Prevalence of toxigenic C. difficile carriers on hospital admission
During the study period, there were 1,045 older patients (≥ 65 years) out of 1,586 patients admitted to the Division of Infectious Diseases, Department of Internal Medicine, at our hospital. Of 1,045 patients, PCR screening for toxigenic C. difficile infection was conducted in 628 (60.1%) (Fig. 1). PCR screening was not performed in 447 patients because stool samples or rectal swab specimens could not be collected within 48 h of admission. Some patients refused rectal swabs or stool examinations.
Of the 628 patients, 101 (16.1%) were toxigenic C. difficile carriers. The monthly prevalence of toxigenic C. difficile carriage at admission ranged from 4.0 to 15.9 per 10,000 patient-days, with an average of 8.8 per 10,000 patient-days during the study period (Fig. 2). The demographic and clinical characteristics of the screened patients are shown in Table 1. Of 101 patients with positive PCR test results, 55 (54.5%) were diagnosed with symptomatic CDI (45 healthcare-associated and 10 community-acquired CDIs) and received antibiotic therapy for CDI. In particular, nine patients were diagnosed with symptomatic CDI at admission, and 24, including these patients, started CDI treatment within 48 hours of admission.
Table 1
Comparison of demographic and clinical characteristics between toxigenic Clostridioides difficile carriers and toxigenic C. difficile non-carriers at the time of hospital admission in the derivation group
| Total (n = 202) | CD toxin negative (n = 101) | CD toxin positive (n = 101) | P-value |
Age (years), median (IQR) | 79 (73–85) | 78 (72–84) | 80 (74–85) | 0.357 |
Male, n (%) | 99 (49.0) | 47 (46.5) | 52 (51.5) | 0.482 |
Admission route, n (%) | | | | |
Emergency room | 182 (90.1) | 91 (90.1) | 91 (90.1) | 1.000 |
Outpatient clinic | 20 (9.9) | 10 (9.9) | 10 (9.9) | 1.000 |
History of CDI before admission, n (%) | 3 (0.01) | 0 (0) | 3 (3.0) | - |
Pre-admission route, n (%) | | | | < 0.001 |
Home | 104 (51.5) | 66 (65.3) | 38 (37.6) | < 0.001 |
Acute care hospital | 38 (18.8) | 16 (15.8) | 22 (21.8) | 0.280 |
Nursing hospital | 44 (21.8) | 11 (10.9) | 33 (32.7) | < 0.001 |
Nursing facility | 16 (7.9) | 8 (7.9) | 8 (7.9) | 1.000 |
MDR acquisition during admission, (%) | | | |
CRE | 7 (3.5) | 2 (2.0) | 5 (5.0) | 0.248 |
VRE | 20 (9.9) | 9 (8.9) | 11 (10.9) | 0.638 |
CRAB | 29 (14.4) | 7 (6.9) | 22 (21.8) | 0.003 |
CRPA | 15 (7.4) | 6 (5.9) | 9 (8.9) | 0.421 |
MRSA | 42 (20.8) | 15 (14.9) | 27 (26.7) | 0.037 |
Total MDROs | 69 (34.2) | 24 (23.8) | 45 (44.6) | 0.002 |
Infectious diseases at the time of admission, n (%) | | | |
Urinary tract infections | 132 (65.3) | 57 (56.4) | 75 (74.3) | 0.008 |
Pneumonia | 95 (47) | 39 (38.6) | 56 (55.4) | 0.017 |
Skin and soft tissue infections | 16 (7.9) | 9 (8.9) | 7 (6.9) | 0.602 |
Central nervous system infections | 6 (3) | 1 (1.0) | 5 (5.0) | 0.097 |
Bone and joint infections | 17 (8.4) | 3 (3.0) | 14 (13.9) | 0.005 |
Intra-abdominal infections | 68 (33.7) | 29 (28.7) | 39 (38.6) | 0.137 |
Septic shock | 58 (28.7) | 18 (17.8) | 40 (39.6) | 0.001 |
Comorbidities, (%) | | | | |
Cardiovascular diseases | 157 (77.7) | 79 (78.2) | 78 (77.2) | 0.866 |
Neurologic diseases | 107 (52.9) | 46 (45.5) | 61 (60.4) | 0.034 |
Malignant diseases | 39 (19.3) | 19 (18.8) | 20 (19.8) | 0.859 |
Trauma | 26 (12.8) | 11 (10.9) | 15(14.9) | 0.401 |
