Prevention of antibiotic-associated diarrhea (AAD).

As an example, we first examine the evidence for different probiotics for the prevention of antibiotic-associated diarrhea. AAD is defined as diarrhea (typically ≥3 loose/watery stools/day for at least two days) that occurs during the admission of oral or intravenous antibiotics (immediate onset AAD) or ≤4–8 weeks post-antibiotic discontinuation (delayed onset AAD). AAD may occur commonly in 5–50% of patients after use of almost any type of antibiotic, which disrupts the protective normal microbiome of the host. The consequences of AAD may include prolonged hospitalization stays for inpatients, increased healthcare costs, higher risk of acquiring other nosocomial infections and poor compliance (especially for outpatients), leading to inadequate cure rates due to shortened antibiotic treatment [33]. Of the 59 trials (61 study arms) testing probiotics for the prevention of AAD, there were eight different types of probiotics tested with ≥ 2 RCTs. Determining which probiotic is best for AAD is challenging due to differences in probiotic strains, doses and types of study populations. However, as shown in Table 2, of the eight different types of probiotics, three probiotic types [S. boulardii CNCM I-745, a mixture of three Lactobacilli strains (L. acidophilus CL1285 +L. casei Lbc80r + L. rhamnosus CLR2) and another single strain probiotic (L. casei DN114001) had strong evidence for the prevention of AAD, while one showed only moderate evidence (E. faecalis SF68). Effective probiotics were typically started 1–2 days of antibiotic admission with daily doses ranging from 10⁷−10¹⁰ per day.and continued for 1–4 weeks after antibiotics were discontinued. Four other probiotics had more studies with non-significant findings compared to significant studies and thus should not be considered as effective therapies for the prevention of AAD until further confirmatory trials show significant efficacy. These trials are described in detail in supporting information (S1 Dataset).

Prevention of Clostridium difficile infections (CDI).

Clostridium difficile accounts for approximately one-third of the cases of AAD, but may result in more severe disease (colitis, pseudomembranous colitis toxic megacolon), higher mortality rates and is the leading cause of nosocomial infections [34]. The incidence of CDI continues to increase in hospitals and long-term care facilities, causing over 14,000 deaths in the United States each year and adding $1 billion in healthcare costs [35]. CDI is diagnosed based on both the symptomatic status of the patient and the presence of at least one of the toxins of C. difficile in the stool. Probiotic strains or mixtures found effective for preventing AAD have been assessed in RCTs for their ability to prevent CDI, but only a few have shown promise [36].

There were only two studies testing probiotics as primary prevention therapies in randomized controlled trials. Most of the data on CDI was extracted from randomized trials for AAD that reported CDI as a secondary outcome. CDI occurs less frequently than AAD and thus these trials were not powered to assess CDI and often suffered from a limited number of CDI cases and an inability to determine if the probiotic was effective for CDI [37]. As shown in Table 2, several of the probiotics had more or equal numbers of non-significant outcomes (S. boulardii CNCM I-745, L. rhamnosus GG, L. casei DN114001 and the mixture of three Lactobacilli strains (L. acidophilus CL1285, L. casei Lbc80r, L. rhamnosus CLR2) by the number of individual studies. As the finding of non-significance may be due to the insufficient power in these studies, two meta-analyses of the pooled data provided more power and determined significant efficacy was present for three of the probiotics (S. boulardii CNCM I-745 and L. casei DN114001 and the mix of Lactobacillus acidophilus CL1285, L. casei LBC80R and L. rhamnosus CLR2) [38,39].

Another strategy we found in the literature was the use of quasi-experimental interventions, but we did not include them in evaluating the strength of the evidence. Primary prevention of CDI has typically rested upon enhanced infection control bundles and antibiotic stewardship programs [37]. Recently, facilities have added a specific probiotic to their infection control bundles to test if this would reduce the incidence of CDI. Only one type of probiotic had multiple studies to confirm their findings. From a RCT testing two different doses of a mixture of Lactobacillus acidophilus CL1285, L. casei LBC80R and L. rhamnosus CLR2, a significant reduction of CDI for both the low-dose (9.4%, p = 0.03) and the high dose (1.2%, p = 0.002) compared with the control group (23.8%) was found [40]. As a consequence, one hospital in Canada which was experiencing a persistent large CDI outbreak gave this three-strain Lactobacilli mixture to all inpatients receiving antibiotics and found a significant reduction of CDI rates, which was sustained over 10 years of the program [41]. The reduction of CDI rates was also observed at other Canadian hospitals who added the three-strain Lactobacilli formulation to their infection control programs [42, 43].

