The consequences of Shigella medically-attended diarrhoea and other leading pathogens among young children living in high-burden settings: a multi-country prospective cohort study

Summary

Background

Shigella is a leading cause of diarrheal illness in children in low- and middle-income countries, with evidence of contribution to prolonged diarrhoea, hospitalization, and impaired growth.

Methods

In this prospective cohort study, we analysed data from 8756 children aged 6–35 months with acute (onset within the last seven days after at least two diarrhoea-free days) medically attended diarrhoea (defined as three or more abnormally loose or watery stools with or without blood in the previous 24 h) enrolled between June 2022–August 2024 from health facilities in Bangladesh (1341 children from 3 health facilities), Kenya (1390 children from 6 health facilities), Malawi (1325, 1), Mali (1341, 4), Pakistan (1365, 6), Peru (1058, 5), and The Gambia (937, 2). Eligible children resided in the pre-defined catchment area and were accompanied by a consenting caregiver. We evaluated associations between Shigella-attributed diarrhoea (compared to controls with no detected diarrheagenic pathogen) and three primary outcomes: prolonged/persistent diarrhoea, hospitalization, and linear growth faltering assessed at 4 weeks and 3 month follow-up visits. Pathogen attribution was based on culture or pathogen-specific quantitative PCR thresholds. Poisson regression and linear regression with generalised estimating equations were used to assess associations, stratified by antibiotic treatment and adjusted for sociodemographic and clinical variables.

Findings

Shigella was the most prevalent attributed pathogen (20.6%). It was the only pathogen associated with both prolonged (≥seven days, adjusted Prevalence Ratio (aPR): 1.56, 95% CI: 1.34–1.82) and persistent (≥14 days, aPR: 1.66, 95% CI:1.03–2.68) diarrhoea. No associations were identified between Shigella and hospitalization or linear growth faltering. However, children with Shigella who did not receive effective antibiotics experienced greater decrements in linear growth (length-for-age z-score (LAZ)/height-for-Age Z-score (HAZ): −0.05, 95% CI: −0.09, −0.01) in the three months following infection compared to those without any pathogen detected. Rotavirus (adjusted Risk Ratio (aRR): 2.40, 95% CI: 1.53–3.76) and Cryptosporidium (aRR: 2.33, 95% CI: 1.35–4.04) were associated with hospitalization, and Cryptosporidium was linked to subsequent stunting (aRR: 1.36, 95% CI: 1.01–1.84).

Interpretation

Shigella contributes to prolonged and persistent diarrhoea and, when untreated, can contribute to linear growth faltering. Pathogen-specific diagnosis, antibiotic access, and Shigella vaccination could reduce prolonged illness and growth impairment, with meaningful population-level benefits in low- and middle-income settings.

Funding

Gates Foundation.

Research in context.

Evidence before this study

Beyond acute gastrointestinal illness, enteropathogens can cause enduring adverse health effects in children under the age of five years in low- and middle-income countries. These include the subsequent development of prolonged and persistent diarrhoea, the attenuation of linear growth in young children, hospitalization, and mortality not clearly linked to the gastrointestinal illness. Literature reviews for the association of symptomatic enteric infections and diarrhoea and these outcomes were carried out in PubMed, Embase, and Web of Science on April 21, 2026 restricted to primary research studies done in populations under five years of age. Search terms included diarrhoea and diarrhoeal diseases and the specified outcome without language filters. The search associating symptomatic enteric infections with the subsequent development of persistent diarrhoea was restricted to the last 10 years (since April 21, 2015) as the epidemiology of persistent diarrhoea has changed notably over the last 25 years but there were no date restrictions applied for growth. There were 47 articles identified that associated symptomatic enteric infections with growth, and 397 articles associating one of 14 enteropathogens with the development persistent diarrhoea. Most data have been derived from relatively small studies in diverse settings and findings were highly heterogeneous. Few have evaluated outcomes among children presenting for care across diverse settings using standardised methods or examined how antibiotic treatment modifies these effects.

Added value of this study

This large, multi-country study of medically attended diarrhoea demonstrated that Shigella was the most prevalent pathogen and the only one significantly associated with both prolonged (≥7 days) and persistent (≥14 days) diarrhoea. While no overall association was observed with hospitalization or growth faltering, children with Shigella who did not receive an effective antibiotic had modest but measurable declines in linear growth relative to children who received an effective antibiotic.

Implications of all the available evidence

Preventing Shigella through vaccination could reduce prolonged illness and support child growth in low-resource settings. Phase 3 Shigella vaccine efficacy trials and post-licensure effectiveness studies are needed to license Shigella vaccines for children and to assess benefits on long-term outcomes. In the interim, strengthening diagnostic capabilities and effective treatment of Shigella and other enteric pathogens remains essential.

Introduction

Shigella is a primary cause of severe diarrhoea in children under five years worldwide and is associated with approximately 93,831 deaths (95% CI, 35,860–185,931) and 8.4 million (3.3–16.5 million) disability-adjusted life-years (DALYs) each year.¹ Several multi-country diarrhoea aetiology studies across Africa and Asia have highlighted Shigella as a leading cause of acute diarrhoea, with 13–30% of moderate to severe (MSD) cases being attributed to Shigella in children under five years.2, 3, 4, 5 Shigella is also associated with persistent diarrhea^4,6 and linear growth faltering,⁶ outcomes that have been elusive to combat via water, sanitation, and hygiene interventions.⁷ These findings, together with the success of rotavirus vaccination, have contributed to renewed global interest in developing Shigella vaccines for children living in high Shigella burden settings.⁸ The World Health Organization (WHO) has identified the development of a safe and effective Shigella vaccine as a public health priority, particularly for children in low- and middle-income countries (LMICs).^9,10

Beyond reducing acute diarrhoea, a Shigella vaccine may play a critical role in shaping child health outcomes over the medium and long term. Persistent diarrhoea (≥14 days) increases the risk of dehydration and mortality, compromises nutritional status, and predisposes children to recurrent enteric infections.¹¹ Prolonged diarrhoea (≥seven days) disrupts nutrient absorption and gut integrity leading to undernutrition and impaired immune responses. In turn, this contributes to growth faltering, cognitive delays, and heightened vulnerability to other childhood illnesses. These cascading effects underscore the disproportionate and lasting burden of Shigella infections, providing a strong rationale for prevention through vaccination. Simulation models estimate that effective prevention of Shigella could avert 43 million stunting episodes, 600,000 deaths, and $4.4 billion in healthcare costs over 20 years.¹²

The Enterics for Global Health Shigella Surveillance Study (EFGH) was a health facility-based diarrhoea surveillance initiative designed to estimate the incidence, cost, and consequences of Shigella diarrhoea among children aged 6–35 months living in LMICs where vaccine licensure trials and early vaccine adoption may occur.^13,14 Here, we report on the associations between Shigella and longer term-outcomes, such as longer duration diarrhoea, hospitalization, and linear growth, alongside the associations of these same outcomes with the four other most prevalent diarrheal aetiologies. This analysis describes key clinical outcomes of Shigella which, combined with other evidence on Shigella burden,¹⁵ informs the value proposition of Shigella vaccines.

