Targeted literature review of the burden of extraintestinal pathogenic Escherichia Coli among elderly patients in Asia Pacific regions

Abstract

Background

Extraintestinal Pathogenic Escherichia coli (ExPEC) is a leading cause of invasive disease, including bacteremia and sepsis. Invasive ExPEC disease (IED) has the potential to complicate the clinical treatment of other conditions and is associated with an increased mortality, hospitalization, and worse outcomes. Older adults and individuals with comorbid conditions are at higher risk of IED. ExPEC is of particular concern in the Asia Pacific region due to aging populations and rising antimicrobial resistance.

Objectives

This study aimed to synthesize most recent data on the epidemiology, clinical and economic burden of IED in the elderly/high risk populations in China, Japan, South Korea, Taiwan, and Australia.

Methods

A targeted literature review was conducted using Embase, Medline, as well as local scientific databases. We included studies published in English and local languages published from January 1, 2010 to October 7, 2020 that were relevant to the research objectives. Studies were narratively synthesized.

Results

A total of 1,047 studies were identified and 34 of them were included in this review. ExPEC accounted for 46.0% (1,238/2,692) of bacteria-related invasive diseases in patients aged above 60 years in South Korea, followed by China (44.4% (284/640)), Taiwan (39.0% (1,244/3,194)), and Japan (18.1% (581/3,206)), while Australia reported ExPEC out of all pathogens (54.7% (4,006/7,330)) in general adults. Comorbidities such as diabetes or cancer were common in these patients. Studies reported increases in length-of-stay, and in-hospital 30-day all-cause mortality related to ExPEC associated bacteremia was between 9% to 12%. From a cost perspective, a 3-fold increase in sepsis-associated cost was reported in South Korea between 2005 and 2012. In Australia, antimicrobial resistance contributed to an additional cost of AUD $5.8 million per year (95% uncertainty interval [UI], $2.2–$11.2 million) in the treatment of bloodstream infections (BSIs).

Conclusion

ExPEC was a major cause of blood stream infection across China, Japan, South Korea, Taiwan, and Australia. Both the clinical and economic burden associated to ExPEC infections as well as the antimicrobial resistance observed in the elderly call for preventive and curative actions in these regions.

PLAIN LANGUAGE SUMMARY

Extraintestinal Pathogenic Escherichia coli (ExPEC) is a leading cause of invasive disease, including bacteremia and sepsis.

A targeted literature review included the most recent data from 34 published studies on the epidemiology and clinical and economic burden of IED in the elderly/high risk populations in China, Japan, South Korea, Taiwan, and Australia.

ExPEC accounted for 46.0% (1,238/2,692) of bacteria-related invasive diseases in patients aged above 60 years in South Korea, followed by China (44.4% (284/640)), Taiwan (39.0% (1,244/3,194)), and Japan (18.1% (581/3,206)), while Australia reported ExPEC out of all pathogens (54.7% (4,006/7,330)) in general adults. Studies reported increases in length-of-stay and in-hospital 30-day all-cause between 9% to 12%. These factors, along with antimicrobial resistance observed in the elderly, call for preventive and curative actions in these regions.

Data for costs associated with ExPEC induced BSI or sepsis in this region are limited, but evidence shows increasing expenditures.

Keywords:

• Asia Pacific
• Escherichia coli
• invasive ExPEC disease
• targeted literature review

JEL classification codes:

• I10
• I1
• I
• I12

Introduction

Escherichia coli (E. coli) is the most common gram-negative species causing extraintestinal infection in the community, as well as in the ambulatory, long-term care, and hospital settings. E. coli strains of biological significance to humans are broadly categorized as commensal, intestinal pathogenic (InPEC), and extraintestinal pathogenic E. coli (ExPEC)¹.

ExPEC strains are E. coli that can cause urinary tract, bloodstream, prostate, and other type of infections at non-intestinal sites and are responsible for the majority of human extraintestinal infections globally². ExPEC is the most common cause of bacteremia, which primarily affects the elderly, and is a frequent cause of meningitis in neonates. Global morbidity and mortality rates due to diseases induced by ExPEC infections are significant and increasing over time³.

The clinical management of E. coli infections is challenging due to the emergence of multi-drug resistance, which frequently leads to treatment failure, increased rates of hospitalization, length-of-stay, morbidity, mortality, and healthcare and societal costs³. In a recent review of the global burden of bacterial antimicrobial resistance (AMR), E. coli was found to be the leading pathogen for deaths associated with AMR⁴. Other studies found that antimicrobial-resistant ExPEC infections led to higher hospitalization rates and infectious complications after prostate biopsy, and ExPEC was identified as a risk factor for poor outcome in patients with bacteremia^5,6. Consequently, there is an urgent need for effective preventive measures, such as vaccination, to reduce the global burden of invasive disease caused by ExPEC.

This targeted literature review aims to synthesize evidence on the burden and epidemiology of ExPEC infections in the elderly population across five countries in the Asia Pacific region, namely Taiwan, China, Japan, South Korea, and Australia.

Methods

Study question

The study aimed to determine the epidemiological and medical burden of ExPEC infection in older adults (≥60 years) in China, Japan, South Korea, Taiwan, and Australia. Given that ExPEC infection is pervasive across all age groups and studies normally include patients with a mixture of ages, this review will also include those studies in which the mean/median age of the included patients is around 60 years old. The research was framed according to the PICOS framework (Population, Intervention, Comparator, Outcomes and Study type) outlined in Table 1.