Prevention of adverse reactions to H. pylori therapy.

H. pylori infections may lead to serious consequences, but the onset of symptoms (dyspepsia, gastric cancer, etc.) has a long incubation period (10–20 years), so efforts to eradicate carriage of the pathogen has been the standard focus of treatment. Unfortunately, the treatment (2-3-types of antibiotics, along with a proton-pump inhibitor) has poor compliance due to the common occurrence of adverse reactions to the antibiotic treatment. In contrast to the ability to eradicate H. pylori, probiotics may have a more useful role in reducing the side-effects of H. pylori eradication therapy [18,19]. We included 14 RCTs (and two additional study arms) that reported adverse reaction data in their trials of the treatment of H. pylori. Two probiotics have strong strength of efficacy evidence for preventing adverse events associated with treatments for H. pylori (S. boulardii I-745 and the mixture of L. helveticus R52 + L. rhamnosus R11), while L. rhamnosus GG only showed moderate strength. Most of the adverse reactions were associated with the antibiotic treatment (AAD, nausea, vomiting or abdominal pain), which may result in poor compliance to the eradication therapy. In this case, the use of these three probiotics may be valuable in preventing side effects of therapy, but not be directly effective in eradicating the pathogen (H. pylori) itself.

Prevention of necrotizing enterocolitis (NEC).

NEC remains one of the most common cause of death and morbidity in preterm infants, especially for very low birth weight neonates (<1500 g) and is characterized by bowel wall necrosis and >27% require surgery to treat NEC [44]. Although recent meta-analyses have found significant reductions in NEC in trials with probiotics, recommendations for specific probiotics are lacking as the analysis often pooled together different types of probiotic strains [45, 46]. Of the 17 RCTs, six different probiotic types were tested for NEC prevention. Most did not show strong or moderate evidence of efficacy (Table 2) for reducing NEC. One mixture (B. infantis, B. lactis, Strept. thermophilus) showed strong evidence (two trials with significant efficacy) in preventing NEC (Table 2). However, when lactoferrin was added to L. rhamnosus GG, both trials showed significant reductions in NEC in contrast to two trials when L. rhamnosus GG was given alone (no efficacy found). Additional confirmatory trials with the other types of probiotics are needed before strong recommendations for use should be made.

Prevention of post-surgical infections.

Post-surgerical infections may be amendable to probiotic therapy as the microflora is often disrupted due to common exposures associated with surgery (pre-operative antibiotic exposure, other medications, etc). We found eight RCTs that had sufficient numbers of trials to assess efficacy. One synbiotic (Pediococcus pentosaceus 5–33:3, Leuconostoc mesenteroides 77:1, L. paracasei ssp. paracasei F19, L. plantarum 2362 and four fibers (inulin, oat bran, pectin, starch) had strong evidence for the prevention of post-surgical infections, but the types of surgeries varied (S1 Dataset), while another type (L. plantarum 299v) had only weak evidence. The synbiotic started on the day of surgery and continued 1–2 weeks afterwards and was given at a higher dose (2–4 x 10¹⁰/day) compared to L. plantarum 299v (2 x 10⁹/d).

Prevention of traveler’s diarrhea.

Traveler’s diarrhea affects as many as 24–40 million travelers worldwide and the mean duration ranges from 12 hours and 3.5 days. One incapacitation day results in a loss of $290–490 million dollars of lost revenue. Prevention options are limited and behavioral practices are often overlooked by travelers. Probiotics may offer a method of preventing this disease [47]. We found only five RCTs (two had additional treatment arms) and found strong evidence for S. boulardii CNCM I-745, while L. rhamnosus GG had weak evidence (Table 2). Typically, probiotics (2–5 x 10⁹ per day) were begun a few days before travel, continued during travel and then for 2–5 days afterwards to allow time for intestinal microflora restoration.

Prevention of other diseases.

Five other disease indications had fewer RCTs or did not find any probiotics with strong or moderate evidence. One probiotic (S. boulardii I-745) had strong evidence from three RCTs for the prevention of diarrhea associated with nasogastric tube feedings (Table 2). The evidence for the prevention of four other diseases (allergy, nosocomial infections, respiratory tract infections and urinary tract infections is less robust and no probiotics were found to be strongly or moderately effective.

Treatment of pediatric acute diarrhea.

