Introduction

Gastrointestinal (GI) symptoms triggered by foodborne diseases are often of infectious nature. Although these conditions are usually benign and self-limiting, diarrheal episodes can infrequently cause severe fluid and electrolyte loss acutely resulting in severe dehydration [1].

Acute gastroenteritis (AGE) causing diarrhea and vomiting is a major cause of illness in the USA; an estimated 179 million episodes occur annually [2]. Although AGE is caused by a variety of infectious and noninfectious causes, Noroviruses are the leading cause of epidemic gastroenteritis, detected in approximately 50% of all AGE outbreaks across Europe and the USA [3, 4].

In the USA and elsewhere, infectious diarrhea caused by traveling is a major burden and although rarely fatal, foodborne and waterborne diarrheal illnesses are extremely common. Nevertheless, even in the USA, up to 1 in 6, or 48 million Americans are affected yearly, resulting in 128,000 hospitalizations and nearly 3000 deaths, per CDC data [5]. Also, rotavirus has been the leading cause of global mortality due to diarrhea in all ages (15·2%), closely followed by Shigella (12·5%) and Salmonella infections (6·9%) [6].

Hall et al. [7] analyzed data reported through the national outbreak reporting system, NORS, during 2009 and 2010. Of 4455 outbreaks reported during this time period, 98% were AGE outbreaks associated with approximately 3000 hospitalizations and 168 deaths [7]. The authors reported that a single etiology was implicated in 64% of the outbreaks. Among the single etiologies, Norovirus was the leading cause of the outbreaks (68%) and responsible for 86% of the deaths observed in the study. Salmonella spp. (13%), Shigella spp. (4%), and Shiga toxin-producing Escherichia coli (STEC, 4%) were the next most frequent etiologies responsible for outbreaks, respectively. Salmonella spp. was associated with more than 700 hospitalizations (32%) and was the second most frequent cause of death [7].

According to the CDC, travelers’ diarrhea (TD), a subset of AGE including incidents occurring on cruise ships, is the most predictable travel-related illness in all travelers [8]. Bacterial pathogens account for 80–90% of TD, and intestinal viruses usually account for 5–8% of illnesses. Overall, the most common pathogen is enterotoxigenic Escherichia coli, followed by Campylobacter jejuni, Shigella spp., and Salmonella spp. Although most cases of TD are acute and self-limited, a certain percentage of travelers will develop persistent (> 14 days) gastrointestinal symptoms [9].

Readily available over-the-counter (OTC) medication for symptomatic relief and appropriate oral hydration can be health-saving measures of great convenience for those affected by TD. Bismuth subsalicylate (BSS) is a non-proprietary monograph product that is available in the USA and abroad, over-the-counter (OTC). Of all OTC medications for TD, BSS has the greatest antimicrobial activity [10] against pathogenic bacteria [11,12,13]. A study done by Gump et al. [14] first demonstrated the ability of BSS to decrease the invasiveness of enteropathic bacteria in gastrointestinal epithelial cells. Electron-dense deposits of bismuth were found in Yersinia enterocolitica cells exposed to BSS, indicating that the antibacterial effect of BSS was mediated by its ability to increase permeability of the bacterial cell wall and increase bismuth concentrations in bacterial cells [14, 15]. Other studies had indicated that the exposure of bacteria to BSS resulted in a loss of membrane integrity and, possibly, inactivation of cellular ATP synthesis leading to death of the organism [16]. More recently, Pitz et al. [13] verified in vitro the antimicrobial effects of BSS and bismuth oxychloride (BiOCl) on key pathogens, such as Escherichia coli O104:H21 (surrogate to 2011 German outbreak strain), Salmonella, and Norovirus (NoV). These authors showed that bismuth reduced bacterial growth significantly resulting in less than 10 cfu/ml within 24 h. C. difficile was the most susceptible pathogen to the bismuth challenges in the antibacterial assays. BSS also exhibited significant inhibition on viral invasion of host cells and viral efficacy. Both BSS and bismuth oxychloride (BiOCl, which is formed in the stomach after ingestion of BSS) at low concentration (0.004–0.13 mg/mL) significantly reduced NoV RNA levels, suggesting an in vivo antiviral mechanism. BSS has been shown to also have antiviral activity since it inhibited replication of four strains of rotavirus in tissue culture cells and caused a dose-dependent reduction in the growth of a number of enteric viruses [17, 18].

