It Takes a Village

Multiple risk factors for juvenile idiopathic arthritis (JIA) influence the microbiome, and various differences in the oral and fecal microbiome have been described to date in JIA. This review summarizes what is known and discusses potential implications for future research on the microbiome in JIA.

Key points

•

The most common pediatric rheumatologic condition, juvenile idiopathic arthritis (JIA), is associated with several risk factors linked to changes in the microbiome.
•

Compared to controls, the oral microbiome in JIA shows overrepresentation of bacteria associated with gingivitis/periodontitis and underrepresentation of regulatory bacteria.
•

The fecal microbiome in JIA shows overabundance of Bacteroides and depletion of regulatory bacteria, and this dysbiosis may already be present in early childhood.
•

To date, no definitive microbiome-based dietary changes or treatments have been identified for JIA, but future research should explore this alongside standard-of-care therapy.

Abbreviations

abx	antibiotics
ACPA	anticitrullinated peptide antibodies
AGA	appropriate for gestational age
AUC ROC	area under the curve receive operating characteristics
AUC	area under the curve
b/n	between
BMI	body mass index
CCP	cyclic citrullinated peptide antibodies
CI	confidence interval
cJADAS10	10-joint clinical Juvenile Arthritis Disease Activity Score
CS	corticosteroids
DHA	docosahexaenoic acid
DMARD	disease-modifying antirheumatic drug
dx	diagnosis
E	enthesitis-related arthritis
EEN	exclusive enteral nutrition
EPA	eicosapentaenoic acid
ERA	enthesitis-related arthritis
ESR	erythrocyte sedimentation rate
f/u	follow-up
HLA	human leukocyte antigen
HR	hazard ratio
IBD	inflammatory bowel disease
IBD+	inflammatory bowel disease-associated arthritis
IgA	immunoglobulin A
IQR	interquartile ratio
IRR	incidence risk ratio
IU	international units
JADAS27	27-joint Juvenile Arthritis Disease Activity Score
JIA	juvenile idiopathic arthritis
LGA	large for gestational age
LPS	lipopolysaccharide
mg	milligram
mL	milliliter
MTX	methotrexate
neg	negative
ng	nanogram
NSAID	nonsteroidal anti-inflammatory drug
O	oligoarticular
OR	odds ratio
P	polyarticular
PICRUSt	Phylogenetic Investigation of Communities by Reconstruction of Unobserved States
polyJIA	polyarticular JIA
pos	positive
PS	psoriatic
PUFA	polyunsaturated fatty acids
RA	rheumatoid arthritis
RF	rheumatoid factor
RF+	rheumatoid factor-positive
ROC	receiver operating characteristic
rRNA	ribosomal ribonucleic acid
S	systemic
SD	standard deviation
SEIFA	Socio-Economic Indexes for Areas
SES	socioeconomic status
SGA	small for gestational age
SI	sacroiliitis
sIgA	serum immunoglobulin A
sIgG	serum immunoglobulin G
sJIA	systemic JIA
SpA	spondyloarthritis
spp	species
TNFi	tumor necrosis factor inhibitor
U	undifferentiated
WBC	white blood cells

Introduction

The human microbiome consists of trillions of microorganisms (primarily bacteria), with distinct compositions at locations including the oropharynx, skin, and gut. The microbiome is essential for training the developing immune system. Numerous pediatric-onset and adult-onset autoimmune/inflammatory conditions have been linked to “dysbiosis,” meaning differences in the microbiome composition compared to healthy controls. Beyond genetic contribution, various environmental factors control the makeup of the microbiome, including delivery method, infant feeding/diet, exposure to pathogens, and antimicrobial treatment. Between birth and 3 years old is a theoretically critical period for microbiome development and subsequent immune system training. Effects on the early childhood microbiome may lead to differences in immunologic responses predisposing to autoimmunity, including the most common pediatric rheumatologic condition, juvenile idiopathic arthritis (JIA). ^,

JIA is a heterogenous condition with 7 subtypes, among which the common primary feature is persistent inflammatory arthritis with onset before the age of 16 years. Subtypes per the International League of Associations for Rheumatology are oligoarticular JIA (≤4 involved joints), rheumatoid factor (RF)-negative polyarticular JIA (polyJIA; ≥5 joints), RF-positive polyJIA, enthesitis-related arthritis (ERA), psoriatic JIA, systemic JIA (sJIA), and undifferentiated JIA. Some subtypes are analogous to adult rheumatologic diseases, such as RF-positive polyJIA with seropositive rheumatoid arthritis (RA) and ERA with adult spondyloarthritis (SpA), allowing for extrapolation from the more extensive evidence behind adult disease pathophysiology.

In fact, preceding investigation into JIA, research demonstrated the microbiome’s role in the development of adult inflammatory arthritis, particularly RA and SpA. Both clinical/associative and basic/mechanistic studies support the mucosal origins hypothesis of RA, which proposes that immune responses at mucosal sites (ie, intestine), triggered by local microorganisms, may lead to chronic inflammation and systemic autoimmunity. Evidence supporting this hypothesis includes intestinal dysbiosis in individuals at-risk for and with RA, immunoglobulin A (IgA) isotype anticitrullinated peptide antibodies (ACPA), expansion of IgA-plasmablasts during the at-risk period, and mucosal site ACPA production. ^, Microbiome research in RA and SpA has previously been reviewed in detail. Relevant parallels and comparisons to findings in JIA will be described in later discussion, though are not the focus of this review.

