An overview of diarrheagenic Escherichia coli in Iran: A systematic review and meta-analysis

Abstract

Background:

Diarrheagenic Escherichia coli (DEC) is a common enteric pathogen that causes a wide spectrum of gastrointestinal infections, particularly in developing countries. This is a systematic review and meta-analysis to determine the prevalence of DEC in various geographical regions in Iran.

Materials and Methods:

English (PubMed, Web of Science, Scopus, Embase, Cochrane Library, and Google Scholar) and Persian (IranMedex, SID, Magiran, and Iran Doc) databases were comprehensively searched from January 1990 to April 2017. Study selection and data extraction were performed by two independent reviewers. After assessing heterogeneity among studies, a random effects model was applied to estimate pooled prevalence. Data analyses were done with the Stata software (version 12.0). This meta-analysis was registered with PROSPERO, number CRD42017070411.

Results:

A total of 73 studies with 18068 isolates were eligible for inclusion within the meta-analysis. The results of random effects model showed that the most prevalent DEC pathotypes were enterotoxigenic E. coli (ETEC) (16%; 95% confidence interval [CI]: 11%–23%), enteroaggregative E. coli (11%; 95% CI: 8%–15%), atypical enteropathogenic E. coli (EPEC) (11%; 95% CI: 8%–14%), Shiga toxin-producing E. coli (9%; 95% CI: 6%–13%), diffuse adherent E. coli (6%; 95% CI: 6%–12%), enteroinvasive E. coli (4%; 95% CI: 2%–6%), and typical EPEC (3%; 95% CI: 1%–5%).

Conclusion:

This study showed that DEC infections in the Iranian population have low frequency. Our data suggest that the ETEC pathotype can be regarded as one of the most important etiological agents of diarrhea in this country. However, the prevalence of DEC pathotypes is diverse in different regions of Iran.

INTRODUCTION

Escherichia coli is a ubiquitous commensal inhabitant bacterium present in the gastrointestinal tracts of humans and animals, although some strains have acquired several putative virulence factors that enable it to adapt to new niches and cause a broad spectrum of infections.[1,2] Pathogenic E. coli is one of the major causes of infectious diseases that span from the gastrointestinal tract to extraintestinal sites. Extraintestinal diseases include the urinary tract infection (UTI), septicemia, newborn meningitis, central nervous system, and respiratory system infections.[3,4]

Pathogenic E. coli agents of gastrointestinal infections are significant causes of sporadic outbreaks of diarrhea worldwide, especially in developing countries.[5] Based on virulence factors and phenotypic traits, diarrheagenic E. coli (DEC) strains are generally classified into eight pathotypes: enteropathogenic E. coli (EPEC), enterotoxigenic E. coli (ETEC), enteroinvasive E. coli (EIEC), enterohemorrhagic E. coli (EHEC)/Shiga toxin-producing E. coli (STEC)/verotoxin-producing E. coli (VTEC), enteroaggregative E. coli (EAEC or EAggEC), diffuse adherent E. coli (DAEC), and adherent-invasive E. coli (AIEC).[6,7,8]

The prevalence of pathogenic E. coli which causes gastrointestinal infections has been reported in the literature from different regions of Iran; therefore, it is necessary to conduct a comprehensive analysis. This study presents a systematic review and meta-analysis of the literature on the DEC pathotypes in the Iranian population.

MATERIALS AND METHODS

Search strategy

This study is a systematic review and a meta-analysis study on original research articles that have been published in English or Persian which present the prevalence of DEC in Iran. The databases of PubMed, Web of Science, Scopus, Embase, Cochrane Library, Google Scholar, Iranmedex, SID, and Magiran were searched for studies published from January 1990 to April 2017. The following keywords or medical subject headings in titles or abstracts were used with the help of Boolean operators (“and” or “or”): “diarrheagenic Escherichia coli,” “enterotoxigenic Escherichia coli,” “ETEC,” “enteropathogenic Escherichia coli,” “EPEC,” “enterohemorrhagic Escherichia coli,” “EHEC,” “enteroinvasive Escherichia coli,” “EIEC,” “enteroaggregative Escherichia coli,” “EAEC,” “shiga-toxigenic Escherichia coli,” “shiga toxin-producing Escherichia coli,” “STEC,” “verotoxin-producing Escherichia coli,” “verotoxigenic Escherichia coli,” “VTEC,” “Diffusely Adherent Escherichia coli,” “DAEC,” “Adherent-invasive E. coli,” “AIEC,” and “Iran.” This study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement with the PRISMA 2015 checklist. The planned systematic review and meta-analysis was registered with the International Prospective Register of Systematic Reviews (PROSPERO), registration number CRD42017070411.[9]

Selection studies

To select studies, two authors scrutinized titles and abstracts of all studies returned by the search strategy. Then, they independently evaluated the full text of potentially relevant nonduplicated articles. Any disagreement was resolved by discussion between the two reviewers. When no agreement was reached, a third reviewer was consulted.

