Risk factors and impact of Clostridium difficile recurrence on
haematology patients
Gianni B. Scappaticci
1
, Anthony J. Perissinotti
1
, Jerod L. Nagel
1
, Dale L. Bixby
2
and Bernard L. Marini
1
*
1
Department of Pharmacy Services and Clinical Sciences, University of Michigan Health System and College of Pharmacy, Ann Arbor,
MI, USA;
2
Department of Internal Medicine, Division of Haematology and Oncology, University of Michigan, Ann Arbor, MI, USA
*Corresponding author. Tel: !1-734-647-0175; Fax: !1-734-936-7027; E-mail: bernmari@med.umich.edu
Received 12 September 2016; returned 15 November 2016; revised 15 December 2016; accepted 3 January 2017
Objectives: The incidence of Clostridium difficile infection (CDI) in adults with malignancy is 7%–14% compared
with 1%–2% in the general hospitalized population. Despite the increased incidence of CDI in this population, a
major concern is the propensity of CDI to recur, leading to delays in therapy impacting outcomes. We conducted
a retrospective case–control study to identify risk factors for recurrent CDI (rCDI) and to determine the impact of
rCDI on adult patients with a haematological malignancy.
Methods: Adult haematology patients with CDI from June 2010 to December 2014 were divided into two groups:
rCDI and non-rCDI. Multivariable models using logistic regression were constructed to identify risk factors for
rCDI.
Results: A total of 100 patients in our study yielded a 41% recurrence rate. CDI impacted chemotherapy signifi-
cantly more in the rCDI group (53.7% versus 11.9%, P
,0.001), primarily due to interruptions in established treat-
ment plans (46.3% versus 10.3%, P
,0.001). Risk factors for rCDI identified at index included salvage lymphoma
chemotherapy (OR 9.64, 95% CI 1.02–91.15, P " 0.048) and severe CDI (OR 4.82, 95% CI 1.31–17.66, P " 0.018).
Longitudinal risk factors included exposure to fluoroquinolones (OR 3.96, 95% CI 1.04–15.15, P " 0.044), ceftriaxone
(OR 18.93, 95% CI 1.27–281.95, P " 0.033) and piperacillin/tazobactam (OR 10.4, 95% CI 1.81–59.64, P " 0.009).
Conclusions: Haematology patients exhibit a higher rate of rCDI than general hospitalized patients. Utilization of
this multivariable model to guide index CDI therapy at index may help to decrease the rCDI and prevent delays
or interruptions in chemotherapy.
Introduction
Over the past decade, Clostridium difficile,ananaerobic,Gram-
positive, spore-forming bacterium, has become the leading cause
of nosocomial infections in the USA.
1–3
C. difficile infection (CDI)
can be self-limiting or may progress to serious and life-threatening
fulminant colitis, ileus or toxic megacolon.
4
In addition to this, CDI
increases the duration of hospitalization and is responsible for an
estimated $4.8 billion in excess healthcare costs in the USA.
5
The
incidence of CDI is 1%–2% in the general hospitalized population
and antibiotic exposure remains the leading risk factor for CDI.
1,2
Over the last two decades, outbreaks of resistant C. difficile strains
causing recurrent and remarkably severe CDI have increased in
prevalence.
1,2
Patients with cancer are at an increased risk of CDI because of
their underlying malignancy, depressed immune response and ex-
posure to chemotherapy.
2,6–8
The incidence of CDI in haematology
patients is 7%–14%, indicating that additional risk factors present
in this population may contribute to the increased incidence of
CDI.
9
Although the increased risk in both frequency and severity of
CDI in haematology patients is significant, a major concern is the
propensity of CDI to recur.
3,10
Recent data indicate that the inci-
dence of recurrent CDI (rCDI) is 15%–35% after discontinuation of
therapy against CDI.
11
Management of rCDI is challenging and
associated with higher morbidity, mortality and healthcare ex-
penditures.
11
Despite the high risk of CDI acquisition and incidence
of rCDI in haematology patients, there is a paucity of data analy-
sing risk factors associated with rCDI for this specific patient
population.
2,11,12
Several studies have identified risk factors for rCDI in the gen-
eral hospitalized population.
10
These traditional risk factors
included advanced age (65 years), duration of hospitalization
(14 days), chronic renal insufficiency, elevated white blood cell
count, low serum albumin, use of acid-suppression therapy and
continued use of broad-spectrum antimicrobials.
13–17
However,
these studies were limited because they lacked adequate data to
V
C
The Author 2017. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved.
For Permissions, please email: journals.permissions@oup.com.
1488
J Antimicrob Chemother 2017; 72: 1488–1495
doi:10.1093/jac/dkx005 Advance Access publication 10 February 2017
Downloaded from https://academic.oup.com/jac/article/72/5/1488/2982029 by guest on 21 August 2022
assess the impact of index CDI (iCDI) treatment and specific non-
CDI antimicrobials.
10
Furthermore, these studies focused on fac-
tors immediately preceding rCDI onset, ignoring the possibility that
factors near the onset of iCDI could have contributed to
recurrence.
