The Evolution of Urban C. Difficile Infection
The Evolution of Urban C. Difficile Infection
Overall Cohort We identified 3,096 consecutive patients with a positive CDI toxin assay and diarrhea between 2006 and 2011; 1,189 were classified into the CDI 06–08 cohort and 1,907 into the CDI 09–11 cohort ( Table 1). CDI 09–11 patients were significantly older (P=0.01) and less frequently Caucasian compared with individuals in CDI 06–08. The more recent cohort had a higher Charlson comorbidity score (P=0.02) with greater prevalences of diabetes mellitus (P=0.01), end-stage kidney disease (P<0.01), cirrhosis (P=0.05), and peptic ulcer disease (P=0.03). CDI 09–11 had lower frequencies of percutaneous endoscopic gastrostomy tubes, any cancers, solid tumors, any previous CDI, and surgery within 12 weeks compared with CDI 06–08. There were no significant differences in the recent hospitalization (e.g., within 12 weeks) before diagnosis between time frames.
Severe Cohort. In all, 243 (20.4%) and 382 (20.0%) patients fulfilled criteria for severe CDI in the CDI 06–08 and CDI 09–11 cohorts, respectively. There were no significant differences in demographics, ethnicity, and medical comorbidities when comparing the severe subgroups. There were significantly fewer patients who had surgery within 12 weeks of diagnosis in the severe CDI 09–11 cohort compared with severe CDI 06–08 cohort (P<0.001).
Data for any previous hospitalization within 12 weeks of the diagnosis of CDI were available for less than the entire cohort for each time frame with 1,031 (86.9%) and 1,849 (96.9%) patients from CDI 06–08 and CDI 09–11 cohorts, respectively (Table 2). When comparing CA-CDI with HA-CDI within the time frames, Charlson scores and ICU requirement were seen to be lower for CA-CDI, and previous CDI, WBC, and albumin level at diagnosis were each higher (Table 2). These statistically significant differences were consistent in both time frames, with the exception of previous CDI, which did not achieve significance in CDI 06–08 but did in CDI 09–11. The remaining observed differences were similar in the two time frames. Of note, there were no differences in 30-day mortality or hospital re-admission rate within 30 days of discharge when comparing CA-CDI with HA-CDI within each time frame.
Data for recurrence were available in 1,071 patients (90.1% of the cohort) from CDI 06–08 and 1,876 patients (98.4% of the entire cohort) from CDI 09–11 (Table 3). In CDI 06–08, episodes of recurrent infection were less frequently severe compared with episodes that were not recurrent. There was no difference in severity between recurrent and nonrecurrent disease in CDI 09–11.The use of any metronidazole and metronidazole alone in both time frames was significantly lower in the population with R-CDI, whereas the use of any vancomycin, only vancomycin, and a combination of vancomycin and metronidazole was more frequent in R-CDI compared with NR-CDI. Length of therapy with any IV metronidazole and any PO vancomycin was shorter in those with R-CDI 09–11compared with NR-CDI 09–11. In R-CDI 09–11, ICU requirement and discharge to home was less frequent and discharge to SNFs was more frequent compared with NR-CDI 09–11. There were no significant differences in mortality between recurrent and nonrecurrent disease in either time frame.
When comparing the recurrent subgroups in the two time intervals, there were no significant differences in severity, serologic testing, or treatment patterns. Compared with R-CDI 06–08, R-CDI 09–11 patients were discharged home less frequently and sent to skilled-nursing facilities following discharge more frequently. Although there was a trend toward decreased 30-day mortality within the recurrent cohort of the more modern time frame, this did not achieve statistical significance (18.1 vs. 12.6%, P=0.14).
Overall cohort. Antibiotic exposure during the 3 months before diagnosis was similar between the two groups when comparing cephalosporins, clindamycin, and penicillins; however, exposure to quinolones (P<0.001), oral vancomycin (P<0.001), and oral metronidazole (P<0.01) was less frequent and macrolide exposure (P<0.001) was more frequent in CDI 09–11 compared with CDI 06–08 ( Table 4).
Severe Cohort. In the severe CDI 09–11 cohort compared with the severe CDI 06–08 cohort, there was a higher frequency of exposure to the macrolide (P<0.001) and penicillin (P<0.05) classes of antibiotics, with lower frequencies of quinolone (P<0.05) and oral vancomycin (P<0.05).
