Introduction
In patients with decompensated cirrhosis, spontaneous bacterial peritonitis (SBP) is a common and serious complication with an annual incidence of 1.5%–3.5% in outpatients and 10%–30% in hospitalized patients, Reference Rimola, Soto, Bory, Arroyo, Piera and Rodes1,Reference Fernández, Navasa and Gómez2 as well as a 66.2% mortality rate within one year of the first SBP episode. Reference Arvaniti, D’Amico and Fede3 Antibiotics are often started empirically, but accurate microbiological diagnosis is critical for appropriate treatment given the illness severity of SBP.
Historically, ascites cultures were performed by plating ascitic fluid onto various agars (Blood, MacConkey, and Chocolate) and inoculating into broths (eg, Schaedler, thioglycolates)—with variations by institution and period. Reference Runyon, Canawati and Akriviadis4–Reference Such and Runyon10 However, cultures were negative for these conventional methods more than a third of the time. Reference Runyon, Canawati and Akriviadis4,Reference Runyon and Hoefs7 It was hypothesized that utilizing standard nutrient broth-containing blood culture bottles would improve culture growth rates as they are optimized for low bacterial concentration fluids, as seen in both bacteremia and SBP. Luce et al. demonstrated in patients receiving peritoneal dialysis with peritonitis that direct inoculation into Bactec© blood culture bottles led to higher rates of microbiologic diagnosis when compared to conventional methods (95.9% vs 77.6%; P < 0.01). Reference Luce, Nakagawa, Lovell, Davis, Stinebaugh and Suki8 Runyon et al. found similar results in patients with SBP and culture-negative neutrocytic ascites (CNNA). They demonstrated improved ascites culture yield with direct inoculation of blood culture bottles in a quasi-experimental trial (91% vs 42% for conventional; P < 0.01) Reference Runyon, Umland and Merlin9 and confirmed these findings in a subsequent study with concurrent controls (93% vs 43%; P < 0.0001). Reference Runyon, Canawati and Akriviadis4 Another prospective study by Bobadilla et al. reported similar findings (81% vs 52%; P < 0.05). Reference Bobadilla, Sifuentes and Garcia-Tsao5 Siersema et al. further compared direct blood culture inoculation, the conventional method, and a lysis centrifugation method in a prospective cohort, demonstrating superior culture yield with direct inoculation (79% vs 33% vs 46%, respectively; P < 0.05). Reference Siersema, de Marie, van Zeijl, Bac and Wilson6
All studies in the extant literature showed higher culture yield for direct inoculation of blood culture bottles compared to conventional culture methods, with absolute improvement in culture positivity ranging from 22% to 50%. Reference Runyon, Canawati and Akriviadis4–Reference Siersema, de Marie, van Zeijl, Bac and Wilson6,Reference Runyon, Umland and Merlin9 However, several limitations may decrease their ecological validity and generalizability. First, all had small sample sizes (n = 23–31). Second, most excluded patients with negative cultures and an absolute neutrophil count (ANC) in the 250–500 cells/mL range, which in current medical practice is treated as SBP. Thus, the results do not fully translate to the target population or standard clinical practice. Third, all the studies excluded patients who had received recent antibiotics, which again does not reflect clinical practice where immediate paracenteses are not always performed, and antibiotics are often started empirically prior to ascites culture collection. Last, none of the studies reported demographic information or clinical outcomes of included patients, limiting the generalizability of their findings.
This study aimed to address the above limitations while evaluating if direct inoculation of ascites fluid into bioMérieux© blood culture bottles would increase the percentage of positive ascites cultures in patients with SBP compared to the conventional method.
Methods
Study design
This pre- and post-intervention study was completed at Baylor St. Luke’s Medical Center in Houston, Texas, United States—a quaternary academic medical center. The study population included all adult patients with SBP in the inpatient or emergency department settings between November 2020 and December 2022. SBP was defined as an ascitic fluid ANC ≥ 250 cells/mL in patients with cirrhosis and no identifiable causes of secondary peritonitis. It was approved by the Institutional Review Board of Baylor College of Medicine (IRB H-51376).