As more evidence supports the microbiome’s role in RA/SpA development, interest in its relation to JIA pathogenesis has increased. This review summarizes evidence behind risk/protective factors for JIA related to the microbiome, differences in the oral and fecal microbiome between individuals with JIA and pediatric controls and implications, JIA treatment efforts targeting the microbiome, and future directions for JIA microbiome aimed at prevention, diagnosis, and management.

Microbiome-related risk/protective factors for juvenile idiopathic arthritis

Antibiotics

Antibiotics, particularly in early life, can cause long-lasting microbiome changes. Depletion of beneficial commensal organisms may change immune tolerance and predispose to autoimmunity. To date, 4 studies have assessed the association between antibiotic exposure and future JIA, all showing modest but significant and dose-dependent increased odds ( Table 1 A ). In 2015, Horton and colleagues (United Kingdom) found that individuals with JIA had 2.1 times higher odds of any antibiotic exposure prior to diagnosis compared to controls matched to age at diagnosis, and 3.0 times higher odds of greater than 5 antibiotic courses. Associations were strongest for exposure to antibiotics within 6 to 12 months prior to JIA diagnosis (median 3 years). Adjustment for prior infections showed risk was specific to antibiotics rather than the infection for which they were given. No association with JIA and antiviral/antifungal treatment was noted. Also in 2015, Arvonen and colleagues (Finland) noted a similar dose-dependent relationship between antibiotic exposure among individuals with JIA with odds ratio (OR) of 1.6 (95% confidence interval [CI] 1.3–1.9) when comparing history of any antibiotics to none. This association was similar regardless of age at JIA diagnosis before or after 3 years. Lincosamides (eg, clindamycin) showed the highest risk (OR 6.6 [3.7–11.7]), followed by cephalosporins (OR 1.6 [1.4–1.8]).

Table 1

Observational studies assessing key microbiome-associated protective and risk factors for juvenile idiopathic arthritis

Study	Cases/JIA Cohort ^a^,^b	Controls	Measure	Results (Reference = No Abx)
1A: Antibiotics
Horton et al, 2015 United Kingdom Case–control	N = 152 Subtype :– Age at dx : 3y (2–6)	N = 1520 Matched for age and sex	OR (95% CI) adjusted for autoimmune conditions and prior infection	≥1 course abx before index date (age at dx, matched in controls): OR 2.1 (1.2–3.5) Each added course before index date: OR 1.09 (1.05–1.13) Dose-dependent and duration-dependent : 1–2 courses: OR 1.5 (0.8–2.7); 1–2 wk: OR 1.5 (0.9–2.7) 3–5 courses: OR 2.5 (1.4–4.4); 3–5 wk: OR 2.4 (1.3–4.3) >5 courses: OR 3.0 (1.6–5.6); >5 wk: OR 2.9 (1.6–5.2)
Arvonen et al, 2015 Finland Case–control	N = 1298 Subtype :– Age at dx : 3.8y (0.8–12.9)	N = 5179 Matched for age at dx, sex, and birthplace	OR (95% CI)	≥1 course abx before index date: OR 1.6 (1.3–1.9) ≥1 course abx at age 0–24 mo: OR 1.4 (1.2–1.6) Highest associations with specific abx : Lincosamides, any: OR 6.6 (3.7–11.7) Cephalosporins, any: OR 1.6 (1.4–1.8)
Kindgren et al, 2023 Sweden Prospect. cohort	N = 111 Subtype :– Age at dx : 11.1y ± 5.5	N = 16,489 All children born in southeast Sweden 1997–1999 followed until 2020	OR (95% CI)	≥1 course abx in utero or during first year of life: OR 1.3 (1.1–1.5) ≥3 courses abx in utero or during first year of life: OR 1.6 (1.1–2.4) Effect of abx courses was compounded by : HLA DR3-DQ2: OR 15.3 (3.2–74.5) HLA DR15-DQ602: OR 9.6 (1.9–50.2)
Hestetun et al, 2023 Norway Prospect. cohort	N = 1011 Subtype :– Age at dx :–	N = 535,294 All children born in Norway 2004–2012 followed until 2020	OR (95% CI) adjusted for sex	≥1 course prenatal abx: OR 1.10 (0.96–10.26) ≥1 course abx at age 0–24 mo: OR 1.40 (1.24–1.59) Per course abx at age 0–24 mo: OR 1.08 (1.06–1.09) By ≥ 1 course abx at given age : <1 mo: OR 1.34 (0.95–1.87); 0–6 mo: OR 1.19 (0.92–1.54); 6–12 mo: OR 1.30 (1.08–1.57); 12–24 mo: OR 1.51 (1.33–1.72) By age at first exposure : 0–6 mo: OR 1.19 (0.92–1.54); 6–12mo: OR 1.31 (1.07–1.59); 12–24 mo: OR 1.50 (1.30–1.72)