Table 1. PICOS Framework for ExPEC infections^a.

PICOS	Description
Population	Older patients (≥60 years) with E. coli bacteremia or invasive extra-intestinal pathogenic disease from different E. coli serotypes
	E. coli high risk older patients (≥60 years), inclusive of:
	History of UTI
	Cardiovascular diseases
	Renal dialysis
Intervention	No restriction
Comparator	No restriction
Outcomes	Epidemiology:
	Mortality
	Incidence of sepsis, urosepsis, septic shock
	Prevalence of sepsis, urosepsis, septic shock
	Incidence of E. coli (O serotypes) bacteremia
	Prevalence of E. coli (O serotypes) bacteremia
	Clinical outcomes:
	Hospitalizations and LOS
	ICU incidence and LOS
	Incidence and duration of complications
	Duration of symptoms
	Severity of Invasive Extra-Intestinal Pathogenic E. Coli disease (IED)
	Economic outcomes:
	Cost of Invasive Extra-Intestinal Pathogenic E. Coli disease (IED)
Study type	Observational studies
	Surveillance studies

^aRegion scope: China, Japan, South Korea, Taiwan, and Australia.

Study selection and analysis

A structured search was conducted for English language articles published from January 1, 2010 to October 7, 2020 in Medline and Embase databases (via EMBASE.com). Moreover, the following databases were also searched: Chinese language databases Wanfang and CNKI; the Korean databases Koreamed and kmbase; and the Japanese database Ichushi-Web. Furthermore, supplementary hand searching in Google Scholar was conducted, and review articles were thoroughly cross-referenced to identify additional relevant articles. Finally, databases and reports linked to sentinel surveillance systems for ExPEC and nosocomial infection in each of the regions were searched for data on the incidence of ExPEC infection in the target population.

ExPEC related search strings included but were not limited to “Escherichia coli”, “E. coli”, “extraintestinal pathogenic E.coli”, “bacteremia”, “bacteraemia”, “sepsis”, and “urosepsis”. The outcomes of interest included “incidence”, “mortality”, “prevalence”, “hospitalization”, “survival”, “incidence rate”, “death”, “mechanical ventilation”, “high risk patient”, and “transmission”. The full version of search terms used for Embase is presented in Table 2. Study selection was done based on predefined selection criteria following the PICOS framework. Titles and abstracts of citations were screened and quality checked by two reviewers independently. Those included at the title and abstract screening stage were full text reviewed and further assessed for eligibility.

Table 2. Search strategy and preliminary number of studies in Embase.

PICOS Element	No.	Search terms	Hits
Population	#1	(“escherichia coli”/exp OR “e coli”:ti,ab OR “e. coli”:ti,ab OR “e.coli”:ti,ab OR “enterococcus coli”:ti,ab OR “escherichia coli”:ti,ab OR “extraintestinal pathogenic escherichia coli”/exp OR “extra intestinal pathogenic e. coli”:ti,ab OR “extraintestinal pathogenic e. coli”:ti,ab OR “extraintestinal pathogenic escherichia coli”:ti,ab) AND (“bacteremia”/de OR “bacteremia”:ti,ab OR “bacteremia”:ti,ab OR “bacteriemia”:ti,ab OR “sepsis”/de OR sepsis:ti,ab OR urosepsis:ti,ab OR “urosepsis”/de)	15,076
Outcomes	#2	“incidence”/exp OR “incidence”:ti,ab OR “incidence rate”:ti,ab OR “rate, incidence”:ti,ab OR “prevalence”/exp OR “prevalence”:ti,ab OR “prevalence study”:ti,ab OR hospitalis*:ti,ab OR hospitaliz*:ti,ab OR “mechanical ventila*”:ti,ab OR icu:ti,ab OR “intensive care”:ti,ab OR mortality:ti,ab OR death:ti,ab OR prognos*:ti,ab OR survival:ti,ab OR epidemiolog*:ti,ab OR “immun*”:ti,ab OR “transmission”:ti,ab OR “high risk population”/de OR “high risk patient”/de OR “high risk”:ti,ab OR “cost”/de OR cost:ti OR “cost control”/de OR “health care utilization”/de OR “economic aspect”/de OR “health economics”/de OR “economic burden”/de OR “economic burden”:ti,ab OR “cost of illness”/de	9,034,251
Combining search terms for Population (E.coli) and outcomes	#3	#1 AND #2	9,166
Region limit (Australia)	#4	“australia”/exp OR australia:ti,ab,ad,ff	993,606
Region limit (China)	#5	“china”/exp OR china:ti,ab,ad,ff	2,187,321
Region limit (Japan)	#6	“japan”/exp OR japan:ti,ab,ad,ff	2,077,633
Region limit (Korea)	#7	“korea”/exp OR korea:ti,ab,ad,ff	570,869
Region limits (Taiwan)	#8	“taiwan”/exp OR “taiwan”:ti,ab,ad,ff	311,626
Combining the search to include all 5 regions	#9	#4 OR #5 OR #6 OR #7 OR #8	5,915,929
Combining search terms for population (E. coli) and epidemiological burden, limiting search results to regions of interest.	#10	#3 AND #9	1,391
Limiting articles to those on patients in age ranges of interest	#11	(“infant”/exp OR “newborn”/exp OR “child”/exp OR “preschool child”/exp OR “school child”/exp OR “adolescent”/exp OR child:ti OR infant:ti OR infants:ti OR children:ti) NOT (“adult”/exp OR “young adult”/exp OR “middle aged”/exp OR “aged”/exp OR “very elderly”/exp)	2,522,281
Non-humans	#12	(“animal”/exp OR “nonhuman”/exp OR “human tissue”/exp OR “animal experiment”/exp OR “animal model”/exp OR “in vitro study”/exp OR “animal tissue”/exp) NOT “human”/exp	7,488,992
Removing irrelevant populations	#13	#10 NOT #11	1,188
Removing non humans	#14	#13 NOT #12	894
Limiting to full-text articles on patients in age ranges of interest	#15	#14 AND (“Article”/it OR “Article in Press”/it)	669

