Elevated incidence of infant botulism in a 17-county area of the Mid-Atlantic region in the United States, 2000–2019, including association with soil types

ABSTRACT

We sought to identify counties in the northeastern United States where the incidence of infant botulism (IB) is elevated compared to the nationwide incidence and to assess associations with soil type at the case residence. IB cases were identified through the distribution of the orphan drug Human Botulism Immune Globulin Intravenous for treatment of IB by state and national surveillance systems and were subsequently confirmed by laboratory testing. IB incidence by county was calculated as the number of IB cases divided by the number of live births in the county from 2000 to 2019. Cases were spatially mapped and assigned to soil types using the US Department of Agriculture’s online soils database. Possible association with soil type was evaluated with the Chi-squared test. We identified a rectangular area consisting of 17 contiguous counties in Delaware, Maryland, New Jersey, New York, and Pennsylvania, approximately 80 km by 250 km, in which the 20-year incidence of IB was nearly seven times greater than that of the remaining counties in those five states. Within this area, case residences were strongly associated with certain soil types (P ≤ 0.003). From 2000 to 2019, IB occurred with disproportionate incidence in a rectangular area encompassing the lower Delaware and Raritan River Valley and parts of five adjacent states. Further investigation of the soils in counties from this area could assess whether C. botulinum is more prevalent in certain soil types and whether isolation of C. botulinum is more common in counties with higher IB incidence.

IMPORTANCE

Infant botulism occurs more frequently in 17 counties within and adjacent to the Delaware and Raritan River watersheds. This study should alert physicians and pediatricians in the area to the higher likelihood of encountering cases of this otherwise rare disease that manifests with constipation, poor feeding, loss of head control, weak suck/cry, generalized weakness, and descending bilateral paralysis.

INTRODUCTION

More than 45 years ago, the pioneering studies of infant botulism (IB) in Pennsylvania (PA) by Long and colleagues identified the suburban counties that border metropolitan Philadelphia to the north, south, and west as the home of 43 of the 53 then-known cases in the state (1). The basis of this unusual epidemiological finding was unknown (2). From 2000 to 2019, IB represented 71% of the 3,241 human botulism cases detected in the United States [data about non-IB cases obtained from the Centers for Disease Control and Prevention (CDC) National Notifiable Diseases Surveillance System at https://www.cdc.gov/nndss/data-statistics/infectious-tables/index.html, accessed 23 August 2024].

IB results when swallowed spores of Clostridium botulinum (or related toxigenic clostridia Clostridium baratii and Clostridium butyricum) (3–13) germinate, colonize, and produce botulinum toxin in the large intestine. After absorption from the intestinal tract, the toxin is carried by the bloodstream to the neuromuscular junction, where it enters motor neurons, cleaves soluble N-ethylmaleimide-sensitive factor-attachment protein receptors, and prevents the release of acetylcholine, thus producing flaccid paralysis (14). Clostridial botulinum toxins exist in eight toxin types designated A–H (15–21) that serve as useful epidemiologic and clinical markers. In the Eastern United States, the majority of IB cases are caused by botulinum toxin type B, whereas in the Western United States, the majority are caused by botulinum toxin type A (14, 22). This epidemiologic distribution of the majority of IB cases corresponds to the geographical distribution of C. botulinum toxin types (A or B) in the soil, its natural habitat (22–26). Most IB cases are thought to be acquired when infants swallow airborne carrying C. botulinum spores.

Diagnosis of IB is made based on clinical symptoms prior to laboratory confirmation of the illness. Rapid administration of the antitoxin results in a speedier recovery (27). Treatment for IB consists of meticulous supportive care and Human Botulism Immune Globulin Intravenous (BIG-IV; BabyBIG) (28). BIG-IV is the not-for-profit public service orphan drug developed in 1988–2003 and distributed nationwide by the California Department of Public Health (CDPH) in accordance with the federal Orphan Drug Act and state law (29). Publication of BIG-IV’s randomized clinical trial, education of physicians, and the national availability of BIG-IV that began in 1998 onward resulted in improved recognition of IB cases nationwide (30, 31).