10
In adult haematology patients, a precise understanding of who
is at a higher risk of developing rCDI is an important clinical ques-
tion that currently lacks adequate data to assist clinicians in fully
understanding rCDI risk factors. Data linking rCDI to modifiable risk
factors would allow clinicians to avoid these exposures if possible
and recognition of unmodifiable risk factors may lead to preventa-
tive strategies or targeted initial therapy.
Methods
Hospital
This retrospective study was conducted at the University of Michigan
Health System (UMHS) in Ann Arbor, MI, a 1059 bed, tertiary-care, univer-
sity-affiliated hospital with 47000 admissions annually.
Ethics
Ethics approval was obtained from the Institutional Review Board (IRB) at the
University of Michigan Health System (approval number HUM00106249).
Written informed consent was waived by the IRB.
Study design
Adult (age 18 years) patients with an active haematological malignancy
and CDI from June 2010 to December 2014 were identified from the elec-
tronic medical record (EMR) at UMHS and screened for inclusion in this retro-
spective case–control study. Patients who received an autologous or
allogeneic HSCT were excluded from the study. Additionally, patients with
inadequate follow-up (
,6 months) from the iCDI at UMHS were also
excluded.
Study data were collected using the EMR and managed using the
Research Electronic Data Capture (REDCap
TM
) tools hosted at UMHS.
18
Data
collection points at iCDI included patient demographics, CDI severity, CDI
treatment, CDI treatment duration, traditional CDI risk factors and potential
haematology-specific risk factors. Traditional risk factors for CDI collected
included age
.65 years, antibiotic exposure, acid-suppressive therapy and
history of CDI. Additional potential haematological risk factors that were
evaluated included type of malignancy, neutropenia and duration of neu-
tropenia at iCDI and chemotherapy exposure (regimen, duration and num-
ber of days prior to iCDI). All potential risk factor data at iCDI were collected
starting 14 days prior to iCDI diagnosis.
The study population was divided into two groups for analysis: recurrent
CDI (rCDI) and non-recurrent CDI (non-rCDI); rCDI was defined as a repeat
positive CDI occurring within 6 months after the end of treatment and ces-
sation of symptoms for the iCDI. The 6 month time period for rCDI was
chosen based on prior studies of risk factors for rCDI and patterns of recur-
rence risk over time.
2,19–21
For patients with multiple CDIs, the earliest posi-
tive date was considered the iCDI and all subsequent positive CDIs
30 days after resolution of symptoms were considered recurrences.
Longitudinal data were collected for 6 months after iCDI treatment was
complete and symptoms resolved. Data collection points included trad-
itional risk factors for CDI and haematological risk factors as listed above.
Additionally, progress notes were reviewed for explanations justifying
modifications to treatment plans, which included delays, dose reductions
or discontinuation of chemotherapy.
Definitions
Diagnosisof CDI required symptoms consistent with CDI and a stool sample
test positive for C. difficile in the UMHS Clinical Microbiology Laboratory.
Stool testing was performed using the C. DIFF QUIK CHEK COMPLETE
V
R
test
(Alere/TECHLAB
V
R
, Inc.; Blacksburg, VA, USA) for C. difficile glutamate de-
hydrogenase (GDH) antigen and toxins A or B by enzyme immunoassay
(EIA). All GDH!/toxin# stool tests were subjected to analysis for the tcdB
gene by real-time PCR assay (BD Gene-Ohm
TM
Cdiff Assay; Franklin Lakes,
NJ, USA).
Acid-suppressive therapy was defined as use of a proton pump inhibitor
(PPI) or histamine type-2 receptor antagonist (H2RA) for 2 days. History of
CDI was a previous CDI 1 year from iCDI. Neutropenia was defined as an
absolute neutrophil count (ANC)
,500 cells/mm
3
. CDI was classified as se-
vere in patients with a white blood cell (WBC) count 15000 cells/mm
3
or
serum creatinine (SCr) 1.5% baseline SCr or evidence of ileus.
Haematological malignancies included acute and chronic leukaemias, mul-
tiple myeloma (MM), non-Hodgkin’s lymphoma (NHL), Hodgkin’s lymph-
oma (HL) and myelodysplastic or myeloproliferative neoplasms.
Statistical analyses
Statistical analyses were performed using SPSS software, version 23.0
(SPSS, Inc., Chicago, IL, USA). Demographic data were analysed by descrip-
tive statistics, continuous data were analysed by two-tailed Student’s t-test
and dichotomous data were analysed by Pearson’s v
2
test or Fisher’s exact
test. Mann–Whitney U-tests were performed for non-normally distributed
factors. Exploratory unconditional logistic regression analysis was per-
formed to evaluate variables associated with rCDI. Variables with a P value
0.2 on univariate analysis were initially considered for inclusion in the
multivariable model. For co-linear variables with a P value 0.2 on univari-
ate analysis, only one variable was included in the multivariable model.