Study	Cases/JIA Cohort ^a^,^b	Controls	Measure	Results
1B: Breastfeeding
Mason et al, 1995 USA Case–control	N = 54 Subtype : O: 28, P: 24 Age at dx : 5.4y ± 3.8	N = 79 Playmates	OR (95% CI)	Any breastfeeding (reference none): OR 0.4 (0.20–0.81) Oligoarticular: OR 0.31 (0.10–0.93) Polyarticular: OR 0.60 (0.21–1.70)
Rosenberg et al, 1996 Canada Case–control	N = 137 Subtype : O: 88, P: 49 Age at dx :–	N = 331 Unmatched N = 54 Matched for age, sex, place, and season	Chi-square; OR (95% CI)	Unmatched : Any breastfeeding: 68% cases vs 62% controls, P = .20; no difference by subtype Matched : Any breastfeeding (reference none): Oligoarticular: OR 2.17 (0.87–5.44) Polyarticular: OR 1.17 (0.33–4.20)
Kasapcopur et al, 1998 Turkey Case–control	N = 53 Subtype : O: 26, P: 18, S: 9 Age at dx : 6.4y ± 4	N = 54	Descriptive, mean ± SD	Breastfeeding duration (months): 12.6 ± 10.4 (cases) vs 10.8 ± 10.1 (controls)
Radon et al, 2010 Germany Case–control	N = 238 Subtype : O: 238 Age at dx : 5.7y ± 3.7	N = 832 Undergoing minor surgery	Chi-square; OR (95% CI) adjusted for age, sex, and so forth	Breastfeeding ≥6 mo: 49% cases vs 30% controls, P < .001 Breastfeeding ≥6 mo (reference <6 mo): OR 1.64 (1.18–2.26)
Ellis et al, 2012 Australia Case–control	N = 262 Subtype :– Age at dx : median 6.4y	N = 481 Undergoing minor surgery	OR (95% CI) adjusted for age, sex, and so forth	Any breastfeeding (reference none): OR 0.86 (0.39–1.89)
Shenoi et al, 2016 USA Case–control	N = 225 Subtype : O: 84, P: 87 (9 RF+), S: 11, E: 26, PS: 17 Age at dx :–	N = 138 Playmates	OR (95% CI) adjusted for age and income	No breastfeeding (reference any breastfeeding): OR 0.85 (0.65–1.10)
Koker et al, 2022 Turkey Case–control	N = 324 Subtype : O: 129, P: 82 (30 RF+), S: 44, E: 53, PS: 16 Age at dx : 6y (1–15)	N = 253	Chi-square	Any breastfeeding (reference none): 95% cases vs 89% controls, P = .008 Breastfeeding duration (cases vs controls): <6 mo: 20% vs 14%; 6–12 mo: 14% vs 18%; 12–18 mo: 24% vs 30%; 18–24 mo: 27% vs 29%; ≥24 mo: 15% vs 10%; P = .10
Kindgren et al, 2023	N = 111	N = 16,489	Descriptive, mean ± SD; OR (95% CI)	Total and exclusive breastfeeding duration in months : 5.9 ± 2.9 and 3.7 ± 1.7 (cases) vs 7.1 ± 2.3 and 4.5 ± 1.9 (controls) Exclusive <4 mo (reference ≥4 mo): OR 3.2 (1.3–7.7) Total <8 mo (reference ≥8 mo): OR 4.3 (2–9.3) Effect compounded by : Consumption of fish age <1 y: OR 13.4 (4.6–39.1); HLA DR1-DQ5: OR 6.4 (1.2–35.3); HLA DR5-DQ7: OR 8.8 (1.0–73.3)
Baggett et al, 2024 USA Case–control	N = 195 Subtype : E: 117, PS: 43, IBD: 18, SI: 11, U: 7 Age at dx : 13y ± 0.2	N = 195 Matched for age and sex	Chi-square OR (95% CI) adjusted for delivery method	Any breastfeeding: 69% cases vs 89% controls, P < .001 Breastfeeding ≥6 mo: 28% cases vs 47% controls, P < .001 Breastfeeding ≥6 mo (reference <6 mo): OR 0.47 (0.30–0.72) No effect on presentation disease severity

