Direct hospitalization costs associated with resistant and susceptible Salmonella and Shigella infections in the United States, 2012–2019

Abstract

Antimicrobial-resistant enteric infections may be difficult to treat, associated with severe outcomes, and costly. To determine if resistant enteric infections are more costly, we described direct hospitalization costs for hospital admissions for typhoid fever, salmonellosis, and shigellosis infections by resistance status. We included all inpatient admissions with culture-confirmed infections from approximately 300 hospitals in the Premier Healthcare database during 2012–2019. We classified typhoid fever, salmonellosis, and shigellosis infections as clinically resistant if laboratory testing found resistance to ≥1 antibiotic class recommended for treatment. We examined patient demographic characteristics and underlying chronic conditions. We used the Wilcoxon rank sum test to assess the difference in median hospitalization cost for each patient admission by resistance status. We estimated adjusted cost and length of stay using multivariable linear regression, adjusting for host and healthcare provider characteristics. Among 67 inpatient admissions for typhoid fever, 3,421 for salmonellosis, and 619 for shigellosis, 33% of typhoid fever, 13% of salmonellosis, and 82% of shigellosis infections were clinically resistant. The unadjusted median costs for resistant salmonellosis were higher than susceptible infection-related costs ($7,753 vs $6,909, p=0.002). The unadjusted costs for resistant versus susceptible typhoid fever ($13,709 vs $9,254, p=0.111) and shigellosis were similar ($5,365 vs $5,777, p=0.167). For all three pathogens, the adjusted mean hospitalization costs were not statistically different for resistant and susceptible infections. Adjusted mean length of stay was significantly longer for resistant typhoid fever compared with susceptible infections (6.50 days vs 4.35 days, p=0.001). Although cost differences for resistant infections were not significant, we observed differences by host factors including age and comorbidities, suggesting patient demographic characteristics for resistant infections are impactful on cost. These findings support the premise that public health interventions to limit the spread of resistant infections, particularly among individuals at higher risk for severe illness, could reduce hospitalization costs.

Introduction

Enteric infections caused by Salmonella and Shigella lead to substantial morbidity in the United States. Foodborne illness costs are estimated to be as high as $90 billion annually (Scharff, 2015). Antimicrobial-resistant Salmonella and Shigella infections are serious public health threats; resistance can diminish treatment options and spread to other bacteria (CDC, 2019).

Salmonella enterica can be classified as typhoidal and nontyphoidal. Salmonella Typhi is estimated to cause 13 million typhoid fever infections worldwide and 5,700 infections in the United States annually (CDC, 2024d; Scallan et al., 2011). Extensively drug-resistant (XDR) typhoid fever cases have emerged among international travelers and non-travelers (CDC, 2021, Chatham-Stephens et al., 2019). Nontyphoidal Salmonella (NTS) includes most Salmonella serotypes and causes an estimated 2.4 million salmonellosis infections in the United States annually (Scallan Walter et al., 2025). During 2015–2017, 16% of NTS isolates were resistant to at least one clinically important antibiotic (CDC, 2019). Shigella species cause approximately 450,000 infections in the United States annually (CDC, 2019). During 2015–2017, 17% of Shigella isolates had decreased susceptibility to ciprofloxacin and 14% had decreased susceptibility to azithromycin (CDC, 2019). XDR shigellosis increased from 0% in 2015 to 5% in 2022 (CDC, 2023b).

Substantial healthcare costs have been attributed to resistance (Broughton et al., 2010; Nelson et al., 2021), but direct hospitalization costs associated with resistant typhoid fever, salmonellosis, and shigellosis in the United States remain unknown. Understanding the cost burden of these pathogens could help allocate resources to prevent resistance. We aimed to describe the characteristics of patients hospitalized with antimicrobial-resistant typhoid fever, salmonellosis, and shigellosis, and to determine if they have higher direct medical costs compared to persons hospitalized with susceptible infections.