Intervention
This study examined the effect of changing the method of ascites culture collection on ascites culture positivity. In the pre-intervention period, ascitic fluid was collected in a sterile container and centrifuged at 1500 revolutions per minute for 20 minutes. The sediment was resuspended in 1.5 mL of supernatant and plated onto Blood, Chocolate, MacConkey agars, and Schaedler broth. Plates were incubated for 48 hours at 34–36°C and 5% CO2 (Blood and Chocolate) or 48 hours at 34–36°C in an atmospheric incubator (MacConkey and Schaedler)—initial read for culture growth occurred 18–24 hours post-plating. Cultures were finalized if no growth was observed by 48 hours.
In the post-intervention period, ascitic fluid was directly inoculated into one each of bioMérieux© aerobic and anaerobic blood culture bottles at the bedside and, upon laboratory receipt, immediately incubated inside the blood culture BacT/Alert VIRTUO detection system (BioMérieux©) for 48 hours. The bottle(s) were removed upon microbial growth, and blood culture broth was plated onto Blood, Chocolate, and MacConkey agars. The medical staff was informed of the change in ascites culture collection methodology via printed and electronic memos and face-to-face communication, along with a notification that the microbiology laboratory would reject improperly collected samples.
Patient identification and chart review
All ascites cultures in the study period were identified via the Reporting Workbench (RWB) in Epic (Epic Systems Corporation, Verona, Wisconsin, United States). Ascites cultures were selected if classified as “Body Fluid Culture” with a reported source of “ascites,” “ascitic,” “paracentesis,” or “abdomen.” Cultures were included if the ascites fluid was collected between November 1, 2020 to October 31, 2021 (pre-intervention), or December 1, 2021 to December 31, 2022 (post-intervention) and had an ANC ≥ 250 cells/mL. November 2021 was excluded as a washout period as it was the intervention implementation month. Patients were additionally excluded if they had a prior included ascites culture during the study period, ascites from an etiology other than cirrhosis, peritoneal dialysis or indwelling peritoneal catheter, abscess or fluid collection on imaging, or a surgical finding of a perforated viscus.
The primary outcome was the percentage of positive ascites cultures. Secondary outcomes were days of antimicrobial therapy (DOT), time to ascites culture positivity, days of hospitalization, 30-day readmission rate, and 30-day mortality rate.
DOT was defined as the aggregate sum of days of antimicrobials given to an individual patient as documented within the electronic health record (EHR), within either 72 hours prior to ascites culture collection or up to 7 days after and excluding antibiotics administered for prophylaxis. Time to ascites culture positivity was defined as the time from ascites culture collection to the time of first antimicrobial growth.
Patient demographics and clinical characteristics were collected from the EHR. Demographics collected were age, gender, race, and ethnicity. Clinical features included cirrhosis etiology, the MELD (model for end-stage liver disease, Organ Procurement and Transplant Network) score, and SBP risk factors including SBP prophylaxis before ascites culture, history of SBP, history of variceal hemorrhage, ascitic fluid total protein <1 g/dL, and the use of proton pump inhibitors within 30 days.
Confounding variable data collected included pre-culture antibiotic administration, location of paracentesis, and culture bottle inoculation volumes. Pre-culture antibiotics and the location of paracentesis were collected from the EHR. The locations of paracentesis were the emergency department, interventional radiology (IR) suite, and the intensive-care unit. Data for inoculation volumes were collected from all ascites cultures obtained post-intervention from December 2021 to May 2022, including patients with and without SBP, via the bioMérieux© instrument data management platform (MYLA). Based on manufacturer recommendations, inoculation volumes were categorized as underfilled (<8 mL), appropriately filled (8–10 mL), or overfilled (>10 mL). Reference Runyon, Canawati and Akriviadis4–Reference Siersema, de Marie, van Zeijl, Bac and Wilson6,Reference Runyon, Umland and Merlin9 Any missing volumes were excluded from the analysis.
Organisms from positive ascites cultures were recorded and EHR analyzed by the primary author (TB) to determine if antibiotic regimens for SBP were escalated, de-escalated, or unchanged by culture results. Charts were reviewed for antibiotic changes which occurred after culture results but within 7 days of culture collection and which were documented to be addressing SBP. The final antibiotic regimen was recorded for purposes of determining escalation or de-escalation, in cases where multiple changes occurred. Escalation referred to the addition of more antibiotics (eg, adding vancomycin to an initial regimen of ceftriaxone only) or broadening spectrum of coverage (eg, changing ceftriaxone to cefepime or ertapenem). De-escalation referred to the removal of antibiotics or narrowing spectrum of coverage.