Historically, BSS has been indicated and effectively used for treatment of TD or enteric infection mainly when vomiting occurs [19,20,21,22,23,24,25]. Although the safety and efficacy of BSS are well known [26, 27], some of the research done with BSS resides within the industry and has not been published. The data collected in such studies matched the regulatory requirements of the time of the study and determined a body of evidence for the regulatory approval of BSS.

Given the importance of easy access to non-antibiotic therapy for prevention and treatment of diarrheal disease, this review includes previously unpublished clinical studies regarding BSS safety and efficacy. Meta-analyses of randomized controlled clinical trials were performed with studies specifically designed to capture prevention of manifestation and relief of diarrhea.

Methods

A systematic review of the medical literature was performed to identify randomized, placebo-controlled clinical studies that evaluated the efficacy of BSS for the prevention and treatment of diarrhea. Results were reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The review included searches with Medline (1950–2018), Embase (1980–2018), the Cochrane Central Register of Controlled Trials, and Procter and Gamble’s clinical study archive. Eligible studies were identified with search terms “bismuth,” “subsalicylate,” “travelers’ diarrhea,” “infectious diarrhea,” “treatment,” “therapy,” and “prevention” with no language restrictions imposed. The magnitude of any BSS effect was not used as a study inclusion criterion. Abstracts, articles, and final reports were evaluated independently by each of the five authors. Relevant information needed for meta-analyses, including sample sizes and incidence of diarrhea by treatment group, and was extracted to a database by two independent reviewers and compared for accuracy and completeness. A total of 1869 records were identified through database searching of the terms. Clinical trials and review articles available made 126 publications and 70 Procter and Gamble internal clinical studies with BSS, comprising 196 records for assessment of eligibility. Complete, unrestricted, and open access to all Procter and Gamble studies was given to BY during this review process. Only 14 studies  (3 prevention, 11 treatment) were eligible for efficacy analyses, and 182 were excluded by diverse reasons listed in Fig. 1.  A total of 24 placebo-controlled studies were eligible for integrated safety assessment with a total of 1164 subjects receiving BSS and 1166 receiving placebo.

Fig. 1
figure 1

Flow diagram for meta-analysis

Full size image

A total of three meta-analyses were performed to assess specifically the efficacy of BSS for (1) prevention of TD, (2) treatment of infectious diarrhea in general, and (3) treatment of TD. The endpoint in each meta-analysis was binary: success or failure in the prevention or treatment of diarrhea. Data in each meta-analysis were pooled using the generic inverse variance method with random effects model and expressed as an odds ratio with 95% confidence interval using the Comprehensive Meta-Analysis software version 2.2.064. Heterogeneity was assessed using the Cochran Q-statistic and quantified with the I2 statistic. Per protocol efficacy results from each study, i.e., results for subjects compliant with the study protocol, were included in each meta-analysis.

Risk of Bias Assessment

Risk of bias within individual studies was assessed with respect to selection bias (randomization and allocation concealment), detection bias (blinding of outcomes), attrition bias (incomplete outcomes), and reporting bias (selective reporting). All studies included in the meta-analysis were randomized and blinded, and, therefore, the risk of selection and detection bias was deemed to be low. The risk of reporting bias was also considered to be low, given that all published data and unpublished P&G data were included. Intent-to-treat data were not available for all the studies, and, therefore, the potential for attrition bias exists.