Methods

Data source

Premier Healthcare Database is a U.S. hospital-based, all-payor electronic healthcare database (PINC Applied Sciences, 2023). Data provided included facility characteristics, patient demographics, admission and discharge diagnoses, laboratory and microbiology test results, total (billed) patient hospitalization costs (inclusive of both fixed and variable costs), and patient disposition. We included inpatient admissions from hospitals reporting culture-confirmed Salmonella or Shigella species in any clinical isolate during January 1, 2012–December 31, 2019. We calculated total hospitalization costs for typhoid fever, salmonellosis, and shigellosis. Salmonella Paratyphi was excluded due to the small number of isolates (Supplemental Table 1).

We categorized specimen sources into stool, urine, invasive (including intra-abdominal), and other sources (Supplemental Table 2) (Specimen Source Working Group, 2016). For patients with multiple specimens collected during a hospital admission, the earliest specimen with complete resistance information was included. Patients with multiple hospitalizations (within 12 months for typhoid fever and salmonellosis and 3 months for shigellosis) had only the earliest hospital admission with complete resistance information included.

Drug categorization and antimicrobial susceptibility testing

Patients hospitalized with Salmonella or Shigella were classified as clinically resistant or susceptible by antimicrobial susceptibility status. Minimum inhibitory concentration data were unavailable for all isolates, preventing application of current Clinical and Laboratory Standards Institute (CLSI) breakpoints. Instead, we used clinical laboratory interpretation, which likely reflected the CLSI breakpoint at the time of testing. We defined clinically resistant Salmonella (Typhi and NTS) as resistant to ≥1 of the drug categories used for the treatment of Salmonella: fluoroquinolones (resistant or intermediate results), third generation cephalosporins, penicillins, or folate pathway inhibitors (Supplemental Table 3). Isolates without known resistance to these four classes were classified as susceptible. We defined clinically resistant Shigella as resistant or intermediate to fluoroquinolones, or resistant to penicillins or folate pathway inhibitors, and defined isolates as susceptible if they were susceptible to all 3 of these drug classes. Although macrolides are recommended for treatment, only a few isolates were tested for susceptibility to macrolides and were not included in the clinically resistant definition (Supplemental Table 4). Third generation cephalosporin were also excluded for Shigella because few isolates were tested. Patients with isolates that could not be classified as resistant or susceptible were excluded.

Underlying conditions and Elixhauser Comorbidity Index

Underlying health conditions were identified using International Classification of Diseases (ICD) 9 and 10 codes by the Clinical Classifications Software developed by the Agency for Healthcare Research and Quality (AHRQ) (AHRQ, 2017a; AHRQ, 2021). We categorized codes into 50 chronic conditions and grouped these into 11 categories (Table 2). We excluded acute conditions and those related to the infection (e.g., sepsis, bacterial infection, Salmonella infection). We used the AHRQ Elixhauser Comorbidity Index to predict in-hospital mortality risk with ICD code categorization (AHRQ, 2017b; Elixhauser et al., 1998). Higher scores indicate greater illness severity (HCUP, 2024). Elixhauser AHRQ scores were computed using comorbidity R package v 1.1.0 (Gasparini, 2018) and classified into 4 categories based on the distribution of severity: ≤3, 4–10, 11–14, and ≥15.

