Hospital Ward Antibiotic Prescribing and the Risks of Clostridium difficile Infection

Ward Antibiotic Prescribing and the Risks of Clostridiumdifficile InfectionKevin Brown, PhD; Kim Valenta, PhD; David Fisman, MD, MSc; Andrew Simor, MD; Nick Daneman, MD, MSc

IMPORTANCE Only a portion of hospital-acquired Clostridium difficile infections can be tracedback to source patients identified as having symptomatic disease. Antibiotic exposure is themain risk factor for C difficile infection for individual patients and is also associated withincreased asymptomatic shedding. Contact with patients taking antibiotics within the sameward may be a transmission risk factor for C difficile infection, but this hypothesis has neverbeen tested.

OBJECTIVES To obtain a complete portrait of inpatient risk that incorporates innate patientrisk factors and transmission risk factors measured at the ward level and to investigateward-level rates of antibiotic use and C difficile infection risk.

DESIGN, SETTING, AND PATIENTS A 46-month (June 1, 2010, through March 31, 2014)retrospective cohort study of inpatients aged 18 years or older in a large, acute care teachinghospital composed of 16 wards, including 5 intensive care units and 11 non–intensive care unitwards.

EXPOSURES Patient-level risk factors (eg, age, comorbidities, hospitalization history, andantibiotic exposure) and ward-level risk factors (eg, antibiotic therapy per 100 patient-days,hand hygiene adherence, and mean patient age) were identified from hospital databases.

MAIN OUTCOMES AND MEASURES Incidence of hospital-acquired C difficile infection asidentified prospectively by hospital infection prevention and control staff.

RESULTS A total of 255 of 34 298 patients developed C difficile (incidence rate, 5.95 per10 000 patient-days; 95% CI, 5.26-6.73). Ward-level antibiotic exposure varied from 21.7 to56.4 days of therapy per 100 patient-days. Each 10% increase in ward-level antibioticexposure was associated with a 2.1 per 10 000 (P < .001) increase in C difficile incidence. Theassociation between C difficile incidence and ward antibiotic exposure was the same amongpatients with and without recent antibiotic exposure, and C difficile risk persisted aftermultilevel, multivariate adjustment for differences in patient-risk factors among wards(relative risk, 1.34 per 10% increase in days of therapy; 95% CI, 1.16-1.57).

CONCLUSIONS AND RELEVANCE Among hospital inpatients, ward-level antibiotic prescribing isassociated with a statistically significant and clinically relevant increase in C difficile risk thatpersists after adjustment for differences in patient-level antibiotic use and other patient- andward-level risk factors. These data strongly support the use of antibiotic stewardship as ameans of preventing C difficile infection.

JAMA Intern Med. doi:10.1001/jamainternmed.2014.8273Published online February 23, 2015.

Invited Commentary

Author Affiliations: Division ofEpidemiology, Dalla Lana School ofPublic Health, University of Toronto,Toronto, Ontario, Canada (Brown,Fisman); Department ofAnthropology and School of theEnvironment, McGill University,Montreal, Quebec, Canada (Valenta);Sunnybrook Health Sciences Centre,University of Toronto, Toronto,Ontario, Canada (Simor, Daneman).

Corresponding Author: NickDaneman, MD, MSc, Division ofInfectious Diseases and ClinicalEpidemiology, Sunnybrook HealthSciences Centre, University ofToronto, 2075 Bayview Ave, Toronto,ON, M4N 3M5, Canada([email protected]).

jamanetwork/2015/imd/02_23_2015/ioi140154pap PAGE: right 1 SESS: 8 OUTPUT: Jan 20 12:33 2015

Research

Original Investigation | LESS IS MORE

(Reprinted) E1

A ntibiotic exposure represents the principal risk factorfor Clostridium difficile infection, and existing re-search estimates that inpatients taking antibiotics are,

on average, 60% more likely to acquire the infection.1 Pro-longed antibiotic exposure and exposure to larger antibioticdoses are associated with increased C difficile infection risk,2

and some antibiotics (clindamycin, cephalosporins, and fluo-roquinolones) are associated with a greater risk relative to otherantibiotic classes.3,4 Risk may increase over time with in-creased prescribing of certain antimicrobials.5

Important gaps in knowledge remain with respect to thenatural history of how C difficile bacteria are transmitted amonghospitalized patients. Hospital environments are persis-tently contaminated with C difficile spores, and surfaces inrooms of infected patients are contaminated before, during,and after treatment for C difficile infection.6 Exposure to symp-tomatic patients with C difficile infection has been identifiedas an independent risk factor for transmission.7 However, ex-posure to spores from other symptomatic patients may not ex-plain most new cases of C difficile infection acquired inhospitals.8 In a C difficile outbreak in a long-term care facility,almost half of the residents had asymptomatic colonization,and antibiotic exposure was the primary risk factor for asymp-tomatic colonization.9 Although asymptomatically colo-nized individuals contribute less to environmental contami-nation at an individual level, asymptomatic carriers outnumbersymptomatic patients by a ratio of 3:1 and as such could rep-resent an important source of C diff ic i le infectiontransmission.10

