- Email: [email protected]
Antibiotic use and duration in association with Clostridioides difficile infection in a tertiary academic medical center: A retrospective case-control study
Anaerobe 59 (2019) 126e130
Contents lists available at ScienceDirect
Anaerobe journal homepage: www.elsevier.com/locate/anaerobe
Clostridioides difficile (including epidemiology)
Antibiotic use and duration in association with Clostridioides difficile infection in a tertiary academic medical center: A retrospective casecontrol study Abrar K. Thabit a, b, *, Christy A. Varugehese b, 2, Alexander R. Levine b, 1 a b
Pharmacy Practice Department, Faculty of Pharmacy, King Abdulaziz University, Jeddah, Saudi Arabia Massachusetts General Hospital, Boston, MA, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 17 December 2018 Received in revised form 25 May 2019 Accepted 26 June 2019 Available online 27 June 2019
Previous literature has attributed the use of certain antibiotics to Clostridioides difficile infection (CDI); however, the time from administration to CDI onset is not adequately evaluated. We aimed to determine the type of antibiotics and duration of therapy associated with the highest CDI incidence at a tertiary academic medical center. This was a retrospective, case-control study of adult inpatients who received at least one course of antibiotic treatment. Patients were divided into either cases or controls. For cases, their first episode of CDI was a determining factor. Primary outcome were the types of antibiotics associated with risk of CDI development and the median time of antibiotic usage defining this risk. Of 601 patients who developed CDI, 313 were included as cases while 150 of 291 who received antibiotics but did not develop CDI were included as controls. Cefepime and cefazolin were significantly associated with increased risk for CDI with odds ratios of 3.01 (95% CI, 1.96e4.65; P < 0.001) and 1.71 (95% CI, 1.02 e2.95; P < 0.05), respectively. Cefepime was associated with CDI after a median time of 8 days while CDI may have occurred after 6 days of therapy with cefazolin. Use of antineoplastic agents was significantly associated with CDI (odds ratio, 2.32; 95% CI, 1.35e4.13; P < 0.01). Antibiotic use increased the risk of CDI, particularly with cefepime and cefazolin with a median time to incidence of 8 and 6 days, respectively. CDI risk was also increased with the use of antineoplastic agents. © 2019 Elsevier Ltd. All rights reserved.
Handling Editor: Stuart Johnson Keywords: Clostridioides difficile infection Antibiotic Antineoplastics Cefepime Cefazolin
1. Introduction Clostridioides difficile is the major cause of infectious nosocomial diarrhea in the United States that usually develops after antibiotic therapy, accounting for 20e30% of cases of nosocomial antibioticassociated diarrhea [1]. It is estimated that C. difficile infection (CDI) is associated with an increased burden to the American health care system with a total exceeding $3 billion per year [2]. Two rationales for CDI pathogenesis currently exist. Upon exposure to antibiotics, intestinal flora other than C. difficile becomes suppressed and disrupted allowing C. difficile to overgrow. Exogenous acquisition of C. difficile creates the basis for the second
* Corresponding author. Pharmacy Practice Department, Faculty of Pharmacy, King Abdulaziz University, 7027 Abdullah Al-Sulaiman Rd, Jeddah, 22254-2265, Saudi Arabia. E-mail address: [email protected] (A.K. Thabit). 1 Current affiliation: University of Saint Joseph, Hartford, CT, USA. 2 Current affiliation: Rush University Medical Center, Chicago, IL, USA. https://doi.org/10.1016/j.anaerobe.2019.06.016 1075-9964/© 2019 Elsevier Ltd. All rights reserved.
rationale [3]. As a result, the development of CDI requires two major components, exposure to antibiotic therapy and presence of toxigenic C. difficile [3]. About 96% of patients who had received antibiotic therapy reported symptomatic diarrhea within 14 days where the median time between the initiation of antibiotic therapy and CDI onset was 2e3 days [4]. Antibiotics that are most implicated in causing CDI are clindamycin, fluoroquinolones, and cephalosporins, whereas parenteral aminoglycosides, vancomycin, and metronidazole are less frequently associated with the disease [3]. A healthcare institution may benefit from knowing its specific antibiotic-associated CDI patterns based on the institution's antibiotic formulary and patient population [3,5]. Moreover, by knowing the timeframe between the administration of culprit antibiotics and CDI onset, healthcare providers can probably consider early treatment discontinuation if antibiotics are no longer needed or switch therapy to agents with lower likelihood of association with CDI in case prolonged therapy is needed or expected. In addition to antibiotic use, several factors were reported to
A.K. Thabit et al. / Anaerobe 59 (2019) 126e130
increase the risk of CDI. Such factors include age equal to or greater than 65 years, antineoplastic therapy, infection with human immunodeficiency virus, gastrointestinal (GI) surgery or manipulation of the GI tract (e.g., tube feeding), and the use of proton pump inhibitors (PPIs) or H2-receptor antagonists (H2RAs) [1]. We conducted this study in order to find out the median time from the administration to the onset of CDI of antibiotics mostly associated with the infection within a tertiary academic medical center. 2. Materials and methods 2.1. Study design and setting This retrospective case-control study was conducted at Massachusetts General Hospital (Boston, MA, USA), a tertiary academic medical center, from January 2010 through December 2012. The study was approved by the Institutional Review Board of Partners Healthcare System (protocol # 2013P000958). 