In Uganda, a newborn with sepsis is typically started on ampicillin and gentamicin, the regimen recommended by the World Health Organization for resource-limited settings. But blood culture data from the country's main referral hospitals suggest that in many cases, the bacteria causing the infection are resistant to one or both drugs. The result is a quiet, deadly mismatch between protocol and pathogen.

The Gap Between Guideline and Organism

WHO's first-line recommendation for neonatal sepsis is ampicillin plus gentamicin. Ugandan neonatal units follow this protocol almost universally, because it is affordable, widely available, and backed by decades of use. The guideline was designed for settings where culture capacity is scarce and where empiric therapy must be safe, cheap, and simple.

Yet culture data from Uganda's referral hospitals tell a different story. At Mulago National Referral Hospital in Kampala, isolates from neonatal blood cultures frequently show resistance to ampicillin and gentamicin. Clinicians there describe a familiar bind: they know the first-line drugs may fail, but they lack the microbiology to justify a change, and second-line options are scarce.

The gap is not a failure of individual prescribers. It is a structural mismatch between a global guideline and local epidemiology. As one related piece on this site argued about asthma care, treatment often follows the formulary rather than the diagnostic evidence. In neonatal sepsis, that gap can be fatal within days.

What the Blood Cultures Actually Grow

Klebsiella species and Staphylococcus aureus dominate neonatal blood culture isolates in Uganda. Many Klebsiella strains are resistant to ampicillin and gentamicin, the two drugs that form the backbone of first-line therapy. Studies from Mulago and Mbale Regional Referral Hospital suggest resistance rates above 50% for ampicillin and rising resistance to gentamicin.

Staphylococcus aureus adds another complication. Methicillin-resistant strains are increasingly common, and they are intrinsically resistant to ampicillin. In a newborn, a bloodstream infection with such an organism can progress to shock and death within 24 to 48 hours if the antibiotic does not cover it.

Blood culture capacity remains scarce outside referral hospitals. Most district-level facilities in Uganda cannot perform cultures, so clinicians rely on clinical signs alone. Even where cultures are available, results may take 48 to 72 hours, a delay that is often too long for a sick neonate.

The clinical picture is further muddied by how non-specific neonatal sepsis is at the bedside. A newborn may present with nothing more than poor feeding, a low temperature, or grunting. Respiratory distress in a neonate could be sepsis, could be transient tachypnoea of the newborn, could be a congenital pneumonia, and the two presentations can look identical in the first hours. Clinicians working without a culture are, in effect, treating a differential rather than an organism.

Even the process of getting a culture is fraught. A blood sample from a neonate is small, and the yield depends on technique, timing, and the volume of blood drawn. Contamination from skin flora can muddy the picture, leading to false positives that prompt unnecessary escalation or false negatives that reassure when they should not. In a unit with one incubator and a shared thermometer, the logistics of aseptic collection are not trivial.

The Case for Sticking With First-Line

Defenders of the current guideline point to its practicality. Ampicillin and gentamicin are cheap, heat-stable, and available in most Ugandan health facilities. Broad-spectrum alternatives such as meropenem or vancomycin are expensive, often out of stock, and require intravenous access and monitoring that many district hospitals cannot provide.

There is also the resistance argument. Changing empiric therapy without culture guidance risks driving further resistance. If every neonatal unit switches to a broader regimen, the selection pressure on Klebsiella and other organisms will increase, potentially narrowing future options even further. The guideline, in this view, is a brake on escalation.

Additionally, the evidence base for changing empiric therapy in low-resource settings is thin. Most randomized trials of neonatal sepsis treatment come from high-income countries or from settings with different pathogen profiles. Ugandan clinicians who stick with first-line therapy are not ignoring the data; they are weighing it against the risk of doing something worse.

There is a quieter, practical reason too. A regimen that is familiar to every midwife and intern on the ward gets given on time. A regimen that requires a dilution calculation, a cold chain, or a pharmacy order that may or may not be filled introduces new failure points into a system that can ill afford them.