Due to the unique position of the CDPH Infant Botulism Treatment and Prevention Program (IBTPP) to track BIG-IV utilization at the national level, we were able to identify hospitals in the Greater Philadelphia, PA-Wilmington, Delaware (DE) area to be frequent users of BIG-IV. This frequent use prompted formal epidemiologic analysis. Our objective was to identify any counties in the 14 northeastern US states and the District of Columbia (DC) with IB incidence higher than the US incidence and assess possible association with soil taxonomy in contiguous counties with heightened incidence.

RESULTS

Case occurrence, incidence, and distribution

During the 20-year study period 2000–2019, the incidence of IB nationwide was 2.9 (95% CI 2.7–3.0) cases per 100,000 live births (Table 1). US case counts increased by 37% from 970 cases in 2000–2009 to 1,325 in 2010–2019. In the 14 Northeast and Mid-Atlantic states and DC, the incidence of IB (4.1 cases per 100,000 live births) was significantly higher than the US incidence of 2.9 (Table 1). A total of 830 IB cases (n = 44 toxin type A, n = 778 type B, n = 1 type Ba, n = 3 Bf, and n = 4 type F) occurred during the study period. All cases but one were hospitalized. Cases lived in 184 county or city health jurisdictions of the 523 within these states (Fig. 1). Eighty-eight health jurisdictions had only one case in the 20-year time period and 26 had just two cases. The calculated incidence of IB in 144 health jurisdictions was higher than the 20-year national rate, but the 95% CIs for 121 of them were non-significant. This was in part due to a small birth population size of <20,000 live births from 2000 to 2019 (i.e., on average ≤1,000 live births per year) in 64 of the jurisdictions that resulted in the calculation of extremely wide CIs.

TABLE 1.

US infant botulism incidence (cases per 100,000 live births) in 14 northeastern US states and DC, 2000–2019^a,f

Region/state	County	No. live births	No. IB cases	Incidence (95% CI)b	In 17-county rectangular areac
Total United States		80,453,783	2,295	2.9 (2.7–3.0)	−
Total 14 Northeast States & DC		20,247,842	830	4.1 (3.8–4.4)	−
17-county rectangular area		2,377,494	415	17.5 (15.8–19.3)d	−
DE, MD, NJ, NY, and PA excluding 17-county rectangular area		9,217,498	243	2.6 (2.3–3.0)	−
Delaware		224,841	47	20.9 (15.2–28.0)d	−
	New Castle	136,894	45	32.9 (24.0–44.0)d	Y
Maryland		1,477,715	83	5.6 (4.5–7.0)d	−
	Anne Arundel	137,027	13	9.5 (5.0–16.2)d	N
	Cecil	23,551	9	38.2 (16.5–75.3)d	Y
	Harford	56,018	7	12.5 (5.4–24.6)d	Y
	Washington	34,611	5	14.4 (4.7–33.7)d	N
New Jersey		2,176,726	147	6.8 (5.7–8.0)d	−
	Essex	223,628	18	8.0 (4.8–12.7)d	Y
	Hudson	193,350	13	6.7 (3.6–11.5)d	Y
	Hunterdon	22,484	10	44.5 (21.3–81.8)d	Y
	Mercer	87,543	17	19.4 (11.3–31.1)d	Y
	Middlesex	201,065	12	6.0 (3.1–10.4)d	Y
	Morris	106,127	10	9.4 (4.5–17.3)d	Y
	Somerset	75,485	16	21.2 (11.3–36.2)d	Y
	Union	143,631	13	9.1 (4.8–15.5)d	Y
New York		4,864,059	82	1.7 (1.3–2.1)e	−
	Richmond	111,634	17	15.2 (8.9–24.4)d	Y
	Ulster	33,754	4	11.9 (3.2–30.3)d	N
Pennsylvania		2,851,651	299	10.5 (9.4–11.8)d	−
	Adams	20,587	4	19.4 (5.3–49.7)d	N
	Allegheny	265,176	16	6.0 (3.2–10.3)d	N
	Bucks	120,268	57	47.4 (35.2–62.6)d	Y
	Chester	114,229	21	18.4 (11.4–28.1)d	Y
	Delaware	134,622	34	25.3 (17.6–35.1)d	Y
	Montgomery	182,826	88	48.1 (38.9–59.7)d	Y
	Philadelphia	444,049	28	6.3 (4.2–9.1)d	Y
Ohio		2,872,877	69	2.4 (1.9–3.1)	−
	Richland	29,726	5	16.8 (5.4–39.2)d	N
Connecticut		775,614	9	1.2 (0.5–2.3)e	−
District of Columbia		171,497	7	4.1 (1.6–8.4)	−
Massachusetts		1,500,088	9	0.6 (0.3–1.1)e	−
Maine		263,523	2	0.8 (0.1–2.7)	−
New Hampshire		263,572	1	0.4 (0.0–2.1)e	−
Rhode Island		229,207	3	1.3 (0.3–3.8)	−
Vermont		122,290	1	0.8 (0.0–4.6)	−
Virginia		2,045,716	55	2.7 (2.0–3.5)	−
West Virginia		407,466	16	3.9 (2.1–6.7)	−

