Table 3. CDI Risk Associated with Antimicrobial Exposures During and Within 5 days of the End of Antimicrobial Therapy, for Antibiotic Risk Indexes and Specific Antibiotic Exposures.
| Unadjusted | Adjustedc | ||||||
| Exposure in the preceding 5d | CDI Cases | Follow-up | IRa | IR Ratio | Δ AICb | IR Ratio | Δ AICb |
| (N) | (days) | (95% CI) | (95% CI) | ||||
| Any antibiotic | |||||||
| No | 33 | 104069 | 3.2 | Reference | Reference | ||
| Yes | 94 | 112909 | 8.3 | 2.63 (1.77, 3.90) | 0 | 2.35 (1.53, 3.60) | 0 |
| Number of antibiotics | |||||||
| 0 | 33 | 104069 | 3.2 | Reference | Reference | ||
| 1 | 64 | 73593 | 8.7 | 2.74 (1.80, 4.17) | 2.49 (1.59, 3.92) | ||
| ≥2 | 30 | 39316 | 7.6 | 2.41 (1.47, 3.95) | 1.6 | 2.09 (1.23, 3.55) | 1.6 |
| Antibiotic risk index | |||||||
| None or low-risk | 35 | 106396 | 3.3 | Reference | Reference | ||
| Medium-risk | 12 | 19256 | 6.2 | 1.89 (1.33, 2.67) | 1.79 (1.24, 2.59) | ||
| High-risk | 80 | 91326 | 8.8 | 2.66 (1.79, 3.96) | 2.1 | 2.43 (1.59, 3.74) | 1.3 |
| Class of antibioticd | |||||||
| Penicillins | 24 | 25103 | 9.5 | 3.02 (1.78, 5.10) | 1.4 | 2.77 (1.56, 4.90) | 1.7 |
| Broad-spectrum | 16 | 18876 | 8.5 | 2.67 (1.47, 4.86) | 2.0 | 2.39 (1.26, 4.53) | 2.3 |
| Narrow-spectrum | 8 | 6857 | 11.7 | 3.68 (1.70, 7.97) | 1.1 | 3.64 (1.47, 9.00) | 1.5 |
| Cephalosporins | 61 | 66330 | 9.2 | 2.90 (1.90, 4.43) | 0.5 | 2.70 (1.69, 4.32) | 0.2 |
| 1st & 2nd generation | 31 | 38508 | 8.0 | 2.54 (1.55, 4.15) | 1.9 | 2.36 (1.36, 4.10) | 2.3 |
| 3rd & 4th generation | 32 | 27092 | 11.8 | 3.72 (2.29, 6.06) | −2.8 | 3.40 (2.02, 5.72) | −2.5 |
| Carbapenems | 6 | 5962 | 10.1 | 3.17 (1.33, 7.57) | 1.8 | 2.40 (0.93, 6.23) | 2.3 |
| Fluoroquinolones | 24 | 30286 | 7.9 | 2.50 (1.48, 4.23) | 1.9 | 2.16 (1.26, 3.72) | 2.0 |
| IV vancomycin | 16 | 11883 | 13.5 | 4.25 (2.34, 7.71) | −1.6 | 3.19 (1.62, 6.26) | 1.2 |
Abbreviations: AIC, Akaike's Information Criterion; CDI, Clostridium difficile infection; CI, confidence interval; d, days; IR, incidence rate; IRR, incidence rate ratio.
a
Incidence rate, per 10,000 patient-days.
b
The difference in AIC relative to the reference model (receipt of any antibiotic in the previous 5 days): negative numbers denote an improvement in fit. Δ AIC <−2 was considered a statistically significant improvement in fit at p<0.05.
c
Adjusted for time since hospital admission, age, sex, number of previous hospital admissions, infection pressure, and current or prior ICU admission.
d
Each antibiotic group was assessed in a separate model; the reference group for each model was no receipt of antibiotics in the last 5 days.