In the absence of reliable measures of patient coloniza-tion and environmental contamination, transmission risks canpotentially be estimated as a function of aggregated mea-sures of patient risk factors for colonization, such as meanward- or hospital-level antibiotic prescribing.11 Multilevel mod-els can be used to tease apart the effect of individual-level riskfactors that affect patient susceptibility (direct effects) fromgroup-level effects that affect transmission risks that are in-dependent of individual-level effects (indirect effects).12 Wesought to establish the effect of ward antibiotic-prescribing rateon ward C difficile infection incidence and whether the ef-fects observed extended beyond the direct antibiotic effectson patients’ infection risk.

MethodsEthics StatementStudy approval was obtained from the Research Ethics Boardof Sunnybrook Health Sciences Centre. The board waived theneed for patient consent because there was no contact withpatients and patient anonymity was assured.

Study Design and ParticipantsA retrospective cohort study design was used to assess the as-sociation of individual- and ward-level risk factors with the in-cidence of C difficile infection among patients admitted to Sun-nybrook Hospital, a large, acute care teaching hospital locatedin Toronto, Ontario, Canada. The source cohort consisted of

all patients older than 18 years without a previous C difficileinfection diagnosis who were hospitalized in an acute care wardat Sunnybrook Hospital from June 1, 2010, through March 31,2014. We excluded patients in the hospital’s psychiatry, ob-stetrics, neonatal, and long-term care wards given a low ex-pected event rate of C difficile infection.

Case DefinitionPatients infected with C difficile were identified by the Infec-tion Prevention and Control Department via active surveil-lance during the study period. A C difficile infection case wasdefined as any hospitalized patient with laboratory confirma-tion of a positive toxin assay result together with diarrhea orvisualization of pseudomembranes on sigmoidoscopy, colo-noscopy, or histopathologic analysis.13,14 For the purposes ofcase identification, diarrhea was defined as 3 or more loose orwatery bowel movements in a 24-hour period, which was un-usual or different for the patient, and with no other recog-nized cause. When a patient developed a C difficile infection,the remaining hospitalized days were excluded from the at-risk patient-days. Toxin assays at the hospital have been per-formed by polymerase chain reaction (BD GeneOhm Cdiff; Bec-ton, Dickinson and Company) since September 2009, whichincludes the entire study period.

For C difficile infection case admissions, event time wasthe number of days from hospital admission to symptom on-set or positive toxin assay result for rare cases (<1%) in whichsymptom onset was missing. For noncase admissions, cen-soring time was the number of days from hospital entry untildischarge, study termination, or death. In addition to exclud-ing hospitalized days after C difficile infection, the first 2 daysof each hospital admission were also excluded because pa-tients are not at risk of nosocomial infection at the beginningof a hospital stay.

Antimicrobial Exposure AssessmentPatient antibiotic exposures were drawn from pharmacy dis-pensing records. We examined records for receipt of any an-tibiotic in the prior 10 days but excluded exposure to metro-nidazole, oral vancomycin hydrochloride, or fidaxomicinbecause these may be treatments for C difficile infection.15 An-tibiotic receipt was classified according to the AnatomicalTherapeutic Chemical (ATC) Classification System, 17thedition.16 As per previous work,4 we classified individual pa-tients according to whether they had received a high-risk an-tibiotic (defined as receipt of cephalosporins or carbapen-ems, fluoroquinolones, or clindamycin and other lincosamides;ATC codes: J01D, J01M, and J01FF), had received a medium-risk antibiotic but not a high-risk antibiotic (defined as peni-cillins, sulfonamides and trimethoprim, macrolides and strep-togramins, or aminoglycosides; ATC codes: J01C, J01E, J01FA,J01FG, and J01G), or had received no antibiotics or a low-riskantibiotic only (defined as receipt of tetracyclines; ATC code:J01A).

Patient Risk FactorsPatient age, sex, admission unit (classified as medical, surgi-cal, or oncologic), and number of previous admissions were

jamanetwork/2015/imd/02_23_2015/ioi140154pap PAGE: left 2 SESS: 8 OUTPUT: Jan 20 12:33 2015

Research Original Investigation Clostridium difficile Infection

E2 JAMA Internal Medicine Published online February 23, 2015 (Reprinted) jamainternalmedicine.com

retrieved from hospital administrative records. Any patient re-ceiving insulin or an antidiabetic medication (ATC code: A10)at any point during any hospitalization was considered dia-betic. We also examined the use of antacids (ATC code: A02),chemotherapeutic agents (ATC code: L01), and feeding tubes(gastric, nasogastric, or jejunostomy). To account for the timedelay between transient pharmaceutical exposures and C dif-ficile infection risk, we measured receipt in any of the previ-ous 10 days rather than receipt on a given day.