Pediatric acute diarrhea is a leading cause of morbidity and mortality in developing countries and can be due to viral or bacterial etiologies. Latin-American guidelines now recommend some probiotics should be given along with oral rehydration therapy for the treatment of acute pediatric diarrhea [13]. Probiotics are often used to treat children with acute diarrhea, but the wide variety of probiotics tested has also led to confusion about the best choice of probiotic to use. Grandy et al. compared giving just the single strain of S. boulardii against a mixture containing S. boulardii and three other bacterial probiotic strains: L. acidophilus and L. rhamnosus and a Bifidobacterial species) and compared the duration of diarrhea against a placebo control group [48]. While the single strain S. boulardii group demonstrated a significant reduction in the mean duration of diarrhea compared to the placebo group (-1.1 day, p = 0.04), when S. boulardii was given along with the mixture of three other bacterial strains, it was less effective (-1 day, p = 0.06). Several meta-analyses have been conducted supporting these individual clinical trials and most have focused on just one type of probiotic. Feizizadeh et al. pooled 22 RCT that treated children with acute diarrhea, but only included studies that used the S. boulardii I-745 yeast probiotic [49]. From the pooled data of 17 trials that reported mean duration of diarrhea, S. boulardii significantly reduced the duration by 19.7 hours. Szajewska et al. included trials using L. acidophilus LB and found a significant mean reduction in diarrhea pooled from 4 trials was 21.6 hours [50]. Urbanska et al. limited their meta-analysis to three trials using L. reuteri DSM17938 and also found a significant mean reduction of diarrhea by 24.8 hours [51].

We found the most evidence in this indication, as 59 RCTs assessed eight different probiotics types (one trial tested three different probiotic types) and seven of the eight probiotic types had strong evidence for this disease indication. Two types of probiotics with over 10 RCTs showed strong evidence of efficacy (S. boulardii CNCM I-745 with 25 trials showing efficacy and four with non-significant findings and L. rhamnosus GG with 10 trials showing efficacy and three with non-significant findings). Five other probiotics have fewer total number of trials but also had at least two more RCTs with significant efficacy compared to the number of RCTs with non-significant outcomes: L. reuteri 17938, L. acidophilus LB, L. casei DN114001, and two mixtures (mix of 4 strains of Bacillus clausii (O/C, N/R84, T84, Sin8) and another mix of 8 strains, as shown in Table 2. Another mixture (L. helveticus R52 and L. rhamnosus R11) showed moderate evidence (two found significant efficacy while one did not).

Treatment of inflammatory bowel disease (IBD).

We found 25 RCTs treating chronic IBD, but only one had strong evidence of efficacy (the 8-strain mixture, VSL#3). This multi-strain mixture found significant improvement in IBD symptom scores in eight of ten trials. S. boulardii CNCM I-745 had moderate evidence in that two of three trials found significant improvement in IBD symptoms, but two other probiotics had more trials with no significant improvement (L. rhamnosus GG and E. coli Nissle) and should not be recommended for patients with IBD at this time.

Treatment of irritable bowel syndrome (IBS).

We found 21 RCTs for the control of irritable bowel syndrome (IBS) symptoms (one trial had three dose arms), Table 2. Two probiotics had strong evidence for the improvement of IBS symptoms (L. plantarum 299v and B. infantis 35624). Four other probiotics (L. rhamnosus GG, S. boulardii CNCM I-745, B. lactis 173010 and the 8-strain mix) had an equal number of trials with significant improvement in IBS and trials with no significant improvement and cannot be recommended for IBS (Table 2).

Eradication of H. pylori.

Helicobacter pylori infections are a global concern, with a prevalence ranging from 70–90% in developing countries and 25–50% in developed countries [18,19]. Prolonged H. pylori carriage may result in an onset of symptoms in adults including gastric ulcers, gastritis and gastric cancer. The standard treatment combines two to three antibiotics with a proton-pump inhibitor, but treatment often fails due to poor compliance relating to a high incidence of side-effects of the antibiotic therapy. In addition to the finding that some probiotics did significantly reduce the incidence of side-effects, much to researchers’ surprise, some probiotics were also effective in increasing the eradication rate of H. pylori. McFarland et al. included 25 RCTs in their meta-analysis using six different types of single strain probiotics and found only one sub-group (S. boulardii CNCM I-745) significantly improved H. pylori eradication rates (RR = 1.11, 95% C.I. 1.07–1.16, p<0.05), while five other sub-groups of the same strain had no significant effect (Clostridum butyricum 588, L. rhamnosus GG, L. acidophilus nr, L. reuteri 17938 or L. casei 114001) [18]. Another meta-analysis of 19 RCTs separated probiotic mixtures into six sub-groups of the same multi-strain mixes and found four mixtures significantly increased H. pylori eradication rates (the mix of L. acidophilus La5 and B. animalis lactis Bb12; a mix of L. helveticus R52 and L. rhamnosus R11; a mix of L. acidophilus nr and B. longum nr and E. faecalis nr; and two RCTs of an eight-strain mixture), while two mixtures (a mix of L. acidophilus La5 and B. bifidum Bb12 and a mix of L. acidophilus 2177 and L. casei 27443 and B. longum 8001) had no significant effect on H. pylori eradication rates [19].

