Nearly half of malaria patients in Ugandan hospitals got antibiotics against guidelines

Nearly half of hospitalized patients in Uganda who carried a malaria diagnosis alongside at least one other condition received antibiotics that did not comply with the country’s national prescribing guidelines, according to a secondary analysis of point prevalence survey data published in Discover Medicine. Of 1,092 patient records reviewed from nine regional referral hospitals between October 2020 and November 2023, 464 patients, or 42.5 percent, were given antibiotic prescriptions that failed to meet at least one of the five rights of antimicrobial stewardship: the right patient, drug, dose, route, and duration. Because antibiotics are not routinely indicated for uncomplicated malaria, the findings point to a substantial gap in prescribing practice in a setting where antimicrobial resistance is an escalating threat.

The analysis drew on a national antimicrobial use and consumption surveillance program coordinated by Uganda’s Ministry of Health. Quarterly point prevalence surveys, built on the World Health Organization’s standardized methodology, captured data on patients receiving ongoing antibiotic treatment in five major wards—gynecology, maternity, medical, pediatric, and surgical—at Arua, Gulu, Jinja, Kabale, Lira, Masaka, Mbale, Mbarara, and Soroti regional referral hospitals. From a dataset of 9,008 de-identified patient records, 1,294 records, or 14.4 percent, included a malaria diagnosis. The researchers excluded 138 records with missing data and 64 records in which malaria was the sole diagnosis, because the WHO survey tool does not distinguish between uncomplicated and severe malaria and offers no bacterial indication category against which antibiotic use could be judged. That left 1,092 records, or 12.1 percent of the total, in which appropriateness could be evaluated against a documented co-existing diagnosis.

The population was young and heavily pediatric. The median age was seven years, with an interquartile range of three to sixteen, and 581 patients, or 53.2 percent, were female. Children accounted for the largest share of admissions: 738 patients, or 67.6 percent, were on pediatric wards. Most patients, 1,020 or 93.4 percent, had not undergone surgery before admission, and 122, or 11.2 percent, had been transferred from another hospital. The majority, 670 patients or 61.4 percent, had one additional recorded diagnosis beyond malaria, 315 had two, and 107 had more than two. Community-acquired infections were the leading recorded indication for antibiotic use, accounting for 580 cases, or 53.1 percent.

Prescribing intensity and drug selection revealed further patterns. Some 626 patients, or 57.3 percent, received one antibiotic, 434, or 39.7 percent, received two, and only 32, or 2.9 percent, received more than two. General practitioners wrote 923 prescriptions, or 84.5 percent of the total. Under the WHO’s Access, Watch, and Reserve classification, which ranks antibiotics by their potential to drive resistance, 866 patients, or 79.3 percent, received drugs in the Watch category, while 457, or 41.9 percent, received Access-category antibiotics. No patient received a Reserve-category drug. In addition, 87 prescriptions, or 8.0 percent, involved unclassified antibiotics, chiefly fixed-dose combinations that are not included in the WHO classification at all.

Non-adherence varied markedly across facilities and settings. It was highest at Mbarara and Mbale regional referral hospitals, where 58.5 percent of prescriptions in each failed guideline compliance, and lowest at Jinja, at 29.2 percent. By ward, the medical ward recorded the highest non-adherence at 48 percent. By age group, non-adherence was most common among patients aged five to fifteen, a difference that reached statistical significance with a p-value of 0.024. Facility, ward, age category, and antibiotic indication all varied significantly with adherence status, with p-values below 0.05.

To identify which factors independently predicted guideline non-adherence, the researchers used modified Poisson regression, a technique chosen over logistic regression to avoid overestimating associations when outcomes are common, with robust standard errors clustered by hospital. Three factors remained significant after adjustment. Patients prescribed antibiotics for medical prophylaxis had an adjusted prevalence ratio of 1.81, with a 95 percent confidence interval of 1.47 to 2.24. Those with an infectious syndrome diagnosis had an adjusted prevalence ratio of 2.15, with a confidence interval of 1.76 to 2.63, meaning they were roughly twice as likely to receive a non-adherent prescription. Prescriptions written for other, non-specific indications had an adjusted prevalence ratio of 1.97, with a confidence interval of 1.65 to 2.37. HIV status, malnutrition, and prescriber type showed no significant association with non-adherent prescribing.