^a A complete listing of cases, birth populations, and cumulative incidence by county in all 14 northeastern states and DC is available online as supplementary data in an Excel spreadsheet.

^b 95% CIs calculated according to Szklo and Nieto (32).

^c Refers to counties that are in the 17-county rectangular area of elevated IB incidence (Y, yes; N, no; −, not applicable).

^d Incidence significantly higher than the 95% CI of nationwide IB incidence in the same time period.

^e Incidence significantly lower than the 95% CI of nationwide IB incidence in the same time period.

^f Counties with elevated incidence appear under their respective states.

Fig 1.

(A) Location and toxin type of infant botulism cases in 14 northeastern and Mid-Atlantic US states and the District of Columbia, 2000–2019. Each case is represented by a single dot that covers an area of approximately 215 km² (approximately 85 mi²) that, together with the 20-year study period, ensures case anonymity. Note the density of the dots in the 17 counties adjacent to the Delaware and Raritan River watersheds, and the predominance of botulinum toxin type B cases in this region of the United States. Data were mapped using ArcGIS Pro version 3.2.0 and map layers available from ESRI (Redlands, CA). (B) Incidence of infant botulism by county or health jurisdiction in 14 northeastern and Mid-Atlantic US states and the District of Columbia, 2000–2019. The cross-shading identifies those counties whose incidence was significantly higher than the 95% CI for nationwide IB incidence from 2000 to 2019. Note the checker-shaded zone of 17 contiguous counties of elevated incidence that extends from Harford County, MD to Morris County, NJ. 121 jurisdictions in PA, OH, VA, and WV have apparent but not statistically significant high rates of infant botulism due to small birth populations <20,000 in the time period; see supplementary online data for a complete list. Map created using ESRI basemap layers ArcGIS Pro Version 3.2.0 (Redlands, CA).

Half of the cases (415 of 830) in the northeastern United States lived in a rectangular area encompassing parts of the states of DE, Maryland (MD), New Jersey (NJ), New York (NY), and PA. This area consisted of the 17 contiguous counties of New Castle (DE); Cecil and Harford (MD); Essex, Hudson, Hunterdon, Mercer, Middlesex, Morris, Somerset, and Union (NJ); Richmond (NY); and Bucks, Chester, Delaware, Montgomery, and Philadelphia (PA); and measured approximately 80 km (50 mi) × 250 km (155 mi) in size (Fig. 2). It extended from Harford County in northeastern MD to Morris County in NJ, and IB incidence in these counties varied from 6.0 to 48.1 cases per 100,000 live births (Table 1; Fig. 1B) during the 20-year study period. The incidence in this area was 17.5 cases per 100,000 live births (Table 1; Fig. 1), whereas the incidence in the remaining 161 health jurisdictions of those five states was 2.6 cases per 100,000 live births. Counties with the highest 20-year incidence included: Cecil County, MD (38.2); Hunterdon County, NJ (44.5); and Bucks (45.7) and Montgomery (48.1) counties in PA.

Fig 2.

The 17-county rectangular area of elevated infant botulism incidence in the Delaware and Raritan River watersheds. Many IB cases occurred in the yellow-shaded region where silty loess and fluvial outwash were deposited after the Pleistocene Laurentide ice sheet meltback. Mapped per methods described in Drohan et al. (33).

Six additional jurisdictions in the 14 northeastern US states and DC outside the 17-county rectangular area (Table 1) demonstrated significantly higher IB incidence than US incidence: Anne Arundel (9.5) and Washington (14.4) counties, MD; Ulster County, NY (11.8); Adams (19.4) and Allegheny (6.0) counties, PA; and Richland County, OH (16.8).