2.2. Data sources We obtained a database from Massachusetts General Hospital Infection Control Unit of all patients who developed CDI during the study period (defined as cases). The Department of Pharmacy provided databases of all antibiotics that were dispensed in the hospital during the study period. Patients in these databases were assessed for eligibility as controls (defined as patients who received antibiotics but did not develop CDI). CDI was defined when a polymerase chain reaction (PCR) test and/or toxin assay of C. difficile was positive [6]. Data of primary and secondary endpoints, as well as data of patient characteristics were obtained from the hospital's electronic records. 2.3. Patients Cases and controls were included in a 2:1 ratio. Patients were deemed eligible if they were 18 years of age or older, initially admitted to our hospital, and received at least one antibiotic during the duration of hospital stay. For cases, the CDI episode had to be their first in order to be included in the study. Patients who had a previous history of CDI (i.e., current episode is a recurrence), developed their first CDI immediately before hospital admission, were exposed to antibiotics within one month before hospital admission, or were recently admitted to a hospital within a month were excluded from the study. Both patients of general care and critical care services were included. If a patient was transferred from a critical care unit to a general care unit (or vice versa), the patient was assigned to the unit at which CDI was acquired (within 24 h of transfer). If a patient was admitted through the emergency department (ED), the duration of hospital stay was counted starting from the day of presentation at the ED. 2.4. Outcomes The primary outcome was types of antibiotic therapy associated with the development of CDI. We assessed this outcome by looking at the electronic medication administration records and counting the days of therapy of each antibiotic administered. If antibiotic administration was stopped and reinitiated after several days, all days of antibiotic administration were counted. For cases, antibiotic days were counted until CDI. Secondary outcomes included effects of confounding factors on CDI risk. Confounding factors included age, acid suppression
127
therapy (PPIs and H2-RAs), and antineoplastic agents. These factors were assessed from each patient's electronic records. 2.5. Statistical analysis In order to assess the association of antibiotics and CDI development, odds ratios (OR) and 95% confidence intervals (CI) were estimated using Cox proportional hazards regression, with covariate adjustment for the antibiotic administered. Median time until CDI was calculated in cases for each antibiotic administered. For secondary outcomes, ORs and 95% CIs were estimated using Cox proportional hazards regression, with covariate adjustment for the confounding factor in question. We compared select baseline characteristics between cases and controls by estimating ORs and 95% CIs. The number of antibiotics received by each patient, days of therapy of all antibiotics received, and the total duration of hospital stay were presented as a median of days given the lack of normal distribution of the data as demonstrated by the lack of significance in the Sahpiro-Wilk test for normality. Logistic regression was used to adjust OR of primary outcome for any confounding factor that showed statistical significant association with CDI development. A P value of <0.05 was considered statistically significant. We used STATA software package 10.1 (StataCorp, College Station, TX) and Microsoft Excel 14.0 (Microsoft Inc., Redmond, WA) for our analyses and chart development, respectively. 3. Results 3.1. Patients and characteristics A total of 892 patients were assessed for eligibility (Fig. 1). Of the 601 patients from the cases database, 313 patients were included in the study. Antibiotic treatment before hospital admission was the major cause for exclusion from the study (n ¼ 48) followed by the transfer from an outside hospital (n ¼ 41). One hundred and fifty patients were included as controls and 141 were excluded out of 291 patients who received antibiotics but did not develop CDI. Most of the excluded patients from this group were aged less than 18 years (n ¼ 42). The number of included CDI cases that were detected by PCR was 32 (10.2%), whereas toxigenic assay detected the rest of the included cases (n ¼ 281, 89.8%). Overall, cases and controls did not significantly differ in their baseline characteristics, except for the use of antineoplastic agents and the length of hospital stay with (P < 0.01 for both), respectively (Table 1). More patients in both groups were admitted to general care units; however, this factor was not significantly different. 3.2. Primary outcome A total of 41 different species of antibiotics were administered to the study population. Two antibiotics were significantly associated with increased risk of CDI development. Cefepime showed the highest OR of 3.01 (95% CI, 1.96e4.65; P < 0.001), while cefazolin had an OR of 1.71 (95% CI, 1.02e2.95; P < 0.05) (Fig. 2). The remaining 39 antibiotics did not show significant increase in CDI risk. Ampicillin-sulbactam showed a significant protective effect (P < 0.05) in not increasing the risk of CDI acquisition. Interestingly, antibiotics that are known to increase the risk of CDI development, such as clindamycin and fluoroquinolones, did not show an increased risk in our study; however these findings were not statistically significant. The median duration in days for CDI to occur on all the antibiotics ranged between 2 and 26 days. Once cefepime was administered, CDI was estimated to occur at the eighth day whether the antibiotic was continued or discontinued irrespective of the
128
A.K. Thabit et al. / Anaerobe 59 (2019) 126e130
Fig. 1. Flowchart of study subjects.