The category of other indications deserves particular attention, because it captures prescriptions made when no clear therapeutic or prophylactic category applies. The authors describe it as including antibiotic use for non-specific febrile illness, treatment started to rule out sepsis without documented clinical justification, antibiotics given for malaria with no evidence of bacterial co-infection, and prescriptions initiated solely on the basis of laboratory abnormalities such as elevated white blood cell counts. This pattern points to diagnostic uncertainty as a central driver of inappropriate prescribing. In malaria-endemic regions, the overlapping symptoms of malaria and bacterial infections, including fever, altered consciousness, and systemic inflammation, make it clinically difficult to distinguish a primary malaria episode from a concurrent bacterial infection or both, especially where confirmatory diagnostics are not routinely available.

The discussion in the paper adds important nuance about the pediatric burden. Although most non-adherent prescriptions occurred in the pediatric ward, pediatric admission was not a statistically significant predictor of non-adherence in the multivariable model, with a prevalence ratio of 1.09, a confidence interval of 0.71 to 1.66, and a p-value of 0.778. The high absolute number of non-adherent prescriptions among children therefore reflects the disproportionate representation of pediatric patients in the cohort rather than an independently elevated per-patient risk of inappropriate prescribing. Similarly, while antibiotic prescribing was common among HIV-positive and malnourished patients, the rates were not statistically different from those in patients without these conditions, even though clinicians in such contexts often initiate empiric antibiotics to guard against occult bacterial infections.

The authors also flagged the unclassified fixed-dose combinations as a stewardship concern. These combinations often lack clear clinical indications and safety profiles, and evidence indicates they contribute to irrational prescribing and may accelerate the emergence of resistance. Their circulation and use in Uganda remain largely unregulated, and the researchers argue that strengthening regulation, through enforcing registration and restricting non-recommended combinations, should be a component of the national antimicrobial stewardship and resistance containment strategy.

The study has limitations inherent to its cross-sectional, secondary-analysis design. Variability in diagnosis documentation across sites may have underestimated non-adherence, and some variables had incomplete data, which may have slightly reduced the precision of estimates. Because data were collected only from inpatients admitted for at least 24 hours, outpatient antibiotic use was not captured. The assessment of treatment duration relied on elapsed time from the antibiotic start date to the survey date as a proxy for intended duration, since the WHO tool does not record planned duration, which may have misclassified some prescriptions. Facility-specific antibiogram data were not consistently available, so some prescriptions classified as non-adherent may have been clinically justified by local resistance patterns. The exclusion of patients with malaria as the sole diagnosis introduces bias whose direction is uncertain, and the authors caution that findings should be interpreted specifically for the co-morbid malaria subgroup rather than generalized to all hospitalized malaria patients. Even so, they conclude that targeted interventions, including prescriber training, improved point-of-care diagnostics, routine audits with feedback, and integrating stewardship support into malaria programming, may help optimize antibiotic use and support broader antimicrobial resistance containment in similar settings.

Subject of Research: Guideline non-adherence in antibiotic prescribing for hospitalized malaria patients with co-existing diagnoses in Uganda

Article Title: Antibiotic use among patients with malaria and co-existing diagnoses in tertiary hospitals in Uganda

Article References: Katumba, H., Nakasendwa, S., Twemanye, V., Tumwine, C., Migisha, R., Kwesiga, B., Bulage, L., Mayito, J., Bulwadda, D., Hope, M., Waswa, J. P., Kiggundu, R., Walwema, R., Byonanebye, D. M., Kakooza, F., Ario, A. R., & Kambugu, A. (2026). Antibiotic use among patients with malaria and co-existing diagnoses in tertiary hospitals in Uganda. Discover Medicine, 3(1), Article 163. https://doi.org/10.1007/s44337-026-00734-1

Image Credits: AI Generated

DOI: 10.1007/s44337-026-00734-1

Keywords: antimicrobial resistance, antimicrobial stewardship, malaria, antibiotic prescribing, Uganda, point prevalence survey, WHO AWaRe classification, Uganda Clinical Guidelines, pediatric wards, regional referral hospitals, fixed-dose combinations, diagnostic uncertainty