Statistical analysis

All statistical analyses aside from Elixhauser score were performed using SAS v9.4 (Cary, NC, USA). Differences in demographic and clinical characteristics of patients infected with clinically resistant versus susceptible pathogens were assessed using Chi-square or Fisher’s exact test as appropriate for categorical variables and Wilcoxon rank sum test for continuous variables. The unadjusted cost differences of hospitalization for resistant versus susceptible infections were examined using Wilcoxon rank sum test and reflect observed cost differences without accounting for patient or hospital characteristics. Adjusted estimates for cost and length of stay (LOS) account for potential confounding factors and were obtained using generalized linear models: a gamma distribution with a log link for cost, and a negative binomial distribution with a log link for LOS to account for skewness and overdispersion. We accounted for potential correlation between patients within the same facility (except for typhoid fever due to low count). The model included antimicrobial susceptibility as a main exposure and adjusted for demographic characteristics (sex, age, and race/ethnicity), Elixhauser score of comorbidities, year of hospital discharge, specimen source, and available hospital level characteristics (e.g., urban versus rural classification, teaching versus non-teaching, number of beds, U.S. Census Region, and cost type [procedural versus cost to charge ratio]). P-values <0.05 were considered statistically significant. Hospitalization costs were inflation-adjusted to 2019 using the Bureau of Labor Statistic’s Producer Price Index for general medical and surgical hospitals (https://data.bls.gov/timeseries/pcu622110622110). Total cohort cost was calculated by multiplying the model-estimated individual cost by the sample size. This activity was reviewed by the CDC, deemed not research, and was conducted consistent with applicable federal law and CDC policy¹.

Results

During 2012–2019, we identified 4,319 inpatient admissions to 256 hospitals with culture-confirmed Salmonella or Shigella infection. We excluded 73 (2%) hospitalizations related to the same infection, 58 (1%) with missing cost data, and 81 (2%) with incomplete susceptibility profiles; 4,107 inpatient admissions with culture-confirmed typhoid fever, salmonellosis, or shigellosis were included in the analysis: 67 (2%) typhoid fever, 3,421 (83%) salmonellosis, and 619 (15%) shigellosis patients.

Demographic and clinical characteristics

Approximately 52% of persons hospitalized with typhoid fever or salmonellosis and 43% of persons hospitalized with shigellosis were female. The most common age group for hospitalized patients with typhoid fever and shigellosis was 18–49 years, and ≥65 years for salmonellosis. Most patients (65%) with salmonellosis were non-Hispanic White; 20% of typhoid fever patients and 47% of shigellosis patients were non-Hispanic White. The median LOS was 6 days (interquartile range [IQR]: 3.5-8) for typhoid fever, 4 days (IQR: 3–6) for salmonellosis, and 3 days (IQR: 2–4) for shigellosis.

Antimicrobial susceptibility testing showed that 33% Salmonella Typhi, 13% NTS, and 82% Shigella isolates were clinically resistant (Table 1). Persons hospitalized with resistant shigellosis were more often male compared with susceptible infection (60% versus 42%, p<0.001). Resistance was associated with age for salmonellosis (p=0.029) and with race/ethnicity for typhoid fever (p=0.006); all resistant typhoid infections were among non-Hispanic Black and non-Hispanic patients of other races. Resistance was also associated with race/ethnicity for salmonellosis infections (p<0.001); a lower proportion of resistant infections were among non-Hispanic White patients (55% vs 66%), and a higher proportion of resistant infections were among non-Hispanic Black patients (21% vs 16%) and non-Hispanic other race patients (18% vs 12%) compared with susceptible infections. LOS of resistant salmonellosis was statistically longer (p=0.030) and LOS of resistant shigellosis was statistically shorter (p=0.041) compared with susceptible infections, despite a similar median number of days. Resistance was not associated with Elixhauser scores for typhoid fever or salmonellosis; a significant association was observed for shigellosis (p=0.013).

Table 1.

Demographic and clinical characteristics of patients, by pathogen and resistance type, Premier Healthcare Database, 2012–2019^a

