BOBilly Ogwel

Billy Ogwel

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Billy Ogwel has a total of 158 co-authors. Here are the top co-authors by number of shared publications:

  • Richard Omore

    Kenya Medical Research Institute-Center for Global Health Research (KEMRI-CGHR), Kisumu, Kenya

    20 shared publications
  • Alex O. Awuor

    Kenya Medical Research Institute-Center for Global Health Research (KEMRI-CGHR), Kisumu, Kenya

    12 shared publications
  • Caleb Okonji

    Kenya Medical Research Institute, Center for Global Health Research (KEMRI-CGHR), Kisumu, Kenya

    10 shared publications
  • Stephen Munga

    Kenya Medical Research Institute, Center for Global Health Research (KEMRI-CGHR), Kisumu, Kenya

    9 shared publications
  • Patricia B. Pavlinac

    Department of Global Health, University of Washington, Seattle, USA

    8 shared publications
  • Raphael O. Anyango

    Affiliation Kenya Medical Research Institute-Center for Global Health Research (KEMRI-CGHR), Kisumu, Kenya

    8 shared publications

Latest Publications

  • Open Access

    Towards Early Warning Systems for Shigellosis in Africa: Multi-Horizon Forecasting of Incidence in Rural Western Kenya Using Statistical State-Space and Machine Learning Models

    by Billy Ogwel, Brian O. Onyando, Bryan O. Nyawanda, Alex O. Awuor, Joseph Ogutu, John B. Ochieng, Maria Garcia Quesada, Patricia B. Pavlinac, Max Lau, Richard Omore, et al.

    Introduction: Despite the significant burden of Shigellosis, routine surveillance systems in resource-limited settings lack predictive capabilities to guide proactive public health responses. Consequently, we developed and evaluated the accuracy, uncertainty and computational cost of multi-horizon forecasting models for Shigella incidence among children Shigella incidence using data from three pediatric enteric studies (2008– 2024), supplemented with Kenya Health Information System during inter-study gaps. Incidence estimates incorporated sampling, healthcare-seeking, stool-testing, and pathogen-specific weights, with bootstrap-derived 95% confidence intervals. Predictors included Google Trends health-related indicators, climate, temporal trends, and seasonality. Forecasts at 1-, 3-, and 6-month horizons were evaluated using rolling-origin validation. Following horizon-specific predictor selection using Boruta, 14 state-space, machine-learning, and hybrid models were compared using point-forecast and probabilistic metrics, bootstrap confidence intervals, and Model Confidence Sets. Adaptive conformal prediction recalibrated prediction intervals, while Shapley Additive exPlanations assessed predictor contributions. Results: Mean monthly Shigella incidence was 2.51 per 100 child-years (standard deviation=1.85). Shigella dynamics in Kenya can be forecast beyond seasonal persistence alone for up to six months reducing absolute errors by ∼ 33%. However, optimal model architectures and predictor sets shift systematically across forecasting horizons. Temporal trend and recent incidence were the dominant predictors, while predictor importance shifted with forecast horizon toward syndromic, pharmaceutical, seasonal, and lagged environmental signals. Extreme Gradient Boosting performed best at 1 month (Root mean squared error [RMSE]=1.86), remained competitive at 3 months (RMSE=1.89), whereas exponential smoothing (ETS) performed best at 6 months (RMSE=2.05). Forecast accuracy declined with increasing horizon. ETS and TBATS provided the best-calibrated prediction intervals, whereas machine-learning models generally under-covered. Model Confidence Sets demonstrated horizon-dependent model distinguishability. Conclusion: Horizon-specific model selection balancing predictive accuracy, uncertainty calibration, and computational feasibility could strengthen operational early warning systems for shigellosis. Author Summary: Shigella infections remain an important cause of illness among young children in low-resource settings, yet public health systems often lack tools to anticipate changes in disease burden before they occur. Early warning systems that can reliably predict future Shigella incidence could help health authorities plan surveillance, allocate resources, and prepare for periods of increased transmission. We used 16 years of data from children Shigella incidence 1-, 3-, and 6-months ahead. We combined disease surveillance data with information on seasonal patterns, weather, and health-related internet searches. We found that Shigella incidence could be predicted beyond what would be expected from seasonal patterns alone, with useful forecasts possible up to six months ahead. However, the best forecasting approach depended on how far into the future predictions were required. Machine-learning methods provided the most accurate short-term predictions, while simpler statistical models performed better for longer-term forecasts and provided more reliable estimates of uncertainty. These findings highlight the importance of selecting forecasting methods according to the intended prediction horizon and operational requirements. Such approaches could strengthen early warning and preparedness for childhood Shigella infections in resource-limited settings.