Results

Prevention of Travelers’ Diarrhea Meta-Analysis

The first meta-analysis specifically assessed the diarrhea preventive effects of BSS in generally healthy adults who had no diarrhea at the time of randomization. Five P&G-sponsored studies were identified for consideration. Out of the five studies, three were designed for prevention of TD, and two involved induced diarrhea through viral inoculation (Norwalk agent or Norovirus) or chemical irritation (castor oil consumption)-induced diarrhea. Although BSS had shown efficacy in these two chemical irritation studies, in order to maintain relevance to populations at risk for TD, only the three naturally occurring diarrhea studies were considered for meta-analysis. All three studies (n = 404 subjects, age between 16 and 70 years) followed a similar parallel group, placebo-controlled design in diarrhea-free individuals embarking on a trip to tropical countries (Mexico and African destinations) known for elevated incidence of TD. All subjects initiated a 21-day prophylactic BSS dosing regimen (one study used BSS tablets QID to a total of 2.1 g/day, the second study used oral suspension of 1050 mg QID to a total dose of 4.2 g/day, and the third study used tablets BID to total dose of 2.1 g/day) starting from 1 day before departure or 2 days after arrival at the trip destination. Study details, including the highest dose administered, the criteria for diarrhea diagnosis, and findings, are summarized in Table 1.

Table 1 Study details for trials evaluating the prevention of travelers’ diarrhea of BSS in generally healthy adults
Full size table

The meta-analysis efficacy endpoint was self-assessed diarrhea (presence/absence) during the 21-day treatment period. Meta-analysis results are reported in Fig. 2 in terms of odds ratios and 95% confidence intervals with larger values indicative of greater efficacy for BSS relative to placebo. The data indicate that subjects who dosed BSS had 3.5 times greater odds of not developing TD than subjects who dosed placebo (95% CI 2.1, 5.9; p < 0.001, I2 = 27%).

Fig. 2
figure 2

Meta-analysis of trials evaluating the prevention of traveler’s diarrhea with BSS in generally healthy adults. Heterogeneity: I2 = 27%, p = 0.252

Full size image

Treatment of All Infectious Diarrhea Meta-Analysis

The second meta-analysis focused on the effect of intervention in otherwise generally healthy adults with acute diarrhea (probably infectious) at the time of randomization. A total of 14 clinical studies were identified for consideration. Johnson et al. [28] and two unpublished P&G studies did not have negative controls and, therefore, were excluded. The remaining 11 studies (n = 840 subjects) consisted of one published P&G study and ten unpublished P&G studies, all of which compared BSS to negative control over a 1- to 2-day treatment period. Nine of the studies dosed 4.2 g/day BSS (1400 mg tid), and two studies dosed 2.1 g/day (700 mg tid) BSS. Study details including findings are summarized in Table 2.

Table 2 Study details for trials evaluating intervention in generally healthy adults with diarrhea
Full size table

The endpoint of interest in this meta-analysis was relief of diarrhea, assessed by either the patient or a physician. The odds ratio and 95% confidence interval for each study are reported in Fig. 3, together with the overall summary effect. The results indicate that subjects who dosed BSS had 3.7 times greater odds of experiencing relief of infectious diarrhea than subjects who dosed negative control (95% CI 2.1, 6.3; p < 0.001, I2 = 42%).

Fig. 3
figure 3

Meta-analysis for trials for evaluating intervention in generally healthy adults with infectious diarrhea. Heterogeneity: I2 = 42%, p = 0.069

Full size image

Treatment of Travelers’ Diarrhea Meta-Analysis

The third meta-analysis focused specifically on traveler’s diarrhea, a subset (n = 5) of the 11 studies (DuPont et al. 1977 [29]; Study 3; Study 5; Study 6; and Study 8) included in the earlier meta-analysis to assess treatment of all infectious diarrhea. These five studies were selected to determine the efficacy of BSS for the treatment of TD only. The results (Fig. 4) indicate that subjects who dosed BSS had 3.1 times greater odds of experiencing relief of TD than subjects who dosed negative control (95% CI 1.9, 5.0; p < 0.001, I2 = 0%).