	Typhoid fever N= 67			Salmonellosis N= 3,421			Shigellosis N= 619
	Resistant	Susceptible	p-value	Resistant	Susceptible	p-value	Resistant	Susceptible	p-value
	22 (33)	45 (67)		450 (13)	2,971 (87)		509 (82)	110 (18)
Sex			0.792			0.530			<0.001
Female	>11 (>50)	23 (51)		229 (51)	1,559 (52)		204 (40)	64 (58)
Male	<11 (<50)	22 (49)		221 (49)	1,412 (48)		305 (60)	46 (42)
Age, years			0.159b			0.029			0.074
<5	<11 (<50)	<11 (<24)		55 (12)	358 (12)		40 (8)	13 (12)
5 to 17	<11 (<50)	<11 (<24)		32 (7)	149 (5)		63 (12)	16 (15)
18 to 49	<11 (<50)	23 (51)		126 (28)	692 (23)		254 (50)	41 (37)
50 to 64	<11 (<50)	<11 (<24)		104 (23)	726 (24)		80 (16)	16 (15)
≥65	<11 (<50)	<11 (<24)		133 (30)	1,046 (35)		72 (14)	24 (22)
Race/Ethnicity			0.006b			<0.001			0.266
Non-Hispanic White	0 (0)	13 (30)		245 (55)	1,958 (66)		231 (46)	>57 (>52)
Non-Hispanic Black	<11 (<50)	<11 (<24)		93 (21)	467 (16)		143 (28)	26 (24)
Hispanic	0 (0)	<11 (<24)		27 (6)	169 (6)		68 (13)	<11 (<10)
Non-Hispanic Other	>11 (>50)	25 (57)		79 (18)	359 (12)		62 (12)	14 (13)
Median LOS [IQR], days c	6 [5–8]	5 [3–8]	0.103	4 [3–7]	4 [3–6]	0.030	3 [2–4]	3 [2–5]	0.041
Specimen source			0.913b			0.491			0.111b
Stool	<11 (<50)	<11 (<24)		270 (60)	1,866 (63)		447 (88)	92 (84)
Urine	<11 (<50)	<11 (<24)		35 (8)	249 (8)		19 (4)	<11 (<10)
Invasive	14 (64)	30 (70)		115 (26)	688 (23)		22 (4)	<11 (<10)
Other	<11 (<50)	<11 (<24)		28 (6)	154 (5)		20 (4)	<11 (<10)
Elixhauser score, n (%)			0.579			0.069			0.013
≤3	<11 (<50)	>10 (>22)		124 (28)	687 (23)		147 (29)	36 (33)
4–10	<11 (<50)	<11 (<24)		80 (18)	665 (22)		84 (16)	14 (13)
11–14	<11 (<50)	12 (27)		122 (27)	793 (27)		208 (41)	33 (30)
≥15	<11 (<50)	12 (27)		124 (28)	826 (28)		70 (14)	27 (25)

^aFrequencies may not sum to total N due to missing data; percentages exclude missing values. Cells with 10 or fewer observations were suppressed.

^bProportions were compared using the Chi-square test or Fisher’s exact test, as appropriate.

^cMedian LOS was compared between groups using Wilcoxon Rank Sum Test.

Overall, the most common underlying chronic conditions were cardiovascular and gastrointestinal disease (Table 2). Among those hospitalized with resistant versus susceptible shigellosis, the proportion of patients with an immunosuppressive condition (18% vs 1%, p<0.001) or with a nutritional and metabolic disorder (19% vs 10%, p=0.03) was higher, and the proportion of patients with cardiovascular disease (36% vs 51%, p<0.01) or diabetes mellitus (14% vs 25%, p<0.007) was lower.

Table 2.

Number and proportion^a of patients with chronic underlying conditions^b by pathogen and resistance type, Premier Healthcare Database, 2012–2019