Descriptive epidemiology of IB cases in the 17-county rectangular area

Cases caused by toxin type B predominated (n = 410 cases, 99%), and just five cases (1%) were caused by toxin type A; 53% were female; and 71% were non-Hispanic white, with the remainder being 11% Hispanic, 6% non-Hispanic Asian, 5% non-Hispanic black, and 3% of mixed or other racial background (n = 16 missing race-ethnicity data). In the United States from 2000 to 2019, non-Hispanic black infants represented a lower proportion of cases (2.8%) and Hispanic infants a higher proportion (24.9%) than IB cases from the 17-county rectangular area. The mean age at onset for the 410 type B cases was 16.3 weeks (SD 8.6), which did not differ significantly from that of type B cases nationwide in the same time period (t-test statistic −0.64, P = 0.52). IB cases in the 17-county rectangular area from 2000 to 2019 did not differ in milk feeding regimen from US IB cases (56% exclusively breastfed vs 54% nationwide, 24% breast- and formula-fed vs 26% nationwide, 13% exclusively formula-fed vs 12% nationwide; χ² statistic 1.57, P = 0.67), although a lower proportion was fed honey (2.2% vs 4.3% nationwide, χ² statistic 17.52, P < 0.001), powdered infant formula (12% vs 20% nationwide, χ² statistic 15.31, P < 0.001), and solid foods (19% vs 23% nationwide, χ² statistic 15.06, P < 0.001).

Association of soil taxonomy with cases

In the 17-county rectangular area at three levels of soil taxonomy (Fig. S3; supplementary material), certain soil taxa were associated with IB cases (Table 2). These levels included soil order (χ² statistic 13.94, P = 0.003), soil suborder (χ² statistic 39.87, P < 0.001), and soil great group (χ² statistic 169.99, P < 0.001) (Table 2). Among soil orders, the highest percentage of cases (40%) resided on Alfisols (clay soils with base saturation ≥35%), which covered 33% of the land area, followed by 33% of cases on Ultisols (clay soils with base saturation <35%). Notably, the latter represented 41% of the soil area so there were fewer case residences on Ultisols than expected. In the 17-county rectangular area, soil suborder Udults (Udic moisture regime Ultisol) and Udalfs (Udic moisture regime Alfisol) predominate (58.1% of land area). At the level of soil suborder, Aquults (saturated soils with a high water table in the order Alfisol, 8.0% of land area) were under-represented, accounting for a lower proportion of cases (1.4%) than expected; and Orthents (soils that are not wet, shallow, or excessively sandy in the order Entisol) were over-represented, with 13% of cases and just 8% of the land area (Table 2).

TABLE 2.

Soil area and classification at residences of 361 infant botulism cases in the 17-county rectangular area^a,b

Soil order	Soil type at case residence in 17-county area N (% of cases)	Landscape area in 17-counties of elevated incidence km2 (% of area)
Alfisol	146 (40.4)	4,075 (33.2)
Entisol	48 (13.3)	1,290 (10.5)
Inceptisol	47 (13.0)	1,770 (14.4)
Ultisol	120 (33.2)	5,026 (40.9)
Otherc	0 (0.0)	0.3 (<0.1)
Total	361 (100.0)	12,161 (100.0)
	χ 2=13.9c, df=3c; P = 0.003
Soil Suborder
Alfisols
Aqualf	39 (10.8)	1,054 (8.7)
Udalf	107 (29.6)	3,021 (24.8)
Entisols
Aquent	2 (0.6)	242 (2.0)
Orthent	46 (12.7)	977 (8.0)
Inceptisols
Aquept	10 (2.8)	450 (3.7)
Udept	237 (10.2)	1,005 (8.3)
Ultisols
Aquult	5 (1.4)	975 (8.0)
Udult	115 (31.9)	4,051 (33.3)
Otherc	0 (0.0)	386 (3.2)
Total	351 (100.0)	12,161 (100.0)
	χ2 = 39.9c, df = 7c; P < 0.001
Soil Great Group
Alfisols
Fragiaqualf	39 (10.8)	902 (7.4)
Fragiudalf	56 (15.5)	1,017 (8.4)
Hapludalf	51 (14.1)	2,004 (16.5)
Entisols
Fluvaquent	1 (0.3)	141 (1.2)
Hydraquent	1 (0.3)	30 (0.2)
Udorthent	46 (12.7)	977 (8.0)
Inceptisols
Dystrudept	27 (7.5)	896 (7.4)
Endoquept	10 (2.8)	285 (2.3)
Eutrudept	4 (1.1)	101 (0.8)
Fragiudept	6 (1.7)	8 (0.1)
Ultisols
Fragiaquult	5 (1.4)	595 (4.9)
Fragiudult	17 (4.7)	695 (5.7)
Hapludult	98 (27.1)	3,354 (27.6)
Otherc	0 (0.0)	1,156 (9.5)
Total	351 (100.0)	12,161 (100.0)
	χ 2 = 170.0c, df = 12c; P < 0.001

