Every clinician working in a high-burden African pediatric ward knows the protocol by reflex: broad-spectrum antibiotics, empirical trimethoprim-sulfamethoxazole for Pneumocystis jirovecii pneumonia (PCP), supportive oxygen, and in TB-endemic settings, empirical antitubercular therapy while culture results take days or weeks to return. The protocol is fast, guideline-endorsed, and saves lives. It also, according to new evidence from the EMPIRICAL trial published in The Lancet, may have been systematically overlooking a pathogen that has been sitting in the chest X-ray all along.
Cytomegalovirus. CMV. A virus so common in immunocompromised infants that its presence in bronchoalveolar lavage fluid is sometimes dismissed as colonization rather than causation. The EMPIRICAL trial, published in The Lancet,, asks whether that dismissal has been costing infant lives, and the answer it delivers is uncomfortable enough to force a redesign of how the clinical trials community thinks about empirical treatment strategies in pediatric HIV.
Mortality from severe pneumonia in infants living with HIV reaches 35%, a figure that has refused to move meaningfully despite decades of WHO-guided empirical regimens. That number comes from the EMPIRICAL trial’s own framing of the problem. If broad-spectrum antibiotics and anti-PCP therapy were sufficient, that number should be falling. The fact that it is not points directly at something the standard regimen does not cover.
The Pathogen Hiding in the Protocol
CMV-associated pneumonia appears in 36% of HIV-infected children hospitalized for severe pneumonia, compared with 15% in uninfected children, an odds ratio of 3.0 (95% confidence interval, 1.3 to 7.4), according to a prevalence study published on PubMed. That signal has been in the literature for years. The clinical trial community’s response, with rare exceptions, has been to treat CMV as a co-passenger rather than a co-killer.
The EMPIRICAL trial challenges that assumption operationally, not just epidemiologically. By randomizing infants living with HIV and hospitalized with severe pneumonia in Africa to receive empirical valganciclovir and/or empirical tuberculosis treatment,, the trial asks whether treating CMV presumptively, before diagnostic confirmation, shifts the survival curve. The design itself is the argument: if you wait for CMV confirmation via bronchoalveolar lavage PCR in resource-limited settings where advanced diagnostics may be unavailable, the delay itself may cost lives.
Diagnosing CMV pneumonitis already requires a compatible clinical syndrome plus a positive CMV result on BAL fluid by PCR, shell vial, or conventional culture, as defined in published diagnostic criteria from Open Forum Infectious Diseases. In the settings where these infants are dying, that standard is operationally unreachable. The EMPIRICAL trial’s designers understood this, which is why the empirical arm was constructed the way it was. Rigorous trial design, in this case, meant confronting what the diagnostic pathway actually looks like in Kampala or Maputo, not what it looks like in a tertiary center in London.
What Trial Design Got Right That Standard Care Got Wrong
Here is the counterintuitive read on the EMPIRICAL trial, and it matters for anyone designing studies in pediatric infectious disease: the most scientifically honest move was to stop waiting for diagnostic certainty before treating. The conventional assumption in trial design is that empirical treatment is a methodological compromise, a concession to the absence of better tools. The EMPIRICAL investigators inverted that logic. Empirical treatment, structured within a randomized controlled trial, becomes the mechanism for generating the evidence that diagnostic-first approaches could never produce, because the diagnostics are not available at the point of care where the deaths are happening.
Compare this to the trajectory of empirical TB treatment in pediatric HIV. WHO guidelines explicitly endorse initiating empiric antitubercular therapy in infants and children living with HIV when TB is strongly suspected, with Directly Observed Therapy recommended to protect adherence and reduce relapse. That endorsement came after years of trial evidence in high-burden settings. CMV is now on a parallel trajectory, though the evidence base for empirical CMV treatment in this population is considerably less mature than the evidence base that underpinned WHO’s formalization of empirical TB guidance.