The geographical focus of the review was on Taiwan, China, South Korea, Japan, and Australia. Observational studies and surveillance reports conducted in elderly patients (age ≥60) or patients with underlying conditions that reported the outcomes of interest were included. Case reports, narrative reviews, commentaries, modelling, and review articles were excluded. Included studies were narratively summarized and results descriptively analyzed. No formal quantitative statistical assessment was done as part of this research. Our study did not involve research on human subjects and/or any patient or cohort level data. We utilized only evidence from already published studies. As such, it was deemed unnecessary to seek approval from an institutional review board.

Results

Study selection

The results of the search are summarized in the simplified PRISMA diagram shown in Figure 1. The initial database search identified a total of 1,047 records, while an additional 24 articles were identified via supplementary manual searching. Following screening activities, a total of 34 studies were selected for data extraction^7–40, including two studies focused on economic burden due to ExPEC-associated BSI^39,40.

Figure 1. Simplified PRISMA flow diagram for ExPEC.

Characteristics of included studies

An overview of the studies reporting the prevalence of invasive diseases caused by ExPEC in the regions of interest is shown in Tables 3 and 4. BSI/bacteremia and sepsis are the most reported invasive diseases caused by E. coli, i.e. in 30/32 studies^{7–32,35–38}, while the remaining two studies focused on acute pyelonephritis^33,34. Studies were predominantly hospital-based, and prevalence rate was commonly reported as a proportion of the number of hospitalized patients with ExPEC-associated BSI over the number of patients with confirmed gram-negative pathogens.

Table 3. Prevalence of extra-intestinal pathogenic E. coli in China, Japan, South Korea, Taiwan, and Australia.

Source	Condition	Data period	Age	Pathogen Sample size (n)	E. coli (n)	Proportion of E. coli in all pathogens
China
Quan et al.^7	Bloodstream infection	2013–2014	≥65	919	284	30.9%
Liu et al.^27	Bloodstream infection	2016–2017	49.59 ± 16.35	735^a	329^b	44.7%
Bai et al.^9	Bloodstream infection	2013–2018	59.08 ± 12.30	157	32	20.38%
Feng et al.^10	Bloodstream infection	2012–2018	All ages	1,043	NR	34.23%
Tian et al.^11	Bloodstream infection	1998–2017	All ages	3,878	NR	2013–2017: 40.57%
						2008–2012: 53.24%
						2003–2007: 46.46%
						1998–2002: 34.75%
Jiang et al.^12	Urosepsis	2012–2017	58 (17–76)	Mild: 38^a	Mild: 19^a	Mild: 47.4%
				Severe: 56^a	Severe: 23^a	Severe: 41.1%
Japan
Kosai et al.^15	Bacteremia	2016	73 ± 16	438	247	56.39%
Mitsuboshi et al.^16	Bacteremia	2012–2016	≥65	179	ESBL-EC: 152	84.92%
Miyazato et al.^17	Bacteremia	2009–2015	74.3 ± 16.7	241	106^b	44%
Nagao^18	Bloodstream infection	2008–2012	64.4 (2–84)	2,941	595	20.23%
Takeshita et al.^19	Bloodstream infection	2012–2013	68.2 ± 20.1	CA: 764	CA: 194	CA: 25.4%
			70.4 ± 16.0	CHA: 551	CHA: 135	CHA: 24.5%
			64.6 ± 19.2	HA: 1,891	HA: 252	HA: 13.3%
Hattori et al.^20	Bloodstream infection	2012–2016	≥65	2,105	434	20.6%
South Korea
Park et al.^25	Sepsis	2005–2009	65.0 ± 14.2	1,192	263	22.1%
Son et al.^26	Bloodstream infection	2006–2007	53.7 ± 20.7	CA: 380	CA: 179	CA: 47.1%
				HA: 558	HA: 83	HA: 14.9%
				HCA: 206	HCA: 56	HCA: 27.2%
Liu et al.^27	Bloodstream infection	2017	≥65	2,692	1,238	45.99%
Lee et al.^36	Bloodstream infection	2016–2017	All ages	3,523	1,536	43.6%
Taiwan
Hsieh et al.^35	Bloodstream infection	2008–2013	>61	3,731	1,453^b	38.95%
Lee et al.^36	Bacteremia	2008–2013	74.8 ± 13.5	ESBL-EC: 65	ESBL-EC: 48	ESBL-EC: 73.8%
			68.2 ± 15.3	Non-ESBL-EC: 1,076	Non-ESBL-EC: 778	Non-ESBL-EC: 72.3%
Lee et al.^34	Urosepsis	2002–2007	Mean 69.4	CA: 11	CA: 9	CA: 81.8%
				CHA: 41	CHA: 25	CHA: 60.9%
				HA: 41	HA: 25	HA: 60.9%
Australia
Australian Commission on Safety and Quality in Health Care^38	Sepsis	2015	All ages	CA: 5,204	CA: 3,148	CA: 60.49%
				HA: 1,724	HA: 617	HA: 35.79%

Abbreviations. CA, community associated; HA: healthcare associated; CHA: community acquired healthcare associated.

