During 2004–2008, serial case–control studies were conducted in 22 remote communities in Esmeraldas Province, Ecuador. Complete study design and laboratory procedures for pathogen detection have been described (7). Briefly, each community was visited 4–6 times on a rotating basis; each visit lasted for 15 days, during which all cases of diarrhea were identified by a visit to each household every morning. Household residents with cases had >3 loose stools in a 24-hour period, and controls had no symptoms of diarrhea during the past 6 days. Fecal samples were collected from 3 healthy controls per person with diarrhea. These samples were plated on selective agar media, and 5 lactose-fermenting colonies were screened by PCR for enterotoxigenic Escherichia coli (ETEC), enteropathogenic E. coli, and enteroinvasive E. coli (EIEC). Lactose-negative isolates that were identified as either Shigella spp. or E. coli were also screened by PCR for the same molecular marker used for EIEC. All non–lactose-fermenting pathogens, including P. shigelloides, were biochemically identified by API 20E system (bioMèrieux, Marey l’Etoile, France). Because shigellae are phylogenetically similar to E. coli pathotypes, we combined data from persons infected with E. coli and those infected with shigellae in our analysis. We tested for Giardia lamblia by using an ELISA kit (RIDASCREEN Giardia; R-Biopharm, Darmstadt, Germany), and rotavirus was detected with an enzyme immunoassay kit (RIDA Quick Rotavirus; R-Biopharm). We chose a molecular method for detecting E. coli pathotypes because they cannot be differentiated solely on the basis of biochemical tests; the metabolic homogeneity of P. shigelloides, however, makes this organism easily and clearly identifiable by biochemical test. Similarly, immunologic methods used for Giardia spp. and rotavirus detection are specific and sensitive enough to accurately detect these pathogens, and use of molecular methods would be justified only for deeper analysis. Institutional review board committees at the University of California, Berkeley; University of Michigan; Trinity College; and Universidad San Francisco de Quito approved all protocols.

During March 2004–March 2008, a total of 2,936 fecal samples were collected from persons of all ages (168 [6%] were <1 year of age, 597 [20%] were 1–4 years, 753 [26%] were 5–12 years, 1,362 [46%] were >13 years, and 56 [2%] were missing a birth date), corresponding to 775 cases and 2,161 controls. P. shigelloides was isolated in 253 (8.6%) samples. This number exceeded isolation rates for all of the pathogens analyzed except G. lamblia, which was present in 701 (23.9%) samples. Rotavirus was detected in 225 (7.7%) samples and EIEC/shigellae in 188 (6.4%) samples. P. shigelloides was detected in 11.4% of case-patients with diarrhea (case prevalence), which is more than the 7.2% estimated in the community (weighted control prevalence; Figure). However, once we stratified by persons infected only with P. shigelloides and those infected with P. shigelloides plus >1 of the other marker pathogens for which we tested, single infections with P. shigelloides were almost equally prevalent in the case-patients and in the community; in contrast, co-infections with P. shigelloides and other pathogens were more frequent in persons with diarrhea (Figure).

To determine whether P. shigelloides infection was associated with diarrhea, we estimated risk ratios (RRs) and bootstrapped 95% CIs for single and co-infection exposures (Table). A single infection with P. shigelloides was not associated with diarrhea (RR 1.5, 95% CI 0.9–2.2). Persons co-infected with P. shigelloides and another pathogen, however, had almost 6× the risk for diarrhea than those with no infection (RR 5.6, 95% CI 3.5–9.3) and simultaneous occurrence of P. shigelloides and rotavirus increased the risk for diarrhea to 16.2 (Table). We found no evidence for confounding of the association between P. shigelloides and diarrhea by co-infecting pathogens (RR_crude = RR_MH-pooled; where RR_crude is the unadjusted RR and RR_MH-pooled is the pooled Mantel-Haenszel RR estimate). However, we found some evidence for confounding by age of P. shigelloides co-infection (RR_MH-pooled 4.2 [95% CI 2.1–8.1], compared with the crude estimate of 5.6) but no evidence for confounding by age for single infection with P. shigelloides.

A single infection with P. shigelloides resulted in a moderately increased risk for diarrheal disease, which suggests that this microorganism plays a minor role as a pathogen. This result agrees with findings of previous studies (4,8,9). Analysis of the co-infections, however, suggests that P. shigelloides may be pathogenic in the presence of another pathogen. Specifically, co-infections of P. shigelloides with either rotavirus or pathogenic E. coli (including shigellae) were 16.2× (95% CI 5.5–62.3) and 13.8× (95% CI 3.3–69.3) more likely to result in diarrhea, respectively. We cannot, therefore, rule out the pathogenic capacity of P. shigelloides even though single infection may not be sufficient to cause disease.

This co-infection analysis might be limited by the number of pathogens considered (Giardia spp., rotavirus, pathogenic E. coli, and shigellae). However, the high isolation rates suggest we are detecting the major pathogens in the region. Other pathogens that may be useful to consider, given their attention in the literature, include Entaemobae histolytica and Cryptosporidium spp.

Although we found nothing in the literature that addresses the role of co-infection in the pathogenicity of P. shigelloides, co-infection with enteric pathogens is a well-known phenomenon, especially in tropical regions (6). Co-infection with ETEC and enteropathogenic E. coli increases virulence (10). Other studies have shown that the severity of disease is increased when rotavirus infections occur alongside another infection with another enteric pathogen (11).

P. shigelloides may take advantage of the disruption of the normal gut microbiota and gut physiology because of the concurrent presence of other pathogens, establishing a pathology in the human gut. For example, diarrhea caused by enterotoxins produced by pathogens, such as ETEC, and Vibrio cholerae (12), may remove normal gut microbiota, enabling P. shigelloides to establish an infection. The disruption of gut microbiota that facilitates gut colonization has been demonstrated in murine models infected with Citrobacter rodentium and Salmonella enterica serovar Typhimurium (13).

Most medical literature considers infectious diarrhea as a monopathogenic phenomenon (12,14). In the data presented here, the crude risk ratio suggests that P. shigelloides is pathogenic. When looking at single infections, we found no evidence that P. shigelloides is pathogenic. When looking at co-infection data, however, we found associations between infection and diarrhea. Our findings suggest that multipathogenic infections may play a role in the pathogenesis of infectious diarrhea.