indication	Dose	Schedule
Streptococcal pharyngitis	1.2 million units IM	Single dose
Syphilis (primary/secondary/latent early <1yr)	2.4 million units IM	Single dose
Syphilis (late latent/>1yr)	2.4 million units IM	Weekly x 3 weeks
Rheumatic fever prophylaxis	1.2 million units IM	Every 3-4 weeks

H. pylori Eradication (Amoxicillin component)

Standard first-line Bismuth Quadruple Therapy:

PPI + Bismuth + Tetracycline + Metronidazole x 14 days
If bismuth unavailable, Non-bismuth concomitant therapy (PPI + Amoxicillin + Clarithromycin + Metronidazole x 14 days) - recent 2026 meta-analysis (PMID: 41498306) confirms high eradication rates with this regimen.

6. RESISTANCE MECHANISMS - THE "BIG 3"

Mechanism	Details	Clinical Example
β-lactamase production	Most common. Enzyme cleaves the β-lactam ring	S. aureus, H. influenzae, E. coli. Overcome with clavulanate/sulbactam/tazobactam
Altered PBPs	PBP mutations reduce penicillin binding affinity	MRSA (PBP2a - mecA gene); Penicillin-resistant S. pneumoniae. β-lactamase inhibitors do NOT help here
Efflux pumps / Reduced permeability	Drug pumped out or can't enter the cell	Gram-negative bacteria (P. aeruginosa); combined with β-lactamase

2026 IDSA AMR Guidance note: ESBLs (Extended-Spectrum β-Lactamases) inactivate most penicillins, cephalosporins, and aztreonam. ESBL-producing Enterobacterales generally remain susceptible to carbapenems. (IDSA AMR Guidance 2026)

7. ADVERSE EFFECTS

Effect	Frequency	Details
Hypersensitivity reactions	Most important	Ranges from maculopapular rash (~5%) to urticaria, serum sickness, to life-threatening anaphylaxis (<0.05%)
GI disturbance	Common with Co-amoxiclav	Nausea, diarrhoea - especially clavulanate component. Take with food for Co-amoxiclav
C. difficile	Less common than clindamycin/FQ	Still a risk, especially in elderly/hospitalized
Neurotoxicity (rare)	High-dose IV	Seizures with very high doses (e.g., >20 MU/day penicillin G), especially in renal impairment
Haematological	Rare	Haemolytic anaemia, neutropaenia, thrombocytopaenia with prolonged use
Electrolyte disturbance	IV formulations	High-dose IV piperacillin can cause hypokalaemia

8. PENICILLIN ALLERGY - THE UPDATED PICTURE ⭐ (Critical GP Pearl)

This is one of the most clinically important and commonly mismanaged issues:

~10% of patients self-report penicillin allergy
<1% are truly allergic (CDC 2025)
Most "allergies" are:
- Maculopapular rash from amoxicillin during EBV infection (misattributed allergy)
- Childhood GI side effects
- Family history

Cross-reactivity with Cephalosporins:

Overall cross-reactivity: ~2-7% (likely an overestimate based on older data)
True cross-reactivity is R-group side chain dependent, not just ring structure
Patients with penicillin allergy + cephalosporin with different R-group: cross-reactivity <2%
Patients with history of anaphylaxis to penicillin: cross-reactivity up to 50% - skin test or use alternative
Penicillin vs. carbapenem: <1%
Penicillin vs. monobactam (aztreonam): 0% (except aztreonam-ceftazidime share R-group)

GP action: When you see "penicillin allergy" on a patient's record - clarify the reaction. Encourage de-labelling of patients with low-risk reactions. An inaccurate penicillin allergy label forces use of broader-spectrum antibiotics, driving resistance.

Washington Manual of Medical Therapeutics

9. KEY DRUG INTERACTIONS

Drug	Interaction	Clinical Significance
Probenecid	Blocks tubular secretion → raises penicillin levels	Used intentionally for syphilis/gonorrhoea regimens
Warfarin / Anticoagulants	Broad-spectrum penicillins alter gut flora → reduce Vit K → potentiate anticoagulation	Monitor INR
Oral contraceptives	Theoretical reduction in efficacy (old data, largely dismissed)	Advise per local guidance
Methotrexate	Penicillins reduce renal MTX excretion → toxicity risk	Avoid combination or monitor MTX levels closely
Allopurinol	Increases incidence of ampicillin rash	Rash rate rises from 5% to ~20%

10. PRESCRIBING PEARLS FOR GPS 🩺

Amoxicillin with food - always fine. All other oral penicillins - empty stomach
Amoxicillin for CAP: 1g TDS is now recommended over 500mg TDS due to rising pneumococcal MICs
Co-amoxiclav for bites - ALWAYS the first choice for human/animal bites
Benzathine penicillin for syphilis - never substitute with oral penicillin (incomplete absorption)
Don't use ampicillin empirically for UTI or typhoid - resistance rates too high
Shorter courses work: 5-day amoxicillin is as effective as 7 days for most community infections (2025-2026 guideline shift)
Penicillin allergy de-labelling is a stewardship priority - most self-reported allergies are not true allergies
MRSA coverage: No penicillin covers MRSA. If MRSA suspected, escalate
Enterococcal endocarditis: Ampicillin + Ceftriaxone (preferred over Ampicillin + Gentamicin - same efficacy, less nephrotoxicity)
Renal dose adjustment: Required for most penicillins except Nafcillin (biliary excretion)

11. MEMORY AID

PENICILLINS = "BLAST the Cell Wall"
B - Bactericidal (cell wall synthesis inhibitor)
L - β-Lactam ring is the pharmacophore
A - Allergy is the #1 concern (anaphylaxis possible)
S - Spectrum varies by subclass (narrow → broad)
T - Time-dependent killing (T > MIC is key PK/PD)

Quick Recall Card

	Natural Pen	Aminopenicillins	Co-amoxiclav	Antistaphylococcal
MRSA?	No	No	No	No (only MSSA)
Pseudomonas?	No	No	No	No
H. influenzae?	No	Some strains	Yes	No
S. aureus (MSSA)?	No (resist)	No	Yes	Yes (DOC)
Streptococci?	Yes (DOC)	Yes	Yes	Yes
Spirochetes?	Yes (DOC)	Yes	Yes	No

That's Day 1 - Penicillins - fully covered from mechanism to GP pearls with 2025-2026 updates.

Tomorrow's suggestion: Cephalosporins (natural progression from penicillins, building on β-lactam knowledge) - or tell me which class you'd prefer next! 🎯

Time-dependent killing (T > MIC is key PK/PD)- what does this mean?

Great question - this is a fundamental concept in antibiotic prescribing that directly affects how you dose drugs in practice.

Searching sources...

Finding Sources

Searching Images

antibiotic pharmacodynamics time-dependent concentration-dependent killing MIC graph

Reading File

Excellent - Harrison's 22e and Fishman's Pulmonary have exactly what we need. Here's the full explanation:

⏱️ Time-Dependent Killing (T > MIC) - Explained

First, what is MIC?

MIC = Minimum Inhibitory Concentration

The lowest drug concentration that stops a bacterium from growing under lab conditions
Think of it as the "floor" the drug must stay above to be effective
Every bacterium has its own MIC for every antibiotic
Example: S. pneumoniae may have an MIC of 0.06 mg/L for amoxicillin

The 3 PK/PD Models of Antibiotic Killing

Antibiotics don't all kill bacteria in the same way. There are 3 patterns:

Model	What drives killing	Key parameter	Examples
Time-dependent	How long drug stays above MIC	T > MIC	Penicillins, Cephalosporins, Carbapenems, Macrolides, Linezolid
Concentration-dependent	How high the peak goes vs. MIC	Cmax : MIC	Aminoglycosides, Fluoroquinolones
AUC-dependent	Total drug exposure over time vs. MIC	AUC : MIC	Vancomycin, Daptomycin, Fluoroquinolones

Harrison's Principles of Internal Medicine 22e, p.1211 | Fishman's Pulmonary Diseases, p.2680

Time-Dependent Killing - The Core Concept

For penicillins (and all β-lactams), killing rate does NOT increase with higher doses beyond a ceiling.

Imagine two scenarios:

Scenario A: Give a HUGE dose once
Drug level: ████████████████░░░░░░░░░░░░░░░░░░░  (high peak, then drops below MIC)
                              ↑ MIC threshold
Effect: Kills bacteria during the high phase... then bacteria REGROW when drug drops

Scenario B: Give smaller doses MORE frequently (or spread out the same dose)
Drug level: ████░░████░░████░░████░░████░░████░░
                              ↑ MIC threshold
Effect: Drug stays above MIC for longer → sustained bacterial killing ✅

The key insight: Once drug concentration reaches ~4x the MIC, killing is maximal and saturated - going higher does NOTHING extra. What matters is keeping it above the MIC for enough time.

The Target: How long above MIC is "enough"?

β-Lactam Group	Target (fT > MIC)	Meaning
Penicillins	40-50% of dosing interval	Drug must be above MIC for ≥40-50% of the time between doses
Cephalosporins	60-70% of dosing interval
Carbapenems	≥40% of dosing interval

"f" stands for the free (unbound) fraction - only drug not bound to protein is active.

Fishman's Pulmonary Diseases, p.2680

A Practical Example (Amoxicillin 500mg TDS)

Dose at 8am → 2pm → 8pm (every 8 hours = 480 min dosing interval)

Drug concentration over 8 hours:
         ↑ Peak
         |  \
         |   \
- - - - -|- - --------  ← MIC
         |         \
         |          \___  (falls below MIC near end of interval)

Time above MIC = ~4 hours out of 8 = 50% ✅ (just hits target for penicillins)

This is why amoxicillin is given 3x/day (TDS) and not once daily - unlike, say, azithromycin which has a long half-life and tissue accumulation.