Methods

EFGH study design

The prospective EFGH cohort study enrolled children with a new episode (onset after two diarrhoea free days) of acute (onset within 7 days of enrollment) diarrhoea (≥3 abnormally loose or watery stools in the previous 24 h) with or without the presence of blood presenting to 29 health facilities in seven countries: Bangladesh, Kenya, Malawi, Mali, Pakistan, Peru, and The Gambia. The study sites were not selected based on demographic surveys to represent the children of any particular country or region. However, the children are considered to be broadly representative of young children in LMIC settings with varied environmental exposures, dietary practices, and nutritional and immunologic status in low resource settings globally.

Population

Each country recruited approximately 1400 children with medically-attended diarrhoea (MAD) aged 6–35 months from 2022 to 2024. Children who arrived at participating health facilities with diarrhoea—defined as three or more unusually loose or watery stools, with or without blood, in the prior 24 h—were assessed for study eligibility. To qualify, children had to live within the designated catchment area, be experiencing a new acute diarrheal episode that began within the previous seven days following at least two diarrhoea-free days, and be accompanied by a caregiver who can consent to their child’s participation. This sample size was determined to ensure high precision in Shigella culture-positive incidence estimates assuming Shigella spp. was identified in 4.7% of participants. Participants were followed prospectively at four weeks and three months following enrollment for anthropometric outcomes and vital status checks. Upon completion of follow-up, the same child could be enrolled again if they developed a new diarrhoea episode and presented to care at an EFGH facility. Details of the study sites, health facilities, methodology, and Shigella incidence rates have been previously described.^14,15

Procedures

Demographic information and medical history, including diarrhoea onset dates, severity and comorbidities were ascertained by standardised caregiver questionnaire. Race and ethnicity were not collected in this study because categories were not comparable across the seven country sites. Participant characteristics are therefore presented by country rather than by race or ethnicity. A clinical examination was conducted by a trained study clinician to assess vital signs, WHO-defined dehydration signs, and other clinical symptoms. Three flocked rectal swabs were collected from each child at enrollment prior to antibiotic administration. Two swabs (one in Cary-Blair media, one in modified buffered glycerol saline) were cultured for Shigella spp. and a third swab underwent quantitative PCR.^16,17

Clinical management of cases followed national guidelines, including rehydration and therapeutic zinc. Antibiotics were administered during the enrollment visit based on indications outlined in the WHO syndromic diarrhoea treatment guidelines (including antibiotics for dysentery)¹⁸ and clinician discretion. Antibiotics effective against Shigella were also given if a child with culture-confirmed Shigella was not doing well upon caregiver contact with the Shigella culture results (which occurred 2–7 days after enrollment, depending on culture isolation time). Caregivers completed a daily diarrhoea diary to document the frequency and severity of diarrhoea in the fourteen days following enrollment. During the scheduled week-four and month-three follow-up visits, study clinicians assessed the participant’s condition and documented the child’s health history since the last visit, including recovery from the initial (index) diarrhoea episode, new diarrheal episodes, other illnesses, and medication use using a standardised questionnaire. Information about the index diarrheal episode was compiled using the enrollment visit, diarrhoea diary, and caregiver recall at follow-up visits. If a hospitalization was reported, the date, length of stay, diagnosis, and treatments were collected. If a participant returned to an EFGH health facility for an illness-related unscheduled visit within the three-month follow-up period, a study clinician assessed the child’s condition and documented any treatment provided.

Length (age <24 months) or height (age ≥24 months), weight, and mid-upper arm circumference (MUAC) were measured at enrollment and follow-up visits. Length and height was measured twice to the nearest 0.1 cm using a ShorrBoard, resetting the board between each measurement. Children were weighed with minimal clothing to the nearest 0.1 kg using a digital scale. MUAC was measured to the nearest 0.1 cm using a single-slotted insertion tape. If length/height, weight or MUAC measurements differed by ≥0.5 cm, ≥0.2 kg or ≥0.3 cm, respectively, a third measurement was taken and the average of the closest two measurements was considered the final measurement.¹⁶ Z-scores were calculated using the 2006 WHO growth standards. To ensure measurement reliability and consistency across sites, routine quality control and quality assurance procedures were implemented, including six-monthly standardization exercises for anthropometry teams.

Definitions

Shigella-attributed diarrhoea was defined as Shigella detected by culture isolation or amplification by TaqMan Array card (TAC) qPCR using the ipaH target to a cycle threshold (Ct) of <29.5 as described elsewhere.^16,17 Other possible diarrhoea-attributable enteric pathogens were determined by TAC using the following Ct thresholds): rotavirus <33.6, Heat Stable Enterotoxigenic Escherichia coli (ST-ETEC) <25.6, Cryptosporidium <25.8, and adenovirus <24.2. The TAC assay included primers and probes for both heat-stable (ST) and heat-labile (LT) enterotoxigenic E. coli. Infections were categorized as ST-ETEC if ST alone or both ST and LT were detected, whereas LT-only detections were classified as LT-ETEC. Cut points for these and other aetiologies were established using whole stool results from diarrhoea case control analyses and adjusted for rectal swabs for each pathogen by using a swab specific cut-point.¹⁵ We considered “no detected pathogen” when none of the following pathogens were detected at the lower limit of TAC detection (Ct < 35): adenovirus 40/41, Aeromonas spp., astrovirus, Campylobacter jejuni/coli, cryptosporidium, Cyclospora cayetanensis, Cystoisospora belli, Entamoeba histolytica, norovirus GII, rotavirus, Salmonella enterica, sapovirus, Shigella, ST-ETEC, typical enteropathogenic E. coli, Vibrio cholerae. Prolonged diarrhoea was defined as a diarrhoea episode that lasted ≥seven days in duration, and persistent diarrhoea as an episode lasting ≥14 days. Hospitalization data was collected through caregiver recall and, when available, verified using hospital records (considered the gold standard). Hospitalization was defined as an overnight stay in the hospital where the child remained in the ward from at least 12:00 am to 6:00 am. Breastfeeding history was collected from caregivers during the enrollment health history survey including whether or not the child had been breastfed and if so, the age the child was first given specific liquids or foods, including water, formula, animal milk, tea, juice, rice, and porridge. Exclusive breastfeeding duration was defined using the youngest reported age at introduction of any non-breast-milk food or fluid.