  • Open Access

    Derivation and Validation of Clinical Prediction Models for Cryptosporidium -attributed Acute Diarrhea in African Children

    by Léna Mazza, Ben J Brintz, M Jahangir Hossain, Billy Ogwel, John B Ochieng, Jane Juma, Helen Powell, Adama Mamby Keita, Richard Omore, Samba O Sow, et al.

    Background: Clinical decision-support tools have the potential to guide empiric treatment and prioritize diagnostic testing at the individual-level, and screening strategies for vaccine and therapeutic trials at the population-level. Such tools for the parasitic pathogen Cryptosporidium, a leading cause of morbidity and mortality among children in low- and middle-income countries (LMICs), are lacking. We aimed to develop clinical prediction models to identify Cryptosporidium-attributed acute diarrhea in young children. Methods: For model derivation, we used clinical and demographic data from the Global Enteric Multicenter Study (GEMS) and integrated site-specific weather data from the NOAA database. Random forest (RF) and logistic regression (LR) models were trained and evaluated using 5-fold cross-validation. Results: Using only the top 3 predictors, prediction models achieved a mean area under the curve of 0.77 for RF and 0.75 for LR models. RF classification identified patient age, mean rainfall and mean temperature in the 30 days preceding enrollment as the most important predictors of Cryptosporidium-attributed diarrhea. For external validation, models were trained on the full GEMS dataset and tested on the Vaccine Impact on Diarrhea in Africa (VIDA) dataset, with the LR models having better calibration and net benefit. Conclusions: We derived and externally validated clinical prediction models for identifying Cryptosporidium infections among LMIC children with acute diarrhea. Our findings highlight the value of incorporating weather variables into clinical prediction models for diarrheal diseases, and their potential use for real-time decision support in the context of climate change and extreme weather events.


  • Open Access

    Frequency and correlates of non-receipt of age-appropriate vaccination among children aged 6-35 months with medically attended diarrhea: Findings from the Enterics for Global Health (EFGH) Shigella study, 2022-2024

    by Caren Oreso, Billy Ogwel, Alex O. Awuor, Raphael O. Anyango, Karen Kotloff, M. Jahangir Hossain, Henry Badji, Khuzwayo C. Jere, Latif Ndeketa, Josh Colston, et al.