Statistical analysis
Power analysis was performed prior to data collection; incidence rates for ascites culture positivity were estimated to be 20% for pre-intervention versus 50% for post-intervention collection methods, respectively. The difference of 30% was a conservative estimate based on prior reported data. Reference Runyon, Canawati and Akriviadis4–Reference Siersema, de Marie, van Zeijl, Bac and Wilson6,Reference Runyon, Umland and Merlin9 Given an alpha level of 0.05 and a power of 80%, 44 patients in the pre-intervention group and 35 in the post-intervention group were required.
The Welch’s t test (or Wilcoxon rank sum test as appropriate) was used for continuous variables, while χ2 (or Fisher’s exact as appropriate) was used for categorical variables. Five multiple regression models (Model 1-Model 5) were employed to assess factors associated with the study outcomes (culture positivity, time to positivity, days of hospitalization, 30-day readmission rate, and mortality). Independent factors included the study intervention, patient demographics, clinical characteristics, and paracentesis location. Patients with missing data were excluded from the regression models. Linear regression was used for continuous outcomes (Model 2, Model 3), while logistic regression was used for dichotomous outcomes (Model 1, Model 4, and Model 5). Fitness was assessed via R2/R2 adjusted for linear regressions (Nagelkerke R2 if negative) and Tjur R2 for logistic regressions. Significance was defined as a P value <0.05. Statistical analysis was performed using R version 4.2.1 (R Foundation for Statistical Computing, Vienna, Austria).
Results
Patient screening
A total of 2,928 ascites culture samples were identified from November 2020 to December 2022. Of these, 582 had an ANC ≥ 250 cells/mL. 468 of these 582 were then excluded (Figure 1). 114 patients with SBP remained and were included in the subsequent analysis (61 pre-intervention and 53 post-intervention).
Figure 1. Study flow chart.
SBP = spontaneous bacterial peritonitis.
Patient characteristics
Patient characteristics are presented in Table 1. There were no statistical differences between pre- and post-intervention groups in terms of age, gender, ethnicity, or race. There was a statistically higher proportion of patients with ethanol-induced cirrhosis in the pre-intervention group compared to the post-intervention group (65.6% vs 41.5%, P = 0.010). There were no statistical differences between pre- and post-intervention groups regarding SBP risk factors, SBP prophylaxis, or MELD scores.
Table 1. Characteristics of patients with spontaneous bacterial peritonitis
Q1-Q3 = Quartile 1 to Quartile 3; SBP = spontaneous bacterial peritonitis; PPI = proton pump inhibitor; MELD = model for end-stage liver disease (Organ Procurement and Transplant Network model).
a For statistical comparison purposes, races were combined into White or Caucasian versus Non-White or Caucasian due to low sample size.
b For purposes of statistical comparison, cirrhosis etiologies were combined into cirrhosis secondary to ethanol versus cirrhosis not secondary to ethanol (due to low sample size). If a patient had a cirrhosis etiology attributed to more than one etiology (eg, hepatitis C and ethanol), they were counted as one patient in each category.
Therapeutic antibiotics (not prescribed for SBP prophylaxis) were administered before ascites culture collection in 62.3% vs 60.4% of the pre- and post-intervention group, respectively (P = 0.834). There was no difference in the proportions of paracentesis locations between groups (Table 1, P = 0.584).
Primary outcome
Ascites culture positivity was 15.8% (18/114) for the entire cohort and was similar between pre- and post-intervention groups (Table 2, 11.5% [7/61] vs 20.8% [11/53], P = 0.205). After adjusting for confounders, there was a trend toward significance in ascites culture positivity after the intervention (Supplemental Table 1, Model 1, P = 0.077). Variables significantly associated with ascites culture positivity included the location of paracentesis (non-IR locations vs IR, P = 0.029), prior variceal hemorrhage (P = 0.044), and ascitic fluid protein ≤ 1 g/dL (P = 0.004).
Table 2. Primary and secondary outcomes of patients with spontaneous bacterial peritonitis (Univariate analysis)
Q1-Q3 = Quartile 1 to Quartile 3.
a n = 7 and 11 for pre- and post-intervention groups, respectively, for this secondary outcome as it only compares patients with positive cultures.