Wealth quintiles were calculated based on validated questionnaires and calculation methods derived from the Demographic and Health Survey.¹⁹ For the growth analysis, we calculated length-for-age z-scores (LAZ) or height-for-age z-scores (HAZ) using 2006 WHO growth standards.²⁰ Stunting was classified as a LAZ/HAZ of less than −2. Change in linear growth (ΔLAZ/HAZ) was calculated by the difference in LAZ/HAZ between enrollment and the three-month follow-up visit.

Likely effective antibiotics were antibiotics currently or previously recommended by the WHO for Shigella (ampicillin, azithromycin, ceftriaxone, ciprofloxacin, nalidixic acid, trimethoprim/sulfamethoxazole, pivmecillinam) for which ≥70% of a given country site’s EFGH Shigella culture isolates were susceptible as determined by disk diffusion testing on all available isolates. Ceftazidime, cefpodoxime, and cefuroxime were also considered effective if site-level ceftriaxone susceptibility was ≥70%. Possibly effective antibiotics were the recommended antibiotics at sites with <70% susceptibility and those that have plausible efficacy, but for which we did not test susceptibility: gentamicin, ampiclox, clarithromycin, erythromycin, doxycycline, tetracycline, cefixime, chloramphenicol, amoxicillin, nitrofurantoin, meropenem, amikacin, thiamphenicol, nifuroxazide, furazolidone, levofloxacin, and cefaclor. To retain power and create a clearly untreated group, analyses combined likely and possibly effective antibiotics together. Ineffective antibiotics included flucloxacillin, cloxacillin, entamizole, cefadroxil, metronidazole, and penicillin, or if a child was not given any antibiotics (Supplementary Fig. S1).

Statistical analysis

The detailed statistical analysis plan is available on ClinicalTrials.gov #NCT06047821. Risks of prolonged diarrhoea, persistent diarrhoea, hospitalization (at any point during the three-month follow-up period, during the index episode, and following diarrhoea resolution), and stunting at the three-month visit were compared between children with each pathogen of interest at attributable levels to those without any detected pathogen among children enrolled in EFGH with valid qPCR results and non-missing length/height measurements. The comparison group was defined as no detectable pathogen. We used modified Poisson models with generalised estimating equations accounting for clustering by children enrolled more than once to measure the impact of each of the top five most prevalent pathogens (including Shigella) on diarrhoea duration, risk of hospitalization, and risk of linear growth faltering as assessed by stunting. The Poisson approach, modified with the inclusion of robust (sandwich) standard errors, was selected because it provides valid estimates for binary outcomes even under potential misspecification of the Poisson variance structure. Missing data was excluded from the analysis as most was missing at random. The analysis for stunting excluded participants with stunting at enrollment. The difference in ΔLAZ/HAZ between those with an attributable pathogen to those with no detected pathogen was assessed using linear regression similarly accounting for clustering by child. This approach was used because linear regression estimates associations with continuous outcomes and is robust even under moderate deviations from normality. Robust standard errors were incorporated to account for some heteroskedasticity. All regression analyses additionally stratified Shigella-attributable and ST-ETEC-attributable cases by those that received and did not receive likely or possibly effective antibiotics to disaggregate the impact of antibiotic treatment on the bacterial pathogens (Shigella and ST-ETEC) under investigation. All model assumptions were verified and all multivariate models were adjusted for enrollment country, age at enrollment in months, maternal education, household wealth quintile, and months of exclusive breastfeeding. These variables were selected, a priori, based on hypothesized associations with the pathogens and outcomes (diarrhoea duration, hospitalization, and ΔLAZ/HAZ) of interest. A sensitivity analysis was performed using diarrhoea duration as a continuous variable to ensure the robustness of associations seen with prolonged and persistent diarrhoea. No other sensitivity analyses were conducted. We presented semi-adjusted (adjusted for country of enrollment only) and fully adjusted models to demonstrate the potential impact of confounding and precision variables of the measures of association. Models utilizing LAZ/HAZ data were additionally adjusted for baseline LAZ/HAZ and duration between enrollment and month three measurements. Data were stored in the Research Electronic Data Capture system and analysed using R version 4.2.2. All analysis methods were pre-specified.

Ethics

This study was conducted according to Good Clinical Practice (GCP), including Good Clinical Laboratory Practice (GCLP), the Declaration of Helsinki, IRB and local rules and regulations specific to each EFGH country. This protocol was subject to ethical approval from the Institutional Review Boards (IRBs) at each EFGH site. Written informed consent, in the language of the caregivers’ choosing, was obtained from the parents or caregivers of all study participants prior to their enrollment. The specific names of the authorities that granted ethical approval for the study are available in the Supplementary Materials (p. 3), along with the reference numbers.

Role of the funding source

This study was funded by the Gates Foundation. The funders had no role in the study design, data collection, data analysis, interpretation of the data, or writing the report.

Results

Among 29,355 children screened, 9476 children were enrolled in EFGH and 8756 (92.4%) were included in the diarrhoea duration and hospitalization analyses representing 7985 unique children (597 were enrolled twice, 60 three times, and 18 four times). Reasons for non-participation were reported elsewhere.¹⁵ Among 720 children not included in this analysis, 684 (95.0%) were excluded due to failed qPCR and 36 (5.0%) were excluded for missing a key variable. Failed qPCR included specimens that could not be linked to participant data because of specimen identification errors and specimens with failed assay controls. These exclusions were considered likely to be at random. 7978 children were included in the linear growth analyses, with the additional 778 (8.2%) children excluded due to a missing length or height measurement from a missed visit (4.5% of total) or a measurement recording error (3.7% of total). Of the children included in this analysis, 1341 were from Bangladesh, 1390 from Kenya, 1324 from Malawi, 1341 from Mali, 1365 from Pakistan, 1058 from Peru, and 937 from the Gambia. Over one-third (37.0%) of children were aged 6–11 months, 45.5% were female, and 43.4% came from households classified as being in the lowest two national wealth quintiles (Table 1). Clinically, 12.8% of children experienced dysentery (as abstracted from the clinical record or by caregiver report) with the lowest prevalence of dysentery observed in Mali (6.6%) and highest prevalence in the Gambia (17.8%). At enrollment, 23.6% of children were stunted, 16.1% were wasted, and 20.9% were underweight. The highest proportion of stunted, wasted and underweight children was in Pakistan (38.4%, 33.8%, and 42.5%, respectively).