^a Excludes 54 cases for which soil data were not available or unclassifiable (urban fill).

^b Soil area excludes water, quarries, pits, dams, and unclassifiable filled urban land.

^c “Other” category was excluded from the Chi-squared test because the inclusion of this category would unduly weight the total Chi-squared statistic toward significance. Hence, the exclusion of the “other” category provided a more conservative statistical evaluation of the possible association between the case soil residence type and the distribution of soil types in the case rectangle. See Table S4 supplementary material for calculations of the Chi-squared statistic with and without inclusion of the “other” category.

At the level of the soil great group, IB cases were associated with Fragiudalfs (in soil order Alfisol, soils with a subsurface layer >15 cm that restricts water flow and root penetration), comprising 16% of cases and 8% of land area; and Udorthents [in order Entisol, soils with minimal horizon (layer) development] comprising 13% of cases and 8% of land area (Table 2).

In addition (Table 3), IB cases were significantly associated with soil particle size family (χ² statistic 21.0 P < 0.001). A higher percentage of cases in the 17-county rectangular area resided on soils with coarse particle size (18.5%) than the land area of such soils (12.4%), and a lower proportion (5.2%) on soil containing other particle types (loamy-skeletal or sandy-skeletal, 11.4% of land area).

TABLE 3.

Soil particle size family at infant botulism case residences^a compared to soil acreage^b in the 17-county rectangular area

Soil particle size family at case residence	Number of cases in 17-county area (% of cases)	Landscape area (km2) in 17-county area (% of area with particle size family data)
Coarsec	61 (18.5)	1,361 (12.4)
Fine/clayeyd	251 (76.3)	8,330 (76.2)
Loamy-skeletal or sandy-skeletale	17 (5.2)	1,245 (11.4)
Total	329 (100.0)	10,936 (100.0)
χ2 = 21.0, df = 2; P < 0.001

^a Excludes 86 IB cases on soils with unclassified particle size (e.g., urban land, fill).

^b Acreage excludes water, quarries, gravel pits, and 1,986 km² of unclassified soils.

^c Particles with diameters of 0.1–75 mm (fine sand or coarser, including gravel and artifacts 2–75 mm in diameter which are both cohesive and persistent).

^d Clay with particles <0.002 mm and silt with particles 0.002 to <0.062 mm in diameter.

^e Loamy-skeletal—coarse fragments comprise ≥35% soil mass by volume, with fine earth filling spaces >1 mm; sandy-skeletal—particles > 2 mm comprise >35% to <90% of mass by volume, with fine earth filling spaces > 1 mm.

DISCUSSION

Building on Long’s work (1) and over four decades of curating IB case data and mapping case distribution, it was apparent that the Delaware River Valley was an area with a disproportionately high number of IB cases. With its longer study period and larger number of cases, commonly applied epidemiologic methods in this study confirm the existence of a 17-county rectangular area that encompasses Long’s “ring” (43/53; 81% of then-recognized IB cases adjacent to the core city of Philadelphia) (1), and enlarges its dimensions to a rectangular area approximately 80 km (50 mi) × 250 km (155 mi), spanning five states. IB incidence in this area was six times greater than the nationwide incidence and nearly seven times greater than the incidence of IB in the remaining 161 counties of those five states (Table 1). There were no differences in the diet of IB cases from this area that could make them more susceptible to IB: milk feeding regimen did not differ from US IB cases, although IB cases from this area were less likely to be fed honey and solid food than US IB cases in the month before illness onset. Honey and some non-sterile solid foods could be a source of C. botulinum spores (34) and the introduction of solid foods changes the composition of the intestinal microbiota, at which time a few infants may become susceptible to colonization by C. botulinum (35).