There were 48 positive cultures out of the 1704 samples collected post-intervention from December 2021 to May 2022. These positive samples were classified as underfilled (54.2%), appropriately filled (14.6%), and overfilled (31.3%). The median inoculation volume was 7 mL for positive cultures (IQR 5–12 mL) versus 11.0 mL for negative cultures (IQR 10–13 mL, P < 0.001).
Secondary outcomes
The time to positivity was longer in the post-intervention period (25.9 hours pre-intervention vs 41.5 hours post-intervention, P = 0.057, Table 2) but did not achieve statistical significance. Similarly, no statistical difference was observed in the length of stay or 30-day readmission between the pre- and post-intervention groups (Table 2). Results were similar after adjusting for confounders (Supplemental Table 1, Models 2, 3, and 4). We could not report the regression results for mortality (Model 5) as the model did not converge due to the small number of patients who met this outcome.
Microbiological results
Organisms for the 18 positive cultures are reported in Table 3. Notably, 33.3% (6/18) of positive cultures grew organisms for which ceftriaxone (commonly recommended empiric therapy for SBP) is not recommended. There was no statistical difference between those with positive versus negative culture results in proportion of antibiotic regimen adjustment (p = 0.176, Table 4). Regimen changes are detailed in Supplemental Table 2.
Table 3. Organism and antibiotic regimen adjustment in patients with spontaneous bacterial peritonitis and positive ascites cultures
ESBL = extended spectrum beta-lactamase producing. Escalation refers to the addition of more antibiotics or change to a broader spectrum of coverage, while de-escalation refers to the removal of antibiotics or change to a narrower spectrum of coverage.
Table 4. Antibiotic regimen adjustment in spontaneous bacterial peritonitis patients with positive versus negative cultures
Cultures refer to ascites cultures. Escalation refers to the addition of more antibiotics or change to a broader spectrum of coverage, while de-escalation refers to the removal of antibiotics or change to a narrower spectrum of coverage.
Discussion
This study demonstrated an absolute increase in culture positivity (9.3%) with a trend toward statistical significance after direct inoculation of ascites fluid into blood culture bottles. There are several plausible explanations for why statistical significance was not observed compared to other studies. Reference Runyon, Canawati and Akriviadis4–Reference Siersema, de Marie, van Zeijl, Bac and Wilson6,Reference Runyon, Umland and Merlin9 First, this study may have been underpowered. Although the sample size was higher than the projected amount by power analysis, percentage of positive cultures were significantly lower for both methods than the reported literature. Reference Runyon, Canawati and Akriviadis4–Reference Siersema, de Marie, van Zeijl, Bac and Wilson6,Reference Runyon, Umland and Merlin9 Thus, the true difference in culture positivity in this population may be lower than the estimated 30%.
The lower culture positivity percentage, in turn, may be due to the higher rates of pre-culture antibiotic treatments (61.4% vs 0% in the three studies that commented on pre-antibiotic use). Reference Runyon, Canawati and Akriviadis4,Reference Siersema, de Marie, van Zeijl, Bac and Wilson6,Reference Runyon, Umland and Merlin9 It also may be due to a difference in the underlying prevalence of SBP—this study set a minimum ANC of 250 for both SBP and CNNA, less specific than the minimum ANC of 500 set in prior studies for CNNA. Reference Runyon, Canawati and Akriviadis4–Reference Siersema, de Marie, van Zeijl, Bac and Wilson6,Reference Runyon, Umland and Merlin9 This ANC cutoff was chosen in line with American Association for the Study of Liver Diseases (AASLD) guidelines Reference Biggins, Angeli and Garcia-Tsao11 but does increase the likelihood of inclusion of patients without microbiological evidence of SBP. It could also be due to finalizing cultures at 48 hours and missing slower growing organisms.
An additional reason for the lack of significant improvement with bedside inoculation of blood culture bottles could be improper sample collection. When using bioMérieux© blood culture bottles, the manufacturer reports that the optimal inoculation volume is 8–10 mL, and the culture yield decreases when filled past 10 mL. In a larger sample set looking at all ascites cultures collected in a 6-month period (not exclusively patients with SBP), only 22.1% of culture bottles had volumes in the optimal range. When comparing positive and negative cultures, there was, on average, significantly lower inoculation volume in positive cultures. This suggests a role of human factors that should be addressed to optimize positivity rates.