	Typhoid fever N= 67			Salmonellosis N= 3421			Shigellosis N= 619
	Resistant	Susceptible	p-value	Resistant	Susceptible	p-value	Resistant	Susceptible	p-value
	22 (33)	45 (67)		450 (13)	2971 (87)		509 (82)	110 (18)
Cardiovascular disease	<11 (<50)	15 (33)	0.806	264 (59)	1,827 (61)	0.251	181 (36)	56 (51)	0.003
Gastrointestinal disease	<11 (<50)	19 (42)	0.235	219 (49)	1,505 (51)	0.431	241 (47)	46 (42)	0.292
Pulmonary disease	<11 (<50)	<11 (<24)	0.542	143 (32)	921 (31)	0.740	92 (18)	27 (25)	0.118
Renal disease	<11 (<50)	<11 (<24)	0.223	148 (33)	1,105 (37)	0.077	121 (24)	26 (24)	0.976
Hepatic disease	<11 (<50)	<11 (<24)	0.999	41 (9)	262 (9)	0.839	28 (5)	<11 (<10)	0.686
Diabetes Mellitus	<11 (<50)	<11 (<24)	0.999	123 (27)	804 (27)	0.904	72 (14)	27 (25)	0.007
Neurologic disease	0 (0)	<11 (<24)	0.167	115 (27)	740 (25)	0.767	100 (20)	26 (24)	0.346
Malignancy	0 (0)	<11 (<24)	0.999	>36 (>8)	274 (9)	0.703	11 (2)	<11 (<10)	0.320
Immunosuppressive condition	0 (0)	<11 (<24)	0.999	<11 (<2)	63 (2)	0.635	94 (18)	<11 (<10)	<0.001
Obesity	<11 (<50)	<11 (<24)	0.999	58 (13)	366 (12)	0.732	32 (6)	<11 (<10)	0.469
Nutritional and metabolic disorder	<11 (<50)	<11 (<24)	0.999	106 (24)	758 (26)	0.373	96 (19)	11 (10)	0.026

^aProportions were compared using the Chi-square test or Fisher’s exact test, as appropriate.

^bObserved condition only. Patients can have more than one chronic underlying condition. Cells with 10 or fewer observations were suppressed.

Unadjusted cost

The median cost for resistant typhoid fever was $13,709, compared with $9,254 for susceptible infections (p=0.111, Table 3). The median cost for resistant salmonellosis was $7,753, compared with $6,909 for susceptible infections (p=0.002). The median cost for resistant shigellosis was $5,365, compared with $5,777 for susceptible infections (p=0.167).

Table 3.

Unadjusted median cost^a of a hospital stay for resistant versus susceptible infection, overall and by pathogen type, Premier Healthcare Database, 2012–2019

	Typhoid fever			Salmonellosis			Shigellosis
	Resistant	Susceptible	p- value	Resistant	Susceptible	p- value	Resistant	Susceptible	p- value
N	22	45		450	2971		509	110
Median cost, $	13,709	9,254	0.111	7,753	6,909	0.002	5,365	5,777	0.167
Interquartile range, $	[8,700–18,147]	[6,600–15,573]		[4,946–13,823]	[4,423–11,774]		[3,630–8,101]	[3,943–9,469]

^aMedian cost was compared between groups using Wilcoxon Rank Sum Test.

Adjusted cost

Adjusted mean cost of hospitalization for resistant typhoid fever was $11,208 and $8,733 for susceptible infections, $15,028 for resistant salmonellosis and $14,111 for susceptible infections, and $9,259 for resistant shigellosis and $10,661 for susceptible infections (all p>0.05, Table 4). From 2012 to 2019, total hospitalization cost in this cohort associated with resistant typhoid fever, salmonellosis and shigellosis was approximately $11.7 million while the mean total hospitalization cost for susceptible infections was $40.9 million for this sample population (Table 4).

Table 4.

Estimates of adjusted costs^a,b for a hospital stay for resistant versus susceptible infection, by pathogen and resistant status, adjusting for demographic and clinical characteristics, from multivariable regression, Premier Healthcare Database, 2012–2019