^aNumber of patients.

^bNot reported, calculated based on proportion.

Age is presented as median (range) or mean ± SD.

Table 4. Hospital stay, ICU frequency/stay, and all-cause mortality rate among patients with BSI in China, Japan, Korea, and Australia.

Source	Condition	Age	Number of Patients	Hospital stays (days)	Frequency of ICU	Days of ICU stay	30-day all-cause mortality
China
Jiang et al.^12	Urosepsis	58 (17–76)	Mild: 38	21 (7–68)	18.42% (7/38)	10.14	1.06% (1/94)
			Severe: 56	36 (3–100)	41.07% (23/56)	19.56
Zhang et al.^13	Bacteremia	59.5 (19–85)	AmpC-EC: 51	29.88 ± 17.9	7.8% (4/51)	NR	25.5% (13/51)
			Non-AmpC-EC: 197	30.98 ± 23.5	13.2% (26/197)		12.2% (24/197)
Xiao et al.^14	Bacteremia	62 (47–73)	ESBL-EC: 298	NR	NR	NR	11.1% (33/298)
			Non-ESBL-EC: 206				9.2% (19/206)
Japan
Nagao^18	Bloodstream infection	64.4 (2–84)	2,941	NR	12.4% (366/2,941)	NR	14.1% (415/2,941)^b
Komatsu et al.^21	Bacteremia	72.3 ± 10.8	ESBL-EC: 30	NR	15.65% (18/115)	NR	20% (6/30)
			Non-ESBL-EC: 185				4.7% (4/85)
Hattori et al.^20	Bloodstream infection	≥65	2,105	46.3 ± 56.0	NR	NR	15.15% (319/2,105)^c
Tsuzuki et al.^22	Bloodstream infection	All ages	NA	NA	NA	NA	(See footnote)^a
Haruki et al.^23	Bacteremia	78 (67–84)	ESBL-EC: 24	NR	NR	NR	37.5% (9/24)
			Non-ESBL-EC: 77				15.6% (12/77)
Namikawa et al.^24	Bacteremia	62.5 ± 18.9	ESBL-EC: 31	NR	NR	NR	9.7% (3/31)
		67.6 ± 13.9	Non-ESBL-EC: 98				9.2% (9/98)
South Korea
Park et al.^25	Sepsis	65.0 ± 14.2	1,192	26.5 ± 34.5	(All in ICU)	26.5 ± 34.5	23.0% (274/1,192)^d
Cho et al.^29	Bloodstream infection	67.65 ± 13.99	ESBL-EC-sequence type 131: 34	NR	NR	NR	17.6% (6/34)
		61.57 ± 15.34	ESBL-EC-sequence type non-131: 76				15.8% (12/76)
Yoon et al.^30	Bloodstream infection	69.1 ± 15.7	1492	NR	10.8% 161/1,492)	NR	9.5% (141/1,492)
Ku et al.^31	Bacteremia	72.26 ± 5.81	ESBL-EC: 120	NR	NR	NR	20% (24/120)
Ha et al.^32	Bacteremia & Cancer	55.88 ± 13.39	ESBL-EC: 95	NR	NR	NR	22.1% (21/95)
		59.55 ± 14.54	Non-ESBL-EC: 255				12.2% (31/255)
Kang et al.^33	Acute	All ages	Age ≥65 years: 71	7 (4–11)	NR	NR	2.3% (2/71)
	pyelonephritis		Age <65 years: 88	5 (1–8)			4.2% (3/88)
Lee et al.^34	Acute	68.5 ± 13.9	Septic shock: 54	NR	NR	NR	25.9% (14/54)
	pyelonephritis	67.1 ± 15.2	Non-Septic shock: 154				0% (0/154)
Taiwan
Lee et al.^36	Bacteremia	74.7 ± 13.5	ESBL-EC: 60	16.0 ± 9.7	NR	3.3 ± 8.1	36.7% (22/60)^e
		70.5 ± 14.2	Non-ESBL-EC: 120	14.5 ± 18.0		1.4 ± 4.1	21.7% (26/120)^e
Australia
Australian Commission on Safety and Quality in Health Care^38	Sepsis	All ages	CA: 5,204	NR	NR	NR	CA: 60.49% (3,148/5,204)
			HA: 1,724				HA: 35.79% (617/1,724)

Age and days of hospital/ICU stays are presented as Mean ± SD or Median (interval).

^aNumber of E. coli induced BSI attributable deaths in each year 2011: 9044 (7,101–11,335). 2012: 9650 (7,585–12,080). 2013: 10,896 (8,594–13,589). 2014: 11,621 (9,178–14,471). 2015: 12,587 (9,991–15,595). 2016: 13,356 (10,612–16,532). 2017: 14,016 (11,140–17,344).

^bE. coli associated deaths, with ICU mortality:30.0%, Non-ICU mortality:12.2%.

^cE. coli associated 30-day mortality: 6.45% (28/434).