Study	Cases/JIA Cohort ^a^,^b	Controls	Measure	Results (Reference for OR: Vaginal delivery)
1C: Delivery Method
Carlens et al, 2009 Sweden Case-Control	N = 3334 Subtype :– Age at dx : median 3y	N = 13,336 Matched for sex, birth year, and delivery unit	OR (95% CI)	C-section: OR 1.1 (1.0–1.3)
Ellis et al, 2012	N = 262	N = 481	Logistic regression	C-section: 23% cases vs 27% controls, P > .05 Missing: 66% cases, 56% controls
Shenoi et al, 2014 USA Case-Control	N = 1196 Subtype : O: 453, P: 342 (55 RF+), S: 65, E: 239, PS: 61, U: 36 Age at dx : 33% <5y, 23% 5–9y, 31% 10–14y, and 13% ≥15y	N = 5618 Matched for birth year	Descriptive	C-section: 22% cases vs 9% controls
Sevelsted et al, 2015 Denmark Prospect. cohort	N = 6946 Subtype :– Age at dx :–	N = 1.9 million Term children born 1977–2011 followed to age 15 y	IRR (95% CI) adjusted for age, sex, and so forth	C-section: IRR 1.10 (1.02–1.18) Population attributable risk fraction: 1.34
Horton et al, 2015	N = 152	N = 1520	Chi-square	C-section: 21% cases vs 15% controls, P = .07 Missing: 34% cases, 38% controls
Kristensen et al, 2016 Denmark Prospect. cohort	N = 1626 Subtype :– Age at dx :–	N = 790,569 All children born in Denmark 1997–2012	HR (95% CI) adjusted for gestational age, sex, and so forth	Elective C-section: HR 1.25 (1.04–1.51) Acute C-section: HR 0.99 (0.82–1.20)
Shenoi et al, 2016	N = 225	N = 138	OR (95% CI) adjusted for age and income	C-section: OR 1.08 (0.54–2.14) Forceps/vacuum: OR 6.29 (2.77–14.27)
Bell et al, 2017 USA Case–control	N = 1252 Subtype : O: 468, P: 343 (61 RF+), S: 73, E: 265, PS: 63, U: 40 Age at dx :–	N = 6072 Matched for birth year	OR (95% CI) adjusted for parent age and birth year	C-section: OR 1.08 (0.93–1.26)
Sutton et al, 2022 USA Case–control	N = 1290 Subtype : O: 402, P: 275, E: 263 Age at dx :–	N = 6072 Matched for birth year	Descriptive	C-section: 22% cases vs 19% controls
Kindgren et al, 2023	N = 111	N = 16,489	OR (95% CI)	C-section: OR 1.1 (0.6–2.0); Oligoarticular only: OR 2.7 (1.3–6.0). Effects compounded by presence of HLA DR8-DQ4: OR 7.5 (1.7–33.1); Oligoarticular only: OR 27.8 (5.7–135.5)
Spangmose et al, 2023 Denmark Prospect. cohort	N = 1528 Subtype :– Age at dx :–	N = 1,347,625 All children born in Denmark 1994–2014 followed until 2014	HR (95% CI) adjusted for maternal age, sex, and so forth	C-section: HR 1.23 (1.08–1.41)
Baggett et al, 2024	N = 195	N = 195	Chi-square OR (95% CI)	Vaginal (reference C-section): OR 0.84 (0.52–1.35)

Study	Cases/JIA Cohort ^a^,^b	Controls	Measure	Results
1D: Siblings/Birth Order
Nielsen et al, 1999 Denmark Case–control	N = 220 Subtype : O: 128, P: 64, S: 26, U: 2 Age at dx : Median 6y	N = 880 Matched for age, sex, and county	OR (95% CI) adjusted for income and housing	Only child (reference any siblings): OR 1.6 (1.15–2.23) No difference whether younger or older. Significant independent explanatory variable in oligoarticular and polyarticular JIA alone (not systemic)
Prahalad et al, 2003 USA Case–control	N = 333 Subtype : O: 157, P: 78 (22 RF+), S: 33, E: 49, PS: 4, U: 12 Age at dx :–	N = 3295 Matched for sex and birth year	T-test Chi-square	Birth order (mean ± SD) : 2.5 ± 1.7 (cases) vs 2.6 ± 1.7 (controls), P = .33 Sibship size (mean ± SD) : 3.6 ± 2.0 (cases) vs 3.7 ± 1.9 in (controls), P > .05 First-born : 28% cases vs 24% controls, P = .18 Last-born : 39% cases vs 40% controls, P = .63 Only child : 7% cases vs 10% controls, P = .09 No differences by subtype
Jaakkola 2005 Finland Prospect. cohort	N = 31 Subtype :– Age at dx : <7y	N = 58,841 Singletons born in 1987 in Finland, followed 7 y	OR (95% CI) adjusted for sex, birth order, SES, and maternal age	# prior births (reference 0): 1: OR 0.58 (0.24–1.39) 2: OR 0.60 (0.19–1.95) 3: OR 0.65 (0.16–2.65) 4: OR (unadjusted) 0.76 (0.10–5.76)
Carlens et al, 2009	N = 3334	N = 13,336	OR (95% CI)	# older siblings (reference none): 1–2: OR 1.0 (0.9–1.1) ≥3: OR 0.9 (0.8–1.1) Multiple birth (reference singleton): OR 0.9 (0.7–1.2)
Radon et al, 2010	N = 238	N = 832	Descriptive	≥2 older siblings in 16% cases vs 17% controls ≥2 younger siblings in 11% cases vs 14% controls
Shenoi et al, 2014	N = 1196	N = 5618	Descriptive	# prior live births (cases vs controls) : 0: 43% vs 42%; 1: 35% vs 32%; 2: 14% vs 16%; ≥3: 9% vs 10%
Miller et al, 2015 Australia Case–control	N = 302 Subtype : Nonsystemic Age at dx :–	N = 676 Hospital controls (HC), undergoing minor surgery N = 341 Community controls (CC)	OR (95% CI)	Birth order (reference 1): OR (HC) , OR (CC) : 2: OR 0.78 (0.55–1.12), OR 1.00 (0.64–1.55) 3: OR 0.74 (0.44–1.23), OR 2.02 (1.08–3.77) 4: OR 0.62 (0.3–1.27), OR 4.17 (1.51–11.48 ≥5: OR 0.33 (0.1–1.07), OR 2.44 (0.47–12.79) # siblings living at home born within 18 y (reference 0): 1: HC OR 0.46 (0.28–0.76); CC OR 0.39 (0.19–0.84) 2: OR 0.50 (0.29–0.87); OR 0.52 (0.24–1.15) ≥3: OR 0.25 (0.13–0.48); OR 0.51 (0.21–1.28) Sibling exposure time (reference 0): OR (HC) , OR (CC) : <1 y: OR 0.37 (0.03–3.94); OR 0.06 (0.01–0.51) 1–3 y: OR 0.35 (0.16–0.77); OR 0.18 (0.06–0.50) ≥3 y: OR 0.49 (0.3–0.79); OR 0.54 (0.26–1.10)
Shenoi et al, 2016	N = 225	N = 138	OR (95% CI) adjusted for age and income	Birth order (reference 1): 2: OR 0.86 (0.49–1.52) ≥3: OR 0.87 (0.48–1.57)
Bell et al, 2017	N = 1252	N = 6072	OR (95% CI) adjusted for parent age and birth year	Prior births (reference 0): 1: OR 0.95 (0.82–1.10) 2: OR 0.68 (0.56–0.83) 3: OR 0.71 (0.53–0.96) ≥4: OR 0.72 (0.51–1.02)
Sutton et al, 2022	N = 1290	N = 6072	Descriptive	# older siblings (cases vs controls): 0: 43% vs 42%; 1: 35% vs 33%; ≥2: 22% vs 25%
Koker et al, 2022	N = 324	N = 253	Chi-square	Birth order (cases vs controls) : 1: 42% vs 48%; 2: 31% vs 33%; ≥3: 28% vs 19%; P = .07
Spangmose et al, 2023	N = 1528	N = 1,347,625	HR (95% CI) adjusted for maternal age, sex, and so forth	≥1 prior birth (reference 0): HR 1.14 (1.02–1.28)