Fig. 4
figure 4

Meta-analysis for trials for evaluating intervention in generally healthy adults with travelers’ diarrhea. Heterogeneity: I2 = 0%, p = 0.877

Full size image

Since the inter-study heterogeneity statistics I2 did not exceed 42% across the three meta-analyses, none of the meta-analyses were considered to have substantial heterogeneity.

Adverse Events

The most common adverse events (at least 2% in either placebo or BSS group) experienced by participants in safety studies in adults (24 studies, including the 14 studies used for efficacy meta-analyses) were benign and are described in Table 3. Subjects receiving BSS were significantly more likely to experience a transient discoloration of the feces or tongue; this is caused by a well-established reaction of gastrointestinal hydrogen sulfide with bismuth resulting in bismuth sulfide [27, 30].

Table 3 Most common (≥ 2% in either treatment group) adverse events by MedDRA system organ class and preferred term (24 integrated BSS clinical studies in adults)
Full size table

Discussion

The global prevalence and consequences of acute infectious diarrhea are increasing [7, 9, 31]. This problem has been compounded by the development of antibiotic-resistant bacteria (MRSA, VRE, and ESBLs) and virulent strains (B1/NAP1 CDI) [32,33,34]. Strategies that manage to decrease antibiotic use in hospital patients have been largely successful in reducing resistant bacterial infections from further spread. Concern that resistant bacterial infections will establish themselves in the community is based on facts and ways of mitigating this risk need to be addressed. Use of antibiotic husbandry in the inpatient clinical situation has been shown to be effective for reducing the incidence of Clostridium difficile infections in at least one institution (UPMC) [35]. Clearly, a non-antibiotic treatment strategy in appropriate outpatients remains a priority. Here, we have shown for the first-time industry and academic data that support the use of a readily available antimicrobial medication for the treatment of infectious diarrhea and prevention of TD. Dupont and Steffen in 2016 warned about the ongoing use of antimicrobial agents for the treatment and prevention of travelers’ diarrhea with ESBLs and the need for consensus recommendations [34]. This clearly has been an ongoing and common problem that has yet to be addressed, and the expansion of “newer” agents of MDR with bacteria such as E. coli ST 131 as the latest agent is case in point of the concept of bacterial ecology and expansion. Such increasing worldwide occurrences need to be studied in the context of associated antibiotics used for treatment, locations of acquisition, and the development of personalized treatment methods to manage the expansion of these epidemics, including the consideration of non-antibiotic strategies for prevention and treatment [36, 37].

In our work, our meta-analyses results have shown that individuals using BSS have 3.5 times greater odds for prevention of TD (p < 0.001), and, once diarrhea is manifested, the odds of relief are 3.7 times greater with BSS compared to placebo (p < 0.001). Preventive therapy for those at high risk, and therapeutic approach for subjects with active symptoms, using a convenient OTC medication could have a significant impact on complications associated with infectious diarrhea [32].

BSS exerts its antidiarrheal effects through several different mechanisms that affect the symptoms (anti-secretory), and the cause (microbial) of diarrhea. BSS treats the symptom by inhibiting secretion of fluids while stimulating absorption to help normalize fluid movement [27]. In addition to normalizing fluid movement, BSS decreases the gastrointestinal motility to help relieve the symptoms [28].

In regard to antimicrobial activity, BSS has been shown to be effective at inhibiting growth of common causative organisms of diarrhea, such as E. coli, Salmonella sp., Shigella sp., and Vibrio sp., both in vitro and in vivo [10]. BSS also binds and inactivates bacterial toxins and bile acids [12, 14] that can cause diarrhea. Bismuth compounds even at sub-bacteriostatic and sub-bactericidal concentrations decrease pathogenic bacterial invasion of gut epithelial cells.