Why Does This Matter for Your Prescribing?

1. Frequency matters more than dose size (for β-lactams)

Doubling the amoxicillin dose from 500mg to 1g does NOT double the bacterial kill if you're already above 4x MIC
But increasing frequency (BD → TDS) or extending infusion time does improve efficacy

2. It explains why compliance matters so much

Missing doses of amoxicillin/penicillin → drug level drops below MIC → bacteria regrow
With concentration-dependent drugs (e.g., gentamicin), one big dose a day is fine - but not with penicillins

3. Extended/continuous IV infusions for severe infections

In ICU settings, piperacillin-tazobactam is sometimes given as a 3-4 hour extended infusion (rather than a 30-min bolus) to maximize T > MIC - especially against organisms with higher MICs
This is called pharmacodynamic optimisation

4. Rising MICs = shrinking T > MIC window

As bacteria develop resistance → MIC rises → the same dose stays above MIC for a shorter fraction of the interval → treatment may fail even with the "right" antibiotic
This is why high-dose amoxicillin (80-90 mg/kg/day) is used for AOM in children - to overcome modestly elevated pneumococcal MICs

Contrast with Concentration-Dependent (e.g., Gentamicin)

For gentamicin - the HIGHER the peak above MIC, the MORE it kills. So you give it once daily in a big dose to maximize Cmax:MIC, and the low trough between doses actually reduces nephrotoxicity. Completely opposite logic to penicillins.

Summary - One-Line Memory Peg

🔑 Penicillins: "Keep it dripping, not spiking" - it's the time spent above the MIC that kills bacteria, not how high the peak goes.

	β-Lactams (Pen, Ceph)	Aminoglycosides (Gentamicin)	Vancomycin
Key parameter	T > MIC	Cmax : MIC	AUC : MIC
Strategy	Dose frequently	Dose high, once daily	Monitor trough/AUC
Analogy	Slow sustained rain	One heavy downpour	Total rainfall over 24h

This concept directly underpins why we prescribe penicillins TDS/QDS, why extended infusions help in severe infections, and why missing doses is harmful - all things you'll use in GP practice every day.

Ready for tomorrow's drug? 🎯

Explain me the next two methods concentration and AUC dependent also like above.

Searching sources...

Finding Sources

Reading File

Finding Sources

There's even a textbook figure for this - let me fetch it.

Loading Image

This is the actual Harrison's 22e PK/PD diagram. Let's build the full explanation around it.

📈 Concentration-Dependent & AUC-Dependent Killing - Explained

First, look at this official textbook diagram - it shows all three parameters on a single drug concentration-time curve:

The red dashed line = MIC. The blue curve = drug concentration over time after one dose.

Peak (Cmax/MIC) = how HIGH the peak goes above the MIC
T > MIC = the window of time the drug stays above the MIC (what we covered last time)
AUC/MIC = the entire shaded area under the curve, divided by the MIC

Now let's go through the next two in depth.

Harrison's Principles of Internal Medicine 22e, p.1211

🔺 Method 2 - Concentration-Dependent Killing (Cmax : MIC)

The Core Concept

For these drugs, the HIGHER the peak concentration relative to the MIC, the FASTER and MORE COMPLETELY bacteria are killed. There is NO ceiling effect - doubling the peak genuinely doubles the kill rate.

Low peak dose (e.g. 3mg/kg):
         ↑ Cmax (3x MIC)
         | \
- - - - -|- --------  ← MIC
         |       \___
Effect: Some killing, bacteria may survive and regrow ❌

High peak dose (e.g. 7mg/kg):
              ↑ Cmax (10x MIC)
              |  \
              |   \
- - - - - - - - - --------  ← MIC
                        \___
Effect: Rapid, massive bacterial kill ✅

The higher you drive the peak, the better. The long drug-free trough between doses? That's actually fine - and even beneficial (see PAE below).

Key Parameter

Cmax : MIC ratio - target is typically ≥8-12x

For example, if bacterial MIC for gentamicin = 1 mg/L, you want your Cmax to be at least 8-12 mg/L.

Why the Drug-Free Trough Helps

This is where the brilliant concept of the Post-Antibiotic Effect (PAE) comes in:

PAE = Even after drug concentration drops BELOW the MIC, bacteria remain stunned and suppressed for 1-4 hours before they start regrowing.

Think of it like a boxer who's been knocked down - even after you stop hitting, they don't immediately jump back up.

For aminoglycosides, PAE against Gram-negatives lasts 1-4 hours. This gap also allows the kidney tubular cells time to "flush out" drug and recover - which is why:

Once-daily gentamicin: HIGH peak (kill) → long trough (kidney recovery + PAE still working)
                                                         ↓
                                        LESS nephrotoxicity + BETTER killing ✅

Three-times-daily gentamicin (old approach): lower peaks + sustained drug in kidney
                                                         ↓
                                        MORE nephrotoxicity + LESS killing ❌

This is completely opposite to penicillins. Frequent dosing of aminoglycosides = worse outcome.

Examples of Concentration-Dependent Drugs

Drug	Target Cmax:MIC	Dosing Strategy
Gentamicin	≥8-10x	Once daily (5-7 mg/kg)
Amikacin	≥8-10x	Once daily
Tobramycin	≥8-10x	Once daily
Metronidazole	High Cmax	Standard dosing

Fluoroquinolones (ciprofloxacin, levofloxacin) are a hybrid - they have both Cmax/MIC AND AUC/MIC dependency (covered below).

Prescribing Implication

Gentamicin dose calculation:

Dose = 5-7 mg/kg once daily (not 1.7 mg/kg TDS as in old regimens)
Monitor trough level (drawn just before next dose) - must be <1 mg/L (ensures adequate drug-free interval = kidney recovery)
Monitor peak level if needed - drawn 1 hour after infusion
Fishman's Pulmonary Diseases, p.2684-2686

📊 Method 3 - AUC-Dependent Killing (AUC : MIC)

What is AUC?

AUC = Area Under the Curve

Look at the graph again - it's literally the total area of the blue shaded region under the concentration-time curve. It represents total drug exposure over the full dosing interval - combining both how high it went AND how long it stayed.

Drug concentration
         ↑
         |████
         |██████
         |█████████
- - - - -|████████████- - -  ← MIC
         |          █████████
         |_____________________________→ Time

The AUC = the ENTIRE shaded area (█) above AND below the MIC line
AUC/MIC = this total area ÷ the MIC value

The Core Concept

These drugs are a hybrid - they are influenced by BOTH how high the peak goes AND how long the drug is around. Neither alone is sufficient. It's the total exposure that predicts killing.

Think of it like paying for a gym membership by total hours used - it doesn't matter if you went for one intense 5-hour session or five 1-hour sessions. Total hours = total benefit.

Key Targets

Drug	Organism	AUC/MIC Target
Vancomycin	MRSA, S. aureus	AUC₂₄ 400-600 mg·h/L
Fluoroquinolones	S. pneumoniae	AUC/MIC >30
Fluoroquinolones	Gram-negative bacteria	AUC/MIC >125

Vancomycin - The Classic Example

Vancomycin monitoring has recently shifted away from trough-only monitoring to AUC-guided dosing:

Old approach (trough monitoring):

Just check trough level before next dose → target 15-20 mg/L
Problem: High trough doesn't guarantee adequate AUC, and risks nephrotoxicity

New approach (AUC/MIC monitoring):

Target AUC₂₄ of 400-600 mg·h/L (with MIC typically = 1 mg/L → AUC/MIC of 400-600)
Achieved by calculating dose using 2-point PK sampling or Bayesian software
Lower risk of nephrotoxicity while ensuring efficacy

"Recent evidence strongly suggests moving away from serum trough vancomycin monitoring to a PK/PD target of AUC to MIC ratio. An AUC₂₄ of 400-600 mg·h/L is associated with clinical efficacy."

Harriet Lane Handbook 23e

Fluoroquinolones - The True Hybrid

Fluoroquinolones (ciprofloxacin, levofloxacin, moxifloxacin) actually sit across BOTH concentration-dependent AND AUC-dependent models:

Against S. pneumoniae: AUC/MIC >30 is the target
Against Gram-negatives: AUC/MIC >125 is the target
A Cmax/MIC of ≥12 for pneumococcus is also associated with optimal killing

This is why levofloxacin 500mg once daily works well - one big dose achieves both a good peak AND a sufficient 24-hour AUC, rather than splitting it into smaller doses.

