Impact of an automatic hospital probiotic protocol on Clostridioides (Clostridium) difficile infection (CDI) rates and CDI antibiotic usage in a community hospital setting

Abstract

The aim of the present study was to see how widespread preventative use of the probiotic Saccharomyces boulardii via automatic protocol in hospitalised patients receiving antibacterials affected rates of hospital-associated Clostridioides (Clostridium) difficile infection (HA-CDI). Rates of HA-CDI appeared to be similar between the pre-protocol and protocol periods. Use of CDI treatment antibiotics (oral metronidazole and oral vancomycin) was also similar. Laboratory-confirmed isolation of S. boulardii from sterile body sites was identified in five patients during the protocol versus only one case in the pre-protocol years.

Background

Over the past decade, Clostridioides (Clostridium) difficile has become a significant opportunistic pathogen across North America and Europe (Jones et al., 2013). In response, clinicians and governmental organisations have stressed the need for preventative measures. One increasingly popular method of prevention is the prophylactic use of probiotics in patients taking antibacterials (Evans and Johnson, 2015; Yi et al., 2016; Evans and Safdar, 2015). Several small studies and meta-analyses from across the globe have reported a potential C. difficile infection (CDI) preventative effect with probiotic use when given to patients receiving antibacterials (Johnston et al., 2012, 2018; McFarland, 2006). In an effort to prevent CDIs, our hospital developed an ‘automatic’ probiotic protocol, which essentially provided probiotic therapy to all admitted adult patients receiving antibacterials. The probiotic used in this study was Saccharomyces boulardii. This probiotic was chosen because it was the most commonly used and it is not likely to be killed by concomitant antibacterial therapy (Yi et al., 2016). The objective of this study was to see how widespread preventative use of S. boulardii in hospitalised patients receiving antibacterials affected rates of hospital-associated CDI (HA-CDI).

Methods

St. Clair Hospital is a 330-bed, advanced suburban hospital serving about 500,000 residents in the Pittsburgh, Pennsylvania region of the United States. This study was an observational assessment of the automatic protocol with the primary outcome being a comparison of rates of HA-CDI of the three-year periods before and after the automatic protocol implementation. The pre-protocol period was 2011–2013 and the protocol period was 2014–2016. The automatic probiotic protocol was triggered by all oral and intravenous antibiotic orders placed in the hospital’s electronic health record system (Eclipsys Sunrise®) for adults aged ⩾ 18 years, who were not neutropenic and were expected to receive antibiotic therapy for at least 72 h. Patients had to be able to take the oral probiotic capsules. The probiotic used was S. boulardii (Florastor® - Biocodex), at a dose of 500 mg (two 250-mg capsules) orally twice daily. The protocol was well-explained to providers (prescribers), nurses and pharmacists, who were advised to inform patients of the potential benefit of the probiotic. All probiotic doses appeared on the patient’s daily medication administration report (MAR), so compliance was expected to be very high. Capsules could be taken with or without meals, but had to be swallowed whole and could not be crushed or opened.

In this quality improvement project, we could not capture individual patient-level data given limited hospital computer decision support resources available for research. It is important to point out that there were no other major changes in CDI infection control efforts or lab diagnostics during this time. CDI diagnosis followed the U.S. Centers for Disease Control & Prevention (CDC) definition (https://www.cdc.gov/hai/prevent/cdi-prevention-strategies.html). Laboratory assessment used Cepheid ® (Sunnyvale, CA, USA), a PCR-toxin based assay and culture testing. Rates of HA-CDI infection were determined using the CDC hospital-associated infection definitions (onset after three days) and were expressed per 1000 patient-days. Overall rates of CDI were assessed using ICD-9 code (008.45) for admitted patients. In addition, use of CDI treatment agents (oral vancomycin and oral metronidazole) expressed in defined daily doses (DDD)/1000 patient-days was compared during both phases of the study. Adult census estimates were used to calculate all patient-day denominator data. Patients who developed positive Saccharomyces sterile site cultures were identified by laboratory. Biostatistical analysis was conducted using JMP, version 13 Software (SAS Institute, Cary, NC, USA). A Poisson regression model was utilised for comparisons before and after protocol implementation. This study was granted exempt status by the West Virginia University Institutional Review Board (IRB).