Table 1.

Participant characteristics and baseline clinical features.

	Bangladesh	Kenya	Malawi	Mali	Pakistan	Peru	The Gambia	Overall
Enrollmentsa	1341	1390	1324	1341	1365	1058	937	8756
Recruitment dates	Jun 21, 2022–Jun 20, 2024	Aug 1, 2022–Jul 31, 2024	Aug 3, 2022–Aug 2, 2024	Aug 16, 2022–Aug 15, 2024	Aug 25, 2022–Aug 24, 2024	Aug 9, 2022–Aug 8, 2024	Aug 16, 2022–Aug 15, 2024
Participant demographics
Age
6–11 months	566 (42.2)	559 (40.2)	453 (34.2)	576 (43.0)	442 (32.4)	343 (32.4)	305 (32.6)	3244 (37.0)
12–23 months	580 (43.3)	589 (42.4)	597 (45.1)	587 (43.8)	632 (46.3)	557 (52.6)	481 (51.3)	4023 (45.9)
24–35 months	195 (14.5)	242 (17.4)	274 (20.7)	178 (13.3)	291 (21.3)	158 (14.9)	151 (16.1)	1489 (17.0)
Sex
Female	588 (43.8)	633 (45.5)	616 (46.5)	616 (45.9)	641 (47.0)	456 (43.1)	432 (46.1)	3982 (45.5)
Male	753 (56.2)	757 (54.5)	708 (53.5)	725 (54.1)	724 (53.0)	602 (56.9)	505 (53.9)	4774 (54.5)
Mother’s highest achieved education level
None	170 (12.7)	6 (0.4)	28 (2.1)	280 (20.9)	395 (28.9)	3 (0.3)	324 (34.6)	1206 (13.8)
Less than primary school	109 (8.1)	120 (8.6)	70 (5.3)	299 (22.3)	95 (7.0)	93 (8.8)	37 (3.9)	823 (9.4)
Primary school only	218 (16.3)	487 (35.0)	487 (36.8)	196 (14.6)	243 (17.8)	70 (6.6)	94 (10.0)	1795 (20.5)
Some secondary school	447 (33.3)	376 (27.1)	530 (40.0)	212 (15.8)	321 (23.5)	315 (29.8)	125 (13.3)	2326 (26.6)
Secondary school or greater	397 (29.6)	401 (28.8)	209 (15.8)	124 (9.2)	279 (20.4)	577 (54.5)	40 (4.3)	2027 (23.1)
Quranic school only	0 (0.0)	0 (0.0)	0 (0.0)	230 (17.2)	32 (2.3)	0 (0.0)	317 (33.8)	579 (6.6)
Household wealth quintileb
Quintile 1 (fewest assets)	0 (0.0)	34 (2.4)	151 (11.4)	23 (1.7)	450 (33.0)	486 (45.9)	230 (24.5)	1374 (15.7)
Quintile 2	82 (6.1)	246 (17.7)	420 (31.7)	409 (30.5)	461 (33.8)	401 (37.9)	410 (43.8)	2429 (27.7)
Quintile 3	270 (20.1)	402 (28.9)	513 (38.7)	574 (42.8)	229 (16.8)	164 (15.5)	235 (25.1)	2387 (27.3)
Quintile 4	422 (31.5)	517 (37.2)	185 (14.0)	324 (24.2)	166 (12.2)	7 (0.7)	62 (6.6)	1683 (19.2)
Quintile 5 (most assets)	567 (42.3)	191 (13.7)	55 (4.2)	11 (0.8)	59 (4.3)	0 (0.0)	0 (0.0)	883 (10.1)
Exclusively breastfed for ≥6 months	866 (64.6)	969 (69.7)	688 (52.0)	705 (52.6)	703 (51.5)	693 (65.5)	463 (49.4)	5087 (58.1)
Baseline clinical characteristics
Dysenteryc	196 (14.6)	148 (10.6)	129 (9.7)	89 (6.6)	232 (17.0)	162 (15.3)	167 (17.8)	1123 (12.8)
Dehydrationd
None	1177 (87.8)	653 (47.0)	1275 (96.3)	1179 (87.9)	1156 (84.7)	205 (19.4)	878 (93.7)	6523 (74.5)
Some	162 (12.1)	678 (48.8)	47 (3.5)	152 (11.3)	201 (14.7)	847 (80.1)	44 (4.7)	2131 (24.3)
Severe	2 (0.1)	59 (4.2)	2 (0.2)	10 (0.7)	8 (0.6)	6 (0.6)	15 (1.6)	102 (1.2)
Hospitalized
During index episode	112 (8.4)	58 (4.2)	1 (0.1)	1 (0.1)	9 (0.7)	8 (0.8)	33 (3.5)	222 (2.5)
After resolution of index episode	19 (1.4)	24 (1.7)	13 (1.0)	0 (0.0)	16 (1.2)	6 (0.6)	37 (3.9)	115 (1.3)
Anthropometry
MUAC (cm)	13.6 (12.9, 14.4)	14.6 (13.8, 15.4)	14.4 (13.6, 15.1)	13.6 (12.9, 14.4)	13.1 (12.5, 13.9)	14.6 (13.9, 15.2)	13.6 (13.0, 14.2)	13.9 (13.1, 14.8)
WHZ	−0.6 (−1.4, 0.1)	0.2 (−0.5, 0.9)	−0.1 (−0.8, 0.8)	−1.1 (−1.8, −0.3)	−1.3 (−2.0, −0.6)	−0.1 (−0.8, 0.5)	−1.3 (−2.0, −0.6)	−0.6 (−1.4, 0.2)
WAZ	−1.1 (−1.9, −0.3)	−0.3 (−1.1, 0.4)	−0.7 (−1.5, 0.1)	−1.2 (−1.9, −0.5)	−1.8 (−2.6, −1.0)	−0.7 (−1.4, 0.0)	−1.5 (−2.3, −0.8)	−1.0 (−1.8, −0.2)
LAZ/HAZ	−1.2 (−2.0, −0.4)	−0.9 (−1.8, −0.2)	−1.4 (−2.1, −0.6)	−0.8 (−1.5, −0.1)	−1.6 (−2.5, −0.8)	−1.1 (−1.8, −0.4)	−1.1 (−1.9, −0.3)	−1.2 (−1.9, −0.4)
Stunted	322 (24.0)	255 (18.3)	381 (28.8)	180 (13.4)	524 (38.4)	223 (21.1)	182 (19.4)	2067 (23.6)
Wasted	236 (17.6)	60 (4.3)	61 (4.6)	304 (22.7)	461 (33.8)	46 (4.3)	244 (26.0)	1412 (16.1)
Underweight	273 (20.4)	101 (7.3)	165 (12.5)	302 (22.5)	580 (42.5)	103 (9.7)	307 (32.8)	1831 (20.9)
Antibiotic use
Likely effective antibioticse	64 (4.8)	311 (22.4)	136 (10.3)	816 (60.9)	18 (1.3)	106 (10.0)	157 (16.8)	1608 (18.4)
Possibly effective antibioticsf	1104 (82.3)	588 (42.3)	212 (16.0)	272 (20.3)	572 (41.9)	232 (21.9)	43 (4.6)	3023 (34.5)
Ineffective antibioticsg	14 (1.0)	236 (17.0)	60 (4.5)	139 (10.4)	11 (0.8)	40 (3.8)	32 (3.4)	532 (6.1)
No antibiotics received	159 (11.9)	255 (18.3)	916 (69.2)	114 (8.5)	764 (56.0)	680 (64.3)	705 (75.2)	3593 (41.0)