Eligibility criteria

The manual revision was conducted on all displayed articles, and the first levels of screening were based on information in the titles and/or abstracts. The final level was applied to the original publications. The included articles were confirmed by bacteriological standard methods for identification of E. coli isolates (e.g., the biochemical API 20E identification system or conventional biochemical tests) and molecular method for the detection of DEC genes (e.g., PCR). Studies were excluded from the analysis that did not focus on the prevalence of DEC from human specimens in various region of Iran.

Data extraction

For all articles, the compiled information contains age (e.g., young children, adults, and the elderly), year of study, year of publication, place of study, the source of samples, number of cases analyzed, and the prevalence of E. coli pathotypes. The isolates were divided into five groups depending on the sample status: diarrheic, UTI, bloody diarrhea, healthy feces, and mixed (diarrheic and healthy). To assess the quality of each included study in the meta-analysis, we used the 22-item STROBE checklist (https://strobestatement.org). We categorized all studies in three categories in which obtained based on total score of checklists: >80% of the total score of checklists yes = high, 60%–80% of total score yes - medium, and <60% yes = low.

Statistical analysis

At the beginning of the analysis, we undertook an initial descriptive analysis for describing the essential characteristics of studies. Then, for each study, the prevalence (the number of current positive cases divided into a total number of the sample) and standard error were calculated. When the estimated prevalence for a study tends toward either 0% or 100%, the variance for that study moves toward zero, and as a result, its weight is overestimated in the meta-analysis. Therefore, we conducted the meta-analysis with prevalence estimates that had been transformed using the double arcsine method.[10] To investigate the influence of each individual study on the overall estimate, a sensitivity analysis was performed. Before the quantitative pooling of the results, heterogeneity among studies was assessed using the Cochran's Q-statistic (P < 0.1 as significant) and the I² index (25%, 50%, and 75% as low, moderate, and high heterogeneity; respectively).[11] When heterogeneity was present, we used a random effects model (DerSimonian–Laird method); otherwise, we applied a fixed effects model (Mantel–Haenszel method) to estimate the pooled prevalence. The point prevalence of pathotypes and its 95% confidence interval (CI) were estimated for pooled effect size and each study. Forest plots illustrated the proportion of E. coli pathotypes, along with 95% CI. We used subgroup analysis to assess sources of heterogeneity using categorical variables as type of DEC, location which study performed, the source of sample and also using meta-regression for sample size of studies, quality of studies and year of study publication. The funnel plot (visual method) and Egger's test (P < 0.1 as significant) were used to evaluate the possibility of publication bias among studies. In the case with significant publication bias, a nonparametric trim and fill method was performed to rectify the bias. All statistical analyses were conducted using Stata MP Software version 14.0 (Stata Corp, College Station, TX, USA) with metaprop command.

RESULTS

Study characteristics

In the initial database searches, 786 potential studies were identified. After the primary screening of titles and abstracts, 254 articles were selected for full-text search. Assessment of title and abstract resulted in the identification of 158 duplicate studies. Among them, 96 full-text articles were retrieved to check the eligibility, of which 73 were included. A total of 73 studies with 86 datasets (18068 isolates) were included in the final review and meta-analysis. Figure 1 presents the PRISMA flowchart, which describes the articles identified from the search strategy. These 86 data consisted of diarrhea samples (53), feces of healthy samples (13), acute diarrhea samples (9), UTI samples (6), bloody diarrhea samples (2), persistent diarrhea samples (1), and mixed (diarrheic and healthy feces) samples (1). The exclusion of articles based on the title and abstract of studies was mainly because of the following reasons: duplicated population groups, the articles were based on case report, assessment of specific methods on DEC pathotypes diagnosis, reported DEC pathotypes from animal and environmental samples, narrative reviews studies with inadequate data and generally, and studies that did not focus on the prevalence of even one of the DEC pathotypes from human samples in Iran. Table 1 presents a summary of the characteristics of the 86 included data. A description of the characteristics of each study, reporting the prevalence of DEC pathotypes in the Iranian population, is presented in “Appendix 1.” Quality of study for five studies was moderate and for 68 studies was high; the results of subgroup analysis showed that the prevalence of DEC in the two groups (moderate and high quality) was statistically not significant (P = 0.13).

Figure 1.

Flowchart of study selection

Table 1.