    Complete childhood immunization protects children from long-term health complications and disabilities caused by vaccine-preventable diseases. Enterics for Global Health (EFGH)- Shigella surveillance was a two-year study measuring incidence rates and consequences of Shigella among children aged 6–35 months in seven sites located in Asia, Latin America and Africa. Here, we estimated the prevalence and factors associated with non-receipt of age-appropriate vaccination among children enrolled in the EFGH- Shigella study. In this nested cross-sectional study, we analysed data from 7,932 children aged 6–35 months presenting with medically attended diarrhea (MAD). Vaccines recommended per each country’s national immunization schedule were extracted from medical records and risk factors were collected by caregiver interview and physical exam. We defined age-appropriate vaccines as receipt of the early childhood vaccinations within one month of the recommended age, according to the national immunization schedule, on the immunization card. Poison regression was used to identify independent factors associated with non-receipt of age-appropriate vaccination accounting for all covariates. Over half of enrolled children (51.7%) did not receive all age-appropriate vaccines most commonly in The Gambia (75.2%) and least frequently in Bangladesh (22.3%). Children 12–35 months of age were more likely not have all age appropriate vaccines compared to children 6–11 months (aPR: 1.47, 95%CI 1.39 to 1.54), children who came from households with ≥3 children aged <5 years (aPR:1.07;1.01-1.13), had mothers with low education (aPR:1.18; 1.12-1.24), and were wasted (Moderately: aPR: 1.06; 1.00-1.13; Severely: aPR: 1.13, 1.03-1.24) were more likely to miss all age-appropriate vaccines compared to their counterparts who did not. Non-receipt of age-appropriate vaccination was largely age dependent, driven by mother’s education and severe wasting highlighting the need to design effective strategies that incorporate site complexities to improve timely vaccination targeting vulnerable groups.

  • Open Access

    The Projected Impact of Climate Change on the Burden of Cryptosporidium in Mali, Kenya, and The Gambia

    by Megan Kowalcyk, Billy Ogwel, Kristopher B Karnauskas, Henry Badji, Jane Juma, M Jahangir Hossain, Richard Omore, Karen Kotloff, Andrea G Buchwald

    Background: Cryptosporidium is an important cause of moderate to severe diarrhea among children aged <5 years, with the highest burden in sub-Saharan Africa. Cryptosporidium is highly climate sensitive, yet there has been no published literature projecting the incidence of Cryptosporidium under climate change in sub-Saharan Africa, where vulnerability to climate change is highest. Methods: Utilizing monthly case counts of Cryptosporidium from 2 case-control studies in The Gambia, Kenya, and Mali, we modeled the relationship among monthly temperature, precipitation, and vulnerability factors with Cryptosporidium incidence by study site using bayesian network models. Global climate models were then used to project temperature and precipitation at each site in 2040 and 2055 under 2 emission scenarios. Future Cryptosporidium incidence was estimated by projected climate and future vulnerability scenarios. Results: Global climate models predict increasing temperatures at all sites and increasing rainfall in Kenya under all scenarios. Incidence of Cryptosporidium per 10 000 children is predicted to change by −16.93% (range, −24.84% to 8.92%), −4.34% (range, −24.50% to 9.70%), and 24.81% (range, 10.41%–44.06%) in Kenya, Mali, and The Gambia, respectively, from baseline (2007–2015) to 2055 under high carbon emissions. These estimates are influenced by future vulnerability levels. Conclusions: The effect of climate change on Cryptosporidium incidence will be location and climate specific; however, whether or not vulnerability factors remain stable will have a strong influence on what future Cryptosporidium looks like. Prioritizing development in areas with the highest burden of climate-sensitive health outcomes can mitigate the impact of climate change.


  • Open Access

    Missed Opportunities of Syndrome-Based Diarrhea Management Guidelines to Detect Non-Dysenteric Shigella Infections in Children: Findings from the Enteric for Global Health (EFGH)-Shigella Surveillance study in Kenya: 2022-2024

    by Alex Ondeng Awuor, Billy Ogwel, Caleb Okonji, Raphael O Anyango, Caren Oreso, Lilian Ambila, Brian Otieno, Stephen Munga, John Benjamin Ochieng, Victor Akelo, et al.