Abbreviations: Abx, antibiotics; dx, diagnosis; E, enthesitis-related arthritis, HR, hazard ratio; IBD+, inflammatory bowel disease-associated arthritis, SI: sacroiliitis; IRR, incidence risk ratio; mo, months; O, oligoarticular, OR, odds ratio; P, polyarticular, PS, psoriatic, RF+, rheumatoid factor-positive, S, systemic, SES, socioeconomic status; U, undifferentiated; wk, weeks; y, year.

a Juvenile idiopathic arthritis (JIA) subtype abbreviations .

b Age at diagnosis in years presented as median (interquartile range [IQR]) or mean ± standard deviation (SD).

A 2023 Norwegian birth registry prospective cohort study (Hestetun and colleagues ) found a dose-dependent increased odds of JIA in individuals exposed to antibiotics in the first 2 years of life (OR 1.4 [1.2–1.6]), even after adjustment for multiple potential confounders. The association was maintained on sensitivity analysis looking only at individuals diagnosed with JIA after age 3 years, performed to account for the possibility that individuals nearing diagnosis were receiving antibiotics due to underlying immune dysregulation with increased vulnerability to infection. No association was found between prenatal antibiotic exposure and JIA. Sulfonamides/trimethoprim had one of the highest associations with JIA (OR 2.3 [1.6–3.3]). Lincosamides, grouped with macrolides and streptogramins, had OR 1.7 (1.4–2.0). Antibiotics with prominent anaerobic activity (ie, tetracyclines and beta-lactamase-resistant penicillins) were grouped as “other” along with additional types of antibiotics (OR 2.5 [1.5–4.2]), making conclusions regarding subtype effects challenging.

Lastly, a 2023 Swedish study of a large prospective birth cohort found that any antibiotic exposure prenatally or during the first year of life was associated with 30% increased odds of future JIA (OR 1.3 [1.1–1.5]). With 3 or greater antibiotic courses, there was 60% increased odds of JIA (OR 1.6 [1.1–2.4]). As courses of antibiotics accumulated during the first 5 years of life, odds of JIA increased (up to OR 2.2 [1.4–3.5]). The effect of antibiotic courses on risk of JIA was compounded by the presence of human leukocyte antigen (HLA) alleles DR3-DQ2 (OR 15.3) and/or DR15-DQ602 (OR 9.6).

None of these studies differentiate effect of antibiotics on risk of JIA by subtype. Age differences in cohorts make comparison difficult. Although attempted, it is impossible to completely control for confounding factors and to assess contribution of infection itself versus antibiotics, though results are suggestive. An association with hospitalization for infection during the first year of life has been weakly associated with JIA in 2 studies, but this may be due to increased antibiotic exposure or microbiome changes related to the hospital environment rather than the infection. ^, Inconsistent findings regarding which antibiotics are highest risk make it challenging to determine the mechanism by which antibiotics may be contributing to JIA development. The higher association with antianaerobic coverage (eg, lincosamides and potentially beta-lactamase inhibitor/penicillin combinations ) is interesting given that it has been shown to cause longer lasting effects on the intestinal microbiome.