Examples of AUC-Dependent Drugs

Drug	Primary target	Monitoring
Vancomycin	AUC₂₄/MIC 400-600	Trough OR AUC-guided (preferred now)
Levofloxacin	AUC/MIC >30 (GPC), >125 (GNR)	Clinical response
Ciprofloxacin	AUC/MIC >125 (GNR)	Clinical response
Moxifloxacin	AUC/MIC	Clinical response
Daptomycin	AUC/MIC	Clinical response
Tigecycline	AUC/MIC	Clinical response

Harrison's 22e, p.1211 | Fishman's Pulmonary Diseases, p.2680 | Harriet Lane 23e

🧠 All Three Methods - Side-by-Side Master Summary

	Time-Dependent	Concentration-Dependent	AUC-Dependent
What drives killing	Duration above MIC	Height of peak vs. MIC	Total exposure (peak + duration)
Key parameter	T > MIC	Cmax : MIC	AUC : MIC
Ceiling effect?	YES - killing saturates at 4x MIC	NO - higher peak = more kill	NO - more exposure = more kill
Optimal dosing	Frequent small doses / extended infusion	Infrequent large doses	Once or twice daily
Drug-free trough?	Dangerous (bacteria regrow)	Fine - PAE covers the gap	Fine to have trough periods
PAE?	Minimal (β-lactams)	Long (aminoglycosides)	Moderate
Examples	Penicillins, Cephalosporins, Carbapenems	Gentamicin, Amikacin	Vancomycin, Fluoroquinolones, Daptomycin
Analogy	Steady drip irrigation	One powerful flood	Total rainfall measurement

🎯 The "Why Does Any of This Matter to a GP?" Summary

Scenario	PK/PD Lesson
Patient missed 2 doses of amoxicillin	Dangerous - T > MIC drops, bacteria regrow. Stress compliance
Prescribing ciprofloxacin for UTI	Once daily 500mg fine - AUC/MIC model means one big dose works
Gentamicin in hospital	Once-daily dosing is BETTER than TDS - higher kill, less toxicity
Patient on vancomycin for MRSA	Ask for AUC monitoring, not just trough levels (current 2025 standard)
Levofloxacin 500mg OD vs 250mg BD	Same total dose - but OD achieves higher Cmax/MIC AND similar AUC - OD preferred

Memory Peg - All 3 Together

🅣 Time > MIC       = β-Lactams  → "Keep it STEADY" (drip, drip, drip)
🅒 Cmax/MIC         = Aminoglycosides → "Hit it HARD once" (one big punch)
🅐 AUC/MIC          = Vancomycin/FQ → "Total EXPOSURE matters" (count all hours)

You now have the complete PK/PD framework that underpins how every antibiotic class is dosed. Everything in antibiotic therapy - dosing frequency, monitoring parameters, dose adjustments in renal failure - flows from these three principles. 💡

Create a quick reference PDF comparing all 3 PK/PD models with dosing strategies

Reading File

Good - I'll use ReportLab to create a well-designed PDF. Let me build it now.

Running Command

ls -la /tmp/workspace/ 2>/dev/null || mkdir -p /tmp/workspace && echo "workspace ready"