MUAC = mid-upper arm circumference in centimetres, WHZ = weight-for-height z-score, WAZ = weight-for-age z-score, LAZ/HAZ = length-for-age z-score (age <24 months or if child is unable to stand) or height-for-age Z-score (age ≥ 24 months or able to stand).

^aParticipants may be enrolled again if they seek care for a new diarrhoea episode after completing the three months of study follow-up from their previous enrollment. The 8756 enrollments represent 7985 unique children.

^bWealth index quintiles calculated from country-specific wealth assessments derived from Demographic Health Surveys as detailed in Chakraborty, GHSP, 2016.

^cBlood in the stool as reported by caregiver during the diarrhoeal episode or by clinical diagnosis.

^dBased on IMCI criteria classification of some or severe dehydration. Severe dehydration = At least two of the following signs: lethargy, abnormally sunken eyes, drinks poorly, skin pinch >2 s. Some dehydration = At least two of the following signs: restless/irritable, abnormally sunken eyes, drinks eagerly, skin pinch 1–2 s.

^eLikely effective antibiotics include ampicilliln, nalidixic acid, ceftriaxone, pivmecillinam, azithromycin, ciprofloxacin, or trimethoprim/sulfamethoxazole if site-level susceptibility to each antibiotic is ≥ 70% among Shigella isolates. Ceftazidime, cefpodoxime, and cefuroxime are also included here if ceftriaxone site-level susceptibility is ≥ 70%.

^fPossibly effective antibiotics include ampicillin, nalidixic acid, ceftriaxone, pivmecillinam, azithromycin, ciprofloxacin, or trimethoprim/sulfamethoxazole if site-level susceptibility is <70% among Shigella isolates. Also included is ceftazidime, cefpodoxime, and cefuroxime if site-level ceftriaxone susceptibility is <70%. Regardless of susceptibility, amoxicillin, cefixime, chloramphenicol, doxycycline, erythromycin, tetracycline, clarithromycin, gentamicin, ampiclox, augmentin, amikacin, cefaclor, furazolidone, levofloxacin, meropenem, nifuroxizide, nitrofurantoin, and thiamphenicol are also considered possibly effective.

^gAntibiotics deemed likely ineffective against bacterial enteric pathogens include metronidazole, penicillin, cefadroxil, cloxacillin, entamizole, flucloxacillin, and mupirocin.

Shigella was the most prevalent pathogen, detected at attributable qPCR levels or by culture in 1803 (20.6%) of MAD cases, followed by rotavirus (12.6%), ST-ETEC (9.8%), Cryptosporidium (6.5%), adenovirus (3.7%), norovirus GII (2.7%), C. jejuni/coli (2.2%), and sapovirus (2.1%). No attributable pathogen was found in 49.7% of cases and no detected pathogen in 15.5% of cases. Among 1803 children with Shigella-attributed diarrhoea, 28.4% received a likely effective antibiotic, 34.5% received a possibly effective antibiotic, 6.1% received an ineffective antibiotic, and 31.0% received no antibiotic (Table 1).

Diarrhoea episodes lasted a median of four days (IQR: three, six), with persistent diarrhoea reported in 2.4% and prolonged diarrhoea in 19.6% of enrolled episodes. Persistent diarrhoea was most prevalent in Pakistan (6.0%) and least prevalent in Mali (<0.1%). Shigella was the only enteric pathogen associated with both prolonged and persistent diarrhoea. Over a quarter (26.8%) of children with Shigella had prolonged diarrhoea and 4.9% had persistent diarrhoea, compared to 14.9% and 1.8%, respectively, in children with no pathogen (adjusted prevalence ratio [aPR]_prolonged: 1.56 [95% CI: 1.34, 1.82]; aPR_persistent: 1.66 [95%CI: 1.03, 2.68]); Fig. 1, Supplementary Table S1). ST-ETEC and Cryptosporidium were also associated with prolonged diarrhoea (aPR 1.26 [95%CI: 1.02, 1.57] and aPR 1.55 [95%CI: 1.25, 1.93], respectively), but not with persistent diarrhoea (Fig. 1, Supplementary Table S1). Stratification by treatment with a likely or possibly effective antibiotic did not meaningfully change the associations (Fig. 1, Supplementary Table S1). When diarrhoea duration was examined as a continuous variable, the same three pathogens were associated with longer duration (data not shown).

Fig. 1.

Prevalence of persistent (≥14 days) and prolonged (≥7 days) diarrhoea in the three months following the index MAD episode, in children with and without Shigella-attributed diarrhoea as well as other leading attributable pathogens. Footnotes: Points represent prevalence ratio estimates and horizontal bars represent 95% confidence intervals. The dotted vertical line at 1.0 indicates the null. CI: 95% confidence interval, PR: prevalence ratio, ST-ETEC: heat stable enterotoxigenic E. coli. ∗Persistence is defined by having 14 or more days of diarrhoea during the index diarrhoea episode. ^†Pathogen attribution is determined based on previously established qPCR Ct thresholds. No detectable pathogen indicates that no pathogen was present at any Ct < 35. ^‡Adjusted for enrollment site, age at enrollment, maternal education, breastfeeding history, wealth quintile. §Shigella determined if qPCR attributable at Ct < 29.5 or culture positive. ^¶Antibiotic treatment includes any prescription or administration of likely or possibly effective antibiotics given prior to enrollment if part of the same illness episode, administered at enrollment, prescribed for home use at enrollment, or following Shigella culture positive result. Likely or possibly effective antibiotics include: ampicillin, azithromycin, ceftriaxone, ciprofloxacin, nalidixic acid, trimethoprim/sulfamethoxazole, pivmecillinam, gentamicin, ampiclox, clarithromycin, erythromycin, doxycycline, tetracycline, cefixime, cefuroxime, chloramphenicol, augmentin, amoxicillin, nitrofurantoin, meropenem, amikacin, thiamphenicol, nifuroxazide, furazolidone, levofloxacin, cefaclor, cefpodoxime, ceftazidime. Antibiotics considered not effective include: flucloxacillin, mupirocin, cloxacillin, entamizole, cefadroxil, metronidazole, and penicillin. Participants who did not receive antibiotics are categorized as not effective. ‖Prolonged diarrhoea is defined by having 7 or more days of diarrhoea during the index diarrhoea episode.