Results

Table 1 highlights the occurrences of CDIs and Saccharo-myces culture isolations before and during the probiotic prophylaxis protocol. Rates of HA-CDI appeared to be similar between the pre-protocol and protocol periods, with average yearly rates of 0.988 and 0.895 cases/1000 patient-days, respectively (P = 0.521; Figure 1). Overall cases of CDI which included patients with community-onset (ICD-9 code based) CDI were not statistically different with average rates of 2.582 and 3.487 cases/1000 patient-days in the pre-protocol and protocol periods, respectively (P = 0.165). Use of CDI treatment antibiotics (oral metronidazole and oral vancomycin) was similar overall when comparing three years before protocol to three years under protocol (P = 0.633). Usage rates did not differ for the individual agents, oral metronidazole 14.020 vs. 10.757 DDD/1000 patient-days (P = 0.269) and oral vancomycin 38.743 vs. 39.14 DDD/1000 patient-days (P = 0.938).

Table 1.

Occurrences of C. difficile infections and Saccharomyces culture isolations before and during probiotic prophylaxis protocol.

	Pre-protocol years	During-protocol years	P value
CDI diagnoses (yearly mean)
HA-CDI	69.667	60.000	—
ICD9-CDI	179.667	233.333	—
CDI diagnoses per census (yearly mean)
HA-CDI / 1000 PD	0.988	0.895	0.521
ICD9-CDI /1000 PD	2.582	3.487	0.165
Saccharomyces isolation (number of unique isolates)
Positive culture in any source	2	27	0.022
Positive culture from sterile body fluid	1	5	0.035

CDI, C. difficile infection; HA-CDI, hospital-associated CDI; ICD9-CDI, patients with diagnostic codes for CDI; PD, patient-days.

Figure 1.

Case rates of CDI using ICD-9 and CDC HAI definitions.

The average yearly cost of Saccharomyces capsules during the protocol was $63,000. That translates to about 78,750 probiotic doses/year. This study was a retrospective analysis of a quality improvement project, so prospectively determined screening cultures were not planned. All cultures positive for Saccharomyces were identified through routine cultures that were ordered based on specific disease (presentation) diagnostic assessments. Saccharomyces was recovered in microbiological cultures from two patients before protocol and in 27 patients during the protocol years. Isolation of Saccharomyces from sterile body sites was identified in five patients during the protocol versus only one case in the pre-protocol years (P = 0.035). The sterile sites infected during the protocol were the peritoneal space and blood. All patients with Saccharomyces in these sterile body sites were deemed clinically significant and received antifungal therapy.

Discussion

The automatic protocol was not associated with lower rates of CDI in this quality improvement project. Nor did the use of probiotics appear to affect usage of CDI treatment antibacterials. Recently, other researchers failed to find lower rates of CDI with the Bio-K+ probiotic (Allen et al., 2013; Box et al., 2018). These findings do not necessarily mean that the probiotics did not help some individuals. Sample size may have affected studies like these which included many patients who were not at high risk. Although uncertain about the total exposure of our patients, we did have substantial use of the probiotic during the protocol, as most adult patients admitted to the hospital receiving antibacterials for a three-year period received the probiotic. The inclusion of low-risk patients was suggested as a limitation in other studies not showing a preventive effect. The best examples can be seen in the PLACIDE study conducted in the UK and in many of the small studies included in the various meta-analyses (Allen et al., 2013; Box et al., 2018; Johnston et al., 2018). In addition to having a large proportion of low-risk patients, the PLACIDE study also allowed patients to start probiotics up to a week after the start of antibacterials (Allen et al., 2013). Starting probiotics near the start of antibacterial therapy was previously associated with a beneficial effect on the prevention of CDI (Shen et al., 2017). Our automatic protocol was triggered with the initial order of antibiotics to allow for the earliest initiation of probiotic.

It is unfortunate that we could not assess the desired patient-level data, which would have included patients’ use of specific antibacterials and proton pump inhibitors. Nonetheless, there may have been some protective effect that was not found during our project, as the rates of HA-CDI did not increase with use of the probiotic protocol during a time when CDI was possibly increasing in our region of the country (Barrett and Owens, 2018).

Data from the previous meta-analyses also suggests that there may be a more preventative effect with multispecies probiotics rather than single agents, like S. boulardii (Johnston et al., 2012, 2018). This may be due to increased diversity and the inclusion of symbiotic organisms with complementary functionality. As the science of the microbiome evolves, we imagine that better formulations of probiotic-like agents will be assessed for prevention of CDI. The potential cost and safety of probiotics must also be considered. These non-prescription agents can, on rare occasion, cause infection. Our unique dataset certainly adds to the current patchwork of information on this very complex relationship. Obviously more rigorous assessments, including risk-adjustments and use of more symbiotic combination probiotics, are needed.