    Shigella is a major cause of childhood diarrhea in sub-Saharan Africa. Current World Health Organization (WHO) guidelines recommend empiric antibiotics only for dysentery, yet non-dysenteric presentations account for 40–89% of Shigella infections. We quantified the performance of existing guidelines to identify Shigella infection and evaluated enhanced syndromic criteria for improved management of Shigella cases. We leveraged data from the Enteric for Global Health (EFGH) Kenyan site, which enrolled children aged 6-35 months with medically attended diarrhea, collected rectal swabs at enrolment, and tested these for Shigella using both the culture and TaqMan-array cards (TAC). Shigella positivity was defined as either a culture-confirmed Shigella or a TAC-attributable result (CT or Fisher’s exact as appropriate while continuous variables were compared between groups using the Wilcoxon rank-sum test. A logistic regression model was fitted to identify a parsimonious set of predictors. Out of the enrolled 1400 MAD cases, of whom 175 (12.5%) were Shigella positive, of which 134 (76.5%) being non-dysenteric. Among all enrolled children, 148 (10.6%) were dysenteric and 1,252 (89.4%) were non-dysenteric. Of the non-dysenteric cases, 57/1,252 (4.6%) and 129/1,251 (10.3%) of children were Shigella attributable by culture and TAC, respectively. Compared to Shigella negative cases, positive cases were older (Median age in months [Q1-Q3]: 20 [14–24] vs 13 [8–19], pShigella positive cases were less likely to present with vomiting (66 [37.7%] vs 587 [47.9%], p=0.014) and difficulty in breathing (0 (0%) vs 27 [2.2%], pShigella (area under the ROC curve (AUC) [95% CI]: 0.58 [0.54-0.61]), while Dysentery and Age binary (Shigella -positive cases which are non-dysenteric. This suggests that dysentery alone has limited predictive power, but incorporating additional symptoms increases discriminatory ability by up to 14%, with age alone improving it by 12%. These findings support the need to re-evaluate the current syndromic based guidelines to better detect Shigellosis and strengthen antibiotic stewardship.

  • Open Access

    Performance of fecal inflammatory biomarkers to identify watery shigellosis: Findings from the Enterics for Global Health (EFGH) Shigella surveillance study

    by Billy Ogwel, Farhana Khanam, Henry Badji, Mary Charles, Sonia Qureshi, Bri’ANna Horne, Stephanie A. Brennhofer, James A. Platts-Mills, Khandra Sears, Sharon Tennant, et al.

    Background: Current syndromic guidelines for diarrhea treatment miss watery Shigella cases, leading to undertreatment of children who may benefit. Incorporating fecal inflammatory biomarkers into diagnosis may improve case identification. Methods: We conducted an ancillary analysis using samples from six sites (The Gambia, Kenya, Malawi, Bangladesh, Pakistan, and Peru) from the Enterics for Global Health (EFGH)- Shigella surveillance study, a facility-based hybrid study of children aged 6–35 months with diarrhea. Four fecal biomarkers were quantified by enzyme-linked immunosorbent assays at enrollment: myeloperoxidase, calprotectin, neutrophil gelatinase-associated lipocalin (lipocalin-2), and hemoglobin. An ensemble model with leave-one-site-out cross-validation was used to predict watery shigellosis, incorporating biomarkers and nine clinical and socio-economic predictors. We compared the predictive performance of the algorithm using: a) all predictors (including biomarkers); b) all non-biomarker predictors; c) all predictors (with selected biomarkers). Results: Between June 2022 and August 2024, a total of 4,191/9,476 (44.2%) children presented with watery diarrhea (non-bloody) and had their whole stool tested for the biomarkers and 4,083 stool samples or rectal swabs were tested by qPCR; 735 (18.0%) had Shigella- attributable diarrhea by qPCR. The full model incorporating all 13 predictors achieved an area under the curve (AUC) of 0.75 [95% CI: 0.67–0.78], with a sensitivity of 0.67 and specificity of 0.75. Excluding biomarkers reduced model performance by 8% (AUC 0.67, 95% CI: 0.61-0.70). Adding hemoglobin alone improved the model’s discriminatory ability by 7%, while further adding myeloperoxidase had marginal contribution (1%), and lipocalin-2 (0%) and calprotectin none (0%). Conclusion: Fecal hemoglobin substantially improved prediction scores for watery shigellosis. Consequently, implementation of point-of-care assays for hemoglobin could improve clinical diagnosis in these settings and inform appropriate antibiotic treatment.