Table 1 Baseline characteristics of cases and controls. Characteristic Age, n (%) 65 years <65 years Gender, n (%) Male Female Hospital service, n (%) Critical care General care Number of antibiotics received per patient, median (IQR) DOT of all antibiotics received, median (IQR) Use of acid suppression therapy, n (%) Proton pump inhibitors H2-receptor antagonists Underlying lower GI disease, n (%) Use of antineoplastic agents, n (%) Total duration of hospital stay, median (IQR), days
Cases (n ¼ 313)
Controls (n ¼ 150)
173 (55.3) 140 (44.7)
70 (46.7) 80 (53.3)
182 (58.1) 131 (41.9)
88 (58.7) 62 (41.3)
74 (23.6) 239 (76.4) 3 (2e4) 10 (5e22) 241 (77) 198 (63.2) 43 (13.7) 29 (9.2) 86 (27.4) 22 (13e40.5)
25 (16.7) 125 (83.3) 3 (2e4) 4 (12e21.5) 117 (78) 92 (61.3) 25 (16.6) 14 (9.3) 21 (14) 8 (5e17)
Odds Ratio (95% confidence interval)
P Value
1.41 (0.93e2.12)
0.08
0.97 (0.65e1.45)
0.92
1.54 (0.91e2.67)
0.09
e e 1.4 (0.88e2.2) 1.08 (0.71e1.65) 0.79 (0.45e1.42) 0.99 (0.48e2.1) 2.32 (1.35e4.13) e
0.78 0.38 0.12 0.7 0.4 0.98 0.001 <0.001
DOT, days of therapy; GI, gastrointestinal; IQR, interquartile range.
treatment duration. It may take up to 6 days for CDI to occur after the first dose of cefazolin is administered (Fig. 3). 3.3. Secondary outcomes Fig. 4 shows the impact of factors known to increase the risk of CDI development on our study population. Increased age and presence of lower GI disease did not increase CDI risk, as did the use of acid suppression therapy. The latter was true for both PPIs and H2RAs. The only factor that was considerably associated with increased CDI risk was the use of antineoplastic agents (OR, 2.32; 95% CI, 1.35e4.13; P < 0.01). 3.4. Controlling for antineoplastic therapy ORs of cefepime and cefazolin to increase CDI risk were adjusted for antineoplastic therapy and vice versa. Adjusted OR of cefepime was 1.86 (95% CI, 1.38e2.48; P < 0.001), which indicated that cefepime still carried the risk for association with CDI even without antineoplastic therapy. Similarly, cefazolin adjusted OR was 1.5
(95% CI, 1.02e2.21; P ¼ 0.04). Use of antineoplastic agents showed ORs of 2.08 (95% CI, 1.22e3.55; P ¼ 0.007) and 2.49 (95% CI, 1.47e4.23; P ¼ 0.001) when adjusted for cefepime and cefazolin therapies, respectively. Therefore, the presence of antineoplastic agents as such was significantly associated with CDI regardless of therapies with cefepime and cefazolin.
4. Discussion We found that only cefepime and cefazolin were significantly associated with increasing CDI risk in our institution. These findings are similar to those by Deshpande et al. in their meta-analysis on community-acquired CDI where they found that the cephalosporins class was the third most likely class to increase the risk of CDI development after clindamycin and fluoroquinolones [5]. The major indication for cefazolin in our study population was perioperative prophylaxis. This was confirmed in other studies as well as in ours [7,8]. It may take up to 8 and 6 days for CDI to occur with cefepime and cefazolin, respectively. Having this knowledge about
A.K. Thabit et al. / Anaerobe 59 (2019) 126e130
129
Fig. 2. Effect of antibiotic type on risk of Clostridioides difficile infection development.
Fig. 3. Median time from administration of antibiotics to the incidence of Clostridioides difficile infection in cases. Note:The numbers in between parentheses represent the number of patients received the corresponding antibiotic.
Fig. 4. Effects of age, acid suppression therapy, and use of antineoplastic agents on Clostridioides difficile infection development.