Valganciclovir carries its own regulatory complexity in this population. The FDA approved valganciclovir (Valcyte) for CMV prophylaxis in pediatric kidney and heart transplant patients, not for CMV pneumonia treatment in HIV-positive African infants. The EMPIRICAL trial is therefore operating in a regulatory space where the drug is being repurposed in a population it was never formally studied in for this indication. That is not a criticism of the investigators. That is precisely what investigator-initiated trials funded by mechanisms like EDCTP are designed to do. But it means the path from EMPIRICAL’s findings to a WHO guideline update will require regulators and guideline bodies to engage with off-label evidence in a population where conducting additional confirmatory trials raises profound ethical and logistical questions.
Who Rebuilds the Protocol First
For sponsors and CROs designing pediatric infectious disease trials in sub-Saharan Africa, the EMPIRICAL trial signals something structural about what the evidence base now demands. PCP treatment protocols for children with HIV already specify 21 days of IV trimethoprim-sulfamethoxazole at 15 to 20 mg/kg per day for the trimethoprim component, per NIH pediatric opportunistic infection guidelines. That regimen is codified. CMV empirical treatment in this population has no equivalent codification yet. The EMPIRICAL trial is the data point that makes codification possible, but only if guideline bodies treat it with the same weight they give RCT evidence in adult populations.
The operational implication for trial designers is sharper. Studies recruiting infants with HIV-associated severe pneumonia can no longer be designed as if the pathogen mix is well-characterized. The EMPIRICAL findings demand that future trials in this space either stratify by CMV status at enrollment or include CMV empirical treatment as a background standard of care in both arms, the same way trials in TB-endemic settings now require disclosure of empirical antitubercular therapy use. Failure to account for CMV in protocol design after EMPIRICAL is not a gap in knowledge. It is a methodological blind spot that trial sponsors and ethics committees will increasingly need to justify.
Technology vendors selling point-of-care diagnostics in low-resource settings face a different kind of pressure. The EMPIRICAL trial’s empirical design succeeded partly because confirmatory diagnostics for CMV are not available at the bedside in the settings studied. If that changes, the trial design logic changes with it. Portable, low-cost CMV PCR at the point of care would shift the question from “should we treat empirically” to “can we confirm fast enough to treat targeted.” The EDCTP-supported EMPIRICAL study is, among other things, a market signal for diagnostic developers who want to work in this space.
The EMPIRICAL trial has named the mechanism by which it might finally move. What happens next depends on whether WHO updates its pediatric HIV pneumonia guidance to reflect CMV empirical treatment, whether regulators engage with valganciclovir’s off-label evidence in this population, and whether the next generation of pediatric infectious disease trials is designed with CMV as a primary variable rather than an afterthought.
References
- The Lancet, “The overlooked role of CMV in pneumonia in infants with HIV” (2026)
- EDCTP, “EMPIRICAL study offers new evidence for treating severe pneumonia in infants living with HIV”
- PubMed, CMV-associated pneumonia prevalence in HIV-infected children: OR 3.0 (95% CI, 1.3–7.4)
- NIH Clinical Info, “Mycobacterium tuberculosis: Pediatric Opportunistic Infections Guidelines”
- NIH Clinical Info, “Pneumocystis jirovecii Pneumonia: Pediatric Opportunistic Infections Guidelines”
- FierceBiotech, “FDA approves Valcyte (valganciclovir) for pediatric CMV prevention”
- Open Forum Infectious Diseases, Diagnostic criteria for CMV pneumonitis (BAL PCR, shell vial, conventional culture)
Moe Alsumidaie, MBA, MSF, is founder and Chief Editor of Vanguard Publications, which publishes Clinical Trial Vanguard, Pharma Vanguard and BullScope, and Head of Research at CliniBiz. He has two decades in clinical trial operations and data science, with earlier roles at Genentech, Abbott Vascular and Stanford University Medical Center, and is a guest lecturer in clinical trial sciences at Rutgers University.