Hospitalizations were rare in this cohort, experienced by 3.4% of enrolled children (2.4% during the index episode and 1.2% after resolution). Children with Shigella-attributed diarrhoea had no statistically significant difference in overall hospitalization risk (aRR: 1.50 [95%CI 0.89, 2.52], Fig. 2, Supplementary Table S2) compared to children with no pathogen. Children with rotavirus (aRR: 2.40 [95%CI: 1.53, 3.76]) and Cryptosporidium (aRR: 2.33 [95%CI: 1.35, 4.04]) had more than twice the risk of hospitalization than children with no pathogen. Antibiotic stratification did not impact the association between Shigella and hospitalization risk, but among children with ST-ETEC, those untreated or treated with an ineffective antibiotic did demonstrate a higher hospitalization risk than children with no pathogen (aRR: 2.99 [95%CI: 1.26, 7.09]). Stratifying hospitalizations between those that occurred during the index episode and those occurring after diarrhoea resolution revealed similar associations, with Cryptosporidium maintained as the only pathogen associated with hospitalizations both during and after the index episode (aRR: 2.20 [95%CI: 1.09, 4.44], Supplementary Table S2).

Fig. 2.

Risk of hospitalization in the three months following the index MAD episode, in children with and without Shigella-attributed diarrhoea as well as other leading attributable pathogens. Footnotes: Points represent risk ratio estimates and horizontal bars represent 95% confidence intervals. The dotted vertical line at 1.0 indicates the null. CI: 95% confidence interval, RR: risk ratio, ST-ETEC: heat stable enterotoxigenic E. coli. ∗Pathogen attribution is determined based on previously established qPCR Ct thresholds. No detectable pathogen indicates that no pathogen was present at any Ct < 35. ^†Adjusted for enrollment site, age at enrollment, maternal education, breastfeeding history, wealth quintile. ^‡Shigella determined if qPCR attributable at Ct < 29.5 or culture positive. ^§Antibiotic treatment includes any prescription or administration of likely or possibly effective antibiotics given prior to enrollment if part of the same illness episode, administered at enrollment, prescribed for home use at enrollment, or following Shigella culture positive result. Likely or possibly effective antibiotics include: ampicillin, azithromycin, ceftriaxone, ciprofloxacin, nalidixic acid, trimethoprim/sulfamethoxazole, pivmecillinam, gentamicin, ampiclox, clarithromycin, erythromycin, doxycycline, tetracycline, cefixime, cefuroxime, chloramphenicol, augmentin, amoxicillin, nitrofurantoin, meropenem, amikacin, thiamphenicol, nifuroxazide, furazolidone, levofloxacin, cefaclor, cefpodoxime, ceftazidime. Antibiotics considered not effective include: flucloxacillin, mupirocin, cloxacillin, entamizole, cefadroxil, metronidazole, and penicillin. Participants who did not receive antibiotics are categorized as not effective.

On average, participants returned for their three-month follow-up visit on day 89 (range: 84–127). The median loss in LAZ/HAZ between enrollment and month three was 0.29 (IQR: 0.03, 0.61). LAZ/HAZ declined most sharply among children aged 6–11 months (−0.56 [IQR:0.21, 0.90]), followed by 12–23 months (0.25 [IQR:0.01, 0.50]), with smaller reductions observed in the 24–35-month group (−0.11 [IQR: −0.05, 0.25], Fig. 3).

Fig. 3.

Mean LAZ/HAZ by timepoint stratified by age at enrollment.

There was no overall association between Shigella and linear growth in the three months following diarrhoea presentation compared to those without enteric pathogens detected. However, children who did not receive antibiotic treatment or who were treated with an ineffective antibiotic demonstrated a modestly greater loss in LAZ/HAZ compared to untreated children with no pathogen (mean difference in ΔLAZ/HAZ −0.05 [95%CI: −0.09, −0.01], Fig. 4, Supplementary Table S3). No other pathogens were found to be associated with linear growth. Among children who were not stunted at presentation, only children with Cryptosporidium were at increased risk of becoming stunted three months after enrollment compared to those with no pathogen (aRR: 1.36 [95%CI: 1.01, 1.84]; Table 2).

Fig. 4.

Mean three-month change in LAZ/HAZ by presence of leading etiologies of the index MAD. Footnotes: Points represent estimates for the difference in ΔLAZ/HAZ and horizontal bars represent 95% confidence intervals. The dotted vertical line at 0 indicates the null (no difference). CI: 95% confidence interval, LAZ/HAZ: length-for-age (age <24 months) or height-for-age (age ≥24 months) z-score, SD: standard deviation, ST-ETEC: heat stable enterotoxigenic E. coli. ∗Pathogen attribution is determined based on previously established qPCR Ct thresholds. No detectable pathogen indicates that no pathogen was present at any Ct < 35. ^†Adjusted for enrollment site, age at enrollment, maternal education, breastfeeding history, wealth quintile. ^‡Shigella determined if qPCR attributable at Ct < 29.5 or culture positive. ^§Antibiotic treatment includes any prescription or administration of likely or possibly effective antibiotics given prior to enrollment if part of the same illness episode, administered at enrollment, prescribed for home use at enrollment, or following Shigella culture positive result. Likely or possibly effective antibiotics include: ampicillin, azithromycin, ceftriaxone, ciprofloxacin, nalidixic acid, trimethoprim/sulfamethoxazole, pivmecillinam, gentamicin, ampiclox, clarithromycin, erythromycin, doxycycline, tetracycline, cefixime, cefuroxime, chloramphenicol, augmentin, amoxicillin, nitrofurantoin, meropenem, amikacin, thiamphenicol, nifuroxazide, furazolidone, levofloxacin, cefaclor, cefpodoxime, ceftazidime. Antibiotics considered not effective include: flucloxacillin, mupirocin, cloxacillin, entamizole, cefadroxil, metronidazole, and penicillin. Participants who did not receive antibiotics are categorized as not effective.