the duration can guide healthcare providers to consider stopping antibiotic therapy or switching to antibiotics less likely to be associated with increased CDI risk in case longer treatment durations are needed or expected. Of our secondary outcomes, the use of antineoplastic agents was the only risk factor that was significantly associated with CDI occurrence. A previous study by Ergen et al. showed a similar effect [9]. This is also analogous to the results reported by Blot et al. where cancer chemotherapy without antibiotic therapy was a major risk factor for CDI development [10]. Nonetheless, the addition of antibiotics increased this risk remarkably (P ¼ 0.008) [10]. Our study was limited by several factors. First, some antibiotics were administered more frequently than others in our hospital. For example, cefepime and cefazolin were among the highly utilized antibiotics, with a mean proportion of utilization for all study
130
A.K. Thabit et al. / Anaerobe 59 (2019) 126e130
patients that reached 46% for cefepime (second highly utilized agent) and 21% for cefazolin (fourth highly utilized agent). Second, the acuity of illness within the study population was not adequately captured. For example, patients who were immunocompromised were not specifically identified. However, it is reported that the degree of immunodeficiency mainly predicts the severity of the infection and its duration more than whether a patient is vulnerable to the infection or not [6]. Third, some patients received more than one antibiotic prior to CDI incidence which may have probably created a confounding effect. Nevertheless, the standard clinical practice usually involves the use of more than one antibiotic agent (at least empirically), hence limiting the study to only patients who received one single agent would have been challenging and only result in a very small sample size. Last, the introduction of PCR testing of C. difficile took place in a late phase during our study period, in September 2012. Overcoming this limitation may have helped increasing the sample size of cases given the high sensitivity and specificity of this test (100% and 99.2%, respectively) compared to the toxin assay, which has a sensitivity of 63e94% and a specificity of 75e100% [1,11,12]. As a conclusion, antibiotic use increased the risk of CDI, particularly cefepime and cefazolin after 8 and 6 days of therapy, respectively. On the other hand, ampicillin-sulbactam did not exhibit this risk. Use of antineoplastics was significantly associated with CDI. However, age, use of acid suppression therapy, and the presence of underlying lower GI disease did not increase CDI risk. Funding Not applicable. Conflicts of interest The authors declare that they have no conflict of interest. Ethical approval The study protocol (number 2013P000958) was approved by the Massachusetts General Hospital institutional review board. Informed consent Not applicable. Acknowledgements We thank Irene Goldenshtein, MD and David Hooper, MD (Massachusetts General Hospital Infection Control Unit) for providing the database of CDI cases. We thank Aaron Sacco, and Robert Tewes (MGH Pharmacy) for providing the databases of antibiotics dispensed by the pharmacy during the study period. We
thank Barbra T. Irby, RPh for reviewing the manuscript. We thank Douglas Hayden, PhD and Brian Healy, PhD for providing consults on statistics. This work was conducted with support from Harvard Catalyst | The Harvard Clinical and Translational Science Center (National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health Award 8UL1TR000170-05 and financial contributions from Harvard University and its affiliated academic health care centers). The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard Catalyst, Harvard University and its affiliated academic health care centers, or the National Institutes of Health. We also thank professor Alex van Belkum for critically revising the manuscript.
References [1] L.C. McDonald, D.N. Gerding, S. Johnson, J.S. Bakken, K.C. Carroll, S.E. Coffin, et al., Clinical practice guidelines for Clostridium difficile infection in adults and children: 2017 update by the infectious diseases society of America (IDSA) and society for healthcare epidemiology of America (SHEA), Clin. Infect. Dis. 66 (2018) 987e994. [2] J.A. O'Brien, B.J. Lahue, J.J. Caro, D.M. Davidson, The emerging infectious challenge of Clostridium difficile-associated disease in Massachusetts hospitals: clinical and economic consequences, Infect. Control Hosp. Epidemiol. 28 (2007) 1219e1227. [3] S. Johnson, D.N. Gerding, Clostridium difficile–associated diarrhea, Clin. Infect. Dis. 26 (1998) 1027e1034, quiz 35-6. [4] S.H. Cohen, D.N. Gerding, S. Johnson, C.P. Kelly, V.G. Loo, L.C. McDonald, et al., Clinical practice guidelines for Clostridium difficile infection in adults: 2010 update by the society for healthcare epidemiology of America (SHEA) and the infectious diseases society of America (IDSA), Infect. Control Hosp. Epidemiol. 31 (2010) 431e455. [5] A. Deshpande, V. Pasupuleti, P. Thota, C. Pant, D.D. Rolston, T.J. Sferra, et al., Community-associated Clostridium difficile infection and antibiotics: a metaanalysis, J. Antimicrob. Chemother. 68 (2013) 1951e1961. [6] S.B. Debast, M.P. Bauer, E.J. Kuijper, European society of clinical microbiology and infectious diseases: update of the treatment guidance document for Clostridium difficile infection, Clinical. Microbiol. Infect. : Off. Pub. Europe. Soc. Clin. Microbiol. Infect. Dis. 20 (Suppl 2) (2014) 1e26. [7] J. Yee, C.M. Dixon, A.P. McLean, J.L. Meakins, Clostridium difficile disease in a department of surgery. The significance of prophylactic antibiotics, Arch. Surg. 126 (1991) 241e246 (Chicago, Ill : 1960). [8] G. Privitera, P. Scarpellini, G. Ortisi, G. Nicastro, R. Nicolin, F. de Lalla, Prospective study of Clostridium difficile intestinal colonization and disease following single-dose antibiotic prophylaxis in surgery, Antimicrob. Agents Chemother. 35 (1991) 208e210. [9] E.K. Ergen, H. Akalin, E. Yilmaz, M. Sinirtas, O. Alver, Y. Heper, et al., Nosocomial diarrhea and Clostridium difficile associated diarrhea in a Turkish University Hospital, Med. Maladies Infect. 39 (2009) 382e387. [10] E. Blot, M.C. Escande, D. Besson, F. Barbut, C. Granpeix, B. Asselain, et al., Outbreak of Clostridium difficile-related diarrhoea in an adult oncology unit: risk factors and microbiological characteristics, J. Hosp. Infect. 53 (2003) 187e192. [11] E. de Jong, A.S. de Jong, C.J. Bartels, C. van der Rijt-van den Biggelaar, W.J. Melchers, P.D. Sturm, Clinical and laboratory evaluation of a real-time PCR for Clostridium difficile toxin A and B genes, Eur. J. Clin. Microbiol. Infect. Dis. 31 (2012) 2219e2225. [12] R.M. Humphries, D.Z. Uslan, Z. Rubin, Performance of Clostridium difficile toxin enzyme immunoassay and nucleic acid amplification tests stratified by patient disease severity, J. Clin. Microbiol. 51 (2013) 869e873.