Subgroup meta-analysis and publication bias for diarrheagenic Escherichia coli pathotypes

Characteristics	Subgroup	Number of study	Isolates positive	Isolates tested	Prevalence of pathotype (95% CI)	Heterogeneity		Publication bias test
						I2 (%)	Q-test	Bias	Egger’s
DEC pathotypes	STEC	38	953	9185	9 (6-13)	95.14	<0.001	2.48	0.04
	ETEC	28	878	5669	16 (11-23)	97.29	<0.001	0.64	0.31
	aEPEC	38	923	9459	11 (8-14)	95.91	<0.001	1.97	0.002
	tEPEC	12	144	4240	3 (1-5)	92.45	<0.001	1.08	0.06
	EAEC	27	815	6526	11 (8-15)	94.91	<0.001	0.54	0.17
	EIEC	17	142	3184	4 (2-6)	88.42	<0.001	0.54	0.46
	DAEC	2	35	401	6 (6-12)	NES	NES	NES	NES
	AIEC	-	-	-	-	-	-	-	-
Type of sample	Diarrhea	53	2398	12,234	19 (8-38)	98.74	<0.001	4.35	0.21
	UTI	6	104	866	7 (4-16)	97.11	<0.001	3.41	0.16
	Stool	13	329	2416	12 (7-21)	97.99	<0.001	2.12	0.77
	Acute Dia	9	959	2296	14 (5-31)	99.56	<0.001	2.57	0.435
Quality of studies	High	68	3692	17,426	11 (9.0, 14.0)	96.97	<0.001	5.11	0.021
	Moderate	5	179	642	19 (14.0, 31.0)	89.77	<0.001	NES	NES
Year of publication	<2010	13	1191	5491	11 (8-14)	98.5	<0.001	4.82	0.055
	2010-2015	50	1912	10,095	3 (1-5)	91.3	<0.001	3.57	0.001
	>2015	10	768	2482	11 (8-15)	84.7	<0.001	6.31	0.054
Location*	Shiraz	7	215	943	30 (0.8-58)	98.32	<0.001	2.21	0.69
	Tabriz	5	321	824	36 (12-65)	90.11	<0.001	7.17	0.02
	Tehran	31	1801	10,501	19 (11-28)	98.45	<0.001	2.45	0.47
	Hamadan	4	181	575	47 (17-78)	81.01	<0.001	3.68	0.02
	Kerman	6	278	1558	33 (12-58)	98.61	<0.001	1.70	0.52
	Zanjan	3	203	920	19 (13-27)	73.70	0.003	2.96	0.74
Sample size	<100	23	589	1445	11 (7-25)	98.17	<0.001	8.13	0.091
	100-500	45	3046	11,940	9 (3-21)	98.7	<0.001	4.02	0.001
	>500	5	236	4683	4 (2-9)	88.7	<0.001	3.0	0.031

*Cities which not included in analysis had not enough sample. E. coli=Escherichia coli; EPEC=Enteropathogenic E. coli; aEPEC=Atypical EPEC; tEPEC=Typical EPEC; ETEC=Enterotoxigenic E. coli; STEC=Shiga toxin-producing E. coli; EAEC=Enteroaggregative E. coli; EIEC=Enteroinvasive E. coli; DAEC=Diffuse adherent E. coli; AIEC=Adherent-invasive E. coli; UTI=Urinary tract infection; DEC=Diarrheagenic E. coli; CI=Confidence interval; NES=No enough sample size

Appendix 1.