  • Open Access

    Post-Diarrheal Nutritional Trajectories Among Malnourished Children: A Clustering and Multinomial Modelling Approach

    by Billy Ogwel, Alex O. Awuor, Brian O. Onyando, Raphael Ochieng, M. Jahangir Hossain, Bakary Conteh, Wardah Mujahid, Fariha Shaheen, Vitumbiko Munthali, Thandizo Malemia, et al.

    Although the co-occurrence of diarrhea and malnutrition is well documented, research has largely focused on the acute management of diarrheal illness. Despite its importance, longitudinal evidence characterizing post-diarrheal recovery trajectories is sparse. We sought to characterize post-diarrheal nutritional recovery trajectories among children aged 6–35 months who were malnourished at enrollment using data from the Enterics for Global Health (EFGH) Shigella Surveillance study (2022–2024). EFGH enrolled children aged 6–35 months presenting with medically-attended diarrhea and followed them at 4 weeks and 3 months post-enrollment. This analysis included children with baseline wasting, stunting, or underweight (z-score < −2) and complete anthropometric follow-up. Latent class mixed-effects models were used to identify distinct post-diarrheal growth trajectories based on changes in anthropometric z-scores over time. Multinomial modified Poisson regression models examined associations between baseline factors and trajectory membership. Among 9,480 enrolled children, 16.5% (n=1,561) were wasted, 22.7% (n=2,155) stunted, and 21.0% (n=1,994) underweight at baseline. Wasting showed greater recovery potential (80.8%) compared with stunting (38.5%) and underweight (40.3%). Recovery was shaped by factors across multiple levels. Clinical severity markers ( prolonged diarrhea, dehydration, and hypoxemia) increased the risk of nutritional failure. Age also influenced outcomes: infants were more likely to worsen, whereas older toddlers more often experienced stagnation. Interventions including exclusive breastfeeding, oral rehydration therapy, appropriate antibiotics, and zinc supplementation, improved outcomes, while unimproved sanitation undermined recovery. These findings highlight the need for integrated strategies combining infection control, nutritional rehabilitation, and water, sanitation, and hygiene interventions tailored to the children’s developmental stage. Key Messages: Post-diarrheal nutritional recovery is highly heterogeneous, with wasting showing the greatest potential for improvement, while stunting and underweight often result in persistent growth stagnation. Baseline anthropometric deficits alone are insufficient to predict recovery, highlighting the need for dynamic monitoring and individualized management. Infants are particularly vulnerable to acute nutritional deterioration, while older toddlers frequently experience growth stagnation. Modifiable protective factors including exclusive breastfeeding, ORS, zinc, and appropriate antibiotics, improved outcomes, whereas poor sanitation undermined recovery. Integrated strategies, tailored to a child’s developmental stage, combining clinical care, nutrition, and environmental interventions are critical to support sustained child growth and development.

  • Open Access

    Performance of Fecal Inflammatory Biomarkers to Identify Watery Shigellosis: Findings from the Enterics for Global Health (EFGH) Shigella Surveillance Study

    by Billy Ogwel, Farhana Khanam, Henry Badji, Mary Charles, Sonia Qureshi, Bri’anna Horne, Stephanie A Brennhofer, James A Platts-Mills, Khandra Sears, Sharon Tennant, et al.