Contents lists available at ScienceDirect
Anaerobe journal homepage: www.elsevier.com/locate/anaerobe
Clostridioides difficile (including epidemiology)
Antibiotic use and duration in association with Clostridioides difficile infection in a tertiary academic medical center: A retrospective casecontrol study Abrar K. Thabit a, b, *, Christy A. Varugehese b, 2, Alexander R. Levine b, 1 a b
Pharmacy Practice Department, Faculty of Pharmacy, King Abdulaziz University, Jeddah, Saudi Arabia Massachusetts General Hospital, Boston, MA, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 17 December 2018 Received in revised form 25 May 2019 Accepted 26 June 2019 Available online 27 June 2019
Previous literature has attributed the use of certain antibiotics to Clostridioides difficile infection (CDI); however, the time from administration to CDI onset is not adequately evaluated. We aimed to determine the type of antibiotics and duration of therapy associated with the highest CDI incidence at a tertiary academic medical center. This was a retrospective, case-control study of adult inpatients who received at least one course of antibiotic treatment. Patients were divided into either cases or controls. For cases, their first episode of CDI was a determining factor. Primary outcome were the types of antibiotics associated with risk of CDI development and the median time of antibiotic usage defining this risk. Of 601 patients who developed CDI, 313 were included as cases while 150 of 291 who received antibiotics but did not develop CDI were included as controls. Cefepime and cefazolin were significantly associated with increased risk for CDI with odds ratios of 3.01 (95% CI, 1.96e4.65; P < 0.001) and 1.71 (95% CI, 1.02 e2.95; P < 0.05), respectively. Cefepime was associated with CDI after a median time of 8 days while CDI may have occurred after 6 days of therapy with cefazolin. Use of antineoplastic agents was significantly associated with CDI (odds ratio, 2.32; 95% CI, 1.35e4.13; P < 0.01). Antibiotic use increased the risk of CDI, particularly with cefepime and cefazolin with a median time to incidence of 8 and 6 days, respectively. CDI risk was also increased with the use of antineoplastic agents. © 2019 Elsevier Ltd. All rights reserved.
Handling Editor: Stuart Johnson Keywords: Clostridioides difficile infection Antibiotic Antineoplastics Cefepime Cefazolin
1. Introduction Clostridioides difficile is the major cause of infectious nosocomial diarrhea in the United States that usually develops after antibiotic therapy, accounting for 20e30% of cases of nosocomial antibioticassociated diarrhea [1]. It is estimated that C. difficile infection (CDI) is associated with an increased burden to the American health care system with a total exceeding $3 billion per year [2]. Two rationales for CDI pathogenesis currently exist. Upon exposure to antibiotics, intestinal flora other than C. difficile becomes suppressed and disrupted allowing C. difficile to overgrow. Exogenous acquisition of C. difficile creates the basis for the second
* Corresponding author. Pharmacy Practice Department, Faculty of Pharmacy, King Abdulaziz University, 7027 Abdullah Al-Sulaiman Rd, Jeddah, 22254-2265, Saudi Arabia. E-mail address: [email protected] (A.K. Thabit). 1 Current affiliation: University of Saint Joseph, Hartford, CT, USA. 2 Current affiliation: Rush University Medical Center, Chicago, IL, USA. https://doi.org/10.1016/j.anaerobe.2019.06.016 1075-9964/© 2019 Elsevier Ltd. All rights reserved.
rationale [3]. As a result, the development of CDI requires two major components, exposure to antibiotic therapy and presence of toxigenic C. difficile [3]. About 96% of patients who had received antibiotic therapy reported symptomatic diarrhea within 14 days where the median time between the initiation of antibiotic therapy and CDI onset was 2e3 days [4]. Antibiotics that are most implicated in causing CDI are clindamycin, fluoroquinolones, and cephalosporins, whereas parenteral aminoglycosides, vancomycin, and metronidazole are less frequently associated with the disease [3]. A healthcare institution may benefit from knowing its specific antibiotic-associated CDI patterns based on the institution's antibiotic formulary and patient population [3,5]. Moreover, by knowing the timeframe between the administration of culprit antibiotics and CDI onset, healthcare providers can probably consider early treatment discontinuation if antibiotics are no longer needed or switch therapy to agents with lower likelihood of association with CDI in case prolonged therapy is needed or expected. In addition to antibiotic use, several factors were reported to
A.K. Thabit et al. / Anaerobe 59 (2019) 126e130
increase the risk of CDI. Such factors include age equal to or greater than 65 years, antineoplastic therapy, infection with human immunodeficiency virus, gastrointestinal (GI) surgery or manipulation of the GI tract (e.g., tube feeding), and the use of proton pump inhibitors (PPIs) or H2-receptor antagonists (H2RAs) [1]. We conducted this study in order to find out the median time from the administration to the onset of CDI of antibiotics mostly associated with the infection within a tertiary academic medical center. 2. Materials and methods 2.1. Study design and setting This retrospective case-control study was conducted at Massachusetts General Hospital (Boston, MA, USA), a tertiary academic medical center, from January 2010 through December 2012. The study was approved by the Institutional Review Board of Partners Healthcare System (protocol # 2013P000958). 