Characteristics of studies included in the meta-analysis

Study	Study period	Published time	Age	Location	Source of sample	Number of isolates/samples	Prevalence of DEC pathotype	Study quality
Abbasi et al.[62]	2012	2014a	3-61 years	Shiraz	Diarrhea	52	ETEC=52	20
Abbasi et al.[63]	2012	2014b	1-26 months	Shiraz	Diarrhea	49	aEPEC=13	20
							EHEC=15
Abbasi et al.[64]	2012	2014c	2 months-63 years	Shiraz	Diarrhea	101	EAEC=5	20
Abbasi et al.[25]	2012	2015	1-24 months	Shiraz	Diarrhea	49	EIEC=7	20
Abbasi et al.[27]	2012-2013	2017	2 months-63 years	Shiraz	Diarrhea	101	DAEC=8	20
Abri et al.[65]	-	2015	-	Tabriz	Diarrhea	60	ETEC=10	20
							aEPEC=2
							tEPEC=1
Aein et al.[98]	2011-2012	2014	<5 years	Yasuj	Diarrhea	200	EIEC=16	16
Adeli et al.[104]	2011-2012	2013	-	Lorestan	UTI	100	EHEC=3	16
Ahangarzadeh-Rezaee et al.[22]	2013	2015	<10 years	Tabriz	Diarrhea	194	EAEC=55	21
Akbari et al.[97]	2009	2010	<5 years	Tehran	Diarrhea	39	EIEC=7	18
Akhi et al.[66]	2014	2015	-	Tabriz	Diarrhea, stool	200	EHEC=11	21
							aEPEC=5
Alikhani et al.[21]	-	2006	0-10 years	Tehran, Ilam, Mazandaran	Diarrhea	247	aEPEC=23	22
							tEPEC=35
Alikhani et al.[21]	-	2006	0-10 years	Tehran, Ilam, Mazandaran	Stool	1108	aEPEC=13
							tEPEC=8
Alikhani et al.[67]	2007	2012	Children	Hamadan	Diarrhea	251	ETEC=13	20
							aEPEC=41
							EHEC=7
							EIEC=0
							EAEC=16
Alikhani et al.[68]	2009	2013	12-65 years	Hamadan	Acute diarrhea	40	ETEC=7	20
							aEPEC=19
							EHEC=6
							EAEC=8
Alizade et al.[28]	2010-2012	2014a	<5 years	Kerman	Diarrhea	142	ETEC=52	20
							aEPEC=9
							EHEC=0
							EIEC=2
							EAEC=23
Alizade et al.[38]	2010-2011	2014b	-	Kerman	Diarrhea	96	aEPEC=5	20
							EHEC=0
Alizade et al.[15]	2014	2017	-	Kerman	Stool, HIV	49	ETEC=29	20
							aEPEC=8
							EHEC=13
Alizade et al.[15]	2014	2017	-	Kerman	Stool, thalassemia	68	ETEC=46
							aEPEC=13
							EHEC=23
Alizade et al.[32]	2010	2015	-	Bam	Diarrhea	155	ETEC=52	19
Alizadeh et al.[69]	2003	2007	5 months-90 years	Hamadan	Acute diarrhea	144	ETEC=22	20
							aEPEC=2
							EHEC=15
							EAEC=10
Asadi Karam et al.[40]	-	2010	<5 years	Tehran	Diarrhea	321	aEPEC=17	19
							EHEC=0
Askari Badouei et al.[8]	-	2014	2-10 years	Garmsar	Diarrhea	75	aEPEC=1	21
							EHEC=4
Aslani et al.[13]	-	2008a	-	Central, western, northern	Diarrhea	193	ETEC=14	19
							aEPEC=74
							EHEC=86
							EAEC=19
Aslani et al.[70]	2007-2008	2011	<12 years	Hamadan	Stool	140	EAEC=15	21
Aslani et al.[96]	2005	2008b	Children	Tehran	Diarrhea	70	EHEC=12	20
Aslani et al.[96]	2005	2008b	Children	Tehran	Hemorrhagic colitis	70	EHEC=2
Bafandeh et al.[23]	-	2015	<30->45 years	Zanjan	Diarrhea	350	EAEC=97	21
Bafandeh et al.[23]	-	2015	<30->45 years	Zanjan	Stool	200	EAEC=44
Bagheri et al.[99]	2006-2007	2009	1 month-66 years	Gorgan	Diarrhea	455	EAEC=20	19
Bakhshi et al.[41]	2009-2010	2012	<5 years	Tehran	Acute diarrhea	309	aEPEC=17	21
							tEPEC=11
							EHEC=0
Bakhshi et al.[71]	-	2014	-	Tehran	Diarrhea	120	aEPEC=28	20
Bazzar et al.[100]	-	2011	<5 years	Ardabil	Diarrhea	194	aEPEC=28	19
							tEPEC=2
Bayat et al.[101]	2012-2013	2014	<5 years	Zanjan	Diarrhea	140	aEPEC=19	18
Bonyadian et al.[12]	2007-2008	2010	-	Shahrekord	Diarrhea	58	EHEC=28	20
Bouzari et al.[24]	2000	2007	Children	Tehran	Diarrhea	200	ETEC=22	20
							EHEC=38
							EAEC=40
Bouzari et al.[72]	-	2011	-	Tehran	Diarrhea	500	aEPEC=14	19
							tEPEC=27
Darbandi et al.[43]	2012-2013	2016	-	Khuzestan	Diarrhea	158	ETEC=21	20
							aEPEC=49
							tEPEC=1
							EHEC=50
							EIEC=1
							EAEC=5
Haghi et al.[33]	2011-2012	2014	<5 years	Tabriz	Diarrhea	140	ETEC=65	21
							aEPEC=13
							tEPEC=6
							EHEC=17
							EAEC=36
Haghi et al.[33]	2011-2012	2014	<5 years	Tabriz	Stool	90	ETEC=8
							aEPEC=3
							tEPEC=0
							EHEC=0
							EAEC=7
Hoseinzadeh et al.[73]	2014	2016	-	Sirjan	Diarrhea	110	ETEC=32	19
							EIEC=6
Hosseini Nave et al.[74]	2013-2014	2016	-	Kerman	Diarrhea	620	EIEC=11	20