Ward Risk FactorsUsing hospital bed assignment information, we identified theward occupants for each inpatient day; when a patient was lo-cated in multiple wards on a given day, we considered that pa-tient to be an occupant of the ward on which he or she was lo-cated at noon. We calculated ward-level risk factors thatrepresented mean characteristics of the ward patient popula-tion during the 46-month study period. The following 5 ward-level measures were retrieved from the hospital informationsystem: age (mean age), antibiotic use in days of therapy (DOTs)per 100 patient-days, antacid use (DOTs per 100 patient-days), chemotherapeutic agent use (DOTs per 100 patient-days), and feeding tube use (tube in situ per 100 patient-days). Within each ward, observer nurses measured handhygiene adherence at specific hand hygiene moments (be-fore entering patient room, after leaving patient room, beforeaseptic procedure, and after body fluid contact) on a quar-terly basis through the study period, as per provincialguidelines.17 Adherence was pooled across periods and handhygiene moments and was reported as a percentage of totalhand hygiene opportunities.

Statistical AnalysisPatient Risk FactorsTo estimate the effect of individual risk factors on C difficileinfection risk, we developed a Poisson regression model thataimed to predict the time elapsed from hospital admission tothe occurrence of a first C difficile infection. Our data werestructured in counting process format with one record for eachpatient-day. The crude incidence rate ratio was assessed in aPoisson regression for each of the 12 individual-level risk fac-tors.

Ward Risk FactorsUsing the ward-level risk factors above in addition to ward-level C difficile infection incidence, we developed 5 bivariateinverse-variance–weighted linear mixed-effects regressionmodels to estimate the effect of each ward-level factor on Cdifficile infection incidence, which were fitted using the Har-tung-Knapp-Sidik-Jonkman method.18 We also considered thebest-fitting, 2-covariate, ward-level model by comparing modelAkaike Information Criterion for all 15 two-covariate models.As a sensitivity analysis, we examined the association be-tween ward-level antibiotic use and C difficile infection inci-dence among the 11 non–intensive care unit (ICU) wards sepa-rately, excluding the 5 ICU wards.

To clearly distinguish patient-level and ward-level antibi-otic effects, we measured the association between ward-

level antibiotic prescribing and C difficile risk separately in pa-tients with and without direct recent antibiotic exposure. Wetested whether there was a difference in association of ward-level antibiotic use and C difficile risk between the 2 groupsusing the Δ method.

Multilevel ModelTo assess the independent effect of individual exposures andaggregate ward-level antibiotic exposure, we developed a mul-tilevel Poisson regression model with random intercepts cor-responding to wards. The multilevel model included 8 indi-vidual-level risk factors: time since admission (modeled as aspline with a knot at 5 days for first admission and readmis-sion separately), patient age (per 10-year increase), sex, dia-betes mellitus, and individual exposure to antibiotics, gastricacid inhibitors, chemotherapeutic agents, and presence of afeeding tube. The number of adjustment factors was re-stricted to ensure at least 10 events per covariate,19 and the se-lection of covariates was based on established associations withC difficile infection risk.2,20

Analyses were conducted using R statistical software, ver-sion 3.0.2 (R Foundation for Statistical Computing); the glm,rma, and glmer functions were used for the unadjusted, bi-variate mixed-effects and the multivariate mixed-effects sta-tistical models, respectively.

ResultsInpatient CohortWe identified 34 298 patients who had an acute care hospitalstay that exceeded 2 days at Sunnybrook Hospital from June1, 2010, through March 31, 2014. These patients spent 428 588patient-days in the 16 study wards during the 46-month studyperiod. The median age of the cohort was 68.4 years (inter-quartile range, 54.3-81.0 years), whereas 9718 (28.3%) of the34 298 patients had additional admissions. Patients receivedantibiotics in 21 239 (45.5%) of 46 661 admissions and had feed-ing tubes in 4765 (10.2%) of 46 661 admissions.

Patients Developing C difficile InfectionWe identified 255 patients developing a new-onset C difficileinfection during the 46-month study period (incidencerate, 5.95 per 10 000 patient-days; 95% CI, 5.26-6.73). Caseswere distributed across 3 types of admitting services, with 111among patients admitted via surgery services, 110 admitted viamedicine services, and 34 admitted via oncology services.