Table 2.

Prevalence of stunting at month-three follow-up among those not stunted at enrollment by diarrhoea aetiology.

Pathogena	Enrollment	Month three
	n/N (%)	n/N (%)b	aRRc (CI)
Shigellad No detected pathogen	459/1655 (27.7%) 275/1201 (22.9%)	169/1196 (14.1%) 90/926 (9.7%)	1.17 (0.94, 1.47) Ref
Shigella, with effective antibiotic treatmente No detected pathogen, with effective antibiotic treatmente	285/1054 (27.0%) 136/595 (22.9%)	111/769 (14.4%) 39/459 (8.5%)	1.27 (0.94, 1.72) Ref
Shigella, none or ineffective antibiotic treatmente No detected pathogen, none or ineffective antibiotic treatmente	174/601 (29.0%) 139/606 (22.9%)	58/427 (13.6%) 51/467 (10.9%)	1.10 (0.80, 1.50) Ref
Rotavirus No detected pathogen	188/1035 (18.2%) 275/1201 (22.9%)	134/847 (15.8%) 90/926 (9.7%)	1.19 (0.95, 1.51) Ref
ST-ETEC No detected pathogen	206/800 (25.8%) 275/1201 (22.9%)	80/594 (13.5%) 90/926 (9.7%)	1.07 (0.82, 1.40) Ref
ST-ETEC, with effective antibiotic treatmente No detected pathogen, with effective antibiotic treatmente	113/426 (26.5%) 136/595 (22.9%)	46/313 (14.7%) 39/459 (8.5%)	1.17 (0.82, 1.65) Ref
ST-ETEC, none or ineffective antibiotic treatmente No detected pathogen, none or ineffective antibiotic treatmente	93/374 (24.9%) 139/606 (22.9%)	34/281 (12.1%) 51/467 (10.9%)	0.98 (0.66, 1.45) Ref
Cryptosporidium No detected pathogen	114/524 (21.8%) 275/1201 (22.9%)	50/410 (12.2%) 90/926 (9.7%)	1.36 (1.01, 1.84) Ref
Adenovirus No detected pathogen	66/295 (22.4%) 275/1201 (22.9%)	37/229 (16.2%) 90/926 (9.7%)	1.16 (0.84, 1.62) Ref

Bold text indicates statistical significance. Abbreviations: CI = 95% confidence interval, aRR = adjusted risk ratio, qPCR = quantitative polymerase chain reaction, ST-ETEC = heat stable enterotoxigenic E. coli.

^aPathogen attribution is determined based on previously established qPCR Ct thresholds. No detectable pathogen indicates that no pathogen was present at any Ct < 35.

^bSummarised among those not stunted at enrollment.

^cAdjusted for enrollment site, age, maternal education, breastfeeding history, wealth quintile, baseline LAZ/HAZ, and days from enrollment to follow-up.

^dShigella determined if qPCR attributable at Ct < 29.5 or culture positive.

^eAntibiotic treatment includes any prescription or administration of likely or possibly effective antibiotics given prior to enrollment if part of the same illness episode, administered at enrollment, prescribed for home use at enrollment, or following Shigella culture positive result. Likely or possibly effective antibiotics include: ampicillin, azithromycin, ceftriaxone, ciprofloxacin, nalidixic acid, trimethoprim/sulfamethoxazole, pivmecillinam, gentamicin, ampiclox, clarithromycin, erythromycin, doxycycline, tetracycline, cefixime, cefuroxime, chloramphenicol, augmentin, amoxicillin, nitrofurantoin, meropenem, amikacin, thiamphenicol, nifuroxazide, furazolidone, levofloxacin, cefaclor, cefpodoxime, ceftazidime. Antibiotics considered not effective include: flucloxacillin, mupirocin, cloxacillin, entamizole, cefadroxil, metronidazole, and penicillin. Participants who did not receive antibiotics are categorized as not effective.

Discussion

In this large, multi-country study of MAD in children aged 6–35 months, Shigella emerged as the most prevalent pathogen and greatest contributor to longer duration diarrhoea. While we did not find Shigella to be linked to hospitalization or overall linear growth faltering in the EFGH cohort, the subset of children with inadequately treated Shigella infections did have modest declines in linear growth compared to children with no pathogen. Given the high prevalence of Shigella across all sites, even modest declines in linear growth accumulate at a population level and represent a preventable loss in human potential.²¹

One quarter (26.8%) of children with Shigella-attributed diarrhoea had symptoms lasting at least seven days and 4.9% persisted at least two weeks. This is consistent with previous studies which highlight Shigella as a key driver of more severe diarrhoea and longer diarrheal episodes in young children.^4,6,22 The pathogen’s invasive nature likely contributes to its prolonged clinical course.²³ However, the majority of Shigella cases in our cohort did not present with visible blood in stool, the most common sign of intestinal invasion and indication for treatment according to WHO guidelines. Cryptosporidium and ST-ETEC were also associated with prolonged diarrhoea, and only Cryptosporidium was linked to persistent episodes alongside Shigella. The importance of Cryptosporidium as a contributor to persistent diarrhoea is well-documented, especially in undernourished or immunocompromised populations.^4,24 Persistent diarrhoea is associated with mortality,^25,26 likely through pathways that involve malnutrition.¹¹ Targeting pathogens that are implicated in persistent diarrhoea may interrupt the cycles of diarrhoea and malnutrition that contribute to diarrhoea mortality unaddressed by appropriate rehydration.

Despite its known severity, Shigella was not associated with hospitalization in this cohort, but rotavirus and Cryptosporidium were. This could reflect the unique pathology of these pathogens, although we did not observe major differences in dehydration between these pathogens (24%–28%) which could have explained differences in hospital admission. Rotavirus vaccine information among EFGH participants has been reported elsewhere¹⁵ and ranged from 67% to 98% in the six countries where the vaccine has been introduced. Shigella is second only to rotavirus as a leading cause of diarrhoea hospitalizations in the Global Paediatric Diarrhoea Surveillance (GPDS) Network⁵ which may differ from EFGH in terms of health-facility levels or differences in endemicity or care-seeking between these two surveillance networks. EFGH also enrolled from the point of diarrhoea presentation which could have averted Shigella-related hospitalizations if antibiotics were given early. The low hospitalization rate in EFGH overall (3.4% of diarrhoea cases) may reflect early care-seeking, appropriate case management, or selection bias against children with the most severe comorbidities.