Jafari et al.[75]	-	2001	Children	Tehran	Diarrhea	1200	EHEC=60	21
Jafari et al.[76]	2004-2005	2008	5-60 years	Tehran	Acute diarrhea	808	ETEC=41	21
							aEPEC=11
							EHEC=52
							EAEC=13
Jafari et al.[77]	2003-2005	2009	<5 years	Tehran	Acute diarrhea	305	ETEC=38	21
							aEPEC=70
							EHEC=105
							EAEC=92
Kalantar et al.[19]	2008	2009	<12 years	Sanandaj	Acute diarrhea	99	ETEC=7	21
							aEPEC=59
							EHEC=11
							EIEC=22
Kargar et al.[78]	2011-2012	2014	<5 years	Yasuj	Diarrhea	300	EHEC=104	20
Kargar et al.[79]	2010	2013	<2 years	Shiraz	Diarrhea	285	ETEC=7	19
Kargar and Homayoon et al.[105]	2006-2007	2009	<5 years	Marvdasht	Diarrhea	89	EHEC=1	20
Kermani et al.[26]	-	2010	<5 years	Tehran	Acute diarrhea	341	ETEC=21	20
							aEPEC=18
							EHEC=25
							EIEC=31
							EAEC=60
Kermani et al.[26]	-	2010	<5 years	Tehran	Persist diarrhea	83	ETEC=5
							aEPEC=5
							EHEC=6
							EIEC=5
							EAEC=15
Khorshidi et al.[80]	2009-2010	2011	<5 years	Kashan	Diarrhea	178	EPEC=51	19
Khoshvaght et al.[81]	2011-2012	2014	0-60 months	Zanjan	Diarrhea	140	EAEC=36	21
Khoshvaght et al.[81]	2011-2012	2014	0-60 months	Zanjan	Stool	90	EAEC=7
Mahmoudi-Aznaveh et al.[42]	2012-2013	2016	<5 years	Tehran	Diarrhea	349	aEPEC=30	21
							tEPEC=1
							EHEC=0
Memariani et al.[82]	2011-2013	2015	<10 years	Tehran	Diarrhea	398	aEPEC=26	21
							tEPEC=16
Miri et al.[83]	2012-2013	2017	-	East-Azerbaijan, Gilan, Zanjan, Kurdistan, Tehran, Hamedan	Stool	145	ETEC=10	19
							aEPEC=17
							tEPEC=1
							EHEC=18
							EIEC=0
							EAEC=0
Mirzarazi et al.[84]	-	2015	-	Isfahan	UTI	138	EIEC=2	19
							EAEC=13
Mirzarazi et al.[84]	-	2015	-	Isfahan	Stool	30	EIEC=0
							EAEC=5
Mirsalehian et al.[107]	2003	2004	-	-	Bloody diarrhea	103	EHEC=3	17
Mohammadzadeh et al.[20]	-	2013	-	Tehran	Diarrhea	130	aEPEC=22	20
							tEPEC=28
							EHEC=0
Mohammadzadeh et al.[85]	2013	2015a	-	Tehran	Diarrhea	140	EIEC=5	20
Mohammadzadeh et al.[86]	2013	2015b	Mean age 38 years	Tehran	Diarrhea	140	ETEC=9	20
Mohammadzadeh et al.[86]	2013	2015b	Mean age 34 years	Tehran	Stool	110	ETEC=1
Mohammadzadeh et al.[106]	-	2016	-	Tehran	Diarrhea	140	EAEC=6	19
Mohammadzadeh et al.[106]	-	2016	-	Tehran	Stool	136	EAEC=0
Motallebi et al.[48]	2009-2010	2011	<5 years	Kashan	Diarrhea	178	aEPEC=51	20
Nakhjavani et al.[88]	2009-2010	2013	<10 years	Tehran	Diarrhea	412	aEPEC=16	21
							tEPEC=7
Navidinia et al.[89]	2008-2009	2012	1-12 years	Tehran	UTI	378	EHEC=9	19
Nazarian et al.[90]	2010-2011	2014	<5 years	Tehran, Kerman, Golestan	Diarrhea	261	ETEC=21	21
Nazemi et al.[91]	2010-2011	2012	-	Tehran	UTI	100	ETEC=2	18
							EHEC=17
Nezarieh et al.[92]	2013-2014	2015	1-10 years	Kerman	Diarrhea	322	EAEC=23	18
Pourakbari et al.[18]	2010-2011	2013	1 month-12 years	Tehran	Acute diarrhea	50	ETEC=21	18
							aEPEC=0
							EHEC=14
							EIEC=0
							EAEC=10
Pourakbari et al.[18]	2010-2011	2013	1 month-12 years	Tehran	Stool	50	ETEC=5
							aEPEC=0
							EHEC=3
							EIEC=0
							EAEC=2
Rajaee et al.[14]	2014-2015	2017	-	Shiraz	Diarrhea	306	EHEC=108	20
Salmani et al.[16]	-	2016	-	Tehran, Sanandaj	Diarrhea	200	ETEC=153	20
							aEPEC=10
							EIEC=13
							EAEC=39
							DAEC=21
Salmani et al.[16]	-	2016	-	Tehran, Sanandaj	UTI	100	ETEC=22
							aEPEC=0
							EIEC=6
							EAEC=23
							DAEC=6
Salmanzadeh-Ahrabi et al.[93]	2003	2005	<5 years	Tehran	Acute diarrhea	200	ETEC=31	16
							aEPEC=12
							EHEC=30
							EAEC=48
Salmanzadeh-Ahrabi et al.[93]	2003	2005	<5 years	Tehran	Stool	200	ETEC=0
							aEPEC=10
							EHEC=4
							EAEC=16
Shams et al.[94]	2008-2009	2013	<14 years	Tehran	Diarrhea	947	EHEC=27	19
Sharifi Yazdi et al.[95]	2008	2011	<5 years	Tehran	Diarrhea	39	aEPEC=20	15
							EHEC=9
							EIEC=7
Soleimanifard et al.[103]	-	2015	-	Saveh	UTI	50	EAEC=1	20
Soltan Dallal et al.[102]	2013	2015	<5 years	Tehran	Diarrhea	300	aEPEC=36	20
Zeighami et al.[17]	2011-2012	2014	<5 years	Tabriz	Diarrhea	140	ETEC=65	20
							EHEC=17

