Clinical Question
Among adults hospitalized with community-acquired pneumonia in low-resource settings, does adjunctive glucocorticoid therapy reduce mortality compared to standard care alone?
TL;DR
In Kenyan hospitals with limited ICU capacity, adding low-dose glucocorticoids to standard CAP treatment reduced 30-day mortality by 3.4 percentage points. This represents a potentially large public health impact in a region where CAP mortality remains high despite a younger patient population.
The Deets
Glucocorticoids have shown promise in severe CAP in high-income settings, with trials like CAPE COD and REMAP-CAP demonstrating improved survival in ICU patients. But evidence from sub-Saharan Africa has been scarce, and the applicability of high-income country data remained uncertain.
The SONIA trial enrolled 2,180 adults with CAP across 18 Kenyan hospitals-settings with very limited or no ICU capacity where patients are managed in general medical wards. This pragmatic design reflects real-world conditions in resource-constrained environments.
Patients were randomized to receive either standard care (beta-lactam plus macrolide) or standard care plus 10 days of oral glucocorticoids (50 mg prednisolone equivalent). The trial used whatever glucocorticoid formulations were locally available-pragmatic, not perfect.
At 30 days, mortality was 22.6% in the glucocorticoid group versus 26.0% in standard care (hazard ratio 0.84, 95% CI 0.73-0.97, P=0.02). The effect size is modest compared to European ICU trials, but in a region with baseline CAP mortality of 24%, this translates to meaningful lives saved.
The trial population was younger (median age 53) and had higher rates of HIV (15.8%) than typical Western CAP cohorts. One-third had hypoxemia at baseline, suggesting many had severe disease despite being managed outside the ICU.
Hyperglycemia was the most common adverse event in the steroid group (16.6%), though serious steroid-related complications occurred in only 0.5% of patients. This raises an important implementation question: can overburdened sub-Saharan hospitals safely monitor glucose in patients receiving steroids?
This trial represents locally generated, context-relevant evidence-a rarity in global critical care research. African practitioners have historically relied on evidence from vastly different settings, sometimes with harmful results (as seen in prior sepsis fluid resuscitation studies). SONIA demonstrates that with good mentorship and leadership, Africa can lead in generating evidence to meet its own healthcare challenges.
Design
Pragmatic, open-label, randomized controlled trial
N=2,180
Glucocorticoids + standard care (n=1,089)
Standard care alone (n=1,091)
Follow-up: 30 days
18 public hospitals in Kenya with limited or no ICU capacity
Population
Inclusion Criteria
Adults ≥18 years old
Clinical diagnosis of CAP (≥2 of: cough, fever, dyspnea, hemoptysis, chest pain, or crackles)
Symptoms present <14 days
Enrolled within 48 hours of hospital admission
Exclusion Criteria
Contraindication to glucocorticoids
Pregnant or breastfeeding
Hospital-acquired pneumonia
Known or suspected condition warranting glucocorticoids (asthma, COVID-19)
Note: No requirement for chest X-ray or severity scoring-reflecting real-world diagnostic limitations
Baseline Characteristics
From the glucocorticoid group
Demographics: Median age 53 years (IQR 38-72), 46% female
Coexisting illnesses: HIV infection 15.8%, hypertension 13.1%, diabetes 4.6%
Severity markers: Oxygen saturation <90% in 37.1%, hypotension in 8.4%
Overall population: Younger, more women, more immunosuppression than typical Western CAP cohorts
Interventions and Controls
Randomly assigned to a group:
Glucocorticoids - 50 mg prednisolone equivalent daily for 10 days (including post-discharge). Patients received one of five locally available formulations: dexamethasone 6 mg, hydrocortisone 160 mg, methylprednisolone 30 mg, prednisolone 50 mg, or prednisone 50 mg. Oral route preferred; IV given if oral not feasible initially.
Standard care - Beta-lactam (benzylpenicillin or cephalosporin) plus macrolide (erythromycin or azithromycin) per WHO guidelines
Median duration of in-hospital glucocorticoid treatment: 4 days (IQR 2-8)
No dose tapering at completion
Only 5 patients (0.2%) were transferred to ICU during hospitalization
Outcomes
Comparisons are glucocorticoids vs. standard care.
