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

This study demonstrates that incorporating fecal hemoglobin into a diagnostic model significantly improves the identification of watery *Shigella* diarrhea in children across six low- and middle-income countries, offering a practical, low-cost strategy to guide antibiotic treatment beyond current syndromic guidelines.

The Gist

Imagine a young child comes to a clinic with a bad tummy ache and watery diarrhea. In many parts of the world, doctors follow a strict rulebook: "If you see blood in the poop, give antibiotics. If it's just watery, give fluids and wait."

This rulebook makes sense because most watery diarrhea is caused by viruses (like the flu of the stomach), and antibiotics don't work on viruses. However, there's a sneaky problem: a bacteria called Shigella can also cause watery diarrhea without any visible blood. Because the doctor can't see the blood, they follow the rulebook and don't give antibiotics. But the child actually needs them to get better and stop spreading the infection. It's like trying to find a needle in a haystack, but the needle is invisible.

The Big Question:
Can we find a simple, cheap "smoke detector" in the stool that tells us, "Hey, there's a bacterial fire here, even if you can't see the flames (blood)?"

The Study:
Researchers from six different countries (Kenya, Bangladesh, The Gambia, Malawi, Pakistan, and Peru) teamed up to test this idea. They looked at over 4,000 children with watery diarrhea. They took their stool samples and tested them for four different "inflammation markers"—chemicals that the body releases when it's fighting an infection. Think of these markers as different types of smoke:

1. Myeloperoxidase (A chemical from white blood cells)
2. Calprotectin (Another immune signal)
3. Lipocalin-2 (A protein that fights bacteria)
4. Hemoglobin (Tiny traces of blood that are too small to see with the naked eye)

The "Aha!" Moment:
The researchers built a computer model (like a smart calculator) to see which combination of these "smoke detectors" and simple facts (like the child's age and how many times they pooped) could best predict Shigella.

Here is what they found:

• The "Invisible Blood" Detector Won: The most important clue wasn't the complex immune chemicals; it was Hemoglobin. Even though the parents and doctors couldn't see blood, the stool contained tiny, invisible traces of it. This single marker improved the detection of Shigella by 7%.
• The Other Detectors Were Noisy: The other three markers (the immune chemicals) didn't add much value. They were like smoke detectors that kept going off when you were just cooking toast—they weren't specific enough to the real fire.
• The New Scorecard: The team created a simple 10-point scorecard for doctors. It only asks three questions:
  1. How old is the child?
  2. How many times did they poop?
  3. Is there invisible blood (Hemoglobin) in the stool?

If the score is high enough, the doctor knows to treat the child for Shigella, even if the poop looks perfectly clear.

Why This Matters:
Currently, doctors either treat everyone (which wastes antibiotics and creates super-bugs) or treat no one unless they see blood (which lets sick kids get worse).

This new approach is like upgrading from a "guessing game" to a "smart sensor."

• It's Cheap: Tests for invisible blood (Hemoglobin) are already sold in pharmacies and are very affordable.
• It's Fast: A doctor can get the result in minutes.
• It's Precise: It helps doctors give antibiotics only to the kids who actually have the bacterial infection, saving the others from unnecessary medicine.

The Bottom Line:
By looking for "invisible blood" in the stool, doctors can finally catch the sneaky Shigella bacteria that hides in watery diarrhea. This simple trick could save lives, help kids recover faster, and stop the spread of antibiotic-resistant super-bacteria. It turns a blurry guess into a clear decision.

Technical Summary

1. Problem Statement

• Clinical Gap: Current World Health Organization (WHO) syndromic guidelines for diarrhea treatment recommend antibiotics only for visible bloody diarrhea (dysentery). However, Shigella infections present as watery (non-bloody) diarrhea in 40%–89% of cases, leading to significant undertreatment of children who would benefit from antibiotic therapy.
• Diagnostic Limitations: Gold-standard stool culture has low sensitivity and slow turnaround times (48 hours to 5 days). Molecular diagnostics (qPCR) are accurate but often too costly and technically complex for routine use in low-resource settings.
• Objective: The study aimed to evaluate whether fecal inflammatory biomarkers, when combined with clinical data, could improve the point-of-care identification of watery Shigella diarrhea, thereby enabling targeted antibiotic treatment and reducing antimicrobial resistance.