Individual Patient Characteristics and the Risk of InfectionThe incidence rates for patients with and without individual-level risk factors are given in Table 1. The individual-level riskfactors associated with C difficile infection were age, readmis-sion, direct exposure to antibiotics, and use of a feeding tube.Each 10-year increase in age was associated with a 1.07-foldincrease in C difficile infection risk (95% CI, 1.00-1.17). Havinga previous admission was associated with a 1.42-fold in-crease in risk (95% CI, 1.10-1.82).


Clostridium difficile Infection Original Investigation Research

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine Published online February 23, 2015 E3

Ward Characteristics and C difficile Infection IncidenceThe 16 study wards included 2 level II ICUs (patients requir-ing detailed observation) and 3 level III ICUs (patients requir-ing advanced respiratory support), 2 cardiology wards, 4 in-ternal medicine wards, 4 surgery wards, and 1 oncology ward.Ward-level characteristics are given in Table 2. The rate of an-tibiotic use in wards varied from 21.7 DOTs per 100 patient-days in ward 6 to 56.4 DOTs per 100 patient-days in ICU 3 (me-dian, 30.6 DOTs per 100 patient-days; interquartile range, 26.6-36.9 DOTs). Mean antibiotic use in the 5 ICUs was 47.2 DOTsper 100 patient-days compared with 30.9 DOTs per 100 patient-days in non-ICU wards (P < .001).

At the ward level, antibiotic use was the strongest predic-tor of C difficile infection incidence (Figure 1). Each 10% in-

crease in ward-level antibiotic use was associated with an in-creased incidence of C difficile infection of 2.1 per 10 000patient-days (slope = 2.1, P < .001, R2 = 0.50). The largest nega-tive outlier in the association was ICU 5, which was the hos-pital burn ICU. Rate of ward-level feeding tube exposure wasmarginally associated with C difficile infection (slope = 0.59,P = .10, R2 = 0.11). Other ward-level factors, including hand hy-giene adherence, mean inpatient age, and rates of antacid useand chemotherapeutic agent use, were not significantly asso-ciated with C difficile infection incidence. The addition of anyof the other 4 ward-level factors to the model with ward-levelantibiotic use did not alter the association between ward-level antibiotic use and C difficile infection incidence (data notshown). When we examined the association between ward-

Table 1. Individual-Level Risk Factors and Clostridium difficile Infection Incidence

Risk FactorNo. ofCases

No. ofPatient-days

Incidence per 10 000Patient-days (95% CI) Relative Risk (95% CI)

Age, y 1.07 (1.00-1.17)a

<50 35 66 530 5.3 (3.8-7.3)

50-59 24 55 321 4.3 (2.9-6.5)

60-69 43 80 493 5.3 (4.0-7.2)

70-79 63 93 691 6.7 (5.3-8.6)

≥80 90 132 553 6.8 (5.5-8.3)

Sex

Male 140 235 784 5.9 (5.0-7.0) 1.00 (0.78-1.27)

Female 115 192 804 6.0 (5.0-7.2) 1.00 [Reference]

Hospitalization

First hospitalization 150 286 846 5.2 (4.5-6.1) 1.00 [Reference]

Additionaladmission(s)

105 141 742 7.4 (6.1-9.0) 1.42 (1.10-1.82)

Admission service

Medicine 110 188 869 5.8 (4.8-7.0) 1.00 [Reference]

Oncology 34 48 953 6.9 (5.0-9.7) 1.19 (0.81-1.88)

Surgery 111 190 766 5.8 (4.8-7.0) 1.00 (0.77-1.34)

Diabetes mellitus

No 176 297 978 5.9 (5.1-6.8) 1.00 [Reference]

Yes 79 130 610 6.0 (4.9-7.5) 1.02 (0.79-1.34)

Any antibiotic

No 63 191 604 3.3 (2.6-4.2) 1.00 [Reference]

Yes 192 236 984 8.1 (7.0-9.3) 2.46 (1.85-3.28)

Antibiotic risk index

None or low 65 196 073 3.3 (2.6-4.2) 1.00 [Reference]

Medium 23 47 663 4.8 (3.2-7.3) 1.46 (0.90-2.34)

High 167 184 852 9.0 (7.8-10.5) 2.73 (2.05-3.63)

Antacid exposure inprevious 10 d

No 65 115 764 5.6 (4.4-7.2) 1.00 [Reference]

Yes 190 312 824 6.1 (5.3-7.0) 1.08 (0.82-1.43)

Chemotherapeuticagent exposure inprevious 10 d

No 214 375 439 5.7 (5.0-6.5) 1.00 [Reference]

Yes 41 53 149 7.7 (5.7-10.5) 1.35 (0.97-1.89)

Feeding tube in situ inprevious 10 d

No 176 330 733 5.3 (4.6-6.2) 1.00 [Reference]

Yes 79 97 855 8.1 (6.5-10.1) 1.52 (1.16-1.98) a Per 10-year increase in age.




level antibiotic use and C difficile infection incidence amongthe 11 non-ICU wards, the association remained statistically sig-nificant (slope = 4.1, P = .03, R2 = 0.41).