Breastfeeding

Compared to formula-fed babies, breastfed babies have a unique intestinal microbiome, including enrichment of Bifidobacteria and Lactobacillus . ^, This may have a beneficial downstream effect on immune development: breastfeeding is associated with lower risk of multiple autoimmune conditions including type I diabetes mellitus, celiac disease, and inflammatory bowel disease (IBD). ^, Breastfeeding has been proposed as a potential protective factor for JIA, with some conflicting results (see Table 1 B). A small 1995 US case–control study demonstrated that children with oligoJIA and polyJIA had 60% lower odds of breastfeeding compared to controls. Similar results were found in a 2024 US study showing 53% lower odds of breastfeeding greater than 6 months among children with juvenile-SpA. However, other studies (from Canada, Turkey, Australia, and United States) showed no significant difference in breastfeeding in JIA versus controls. ^, ^, ^, A German study noted the opposite association, with children with JIA demonstrating 60% higher odds of having breastfed versus controls. Significant limitations exist with the case–control study design, including retrospective data collection leading to recall bias. The 2023 prospective birth cohort study from Sweden (Kindgren and colleagues, described earlier), addressed this and found that breastfeeding for less than 4 months and less than 8 months were associated with higher odds of developing JIA (OR 3.2 [1.3–7.7] and 4.3 [2.0–9.3], respectively).

Birth History: Delivery Method and beyond

Many birth factors may play a role in neonatal microbiome establishment, including delivery method, prematurity, parental age at birth, and birth weight, and the impact may persist for years. Infants born by C-section show decreased colonization with vaginal flora, such as Bifidobacteria and Bacteroides and are more likely to be colonized with potential pathogens such as Klebsiella , Enterococcus , and Enterobacter . Multiple studies have shown a modest but significant association between delivery by C-section and JIA, ^, ^, ^, ^, including 3 prospective cohort studies from Denmark and Sweden (see Table 1 C, D). Two US case–control studies did not demonstrate this association. ^, No consistent trends in association with gestational age, parental age at birth, or birth weight are seen ^, ^, ^, ( Table 2 ).

Table 2

Demographic, birth-related, and miscellaneous risk/protective factors in juvenile idiopathic arthritis (JIA)

Study ^a	Cases/JIA Cohort ^b^,^c	Controls	Measure ^d	Results (Reference: Male unless Otherwise Specified)
Sex (Section 2.3)
Jaakkola, 2005 Finland Prospect. cohort	N = 31 Subtype ^a: unstated Age at dx ^b: <7y	N = 58,841 Singletons born in 1987 in Finland, followed 7 y	OR (95%) adjusted for sex, birth order, SES, and maternal age	Female: OR 3.03 (1.36–6.76)
Radon, 2010 Germany Case–control	N = 238 Subtype : O: 238 Age at dx : 5.7y ± 3.7	N = 832 Minor surgery	OR (95% CI) adjusted for age, sex, and so forth	Male (reference female): OR 0.47 (0.34–0.66)
Ellis, 2012 Australia Case–control	N = 262 Subtype : unstated Age at dx : median 6.4y	N = 481 Minor surgery	Logistic regression	Female: 67.2% cases vs 39.7% controls, P < .05
Shenoi, 2014 USA Case–control	N = 1196 Subtype : O: 453, P: 342 (55 RF+), S: 65, E: 239, PS: 61, U: 36 Age at dx : 33% <5y, 23% 5–9y, 31% 10–14y, 13% ≥15y	N = 5618 Matched for birth year	Descriptive, %	Female: 67.7% cases vs 47.8% controls
Miller, 2015 Australia Case–control	N = 302 Subtype : Nonsystemic Age at dx : unstated	N = 676 Hospital N = 341 Community	Descriptive, %	Female: 67% cases vs 41% hospital controls (undergoing minor surgery) and 54% community controls
Shenoi, 2016 USA Case–control	N = 225 Subtype : O: 84, P: 87 (9 RF+), S: 11, E: 26, PS: 17 Age at dx : unstated	N = 138 Playmates	OR (95% CI) adjusted for age and income	Female: OR 1.80 (1.25–2.59)
Thorsen, 2016 Denmark Case–control	N = 300 Subtype : O: 202, P: 98 (14 RF+) Age at dx : O: 5y (3–9), P: 8.5y (3.5–12)	N = 300 Matched for birth date	OR (95% CI) adjusted for ethnicity, birth weight, maternal age, and so forth	Female: OR 2.4 (1.7–3.5)
Bell, 2017 USA Case–control	N = 1252 Subtype : O: 468, P: 343 (61 RF+), S: 73, E: 265, PS: 63, U: 40 Age at dx : unstated	N = 6072 Matched for birth year	Descriptive, %	Female: 67% cases vs 48% controls
Sutton, 2022 USA Case–control	N = 1290 Subtype : O: 402, P: 275, E: 263 Age at dx : unstated	N = 6072 Matched for birth year	Descriptive, %	Female: 68% cases vs 48% controls
Koker, 2022 Turkey Case–control	N = 324 Subtype : O: 129, P: 82 (30 RF+), S: 44, E: 53, PS: 16 Age at dx : 6y (1–15)	N = 253	Chi-square	Female 65% cases vs 61% controls, P = .3
Kindgren, 2023 Sweden Prospect. cohort	N = 111 Subtype : unstated Age at dx : 11.1y ± 5.5	N = 16,489 All children born in SE Sweden 1997–1999 followed until 2020	Chi-square	Female: 48% controls vs 66% cases, P < .001
Spangmose, 2023 Denmark Prospect. cohort	N = 1528 Subtype : unstated Age at dx : unstated	N = 1,347,625 All children born in Denmark 1994–2014 followed until 2014	HR (95% CI) adjusted for maternal age, sex, and so forth	Female: HR 1.88 (1.69–2.09)