    Background: Current syndromic guidelines for diarrhea treatment miss watery Shigella cases, leading to undertreatment of children who may benefit. Incorporating fecal inflammatory biomarkers into diagnosis may improve case identification. Methods: We conducted an ancillary analysis using samples from six sites (The Gambia, Kenya, Malawi, Bangladesh, Pakistan, and Peru) from the Enterics for Global Health (EFGH)- Shigella surveillance study, a facility-based hybrid study of children aged 6-35 months with diarrhea. Four fecal biomarkers were quantified by enzyme-linked immunosorbent assays at enrollment: myeloperoxidase, calprotectin, neutrophil gelatinase-associated lipocalin (lipocalin-2), and hemoglobin. An ensemble model with leave-one-site-out cross-validation was used to predict watery shigellosis, incorporating biomarkers and nine clinical and socio-economic predictors. We compared the predictive performance of the algorithm using: a) all predictors (including biomarkers); b) all non-biomarker predictors; c) all predictors (with selected biomarkers). Results: Between June 2022 and August 2024, a total of 4,191/9,476 (44.2%) children presented with watery diarrhea and had their whole stool tested for the biomarkers and 4,083 stool samples or rectal swabs were tested by qPCR; 735 (18.0%) had Shigella- attributable diarrhea by qPCR. The full model incorporating all 13 predictors achieved an area under the curve (AUC) of 0.75 [95% CI: 0.67–0.78], with a sensitivity of 0.67 and specificity of 0.75. Excluding biomarkers reduced model performance by 8% (AUC 0.67, 95% CI: 0.61-0.70). Adding hemoglobin alone improved the model’s discriminatory ability by 7%, while further adding myeloperoxidase had marginal contribution (1%), and lipocalin-2 (0%) and calprotectin none (0%). Conclusion: Fecal hemoglobin substantially improved prediction scores for watery shigellosis. Consequently, implementation of point-of-care assays for hemoglobin could improve clinical diagnosis in these settings and inform appropriate antibiotic treatment. Author Summary: Diarrhea remains a leading cause of illness in young children, with the bacteria Shigella being a major contributor. Current treatment guidelines recommend antibiotics only when diarrhea is visibly bloody. However, many children with Shigella infection have non-bloody, watery diarrhea and are therefore not treated, despite needing care. In this study, we explored whether simple markers of gut inflammation found in stool could improve the identification of Shigella infection. We analyzed stool samples from over 4,000 children across six countries in Africa, Asia, and South America and combined these markers with basic clinical information. We found that detecting small, invisible amounts of blood in stool (fecal hemoglobin) improved identification of non-bloody Shigella diarrhea by 7%. Other inflammation markers added little benefit. By combining fecal hemoglobin with child’s age and stool frequency, we developed a simple score that matched performance of the full model enhancing its practical use. This approach outperformed current guidelines, which only treat bloody diarrhea, and was more accurate than treating most children without testing. Importantly, fecal hemoglobin tests are inexpensive and commercially available. Using them at the point of care could help health workers better target antibiotics, improve child outcomes, and promote responsible antibiotic use in low-resource settings.

  • Open Access

    Temporal patterns and risk factors of diarrheal comorbidity among children aged < 5 years in rural western Kenya: Evidence from three consecutive enteric studies, 2008–2024

    by Billy Ogwel, Bryan O. Nyawanda, Brian O. Onyando, Alex O. Awuor, Caleb Okonji, Raphael O. Anyango, Caren Oreso, Catherine Sonye, John B. Ochieng, Stephen Munga, et al.