Chronic renal diseases | 61 (30.1) | 36 (35.6) | 25 (24.8) | 0.092 |
Chronic liver diseases | 26 (12.8) | 12 (11.9) | 14 (13.9) | 0.674 |
Chronic pulmonary diseases | 63 (31.1) | 23 (22.8) | 40 (39.6) | 0.010 |
Connective tissue diseases | 9 (4.4) | 1 (1.0) | 8 (7.9) | 0.017 |
Metabolic diseases | 101 (5) | 48 (47.5) | 53 (52.5) | 0.482 |
Haematologic diseases | 145 (71.7) | 60 (59.4) | 85 (84.2) | < 0.001 |
Charlson’s comorbidity score, median (IQR) | 3 (2–4) | 4 (3–5) | 2 (1–4) | < 0.001 |
Predisposing factors within 30 days, n (%) | | | |
Recent surgery | 16 (7.9) | 4 (4.0) | 12 (11.9) | 0.037 |
Recent admission | 93 (46) | 29 (28.7) | 64 (63.4) | < 0.001 |
Use of corticosteroids | 36 (17.8) | 16 (15.8) | 20 (19.8) | 0.462 |
Use of antibiotics | 97 (48) | 30 (29.7) | 67 (66.3) | < 0.001 |
Intensive care unit stay | 35 (17.3) | 13 (12.9) | 32 (31.7) | 0.001 |
Foley catheterization | 85 (42) | 29 (28.7) | 56 (55.4) | < 0.001 |
Central venous catheterization | 43 (21.2) | 15 (14.9) | 28 (27.7) | 0.025 |
Nasogastric tube | 61 (30.1) | 21 (20.8) | 40 (39.6) | 0.004 |
Percutaneous drainage, | 29 (14.3) | 11 (10.9) | 18 (17.8) | 0.160 |
Mechanical ventilation | 17 (8.4) | 4 (4.0) | 13 (12.9) | 0.023 |
Tracheostomy | 16 (7.9) | 6 (5.9%) | 10 (9.9%) | 0.297 |
Haemodialysis | 7 (3.4) | 1 (1.0) | 6 (5.9) | 0.054 |
Status of bed-ridden status | 78 (38.6) | 28 (27.7) | 50 (49.5) | 0.001 |
Sore sites | 67 (33.1) | 21 (20.8) | 46 (45.5) | < 0.001 |
Proton-pump inhibitors | 95 (47) | 31 (30.7) | 64 (63.4) | < 0.001 |
Use of probiotics | 80 (39.6) | 21 (20.8) | 59 (58.4) | < 0.001 |
Anaemia (haemoglobin ≤ 10 g/dL) | 131 (64.9) | 29 (28.7%) | 64 (63.4) | < 0.001 |
Length of hospital stay, (days), median (IQR) | 14 (9–24) | 11 (7–18) | 16 (10–28) | 0.236 |
In-hospital mortality, n (%) | 14 (6.9) | 6 (5.9) | 8 (7.9) | 0.580 |
CD, Clostridiodes difficile; CDI, Clostridiodes difficile infection; CRAB, carbapenem-resistant Acinetobacter baumannii; CRE, carbapenem-resistant Enterobacteriaceae; CRPA, carbapenem-resistant Pseudomonas aeruginosa; IQR, interquartile range; MDR, multidrug-resistant; MDROs, multidrug-resistant microorganisms; MRSA, methicillin-resistant Staphylococcus aureus; VRE, vancomycin-resistant enterococci |
Construction Of The Clinical Prediction Model
Patients with toxigenic C. difficile infection were more likely to reside in a facility than at home and be exposed to medical procedures 30 days before admission than those without toxigenic C. difficile infection (Table 1). Carriers had more neurological, chronic pulmonary, connective tissue, and hematologic diseases than non-carriers (Table 1). A history of CDI 90 days before admission was available in only three patients with toxigenic C. difficile infection. Multidrug-resistant microorganisms, such as carbapenem-resistant Acinetobacter baumannii or methicillin-resistant Staphylococcus aureus were more commonly isolated from clinical samples during hospitalization in carriers than in non-carriers (Table 1).
In the multivariable logistic regression model, septic shock, connective tissue diseases, anaemia, recent use of antibiotics, and recent use of proton-pump inhibitors were significant risk factors associated with toxigenic C. difficile infection at the time of admission in older patients admitted to the Division of Infectious Diseases, Department of Internal Medicine (Table 2).