The antimicrobial effect of BSS in humans was demonstrated in a study conducted in volunteers who were given either BSS or placebo 8 and 2 h prior to and for 3 days after receiving a dose of enterotoxigenic E. coli [38]. Enterotoxigenic E. coli was recovered from 2 of 14 patients who received BSS and from 13 of 15 patients who received placebo. Furthermore, antibacterial effect of BSS was detected in a placebo-controlled study conducted in 30 infants with diarrhea [39]. Clearance of enterotoxigenic E. coli was observed in all BSS-treated patients (15/15) but not in placebo-treated patients (9/15).

The isolation of bacterial pathogens from the stools of patients with infectious diarrhea is reduced in patients treated with BSS compared to placebo-treated controls. In the study done by Dupont et al. (1977) [29] in Mexico for the treatment of traveler’s diarrhea with BSS, pathogens were isolated from the feces of 4 of 12 patients receiving BSS compared to 27 of 38 in placebo. In a study done by Graham et a1 (1983) [38], in a metabolic unit with monitored healthy (n = 14) volunteers given BSS prophylactically before administration of ETEC orally, recovery of toxigenic E coli from stools was seen in 2 of 14 patients given BSS compared to recovery from 13 of 15 stools in the placebo-treated group. Similarly, in the prophylaxis study by Steffen et al. [40] among travelers to Africa, pathogens could not be isolated in the stools of any of the six patients given BSS, but positive cultures were found in 6 of 12 subjects receiving placebo.

These studies support the clinical antimicrobial activity of BSS. Although BSS inhibits bacterial pathogens in vitro at the concentrations similar to what is achieved by recommended oral doses, this antimicrobial activity is not sufficient to significantly change the microbial composition of the normal microbiota detected in feces. While the overall concentrations and types of colonic bacteria are potentially unchanged by oral ingestion of OTC doses of BSS, there are some indications that fermentation of lactose and raffinose is reduced by colonic bacteria in patients treated with BSS [18, 41].

State-of-the-art quantitative PCR assays for pathogens causing TD have shown more recently that enteropathogen E. coli (47%), enteroaggregative E. coli (46%), enterotoxigenic E. coli (22%), enterohemorrhagic E. coli (7%), campylobacter (6%), Shigella (2%), or enteroinvasive E. coli (2%), and salmonella (2%) were detectable from stools of subjects with TD [42,43,44].

Recently, research was conducted using a time-kill assay with pure cultures of six pathogens of the gastrointestinal microflora including E. coli 0157:H7. Transmission electron microscopy was used to learn how BSS interacts with bacteria at the cellular level. Within a 0.5 h of BSS exposure, bismuth was detected on the bacterial cell membrane and inside the cells. Results from the study illustrate that pure BSS salt, Pepto-Bismol, and the hydrolyzed forms (bismuth hydroxide and bismuth carbonate) either inhibited bacterial growth or eradicated the bacteria for all species tested within a 24-h period [13].

BSS when given prophylactically in much lower daily doses than required for treatment can significantly decrease the incidence of TD. The fact that the bismuth component of BSS is an agent that is very poorly absorbed favors a localized antimicrobial effect. The meta-analysis of the efficacy showed that subjects receiving BSS immediately after arriving in the foreign country and before having any symptoms or signs of diarrhea had 3.5 times greater odds of not developing TD than subjects taking placebo as preventative therapy. Corresponding to this clinical preventive efficacy are the findings that BSS even at bacteriostatic and sub-bactericidal concentrations was shown to decrease pathogenic bacteria attachment and invasion of gut epithelial cells [16]. Therefore, utilization of BSS may also help to avoid early unnecessary antibiotic therapy for acute and traveler’s diarrhea, a common practice in developing countries. A recent study done by the Center for Disease Control (USA) with 440 outpatients with acute diarrhea and randomized to BSS or placebo in Pakistan showed that BSS was able to reduce the use of antibiotic by 45% independent of the infectious etiology and by 74% of bacterial gastroenteritis [45].