	Typhoid feverc			Salmonellosis			Shigellosis
	Coefficient	95% CI	p-value	Coefficient	95% CI	p-value	Coefficient	95% CI	p-value
Resistant status
Resistant	1.28	(0.98–1.68)	0.068	1.06	(0.92–1.23)	0.386	0.87	(0.70–1.08)	0.214
Susceptible	Ref			Ref			Ref
Sex
Female	0.98	(0.75–1.29)	0.900	0.99	(0.93–1.06)	0.861	0.91	(0.82–1.00)	0.053
Male	Ref			Ref			Ref
Age, years
<5	1.02	(0.51–2.04)	0.960	1.07	(0.85–1.34)	0.571	0.78	(0.60–1.00)	0.047
5–17	1.34	(0.96–1.87)	0.086	1.12	(0.92–1.38)	0.264	0.80	(0.67–0.95)	0.012
18–49	Ref			Ref			Ref
50–64	1.21	(0.70–2.10)	0.493	1.19	(1.07–1.31)	0.001	1.24	(1.02–1.51)	0.027
≥65	1.09	(0.58–2.05)	0.786	1.18	(1.06–1.32)	0.003	1.23	(1.00–1.50)	0.046
Race/Ethnicity
Non-Hispanic White	Ref			Ref			Ref
Non-Hispanic Black	1.04	(0.61–1.77)	0.893	1.00	(0.90–1.12)	0.950	1.13	(0.94–1.36)	0.191
Hispanic	1.83	(0.78–4.31)	0.164	0.99	(0.84–1.17)	0.883	0.85	(0.71–1.01)	0.058
Non-Hispanic Other	1.38	(0.95–1.99)	0.087	0.99	(0.80–1.22)	0.923	0.97	(0.80–1.18)	0.761
Elixhauser score, quartile
≤3	Ref			Ref			Ref
4–10	1.82	(1.03–3.19)	0.038	1.05	(0.94–1.16)	0.374	1.07	(0.86–1.33)	0.567
11–14	1.24	(0.90–1.70)	0.182	0.97	(0.86–1.08)	0.555	1.04	(0.93–1.15)	0.490
≥15	1.64	(1.18–2.27)	0.003	1.65	(1.45–1.87)	<0.001	1.54	(1.22–1.93)	<0.001
Year	1.01	(0.94–1.08)	0.839	1.00	(0.98–1.02)	0.873	1.01	(0.98–1.05)	0.350
Specimen source
Stool	Ref			Ref			Ref
Urine	1.23	(0.72–2.08)	0.446	1.59	(1.38–1.82)	<0.001	1.09	(0.78–1.53)	0.614
Invasive	1.29	(0.91–1.84)	0.155	2.01	(1.81–2.23)	<0.001	1.78	(1.18–2.68)	0.006
Other	1.26	(0.81–1.97)	0.300	2.32	(1.58–3.40)	<0.001	1.17	(0.78–1.76)	0.441
Estimated adjusted cost, Mean $ (95% CI)d
Resistant	11,208 (8,628–14,560)			15,028 (12,615–17,902)			9,259 (7,321–11,711)
Susceptible	8,733 (6,684–11,409)			14,111 (12,480–15,956)			10,661 (7,912–14,367)
Total cost (based on mean adjusted cost), $
Resistant	246,576			6,762,600			4,712,831
Susceptible	392,985			39,383,801			1,172,710

^aHealthcare providers characteristics adjusted for in the model included (urban versus rural classification, teaching versus non-teaching facility, number of hospital beds, U.S. Census Region, and cost type (procedural versus ratio of cost to charges).

^bThe total cost of hospitalization was adjusted for inflation from 2012 to 2019 based on the Bureau of Labor Statistic’s Producer Price Index (www.bls.gov/ppi/) prior to fitting in model.

^cCorrelation between hospitalization within the same healthcare provider were not considered for typhoid fever due to low cell count.

^dP-values for the comparison of estimated adjusted mean costs between resistant and susceptible groups are presented in the ‘Resistance Status’ row of the table.