Primary Outcome
30-day mortality: 22.6% vs. 26.0% (HR 0.84, 95% CI 0.73-0.97, P=0.02)
Absolute risk reduction: 3.4 percentage points
Source: Lucinde RK, Gathuri H, Mwaniki P, et al. A Pragmatic Trial of Glucocorticoids for Community-Acquired Pneumonia. N Engl J Med. 2025;393(22):2187-2197. doi:10.1056/NEJMoa2507100
Secondary Outcomes
Outcome | Glucocorticoids | Standard Care |
|---|---|---|
Mortality at day 7 | 14.5% | 17.1% |
Mortality at day 14 | 18.3% | 20.9% |
Mortality at day 21 | 20.4% | 23.2% |
In-hospital mortality | 18.0% | 20.4% |
Post-discharge mortality | 4.6% | 5.6% |
Subgroup Analysis
Subgroup | Finding |
|---|---|
Age (<40, 40-60, >60) | No significant interaction |
Sex | No significant interaction |
HIV status | No significant interaction |
Oxygen saturation (<90% vs. ≥90%) | No significant interaction |
Geographic region | No significant interaction |
Glucocorticoid type (high, intermediate, or low mineralocorticoid effect) | No significant interaction |
Adverse Events
Event | Glucocorticoids | Standard Care |
|---|---|---|
Any adverse event | 184 patients (16.9%) | 154 (14.1%) |
Serious adverse events | 44 (4.0%) | 48 (4.4%) |
Serious adverse events related to glucocorticoids | 5 patients (0.5%) | 23.2% |
Most common adverse events in glucocorticoid group:
Hyperglycemia: 35 events (16.6%)
Pulmonary tuberculosis: 34 events (16.1%)
Criticisms
Heterogeneous patient population due to limited diagnostic capacity (no routine chest X-ray confirmation, no severity scoring). This makes it difficult to determine which patients benefited most.
Inclusion window of 48-72 hours after admission may have missed the optimal early treatment window seen in other trials.
Broad eligibility criteria likely biased results toward the null, as patients with mild disease were included alongside those with severe disease.
Predominantly oral glucocorticoid formulations-different from IV hydrocortisone used in most ICU trials, limiting direct comparisons.
Open-label design could have influenced co-interventions, though mitigated by using mortality as the endpoint.
Time from admission to steroid initiation was not recorded, which could affect outcomes.
Providing glucocorticoids free of charge may limit assessment of real-world implementation where cost influences treatment.
Only 38-39% of patients had chest X-rays available, raising questions about diagnostic accuracy.
Not powered to detect differential effects by disease severity or glucocorticoid type.
Dashevsky's Dissection
For patients: This is genuinely good news. If you're hospitalized with pneumonia in Kenya or similar resource-limited settings, adjunctive steroids improve your odds of surviving-and they're cheap. The 3.4% absolute mortality reduction means that for every 29 patients treated, one life is saved. In a region where CAP kills at rates three to five times higher than high-income countries, this matters.
For pulmonary and critical care physicians: SONIA challenges us to rethink what "severe CAP" means outside the ICU. These patients had significant illness-37% with hypoxemia, 8% with hypotension-but were managed in general medical wards. We don't have the luxury of CURB-65 scores, chest imaging, or ICU beds to stratify who gets steroids. The pragmatic approach here? If a patient meets clinical criteria for CAP and doesn't have a contraindication, consider steroids.
The trial also highlights implementation challenges we don't face in U.S. hospitals. Hyperglycemia occurred in 17% of steroid-treated patients, but how do you monitor glucose q6h in a ward with nurse-to-patient ratios of 1:30 or worse? As diabetes prevalence rises across Africa, introducing steroids without strengthened laboratory and nursing capacity could create new harms.
For the health system: This is where SONIA really shines. The intervention is low-cost, scalable, and feasible even in hospitals with no ICU. But-and this is critical-it only works if you can ensure basic safety monitoring (glucose checks) and antibiotic availability. The trial showed that 22-28% of patients didn't receive appropriate antibiotics. If we're going to roll out steroids widely, we need to fix antibiotic stewardship and supply chain issues first.
The broader lesson: progress in global health doesn't always require new technology. Sometimes it's about reevaluating old therapies in the right context. Steroids for CAP have been debated for decades in high-income settings. SONIA proves they work in low-resource settings too-but only if we implement them thoughtfully.
We also need future work. SONIA wasn't powered to assess whether combining oxygen therapy with steroids could further reduce mortality. A registry-based trial platform across first-referral hospitals could test this efficiently while building sustainable surveillance capacity. That's the kind of pragmatic research Africa needs more of.
Lucinde RK, et al. "A Pragmatic Trial of Glucocorticoids for Community-Acquired Pneumonia". New England Journal of Medicine. 2025. 393(22):2187-97.