We separately measured the association between C diffi-cile infection incidence and ward antibiotic exposure rateamong patients recently exposed and those not recently ex-posed to antibiotics (Figure 2). Each 10% increase in ward-level antibiotic use was associated with a 1.8 per 10 000 in-crease (slope = 1.8, P = .005, R2 = 0.50) in the incidence of Cdifficile infection among patients without direct recent expo-sure and a 1.6 per 10 000 increase (slope = 1.6, P = .05, R2 = 0.14)among patients with direct recent exposure to antibiotics. Theeffect of ward-level antibiotic exposure on C difficile infec-tion incidence did not differ significantly between patients di-rectly using or not directly using antibiotics (P = .16).

Ward-Level Antibiotic Use and C difficile Infection Incidence:Multilevel ModelAfter adjustment for patient characteristics, the ward-level an-tibiotic exposure remained associated with C difficile infec-tion risk (Table 3). Each 10% increase in ward antibiotic expo-sure rate was associated with a 1.34-fold increase in C difficileinfection risk (95% CI, 1.16-1.57).

DiscussionIn this 46-month cohort study of C difficile infection risk, wefound that ward-level antibiotic exposure is the main risk fac-tor for infection. The effect of antibiotic prescribing reachesbeyond individual-level antibiotic use, such that all patients,irrespective of whether they receive antibiotics directly, are athigher risk of C difficile infection in high antibiotic-

prescribing wards. Ward-level C difficile infection risk was notconfounded by other ward-level aggregate patient character-istics, including antacid use, chemotherapy, feeding tube pres-ence, age, or crowding, or by individual-level patient comor-bidities or antibiotic exposures.

This is the first study, to our knowledge, to consider ward-level antibiotic exposure as a risk factor for C difficile infec-tion. In a previous multilevel study considering individual- andhospital-level risk factors, Pakyz et al11 found that hospital-level antibiotic exposure rates were not a significant predic-tor of hospital-level C difficile infection incidence. This find-ing suggests that hospital-level antimicrobial use may not differmeaningfully across centers or that factors that were not con-sidered, such as infection control practices or C difficile diag-nostic testing rate,21 may have confounded an underlying as-sociation.

We hypothesize that the marked effects of ward-level an-tibiotic exposure rate are likely explained by an increase in thenumber of patients colonized with, and shedding, C difficilein wards with high rates of antibiotic use. This high preva-lence of antibiotic use would increase environmental contami-nation and the incidence of C difficile infection. This mecha-nism is supported by research indicating that antibioticexposure is the principal risk factor for C difficile colonization10

and that approximately half of C difficile strains among C dif-ficile infection cases in hospitals cannot be genetically linkedto previously identified symptomatic patients.8 The hospitalburn center was the only outlier, with lower-than-expected Cdifficile infection incidence given its high ward-level antibi-otic use. The burn center is unique in that it had a low nurse-patient staffing ratio, single-bed rooms, and a younger pa-tient population, which is consistent with findings from a priorstudy.22 Our multilevel statistical model revealed that younger

Table 2. Variation in Ward-Level Characteristics and Clostridium difficile Infection Incidence

Variable

Intensive Care Unit Non–Intensive Care Unit Ward

1 2 3 4 5a 6 7 8 9 10 11 12 13 14 15 16Patient-days, inthousands

12.4 7.6 20.8 13.1 11.2 18.4 36.9 26.8 39.8 33.9 20.7 38.6 39.6 39.1 36.3 33.6

Hand hygiene, % 87.2 89.5 84.6 85.4 92.9 84.8 85.5 88.1 88.4 85.6 86.9 90.4 86.3 87.0 86.2 87.3

Mean age, y 60.4 68.4 63.8 69.5 50.3 70.9 71.4 76.7 76.7 74.6 76.1 64.0 67.9 58.2 66.0 64.0

Medicationreceipt, DOTs per100 patient-days

Antibiotics 34.3 50.7 56.4 47.4 44.6 21.7 27.9 33.8 26.6 30.6 33.0 39.6 26.3 29.2 35.1 36.9

Antacids 50.9 53.8 94.4 93.7 76.1 71.3 80.9 59.6 55.0 66.8 56.1 62.8 71.2 58.6 56.1 49.6

Chemotherapeuticagents

13.6 8.5 6.7 2.6 3.5 0.8 1.9 1.9 1.5 1.4 1.9 9.8 10.4 14.0 12.3 14.5

Feeding tube,tube-days per100patient-days

29.6 23.5 85.2 52.5 38.5 12.5 4.0 15.2 13.8 4.8 4.6 4.2 10.5 15.4 3.3 6.0

C difficileinfections

No. 8 15 22 14 3 9 11 15 18 23 10 27 12 19 27 22

Incidence per10 000patient-days

6.4 19.7 10.6 10.7 2.7 4.9 3.0 5.6 4.5 6.8 4.8 7.0 3.0 4.9 7.4 6.6

Abbreviation: DOTs, days of therapy.a Burn intensive care unit.




age and patient pharmaceutical exposures did not com- pletely account for the lower-than-expected incidence in theburn ICU, suggesting that other patient or ward characteris-tics may have been be at play.

