Pneumonia in Pakistan: the national guideline changed in March 2026, and the macrolide is gone

If you learned the Pakistan Chest Society's pneumonia guideline, the first-line box for a previously healthy outpatient read amoxicillin, or a macrolide. Doctors reading it against Pakistani pneumococcal data had to work out for themselves that only one of those two still held.

The Society published a new guideline in March 2026 — Community Acquired Bacterial Pneumonia — and it makes the call itself. The outpatient first line is now amoxicillin or doxycycline, and macrolide monotherapy is ruled out by name:

Avoid macrolide monotherapy due to high resistance (macrolide resistance >30% in S. pneumoniae).

Azithromycin has not vanished. It stays as the second agent beside a beta-lactam in admitted patients. What it is no longer is a drug you send an outpatient home on by itself.

The new algorithm, in full

A · Outpatient, low risk — CRB-65 or CURB-65 of 0–1, no drug-resistant-pathogen risk:

• Amoxicillin 1 g PO three times daily, or doxycycline 100 mg PO twice daily, for 5–7 days
• Co-amoxiclav 625 mg PO three times daily where there are mild comorbidities
• Cefpodoxime 200 mg PO twice daily for 5 days

B · Outpatient, moderate risk or drug-resistant-pathogen risk factors:

• Co-amoxiclav PO, or doxycycline 100 mg PO twice daily
• Levofloxacin 750 mg PO once daily for 5–7 days
• Cefuroxime 500 mg PO twice daily plus azithromycin 500 mg once daily
• Cefpodoxime 200 mg PO twice daily for 5 days

C · Inpatient, non-ICU: ceftriaxone 1–2 g IV once daily plus azithromycin 500 mg IV or PO once daily; levofloxacin 750 mg IV once daily if beta-lactam allergic.

D · Severe or ICU, or drug-resistant-pathogen risk: cefepime 2 g IV eight-hourly plus azithromycin 500 mg IV once daily; or piperacillin-tazobactam 4.5 g IV six-hourly plus levofloxacin 750 mg IV once daily; or carbapenem monotherapy.

Add vancomycin 15 mg/kg IV twice daily or linezolid 600 mg IV twice daily only if MRSA is suspected. If Pseudomonas is suspected, avoid ceftriaxone and use an anti-pseudomonal beta-lactam.

The branch point is resistance risk, not comorbidity

This is the structural change in the 2026 edition. What decides whether your patient goes into box A or box B is not simply "is there a comorbidity" — it is whether they carry risk factors for a drug-resistant pathogen:

• Hospitalisation within the past 90 days
• Antibiotics within the past 3 months
• Chronic lung disease — COPD, bronchiectasis
• Structural lung disease, chronic kidney disease, or malignancy
• Tube feeding or aspiration risk
• Known colonisation with a multidrug-resistant organism

Two or more of these call for extended-spectrum or dual therapy. The reason is a number worth remembering: Pakistani surveillance puts drug-resistant organisms — ESBL-producing Klebsiella pneumoniae or Pseudomonas aeruginosa — in 40–45% of hospitalised CAP patients. The guideline attributes part of that to healthcare-associated strains being misclassified as community-acquired, which is precisely why it asks for a deliberate risk assessment instead of a default regimen.

Note the second-order point. Antibiotics are sold over the counter throughout Pakistan, so "antibiotics in the past 3 months" is a question the patient may answer wrongly unless you ask about pharmacy purchases specifically. That single question can move a patient from box A to box B.

What the resistance data says

The guideline's own figures: macrolide resistance in S. pneumoniae "exceeds 25%, which is the threshold used in international guidelines to discourage monotherapy" — and it observes that macrolides "remain commonly prescribed empirically, particularly in outpatient settings". For H. influenzae: ampicillin resistance 18–25%, macrolide resistance around 15–20%, co-amoxiclav and cefuroxime above 95% susceptible.

The published country susceptibility series, 2015–17, CLSI breakpoints:

For Haemophilus influenzae over the same period: co-amoxiclav 100%, levofloxacin 77.1%, moxifloxacin 75.4%, co-trimoxazole 41%.

The line that gets misread

Penicillin sits at 23.4% susceptible in the same dataset where amoxicillin is "high, maintained". Reading those two lines together is the commonest misinterpretation of a pneumococcal sensitivity report, and it leads people to abandon a drug that still works.

The gap is a breakpoint artefact. CLSI applies stricter thresholds to oral penicillin than to amoxicillin at adequate dose in non-meningeal infection. Different drug, different breakpoint, same organism. Do not discard amoxicillin because the penicillin line looks bad.

The fluoroquinolone problem is not microbiological

Respiratory fluoroquinolones work well enough against these organisms — 75–77% against H. influenzae — and the 2026 algorithm offers levofloxacin in three of its four boxes. The guideline attaches its own caution: avoid them in patients at risk of QT prolongation or with a cardiovascular history. It also lists fluoroquinolone overuse as one of four markers of inappropriate empiric therapy.