Although BSS can have significant efficacy in the eradication of H pylori, by itself, it is not expected that the amounts ingested would be sufficient to sterilize the GI tract in a 48-h dosing regimen. Gorbach et al. [18] showed that among patients being prepared for colonoscopy the various populations of normal microbes from fecal samples were relatively unchanged in culture by a 1- or 2-day course of BSS. Therefore, BSS may be more effective against invading microbes than those colonizing the GI tract.

These recent data with new generation of diagnostics make the case for BSS as a potential first-line therapy for TD. Moreover, recent non-bacterial infectious epidemics such as recently well-publicized cruise-ship epidemics with Norovirus make TD situations particularly likely to have single organism etiologies. The susceptibility of these viral infectious agents to BSS is well documented [13]. Exceptions to BSS therapy should include more severely ill, febrile individuals with evidence of systemic signs such as dehydration, marked tachycardia, or inability to maintain oral intake to balance fluid losses by sweat and diarrhea.

BSS has a long history of safety when used as over-the-counter or prescribed medication [46, 47]. A previous review (internal P&G data) of publically available post-marketing AE reports from consumers utilizing over-the-counter BSS (Pepto-Bismol®, P&G Cincinnati, Ohio) over a period of more than 17 years found that the post-market AE profile of BSS was similar to the AE profile of clinical trials observed in this study. Furthermore, in patients with Helicobacter pylori-positive gastritis, BSS is commonly administered for H pylori eradication usually for 2 weeks due to its efficacy and benign AE profile [48]. Therefore, these clinical studies support that BSS, when administered in accordance with current OTC labeling, is safe, well tolerated. Although cases of overdosing with bismuth subsalicylate can lead to toxicities [46, 49, 50], such as tinnitus and even life threatening salicylism and bismuth neurotoxicity, their occurrence is rare.

A clinical study demonstrated that orally administered bismuth salts were in general poorly absorbed and that BSS in particular had significantly lower potential for systemic bismuth exposures than other bismuth salts such as colloidal bismuth subcitrate or bismuth subgallate [51].

From the integrated safety analysis, the most common AEs experienced by subjects receiving BSS were discoloration of the feces or tongue; this is a benign and well-established condition associated with the use of bismuth salts due to formation of bismuth sulfide (a black color salt) from bismuth reaction with hydrogen sulfide present in the feces or food residues. Other non-serious gastrointestinal events such as constipation, diarrhea, nausea, and vomiting were conditions for which the population was being treated and were present in both placebo and BSS treatment groups similarly.

The limitations of our analyses comprise elements of individual study design and population samples and endpoints. Although all studies followed similar definitions of diarrhea which was the primary endpoint analyzed, not all studies included assessment of other symptoms commonly associated with diarrhea, such as nausea, vomiting, abdominal discomfort, gases, and pain which would have enriched the clinical aspects of the benefits brought by BSS treatment or prevention of diarrhea. Pathogen analysis confirming infectious nature of diarrhea was not done in studies used to compare doses and different concentrations of the active BSS and therefore was not included in our analysis or discussion of efficacy. Nevertheless, it is well accepted that traveler’s diarrhea is a subset of infectious gastroenteritis.

Overall, the current meta-analysis with BSS shows significant benefits in prevention and treatment of TD with the additional importance of recognizing non-antibiotic over-the-counter treatment as a potential first-line treatment for a highly prevalent condition that has been suspected of largely contributing to antibiotic resistance. We agree with the work of DuPont and others that the management and prevention of TD are not a simple matter [34]. Clearly, research into the role of all agents used in this global illness needs to be stimulated, particularly with review and consensus development as well as new research initiatives such as personalized medicine and antibiotic husbandry.