The independent association of ward antibiotic exposurewith C difficile infection risk most likely reflects the noninde-pendence of communicable disease cases. Communicable dis-eases differ from other classes of disease because a case is alsoa risk factor.23 In the context of C difficile infection, this state-ment means that an increase in disease-related force of infec-tion could occur via antibiotic exposure in individuals whonever themselves become symptomatic cases.24 Such indi-rect effects are well recognized with communicable diseasecontrol interventions, and indeed this effect may be concep-tualized as an inverse of herd immunity seen with vaccines.25

Analogously, beneficial herd effects would logically be seen inwards with reduced antibiotic prescribing, as was observed inour study. A previous meta-analysis26 of antimicrobial stew-ardship interventions lends credibility to this explanation be-cause these interventions have produced substantial reduc-tions in C difficile infection incidence with only smallreductions in antibiotic prescribing.

As such, the principal clinical implication of this study isthat aggregate ward-level antibiotic use should be subject tosurveillance by infection control and stewardship personnel.

Figure 1. Association of Ward-Level Exposures With Ward Clostridium difficile Infection (CDI) Incidence

25

20

15

10

5

25

20

15

10

5

25

20

15

10

5

020 453025 5535 40 50 60

CDI I

ncid

ence

per

10

000

Patie

nt-d

ays

Ward Antibiotic Use, DOTs per 100 Patient-days

Burn intensive care unitOther intensive care unitsNon-intensive care wards

0

25

20

15

10

5

84 8886 9290 94

CDI I

ncid

ence

per

10

000

Patie

nt-d

ays

Ward Hand Hygiene Adherence, %

040 6050 80 9070 100

CDI I

ncid

ence

per

10

000

Patie

nt-d

ays

Ward Antacid Use, DOTs per 100 Patient-days

00 40 8020 60 100

CDI I

ncid

ence

per

10

000

Patie

nt-d

ays

Ward Feeding Tube Use, Tubes in Situ per 100 Patient-days

A, Antibiotic use; B, hand hygiene; C, antacid use; and D, feeding tube use. Each symbol represents a hospital ward. The size of the symbols is proportional to theamount of follow-up time on each ward. DOTs indicates days of therapy.

Figure 2. Ward Clostridium difficile Infection (CDI) Incidence andAntibiotic Use Across Hospital Wards and Among Patients With andWithout Direct Antibiotic Exposure

20

15

10

5

25

0

20 453025 5535 40 50 60

CDI I

ncid

ence

per

10

000

Patie

nt-d

ays

Ward Antibiotic Use, DOTs per 100 Patient-days

Patients with direct antibiotic exposurePatients without direct antibiotic exposure

69

10 124

5

3

2

16

1514

137

11 1

8

15 4

121614

13

96 7

10

11

18

5

3

2

Each pair of numbered symbols represents the incidence of C difficile infectionamong the subset of patients who received antibiotics (diamonds) and thosewho did not (circles) within a given ward. For correspondence of wardidentifiers, see Table 2. DOTs indicates days of therapy.




Hospital antimicrobial stewardship programs consistentlyachieve substantial reductions (22%-36%) in overall antibi-otic use,27 and such interventions reduce C difficile infectionincidence by 50%.26 Because almost all antibiotics are associ-ated with increased C difficile infection risk,4 antimicrobialstewardship initiatives aiming to reduce infection incidenceshould aim to reduce overall antimicrobial exposure in addi-tion to reducing use of specific high-risk agents. Further-more, our results suggest that hand hygiene with soap and wa-

ter should be considered before and after caring for patientsusing antibiotics, especially in ICU wards with high levels ofantibiotic use.

Like any observational study, ours was subject to a num-ber of limitations, including confounding by unmeasured pa-tient characteristics, such as comorbidities, and outcome as-certainment bias related to potential systematic differences inphysicians’ vigilance for detecting milder cases of infection.This was a single-hospital study, and the overall number ofwards at our study hospital was small (n = 16). Furthermore,our study was subject to limitations because of incomplete fol-low-up information on patients after hospital discharge. Weconsidered patients who were discharged as censored, but pa-tient censoring may not have been independent of the studyoutcome.28

ConclusionsOur 46-month study of inpatient C difficile infection risk across16 wards of a large tertiary care hospital found a strong asso-ciation between ward antibiotic prescribing and C difficile in-fection incidence that affected patients with and without re-cent antibiotic exposure. Future studies of C difficile infectionetiology should seek to quantify patient, ward, and airbornecontamination with C difficile spores to more clearly describethe mechanisms that link ward-level antimicrobial use and in-fection incidence. These findings strongly support the fur-ther funding and development of hospital antibiotic steward-ship programs.