^dIn-hospital mortality: 28.0% (334/1192), Sepsis-related mortality: 23.6% (281/1,192).

^eSepsis-related mortality: ESBL-EC: 33.3% (20/60), Non-ESBL-EC: 19.2% (23/120), p = 0.04.

Abbreviations. CA, community associated sepsis; HA, healthcare associated sepsis.

Epidemiology of invasive diseases caused by ExPEC

The highest proportion of elderly patients with invasive diseases (BSI/bacteremia or sepsis) caused by E. coli were reported in studies from South Korea (up to 46.0% (1,238/2,692))²⁷, followed by China (44.4% (284/640))⁷, Taiwan (39.0% (1,244/3,194))³⁵, and Japan (18.1% (581/3,206))¹⁹. In Australia, such data was only available for the general adult population, being 54.7% (4,006/7,330)³⁸.

South Korea

Four studies focusing on South Korea reported sepsis/BSIs due to E. coli, including two derived from the Korean Global AMR Surveillance System (Kor-GLASS)^27,28. Both studies, Lee et al.²⁸ and Liu et al.²⁷, examined blood samples collected from patients in sentinel hospitals and identified E. coli as the leading cause of infection, with a prevalence of 43.6% (1,536/3,523) and 42.9% (1,772/4,129), respectively. In Liu et al.²⁷, where the number of patients with bacteremia was stratified by age and specific pathogens, E. coli infection peaked among patients over 85 years old (51% (229/447)).

In the other two Korean studies, the prevalence of E. coli among cases of invasive disease differed due to differences in the underlying population^25,26. Park et al.²⁵ included 1,192 adult patients admitted into the intensive care unit (ICU) due to sepsis from 2005 to 2009, from which E. coli was detected in 22.1% (mean age = 65 years). Among the 422 patients who developed bacteremia, E. coli was present in 37.7% of them. A separate study by Son et al.²⁶ analyzed 380 community-acquired (mean age = 57.0 years), 558 hospital-acquired (mean age = 47.3 years), and 206 healthcare-associated cases of bacteremia (mean age = 53.7 years), and identified E. coli as the leading cause in the first (47.1%) and third (27.2%) group.

Taiwan

The incidence rate of ExPEC infections at a regional level in Taiwan was available from a study by Hsieh et al.³⁵. The authors examined 104,641 blood cultures from adult patients with BSIs from a regional hospital in northern Taiwan during 2008–2013. E. coli was the most commonly identified organism (39.0% (1,244/3,194)) in bacteremia cases in patients aged over 61 years³⁵. In another study by Lee et al.³⁶, the authors included adult patients (n = 1,141, mean age >68) with bacteremia caused by E. coli, Klebsiella species or Proteus mirabilis, from which 72.4% were found to be caused by E. coli. This high prevalence could be biased due to the exclusion of patients infected with other pathogens different than the above-mentioned.

China

In China, a multicenter study by Quan et al.⁷ examined a total of 919 consecutive episodes of community-onset bloodstream infections (COBSIs) from 28 tertiary hospitals across the country. E. coli was identified in 69.64% (640/919) of cases, and 44.4% (284/640) were from patients over 65 years old. It should be noted that this was the rate of elderly patients among all age groups with bloodstream infections, specifically caused by E. coli and K. pneumoniae. Additionally, studies reporting regional level data showed that the prevalence of E.coli among cases of BSI varied from 20% to 53% across different provinces or cities, including Hunan, Beijing, Lanzhou, and Wuhan^8–12.

Japan

The proportion of hospitalized patients with E. coli BSI in Japan was similar to other regions in three multicenter studies^18–20. Takeshita et al.¹⁹ reported that 18.1% (581/3,206) of hospitalized elderly patients with BSI had E. coli as the causative agent. The prevalence of E. coli was slightly higher (20.23%) in Nagao¹⁸, which included 2,941 hospitalized patients with BSI. Similarly, a single center study of 2,105 hospital patients with BSIs by Hattori et al.²⁰, reported that 20.6% of the infections were due to E. coli. In contrast, higher prevalence of E. coli was found in three studies with smaller sample size (n = 179–438). In these studies, E. coli was identified in 44–85% of patients with gram-negative bacteremia or extended-spectrum β-lactamase (ESBL) producing bacteremia^15–17.

Australia

In Australia, epidemiological data came from the Sepsis Outcome Programs 2015 Report by the Australian Commission on Safety and Quality in Health Care. The document included common prevalence of E. coli in BSI patients across 32 institutions, and in each state and territory of Australia. E. coli was the most common organism causing bacteremia in the country, accounting for 54.7% (4,006/7,330) of all BSI episodes reported³⁸.

Clinical burden of ExPEC in Asia Pacific

Patient comorbidities

Comorbidities including cancer, diabetes, and kidney diseases were frequently found in patients with BSI from all regions of study except for Australia. Cancer was present in 15–40% (median across studies: 33.74%) of the patients across all studies, between 5% and 26% (median across studies: 15%) had kidney diseases, while 3% to 42% (median across studies: 20.51%) were found to have diabetes.

When studies from individual regions were assessed, in the Quan et al.⁷ study from China, diabetes was the most common underlying disease in 27.8% (178/640) of patients with BSI, followed by urinary system disease (e.g. obstructive urinary tract disease, renal failure) in around 13% (82/640), chronic kidney disease (7.97%), and autoimmune disease (5.78%).