By demographic and clinical characteristics, typhoid fever hospitalization cost was 82% higher for patients with Elixhauser scores of 4–<11 (p=0.038) and 64% higher with those ≥15 (0.003) compared with those with scores ≤3. Salmonellosis hospitalization cost was higher for patients aged 50–64 years (p=0.001) and aged ≥65 years (p=0.003) compared with patients aged 18–49 years, for those with Elixhauser scores of ≥15 compared with those with scores ≤3 (p<0.001), and for those with urine, invasive, or other specimen source compared with stool (all p<0.001). Shigellosis hospitalization cost was higher for patients aged 50–64 years (p=0.027) and ≥65 years (p=0.046) compared with patients aged 18–49 years, for those with Elixhauser scores of ≥15 compared with those with scores ≤3 (p<0.001), and for those with an invasive specimen source compared with stool (p=0.006). Shigellosis hospitalization costs were lower for patients aged <5 years (p=0.047) and 5–17 years (p=0.012) compared with patients aged 18–49 years.

Adjusted LOS

When LOS was the outcome in multivariable regression, the mean LOS estimated was significantly longer for resistant typhoid fever compared with susceptible infections (6.50 days vs 4.35 days, p = 0.001, Supplemental Table 5). For typhoid fever, Elixhauser scores of 4–<11 (p=0.004) and ≥15 (p=0.006, compared with those with scores ≤3), and invasive specimens (p=0.020, compared with stool) were associated with longer LOS, and persons aged 50–64 years (p=0.035, compared with 18–49 years) were associated with shorter LOS. For salmonellosis and shigellosis hospitalizations, LOS was not significantly different by resistance status, and similar characteristics that were associated with higher cost were also associated with longer LOS.

Discussion

This study provides important data on direct hospitalization costs in the United States for resistant enteric pathogens. In our analysis, 33% of typhoid fever, 13% of salmonellosis, and 82% of shigellosis patients had clinically resistant infections. Unadjusted hospitalization costs were significantly higher for resistant salmonellosis than for susceptible infection; whereas costs for resistant typhoid fever and shigellosis were similar to susceptible infection. After adjusting for covariates, cost differences were no longer significant, suggesting that host factors including age and comorbidities may explain some of the observed differences. These findings imply that patient demographic characteristics may impact hospitalization costs more than resistance.

Studies outside of the United States have found that direct medical costs of resistant typhoid fever and salmonellosis are higher than susceptible infections (Broughton et al., 2010; Mejia et al., 2020). There is limited evidence on the cost of resistant shigellosis. Studies of non-enteric pathogens have linked resistant infections to longer hospital LOS and higher costs (Kadam et al., 2024; Serra-Burriel et al., 2020). Higher costs may partly reflect treatment failure when pathogens are resistant to first-line antibiotics, requiring hospitalization (Crump et al., 2003; Park et al., 2019). Demographic factors, including age and race/ethnicity have also been associated with increased costs for treating resistant infections (Thorpe et al., 2018).

Host factors (e.g., multiple comorbidities) contributed to increased hospitalization costs in our multivariable model, consistent with previous work linking higher costs and longer LOS for resistant infections (Cummings et al., 2016; GBD 2017 Collaborators, 2019; Trevejo et al., 2003). Although we could not adjust for strain factors (e.g., serotype and virulence markers), they can also contribute to increased costs. Resistance and virulence may be linked through genes on a common plasmid. A study of quinolone-resistant salmonellosis concluded that increased LOS was due to the virulence of the infection rather than antibiotic treatment failure (Broughton et al., 2010). Treatment failure or more virulent infection could result in higher costs through additional treatment requirements, complications, and longer LOS.

Typhoid fever has a high clinical burden, causing severe illness that often results in extraintestinal infection and requires antibiotic treatment (Chatham-Stephens et al., 2019). During 2015–2020, 82% of patients reported to the National Typhoid and Paratyphoid Fever Surveillance System were hospitalized (CDC, 2024d). Although fluoroquinolones were recommended as a first line typhoid fever treatment during part of our study period, they are not currently recommended for empiric treatment due to high rates of nonsusceptibility and risk of treatment failure (CDC, 2019; Chatham-Stephens et al., 2019; Rupali et al., 2004). International travel, especially to Southeast Asia, may increase the risk of resistant typhoid fever infection (Francois Watkins et al., 2025). We lacked information on travel or vaccination status information, which could influence costs independent of antimicrobial resistance. Additionally, typhoid fever is uncommon in the United States, and the small sample size limited statistical power. As a result, the typhoid model did not account for correlation between hospitalizations within the same healthcare facility due to low cell counts. Despite these limitations, we found that adjusted hospital LOS was significantly longer for resistant infections compared with susceptible infections.