The bigger reason to hesitate in Pakistan has nothing to do with pneumococcus. Fluoroquinolones have antimycobacterial activity, and Pakistan carries one of the world's highest tuberculosis burdens. An empiric respiratory fluoroquinolone in a patient whose "pneumonia" is actually TB will partially treat it — blunting the clinical picture, delaying the diagnosis, and selecting for fluoroquinolone-resistant TB.

No international CAP guideline carries that caveat, because they are written for settings where TB is rare. Subacute onset, weight loss, night sweats, haemoptysis, upper-lobe or cavitating changes, or failure to respond to an appropriate antibiotic should all prompt sputum testing for TB rather than a second course of antibiotics.

Before any of that: is it pneumonia?

A demonstrable infiltrate on chest radiograph or other imaging, with or without supporting microbiology, is required for the diagnosis. A febrile cough with crackles and a clear film is not pneumonia. Treating it as such accounts for a large share of the avoidable antibiotic use behind those susceptibility figures.

Severity: CURB-65, CRB-65, and where both under-call it

CURB-65 — confusion, urea above 7 mmol/L, respiratory rate 30 or more, systolic below 90 or diastolic 60 or below, age 65 or over. Score 0–1 is low risk and outpatient management; 2 means considering a short stay or close monitoring; 3–5 is high risk, hospitalisation or ICU.

CRB-65 drops the urea, and the guideline names it as particularly useful in rural Pakistani setups and Basic Health Units without immediate lab access — a score you can complete with a watch and a blood-pressure cuff.

Then the caveats, which are the most locally useful paragraph in the chapter:

• Young patients with multi-lobar disease may be under-triaged by CURB-65.
• Malnourished or diabetic patients may do worse despite low scores.
• CRP and procalcitonin are rarely available at point of care here, so biomarker-assisted triage is mostly theoretical.

Oxygen saturation below 90%, confusion, or a respiratory rate of 30 or more should trigger escalation regardless of score. Some hospitals in Karachi now pair CRB-65 with SpO₂ below 90% as a single red flag for immediate referral.

And a number that should make anyone cautious about treating the score as the decision: a nationwide study found the CURB-65 recommendation and the physician's actual admission decision disagreed in almost four out of every ten patients.

How long, and what the audit found

Outpatients: 5–7 days if stable and improving. Inpatients: 7–10 days depending on severity. Courses of five days or fewer are reserved for mild to moderate disease with procalcitonin guidance and rapid recovery — which in most Pakistani hospitals means it is not an option.

The guideline cites a 2022 multi-centre audit that is worth reading twice:

• 39% of CAP patients received empiric therapy that did not match the guideline
• Patients on inappropriate empiric treatment had 2.5 times the mortality
• Duration exceeded recommended limits in 42% of cases
• Where stewardship programmes were implemented, inappropriate antibiotic use fell 25% and stays got shorter

In-hospital mortality in a 2022–2024 five-province cohort was 9.5% overall — 11.2% in Sindh against 7.8% in Punjab.

Three things about antibiotics here that no international guideline accounts for

Antibiotics are sold without prescription. The Drug Act of 1967 prohibits it, but enforcement is inadequate and most classes are obtainable on a simple request. So prior antibiotic exposure — which now moves your patient between algorithm boxes — has to be asked about specifically.

Pakistan is the third-highest antibiotic-consuming low- and middle-income country. That consumption is the engine behind the numbers above.

Substandard products are a real confounder. Antibiotics account for 16.9% of all products reported as substandard or falsified in Pakistan. A patient who has apparently failed a correct antibiotic at a correct dose may have received an inadequate product rather than harboured a resistant organism.

On sending cultures

"Empiric regimens are often not adjusted based on culture or clinical response" is one of the guideline's four markers of inappropriate therapy. The national susceptibility picture this guideline rests on is thin precisely because cultures are sent inconsistently, and thinner still outside Karachi, Lahore and Islamabad. Sending them in admitted patients — and acting on them at 48–72 hours — is how the next edition gets better numbers.

Bottom line

Confirm the infiltrate. Score with CURB-65, or CRB-65 if there is no lab, and escalate on saturation, confusion or respiratory rate whatever the score says. Count the drug-resistant-pathogen risk factors before you choose, and ask specifically about over-the-counter purchases. For a low-risk outpatient the answer is amoxicillin or doxycycline — not azithromycin. Do not abandon amoxicillin because penicillin looks resistant. Think twice before a respiratory fluoroquinolone, because of TB rather than pneumococcus. Stop on time. And send cultures in admitted patients, because someone has to.

Our pneumonia pathway walks the March 2026 algorithm step by step, with the risk factors, doses and durations in one place. Also in the section: tuberculosis for the differential above, urinary tract infection, type 2 diabetes and hypertension.

ReviseFCPS1 is built by doctors in Pakistan, for doctors in Pakistan. All our clinical tools and pathways are free. If you are preparing for FCPS Part 1, start with a free account — 10,176 doctor-verified MCQs and 8,885 real exam recalls, full access from PKR 500.