Study	Cases/JIA cohort ^a^,^b	Controls	Measure	Results
Race/Ethnicity (Section 2.3)
Ellis, 2012 Australia	N = 262	N = 481	Logistic regression	Child has 4 Caucasian grandparents : 87% cases vs 79% controls, P < .05; Born in Victoria , Australia : 90% of cases vs 79% of controls, P < .05
Shenoi, 2014 USA Case–control	N = 1196	N = 5618	Descriptive, %	Race/ethnicity (cases vs controls): White: 85% vs 78%; African American: 1.5% vs 4%; Hispanic: 7% vs 9%; Asian: 3% vs 5%; American Indian: 2.5% vs 2%; Hawaiian/Pacific Islander: 0.4% vs 0.3%; Other: 1.5% vs 1.3%
Shenoi, 2016 USA	N = 225	N = 138	OR (95% CI) adjusted for age and income	Non-Caucasian (reference Caucasian): OR 1.19 (0.51–2.75) Multiracial (reference Caucasian): OR 0.42 (0.30–0.58) Hispanic/Latino (reference no): OR 1.39 (0.88–2.18)
Thorsen, 2016 Denmark	N = 300	N = 300	OR (95% CI) adjusted for sex, birth weight, and so forth	Ethnic Dane (reference: other): OR 3.2 (1.6–6.6)
Sutton, 2022 USA	N = 1290	N = 6072	Descriptive, %	Race/ethnicity (cases vs controls): White: 84% vs 78%; Black: 2% vs 4%; Native American: 3% vs 2%; Asian/Pacific Islander: 5% vs 7%; Hispanic: 6% vs 9%; Missing: 4% vs 2%
Beesley, 2023 England Cross-sectional	N = 795 Subtype : unstated Age at dx : unstated	N/A	Incidence ratio by ethnicity; and prevalence (rate per 100,000)	Ratio of the proportion of incident JIA cases in December 2018 by ethnic group compared to the general population : White, 1.1:1; Mixed, 0.5:1; Asian, 0.5:1; and Black, 0.6:1 Prevalence of JIA among people <16 in December 2018 by ethnic group : White, 67.6 (CI 62.5–73.0); Mixed, 29.0 (19.3–42.0); Asian, 38.5 (29.7–48.9); and Black, 42.0 (29.1–58.7)
Baggett, 2024 USA Case–control	N = 195 Subtype : E: 117, PS: 43, IBD: 18, SI: 11, U: 7 Age at dx : 13y ± 0.2	N = 195 Matched for age and sex	Chi-square OR (95% CI) adjusted for delivery method	White (reference: other): OR 1.20 (0.68–2.11) No effect on disease severity at presentation

Study	Cases/JIA cohort ^a^,^b	Controls	Measure	Results (Reference: 37–42 wk)
Gestational Age (Section 2.4)
Carlens, 2009 Sweden Case–control	N = 3334 Subtype : Unstated Age at dx : median 3y	N = 13,336 Matched for sex, birth year, and delivery unit	OR (95% CI)	<37 wk: OR 0.8 (0.7–1.0) >42 wk: OR 1.2 (1.03–1.3)
Ellis, 2012 Australia	N = 262	N = 481	Logistic regression	Gestational age in weeks (mean ± SD): 39.3 ± 1.8 in cases vs 39.0 ± 2.3 in controls, P > .05
Shenoi, 2014 USA	N = 1196	N = 5618	Descriptive, %	<37 wk: 7% of cases vs 6% of controls 37–42 wk: 85% of cases vs 84% of controls ≥42 wk: 8% of cases vs 10% of controls
Horton, 2015 United Kingdom Case–control	N = 152 Subtype : Unstated Age at dx : 3y (2–6)	N = 1520 Matched for age and sex	Chi-square	<37 wk: 3% cases vs 1.7% controls, P = .18
Miller, 2015 Australia	N = 302	N = 676 Hospital N = 341 Community	Descriptive, %	Gestational age (cases vs hospital and community controls): <37 wk: 3% vs 8% and 7%; 37–42 wk: 86% vs 72% and 84% ≥42 wk: 11% vs 5% and 8%; Missing: 0.3% vs 15% and 0.6%
Shenoi, 2016 USA	N = 225	N = 138 Playmates	OR (95% CI) adjusted for age and income	<37 wk: OR 1.8 (1.2–2.7) ≥42 wk: OR 0.73 (0.37–1.47)
Thorsen, 2016 Denmark	N = 300	N = 300 Matched for birth date	OR (95% CI) adjusted for sex, ethnicity, and so forth	<37 wk: OR 0.9 (0.4–1.8) ≥42 wk: OR 1.1 (0.6–2.0)
Bell, 2017 USA	N = 1252	N = 6072	OR (95% CI) adjusted for birth year and parent age	<37 wk: OR 0.99 (0.76–1.29) ≥42 wk: OR 0.88 (0.70–1.10)
Franca, 2018 Brazil Case–control	N = 66 Subtype : O: 20, P: 27 (4 RF+), S: 17, PS: 1, U: 1 Age at dx : 6.6y ± 3.8	N = 124 Matched for age and sex	OR (95% CI) adjusted for maternal job, smoking, and so forth	<37 wk: OR 2.0 (0.6–6.9)
Sutton, 2022 USA	N = 1290	N = 6072	Descriptive, %	Gestational age (cases vs controls): <37 wk: 7% vs 7%; 37–42 wk: 85% vs 83%; ≥42 wk: 8% vs 10%; Missing: 3% vs 3%
Kindgren, 2023	N = 111	N = 16,489	Mann–Whitney	Gestational age in weeks (mean ± SD): 39.5 ± 2.0 in cases vs 39.7 ± 2.4 in controls, P = .25