Overall.CDI 09–11 had lower peak WBC (P<0.001), and WBC at diagnosis (P<0.001), with a higher creatinine at diagnosis (P=0.002) than did CDI 06–08 ( Table 5). Computed tomography (CT) scan (P<0.01) and colonoscopy were performed less frequently during 2009–2011 (P<0.01); however, if a colonoscopy was performed, pseudomembranes were more likely to be identified (P<0.001) compared with 2006–2008.
Severe. In severe CDI 09–11 compared with CDI 06–08, peak WBC (P<0.01) and WBC at diagnosis were lower (P<0.001), without significant differences in creatinine and albumin. CT scans of the abdomen and pelvis (P<0.05) were performed more frequently, and if a colonoscopy was performed pseudomembranes were seen more frequently in the more recent time frame compared with CDI 06–08.
Overall Cohort. Treatment patterns varied between the older and newer cohorts. CDI 09–11 patients were treated with any oral metronidazole less frequently (81.4 vs. 71.8%, P<0.001) and for a shorter duration (5.1±4.8 vs. 6.2±5.9 days, P<0.001; Table 6). Although similar percentages of patients received IV metronidazole during the two time frames, the duration of treatment was longer in CDI 09–11 (6.4±6.2 vs. 5.5±6.3 days, P<0.05). Patients treated with only metronidazole (oral or IV) was similar between the cohorts, but again the length of therapy with oral metronidazole was shorter (5.1±4.7 vs. 6.1±5.6 days, P<0.001) and longer with IV metronidazole (4.3±4.7 vs. 3.5±3.8 days, P<0.05). Patients were treated with only oral vancomycin therapy more frequently during the CDI 09–11 time frame (4.8 vs. 2.8%, P<0.01), but for a similar duration as CDI 06–08. The frequencies of combination therapy with any metronidazole and vancomycin were similar, but patients were switched from any metronidazole therapy to oral vancomycin almost twice as frequently (11.4 vs. 6.7%, P<0.001) during the more recent time frame.
Severe Cohort. When comparing CDI 09–11 with CDI 06–08, any oral metronidazole was used less frequently (P < 0.001) with shorter duration of therapy (P < 0.05). Oral metronidazole as the lone therapy also was used for less time (P < 0.01). The more modern cohort had higher rates of only oral vancomycin usage (P < 0.01) and switching from vancomycin to metronidazole (P < 0.01).
Overall Cohort. Compared with CDI 06–08, patients from 2009 through 2011 were discharged more frequently on treatment for CDI and to SNF, with a higher frequency of patients being re-admitted for any cause within 30 days of discharge; no differences were observed in the subset of patients being re-admitted within 30 days with CDI (Table 7). Patients required ICU level of care less frequently during the more modern time frame. Mortality rates in the CDI 09–11 cohort were lower in both the 30-day (13.1 vs. 17.1%, P<0.01) and 6-month (34.8 vs. 38.4%, P<0.05) end point measures. Through Kaplan–Meier analysis, survival was shown to be lower in CDI 06–08 within 30 days of diagnosis (P<0.01). CDI 09–11 had significantly lower mortality rates in the 70- to 79-year-old cohort (13.3 vs. 22.0%, P<0.01) and in the 80- to 89-year-old cohort (18.1 vs. 29.2%, P<0.001) compared with CDI 06–08. There were no significant differences when comparing the other decades of life (Figure 1a).
(Enlarge Image)
Figure 1.
(a) The 30-day mortality by decade (entire cohort): CDI 09–11 had significantly lower mortality rates in the 70- to 79-year-old cohort (13.3 vs. 22.0%, P<0.01) and in the 80- to 89-year-old cohort (18.1 vs. 29.2%) compared with CDI 06–08. (b) Mortality by decade (severe CDI): severe CDI 09–11 had significantly lower mortality rates in the 70- to 79-year-old cohort (24.3 vs. 39.5%, P=0.03), 80- to 89-year-old cohort (25.5 vs. 46.4%), and the 90- to 99-year-old cohort (34.1 vs. 64.7%, P=0.03) compared with CDI 06–08.