In Japan, the presence of a solid tumor was found in 31–40% of patients with BSI according to three large sample studies included in this review, Takeshita et al. (n = 3,206), Nagao (n = 2,941), and Hattori et al. (n = 2,105)^18–20.

With regards to South Korea, 71.1% of the patients with BSI in the study by Park et al.²⁵ had at least one type of comorbidity (i.e. cardiovascular disease = 37.85%, metabolic disorder = 28.1%, cancer = 14.68%). On the other hand, solid tumor (33.74%) and diabetes (18%) were the most common underlying disease in Son et al.²⁶.

Finally, in the Lee et al.³⁶ study from Taiwan, hypertension (49.4%, 564/1,141), diabetes mellitus (38.9%, 444/1,141), and malignancy (26.9%, 307/1,141) were the most common comorbidities in patients with bacteremia. Similar to E. coli prevalence, the high rate of comorbidities in this study could be biased due to the selection of patients.

Hospitalizations and ICU frequency

Apart from reporting the burden of E. coli BSI, several studies included hospitalization rates, ICU care, and in-hospital mortality associated with this disease (Table 4).

The longest mean hospital stay (mean = 46.3 days, SD = 56.0) was seen in the Japanese study by Hattori et al.²⁰, although this number was based on analysis of a mixture of patients infected with a variety of gram-negative bacteria and not exclusively with E. coli. In comparison, Chinese and Korean studies showed shorter hospital stays (means <31 days across studies). But the shortest hospitalization time was observed in Taiwan, with a mean length of 15 days in patients with BSI due to E. coli. Interestingly, Kang et al.³³ found that patients aged ≥65 years had significantly longer hospital stays than those under this age (median 7 days vs. 5 days, p = 0.009), and diabetes mellitus was significantly associated with a longer length-of-stay (OR = 5.34, 95% CI = 1.76–16.2, p = 0.003).

Information on length-of-stay in Australian patients with E. coli induced BSI was limited. In this regard, Wozniak et al.⁴⁰ conducted a population-based modelling study to investigate the health and economic burden of antimicrobial-resistant infections in Australian hospitals. They found that BSI caused by ceftriaxone-resistant E. coli led to an additional 4.89 days (95% UI = 1.1–8.7) of hospital stay compared to patients susceptible to E. coli infections.

With regards to ICU admission in patients with BSI, several factors such as severity of infection, underlying diseases, and older age could influence its rate. For example, severe urosepsis was associated with a significantly higher rate of ICU admission compared to non-urosepsis hospitalizations (41.07% vs. 18.42, p = 0.0208) in the study by Jiang et al.⁴¹, although this condition did not impact the length of stay as the days of ICU stay were not significantly different (p = 0.8051).

30-Day mortality

E. coli associated BSI has been shown to lead to considerable mortality, with some evidence suggesting that it has increased in recent years (Table 4)²². The highest 30-day mortality in patients suffering from E. coli BSI was reported in Taiwan (19.2–33.3%)³⁶, followed by China (12.2–25.5%)¹³, South Korea (23.6%)²⁵, Australia (8.4–21.3%)³⁸, and Japan (6.45%)²⁰. However, this number in Japan is on the rise according to Tsuzuki et al.²², in which a 40% increase in BSI attributable deaths caused by E. coli was found during 2011–2017, from 7.1 per 100,000 population to 11.1 per 100,000 population (p < 0.001). Finally, the case-fatality rate further increased when only patients with E. coli BSIs at the ICU were considered, reaching 30% according to another Japanese study by Nagao¹⁸.

When 30-day mortality was compared between hospital-onset and community-onset BSI, the first one was associated with a higher mortality rate than the second one. These results were similar across different regions analyzed, including Australia, Korea, Japan, and Taiwan^19,26,37,38.

When variant-specific results were analyzed, the Chinese study Zhang et al.¹³ found that bacteremia due to AmpC β-lactamase-producing E. coli was associated with a higher 30-day mortality compared with that induced by non-AmpC β-lactamase-producing E. coli (25.5% vs. 12.2%, p = 0.018) in cancer patients. Another Chinese study, Xiao et al.¹⁴, analyzed the risk factors associated with ESBL-E. coli induced BSI, and no significant difference was found in 30-day mortality between the ESBL group and the non-ESBL group (11.1% vs. 9.2%, p = 0.642). In this case, a significant higher risk of mortality was associated to a respiratory tract originated E. coli infection (OR = 2.050; p = 0.031) and prior use of carbapenems (OR = 3.491; p = 0.003). Similar results were found in the Japanese study by Namikawa et al.²⁴, where 30-day mortality did not differ between the two groups (9.7% vs 9.2%). However, two other Japanese studies found that ESBL-E. coli BSI led to higher mortality than non-ESBL-E. coli BSI, but they all suggested that this finding should be further investigated as differences in baseline characteristics between groups could have impacted survival^21,23,24. For example, in Haruki et al.²³, the ESBL group had a significantly higher proportion of patients with chronic renal failure than the non-ESBL group, and further sub-analyses suggested that antibiotic use could have impacted the outcomes. Finally, in South Korea, Ha et al.³² and Cho et al.²⁹ focused specifically on elderly patients with ESBL-E. coli induced BSI and found similar 30-day mortality rates (14.9%, 12.6%, and 10.9%). In contrast, Ku et al.³¹ reported an all-cause 28-day mortality of 20% for the same type of patients.