Study	Cases/JIA cohort ^a^,^b	Controls	Measure	Results
Parental Age at Birth (Section 2.4)
Prahalad, 2003 USA Case–control	N = 333 Subtype : O: 157, P: 78 (22 RF+), E: 49, S: 33, PS: 4, U: 12 Age at dx : unstated	N = 3295 Matched for sex and birth year	t-test	Maternal age (mean in year ± SD): 26.9 ± 5.8 in cases vs 26.4 ± 5.5 in controls; P > .05
Jaakkola, 2005 Finland	N = 31	N = 58,841	OR (95%) adjusted for sex, SES, and so forth	Maternal age (reference <19 y): 20–24 y: OR 1.36 (0.17–11.2); 25–29 y: 1.41 (0.18–11.4); 30–34 y: 1.70 (0.20–12.2); 35–39 y: 1.14 (0.11–12.2)
Carlens, 2009	N = 3334	N = 13,336	OR (95% CI)	Maternal age: (reference 25–29 y): <25 y: OR 1.0 (0.9–1.1) 30–34 y: OR 1.0 (0.9–1.1) ≥35 y: OR 1.0 (0.9–1.2
Ellis, 2012 Australia	N = 262	N = 481	OR (95% CI) adjusted for age, sex, and so forth	Maternal age (based on 1 unit ↑): OR 1.08 (1.02–1.15) Paternal age (based on 1 unit ↑): OR 1.06 (1.01–1.11)
Shenoi, 2014 USA	N = 1196	N = 5618	Descriptive, %	Maternal age (cases vs controls): <20 y: 8% vs 11%; 20–34 y: 76.8% vs 76.4%; ≥35 y: 15% vs 12.2%
Miller, 2015 Australia	N = 302	N = 676 Hospital N = 341 Community	Descriptive, %	Maternal age (cases vs hospital controls and community controls): <25 y: 8% vs 19% and 9%; 25–29 y: 31% vs 31% and 26%; 30–34 y: 35% vs 31% and 36%; ≥35 y: 22% vs 16% and 23%; Missing data: 4%, 3%, 18.5%
Shenoi, 2016 USA	N = 225	N = 138 Playmates	OR (95% CI) adjusted for age and income	Maternal age ≥35 y (reference: 20–34 y): OR 0.59 (0.34–1.01)
Thorsen, 2016 Denmark	N = 300	N = 300 Matched for birth date	OR (95% CI) adjusted for sex, ethnicity, and so forth	Maternal age (reference <25 y): 25–34 y: OR 1.2 (0.7–2.0) ≥35 y: OR 1.1 (0.6–2.3)
Bell, 2017 USA	N = 1252	N = 6072	OR (95% CI) adjusted for birth year and parent age	Maternal age (cases vs controls): <20 y: 9% vs 11%; 20–24 y: 19% vs 25%; 25–29 y: 31% vs 29%; 30–34 y: 27% vs 22%; ≥35y: 15% vs 12%
Sutton, 2022 USA	N = 1290	N = 6072	Descriptive, %	Maternal age (cases vs controls): <20 y: 8% vs 11%; 20–34 y: 77% vs 76%; ≥35 y: 15% vs 12%
Koker, 2022 Turkey	N = 324	N = 253	Chi-square	Maternal age (cases vs controls): <20 y: 10% vs 17%; 20–34 y: 76% vs 64%; ≥35 y: 13% vs 19%, P = .002 Paternal age (cases vs controls): <20 y: 0% vs 3%; 20–34 y: 74% vs 65%; ≥35 y: 26% vs 32%, P = .001
Kindgren, 2023 Sweden Prospect. cohort	N = 111 Subtype : unstated Age at dx : 11.1y ± 5.5	N = 16,489 All children born in SE Sweden 1997–1999 followed until 2020	Mann–Whitney	Maternal age in years (mean ± SD): 28 ± 5 in cases vs 30 ± 5 in controls, P = .008 Paternal age in years (mean ± SD): 31 ± 5 in cases vs 32 ± 5 in controls, P = .07
Spangmose, 2023 Denmark	N = 1528	N = 1,347,625	HR (95% CI) adjusted for sex and so forth	Maternal age (reference <25 y): 25–29 y: HR 1.08 (0.89–1.29); 30–34 y: HR 0.97 (0.80–1.17); ≥35 y: HR 0.96 (0.78–1.19)