    Sub-Saharan Africa bears the highest burden of diarrhea, often complicated by comorbidities that delay diagnosis, hinder treatment, and worsen outcomes. As the epidemiology of diarrheal disease evolves, understanding comorbidity patterns is critical for effective public health responses. We examined the temporal patterns and risk factors of diarrheal comorbidity in Kenyan children aged < 5. We conducted secondary pooled analysis with a retrospective cohort design leveraging data from the Global Enteric Multicenter Study (GEMS, 2008–2012), the Vaccine Impact on Diarrhea in Africa (VIDA, 2015–2018), the Enteric for Global Health (EFGH) Shigella surveillance study (2022–2024). The outcome was comorbidity count, defined by Integrated Management of Childhood Illnesses case definitions and clinician diagnoses of ten conditions: malaria, bacterial infection, pneumonia, severe acute malnutrition (SAM), meningitis, acute febrile illness (AFI), respiratory Illness (non-pneumonia), anemia, stunting and wasting. Temporal trends were assessed using descriptive statistics and the Cochran-Armitage trend test. Risk factors were identified using generalized estimating equations with a Poisson distribution, adjusting for clustering. We analyzed data from 4,148 children with moderate-to-severe diarrhea; 90.3% had ≥ one comorbidity, with a declining trend across studies: GEMS (92.9%), VIDA (89.3%), and EFGH (86.6%). Pneumonia (49.5%), malaria (48.3%), and stunting (24.7%) were most common comorbidities. The proportion of children with only one comorbidity increased (28.9% [2008] to 49.7% [2024]), while multiple comorbidities declined. Traditional comorbidities (malaria, pneumonia, wasting, SAM) significantly decreased, while AFI, anemia, and non-pneumonia respiratory illness increased. Multivariable analysis identified older age, lower caregiver education, dehydration, vomiting, 3-month lagged rainfall and temperature, and high respiratory rate as drivers of higher comorbidity counts, while female sex was associated with fewer comorbidities. Despite the high prevalence, we observed a 20–23% decline in comorbidity burden and a fundamental shift in disease profiles. Our findings support the need for a shift from single-disease control to integrated disease management.


  • Open Access

    The Accuracy of Caregiver’s ‘hot to touch’ assessment in paediatric healthcare among children aged 6-35 months with medically-attended diarrhea: Findings from the EFGH- Shigella surveillance in Kenya, Malawi, Bangladesh and Peru, 2022-2024

    by Raphael O Anyango, Billy Ogwel, Olivia Schultes, Caren Oreso, Brian O Onyando, Donnie Mategula, Desiree Witte, Wagner Valentino Shapiama Lopez, Pablo Penataro Yori, Taufiqur Rahman Bhuiyan, et al.

    Introduction: Confidence in caregivers’ assessment of fever in their children, compared to thermometer readings, could help guide prompt care seeking and appropriate treatment in settings where access to reliable diagnostic tools is limited. Here, we evaluated the accuracy and drivers of caregiver-reported ‘hot-to-touch’ fever compared to digital thermometry among children in the Enterics for Global Health (EFGH) Shigella surveillance study. Methods: Children aged 6–35 months with medically attended diarrhea (MAD) enrolled in Kenya, Malawi, Bangladesh, and Peru between August 2022 and August 2024 were included. We trained caregivers to assess and record daily ‘hot-to-touch’ (subjective fever measurement) and digital (thermometer) axillary temperature (fever defined as ≥37.5°C) readings over for 14 days post-enrolment. We calculated site specific and overall accuracy of ‘hot-to-touch’ compared to thermometer-measured fever and used multivariable Poisson regression to identify factors associated with accurate detection. Results: The accuracy of caregiver-reported ‘hot-to-touch’ assessment ranged from 62.1% to72.0% overall and was highest in Bangladesh (83.2%–96.1%) and lowest in Malawi (47.4%–53.4%) over the 14 day assessment period. Accuracy was higher in children with chest indrawing (aPR=1.29, 95% CI: 1.04–1.60) and low respiratory rate (aPR=1.20, 95% CI: 1.11–1.29) and in children from wealthier households (Quintile 5: aPR=1.21, 95% CI: 1.01-1.44). Accuracy was lower among caregivers from households with ≥3 children (aPR=0.88, 95% CI: 0.79–0.99) and for children with low heart rate (aPR=0.76, 95% CI: 0.61–0.94). Conclusion: Suboptimal accuracy of hot-to-touch compared to digital thermometers in detecting fever in this study supports the need for digital thermometer use and context-specific strategies to enhance early detection of fever, particularly in communities living in resource-poor settings.