Through multivariate analysis, in the CDI 06–08 cohort, age (hazard ratio (HR): 1.04±0.01, P<0.001), WBC (1.02±0.01, P<0.01), albumin level (0.41±0.10, P<0.001), any glucocorticoid exposure in the 3 months before diagnosis (1.59±0.20, P<0.01), and discharge to SNF (HR: 1.77±0.17, P<0.001) were associated with increased 30-day mortality. In the CDI 09–11 cohort, age (HR: 1.02±0.01, P<0.001), Charlson score (HR: 1.04±0.001, P<0.001), WBC (HR: 1.04±0.01, P<0.001), albumin (0.44±0.01, P<0.001), international normalized ratio at the time of diagnosis (HR: 1.07±0.03, P<0.02), glucocorticoid exposure (HR: 1.17±0.08, P<0.001), and discharge to SNF (HR: 2.07±0.15, P<0.001) were all associated with increased 30-day mortality.
Severe Cohort. Patients in the CDI 09–11 cohort were discharged on treatment (P<0.01) and admitted within 30 days more frequently compared with patients in the CDI 06–08 cohort. There was no significant difference in re-admission for R-CDI (P=0.93) during the time frames. Patients with severe CDI required ICU level of care less frequently during the more recent time frame (P<0.01). Thirty-day mortality of severe CDI was lower in 2009–2011 compared with 2006–2008 (31.3 vs. 23.3%, P<0.05). Through Kaplan–Meier analysis, survival was shown to be lower in CDI 06–08 within 30 days of diagnosis (P<0.05). Severe CDI 09–11 had significantly lower mortality rates in the 70- to 79-year-old cohort (24.3 vs. 39.5%, P=0.03), 80- to 89-year-old cohort (25.5 vs. 46.4%), and the 90- to 99-year-old cohort (34.1 vs. 64.7%, P=0.03) compared with CDI 06–08 (Figure 1b).
Age (1.06±0.01, P<0.001), history of solid cancer (1.73±0.26, P<0.05), and colonoscopy at or around the time of diagnosis (0.11±1.0, P<0.05) were significant multivariate predictors of 30-day mortality in the CDI 06–08 cohort. Through multivariate analysis, significant predictors of 30-day mortality in patients in the severe CDI 09–11 included age (HR: 1.03±0.01, P<0.001), Charlson score (1.17±0.05, P<0.001), the diagnosis of lymphoma in the past (6.20±0.54, P<0.001), WBC (1.02±0.01, P<0.001), and albumin (0.38±0.26, P<0.001).
Results
Patient and Group Characteristics
Overall Cohort We identified 3,096 consecutive patients with a positive CDI toxin assay and diarrhea between 2006 and 2011; 1,189 were classified into the CDI 06–08 cohort and 1,907 into the CDI 09–11 cohort ( Table 1). CDI 09–11 patients were significantly older (P=0.01) and less frequently Caucasian compared with individuals in CDI 06–08. The more recent cohort had a higher Charlson comorbidity score (P=0.02) with greater prevalences of diabetes mellitus (P=0.01), end-stage kidney disease (P<0.01), cirrhosis (P=0.05), and peptic ulcer disease (P=0.03). CDI 09–11 had lower frequencies of percutaneous endoscopic gastrostomy tubes, any cancers, solid tumors, any previous CDI, and surgery within 12 weeks compared with CDI 06–08. There were no significant differences in the recent hospitalization (e.g., within 12 weeks) before diagnosis between time frames.
Severe Cohort. In all, 243 (20.4%) and 382 (20.0%) patients fulfilled criteria for severe CDI in the CDI 06–08 and CDI 09–11 cohorts, respectively. There were no significant differences in demographics, ethnicity, and medical comorbidities when comparing the severe subgroups. There were significantly fewer patients who had surgery within 12 weeks of diagnosis in the severe CDI 09–11 cohort compared with severe CDI 06–08 cohort (P<0.001).
CA-CDI vs. HA-CDI
Data for any previous hospitalization within 12 weeks of the diagnosis of CDI were available for less than the entire cohort for each time frame with 1,031 (86.9%) and 1,849 (96.9%) patients from CDI 06–08 and CDI 09–11 cohorts, respectively (Table 2). When comparing CA-CDI with HA-CDI within the time frames, Charlson scores and ICU requirement were seen to be lower for CA-CDI, and previous CDI, WBC, and albumin level at diagnosis were each higher (Table 2). These statistically significant differences were consistent in both time frames, with the exception of previous CDI, which did not achieve significance in CDI 06–08 but did in CDI 09–11. The remaining observed differences were similar in the two time frames. Of note, there were no differences in 30-day mortality or hospital re-admission rate within 30 days of discharge when comparing CA-CDI with HA-CDI within each time frame.