Salmonellosis is a leading cause of foodborne illness in the United States (CDC, 2024a). Children under five years are most likely to acquire infection, and infants and adults ≥65 years have a higher risk of hospitalization (Shane et al., 2017). Most illnesses are self-limiting; approximately 28% of patients are hospitalized (CDC, 2023a). Higher unadjusted costs of resistant hospitalized salmonellosis may be because resistant infections result in more severe health outcomes with higher risk of hospitalization compared with susceptible infections (Francois Watkins et al., 2024; Parisi et al., 2018). Our finding that host factors are more influential for costs than resistance alone highlights the importance of preventing resistant infections among individuals at higher risk for severe infection.

Most shigellosis illnesses are self-limiting, and approximately 25% of patients in the United States are hospitalized (CDC, 2023a). In our study, only 18% of Shigella infections were susceptible to tested antibiotics, and more patients with resistant shigellosis were men, or had immunosuppressive conditions or nutritional metabolic disorders compared with patients with susceptible shigellosis. Certain groups may be more likely to have resistant infections, including international travelers, men who have sex with men (MSM), people with weakened immune systems, and people who live in congregate settings (CDC, 2024b; CDC, 2024c). Regardless, resistant shigellosis resulted in higher overall cost, and infections have been increasing (Baker and Scott, 2023; CDC, 2023b; CDC, 2024b), emphasizing the need for strategies to prevent resistant infections.

This study has several limitations. First, we did not account for the entire medical cost of disease management and treatment, did not attempt to attribute costs to specific complications, and could not evaluate daily resource utilization. We examined costs using inpatient data for one hospitalization, and costs associated with previous or subsequent outpatient or hospitalized care were not included. Also, hospitalized patients likely have a severe infection, so our findings may not be generalizable to mild or moderate infections treated in an outpatient setting or to those who did not seek healthcare. Second, we did not examine which drug(s) the patients were treated with or whether treatment failure played a role in cost of admission or LOS. Another limitation is the possibility of error or variability in the documentation of comorbidities that we used to inform the Elixhauser scores used in our cost and LOS models, potentially resulting in parameter estimates that are biased or inconsistent, and inaccurate statistical inferences. Our analysis did not include cases who may have developed resistance during hospitalization or whose resistance was not apparent on the first specimen. Additionally, we were not able to apply current CLSI breakpoints to all isolates and had to rely on the interpretation from the clinical laboratory, and we were unable to look at multidrug resistance. During the study period, guidance from CLSI was to only report susceptibility results for ampicillin, a fluoroquinolone, and trimethoprim-sulfamethoxazole, in part because there were no clinical breakpoints defined by CLSI for some treatment drugs (e.g., azithromycin) (CLSI, 2019). Therefore, results for susceptibility to all recommended treatment drugs were not available and we may have misclassified resistant infections as susceptible. This is likely most significant for shigellosis; in 2019, 36% of Shigella isolates tested by NARMS were resistant to azithromycin (CDC, 2025).

Conclusion

Although resistant Salmonella and Shigella hospitalization costs were not significantly higher when adjusted for host and healthcare factors, antimicrobial-resistant infections still impose substantial medical costs on patients and the healthcare system (Hofer, 2019; Johnston et al., 2019; Nelson et al., 2021; Thorpe et al., 2018; Wozniak et al., 2022). Public health prevention measures, including improving food safety, judicial use of antibiotics, and enhancing surveillance in at-risk populations, remain essential to reduce incidence and lessen the burden on both patients and the U.S. healthcare system. Future studies investigating potential effect modification by patient characteristics or incorporating detailed clinical management and resource utilization data could better characterize specific factors affecting costs.