Table 2
Multivariable logistic regression analysis of risk factors associated with toxigenic Clostridioides difficile carriage at the time of hospital admission in the derivation group (final model)
Risk factors | ß-coefficient | Standard error | Odds ratio | 95% confidence interval | P-value |
Intercept | -1.93 | 0.35 | | | < 0.0001 |
Septic shock | 0.87 | 0.39 | 2.39 | 1.13–5.09 | 0.0234 |
Connective tissue diseases | 3.03 | 1.14 | 20.64 | 2.23–191 | 0.0077 |
Anaemia | 0.87 | 0.36 | 2.39 | 1.18–4.85 | 0.0161 |
Use of antibiotics | 1.39 | 0.34 | 4.03 | 2.07–7.86 | < 0.001 |
Use of proton-pump inhibitors | 0.72 | 0.34 | 2.06 | 1.05–4.05 | 0.0355 |
We calculated the predictive probability of toxigenic C. difficile carrier using the following formula:
$$\text{P}(\text{Y}=\text{y}\text{e}\text{s})=\frac{\text{e}\text{x}\text{p}(-1.93+0.87\text{A}+3.03\text{B}+0.87\text{C}+1.39\text{D}+0.72\text{E})}{1+\text{e}\text{x}\text{p}(-1.93+0.87\text{A}+3.03\text{B}+0.87\text{C}+1.39\text{D}+0.72\text{E})}$$
A: If septic shock, positive = 1 / negative = 0, B: If connective tissue diseases, positive = 1 / negative = 0, C: If anemia, positive = 1 / negative = 0, D: If recent use of antibiotics, positive = 1 / negative = 0, E: If recent use of proton-pump inhibitors, positive = 1 / negative = 0.
When a cut-off value of ≥ 0.45 was applied to the clinical prediction model, the AUC value was 0.80, with a 95% confidence interval (CI) of 0.74–0.86 in the derivation cohort. The sensitivity, specificity, positive predictive value, and negative predictive value of this prediction rule were 75.3% (95% CI, 65.7–83.3), 74.3% (95% CI, 64.6–82.4), 74.5% (95% CI, 67.3–80.6) and 75.0% (95% CI: 67.7–81.1), respectively (Table 3). As shown in Table 3, the accuracy, sensitivity, and specificity of the test data were similar to those of the training data. The Hosmer–Lemeshow goodness-of-fit test result for the final model was P = 0.58. Therefore, there was no evidence of a lack of fit in the final model.
Table 3
The area under the ROC curve and the probability of toxigenic Clostridioides difficile carriage at hospital admission based on a cut-off value in the derivation population with cross-validation and in the validation cohort
Variable | AUC ± SE | 95% CI | Cut-off values | Sensitivity | 95% CI | Specificity | 95% CI | PPV | 95% CI | NPV | 95% CI |
Training dataset | 0.80 ± 0.031 | 0.74–0.86 | 0.45 | 75.25 | 65.7–83.3 | 74.3 | 64.6–82.4 | 74.5 | 67.3–80.6 | 75.0 | 67.7–81.1 |
Cross-validation (Training dataset) | 0.76 ± 0.035 | 0.70–0.82 | 0.47 | 75.25 | 65.7–83.3 | 74.3 | 64.6–82.4 | 74.5 | 67.3–80.6 | 75.0 | 67.7–81.1 |
Test dataset | 0.84 ± 0.035 | 0.78–0.90 | 0.45 | 78.26 | 56.3–92.5 | 70.8 | 62.7–78.0 | 29.5 | 23.1–36.8 | 95.4 | 90.5–97.8 |
AUC, area under the receiver operating characteristic curve; CI, confidence interval; NPV, negative predictive value; PPV, positive predictive value; SE, standard error. |
Validation Of The Clinical Prediction Model
The validation cohort included 170 patients who underwent PCR screening upon admission for toxigenic C. difficile carriage between April 2021 and October 2021. PCR screening was not performed in 33 patients because stool samples or rectal swab specimens could not be collected within 48 h of admission. This prediction risk model showed an AUC of 0.84 (95% CI, 0.78–0.90) in the validation dataset. The sensitivity, specificity, positive predictive value, and negative predictive value of this prediction rule were 78.3% (95% CI, 56.3–92.5), 70.75% (95% CI, 62.7–78.0), 29.5% (95% CI, 23.1–36.8), and 95.4% (95% CI, 90.5–97.8), respectively (Table 3). Of the 170 patients, 23 were positive for C. difficile toxin and 147 were negative. Our prediction rule confirmed that 27 were false positives and six were false negatives.