2.2. Data sources We obtained a database from Massachusetts General Hospital Infection Control Unit of all patients who developed CDI during the study period (defined as cases). The Department of Pharmacy provided databases of all antibiotics that were dispensed in the hospital during the study period. Patients in these databases were assessed for eligibility as controls (defined as patients who received antibiotics but did not develop CDI). CDI was defined when a polymerase chain reaction (PCR) test and/or toxin assay of C. difficile was positive [6]. Data of primary and secondary endpoints, as well as data of patient characteristics were obtained from the hospital's electronic records. 2.3. Patients Cases and controls were included in a 2:1 ratio. Patients were deemed eligible if they were 18 years of age or older, initially admitted to our hospital, and received at least one antibiotic during the duration of hospital stay. For cases, the CDI episode had to be their first in order to be included in the study. Patients who had a previous history of CDI (i.e., current episode is a recurrence), developed their first CDI immediately before hospital admission, were exposed to antibiotics within one month before hospital admission, or were recently admitted to a hospital within a month were excluded from the study. Both patients of general care and critical care services were included. If a patient was transferred from a critical care unit to a general care unit (or vice versa), the patient was assigned to the unit at which CDI was acquired (within 24 h of transfer). If a patient was admitted through the emergency department (ED), the duration of hospital stay was counted starting from the day of presentation at the ED. 2.4. Outcomes The primary outcome was types of antibiotic therapy associated with the development of CDI. We assessed this outcome by looking at the electronic medication administration records and counting the days of therapy of each antibiotic administered. If antibiotic administration was stopped and reinitiated after several days, all days of antibiotic administration were counted. For cases, antibiotic days were counted until CDI. Secondary outcomes included effects of confounding factors on CDI risk. Confounding factors included age, acid suppression
127
therapy (PPIs and H2-RAs), and antineoplastic agents. These factors were assessed from each patient's electronic records. 2.5. Statistical analysis In order to assess the association of antibiotics and CDI development, odds ratios (OR) and 95% confidence intervals (CI) were estimated using Cox proportional hazards regression, with covariate adjustment for the antibiotic administered. Median time until CDI was calculated in cases for each antibiotic administered. For secondary outcomes, ORs and 95% CIs were estimated using Cox proportional hazards regression, with covariate adjustment for the confounding factor in question. We compared select baseline characteristics between cases and controls by estimating ORs and 95% CIs. The number of antibiotics received by each patient, days of therapy of all antibiotics received, and the total duration of hospital stay were presented as a median of days given the lack of normal distribution of the data as demonstrated by the lack of significance in the Sahpiro-Wilk test for normality. Logistic regression was used to adjust OR of primary outcome for any confounding factor that showed statistical significant association with CDI development. A P value of <0.05 was considered statistically significant. We used STATA software package 10.1 (StataCorp, College Station, TX) and Microsoft Excel 14.0 (Microsoft Inc., Redmond, WA) for our analyses and chart development, respectively. 3. Results 3.1. Patients and characteristics A total of 892 patients were assessed for eligibility (Fig. 1). Of the 601 patients from the cases database, 313 patients were included in the study. Antibiotic treatment before hospital admission was the major cause for exclusion from the study (n ¼ 48) followed by the transfer from an outside hospital (n ¼ 41). One hundred and fifty patients were included as controls and 141 were excluded out of 291 patients who received antibiotics but did not develop CDI. Most of the excluded patients from this group were aged less than 18 years (n ¼ 42). The number of included CDI cases that were detected by PCR was 32 (10.2%), whereas toxigenic assay detected the rest of the included cases (n ¼ 281, 89.8%). Overall, cases and controls did not significantly differ in their baseline characteristics, except for the use of antineoplastic agents and the length of hospital stay with (P < 0.01 for both), respectively (Table 1). More patients in both groups were admitted to general care units; however, this factor was not significantly different. 3.2. Primary outcome A total of 41 different species of antibiotics were administered to the study population. Two antibiotics were significantly associated with increased risk of CDI development. Cefepime showed the highest OR of 3.01 (95% CI, 1.96e4.65; P < 0.001), while cefazolin had an OR of 1.71 (95% CI, 1.02e2.95; P < 0.05) (Fig. 2). The remaining 39 antibiotics did not show significant increase in CDI risk. Ampicillin-sulbactam showed a significant protective effect (P < 0.05) in not increasing the risk of CDI acquisition. Interestingly, antibiotics that are known to increase the risk of CDI development, such as clindamycin and fluoroquinolones, did not show an increased risk in our study; however these findings were not statistically significant. The median duration in days for CDI to occur on all the antibiotics ranged between 2 and 26 days. Once cefepime was administered, CDI was estimated to occur at the eighth day whether the antibiotic was continued or discontinued irrespective of the
128
A.K. Thabit et al. / Anaerobe 59 (2019) 126e130
Fig. 1. Flowchart of study subjects.