Running Command

mkdir -p /tmp/workspace/pkpd-reference && cat > /tmp/workspace/pkpd-reference/generate_pdf.py << 'PYEOF' from reportlab.lib.pagesizes import A4 from reportlab.lib import colors from reportlab.lib.units import cm, mm from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle from reportlab.lib.enums import TA_CENTER, TA_LEFT, TA_JUSTIFY from reportlab.platypus import ( SimpleDocTemplate, Paragraph, Spacer, Table, TableStyle, HRFlowable, KeepTogether ) from reportlab.graphics.shapes import Drawing, Rect, Line, String, Polygon, Circle from reportlab.graphics import renderPDF from reportlab.graphics.charts.lineplots import LinePlot from reportlab.graphics.widgets.markers import makeMarker import math # ── Colour palette ────────────────────────────────────────────── C_NAVY = colors.HexColor('#1B2A4A') C_TEAL = colors.HexColor('#0F7C8A') C_AMBER = colors.HexColor('#E07B2A') C_GREEN = colors.HexColor('#2A7A3B') C_LRED = colors.HexColor('#C0392B') C_LTBLUE = colors.HexColor('#D6EAF8') C_LTAMBER = colors.HexColor('#FEF9E7') C_LTGREEN = colors.HexColor('#EAFAF1') C_LTYELL = colors.HexColor('#FDFEFE') C_GREY = colors.HexColor('#7F8C8D') C_LGREY = colors.HexColor('#F2F3F4') C_WHITE = colors.white C_BLACK = colors.HexColor('#1C1C1C') C_MICRED = colors.HexColor('#E74C3C') W, H = A4 # 595.27 x 841.89 pts # ── Document ───────────────────────────────────────────────────── doc = SimpleDocTemplate( '/tmp/workspace/pkpd-reference/PKPD_Antibiotic_Reference.pdf', pagesize=A4, leftMargin=1.5*cm, rightMargin=1.5*cm, topMargin=1.5*cm, bottomMargin=1.8*cm, title='Antibiotic PK/PD Models – Quick Reference', author='Orris Medical AI', subject='Pharmacokinetics / Pharmacodynamics' ) styles = getSampleStyleSheet() def sty(name, **kw): s = ParagraphStyle(name, **kw) return s S_TITLE = sty('Title2', fontName='Helvetica-Bold', fontSize=20, textColor=C_WHITE, alignment=TA_CENTER, spaceAfter=2) S_SUBTITLE = sty('Sub', fontName='Helvetica', fontSize=10, textColor=C_WHITE, alignment=TA_CENTER, spaceAfter=0) S_SECHEAD = sty('SecHead', fontName='Helvetica-Bold', fontSize=12, textColor=C_WHITE, alignment=TA_CENTER, spaceAfter=4) S_BODY = sty('Body2', fontName='Helvetica', fontSize=8.5, textColor=C_BLACK, leading=12, spaceAfter=3) S_BOLD = sty('Bold2', fontName='Helvetica-Bold', fontSize=8.5, textColor=C_BLACK, leading=12) S_SMALL = sty('Small', fontName='Helvetica', fontSize=7.5, textColor=C_GREY, leading=10) S_CELLHEAD = sty('CH', fontName='Helvetica-Bold', fontSize=8, textColor=C_WHITE, alignment=TA_CENTER) S_CELL = sty('CL', fontName='Helvetica', fontSize=8, textColor=C_BLACK, alignment=TA_LEFT, leading=11) S_CELLC = sty('CC', fontName='Helvetica', fontSize=8, textColor=C_BLACK, alignment=TA_CENTER, leading=11) S_PEARLS = sty('Pearl', fontName='Helvetica', fontSize=8, textColor=C_BLACK, leading=12, leftIndent=6) # ── Curve drawing helper ────────────────────────────────────────── def make_curve_drawing(model, w=160, h=90): """Draw a mini PK curve for each model.""" d = Drawing(w, h) pad_l, pad_b, pad_r, pad_t = 18, 14, 8, 8 gw = w - pad_l - pad_r gh = h - pad_b - pad_t mic_y = pad_b + gh * 0.28 # MIC line position def tx(t): return pad_l + t * gw # t in [0,1] def cy(c): return pad_b + c * gh # c in [0,1] # Axes d.add(Line(pad_l, pad_b, pad_l, pad_b+gh, strokeColor=C_GREY, strokeWidth=0.7)) d.add(Line(pad_l, pad_b, pad_l+gw, pad_b, strokeColor=C_GREY, strokeWidth=0.7)) # Axis labels d.add(String(2, pad_b+gh//2, 'Conc.', fontSize=5.5, fillColor=C_GREY, textAnchor='middle')) d.add(String(pad_l+gw//2, 1, 'Time', fontSize=5.5, fillColor=C_GREY, textAnchor='middle')) # MIC dashed line seg = 4 x = pad_l while x < pad_l + gw: d.add(Line(x, mic_y, min(x+seg, pad_l+gw), mic_y, strokeColor=C_MICRED, strokeWidth=0.9, strokeDashArray=[3,2])) x += seg*2 d.add(String(pad_l+gw+1, mic_y-2, 'MIC', fontSize=5, fillColor=C_MICRED, textAnchor='start')) if model == 'time': # Two doses showing T>MIC windows for offset in [0, 0.47]: pts = [] for i in range(60): t = i/59 raw = math.exp(-6*(t-0.08)**2) if t > 0.04 else 0 c = 0.55 * raw pts.append((tx(t*0.44 + offset), cy(c))) for i in range(len(pts)-1): d.add(Line(pts[i][0], pts[i][1], pts[i+1][0], pts[i+1][1], strokeColor=C_TEAL, strokeWidth=1.4)) # T>MIC bracket x1, x2 = tx(0.07), tx(0.38) d.add(Line(x1, mic_y+2, x2, mic_y+2, strokeColor=C_TEAL, strokeWidth=1.2, strokeDashArray=[2,1])) d.add(String((x1+x2)/2, mic_y+4, 'T > MIC', fontSize=5.5, fillColor=C_TEAL, textAnchor='middle')) elif model == 'conc': # Single high spike pts = [] for i in range(80): t = i/79 if t < 0.05: c = t/0.05 * 0.92 else: c = 0.92 * math.exp(-5*(t-0.05)) pts.append((tx(t), cy(c))) for i in range(len(pts)-1): d.add(Line(pts[i][0], pts[i][1], pts[i+1][0], pts[i+1][1], strokeColor=C_AMBER, strokeWidth=1.4)) # Cmax arrow peak_x = tx(0.05) peak_y = cy(0.92) d.add(Line(peak_x+2, peak_y, peak_x+2, mic_y, strokeColor=C_AMBER, strokeWidth=1, strokeDashArray=[2,2])) d.add(String(peak_x+4, (peak_y+mic_y)/2, 'Cmax/MIC', fontSize=5.5, fillColor=C_AMBER, textAnchor='start')) # PAE bracket # find where curve drops below MIC pae_start = tx(0.42) pae_end = tx(0.62) d.add(Line(pae_start, mic_y-5, pae_end, mic_y-5, strokeColor=C_GREEN, strokeWidth=1)) d.add(Line(pae_start, mic_y-3, pae_start, mic_y-7, strokeColor=C_GREEN, strokeWidth=0.8)) d.add(Line(pae_end, mic_y-3, pae_end, mic_y-7, strokeColor=C_GREEN, strokeWidth=0.8)) d.add(String((pae_start+pae_end)/2, mic_y-12, 'PAE', fontSize=5.5, fillColor=C_GREEN, textAnchor='middle')) elif model == 'auc': # Moderate curve, shade AUC pts = [] for i in range(80): t = i/79 if t < 0.07: c = t/0.07 * 0.72 else: c = 0.72 * math.exp(-3.5*(t-0.07)) pts.append((tx(t), cy(c))) # Fill AUC area (simplified - draw polygon) poly_pts = [(pad_l, pad_b)] for p in pts: poly_pts.append(p) poly_pts.append((tx(1), pad_b)) poly_flat = [] for p in poly_pts: poly_flat.extend(p) d.add(Polygon(poly_flat, fillColor=colors.HexColor('#D5F5E3'), strokeColor=None, fillOpacity=0.5)) for i in range(len(pts)-1): d.add(Line(pts[i][0], pts[i][1], pts[i+1][0], pts[i+1][1], strokeColor=C_GREEN, strokeWidth=1.4)) # AUC label inside shaded region d.add(String(tx(0.35), cy(0.18), 'AUC/MIC', fontSize=5.5, fillColor=C_GREEN, textAnchor='middle')) return d # ── PAGE BACKGROUND helper (via canvas) ────────────────────────── def header_footer(canvas, doc): canvas.saveState() # Top header bar canvas.setFillColor(C_NAVY) canvas.rect(0, H-2.6*cm, W, 2.6*cm, fill=1, stroke=0) # Title canvas.setFillColor(C_WHITE) canvas.setFont('Helvetica-Bold', 18) canvas.drawCentredString(W/2, H-1.5*cm, 'Antibiotic PK/PD Models – Quick Reference') canvas.setFont('Helvetica', 9) canvas.drawCentredString(W/2, H-2.1*cm, 'For General Practitioners | Time-Dependent · Concentration-Dependent · AUC-Dependent') # Bottom bar canvas.setFillColor(C_NAVY) canvas.rect(0, 0, W, 1.2*cm, fill=1, stroke=0) canvas.setFillColor(C_WHITE) canvas.setFont('Helvetica', 7.5) canvas.drawString(1.5*cm, 0.45*cm, 'Sources: Harrison\'s Principles 22e · Goodman & Gilman · Fishman\'s Pulmonary Diseases · Harriet Lane 23e') canvas.drawRightString(W-1.5*cm, 0.45*cm, 'Orris Medical AI | August 2026') canvas.restoreState() # ── Build content ───────────────────────────────────────────────── story = [] story.append(Spacer(1, 2.8*cm)) # space below header bar # ── INTRO ROW ──────────────────────────────────────────────────── intro_text = ( "Antibiotic efficacy depends not only on which drug you prescribe, but on how you dose it. " "Three pharmacodynamic (PD) models govern bacterial killing, each requiring a different dosing strategy. " "Understanding these models explains dosing frequency, monitoring parameters, and why the same total daily " "dose given differently can succeed or fail." ) intro_style = ParagraphStyle('intro', fontName='Helvetica', fontSize=9, textColor=C_BLACK, leading=13, alignment=TA_JUSTIFY, borderColor=C_TEAL, borderWidth=1, borderPadding=8, backColor=colors.HexColor('#EAF7FB')) story.append(Paragraph(intro_text, intro_style)) story.append(Spacer(1, 0.35*cm)) # ══════════════════════════════════════════════════════════════════ # SECTION CARDS – one per model # ══════════════════════════════════════════════════════════════════ models = [ { 'color': C_TEAL, 'light': colors.HexColor('#EAF7FB'), 'emoji': '⏱', 'title': 'Model 1 – Time-Dependent Killing', 'param': 'T > MIC', 'param_full': 'Time drug concentration stays above MIC', 'what': 'Killing rate saturates once concentration reaches ~4× MIC. Going higher adds nothing. Only staying ABOVE the MIC longer increases effect.', 'target': 'Penicillins: 40–50% · Cephalosporins: 60–70% · Carbapenems: ≥40% of dosing interval above MIC', 'drugs': [ ('Amoxicillin / Pen V', 'TDS or QDS', 'Must maintain T>MIC; TDS > BD'), ('Benzylpenicillin (IV)', 'q4–6h or CI', 'Short t½; continuous infusion in ICU'), ('Cephalexin', 'QDS', 'Higher freq than cefixime (longer t½)'), ('Ceftriaxone', 'OD–BD', 'Long t½ 6–9h allows once daily'), ('Meropenem (IV)', 'q8h or ext. infusion', 'Extended 3–4h infusion maximises T>MIC'), ('Piperacillin-Tazobactam', 'q6–8h or CI', 'Extended infusion for resistant organisms'), ], 'strategy': 'Dose FREQUENTLY or use extended/continuous IV infusions to keep drug above MIC.', 'trap': 'Doubling the dose without increasing frequency does NOT improve killing. Missing doses causes bacterial regrowth.', 'pae': 'Minimal PAE against Gram-negatives → cannot rely on drug-free interval.', 'monitoring': 'Clinical response; ensure compliance; extend infusion time for high-MIC organisms.', 'analogy': '🚿 Steady drip irrigation – constant water supply, not one big flood', 'curve': 'time', }, { 'color': C_AMBER, 'light': colors.HexColor('#FEF5E7'), 'emoji': '🔺', 'title': 'Model 2 – Concentration-Dependent Killing', 'param': 'Cmax : MIC', 'param_full': 'Peak drug concentration relative to MIC', 'what': 'The HIGHER the peak above MIC, the faster and more completely bacteria are killed. No ceiling – higher peak = better kill. Drug-free trough is safe (even beneficial).', 'target': 'Cmax/MIC ≥ 8–12× for aminoglycosides · Once-daily dosing achieves optimal peak', 'drugs': [ ('Gentamicin', '5–7 mg/kg OD (IV)', 'OD > TDS – higher peak, less nephrotoxicity'), ('Amikacin', '15–20 mg/kg OD (IV)', 'Monitor trough <5 mg/L; peak 56–64 mg/L'), ('Tobramycin', '5–7 mg/kg OD (IV)', 'Similar to gentamicin'), ('Metronidazole', 'Standard dosing', 'Some concentration-dependent features'), ], 'strategy': 'Give INFREQUENT but LARGE doses to maximise peak (Cmax). Once-daily > multiple small doses.', 'trap': 'Old TDS aminoglycoside dosing = lower peaks (less killing) + sustained trough (more nephrotoxicity). Avoid.', 'pae': 'Prolonged PAE 1–4h against Gram-negatives → drug-free trough is SAFE and reduces kidney exposure.', 'monitoring': 'Trough level before next dose: target <1 mg/L (gentamicin). Peak 1h post-infusion if needed.', 'analogy': '🥊 One powerful knockout punch – intensity matters, not duration', 'curve': 'conc', }, { 