Initial therapy affects duration of diarrhoea in critically ill patients with Clostridioides difficile infection (CDI)

We performed a retrospective cohort analysis in all critically ill patients diagnosed with CDI in our tertiary care centre to assess risk factors for mortality as primary end point and outcome depending on choice of CDI therapy. In our study population, 2189 patients who developed diarrhoea (total of 36.477 patients admitted to the 12 ICUs of the university hospital in Hamburg, Germany, during the study period (January 2010 until September 2015)) were analysed. The total number of beds in our ICU department is 140; specialties include 5 interdisciplinary ICUs, 1 surgical ICU, 1 internal medicine ICU, 1 neurological and 1 neurosurgical ICU and 3 cardiovascular ICUs.

Overall, 3188 stool specimens were sent to the microbiology laboratory. Patients tested represented 6.0% of all patients admitted. Ninety-four samples (2.9%) were not processed by the laboratory. C. difficile testing was requested in 2209 samples from 1241/36.477 (3.4%) patients and performed as described below. Testing yielded positive results in 242 (glutamate dehydrogenase (GDH) antigen only) and 179 (GDH antigen and toxin A/B; 8.1% of tested samples) samples. In patients negative for GDH antigen determined by enzyme immunoassay (EIA) and positive toxin A/B (EIA), C. difficile PCR/C. difficile culture was performed. Finally, 144 patients (0.4% of all ICU patients; 6.6% of patients with diarrhoea) were identified as being tested positive for C. difficile (EIA for C. difficile-GDH and toxin A/B) since some had been tested multiple times. We extracted test results and patient characteristics from our hospital and laboratory information systems and performed individual chart reviews for ICU patients with positive stool cultures. If patients had multiple ICU admissions during one hospital stay, only ICU admissions with CDI diagnosis were included in the analysis. Hundred thirty-two patients out of 144 (91.7%) received specific CDI therapy which was initiated after the positive test result. Twelve patients did not receive a specific CDI therapy on the ICU due to instant death after diagnosis (n = 1), decision to limitation of therapy (n = 1), lack of symptoms (n = 2) or diagnosis after dismissal (n = 4). In 4 patients, the reasons for refraining from CDI therapy were unclear (Fig. 1 and Additional file 1: Figure S1). First-line therapy was defined as the CDI-specific antimicrobial treatment for at least 48 h after CDI diagnosis.

This retrospective analysis was performed in accordance to the local regulations of the ethics committee (General Medical Council Hamburg, Ärztekammer Hamburg, reference number WF 11/16).

CDI was defined as a positive stool test for C. difficile GDH and toxin A/B (via EIA) or positive PCR for toxigenic C. difficile in combination with a documentation of matching clinical symptoms (diarrhoea, abdominal discomfort). Time point of CDI diagnosis was defined as the date of receiving the positive stool test result.

A severe episode of CDI was defined by fulfilling any one or more of the following criteria at the time point of diagnosis according to literature: serum creatinine concentration > 1.5 mg/dl and > 15,000 white blood cells per μL according to the clinical practice guidelines by the Infectious Diseases Society of America (IDSA) [17].

For detection of CDI, the C. diff Quick Check Complete EIA (TechLab; Blacksburg, VA, USA) had been used for C. difficile glutamate dehydrogenase antigen (GDH) and toxin A/B testing of non-formed stool samples as recommend by the manufacturer. GDH-positive, toxin A/B-negative samples had been retested by Xpert C. difficile PCR (Cepheid, Sunnyvale, CA).

All continuous variables are reported as median and 25–75% interquartile range (IQR). Categorical variables were compared via chi-square analysis or Fisher’s exact, as appropriate. Metric variables were compared via Mann-Whitney U test. Multivariate logistic regression analysis was performed to assess effect of initial medical treatment on duration of diarrhoea > 5 days. Cox regression proportional hazard analysis was performed to assess predictors of mortality. SPSS 24 for Windows (SPSS, Inc., Chicago, IL) was used for statistical analysis. All p values reported are two sided, and p < 0.05 was considered significant.

The study population was 132 patients treated with CDI-specific antibiotics. Median age of all patients was 70 years (IQR 59–77), and 71.2% were male. 72.7% of the patients needed mechanical ventilation (invasive and non-invasive) during their ICU stay (Table 1); median duration of mechanical ventilation was 11 days (IQR 2–25). 78.8% of the patients received vasopressors for a median of 4 days (IQR 1–12). Renal replacement therapy was necessary in 24.2% of the patients for a median of 9 days (IQR 3–28.5). Main admission diagnoses were sepsis (30.3%) and postoperative admission (29.5%) and cardiac failure (16.7%). In five patients (3.8%), CDI was the leading diagnosis due to severe symptoms; all patients analysed developed CDI during their ICU stay. Detailed patients’ characteristics are illustrated in Table 1.