Additional Korean studies also assessed risk factors associated with mortality. In Yoon et al.³⁰, the leading risk factor for mortality in patients with E. coli BSI (30-day mortality of 9.5%, n = 1,492) was an ICU admission, accounting for 31.9% of these early deaths. Nosocomial origin of the infection was also reported in 36.9% of these patients. In this same line, Park et al.²⁵ reported higher in-hospital mortality (28.0%, 334/1192) and sepsis-related mortality (23.6%, 281/1192) among ICU patients, although this study was not restricted to patients with E. coli. Finally, a higher mortality rate was found in older patients (age >65) with acute pyelonephritis due to E. coli. Kang et al.³³ reported a 14-day mortality of 4.2% vs. 2.3% when comparing older to younger patients. This number increased considerably in a different study, by Lee et al.³⁴, where 25.9% of 54 bacteremic acute pyelonephritis patients with septic shock due to E. coli died within 7 days of admission, although results in this study were not stratified by age and the sample size was limited.

Antimicrobial resistance

Antimicrobial resistance (AMR) in ExPEC infections was found to be a common issue across China, Japan, South Korea, and Taiwan, with a resistance rate across studies of 30% to >90% to most antibiotics such as ampicillin (>65%), piperacillin (>55%), cefazolin (>42%), and cefotaxime (>35%). In contrast, the AMR rate is relatively lower in Australia, with the highest AMR rate to ampicillin being approximately 60% and under 16% to the others.

Resistance rates of E. coli to different antibiotics were considerably high. Resistance to ampicillin and piperacillin was reported in China (90% to both)¹¹, South Korea (65% and 55%, obtained from the Korean governmental antimicrobial resistance AMR surveillance system), and Taiwan (70% for ampicillin)⁴², while Australia showed slightly lower resistance to ampicillin (50.9% for community onset and 61.6% for hospital onset sepsis) and piperacillin–tazobactam (2.1% and 6.7%, for community and hospital onset sepsis, respectively)³⁸. A continuous decrease in sensitivity of E. coli to other antibiotics was also observed in the same regions, as reported in the Chinese study by Tian et al.¹¹, the Korean governmental antimicrobial resistance AMR surveillance system (42% to cefazolin, 40% to ciprofloxacin, 35% to cefotaxime), and the study by Chen et al.⁴² from Taiwan (51% to trimethoprim/sulfamethoxazole, 56% to ampicillin/sulbactam). In contrast, Australia showed lower resistance rates for several antibiotics including ceftriaxone (9.1–15.8%), ciprofloxacin (11.3–15.1%), amoxicillin–clavulanate (7.7–13%), mainly attributable to the conservative use of antibiotics³⁸. This is in line with a Korean analysis that found a positive correlation between consumption of antimicrobials such as fluoroquinolone, cefoxitin, and cefotaxime with the resistance rates of E. coli to these agents⁴³.

When AMR in specific E. coli strains were assessed, these numbers increased significantly in all regions and for most antimicrobials tested. ESBL-E. coli isolates were highly resistant to different antimicrobials (e.g. cefazolin, ceftriaxone, ceftazidime, cefotaxime, cefepime, levofloxacin, ciprofloxacin, trimethoprim/sulfamethoxazole) in China and Japan, with AMR rates of over 63% for the most effective agents and over 90% for the less effective ones¹². In addition, prevalence of ESBLs was very high (80.95%) in the Chinese study by Jiang et al.¹².

Economic burden of ExPEC in the Asia Pacific

Costs associated with ExPEC induced BSI or sepsis were not comprehensively documented across the five regions, with limited data only available for South Korea and Australia. Kim et al.³⁹ estimated that the costs of sepsis in Korea have tripled between 2005 and 2012, growing 311.8% from 4,252 × 100,000,000 KRW to 13,226.5 × 100,000,000 KRW. In Australia, the population-based modeling study conducted by Wozniak et al.⁴⁰ found that an additional AUD $5.8 million (95% UI: $2.2–$11.2 million) per year was spent to treat ceftriaxone-resistant E. coli BSIs.

Discussion

To our knowledge this is the first review of the overall burden of ExPEC infections in the five Asia Pacific countries studied. Findings on the epidemiology of BSIs caused by ExPEC in China, Japan, South Korea, Taiwan, and Australia indicated higher incidence among older patients, which may further increase in the future with aging and growing trends in pathogen resistance. Importantly, our review identified E. coli as the most common organism causing bacteremia among the elderly across several studies in the regions of interest.

These findings are in line with evidence from outside the Asia Pacific region. Older populations were also shown to be more susceptible to ExPEC infections according to surveillance data in the UK in 2018 (150/100,000 in age = 65–74 years and 430/100,000 in age >75 years)⁴⁴. A recent review⁴⁵ on the epidemiology of ExPEC induced BSIs which covered western countries found that, overall, E. coli accounted for 27% of documented bacteremia episodes. Furthermore, the review suggested that ExPEC infection is a major health issue in older adults and at-risk groups, with approximately three quarters of ExPEC BSIs being contracted in the community.

Our review also highlighted the complications and comorbidities accompanied to ExPEC infections, which is reflected in the long hospital and ICU stays as well as the considerable in-hospital and ICU mortality rates. Although not all the studies assessed trends for mortality, evidence from Japan shows a significant increase in E. coli BSI-related mortality which may be under-documented in the other regions. This is expected to some extent as elderly patients suffer in general from higher prevalence of cancer, diabetes, renal, and other issues.