'color': C_GREEN, 'light': colors.HexColor('#EAFAF1'), 'emoji': '📊', 'title': 'Model 3 – AUC-Dependent Killing', 'param': 'AUC : MIC', 'param_full': 'Total drug exposure (area under curve) relative to MIC', 'what': 'Total drug exposure over 24h drives killing – accounts for BOTH peak height AND duration. Neither alone is sufficient; the integrated area under the concentration-time curve predicts efficacy.', 'target': 'Vancomycin: AUC₂₄/MIC 400–600 · Fluoroquinolones vs S. pneumoniae: >30 · vs Gram-negatives: >125', 'drugs': [ ('Vancomycin (IV)', 'Individualised dosing', 'AUC-guided monitoring now preferred over trough'), ('Levofloxacin', '500 mg OD', 'OD better than BD – higher AUC/MIC & Cmax/MIC'), ('Ciprofloxacin', '500–750 mg BD', 'AUC/MIC >125 for Gram-negatives'), ('Moxifloxacin', '400 mg OD', 'Superior tissue penetration; AUC-driven'), ('Daptomycin (IV)', '4–6 mg/kg OD', 'AUC/MIC + Cmax/MIC; OD dosing'), ('Tigecycline (IV)', '50 mg BD after LD', 'AUC-dependent; broad spectrum'), ], 'strategy': 'Optimise TOTAL EXPOSURE. OD or BD dosing achieves adequate AUC while maintaining practical schedule.', 'trap': 'Vancomycin: old trough-only monitoring (target 15–20 mg/L) is now replaced by AUC/MIC-guided dosing to balance efficacy and reduce nephrotoxicity.', 'pae': 'Moderate PAE; the drug-free interval is tolerated due to residual post-antibiotic suppression.', 'monitoring': 'Vancomycin: AUC₂₄ target 400–600 mg·h/L (2-point sampling or Bayesian). Fluoroquinolones: clinical response.', 'analogy': '🌧 Total rainfall over 24h – both the downpour and the drizzle count', 'curve': 'auc', }, ] for m in models: col = m['color'] bg = m['light'] items = [] # ── Section header bar ────────────────────────────────────── hdr_data = [[Paragraph(f"{m['title']}", S_SECHEAD)]] hdr_tbl = Table(hdr_data, colWidths=[doc.width]) hdr_tbl.setStyle(TableStyle([ ('BACKGROUND', (0,0), (-1,-1), col), ('ROWBACKGROUNDS', (0,0), (-1,-1), [col]), ('TOPPADDING', (0,0), (-1,-1), 6), ('BOTTOMPADDING', (0,0), (-1,-1), 6), ('LEFTPADDING', (0,0), (-1,-1), 8), ('ROUNDEDCORNERS', [4,4,0,0]), ])) items.append(hdr_tbl) # ── Two-column layout: curve + key facts ──────────────────── curve_d = make_curve_drawing(m['curve'], w=150, h=85) facts_para = Paragraph( f"Key Parameter: {m['param']} " f"{m['param_full']} " f"How it works: {m['what']} " f"Target: {m['target']}", ParagraphStyle('facts', fontName='Helvetica', fontSize=8, leading=11.5, textColor=C_BLACK, spaceAfter=2) ) two_col = Table( [[curve_d, facts_para]], colWidths=[155, doc.width - 155] ) two_col.setStyle(TableStyle([ ('BACKGROUND', (0,0), (-1,-1), bg), ('VALIGN', (0,0), (-1,-1), 'MIDDLE'), ('TOPPADDING', (0,0), (-1,-1), 6), ('BOTTOMPADDING', (0,0), (-1,-1), 6), ('LEFTPADDING', (0,0), (0,-1), 6), ('RIGHTPADDING', (-1,0), (-1,-1), 8), ])) items.append(two_col) # ── Drug table ─────────────────────────────────────────────── drug_rows = [[ Paragraph('Drug', S_CELLHEAD), Paragraph('Dosing Schedule', S_CELLHEAD), Paragraph('Clinical Note', S_CELLHEAD), ]] for drug, dose, note in m['drugs']: drug_rows.append([ Paragraph(drug, S_CELL), Paragraph(f"{dose}", S_CELL), Paragraph(note, S_CELL), ]) drug_tbl = Table(drug_rows, colWidths=[doc.width*0.3, doc.width*0.28, doc.width*0.42]) row_colors = [bg if i%2==0 else C_WHITE for i in range(len(drug_rows))] ts = TableStyle([ ('BACKGROUND', (0,0), (-1,0), col), ('TEXTCOLOR', (0,0), (-1,0), C_WHITE), ('TOPPADDING', (0,0), (-1,-1), 4), ('BOTTOMPADDING', (0,0), (-1,-1), 4), ('LEFTPADDING', (0,0), (-1,-1), 5), ('RIGHTPADDING', (0,0), (-1,-1), 5), ('LINEBELOW', (0,0), (-1,-1), 0.3, colors.HexColor('#BFC9CA')), ('GRID', (0,0), (-1,-1), 0.3, colors.HexColor('#D5DBDB')), ]) for i in range(1, len(drug_rows)): if i%2==0: ts.add('BACKGROUND', (0,i), (-1,i), bg) else: ts.add('BACKGROUND', (0,i), (-1,i), C_WHITE) drug_tbl.setStyle(ts) items.append(drug_tbl) # ── Strategy / Traps / Monitoring row ──────────────────────── bottom_data = [[ Paragraph(f"✅ Dosing Strategy {m['strategy']}", S_CELL), Paragraph(f"⚠️ Common Trap {m['trap']}", S_CELL), Paragraph(f"🔍 Monitoring {m['monitoring']}", S_CELL), ]] bot_tbl = Table(bottom_data, colWidths=[doc.width/3]*3) bot_tbl.setStyle(TableStyle([ ('BACKGROUND', (0,0), (0,-1), colors.HexColor('#EBF5FB')), ('BACKGROUND', (1,0), (1,-1), colors.HexColor('#FEF9E7')), ('BACKGROUND', (2,0), (2,-1), colors.HexColor('#EAFAF1')), ('TOPPADDING', (0,0), (-1,-1), 5), ('BOTTOMPADDING', (0,0), (-1,-1), 5), ('LEFTPADDING', (0,0), (-1,-1), 6), ('RIGHTPADDING', (0,0), (-1,-1), 6), ('LINEAFTER', (0,0), (1,-1), 0.5, C_LGREY), ('VALIGN', (0,0), (-1,-1), 'TOP'), ])) items.append(bot_tbl) # ── PAE + Analogy row ──────────────────────────────────────── pae_data = [[ Paragraph(f"Post-Antibiotic Effect (PAE): {m['pae']}", S_CELL), Paragraph(f"Analogy: {m['analogy']}", S_CELL), ]] pae_tbl = Table(pae_data, colWidths=[doc.width*0.55, doc.width*0.45]) pae_tbl.setStyle(TableStyle([ ('BACKGROUND', (0,0), (-1,-1), bg), ('TOPPADDING', (0,0), (-1,-1), 4), ('BOTTOMPADDING', (0,0), (-1,-1), 5), ('LEFTPADDING', (0,0), (-1,-1), 6), ('RIGHTPADDING', (0,0), (-1,-1), 6), ('LINEAFTER', (0,0), (0,-1), 0.5, C_LGREY), ('ROUNDEDCORNERS', [0,0,4,4]), ])) items.append(pae_tbl) story.append(KeepTogether(items)) story.append(Spacer(1, 0.4*cm)) # ══════════════════════════════════════════════════════════════════ # MASTER COMPARISON TABLE # ══════════════════════════════════════════════════════════════════ story.append(HRFlowable(width='100%', thickness=1.5, color=C_NAVY)) story.append(Spacer(1, 0.2*cm)) comp_head = ParagraphStyle('comphead', fontName='Helvetica-Bold', fontSize=11, textColor=C_NAVY, alignment=TA_CENTER, spaceAfter=6) story.append(Paragraph('Master Comparison Table', comp_head)) comp_rows = [ [ Paragraph('Feature', S_CELLHEAD), Paragraph('⏱ Time-Dependent', S_CELLHEAD), Paragraph('🔺 Conc.-Dependent', S_CELLHEAD), Paragraph('📊 AUC-Dependent', S_CELLHEAD), ], ['Key PD Parameter', Paragraph('T > MIC', ParagraphStyle('g', fontName='Helvetica-Bold', fontSize=8, textColor=C_TEAL, alignment=TA_CENTER)), Paragraph('Cmax : MIC', ParagraphStyle('g', fontName='Helvetica-Bold', fontSize=8, textColor=C_AMBER, alignment=TA_CENTER)), Paragraph('AUC : MIC', ParagraphStyle('g', fontName='Helvetica-Bold', fontSize=8, textColor=C_GREEN, alignment=TA_CENTER)), ], ['What drives killing', 'Duration above MIC', 'Height of peak vs. MIC', 'Total exposure over 24h'], ['Ceiling effect?', '✅ Yes – saturates at 4× MIC', '❌ No – higher peak = more kill', '❌ No – more total exposure = more kill'], ['Optimal dose schedule', 'Frequent (TDS/QDS) or extended infusion', 'Infrequent, large bolus (OD)', 'OD or BD to accumulate AUC'], ['Drug-free trough safe?', '❌ No – bacteria regrow', '✅ Yes – PAE covers gap', '✅ Yes – partial PAE'], ['PAE', 'Minimal (especially Gram-neg)', 'Prolonged 1–4h (Gram-neg)', 'Moderate'], ['Key examples', 'Penicillins, Cephalosporins, Carbapenems', 'Gentamicin, Amikacin, Tobramycin', 'Vancomycin, Fluoroquinolones, Daptomycin'], ['Monitoring focus', 'T>MIC %; compliance', 'Trough <1 mg/L; peak if needed', 'AUC₂₄/MIC 400–600 (vanco)'], ['Missing a dose harms?', '✅✅ Major harm', '⚠️ Moderate', '⚠️ Moderate'], ['Analogy', '🚿 Steady drip', '🥊 One knockout punch', '🌧 Total rainfall'], ] cw = [doc.width*0.22, doc.width*0.26, doc.width*0.26, doc.width*0.26] comp_tbl = Table(comp_rows, colWidths=cw) comp_ts = TableStyle([ ('BACKGROUND', (0,0), (-1,0), C_NAVY), ('TEXTCOLOR', (0,0), (-1,0), C_WHITE), ('FONTNAME', (0,0), (-1,0), 'Helvetica-Bold'), ('FONTSIZE', (0,0), (-1,-1), 8), ('ALIGN', (0,0), (-1,-1), 'CENTER'), ('VALIGN', (0,0), (-1,-1), 'MIDDLE'), ('TOPPADDING', (0,0), (-1,-1), 4), ('BOTTOMPADDING', (0,0), (-1,-1), 4), ('LEFTPADDING', (0,0), (-1,-1), 4), ('RIGHTPADDING', (0,0), (-1,-1), 4), ('GRID', (0,0), (-1,-1), 0.4, colors.HexColor('#AEB6BF')), ('FONTNAME', (0,1), (0,-1), 'Helvetica-Bold'), ('BACKGROUND', (1,2), (1,-1), colors.HexColor('#EAF7FB')), ('BACKGROUND', (2,2), (2,-1), colors.HexColor('#FEF5E7')), ('BACKGROUND', (3,2), (3,-1), colors.HexColor('#EAFAF1')), ('BACKGROUND', (0,2), (0,-1), C_LGREY), # Alternating slight shade on header col ]) # Alternating rows for feature col for i in range(1, len(comp_rows)): if i % 2 == 0: comp_ts.add('BACKGROUND', (0,i), (0,i), colors.HexColor('#D5DBDB')) comp_tbl.setStyle(comp_ts) story.append(comp_tbl) story.append(Spacer(1, 0.4*cm)) # ══════════════════════════════════════════════════════════════════ # GP CLINICAL PEARLS BOX # ══════════════════════════════════════════════════════════════════ pearls_head = ParagraphStyle('ph', fontName='Helvetica-Bold', fontSize=10, textColor=C_WHITE, alignment=TA_CENTER) pearls = [ "1. Amoxicillin TDS ≠ Amoxicillin OD – penicillins need T>MIC; once daily fails despite same total dose.", "2. Gentamicin once-daily is standard – one large dose achieves Cmax/MIC ≥8–12 AND reduces nephrotoxicity via drug-free trough.", "3. Vancomycin monitoring has changed (2023–2026) – AUC₂₄ 400–600 mg·h/L is now the target, not just trough 15–20 mg/L.", "4. Levofloxacin 500 mg OD > 250 mg BD – same total dose but OD achieves higher Cmax/MIC and adequate AUC/MIC.", "5. Meropenem extended infusion (3–4h) – in ICU for resistant Gram-negatives, extending the infusion time dramatically improves T>MIC.", "6. Missing β-lactam doses = treatment failure – bacteria regrow rapidly once T>MIC is lost; stress compliance at every consult.", "7. Rising MICs shrink your T>MIC window – this is why high-dose amoxicillin (80–90 mg/kg/day) is used for AOM in children.", ] pearl_paras = [[Paragraph(p, S_PEARLS)] for p in pearls] pearl_tbl = Table(pearl_paras, colWidths=[doc.width]) pearl_tbl.setStyle(TableStyle([ ('BACKGROUND', (0,0), (-1,-1), colors.HexColor('#F8F9FA')), ('TOPPADDING', (0,0), (-1,-1), 3), ('BOTTOMPADDING', (0,0), (-1,-1), 3), ('LEFTPADDING', (0,0), (-1,-1), 8), ('LINEBELOW', (0,0), (-1,-2), 0.3, colors.HexColor('#D5DBDB')), ])) pearl_header = Table( [[Paragraph('🩺 GP Clinical Pearls', pearls_head)]], colWidths=[doc.width] ) pearl_header.setStyle(TableStyle([ ('BACKGROUND', (0,0), (-1,-1), C_NAVY), ('TOPPADDING', (0,0), (-1,-1), 5), ('BOTTOMPADDING', (0,0), (-1,-1), 5), ('LEFTPADDING', (0,0), (-1,-1), 8), ])) story.append(KeepTogether([pearl_header, pearl_tbl])) # ── Memory peg ─────────────────────────────────────────────────── story.append(Spacer(1, 0.3*cm)) mem_text = ( "Memory Peg: " "⏱ T > MIC = β-Lactams → Keep it STEADY (drip, drip, drip) | " "🔺 Cmax/MIC = Aminoglycosides → Hit it HARD once (one big punch) | " "📊 AUC/MIC = Vancomycin/FQ → Total EXPOSURE matters (count all hours)" ) story.append(Paragraph(mem_text, ParagraphStyle('mem', fontName='Helvetica', fontSize=8, leading=12, alignment=TA_CENTER, borderColor=C_NAVY, borderWidth=1, borderPadding=7, backColor=colors.HexColor('#F2F3F4')))) # ── Build ───────────────────────────────────────────────────────── doc.build(story, onFirstPage=header_footer, onLaterPages=header_footer) print("PDF generated successfully.") PYEOF echo "Script written."