EHEC=Enterohemorrhagic E. coli; E. coli=Escherichia coli; EPEC=Enteropathogenic E. coli; aEPEC=Atypical EPEC; tEPEC=Typical EPEC; ETEC=Enterotoxigenic E. coli; STEC=Shiga toxin-producing E. coli; EAEC=Enteroaggregative E. coli; EIEC=Enteroinvasive E. coli; DAEC=Diffuse adherent E. coli; UTI=Urinary tract infection; DEC=Diarrheagenic E. coli

Sensitivity analysis

At first, the sensitivity analysis was conducted to evaluate whether individual studies influenced pooled prevalence. The result indicated that no study substantially influenced the pooled prevalence of pathotypes of DEC, which indicates that all studies need to be included for further research.

Heterogeneity

We performed a fixed effects model to estimate the pooled prevalence of pathotypes of DEC, 17.0% (95% CI: 16.9%–18.0%; I² = 98.9, P < 0.001). The results showed that there was substantial statistical heterogeneity among the effect size of studies. To assess the causes of the heterogeneity, five subgroup analyses were undertaken. The results showed that characteristics such as sample size, quality of studies, and year of publication are possible sources of heterogeneity [Table 1]; however, after using these variables in multiple meta-regression, the results indicated that sample size (P = 0.031) can be one of the causes of heterogeneity [Table 2]; so that, studies with small sample size were affected than studies with large sample [Figure 2].

Table 2.

Results of multiple meta-regression to assess the source of heterogeneity

Variables	Coefficient	SE	t	P
Sample size	−0.0004	0.0014	−2.94	0.004
Quality of studies	−0.002	0.003	−0.70	0.484
Year of publication	−0.006	0.003	−0.62	0.540
Location	−0.002	0.006	−0.33	0.745

SE=Standard error

Figure 2.

Relationship between sample size and effect size for included studies

Shiga toxin-producing Escherichia coli

The presence of STEC was examined in 9185 isolates in 38 publications. Overall, 953 of 9185 E. coli isolates were positive for the STEC pathotype. According to a random effects model (I² = 95.14, P < 0.001), the pooled prevalence was estimated to be 9% (95% CI: 6%–13%) [Table 1]. The prevalence of STEC isolated from diarrheic samples was 21/43 (48.8%), from acute diarrhea samples was 9/43 (20.9%), from healthy stool samples was 6/43 (13.9%), from UTI samples was 3/43 (6.9%), from bloody diarrhea samples was 2/43 (4.6%), and from persistent diarrhea and mixed samples each one was 1/43 (2.3%). Based on previous reports from various regions in Iran, a high prevalence of the STEC pathotype of E. coli was detected in Shahrekord (central; 48.3%);[12] central, western, and northern (44.5%);[13] and Shiraz (south; 35.3%)[14] [Appendix 1 and Figure 3].

Figure 3.

Forest plot of prevalence for seven pathotypes of diarrheagenic Escherichia coli in Iran. Diamonds indicate the 95% confidence interval for each pathotypes

Enterotoxigenic Escherichia coli

The presence of ETEC was investigated in 5669 isolates in 28 publications. A random effects model (I² = 97.21, P < 0.001) showed that the pooled prevalence of ETEC was 16% (95% CI: 11%–23%) [Table 1]. The prevalence of ETEC in dependent of diarrheic sample status was 17/34 (50%) with 8/34 (23.5%) in acute diarrhea samples, 7/34 (20.6%) in healthy samples, 2/34 (5.9%) in UTI samples, and 1/34 (2.9%) in the persistent diarrhea sample. The highest report of the ETEC was 64.1%,[15] 58.3%,[16] and 46.4%[17] of E. coli isolates from Kerman (south-east), Tehran and Sanandaj (capital and west), and Tabriz (north-west), respectively [Appendix 1 and Figure 3].

Enteropathogenic Escherichia coli

The presence of atypical EPEC (aEPEC) was investigated in 9459 isolates in 38 publications. In total, the pooled prevalence of aEPEC (923 out of 9459) based on a random effects model (I² = 95.9, P < 0.001) was estimated to be 11% (95% CI: 8%–14%) [Table 1]. aEPECs were found more often in diarrheic (25/44; 56.8%) samples than in acute diarrhea (10/44; 22.7%), feces of healthy (6/44; 13.6%), UTI (1/44; 2.3%), mixed (1/44; 2.3%), and persistent diarrhea (1/44; 2.3%) samples. The prevalence of aEPEC ranged from 0.00% among E. coli isolated from children (Tehran, capital)[18] to 59.6% among E. coli isolated from acute diarrhea (Sanandaj, west).[19] Egger's test and funnel plot showed a significant publication bias for STEC (bias = 2.48, P = 0.04) and aEPEC (bias = 1.97, P = 0.002), but for other pathotypes, there was no publication bias [Appendix 1 and Figure 3].