There is also no clear explanation as to why IB incidence was higher than the nationwide incidence in the six other health jurisdictions of Anne Arundel and Washington counties, MD; Ulster County, NY; Richland County, OH; and Adams and Allegheny counties, PA. All of them contain river basins. Anne Arundel is part of the Piedmont region like the 17-county rectangular area; Washington County is part of the Great Appalachian Valley bounded by the Potomac River; and both Allegheny County (where the Allegheny, Monongahela and Ohio rivers meet in the city of Pittsburgh) and Ulster County (with watersheds draining into the Delaware and Hudson rivers) are on the Allegheny Plateau, the latter in its glaciated section (36). Three of the five cases in Richland County, OH resided on soil great group Epiaqualf, a soil associated with glacial moraines.

In the 17-county rectangular area, residence locations of IB cases were disproportionately associated with specific soil types. This finding was unexpected, although it has long been known that the natural habitat of C. botulinum worldwide is soil (23–26). Soils associated with IB case residences included those in the orders Alfisol and Entisol, in soil suborder Orthent, and the soil great groups Fragiudalf and Udorthent (Table 2; Fig. S3). Alfisols are organically rich soils that often formed under deciduous forests, accumulate soluble minerals and clays in the subsoil horizon, and contain high concentrations of basic cations [Ca(OH)₂, Mg(OH)₂, KOH, NaOH] that increase alkalinity. Entisols are a diverse group of “young” soils frequently present in floodplains with alluvial deposits and lack features of soil development (weathering, deposition of humic material, and enrichment due to animal and plant life) (37). Also, soils in coarse particle families were associated with IB cases. They are generally sandy and drain better than soils that are clay rich, and represent a high proportion of soils in CT, ME, MA, NH, NY, RI, and VT (38), where IB incidence is low. It is unknown if well-drained, more granular soils facilitate the growth of C. botulinum bacteria and/or dispersal of its spores into the air; and they represented only 12% of the landscape area in the 17 counties.

The predominance of proteolytic type B C. botulinum in this region and the relatively high organic content of the soils in these river valleys suggest that the soils of this area support its growth (24). By contrast, type A C. botulinum strains are typically isolated from soils with relatively low organic content (e.g., arid regions encompassing southern California, parts of Arizona and New Mexico, and west Texas), where pH is neutral to alkaline (38, 39). Acidic environments (pH <4.6) inhibit C. botulinum growth (39). Soil pH across the 17-county rectangular area has a surface pH well above 4.6 up to 6.3, whereas much of the northeastern United States, including most of New York state and the Eastern Seaboard, has acidic soil with pH of about 4.2 (38, 40), unless lime is applied to the soil locally. Spore germination conditions differ among proteolytic type A and B C. botulinum strains, with type B spore germination, stimulated more than type A in the presence of the amino acids L-alanine/L-lactate and optimal in bicarbonate buffers KHCO₃ and NaHCO₃ (41), reviewed by Shen et al. (42). One possible explanation for the association of type B cases with soil type is better spore germination and outgrowth of proteolytic type B C. botulinum in the pH-neutral soils of the 17-county rectangular area.

Organically rich Alfisols, which historically have been used for agriculture, were often extensively limed and fertilized (37, 43). In Argentina, Lúquez et al. found a higher prevalence of C. botulinum in soils under cultivated/urbanized/industrialized use than in virgin soils (44). Soil cultivation/disruption and the addition of fertilizers and minerals may inadvertently promote the growth of C. botulinum and the dispersal of spores. Isolation of C. botulinum from soil paradoxically does not always correlate with IB incidence. Despite the fact that all IB cases in Argentina from 1976 to 2006 were type A C. botulinum (45), type A strains represented a lower proportion (56.7%) of soil samples than IB case recognition would suggest (44). Luquez et al. identified type B C. botulinum in 15.3% of positive soil samples and not a single type B IB case occurred in Argentina from 1982 to 2005 (44, 46). The incidence of IB in Argentina (2.2 cases per 100,000 live births from 1982 to 2005) correlated with the proportion of soil samples positive for C. botulinum in the Northeast and West regions but not elsewhere, including the South region where IB incidence is high (8.8 cases per 100,000 live births) but only 14.7% of soil samples tested positive for C. botulinum (46). Similar to the results of the soil study in Argentina, Parry et al. isolated C. botulinum more frequently from agricultural than virgin soils in NY, United States (47). Over the past 30 years former agricultural regions in MD, NY, NJ, and PA have been increasingly developed as suburban residential communities around large cities (48). This changing land-use pattern and its associated soil disruption activities may have contributed to the greater incidence of IB in the 17-county rectangular area.