2. Methodology

• Study Design: An ancillary analysis of the Enterics for Global Health (EFGH)-Shigella surveillance study, a facility-based hybrid study conducted across six low-resource sites: Bangladesh, Kenya, Malawi, Pakistan, Peru, and The Gambia.
• Population: 4,191 children aged 6–35 months presenting with medically-attended watery (non-bloody) diarrhea between June 2022 and August 2024.
• Data Collection:
  • Clinical/Socio-economic Predictors: Age, sex, height-for-age Z-score, diarrhea duration, stool frequency, vomiting, dehydration, fever, transmission season, and household wealth.
  • Biomarkers: Fecal concentrations of myeloperoxidase (MPO), calprotectin, neutrophil gelatinase-associated lipocalin (lipocalin-2/NGAL), and hemoglobin were quantified using ELISA.
  • Reference Standard: Pathogen detection via quantitative PCR (qPCR) using TaqMan Array Cards (TAC) on rectal swabs or whole stool. Shigella attribution was defined by specific cycle threshold (Ct) values.
• Statistical Analysis:
  • Modeling: A SuperLearner ensemble machine learning approach was used, combining multiple base learners (GLM, Random Forest, SVM, XGBoost, etc.) to predict outcomes.
  • Validation: A leave-one-site-out cross-validation strategy was employed to ensure external validity across geographic locations.
  • Scenarios: Models were compared using: (a) all 13 predictors (clinical + biomarkers), (b) clinical predictors only, and (c) incremental addition of biomarkers.
  • Clinical Score: A parsimonious 10-point clinical score was developed using the AutoScore framework based on the most significant predictors.

3. Key Results

• Prevalence: Among the 4,083 children with valid data, 18.0% (735) had Shigella-attributable diarrhea confirmed by qPCR.
• Individual Biomarker Performance:
  • Fecal Hemoglobin demonstrated the highest discriminatory ability for Shigella (AUC = 0.65).
  • Lipocalin-2 showed the lowest performance (AUC = 0.59).
  • Biomarkers performed poorly for non-Shigella bacterial diarrhea (AUCs ~0.51–0.54).
• Model Performance (Watery Shigellosis):
  • Full Model (All 13 predictors): Achieved an AUC of 0.75 (95% CI: 0.67–0.78), with 67% sensitivity and 75% specificity.
  • Impact of Biomarkers: Excluding biomarkers reduced the AUC to 0.67 (an 8% drop).
  • Incremental Contribution: Adding Hemoglobin alone improved the model's AUC by 7% (statistically significant). Adding MPO, Calprotectin, or Lipocalin-2 provided negligible additional improvement (0–1%).
  • Site Variability: Performance varied by site, with the highest AUC in Kenya (0.78) and lowest in Peru (0.66).
• Clinical Score: A simplified 3-variable score (Age, Hemoglobin status, Stool Frequency) achieved an AUC of 0.75 with a treatment threshold of ≥6 points, matching the performance of the complex full model.
• Other Bacterial Etiologies: The models showed minimal predictive value for non-Shigella bacterial diarrhea (AUC ~0.55), indicating specificity to Shigella pathophysiology.

4. Key Contributions

1. Validation of Fecal Hemoglobin: The study identifies fecal hemoglobin as the single most critical biomarker for detecting watery Shigella diarrhea, outperforming traditional inflammatory markers like MPO and calprotectin in this specific context.
2. Superiority Over Current Guidelines: The biomarker-enhanced model outperforms current WHO syndromic guidelines (which only treat dysentery) by identifying a significant subset of watery Shigella cases that would otherwise be missed.
3. Practical Clinical Tool: Development of a simple, 10-point clinical score that relies on age, stool frequency, and a dichotomized hemoglobin test. This score is designed for low-resource settings where complex machine learning models are not feasible.
4. Robust Multi-Site Validation: By utilizing a leave-one-site-out approach across six diverse countries, the study demonstrates the generalizability of the findings across different epidemiological settings.

5. Significance and Implications

• Antibiotic Stewardship: Implementing point-of-care fecal hemoglobin tests (which are inexpensive and commercially available, e.g., iFOBT/FIT) could allow health workers to target antibiotics specifically to children with watery Shigella diarrhea. This reduces unnecessary antibiotic use in viral cases while ensuring treatment for bacterial cases.
• Clinical Impact: The proposed approach offers a scalable strategy to improve child health outcomes in low-resource settings by bridging the gap between syndromic guidelines and precise etiological diagnosis.
• Future Directions: The findings support the integration of fecal hemoglobin into triage algorithms and mobile clinical decision support systems (CDSS) to guide rational antibiotic prescribing globally.

Conclusion: The study concludes that while current guidelines miss many Shigella cases, the integration of a simple fecal hemoglobin test with basic clinical variables significantly improves the detection of watery shigellosis, offering a viable path toward optimized antibiotic use in global health.