The presence of typical EPEC (tEPEC) was examined in 4240 isolates in 12 publications. The results of a random effects mode showed that (I² = 92.45, P < 0.001) the pooled effect prevalence of tEPEC was estimated to be 3% (95% CI: 1%–5%) from 144 E. coli isolates which were positive for tEPEC pathotype [Table 1]. This pathotype was identified in diarrheic (10/14; 71.4%) samples than in the feces of healthy (3/14; 21.4%) and acute diarrhea (1/14; 7.1%) samples. The percentage of tEPEC is higher in Tehran (central; 21.5%)[20] and Tehran, Ilam, and Mazandaran (central, west, and north of Iran; 14.2%)[21] [Appendix 1 and Figure 3].

Enteroaggregative Escherichia coli

The presence of EAEC was investigated in 6526 isolates in 27 publications. The EAEC was present in 815 isolates. Based on a random effects model (I² = 92.45, P < 0.001), the pooled estimate of prevalence was obtained 11% (95% CI: 8%–15%) [Table 1]. The prevalence of EAEC in diarrheic samples was 42.8% (15/35), in healthy stool was 25.7% (9/35), in acute diarrhea was 20% (7/35), in UTI was 8.6% (3/35), and in persistent diarrhea was 2.8% (1/35). The percentages of EAEC vary with the geographical location of the patients: 28.3% in Tabriz,[22] 25.6% in Zanjan,[23] and 20% in Tehran[24] [Appendix 1 and Figure 3].

Enteroinvasive Escherichia coli

The presence of EIEC was examined in 3184 isolates in 17 publications. A random effects model showed that the pooled prevalence of EIEC was 4% (95% CI: 2%–6%) from 142 E. coli isolates which were positive for the EIEC pathotype [Table 1]. The prevalence of the EIEC pathotype isolated from diarrheic samples was 12/21 (57.1%), and it was 3/21 (14.3%) from acute diarrhea samples, 3/21 (14.3%) from healthy stool samples, 2/21 (9.5%) from UTI samples, and 1/21 (4.8%) from persistent diarrhea. EIEC was found in 22.2%, 14.3%, and 8.5% of the isolates in Sanandaj (west),[19] Shiraz (south),[25] and Tehran (central),[26] respectively [Appendix 1 and Figure 3].

Diffusely adherent Escherichia coli

The presence of DAEC was evaluated in 401 isolates in two publications. To calculation of heterogeneity, there were not enough studies (n = 2). According to a fixed effects model, the pooled prevalence of DAEC was estimated at 6% (95% CI: 6%–12%) based on 35 isolates [Table 1]. DAEC was found more often in isolates from diarrheic isolates (29/35; 82.8%) as compared to isolates from UTI isolates (6/35; 17.1%). The prevalence of DAEC ranged from about 8% (Shiraz)[27] to 9% (Tehran and Sanandaj)[16] in Iran [Appendix 1 and Figure 3].

Adherent-invasive Escherichia coli

There are no data about the prevalence of the AIEC pathotype until April 2017 in Iran.

Publication bias

Funnel plot depicted for 73 studies. It seems that there is an asymmetric in funnel plot and small studies (smaller precision) trending to have large effect t size (prevalence) [Figure 4], this was cross-checked by Egger's regression test in which the results of this test confirm that a publication bias among studies (t = 4.05, P < 0.001). Furthermore, publication bias was assessed in each of subgroup using Egger's test; the results indicated that there was a considerable publication bias in some variables such as quality of studies, year of publication, and sample size [Table 1]. A trim and fill method was performed to rectify the detected publication bias and then the pooled effect size for each subgroup (with P < 0.1) was filled.

Figure 4.

Funnel plot to display publication bias among 73 studies included in meta-analysis