R-CDI vs. NR-CDI
Data for recurrence were available in 1,071 patients (90.1% of the cohort) from CDI 06–08 and 1,876 patients (98.4% of the entire cohort) from CDI 09–11 (Table 3). In CDI 06–08, episodes of recurrent infection were less frequently severe compared with episodes that were not recurrent. There was no difference in severity between recurrent and nonrecurrent disease in CDI 09–11.The use of any metronidazole and metronidazole alone in both time frames was significantly lower in the population with R-CDI, whereas the use of any vancomycin, only vancomycin, and a combination of vancomycin and metronidazole was more frequent in R-CDI compared with NR-CDI. Length of therapy with any IV metronidazole and any PO vancomycin was shorter in those with R-CDI 09–11compared with NR-CDI 09–11. In R-CDI 09–11, ICU requirement and discharge to home was less frequent and discharge to SNFs was more frequent compared with NR-CDI 09–11. There were no significant differences in mortality between recurrent and nonrecurrent disease in either time frame.
When comparing the recurrent subgroups in the two time intervals, there were no significant differences in severity, serologic testing, or treatment patterns. Compared with R-CDI 06–08, R-CDI 09–11 patients were discharged home less frequently and sent to skilled-nursing facilities following discharge more frequently. Although there was a trend toward decreased 30-day mortality within the recurrent cohort of the more modern time frame, this did not achieve statistical significance (18.1 vs. 12.6%, P=0.14).
Medication Exposures
Overall cohort. Antibiotic exposure during the 3 months before diagnosis was similar between the two groups when comparing cephalosporins, clindamycin, and penicillins; however, exposure to quinolones (P<0.001), oral vancomycin (P<0.001), and oral metronidazole (P<0.01) was less frequent and macrolide exposure (P<0.001) was more frequent in CDI 09–11 compared with CDI 06–08 ( Table 4).
Severe Cohort. In the severe CDI 09–11 cohort compared with the severe CDI 06–08 cohort, there was a higher frequency of exposure to the macrolide (P<0.001) and penicillin (P<0.05) classes of antibiotics, with lower frequencies of quinolone (P<0.05) and oral vancomycin (P<0.05).
Evaluation of Disease
Overall.CDI 09–11 had lower peak WBC (P<0.001), and WBC at diagnosis (P<0.001), with a higher creatinine at diagnosis (P=0.002) than did CDI 06–08 ( Table 5). Computed tomography (CT) scan (P<0.01) and colonoscopy were performed less frequently during 2009–2011 (P<0.01); however, if a colonoscopy was performed, pseudomembranes were more likely to be identified (P<0.001) compared with 2006–2008.
Severe. In severe CDI 09–11 compared with CDI 06–08, peak WBC (P<0.01) and WBC at diagnosis were lower (P<0.001), without significant differences in creatinine and albumin. CT scans of the abdomen and pelvis (P<0.05) were performed more frequently, and if a colonoscopy was performed pseudomembranes were seen more frequently in the more recent time frame compared with CDI 06–08.
CDI Treatment Patterns
Overall Cohort. Treatment patterns varied between the older and newer cohorts. CDI 09–11 patients were treated with any oral metronidazole less frequently (81.4 vs. 71.8%, P<0.001) and for a shorter duration (5.1±4.8 vs. 6.2±5.9 days, P<0.001; Table 6). Although similar percentages of patients received IV metronidazole during the two time frames, the duration of treatment was longer in CDI 09–11 (6.4±6.2 vs. 5.5±6.3 days, P<0.05). Patients treated with only metronidazole (oral or IV) was similar between the cohorts, but again the length of therapy with oral metronidazole was shorter (5.1±4.7 vs. 6.1±5.6 days, P<0.001) and longer with IV metronidazole (4.3±4.7 vs. 3.5±3.8 days, P<0.05). Patients were treated with only oral vancomycin therapy more frequently during the CDI 09–11 time frame (4.8 vs. 2.8%, P<0.01), but for a similar duration as CDI 06–08. The frequencies of combination therapy with any metronidazole and vancomycin were similar, but patients were switched from any metronidazole therapy to oral vancomycin almost twice as frequently (11.4 vs. 6.7%, P<0.001) during the more recent time frame.
Severe Cohort. When comparing CDI 09–11 with CDI 06–08, any oral metronidazole was used less frequently (P < 0.001) with shorter duration of therapy (P < 0.05). Oral metronidazole as the lone therapy also was used for less time (P < 0.01). The more modern cohort had higher rates of only oral vancomycin usage (P < 0.01) and switching from vancomycin to metronidazole (P < 0.01).