For the secondary outcomes, no significant differences were observed. This could be related to the absence of a statistically significant change in the primary outcome or the study’s lack of power to detect differences in secondary outcomes. Of note, there was a trend toward increased time to positivity with an average increase of ∼16 hours, although this was not significant on the initial analysis (P = 0.057) or after adjusting for the other independent variables (P = 0.182). This difference may be because bioMérieux© recommends the addition of defibrinated horse blood for non-blood culture samples, which was not part of our institution’s lab protocol at the time of this study. Notably, this additive assists in the growth of fastidious organisms (eg, Haemophilus influenza) and it is unclear whether it impacts bacterial growth times and yield of non-fastidious organisms. Additionally, there may be a delay in the inoculation of ascitic fluid into the blood culture bottles at the bedside, depending on procedural logistics. Unfortunately, this was not a reported measurement that could be analyzed at our institution.
A proposed benefit of increased culture yield is the appropriate tailoring of antibiotic regimens, both in terms of de-escalation (for antibiotic stewardship) and escalation (to prevent progression of the infection). We did not demonstrate a statistically significant change in antibiotic regimen de-escalation. However, we were not powered for this outcome and the 18.4% absolute increase in de-escalation (53.6% relative increase) in culture-positive patients indicates there may be some validity to this concept. Cultures also captured a significant portion of organisms resistant to standard empiric therapy (33.3%), all of which were adjusted to an appropriate regimen after culture results. Our study has several limitations. First, it has a quasi-experimental approach without a control group, making it difficult to assess the role of external factors that could have confounded the results. We have attempted to adjust for that by including patients’ demographics and characteristics in multiple regression models. Second, examination of confounding variables revealed a large proportion of ascites cultures were inoculated with inappropriate volumes (77.9%), which can be reasonably presumed to lower the percentage of positive cultures. Last, as a single-center study, it may not accurately reflect patient populations of other practice settings.
This study has several advantages. First, the sample size is substantially larger than most studies reported in the literature (n = 114 vs n = 23–31). Reference Runyon, Canawati and Akriviadis4–Reference Siersema, de Marie, van Zeijl, Bac and Wilson6,Reference Runyon, Umland and Merlin9 Second, the study population more accurately reflects patients observed in clinical practice. Specifically, it includes patients with CNNA and an ANC of 250–500 cells/uL—an SBP variant treated the same as culture-positive SBP but excluded from prior studies. Our study also has a high proportion of patients treated with antibiotics pre-culture collection (61.4% vs 0% in other studies), Reference Runyon, Canawati and Akriviadis4,Reference Siersema, de Marie, van Zeijl, Bac and Wilson6,Reference Runyon, Umland and Merlin9 which, due to the morbidity associated with delay of antibiotics and logistic limitations for obtaining an expedient paracentesis, is likely a factor that applies to other healthcare systems as well. We also adjusted for demographics and clinical characteristics. Although there were no noteworthy differences between pre- and post-intervention groups, these data are absent from prior related studies. Reference Runyon, Canawati and Akriviadis4–Reference Siersema, de Marie, van Zeijl, Bac and Wilson6,Reference Runyon, Umland and Merlin9 Thus, this study provides novel data that may be used as a baseline for comparison in future investigations. For outcomes in particular, prior studies imply that improved microbiologic diagnosis will improve patient-centered outcomes, but this has yet to be explicitly demonstrated. This study reports no improvements in clinically relevant outcomes.
Our study is the first to demonstrate no statistically significant improvement in ascites culture positivity with a change to direct inoculation of blood culture bottles at the bedside. This finding may be due to procedural errors (eg, inappropriate inoculation volumes) or the study being underpowered (for a population with ∼60% pre-culture antibiotics), but both possibilities highlight important considerations for any institution planning to transition methods of ascites culture collection. Our results suggest that if direct inoculation improves culture yields, the effect size is significantly smaller than previously reported when applied to realistic clinical settings. Future investigations are needed to optimize ascitic culture yield and determine if this improves patient- and hospital-centered outcomes.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/ash.2024.84
Data availability
Based on our IRB protocol, data collected and utilized in this study will not be publicly available.
Acknowledgements
None.
Financial support
None reported.
Potential competing of interests
All authors report no conflicts of interest relevant to this article.
Open access