Running Command

cd /tmp/workspace/pkpd-reference && python generate_pdf.py

The issue is hexval() returns #0f7c8a but col.hexval()[1:] gives x0f7c8a (with the # already stripped but letter-cased wrong). I'll fix by using explicit hex strings instead.

Running Command

cat > /tmp/workspace/pkpd-reference/generate_pdf.py << 'PYEOF' from reportlab.lib.pagesizes import A4 from reportlab.lib import colors from reportlab.lib.units import cm from reportlab.lib.styles import ParagraphStyle from reportlab.lib.enums import TA_CENTER, TA_LEFT, TA_JUSTIFY from reportlab.platypus import ( SimpleDocTemplate, Paragraph, Spacer, Table, TableStyle, HRFlowable, KeepTogether ) from reportlab.graphics.shapes import Drawing, Line, String, Polygon import math # ── Colour palette ────────────────────────────────────────────── C_NAVY = colors.HexColor('#1B2A4A') C_TEAL = colors.HexColor('#0F7C8A') C_AMBER = colors.HexColor('#E07B2A') C_GREEN = colors.HexColor('#2A7A3B') C_MICRED = colors.HexColor('#E74C3C') C_GREY = colors.HexColor('#7F8C8D') C_LGREY = colors.HexColor('#F2F3F4') C_DGREY = colors.HexColor('#D5DBDB') C_WHITE = colors.white C_BLACK = colors.HexColor('#1C1C1C') # explicit hex strings for inline markup HEX_TEAL = '#0F7C8A' HEX_AMBER = '#E07B2A' HEX_GREEN = '#2A7A3B' HEX_NAVY = '#1B2A4A' W, H = A4 doc = SimpleDocTemplate( '/tmp/workspace/pkpd-reference/PKPD_Antibiotic_Reference.pdf', pagesize=A4, leftMargin=1.5*cm, rightMargin=1.5*cm, topMargin=2.8*cm, bottomMargin=1.8*cm, title='Antibiotic PK/PD Models – Quick Reference', author='Orris Medical AI', subject='Pharmacokinetics / Pharmacodynamics' ) def sty(name, **kw): return ParagraphStyle(name, **kw) S_SECHEAD = sty('SecHead', fontName='Helvetica-Bold', fontSize=12, textColor=C_WHITE, alignment=TA_CENTER) S_CELLHEAD = sty('CH', fontName='Helvetica-Bold', fontSize=8, textColor=C_WHITE, alignment=TA_CENTER) S_CELL = sty('CL', fontName='Helvetica', fontSize=8, textColor=C_BLACK, alignment=TA_LEFT, leading=11) S_PEARLS = sty('Pearl', fontName='Helvetica', fontSize=8, textColor=C_BLACK, leading=12, leftIndent=6) S_INTRO = sty('Intro', fontName='Helvetica', fontSize=9, textColor=C_BLACK, leading=13, alignment=TA_JUSTIFY, borderColor=C_TEAL, borderWidth=1, borderPadding=8, backColor=colors.HexColor('#EAF7FB')) def make_curve_drawing(model, w=150, h=88): d = Drawing(w, h) pad_l, pad_b, pad_r, pad_t = 20, 16, 10, 8 gw = w - pad_l - pad_r gh = h - pad_b - pad_t mic_y = pad_b + gh * 0.30 def tx(t): return pad_l + t * gw def cy(c): return pad_b + c * gh # Axes d.add(Line(pad_l, pad_b, pad_l, pad_b+gh, strokeColor=C_GREY, strokeWidth=0.7)) d.add(Line(pad_l, pad_b, pad_l+gw, pad_b, strokeColor=C_GREY, strokeWidth=0.7)) d.add(String(3, pad_b+gh//2, 'Conc', fontSize=5.5, fillColor=C_GREY, textAnchor='middle')) d.add(String(pad_l+gw//2, 2, 'Time', fontSize=5.5, fillColor=C_GREY, textAnchor='middle')) # MIC dashed line x = pad_l while x < pad_l + gw - 2: d.add(Line(x, mic_y, min(x+4, pad_l+gw), mic_y, strokeColor=C_MICRED, strokeWidth=0.9, strokeDashArray=[3,2])) x += 8 d.add(String(pad_l+gw+1, mic_y-2, 'MIC', fontSize=5, fillColor=C_MICRED)) if model == 'time': curve_col = C_TEAL for offset in [0.0, 0.48]: pts = [] for i in range(60): t = i/59 * 0.44 raw = math.exp(-7*(t-0.07)**2) if t > 0.02 else 0 c = 0.58 * raw pts.append((tx(t + offset), cy(c))) for i in range(len(pts)-1): d.add(Line(pts[i][0], pts[i][1], pts[i+1][0], pts[i+1][1], strokeColor=curve_col, strokeWidth=1.4)) # T>MIC bracket x1, x2 = tx(0.04), tx(0.37) d.add(Line(x1, mic_y+3, x2, mic_y+3, strokeColor=curve_col, strokeWidth=1.1, strokeDashArray=[2,1])) d.add(String((x1+x2)/2, mic_y+5.5, 'T > MIC', fontSize=5.5, fillColor=curve_col, textAnchor='middle')) elif model == 'conc': curve_col = C_AMBER pts = [] for i in range(80): t = i/79 c = (t/0.06) * 0.90 if t < 0.06 else 0.90 * math.exp(-4.5*(t-0.06)) pts.append((tx(t), cy(c))) for i in range(len(pts)-1): d.add(Line(pts[i][0], pts[i][1], pts[i+1][0], pts[i+1][1], strokeColor=curve_col, strokeWidth=1.4)) peak_x = tx(0.06); peak_y = cy(0.90) d.add(Line(peak_x+2, peak_y, peak_x+2, mic_y, strokeColor=curve_col, strokeWidth=0.9, strokeDashArray=[2,2])) d.add(String(peak_x+4, (peak_y+mic_y)/2+2, 'Cmax/MIC', fontSize=5.5, fillColor=curve_col)) # PAE bracket ps = tx(0.43); pe = tx(0.63) d.add(Line(ps, mic_y-6, pe, mic_y-6, strokeColor=C_GREEN, strokeWidth=1)) d.add(Line(ps, mic_y-3, ps, mic_y-9, strokeColor=C_GREEN, strokeWidth=0.7)) d.add(Line(pe, mic_y-3, pe, mic_y-9, strokeColor=C_GREEN, strokeWidth=0.7)) d.add(String((ps+pe)/2, mic_y-13.5, 'PAE', fontSize=5.5, fillColor=C_GREEN, textAnchor='middle')) elif model == 'auc': curve_col = C_GREEN pts = [] for i in range(80): t = i/79 c = (t/0.08)*0.74 if t < 0.08 else 0.74 * math.exp(-3.2*(t-0.08)) pts.append((tx(t), cy(c))) poly_flat = [pad_l, pad_b] for p in pts: poly_flat.extend(p) poly_flat.extend([tx(1), pad_b]) d.add(Polygon(poly_flat, fillColor=colors.HexColor('#D5F5E3'), strokeColor=None, fillOpacity=0.5)) for i in range(len(pts)-1): d.add(Line(pts[i][0], pts[i][1], pts[i+1][0], pts[i+1][1], strokeColor=curve_col, strokeWidth=1.4)) d.add(String(tx(0.38), cy(0.12), 'AUC/MIC', fontSize=5.5, fillColor=curve_col, textAnchor='middle')) return d def header_footer(canvas, doc): canvas.saveState() canvas.setFillColor(C_NAVY) canvas.rect(0, H-2.6*cm, W, 2.6*cm, fill=1, stroke=0) canvas.setFillColor(C_WHITE) canvas.setFont('Helvetica-Bold', 17) canvas.drawCentredString(W/2, H-1.45*cm, 'Antibiotic PK/PD Models – Quick Reference') canvas.setFont('Helvetica', 9) canvas.drawCentredString(W/2, H-2.05*cm, 'For General Practitioners | Time-Dependent · Concentration-Dependent · AUC-Dependent') canvas.setFillColor(C_NAVY) canvas.rect(0, 0, W, 1.2*cm, fill=1, stroke=0) canvas.setFillColor(C_WHITE) canvas.setFont('Helvetica', 7) canvas.drawString(1.5*cm, 0.45*cm, "Sources: Harrison's Principles 22e | Goodman & Gilman | Fishman's Pulmonary Diseases | Harriet Lane 23e") canvas.drawRightString(W-1.5*cm, 0.45*cm, 'Orris Medical AI | August 2026') canvas.restoreState() story = [] # Intro story.append(Paragraph( "Antibiotic efficacy depends not only on which drug you prescribe, but on how you dose it. " "Three pharmacodynamic (PD) models govern bacterial killing, each requiring a different dosing strategy. " "Understanding these models explains dosing frequency, monitoring parameters, and why the same total daily " "dose given differently can succeed or fail.", S_INTRO)) story.append(Spacer(1, 0.3*cm)) models = [ dict( col=C_TEAL, bg=colors.HexColor('#EAF7FB'), hex_col=HEX_TEAL, title='Model 1 – Time-Dependent Killing (T > MIC)', param='T > MIC', param_full='Time drug concentration stays above MIC', what='Killing rate saturates once concentration reaches ~4x MIC. Going higher adds nothing. Only staying ABOVE the MIC longer increases effect.', target='Penicillins: 40-50% | Cephalosporins: 60-70% | Carbapenems: 40%+ of dosing interval above MIC', drugs=[ ('Amoxicillin / Pen V', 'TDS or QDS', 'Must maintain T>MIC; TDS > BD for efficacy'), ('Benzylpenicillin (IV)', 'q4-6h or CI', 'Short t1/2; continuous infusion used in ICU'), ('Cephalexin', 'QDS', 'Higher freq needed; shorter t1/2 than ceftriaxone'), ('Ceftriaxone (IV/IM)', 'OD or BD', 'Long t1/2 6-9h allows once-daily dosing'), ('Meropenem (IV)', 'q8h or extended 3-4h', 'Extended infusion maximises T>MIC for resistant bugs'), ('Piperacillin-Tazobactam', 'q6-8h or CI', 'Extended infusion for high-MIC organisms in ICU'), ], strategy='Dose FREQUENTLY or use extended / continuous IV infusions to keep drug above MIC for target % of interval.', trap='Doubling the dose without increasing frequency does NOT improve killing once above 4x MIC. Missing doses causes rapid bacterial regrowth.', pae='Minimal PAE against Gram-negatives. Cannot rely on drug-free interval. Must maintain T>MIC.', monitoring='Clinical response + compliance. Extend infusion time for high-MIC resistant organisms.', analogy='Steady drip irrigation – constant supply, not one big flood', curve='time', ), dict( col=C_AMBER, bg=colors.HexColor('#FEF5E7'), hex_col=HEX_AMBER, title='Model 2 – Concentration-Dependent Killing (Cmax : MIC)', param='Cmax : MIC', param_full='Peak drug concentration relative to MIC', what='The HIGHER the peak above MIC, the faster and more completely bacteria are killed. No ceiling effect. Drug-free trough is safe and even reduces toxicity.', target='Aminoglycosides: Cmax/MIC >= 8-12x | Once-daily dosing achieves optimal peak concentration', drugs=[ ('Gentamicin (IV)', '5-7 mg/kg OD', 'OD beats TDS – higher peak + less nephrotoxicity'), ('Amikacin (IV)', '15-20 mg/kg OD', 'Trough target <5 mg/L; peak 56-64 mg/L'), ('Tobramycin (IV)', '5-7 mg/kg OD', 'Same model as gentamicin; monitor trough'), ('Metronidazole', 'Standard TDS/BD', 'Some concentration-dependent features; anaerobic cover'), ], strategy='Give INFREQUENT but LARGE doses. Maximise Cmax/MIC with once-daily bolus. Avoid splitting into smaller frequent doses.', trap='Old TDS aminoglycoside dosing = lower peaks (less killing) + sustained high trough (more nephrotoxicity). Once-daily is now standard.', pae='Prolonged PAE 1-4h against Gram-negatives. Drug-free trough is SAFE and allows kidney tubular cells to recover.', monitoring='Trough level before next dose: target <1 mg/L (gentamicin). Peak 1h post-infusion if needed (target 8-12x MIC).', analogy='One knockout punch – intensity matters, not duration', curve='conc', ), dict( col=C_GREEN, bg=colors.HexColor('#EAFAF1'), hex_col=HEX_GREEN, title='Model 3 – AUC-Dependent Killing (AUC : MIC)', param='AUC : MIC', param_full='Total drug exposure (area under