Table 1 Baseline characteristics of cases and controls. Characteristic Age, n (%) 65 years <65 years Gender, n (%) Male Female Hospital service, n (%) Critical care General care Number of antibiotics received per patient, median (IQR) DOT of all antibiotics received, median (IQR) Use of acid suppression therapy, n (%) Proton pump inhibitors H2-receptor antagonists Underlying lower GI disease, n (%) Use of antineoplastic agents, n (%) Total duration of hospital stay, median (IQR), days
Cases (n ¼ 313)
Controls (n ¼ 150)
173 (55.3) 140 (44.7)
70 (46.7) 80 (53.3)
182 (58.1) 131 (41.9)
88 (58.7) 62 (41.3)
74 (23.6) 239 (76.4) 3 (2e4) 10 (5e22) 241 (77) 198 (63.2) 43 (13.7) 29 (9.2) 86 (27.4) 22 (13e40.5)
25 (16.7) 125 (83.3) 3 (2e4) 4 (12e21.5) 117 (78) 92 (61.3) 25 (16.6) 14 (9.3) 21 (14) 8 (5e17)
Odds Ratio (95% confidence interval)
P Value
1.41 (0.93e2.12)
0.08
0.97 (0.65e1.45)
0.92
1.54 (0.91e2.67)
0.09
e e 1.4 (0.88e2.2) 1.08 (0.71e1.65) 0.79 (0.45e1.42) 0.99 (0.48e2.1) 2.32 (1.35e4.13) e
0.78 0.38 0.12 0.7 0.4 0.98 0.001 <0.001
DOT, days of therapy; GI, gastrointestinal; IQR, interquartile range.
treatment duration. It may take up to 6 days for CDI to occur after the first dose of cefazolin is administered (Fig. 3). 3.3. Secondary outcomes Fig. 4 shows the impact of factors known to increase the risk of CDI development on our study population. Increased age and presence of lower GI disease did not increase CDI risk, as did the use of acid suppression therapy. The latter was true for both PPIs and H2RAs. The only factor that was considerably associated with increased CDI risk was the use of antineoplastic agents (OR, 2.32; 95% CI, 1.35e4.13; P < 0.01). 3.4. Controlling for antineoplastic therapy ORs of cefepime and cefazolin to increase CDI risk were adjusted for antineoplastic therapy and vice versa. Adjusted OR of cefepime was 1.86 (95% CI, 1.38e2.48; P < 0.001), which indicated that cefepime still carried the risk for association with CDI even without antineoplastic therapy. Similarly, cefazolin adjusted OR was 1.5
(95% CI, 1.02e2.21; P ¼ 0.04). Use of antineoplastic agents showed ORs of 2.08 (95% CI, 1.22e3.55; P ¼ 0.007) and 2.49 (95% CI, 1.47e4.23; P ¼ 0.001) when adjusted for cefepime and cefazolin therapies, respectively. Therefore, the presence of antineoplastic agents as such was significantly associated with CDI regardless of therapies with cefepime and cefazolin.
4. Discussion We found that only cefepime and cefazolin were significantly associated with increasing CDI risk in our institution. These findings are similar to those by Deshpande et al. in their meta-analysis on community-acquired CDI where they found that the cephalosporins class was the third most likely class to increase the risk of CDI development after clindamycin and fluoroquinolones [5]. The major indication for cefazolin in our study population was perioperative prophylaxis. This was confirmed in other studies as well as in ours [7,8]. It may take up to 8 and 6 days for CDI to occur with cefepime and cefazolin, respectively. Having this knowledge about
A.K. Thabit et al. / Anaerobe 59 (2019) 126e130
129
Fig. 2. Effect of antibiotic type on risk of Clostridioides difficile infection development.
Fig. 3. Median time from administration of antibiotics to the incidence of Clostridioides difficile infection in cases. Note:The numbers in between parentheses represent the number of patients received the corresponding antibiotic.
Fig. 4. Effects of age, acid suppression therapy, and use of antineoplastic agents on Clostridioides difficile infection development.
the duration can guide healthcare providers to consider stopping antibiotic therapy or switching to antibiotics less likely to be associated with increased CDI risk in case longer treatment durations are needed or expected. Of our secondary outcomes, the use of antineoplastic agents was the only risk factor that was significantly associated with CDI occurrence. A previous study by Ergen et al. showed a similar effect [9]. This is also analogous to the results reported by Blot et al. where cancer chemotherapy without antibiotic therapy was a major risk factor for CDI development [10]. Nonetheless, the addition of antibiotics increased this risk remarkably (P ¼ 0.008) [10]. Our study was limited by several factors. First, some antibiotics were administered more frequently than others in our hospital. For example, cefepime and cefazolin were among the highly utilized antibiotics, with a mean proportion of utilization for all study
130
A.K. Thabit et al. / Anaerobe 59 (2019) 126e130