Median stay on ICU was 14 days (IQR 6–29); overall median hospital stay was 37.5 days (IQR 18–61).

Most patients received proton pump inhibitors (PPI) during ICU stay (95.5%). Furthermore, 25% of all patients were given immunosuppressive drugs, mostly steroids in a dosage > 10 mg/day (21.2%). Other immunosuppressive medications included calcineurin inhibitors (9.1%), mycophenolic acid (3%) and azathioprine (3%).

The median time between ICU admission and diagnosis of CDI was 13.5 days (IQR 5–28) (Table 2). 18.9% (n = 25) of all patients suffered from recurrent CDI. Severe CDI was present in 60.6% of all patients. Toxic megacolon occurred in one patient. Overall, two patients (1.5%) required colectomy due to fulminant CDI.

First-line therapy within 48 h after diagnosis was metronidazole IV in 37.1% (n = 49) of all patients, 35.6% (n = 47) received metronidazole orally and 18.2% (n = 24) were given vancomycin orally. Only 9.1% were treated with a combination therapy (n = 12) initially (Fig. 1).

Severity of CDI did not influence first-line therapy in these patients since equal parts in both patient groups with severe (n = 80, 60.6%) and non-severe CDI (n = 52, 39.4%) were treated with metronidazole IV (37.7 vs. 36.5%), metronidazole orally (33.8 vs. 38.5%), vancomycin orally (20 vs. 15.4%) and combination therapy (8.8 vs. 9.6%) (Additional file 1: Table S1). After 2014, treatment choices started to change; here only 18.9% (n = 7/34) of patients with severe CDI received metronidazole IV.

A detailed overview of all CDI medications over time is illustrated in Additional file 1: Figure S1. Briefly, 62.1% (n = 82/132) received only one substance (metronidazole or vancomycin), whereas 37.9% (n = 50/132) received a combination of drugs, mostly metronidazole IV plus vancomycin orally (15.9%, n = 21/132).

Therapy was adjusted over the course of disease: most patients obtained metronidazole intravenously (IV) (53.8%, n = 71); 45.5% (n = 60) were treated with metronidazole orally while 43.2% (n = 57) received vancomycin orally. Only 2 patients (1.5%) obtained vancomycin per rectum (VPR). Two patients were additionally treated with teicoplanin or fidaxomicin orally, respectively (Additional file 1: Figure S1). Median length of CDI therapy was 8 days (IQR 4–11 days).

Median length of diarrhoea was 5 days (IQR3–8). Half of the patients were suffering from prolonged diarrhoea > 5 days (n = 66, 50%, see Additional file 1: Table S3). First-line metronidazole IV was the only CDI-specific therapy associated significantly with prolonged diarrhoea > 5 days (p = 0.020) as shown in Fig. 2. Furthermore, metronidazole IV as initial therapy was associated with prolonged diarrhoea independently of gender, age, severity of CDI, SAPS on diagnosis and presence of sepsis (OR 2.499, 95% CI 1.150–5.431, p = 0.021) as illustrated in Table 3. Patients treated with metronidazole IV exhibited increased 28-day (34.7 vs. 27.7%, p = 0.141) and 90-day mortality (40.8 vs. 26.5%, p = 0.088) compared to patients treated with a different regime, although not statistically significant. Increased mortality in this patient subgroup rather shows the overall increased morbidity also represented by higher rates of renal replacement therapy (32.7 vs. 19.3%, p = 0.083) and more frequent presence of sepsis (34.7 vs. 27.7%, p = 0.399) (Additional file 1: Table S3).

Twenty-eight- and 90-day mortality rates were 27.3% and 31.8%, respectively. Apart from severity of illness, impaired kidney and liver function, presence of sepsis not related to CDI, neutropenia and necessity of parenteral nutrition were associated with significantly higher mortality rates as illustrated in Table 1 and Additional file 1: Table S2. Patients receiving immunosuppressants (p = 0.007), especially steroids > 10 mg per day (p = 0.002) during their ICU stay, exhibited increased 28-day mortality as illustrated in Table 1. We observed a trend of higher mortality in patients treated with metronidazole IV initially as single-agent therapy (28-day mortality 34.7% and 90-day mortality 40.8%) (Additional file 1: Table S4); otherwise, none of the employed CDI-therapy regimens had a relevant impact on mortality. Furthermore, both patients who received vancomycin enemas died within 90 days due to severe course of disease.

Multivariate Cox regression analysis identified presence of immunosuppressive therapy (HR 2.118, 95% CI 1.068–4.198, p = 0.032) and severity of illness represented by SOFA score (HR 1.216, 95% CI 1.109–1.334, p = < 0.001) as independent predictors of 28-day mortality.