ARTICLE INFORMATION

Accepted for Publication: November 8, 2014.

Published Online: February 23, 2015.doi:10.1001/jamainternmed.2014.8273.

Author Contributions: Drs Brown and Danemanhad full access to all the data in the study and takeresponsibility for the integrity of the data and theaccuracy of the data analysis.Study concept and design: Brown, Valenta, Fisman,Daneman.Acquisition, analysis, or interpretation of the data:Brown, Simor, Daneman.Drafting of the manuscript: Brown.Critical revision of the manuscript for importantintellectual content: All authors.Statistical analysis: Brown.Obtained funding: Brown, Fisman.Administrative, technical, or material support:Brown, Valenta, Simor, Daneman.Study supervision: Fisman, Daneman.

Conflict of Interest Disclosures: Dr Simor reportedreceiving honoraria for speaking on behalf ofOptimer Pharmaceuticals Inc Canada (now CubistPharmaceuticals). No other disclosures werereported.

Funding/Support: This study was supported by adoctoral research award from the CanadianInstitutes for Health Research (Dr Brown).

Role of the Funder/Sponsor: The funding sourcehad no role in the design and conduct of the study;collection, management, analysis, and

interpretation of the data; preparation, review, orapproval of the manuscript; and the decision tosubmit the manuscript for publication.

Additional Contributions: Marion Elligsen,BScPhm, developed the infection prevention andcontrol database and aided in antimicrobial coding.Dariusz Pajak, BASc, CPHI(C), provided importantinformation on hospital ward configuration.

REFERENCES

1. Owens RC Jr, Donskey CJ, Gaynes RP, Loo VG,Muto CA. Antimicrobial-associated risk factors forClostridium difficile infection. Clin Infect Dis. 2008;46(suppl 1):S19-S31.

2. Stevens V, Dumyati G, Fine LS, Fisher SG, vanWijngaarden E. Cumulative antibiotic exposuresover time and the risk of Clostridium difficileinfection. Clin Infect Dis. 2011;53(1):42-48.

3. Slimings C, Riley TV. Antibiotics andhospital-acquired Clostridium difficile infection:update of systematic review and meta-analysis.J Antimicrob Chemother. 2014;69(4):881-891.

4. Brown KA, Khanafer N, Daneman N, Fisman DN.Meta-analysis of antibiotics and the risk ofcommunity-associated Clostridium difficileinfection. Antimicrob Agents Chemother. 2013;57(5):2326-2332.

5. Baxter R, Ray GT, Fireman BH. Case-controlstudy of antibiotic use and subsequent Clostridium

difficile–associated diarrhea in hospitalizedpatients. Infect Control Hosp Epidemiol. 2008;29(1):44-50.

6. Sethi AK, Al-Nassir WN, Nerandzic MM,Bobulsky GS, Donskey CJ. Persistence of skincontamination and environmental shedding ofClostridium difficile during and after treatment of C.difficile infection. Infect Control Hosp Epidemiol.2010;31(1):21-27.

7. Dubberke ER, Reske KA, Olsen MA, et al.Evaluation of Clostridium difficile–associateddisease pressure as a risk factor for Cdifficile–associated disease. Arch Intern Med. 2007;167(10):1092-1097.

8. Eyre DW, Cule ML, Wilson DJ, et al. Diversesources of C. difficile infection identified onwhole-genome sequencing. N Engl J Med. 2013;369(13):1195-1205.

9. Riggs MM, Sethi AK, Zabarsky TF, Eckstein EC,Jump RLP, Donskey CJ. Asymptomatic carriers are apotential source for transmission of epidemic andnonepidemic Clostridium difficile strains amonglong-term care facility residents. Clin Infect Dis.2007;45(8):992-998.

10. Guerrero DM, Becker JC, Eckstein EC, et al.Asymptomatic carriage of toxigenic Clostridiumdifficile by hospitalized patients. J Hosp Infect. 2013;85(2):155-158.