Some of the studies we identified reported issues with antimicrobial resistance (AMR) and the associated consequences. This deserves particular attention as the emergence of AMR in the Enterobacter species has been reported to be associated with significantly (p < 0.001) longer hospitalization and higher treatment cost, as well as a higher risk of death (relative risk = 5.02; p = 0.01) compared with susceptible strains⁴⁶. Furthermore, recent work by Tsuzuki et al. studying the population-level burden of disease associated with AMR BSIs in Japan found that 137.9 (95% CI = 130.7–145.2) disability-adjusted life-years (DALYs) per 100,000 population were attributable to BSIs caused by nine antimicrobial-resistant bacteria in 2018. Of those, methicillin-resistant Staphylococcus aureus (MRSA), fluoroquinolone-resistant E. coli (FQREC), and third-generation cephalosporin-resistant E. coli (3GREC) accounted for 87.2% overall burden, which was particularly high in those aged 65 or older⁴⁷. AMR also places additional burden on the clinical practice by contributing to more infections, undermining the outcome of the treatment, inducing extra expenses, and interrupting the capacity of healthcare infrastructures⁴⁸. A recent study of the economic burden associated with AMR in gram-negative bacteria estimated that the additional costs associated with hospitalizations in Japan were USD $1.1 billion⁴⁹. For all these reasons, managing AMR has become an important component to the achievement of sustainable development as highlighted by the United Nation (UN). The UN introduced a new sustainable development goal (SDG) indicator specifically related to E. coli induced BSIs in March 2020, along with several recommended approaches to cope with AMR such as infection prevention and control programs⁵⁰.

It has been argued that appropriate prevention strategies could help reduce the pressure which ExPEC causes on healthcare systems and individual patient suffering⁴⁵. Such prevention strategies, including vaccinations, have been effective in reducing the burden of other infections in the elderly, such as influenza, pneumococcal infections, and herpes zoster. In addition, vaccination also reduces usage of antibiotics, therefore contributes to preventing expansion of drug-resistant pathogens. A recent Japanese study estimated that reducing AMR levels of gram-negative pathogens including E. coli had the potential of increasing clinical and economic gains⁴⁹. With aging populations in many of the Asia Pacific regions, especially in Japan, where 28% of the population is over 65 years old and the aging trend is still increasing⁵¹, preventing ExPEC infections is likely to have significant implications in public health and the associated economy in the long-run.

Finally, this review has several limitations. Most notably, data were overall limited despite the number of studies identified. Secondly, the considerable heterogeneity in study design and regional medical practice limits comparability across studies. The different selection criteria of patients in the studies deserve special attention when interpreting the findings to avoid overestimation or underestimation of disease burden. Thirdly, E. coli has been classified into three serotypes (O:H:K), with the O-groups being the most common ones (N = 186)⁵². However, none of the included studies reported the serotyping of the E. coli identified.

Conclusion

This review found that ExPEC was a major cause of blood stream infections across China, Japan, South Korea, Taiwan, and Australia. Our findings provide evidence on the significant clinical and economic consequences of such infections which are largely linked to the acute care for patients as well as AMR. With length of hospital stay and ICU stay exceeding 20 days in many of the studies, the average cost per patient would be considerable. In the Asia Pacific region, with an aging population and increasing difficulty to treat ExPEC infections, the burden to the healthcare system is expected only to increase. Considering the diverse characteristics of different serotypes, an in-depth exploration would facilitate the development of vaccination strategies. Further work into the evidence gaps identified in this review in Table 5 and the implications of ExPEC prevention are likely to provide key information for policy makers.

Table 5. Evidence gap analysis for ExPEC in patients aged 60 years and over in China, Japan, South Korea, Taiwan, and Australia.

Transparency

Declaration of financial/other relationships

LHP, KL, MCN: Employees of Janssen. PA, XJ are employees of Amaris Consulting, which received funding from Janssen Asia Pacific during the conduct of the study and has no other funding, financial relationships, or conflict of interest to disclose.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Abbreviations
ExPEC	=	Extraintestinal pathogenic Escherichia coli
IED	=	Invasive ExPEC disease
AUD	=	Australian dollar
UI	=	Uncertainty interval
BSIs	=	Bloodstream infections
E. coli	=	Escherichia coli
InPEC	=	Intestinal pathogenic Escherichia coli
AMR	=	Antimicrobial resistance
PICOS	=	Population, Intervention, Comparator, Outcomes and Study type
ICU	=	Intensive care unit
COBSIs	=	Community-onset bloodstream infections
ESBL	=	Extended-spectrum β-lactamase
OR	=	Odds ratio
KRW	=	Korean Won
DALYs	=	Disability-adjusted life-years
MRSA	=	Methicillin-resistant Staphylococcus aureus
FQREC	=	Fluoroquinolone-resistant E. coli
3GREC	=	Third-generation cephalosporin-resistant E. coli
USD	=	US dollar
UN	=	United Nation
SDG	=	Sustainable development goal
PRISMA	=	Preferred Reporting Items for Systematic Reviews and Meta-Analyses
SD	=	Standard deviation
CI	=	Confidence interval
UK	=	United Kingdom

Acknowledgements

Jun Feng and David Wu, both employees of Janssen, provided valuable review and feedback. Xue Chen of Amaris provided additional support.

Additional information

Funding

Funding was provided by Janssen Asia Pacific.