patients that reached 46% for cefepime (second highly utilized agent) and 21% for cefazolin (fourth highly utilized agent). Second, the acuity of illness within the study population was not adequately captured. For example, patients who were immunocompromised were not specifically identified. However, it is reported that the degree of immunodeficiency mainly predicts the severity of the infection and its duration more than whether a patient is vulnerable to the infection or not [6]. Third, some patients received more than one antibiotic prior to CDI incidence which may have probably created a confounding effect. Nevertheless, the standard clinical practice usually involves the use of more than one antibiotic agent (at least empirically), hence limiting the study to only patients who received one single agent would have been challenging and only result in a very small sample size. Last, the introduction of PCR testing of C. difficile took place in a late phase during our study period, in September 2012. Overcoming this limitation may have helped increasing the sample size of cases given the high sensitivity and specificity of this test (100% and 99.2%, respectively) compared to the toxin assay, which has a sensitivity of 63e94% and a specificity of 75e100% [1,11,12]. As a conclusion, antibiotic use increased the risk of CDI, particularly cefepime and cefazolin after 8 and 6 days of therapy, respectively. On the other hand, ampicillin-sulbactam did not exhibit this risk. Use of antineoplastics was significantly associated with CDI. However, age, use of acid suppression therapy, and the presence of underlying lower GI disease did not increase CDI risk. Funding Not applicable. Conflicts of interest The authors declare that they have no conflict of interest. Ethical approval The study protocol (number 2013P000958) was approved by the Massachusetts General Hospital institutional review board. Informed consent Not applicable. Acknowledgements We thank Irene Goldenshtein, MD and David Hooper, MD (Massachusetts General Hospital Infection Control Unit) for providing the database of CDI cases. We thank Aaron Sacco, and Robert Tewes (MGH Pharmacy) for providing the databases of antibiotics dispensed by the pharmacy during the study period. We
thank Barbra T. Irby, RPh for reviewing the manuscript. We thank Douglas Hayden, PhD and Brian Healy, PhD for providing consults on statistics. This work was conducted with support from Harvard Catalyst | The Harvard Clinical and Translational Science Center (National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health Award 8UL1TR000170-05 and financial contributions from Harvard University and its affiliated academic health care centers). The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard Catalyst, Harvard University and its affiliated academic health care centers, or the National Institutes of Health. We also thank professor Alex van Belkum for critically revising the manuscript.
References [1] L.C. McDonald, D.N. Gerding, S. Johnson, J.S. Bakken, K.C. Carroll, S.E. Coffin, et al., Clinical practice guidelines for Clostridium difficile infection in adults and children: 2017 update by the infectious diseases society of America (IDSA) and society for healthcare epidemiology of America (SHEA), Clin. Infect. Dis. 66 (2018) 987e994. [2] J.A. O'Brien, B.J. Lahue, J.J. Caro, D.M. Davidson, The emerging infectious challenge of Clostridium difficile-associated disease in Massachusetts hospitals: clinical and economic consequences, Infect. Control Hosp. Epidemiol. 28 (2007) 1219e1227. [3] S. Johnson, D.N. Gerding, Clostridium difficile–associated diarrhea, Clin. Infect. Dis. 26 (1998) 1027e1034, quiz 35-6. [4] S.H. Cohen, D.N. Gerding, S. Johnson, C.P. Kelly, V.G. Loo, L.C. McDonald, et al., Clinical practice guidelines for Clostridium difficile infection in adults: 2010 update by the society for healthcare epidemiology of America (SHEA) and the infectious diseases society of America (IDSA), Infect. Control Hosp. Epidemiol. 31 (2010) 431e455. [5] A. Deshpande, V. Pasupuleti, P. Thota, C. Pant, D.D. Rolston, T.J. Sferra, et al., Community-associated Clostridium difficile infection and antibiotics: a metaanalysis, J. Antimicrob. Chemother. 68 (2013) 1951e1961. [6] S.B. Debast, M.P. Bauer, E.J. Kuijper, European society of clinical microbiology and infectious diseases: update of the treatment guidance document for Clostridium difficile infection, Clinical. Microbiol. Infect. : Off. Pub. Europe. Soc. Clin. Microbiol. Infect. Dis. 20 (Suppl 2) (2014) 1e26. [7] J. Yee, C.M. Dixon, A.P. McLean, J.L. Meakins, Clostridium difficile disease in a department of surgery. The significance of prophylactic antibiotics, Arch. Surg. 126 (1991) 241e246 (Chicago, Ill : 1960). [8] G. Privitera, P. Scarpellini, G. Ortisi, G. Nicastro, R. Nicolin, F. de Lalla, Prospective study of Clostridium difficile intestinal colonization and disease following single-dose antibiotic prophylaxis in surgery, Antimicrob. Agents Chemother. 35 (1991) 208e210. [9] E.K. Ergen, H. Akalin, E. Yilmaz, M. Sinirtas, O. Alver, Y. Heper, et al., Nosocomial diarrhea and Clostridium difficile associated diarrhea in a Turkish University Hospital, Med. Maladies Infect. 39 (2009) 382e387. [10] E. Blot, M.C. Escande, D. Besson, F. Barbut, C. Granpeix, B. Asselain, et al., Outbreak of Clostridium difficile-related diarrhoea in an adult oncology unit: risk factors and microbiological characteristics, J. Hosp. Infect. 53 (2003) 187e192. [11] E. de Jong, A.S. de Jong, C.J. Bartels, C. van der Rijt-van den Biggelaar, W.J. Melchers, P.D. Sturm, Clinical and laboratory evaluation of a real-time PCR for Clostridium difficile toxin A and B genes, Eur. J. Clin. Microbiol. Infect. Dis. 31 (2012) 2219e2225. [12] R.M. Humphries, D.Z. Uslan, Z. Rubin, Performance of Clostridium difficile toxin enzyme immunoassay and nucleic acid amplification tests stratified by patient disease severity, J. Clin. Microbiol. 51 (2013) 869e873.