Our analysis has several limitations. For 13% of cases (n = 54), the soil type was unknown or urban landfill that could not be classified. Also, soil type was categorized only at the patient’s residence location. The analysis assumed that infants spent most of their time in and around their homes. Other environments to which the patient may have been exposed, such as family vacation homes, daycare facilities, or farms were not considered. Exposure to airborne C. botulinum spores may have occurred at places away from the family residence.

Another limitation of the analysis of soil taxa was that no soil samples were collected from case residences and tested for the presence of C. botulinum. Additional studies are needed. For example, a case-control study using soils collected at the residences of IB cases and age-matched healthy infants in the 17-county rectangular area could test the hypotheses that (i) C. botulinum detection in soil collected at the residence is associated with being an IB case or (ii) certain soil taxa are associated with detection of C. botulinum. Our unpublished data of soil types associated with cases and county controls in our California IB case-control study (49) showed no difference in soil types between cases and controls, albeit the soil types in California differ considerably from those in the Mid-Atlantic United States. It is unknown if human populations in the 17-county rectangular area are geographically clustered in areas that are associated with specific soil taxa since we do not have information about soil taxa for the birth population.

While there may be other explanations for the heightened occurrence of IB in the Delaware and Raritan River watersheds, an association with soil-related characteristics seems highly plausible for a soil-dwelling bacterium. What has been recognized about this area, also described as the Piedmont region (36), is that the retreat of the Wisconsin-stage continental Laurentide ice sheet, beginning approximately 25,000 years ago in the Pleistocene epoch, deposited glacial debris along a roughly east-west line across PA, NJ, and eastward onto what is now Long Island (Fig. 2) (50). Both Richland County, OH and Allegheny County, PA (which includes the city of Pittsburgh), where the incidence of IB was significantly higher than US incidence, are also located downstream from the terminal moraines of the former Pleistocene Laurentide ice sheet. Glacial debris deposited following the meltback of the ice sheet consisted of outwash, loess, tills, and the like that were enriched in chemical bases (e.g., NaOH, KOH), thereby increasing soil alkalinity. In the 17-county rectangular area, these glacial deposits were distributed southward along the Delaware and Raritan River watersheds by erosion and outflow from the lakes and swamps that formed behind the terminal moraines of the melting glacier (Fig. 2). The identification of a 17-county rectangular area in these watersheds with a disproportionately elevated occurrence of type B IB suggests that the soils in this region contain nutrients specifically supportive to the growth of type B C. botulinum bacteria and/or possess physical properties that favor the survival of C. botulinum spores.

MATERIALS AND METHODS

Definitions

A case was defined as a patient with clinical signs of IB (28) whose stool or enema sample subsequently tested positive for botulinum toxin and/or from which C. botulinum was isolated. Testing was done at state health departments or the CDC. The onset of illness was defined as the date of first contact with a medical professional for symptoms of the illness that was later confirmed to be IB by the mouse bioassay or culture. Age at onset was calculated as the difference between date of birth and date of illness onset. Soil taxa (taxon) refer to any of the following hierarchical class ranks of Soil Taxonomy: soil order, suborder, and great group.

Case ascertainment

Nationwide distribution of BIG-IV as a Treatment Investigational New Drug began in June 1998 and continued after its licensure under the name BabyBIG in October 2003. All laboratory-confirmed IB cases that occurred in the United States in the 20 years between January 1, 2000 and December 31, 2019 were included in this analysis. Cases were identified through requests for BIG-IV and annual reconciliation of cases recorded by the IBTPP with CDC’s nationwide list. Human subjects’ approval was not necessary for this study since the data were collected as part of regular surveillance activities and the 20-year time period at the geographic level of the county ensures anonymity.

Case characteristics

The 830 cases included in this report came from DC and the 14 northeastern US states of Connecticut (CT), DE, Maine (ME), MD, Massachusetts (MA), New Hampshire (NH), NJ, NY, OH, PA, Rhode Island (RI), Vermont (VT), Virginia (VA), and West Virginia (WV); 759 were treated with BIG-IV and 71 were untreated. The patient’s age, race, sex, month of onset, and maternal residence address at illness onset were obtained from IBTPP, state, or CDC records. In instances of mixed-race ethnicity, the infant race was coded as that of the mother.