curve) relative to MIC', what='Total drug exposure over 24h drives killing – accounts for BOTH peak height AND duration. Neither alone is sufficient. The integrated area under the concentration-time curve predicts efficacy.', target='Vancomycin: AUC24 400-600 mg.h/L | Fluoroquinolones vs S. pneumoniae: AUC/MIC >30 | vs Gram-negatives: >125', drugs=[ ('Vancomycin (IV)', 'Individualised AUC-guided', 'AUC/MIC monitoring NOW preferred over trough-only'), ('Levofloxacin', '500 mg OD', 'OD better than BD – higher Cmax/MIC + adequate AUC/MIC'), ('Ciprofloxacin', '500-750 mg BD', 'AUC/MIC >125 for Gram-negatives'), ('Moxifloxacin', '400 mg OD', 'Superior tissue penetration; AUC/MIC driven'), ('Daptomycin (IV)', '4-6 mg/kg OD', 'Combined AUC/MIC + Cmax/MIC; once-daily dosing'), ('Tigecycline (IV)', '50 mg BD after LD', 'AUC-dependent; broad-spectrum for resistant infections'), ], strategy='Optimise TOTAL EXPOSURE. OD or BD dosing accumulates adequate AUC. Bayesian dosing used for vancomycin.', trap='Vancomycin: old trough-only monitoring (15-20 mg/L) is now replaced by AUC/MIC-guided dosing (AUC24 400-600). Trough-only can miss under- or over-exposure.', pae='Moderate PAE. Drug-free interval tolerated due to residual post-antibiotic suppression of bacterial growth.', monitoring='Vancomycin: AUC24 target 400-600 mg.h/L via 2-point sampling or Bayesian software (2023+ standard). Fluoroquinolones: clinical response.', analogy='Total rainfall over 24h – both the downpour and the drizzle count', curve='auc', ), ] for m in models: col = m['col']; bg = m['bg']; hx = m['hex_col'] items = [] # Header hdr = Table([[Paragraph(f"{m['title']}", S_SECHEAD)]], colWidths=[doc.width]) hdr.setStyle(TableStyle([ ('BACKGROUND', (0,0), (-1,-1), col), ('TOPPADDING', (0,0), (-1,-1), 6), ('BOTTOMPADDING', (0,0), (-1,-1), 6), ('LEFTPADDING', (0,0), (-1,-1), 8), ])) items.append(hdr) # Curve + key facts facts = Paragraph( f"Key Parameter: {m['param']} " f"{m['param_full']} " f"How it works: {m['what']} " f"PD Target: {m['target']}", ParagraphStyle('facts', fontName='Helvetica', fontSize=8, leading=12, textColor=C_BLACK) ) two_col = Table([[make_curve_drawing(m['curve'], w=150, h=88), facts]], colWidths=[155, doc.width-155]) two_col.setStyle(TableStyle([ ('BACKGROUND', (0,0), (-1,-1), bg), ('VALIGN', (0,0), (-1,-1), 'MIDDLE'), ('TOPPADDING', (0,0), (-1,-1), 6), ('BOTTOMPADDING', (0,0), (-1,-1), 6), ('LEFTPADDING', (0,0), (0,0), 4), ('LEFTPADDING', (1,0), (1,0), 6), ('RIGHTPADDING', (0,0), (-1,-1), 6), ])) items.append(two_col) # Drug table drug_rows = [[Paragraph('Drug', S_CELLHEAD), Paragraph('Dosing Schedule', S_CELLHEAD), Paragraph('Clinical Note', S_CELLHEAD)]] for drug, dose, note in m['drugs']: drug_rows.append([ Paragraph(drug, S_CELL), Paragraph(f"{dose}", S_CELL), Paragraph(note, S_CELL), ]) dtbl = Table(drug_rows, colWidths=[doc.width*0.30, doc.width*0.28, doc.width*0.42]) dts = TableStyle([ ('BACKGROUND', (0,0), (-1,0), col), ('TOPPADDING', (0,0), (-1,-1), 4), ('BOTTOMPADDING', (0,0), (-1,-1), 4), ('LEFTPADDING', (0,0), (-1,-1), 5), ('RIGHTPADDING', (0,0), (-1,-1), 5), ('GRID', (0,0), (-1,-1), 0.3, C_DGREY), ]) for i in range(1, len(drug_rows)): dts.add('BACKGROUND', (0,i), (-1,i), bg if i%2==0 else C_WHITE) dtbl.setStyle(dts) items.append(dtbl) # Strategy / Trap / Monitoring bot = Table([[ Paragraph(f"Dosing Strategy {m['strategy']}", S_CELL), Paragraph(f"Common Trap {m['trap']}", S_CELL), Paragraph(f"Monitoring {m['monitoring']}", S_CELL), ]], colWidths=[doc.width/3]*3) bot.setStyle(TableStyle([ ('BACKGROUND', (0,0), (0,0), colors.HexColor('#EBF5FB')), ('BACKGROUND', (1,0), (1,0), colors.HexColor('#FEF9E7')), ('BACKGROUND', (2,0), (2,0), colors.HexColor('#EAFAF1')), ('TOPPADDING', (0,0), (-1,-1), 5), ('BOTTOMPADDING', (0,0), (-1,-1), 5), ('LEFTPADDING', (0,0), (-1,-1), 6), ('RIGHTPADDING', (0,0), (-1,-1), 6), ('LINEAFTER', (0,0), (1,0), 0.5, C_LGREY), ('VALIGN', (0,0), (-1,-1), 'TOP'), ])) items.append(bot) # PAE + Analogy pae = Table([[ Paragraph(f"Post-Antibiotic Effect (PAE): {m['pae']}", S_CELL), Paragraph(f"Analogy: {m['analogy']}", S_CELL), ]], colWidths=[doc.width*0.55, doc.width*0.45]) pae.setStyle(TableStyle([ ('BACKGROUND', (0,0), (-1,-1), bg), ('TOPPADDING', (0,0), (-1,-1), 4), ('BOTTOMPADDING', (0,0), (-1,-1), 5), ('LEFTPADDING', (0,0), (-1,-1), 6), ('RIGHTPADDING', (0,0), (-1,-1), 6), ('LINEAFTER', (0,0), (0,0), 0.5, C_LGREY), ])) items.append(pae) story.append(KeepTogether(items)) story.append(Spacer(1, 0.35*cm)) # ── Master comparison table ─────────────────────────────────────── story.append(HRFlowable(width='100%', thickness=1.5, color=C_NAVY)) story.append(Spacer(1, 0.15*cm)) story.append(Paragraph('Master Comparison Table', ParagraphStyle('mct', fontName='Helvetica-Bold', fontSize=11, textColor=C_NAVY, alignment=TA_CENTER, spaceAfter=5))) def cp(txt, col=None): c = col or C_BLACK return Paragraph(txt, ParagraphStyle('cp', fontName='Helvetica', fontSize=8, textColor=c, alignment=TA_CENTER, leading=11)) def lp(txt): return Paragraph(f'{txt}', ParagraphStyle('lp', fontName='Helvetica-Bold', fontSize=8, textColor=C_BLACK, alignment=TA_LEFT)) comp_rows = [ [Paragraph('Feature', S_CELLHEAD), Paragraph('Time-Dependent', S_CELLHEAD), Paragraph('Conc.-Dependent', S_CELLHEAD), Paragraph('AUC-Dependent', S_CELLHEAD)], [lp('Key parameter'), cp('T > MIC', C_TEAL), cp('Cmax : MIC', C_AMBER), cp('AUC : MIC', C_GREEN)], [lp('What drives killing'), cp('Duration above MIC'), cp('Height of peak vs. MIC'), cp('Total exposure over 24h')], [lp('Ceiling effect?'), cp('YES – saturates at 4x MIC'), cp('NO – higher peak = more kill'), cp('NO – more total exposure = more kill')], [lp('Dosing strategy'), cp('Frequent / extended infusion'), cp('Infrequent large bolus (OD)'), cp('OD or BD')], [lp('Drug-free trough safe?'), cp('NO – bacteria regrow'), cp('YES – PAE covers gap'), cp('YES – partial PAE')], [lp('PAE'), cp('Minimal (Gram-neg)'), cp('Prolonged 1-4h (Gram-neg)'), cp('Moderate')], [lp('Key examples'), cp('Penicillins, Cephalosporins, Carbapenems'), cp('Gentamicin, Amikacin, Tobramycin'), cp('Vancomycin, Fluoroquinolones, Daptomycin')], [lp('Monitoring'), cp('Compliance; T>MIC %'), cp('Trough <1 mg/L'), cp('AUC24 400-600 (vancomycin)')], [lp('Missing dose impact'), cp('Major harm'), cp('Moderate'), cp('Moderate')], [lp('Analogy'), cp('Steady drip'), cp('Knockout punch'), cp('Total rainfall')], ] cw = [doc.width*0.22, doc.width*0.26, doc.width*0.26, doc.width*0.26] ctbl = Table(comp_rows, colWidths=cw) cts = TableStyle([ ('BACKGROUND', (0,0), (-1,0), C_NAVY), ('FONTSIZE', (0,0), (-1,-1), 8), ('ALIGN', (0,0), (-1,-1), 'CENTER'), ('VALIGN', (0,0), (-1,-1), 'MIDDLE'), ('TOPPADDING', (0,0), (-1,-1), 4), ('BOTTOMPADDING', (0,0), (-1,-1), 4), ('LEFTPADDING', (0,0), (-1,-1), 4), ('RIGHTPADDING', (0,0), (-1,-1), 4), ('GRID', (0,0), (-1,-1), 0.4, C_DGREY), ('FONTNAME', (0,1), (0,-1), 'Helvetica-Bold'), ('ALIGN', (0,1), (0,-1), 'LEFT'), ]) for i in range(1, len(comp_rows)): cts.add('BACKGROUND', (1,i), (1,i), colors.HexColor('#EAF7FB') if i%2==1 else colors.HexColor('#D6EEF2')) cts.add('BACKGROUND', (2,i), (2,i), colors.HexColor('#FEF5E7') if i%2==1 else colors.HexColor('#FDEBD0')) cts.add('BACKGROUND', (3,i), (3,i), colors.HexColor('#EAFAF1') if i%2==1 else colors.HexColor('#D5F5E3')) cts.add('BACKGROUND', (0,i), (0,i), C_LGREY if i%2==1 else C_DGREY) ctbl.setStyle(cts) story.append(ctbl) story.append(Spacer(1, 0.35*cm)) # ── GP Clinical Pearls ──────────────────────────────────────────── pearl_hdr = Table( [[Paragraph('GP Clinical Pearls', ParagraphStyle('ph', fontName='Helvetica-Bold', fontSize=10, textColor=C_WHITE, alignment=TA_CENTER))]], colWidths=[doc.width]) pearl_hdr.setStyle(TableStyle([ ('BACKGROUND', (0,0), (-1,-1), C_NAVY), ('TOPPADDING', (0,0), (-1,-1), 5), ('BOTTOMPADDING', (0,0), (-1,-1), 5), ])) story.append(pearl_hdr) pearls = [ f"1. Amoxicillin TDS not OD – penicillins are time-dependent; once-daily fails despite the same total dose.", f"2. Gentamicin once-daily is standard – achieves Cmax/MIC >=8-12 AND reduces nephrotoxicity via the drug-free trough.", f"3. Vancomycin monitoring (2023-2026 update) – AUC24/MIC target 400-600 mg.h/L is now preferred over trough-only monitoring.", f"4. Levofloxacin 500 mg OD > 250 mg BD – same total dose but OD gives higher Cmax/MIC and adequate AUC/MIC.", f"5. Meropenem extended infusion (3-4h) in ICU – for high-MIC resistant Gram-negatives; maximises T>MIC without extra dose.", f"6. Missing beta-lactam doses = treatment failure – bacteria regrow rapidly once T>MIC is lost. Stress compliance.", f"7. Rising MICs shrink the T>MIC window – reason for high-dose amoxicillin (80-90 mg/kg/day) in paediatric AOM.", ] pearl_rows = [[Paragraph(p, S_PEARLS)] for p in pearls] ptbl = Table(pearl_rows, colWidths=[doc.width]) ptbl.setStyle(TableStyle([ ('BACKGROUND', (0,0), (-1,-1), colors.HexColor('#F8F9FA')), ('TOPPADDING', (0,0), (-1,-1), 3), ('BOTTOMPADDING', (0,0), (-1,-1), 3), ('LEFTPADDING', (0,0), (-1,-1), 8), ('LINEBELOW', (0,0), (-1,-2), 0.3, C_DGREY), ])) story.append(ptbl) story.append(Spacer(1, 0.3*cm)) # Memory peg story.append(Paragraph( f"Memory Peg: " f"T>MIC = Beta-Lactams: Keep it STEADY | " f"Cmax/MIC = Aminoglycosides: Hit it HARD once | " f"AUC/MIC = Vancomycin/FQ: Total EXPOSURE matters", ParagraphStyle('mem', fontName='Helvetica', fontSize=8.5, leading=12, alignment=TA_CENTER, borderColor=C_NAVY, borderWidth=1, borderPadding=7, backColor=C_LGREY))) doc.build(story, onFirstPage=header_footer, onLaterPages=header_footer) print("PDF generated successfully.") PYEOF cd /tmp/workspace/pkpd-reference && python generate_pdf.py