In our review, we found of 33 RCTs (two had an additional treatment arm) for the eradication of H. pylori (details found S1 Dataset). Two probiotic mixtures had either strong evidence (L. helveticus R52 and L. rhamnosus R11) or moderate evidence (L. acidophilus La5 and Bifido. lactis Bb12) for the eradication of H. pylori, while four other types of probiotics (S. boulardii CNCM I-745, L. rhamnosus GG, L. acidophilus LB and C. butyricum 588 had weak to no evidence of efficacy to eradicate H. pylori.

Treatment of Clostridium difficile infections.

Probiotics have also been tested for as an adjunct treatment for CDI, given in addition to standard antibiotic treatments (vancomycin or metronidazole) and using the recurrence of CDI as an outcome. As many as 20–30% of patients with CDI may experience a recurrence of CDI and some may continue to have CDI episodes over a period of years [35]. Only two types of probiotics had at least two RCTs for CDI. Two trials found S. boulardii I-745 significantly reduced CDI recurrence rates [52,53], but the effect was strongest in those with recurrent CDI [52]. Two other RCTs with L. rhamnosus GG did not find any significant reductions in CDI recurrences [54, 55].

Treatment of other diseases.

Three other diseases had evidence from ≥2 RCTs (adult acute diarrhea, pediatric colic and constipation), that show promising results for four types of probiotics (Table 2), but the number of studies is sparse. One probiotic (L. reuteri 17938) had strong evidence from four RCTs for the treatment of pediatric colic.

Formulation.

Probiotics are also available in a wide range of formulations (from yogurts to fermented beverages, to powders in sachets or lyophilized capsules or tablets), as shown in Table 1. There is scant evidence comparing which type of formulation may be more effective. The choice of formulation may be based on shelf-life, in that lyophilized capsules maintain high concentrations longer than probiotics in dairy products and enteric-coated capsules show higher survival rates compared to non-enteric coated capsules [56]. Probiotic capsules requiring refrigeration are heat-dried (not lyophilized) and thus not stable at room temperature, limiting their portability. In addition, if the patient is lactose-intolerant, yogurts or other types of fermented dairy products may be inadvisable.

Daily dose.

Probiotics have been defined by having a health benefit, but also that it be given as an ‘adequate amount’, but the specific dose was not specified in guidelines [23]. The question of whether probiotics have a dose-response, similar to other medications, has been explored. Several reviews and meta-analyses have examined if there is a threshold for an effective oral dose of probiotics. One meta-analysis of 20 RCTs for the prevention of pediatric antibiotic-associated diarrhea (AAD) gave probiotics in doses ranging from 6 x 10⁶ to 3 x 10¹⁰ colony-forming units (cfu) per day and found a lower rate of AAD in those children given probiotics (~8%) compared to controls (17%), but the effect did not significant improve with increasing daily doses [17]. A meta-analysis by Goldenberg et al. analyzed 22 randomized controlled trials testing various probiotics for the prevention of pediatric AAD found an overall dose effect of increased efficacy for probiotic doses ≥ 5 x 10⁹ cfu/day, but pooled data below that dose were ineffective [57]. Ouwehand reviewed the literature and confirmed this dose-response (breakpoint of 10¹⁰ cfu/day) from seven meta-analyses of probiotics for AAD [58]. However, this review found a dose-response only for AAD and the reduction of blood pressure (breakpoint of >10¹¹cfu/day), but not for any other disease indication, including C. difficile infections, necrotizing enterocolitis, prevention of atopic dermatitis, prevention of colorectal cancer, or the treatment of irritable bowel syndrome. Unfortunately, all these reviews combined different probiotic strains into dose groups, thus the findings may have been confounded by the type of probiotic product. Clear dose-response behavior is observed in pharmacokinetic studies of specific strains in dose-ranging studies, typically finding more fecal recovery of the probiotic as the oral dose is increased, but these studies did not document the effect of dose on clinical efficacy [58]. As most daily doses used in probiotic clinical trials are based on kinetic dose recovery studies, a lower limit where a loss of efficacy may be observed (<10⁶ cfu/day) is often not used in RCTs, which might explain why a demonstrable dose-response has not been documented.