11. Pakyz AL, Jawahar R, Wang Q, Harpe SE.Medication risk factors associated withhealthcare-associated Clostridium difficile infection:

Table 3. Patient- and Ward-Level Risk Factors for Clostridium difficileInfection From a Multilevel Model

Risk Factor Relative Risk (95% CI)Patient-level risk factors

Age (per 10-y increment) 1.12 (1.03-1.21)

Male sex 0.98 (0.76-1.25)

Diabetes mellitus 1.00 (0.76-1.31)

Admission unit

Oncology 0.97 (0.64-1.46)

Surgery 1.00 (0.75-1.32)

Medication history in previous 10 d

Antibiotics 2.02 (1.50-2.72)

Antacids 0.84 (0.62-1.13)

Chemotherapeutic agents 1.48 (1.04-2.11)

Feeding tube in situ in previous 10 d 1.14 (0.82-1.58)

Ward-level risk factors

Antibiotic exposure rate (per 10% increase) 1.34 (1.16-1.57)




a multilevel model case-control study among 64 USacademic medical centres. J Antimicrob Chemother.2014;69(4):1127-1131.

12. Diez Roux AV, Aiello AE. Multilevel analysis ofinfectious diseases. J Infect Dis. 2005;191(suppl 1):S25-S33.

13. Provincial Infectious Diseases AdvisoryCommittee. Best Practices Document for theManagement of Clostridium difficile in All HealthCare Settings Protecting Patients and Staff. Toronto,ON: Ministry of Health and Long-Term Care; 2007.

14. McDonald LC, Coignard B, Dubberke E, Song X,Horan T, Kutty PK; Ad Hoc Clostridium difficileSurveillance Working Group. Recommendations forsurveillance of Clostridium difficile–associateddisease. Infect Control Hosp Epidemiol. 2007;28(2):140-145.

15. Zar FA, Bakkanagari SR, Moorthi KMLST, DavisMB. A comparison of vancomycin andmetronidazole for the treatment of Clostridiumdifficile–associated diarrhea, stratified by diseaseseverity. Clin Infect Dis. 2007;45(3):302-307.

16. WHO Collaborating Centre for Drug StatisticsMethodology, Norwegian Institute of Public Health.Guidelines for ATC classification and DDDassignment 2014. Oslo, Norway: WHO CollaboratingCentre for Drug Statistics Methodology, NorwegianInstitute of Public Health; 2013. http://www.whocc.no/filearchive/publications/2014guidelines.pdf.Accessed June 5, 2014.

17. Sax H, Allegranzi B, Chraïti M-N, Boyce J, LarsonE, Pittet D. The World Health Organization handhygiene observation method. Am J Infect Control.2009;37(10):827-834.

18. IntHout J, Ioannidis JP, Borm GF. TheHartung-Knapp-Sidik-Jonkman method for randomeffects meta-analysis is straightforward andconsiderably outperforms the standardDerSimonian-Laird method. BMC Med Res Methodol.2014;14(1):25.

19. Peduzzi P, Concato J, Feinstein AR, Holford TR.Importance of events per independent variable inproportional hazards regression analysis, II:accuracy and precision of regression estimates.J Clin Epidemiol. 1995;48(12):1503-1510.

20. Bignardi GE. Risk factors for Clostridium difficileinfection. J Hosp Infect. 1998;40(1):1-15.

21. Brown KA, Fisman DN, Daneman N. HospitalClostridium difficile infection testing rates: is “don’task, don’t tell” at play? Infect Control Hosp Epidemiol.2014;35(7):911-912.

22. Crabtree SJ, Robertson JL, Chung KK, et al.Clostridium difficile infections in patients withsevere burns. Burns. 2011;37(1):42-48.

23. Giesecke J. Modern Infectious DiseaseEpidemiology. 2nd ed. New York, NY: OxfordUniversity Press; 2002.

24. Vynnycky E. An Introduction to InfectiousDisease Modelling. New York, NY: Oxford UniversityPress; 2010.

25. Halloran ME, Longini IM Jr, Struchiner CJ.Design and interpretation of vaccine field studies.Epidemiol Rev. 1999;21(1):73-88.

26. Feazel LM, Malhotra A, Perencevich EN, KaboliP, Diekema DJ, Schweizer ML. Effect of antibioticstewardship programmes on Clostridium difficileincidence: a systematic review and meta-analysis.J Antimicrob Chemother. 2014;69(7):1748-1754.

27. Dellit TH, Owens RC, McGowan JE Jr, et al;Infectious Diseases Society of America; Society forHealthcare Epidemiology of America. InfectiousDiseases Society of America and the Society forHealthcare Epidemiology of America guidelines fordeveloping an institutional program to enhanceantimicrobial stewardship. Clin Infect Dis. 2007;44(2):159-177.

28. Schumacher M, Allignol A, Beyersmann J,Binder N, Wolkewitz M. Hospital-acquiredinfections: appropriate statistical treatment isurgently needed! Int J Epidemiol. 2013;42(5):1502-1508.




Date post:	28-Nov-2023
Category:	Documents
Upload:	mcgill
View:	0 times
Download:	0 times

Hospital Ward Antibiotic Prescribing and the Risks of Clostridium difficile Infection

Documents