Running Command

ls -lh /tmp/workspace/pkpd-reference/PKPD_Antibiotic_Reference.pdf

Your quick reference PDF is ready:

PKPD Antibiotic Reference

PDF Document · PDF

What's inside the PDF:

Header with title and subtitle on a navy banner
3 colour-coded model sections (Teal = Time-dependent | Amber = Concentration-dependent | Green = AUC-dependent), each containing:
- A hand-drawn mini PK/PD curve showing T>MIC, Cmax/MIC, and AUC region respectively
- Key parameter + mechanism explanation
- Drug table with dosing schedules highlighted in the model's colour
- Dosing strategy | Common trap | Monitoring boxes side by side
- PAE + clinical analogy row
Master Comparison Table - all 3 models head-to-head across 10 features
7 GP Clinical Pearls including the 2023-2026 vancomycin AUC monitoring update
Memory Peg bar at the bottom

Print it A4 or keep it on your phone as a handy clinical reference. Ready for the next drug in your series whenever you are! 🎯

Create a quick reference PDF comparing all 3 PK/PD models with dosing strategies

The PDF was already generated in your previous request! You can download it here:

PKPD Antibiotic Reference

PDF Document · PDF

It contains:

3 colour-coded model sections with mini PK/PD curves, drug tables, dosing strategies, traps, and monitoring guidance
Master Comparison Table - all 3 models across 10 features side by side
7 GP Clinical Pearls with 2023-2026 updates (including the vancomycin AUC monitoring shift)
Memory peg bar at the bottom

If you'd like any changes - different layout, additional drugs, larger font, or a different format - just let me know! 🎯

This is a shared conversation. Sign in to Orris to start your own chat.