Outcomes and Predictors of 30-day Mortality
Overall Cohort. Compared with CDI 06–08, patients from 2009 through 2011 were discharged more frequently on treatment for CDI and to SNF, with a higher frequency of patients being re-admitted for any cause within 30 days of discharge; no differences were observed in the subset of patients being re-admitted within 30 days with CDI (Table 7). Patients required ICU level of care less frequently during the more modern time frame. Mortality rates in the CDI 09–11 cohort were lower in both the 30-day (13.1 vs. 17.1%, P<0.01) and 6-month (34.8 vs. 38.4%, P<0.05) end point measures. Through Kaplan–Meier analysis, survival was shown to be lower in CDI 06–08 within 30 days of diagnosis (P<0.01). CDI 09–11 had significantly lower mortality rates in the 70- to 79-year-old cohort (13.3 vs. 22.0%, P<0.01) and in the 80- to 89-year-old cohort (18.1 vs. 29.2%, P<0.001) compared with CDI 06–08. There were no significant differences when comparing the other decades of life (Figure 1a).
(Enlarge Image)
Figure 1.
(a) The 30-day mortality by decade (entire cohort): CDI 09–11 had significantly lower mortality rates in the 70- to 79-year-old cohort (13.3 vs. 22.0%, P<0.01) and in the 80- to 89-year-old cohort (18.1 vs. 29.2%) compared with CDI 06–08. (b) Mortality by decade (severe CDI): severe CDI 09–11 had significantly lower mortality rates in the 70- to 79-year-old cohort (24.3 vs. 39.5%, P=0.03), 80- to 89-year-old cohort (25.5 vs. 46.4%), and the 90- to 99-year-old cohort (34.1 vs. 64.7%, P=0.03) compared with CDI 06–08.
Through multivariate analysis, in the CDI 06–08 cohort, age (hazard ratio (HR): 1.04±0.01, P<0.001), WBC (1.02±0.01, P<0.01), albumin level (0.41±0.10, P<0.001), any glucocorticoid exposure in the 3 months before diagnosis (1.59±0.20, P<0.01), and discharge to SNF (HR: 1.77±0.17, P<0.001) were associated with increased 30-day mortality. In the CDI 09–11 cohort, age (HR: 1.02±0.01, P<0.001), Charlson score (HR: 1.04±0.001, P<0.001), WBC (HR: 1.04±0.01, P<0.001), albumin (0.44±0.01, P<0.001), international normalized ratio at the time of diagnosis (HR: 1.07±0.03, P<0.02), glucocorticoid exposure (HR: 1.17±0.08, P<0.001), and discharge to SNF (HR: 2.07±0.15, P<0.001) were all associated with increased 30-day mortality.
Severe Cohort. Patients in the CDI 09–11 cohort were discharged on treatment (P<0.01) and admitted within 30 days more frequently compared with patients in the CDI 06–08 cohort. There was no significant difference in re-admission for R-CDI (P=0.93) during the time frames. Patients with severe CDI required ICU level of care less frequently during the more recent time frame (P<0.01). Thirty-day mortality of severe CDI was lower in 2009–2011 compared with 2006–2008 (31.3 vs. 23.3%, P<0.05). Through Kaplan–Meier analysis, survival was shown to be lower in CDI 06–08 within 30 days of diagnosis (P<0.05). Severe CDI 09–11 had significantly lower mortality rates in the 70- to 79-year-old cohort (24.3 vs. 39.5%, P=0.03), 80- to 89-year-old cohort (25.5 vs. 46.4%), and the 90- to 99-year-old cohort (34.1 vs. 64.7%, P=0.03) compared with CDI 06–08 (Figure 1b).
Age (1.06±0.01, P<0.001), history of solid cancer (1.73±0.26, P<0.05), and colonoscopy at or around the time of diagnosis (0.11±1.0, P<0.05) were significant multivariate predictors of 30-day mortality in the CDI 06–08 cohort. Through multivariate analysis, significant predictors of 30-day mortality in patients in the severe CDI 09–11 included age (HR: 1.03±0.01, P<0.001), Charlson score (1.17±0.05, P<0.001), the diagnosis of lymphoma in the past (6.20±0.54, P<0.001), WBC (1.02±0.01, P<0.001), and albumin (0.38±0.26, P<0.001).
Source...