Identification and definition of counties with elevated IB incidence

County-specific birth data were obtained from individual state health department websites or by request to state vital statistics units. For the 20-year period 2000–2019, the cumulative incidence of IB for each county in the 14 northeastern states plus DC (Fig. 1) was calculated using the number of cases as the numerator and the total number of live births in the same time period as the denominator, with asymptotic 95% CIs calculated according to the method of Szklo and Nieto (32).

A county was considered to have elevated IB incidence if the lower bound of the 95% CI exceeded the upper limit of the 95% CI for US nationwide incidence in the same time period. Contiguous counties with elevated incidence were designated as members of a region with elevated IB occurrence (Fig. 2).

Residence soil type of cases

Soil data were acquired using the Web Soil Survey user interface to access the State Soil Geographic Database (SSURGO), Natural Resources Conservation Service (NRCS), US Department of Agriculture (USDA) (51), which includes information on the spatial extent of soil map units and associated soil characteristics. For this study, tabular data of interest included soil order, suborder, great group, and particle size family (fine, coarse, or other) of the dominant soil component in each mapping unit associated with a case residence address in any area with elevated IB incidence. Cases were electronically geocoded using ArcGIS Pro 3.2.0 (ESRI, Redlands, CA). Cases without an address were coded to the ZIP code of county residence (n = 1). Fifty-four cases that could not be classified as soil taxa from the SSURGO database due to soil descriptors such as “urban fill,” “landfill,” or “cut/fill” were excluded from the analysis of soil taxa.

Soil taxonomy

Soil taxonomy is a hierarchical system with six levels. Soil order is the highest, followed in descending rank by suborder, great group, subgroup, family, and series (52). More than 19,000 soil series exist in the United States (52). Soils that differ in their developmental, chemical, or physical characteristics (e.g., shallow vs deep vs bedrock) may be assigned to different soil series, and each soil series can be classified via soil taxonomy.

There are 12 soil orders in the United States, each with a unique developmental pathway and resultant physical and chemical characteristics. As an example of soil order, Alfisols are described as soils with evidence of clay illuviation and accumulation resulting in an argillic horizon (silicate clay layer) and a base saturation of >35% (percent of soil exchange sites containing basic cations) (37).

Within a region of elevated IB incidence, the percentage of land area (km²) for each soil order, suborder, and great group was calculated using the predefined summary report for soil surface area generated from SSURGO tabular data in Microsoft Access (51). Areas designated as water, quarries, gravel pits, dams, or unclassifiable urban fill were excluded from soil area calculations.

Soil particle size family

The area represented by each soil particle size family in any area of elevated incidence was derived by looking up the associated soil’s alphanumeric code and calculating the corresponding soil area.

Statistical methods

Statistical analyses were performed in Microsoft Excel (Redmond, WA) and P-values < 0.05 were considered significant. Mean ages at onset for infant botulism type B cases in the counties with elevated IB incidence and for all other US IB type B cases from 2000 to 2019 were compared using a two-sided t-test for distributions with equal variance and diet variables were compared using a Chi-squared test with significance at α < 0.05. Possible association of cases with specific soil taxonomic criteria in any area of elevated incidence was evaluated with the Chi-squared test, excluding soil orders that fell outside the most common orders, so as to calculate a more conservative Chi-squared statistic. Excluded soils in the 17-county rectangular area amounted to 4.3% of land area and 0% of IB cases. The expected number of cases for the Chi-squared test was calculated by multiplying the total number of cases in counties of elevated incidence by the percentage from the area represented by each soil order, suborder, and great group (calculations shown in Tables S4A through C online); and particle size family in those counties (Table 3).

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.01063-24.

Figure S3. aem.01063-24-s0001.pdf.

Hierarchies and descriptions of soils associated with infant botulism cases in the 17-county rectangular area.

Table S4. aem.01063-24-s0002.xlsx.

Soil classifications.

Table S5. aem.01063-24-s0003.xlsx.

Listing of live births, infant botulism cases, and incidence by county in 14 northeastern US states and DC, 2000-2019.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.