Quality control.

Lack of consistent regulations governing the quality of probiotic products has resulted in varying degrees of product quality. Several studies have found discrepancies from what was stated on the product label versus independent testing of the ingredients including lower levels of microbes, not finding the labeled strain or finding contamination of other strains in the product. In one study, of five “Bacillus clausii” products available in India and Pakistan, 80% had lower Bacilli concentrations when assayed than the numbers listed on the label and were also contaminated by non-B. clausii bacteria [59]. Chen et al. assayed 28 commercial probiotic products available in China and could only confirm 53–83% of the species listed on the labels and, of 16 products labeled with a Bifidobacterial species, none were found in any of the products [60]. Toscano et al. tested 24 probiotic products available in Europe and found 42% did not contain the species listed on the label and four products had no living microbes at all in the product [61]. In contrast, Goldstein et al. investigated five over-the-counter probiotic products and found the concentrations of all the microbes listed on the label were accurate [62]. Currently, an European advisory group (ESPGHAN) of experts has recognized the issue of varying quality and has called for improved quality control on all commercial probiotic products [11].

Manufacturer.

Manufacturing processes, stability over time (shelf-life) and the type of formulation has been found to be significant factors in the effectiveness of the final probiotic product [7, 63]. Grzeskowiak et al. tested 13 different “LGG” products available in different formulations (4 capsulated, 2 infant foods, 3 lyophilized, 4 other types) and, while all had detectable L. rhamnosus GG strains identified, 38% of the products appeared to have lost the ability to inhibit pathogens L. rhamnosus GG normally inhibits and differences in the ability to adhere to the colonic mucosa were also observed for the different L. rhamnosus GG products [64]. Perhaps differences were due to manufacturing processes involved in the different formulations, but this was not investigated further. Quality control studies of probiotic products manufactured by established pharmaceutical-grade companies (such as Bio-K+, Florastor, Culturelle, etc.) have confirmed both accurate species/strain identification in the products, and also confirmed the concentration stated on labels and a lack of contamination by other microbial strains not listed on the labels [62, 65,66]. Probiotics sold on-line or over-the-counter that do not provide a manufacturing source or having a less established fermentation quality control history should be viewed with caution.

Safety.

In the majority of the RCTs, patients are followed for any adverse events associated with the probiotic intervention. Although no serious adverse events were associated with most (80%) trials of probiotics, mild-moderate gastrointestinal symptoms (gas, abdominal bloating, etc.) were reported by some trials, while other trials did not report any adverse events [49,51,63]. Many RCTs simply reported ‘no adverse events were noted’, but did not report specific data. This was also noted in a Cochrane review of 384 studies of probiotics, prebiotics and synbiotics that found while 28% of studies reported no adverse events were noted, 37% failed to report any adverse event data in the results sections [67]. In another review investigating 24 trials for the safety of probiotics in variety of different 'at risk' populations (children, pregnant women, elderly, IBD patients, immunocompromised), only two cases of adverse events were associated with probiotics [68]. However infrequent cases of bacteremia or fungemia have been reported in seriously ill patients or neonates who were taking probiotics and thus, the use of probiotics in hospitalized patients with central catheters or who are immunocomprised may be contra-indicated [15, 69].

Practical guidance when choosing a probiotic product.

The choice of the best probiotic for your patient will continue to be a shifting target, as more and more research and clinical trials are done. Data from this literature review indicates the first consideration is the mode of therapy (prevention or treatment), as different probiotics may be more appropriate for each goal (Fig 2). Different choices may also be more appropriate for pediatric versus adult populations. The most suitable single-strain or probiotic mixture may then be selected from the current knowledge and summary of clinical trials (Table 2). Depending upon availability and regulatory oversight in your country, other considerations may also assist the selection. For example, if two probiotic products are available with similar strains, reviewing how well the product complies with required label information might be useful. As shown in Fig 2, probiotics available as dietary supplements should list the FDA disclaimer, manufacturer, probiotic strains in the product, daily dose and comply with FDA regulated health/function claims. Products not listing these components should be viewed with caution. If the probiotic is regulated as a prescription medicine, seek the advice of a pharmacist or physician with knowledge in this field.