The Case for Revising Empiric Therapy

Critics of the current approach argue that the mortality data demand action. Neonatal sepsis mortality in Uganda is estimated at 20–40%, and many of those deaths occur within the first 48 hours. Waiting for culture results that may never come is, in their view, a luxury that costs lives.

Observational series from Ugandan hospitals suggest that culture-directed treatment reduces mortality compared with blind empiric therapy. Where an antibiogram exists, it often shows that a second-line regimen such as ceftriaxone plus amikacin would cover a majority of isolates. The problem is that such regimens are not routinely available or affordable.

Proponents of revision also point to the ethical dimension. If a clinician knows that local resistance rates are high, continuing a drug that is likely to fail can feel like a violation of the duty of care. The counterargument, that changing without evidence is equally risky, does not resolve the immediate bedside dilemma.

Cost is where the two sides often talk past each other. A course of a broader agent may cost several times what ampicillin and gentamicin do, and in a facility where the pharmacy budget is fixed, that difference is not abstract. It may mean fewer doses of something else for another child on the same ward.

Yet the calculation is not only financial. A neonate who spends an extra week in hospital because of inadequate initial coverage consumes bed days, nursing time, and family resources. Those costs are harder to count but no less real. When a mother has travelled from a distant village and must stay near the hospital, the ripple effects of a failed first-line regimen extend well beyond the pharmacy ledger.

Why the Evidence Remains Contested

Few randomized trials of neonatal sepsis treatment have been conducted in East Africa. Most guidelines rely on extrapolations from other settings or on observational data. The absence of robust local evidence leaves room for reasonable disagreement about the best empiric regimen.

Culture results are often unavailable at the point of care, and even when available, they may not reflect the current resistance landscape. Antibiograms vary widely between hospitals, and a regimen that works in Mulago may fail in Mbale. This heterogeneity makes a single national guideline difficult to justify.

Industry influence and antibiotic access also shape prescribing. Pharmaceutical representatives promote newer, broader-spectrum agents, while procurement systems may favor cheaper first-line drugs. Clinicians are caught between commercial pressure, supply realities, and patient needs. As a related piece on this site noted about asthma inhalers, cost and availability often diverge from clinical need.

There is a methodological wrinkle that rarely makes it into the summaries. Blood cultures taken after antibiotics have already been started often come back negative, and a negative culture in a sick neonate is not reassurance. It may simply mean the sample was drawn too late. Units that count these as "no growth" can inadvertently report a cleaner resistance picture than the ward actually faces.

Another confounder is the distinction between community-acquired and hospital-acquired sepsis. A newborn infected at home may carry a different organism profile from one who has spent a week in a crowded nursery. Yet many studies pool the two, blurring the picture further. Clinicians on the ground know that a baby readmitted from the community is not the same as one who never left the ward, but the data seldom make that separation.

What a Clinician Can Do Now

Push for local antibiogram data in every neonatal unit. Even a small, periodic review of culture results can reveal resistance patterns that inform empiric choices. Without local data, clinicians are flying blind.

Use blood cultures before starting antibiotics whenever feasible. Even if results arrive late, they contribute to the unit's cumulative antibiogram and help the next patient. Document the reasons for empiric choices in the notes, and review them monthly with colleagues.

Advocate for affordable second-line antibiotics at the facility and district level. This includes engaging with national AMR action plans, which often include neonatal sepsis as a priority. The Uganda National Action Plan on Antimicrobial Resistance, for instance, calls for improved surveillance and access to essential antibiotics.

Finally, engage in peer review of sepsis deaths. A monthly mortality review can identify patterns, such as delayed presentation or inadequate initial coverage, that point to actionable changes. As a related piece on this site argued about diabetes care, the solution often lies in the system, not just the prescription.

This article is for informational purposes only and does not constitute personalised medical advice. Clinicians should consult local guidelines and their own clinical judgement.