Children with Shigella-attributable diarrhoea experienced declines in linear growth over the three-month follow-up. This aligns with reported data from the multicountry Antibiotics for Children with Severe Diarrhoea Trial (ABCD),²⁷ the GEMS study,²⁸ and the MAL-ED study,²⁹ all of which found associations between Shigella and stunted growth. In our study, the average decline in LAZ/HAZ in Shigella was not statistically significantly worse than among children with no pathogen, except among untreated cases. This contrasts with earlier studies that showed small but reproducible decrements in linear growth.^30,31 This may be due to progressive improvements in nutritional status and resilience of populations or diagnostics that are increasingly sensitive and detect the infection in milder cases of symptomatic disease as opposed to episodes that are detected by culture alone.

Unlike for linear growth, we did not observe major differences in the relationship between Shigella and diarrhoea duration when stratifying by receipt of antibiotics. This is counter to other studies, including trials of watery and dysenteric diarrhoea, in which antibiotics reduced diarrhoea duration among children with Shigella.^27,32 Although we made efforts to account for differences in antibiotic indication by comparing Shigella to no pathogens within groups of children treated and not treated with antibiotics and for variability in likely effectiveness of the antibiotics, we caution against making causal inference from these observational findings. While this analysis did not investigate the effect of antibiotic treatment on illness outcomes, our findings and future analyses may nonetheless have implications for treatment policies and highlights the need for clinical trial evidence of antibiotics for enteric bacteria such as Shigella.

By preventing longer duration diarrhoea and growth deficits, a Shigella vaccine could contribute to downstream reductions in antibiotic use, healthcare utilization, and linear growth faltering among the majority of children with shigellosis who do not receive effective antimicrobial therapy during their acute illness. In high-burden settings, vaccination is a scalable and equitable intervention that aligns with WHO priorities and child survival goals. However, expectations must be tempered by recognition that vaccine-preventable Shigella infections likely represent a small fraction of progressive linear growth shortfalls seen in LMICs, for which diverse determinants, including consistently inadequate nutrient intake and foetal growth restriction, are unlikely to be altered by disease-specific interventions.³³

While this study is strengthened by the large and diverse cohort comprising seven countries, a longitudinal design enabling tracking of outcomes, high participant retention, standardised data collection, and use of quantitative PCR for pathogen detection, there were several limitations. The study excluded children who did not seek care for the index diarrhoea episode, likely underestimating the true community burden of Shigella and other pathogens. Because they were recruited at the same centre and seeking care for the same illness, the comparison population resembles the case group in health care seeking and sociodemographic context. The comparison group of children with diarrhoea of other or unknown aetiology may have underestimated the consequences of the targeted enteric pathogens because this group was also undergoing a physiologic insult severe enough to stimulate care-seeking for diarrhoea. Those not seeking care may differ systematically from those who do, potentially with worse outcomes. Restricting enrollment to diarrhoea onset within seven days of presentation may have underestimated longer-duration diarrhoea. Finally, while qPCR testing allowed detection of asymptomatic infections, such cases may also contribute to growth faltering and are difficult to interpret in relation to clinical disease. These limitations likely bias our findings towards underestimating the association between Shigella and illness outcomes.

Other limitations may bias our findings in less predictable ways. The use of the quantitative PCR to determine pathogen attribution often yields more than one pathogen in children with acute diarrhoea in LMIC contexts. The use of quantitative methodologies that determine the threshold for disease attributions are an improvement compared to detection-only categorical assignment, but even when using quantitative thresholds, more than one pathogen may be present at or above the level of the quantitative threshold for illness attribution. Pathogen co-occurrence was not uncommon, and the prevalence of multiple-pathogen attribution was highly heterogeneous between sites ranging from 16.8% in Peru to 48.3% in Bangladesh. This problem is not unique to this study; in the GEMS study which also used quantitative cutoffs, 38.9% of diarrhoea cases had two or more attributable pathogens.³ Vaccine probe studies will help elucidate the causal role of Shigella when detected with other enteric pathogens. Finally, we only considered pathogens deemed “attributable” to diarrhoea based on diarrhoea case/asymptomatic control studies and other pathogens, such as giardia and enteroaggregative E. coli, may have been included in the “no detectable pathogen” group.

In this large, multi-country study of medically attended diarrhoea in children aged 6–35 months, Shigella was the most prevalent pathogen and the only one significantly associated with prolonged and persistent diarrhoea, a syndrome that is difficult to treat and has been associated with increased mortality. Although not linked to hospitalization or linear growth faltering overall, children with Shigella who did not receive effective antibiotic treatment experienced modest growth deficits. These findings highlight Shigella’s substantial subacute burden and the importance of timely, effective treatment, including in the absence of dysentery. Strengthening pathogen-specific diagnosis and antibiotic access, alongside Shigella vaccination, could reduce prolonged illness and growth impairment and yield meaningful population-level benefits in low- and middle-income settings.

Contributors

DM, BC, CM, EF, KK, AA, MI, MPO, and SQ contributed equally as first authors. KZ, DN, EH, FQ, PP, MH, RO, KJ, MK contributed equally as senior authors. DM, BC, CM, EF, KK, AA, MI, MPO, SQ, KZ, DN, EH, FQ, PP, MH, RO, KJ, and MK conceptualized the manuscript and contributed to writing and editing. EF conducted the statistical analysis following the statistical analysis plan developed by the EFGH consortium. EF and PP accessed and verified the underlying data. The EFGH protocol was co-developed by members of the EFGH consortium. All authors reviewed and approved the final manuscript. This study was conducted as part of the Enterics for Global Health (EFGH) consortium, a multi-country partnership that co-developed a standardised protocol and surveillance methods, built robust data infrastructure, and prioritized equitable authorship and capacity strengthening through inclusive governance, early-career researcher training, and country-led scientific leadership.

Data sharing statement

The EFGH statistical analysis plan (https://clinicaltrials.gov/study/NCT06047821) and study protocol (https://academic.oup.com/ofid/issue/11/Supplement_1) were made publicly available. The dataset is deidentified and anonymized and can be found here from December 2025: Vivli DOI–PR00011860. Analytic code used for this analysis can be found on GitHub at the following link: https://github.com/efeutzz/Mategula_et_al_2026_Consequences_of_Shigella.

Editor note

The Lancet Group takes a neutral position with respect to territorial claims in published maps and institutional affiliations.

Declaration of interests

PP received support for attending meetings and travel from the Gates Foundation. EF received support for attending meetings and travel from the Gates Foundation. KK received grant support and royalties from the University of Washington/Gates Foundation/Institut Pasteur to her institution, University of Virginia, she received support for travel and attending meetings in the form of hotel reimbursement from University of Virginia. MT received grant support from the University of Washington to her institution, University of Maryland, her spouse is co-author SS. All other authors declare no competing interests.