DISCUSSION

Gastrointestinal infections due to pathogenic E. coli are significant causes of outbreaks and sporadic cases worldwide, particularly in children from countries in South Asia and Sub-Saharan Africa.[3,28] The present systematic review and meta-analysis was designed to estimate the prevalence of DEC pathotypes in the Iranian population according to available data from articles collected from various regions of the country by Iranian researchers. The findings of this study highlight the low frequency of DEC pathotypes in various geographical areas of Iran. The ETEC was identified as one of the most common pathotypes (16%; 95% CI: 11%–23%) among the studied data. ETEC strains are the significant cause of diarrhea in travelers, and there is major morbidity and mortality in children residing in developing countries.[29,30] Annually, this pathogen causes 280–400 million diarrheal episodes in children aged <5 years and causes substantial disease in adults in developing countries, with an estimated 400 million cases per year in people aged over 15 years.[31] Several studies conducted in Iran have also revealed the high prevalence rate of E. coli isolates to ETEC in different parts.[15,16,17,32,33] According to the report of Gupta et al., ETEC was the etiological agent isolated in a median of 13% of diarrheal cases in children of developing countries.[34] Previous studies showed that ETEC is usually a frequent cause of diarrhea in infants younger than 2 years of age.[35,36] In Iran, it was found to be the most common cause of diarrhea in children younger than 5 years of age.[17,28,33,37] This variation may be because children under 3 months became susceptible to a primary infection; ETEC strains to produce STh and LT were most common, whereas at 6–7 months, ETEC strains to produce STp, STpLT, and SThLT were dominant.[35] The susceptibility of young children has also been reported in other settings which have poor hygiene conditions.

VTEC, also known as STEC infections, causes a wide spectrum of clinical manifestations ranging from symptom-free carriage forms of intestinal illnesses to bloody diarrhea. EHEC is a subset of STEC capable of causing hemorrhagic colitis and hemolytic uremic syndrome in humans.[38,39] In earlier studies in Iran, the prevalence of STEC isolates ranged from 0.0% among E. coli isolated from children <5 years of age[28,40,41,42] to 30%–48% among E. coli isolated from patients of unknown age.[12,13,14,15,43] In developing countries, where studies typically focus solely on children, STEC has also been detected as a pathogen of concern in this age group.[44,45] However, in developed countries, studies investigating all ages typically identified STEC more frequently and with greater severity in young children.[46,47]

Unfortunately, despite the existence of many data of aEPEC in Iran (39 publications), there is not sufficient data about the incidence of tEPEC (12 publications). The results of our study indicated that the incidence of aEPEC in the investigated studies varies in different provinces, such that the highest rate of incidence was reported from Sanandaj with 59.6% (west of Iran);[19] central, western, and northern Iran with 38.8%;[13] Khuzestan with 31% (south west of Iran);[43] and Kashan with 28.6% (central of Iran).[48] Previous studies in Northeast India,[49] Ghana,[50] and Kenya[51] showed that EPEC was the most frequently identified bacterial pathogen.

In Iran, moderately high reports are available on the occurrence of EIEC (17 publications). In the present study, 4.5% (95% CI: 2%–6%) of E. coli isolates were confirmed as EIEC, of which the most prevalent were from Sanandaj (west)[19] and Tehran (central)[26]. Other studies in the world also reported less percentage of EIEC (0%–1.5%).[52,53]

EAEC is well known as a diarrhea-causing agent in people of all ages in developing and industrialized countries, most prominently in association with persistent diarrhea.[54] According to our results, the total prevalence of EAEC strains was 11.1% (95% CI: 8%–15%). It is consistent with some studies conducted in Switzerland (10.2%) and Kenya (8.9%).[51,55] Another study in urban and rural South Korea showed that the prevalence of EAEC was 0.3%–3.7% of acute diarrhea samples.[56]

There are very few reports on the epidemiology of DAEC infections in Iran (two publications). However, the frequency of this pathotype was 8.7%. Both studies were conducted in Brazil, and DAEC was significantly associated with diarrhea in children.[57,58] In a study in Iran, DAEC strains were isolated in 6% of patients with UTI.[16] DAEC strains may represent the reservoir for uropathogenic E. coli, since several virulence factors of DAEC, such as adhesins of the Afa/Dr family, are found in uropathogenic E. coli strains.[59] Four E. coli isolates presenting mixed characteristics (mixed EAEC/DAEC genes) were identified.[15] Several studies showed the presence of diarrheagenic isolates presenting mixed characteristics of two different pathotypes.[60,61]

CONCLUSION

By combining the results from 73 studies, this systematic review and meta-analysis identified the lower frequency of DEC infections in Iran. Our data suggest that the ETEC pathotype is one of the most important etiological agents of disease in this country. However, the prevalence of DEC pathotypes is diverse in different regions of Iran. The limitation of the study was associated with high heterogenicity, especially in sample size subgroup. We used subgroup analysis to find the source of heterogeneity, but in all moderator variables, there is a high heterogeneity. Maybe there is a possibility that a moderator variable is missing that does, in fact, explain the heterogeneity (i.e., age and sex). Furthermore, nonuniform study design and various types of publication bias may be the other causes of heterogeneity in these studies. Our study included other limitations such as it cannot fully represent the prevalence of DEC infections in Iran because the extent of DEC has not yet been examined in some regions of the country.

Financial support and sponsorship

The authors would like to thank the Infectious and Tropical Disease Research Center, Hormozgan Health Institute, Hormozgan University of Medical Sciences, Bandar Abbas, Iran, for funding assistance with this project. This study was funded by grant number (960112) and ethical number (1396.65).

Conflicts of interest

There are no conflicts of interest.