Two separate regression models were created: one to assess possible iCDI
risk factors associated with rCDI (forward conditional regression) and a se-
cond model to assess longitudinal risk factors associated with rCDI (binary
regression). A receiver operating characteristic (ROC) curve was generated
for both models to assess the sensitivity and specificity of the models. A P
value of 0.05 was considered statistically significant and all P values were
based on two-tailed tests.
Results
A total of 649 patients with CDI and malignancy were identified in
the EMR from June 2010 to December 2014. Of these patients,
100 (15.4%) met the inclusion criteria (Figure 1). During the
6 month follow-up, 41 patients (41.0%) developed rCDI and 59
(59.0%) did not develop rCDI. In the rCDI cohort, 30 patients
(73.2%) had only one recurrence, whereas 11 patients (26.8%)
experienced two or more episodes of rCDI (Figure S1a, available as
Supplementary data at JAC Online). The first recurrence of CDI
occurred 30 days after the iCDI in 20 of the 41 patients with rCDI
(48.9%). Among the remaining 21 patients, recurrences occurred
after 31–60 days in 13 (31.7%) and 8 (19.5%) recurred
.60 days
but within 6 months after iCDI (Figure S1b).
Age, gender and race were not different between the two
groups (Table 1). Metronidazole was used to treat 61.0% of the pa-
tients in the rCDI group compared with 45.8% of the patients in
the non-rCDI group. The remaining patients were treated with oral
vancomycin (19.5% rCDI versus 28.8% non-rCDI) or a combination
of both agents (19.5% rCDI versus 25.4% non-rCDI) (Table S1). The
median duration of treatment was longer in the non-rCDI group
for oral metronidazole and vancomycin, although this was not
statistically significant (Table S1). Overall, antibiotic exposure and
Recurrent CDI in haematology patients JAC
1489
Downloaded from https://academic.oup.com/jac/article/72/5/1488/2982029 by guest on 21 August 2022
use of acid-suppression therapy were not different between the
two groups, although the non-rCDI group had more patients using
combination PPI and H2RA than the rCDI group (18.0% versus
3.2%, P " 0.044).
Patients in the rCDI group had a significantly higher incidence of
severe CDI (22.0% versus 6.8%, P " 0.027). Metronidazole was
used to treat five of the patients in the rCDI group with severe CDI
compared with two in the non-rCDI group. Oral vancomycin was
used in two patients in the rCDI group with severe CDI and none in
the non-rCDI group, whereas combination therapy was used in
two patients in both groups. Median duration of treatment with
oral metronidazole was 12 days (IQR 11–17) for the rCDI group
with severe CDI compared with 15.5 days (IQR 10–21) in the non-
rCDI group. Duration of combination therapy was numerically
higher in the rCDI group with severe CDI (median 75 days, IQR 21–
129 versus 34.5 days, IQR 12–57), but the difference was not stat-
istically significant (Table S1).
Differences in haematological data at iCDI were also compared
on univariate analysis between the two groups. There was a signifi-
cantly higher proportion of patients with AML in the non-rCDI
group compared with the rCDI group (37.3% versus 17.1%,
P " 0.028). NHL was the most common malignancy, followed by
MM, ALL and CLL. At the time of iCDI, 72% of patients were receiv-
ing active chemotherapy (68.3% rCDI versus 74.6% non-rCDI,
P " 0.491). iCDI occurred at a median of 9.5 days after chemother-
apy in the rCDI group compared with 7.5 days in the non-rCDI
group (P " 0.399). These were not different in the recurrent and
non-recurrent groups. Other haematological data, including
chemotherapy exposure and duration of neutropenia, were not
statistically significantly different between the two groups at index
on univariate analysis (Table 1).
On multivariate analysis, exposure to salvage lymphoma regi-
mens [R-ICE (rituximab ! ifosfamide ! carboplatin ! etoposide),
ICE (ifosfamide ! carboplatin ! etoposide), R-ESHAP (rituximab !
etoposide ! methylprednisolone ! high-dose cytarabine ! cis-
platin) and R-CODOX-M/IVAC (rituximab ! cyclophosphamide !
doxorubicin ! vincristine ! methotrexate/cytarabine ! ifosfamide
! etoposide)] at iCDI was the only independent haematological
predictor of recurrent disease (OR 9.64, 95% CI 1.02–91.15,
P " 0.048). Severe CDI at index was also shown to be an independ-
ent risk factor of rCDI (OR 4.82, 95% CI 1.31–17.66, P " 0.018)
in this model. Many traditional risk factors for rCDI were present on
univariate analysis (P 0.2); however, these differences were
not shown to be significant on multivariable regression
analysis (Table 1). The ROC AUC for this model was 0.712 (95%
CI 0.610–0.814).
Results of the univariate analysis during the 6 month follow-up
revealed that patients in the rCDI group were hospitalized more
(92.7% versus 62.7%, P " 0.001) and exposed to antibiotics more
often than those in the non-rCDI group (85.4% versus 54.2%,
P " 0.001). Antibiotic exposure was led by fluoroquinolones (58.4%
versus 28.8%, P " 0.004) and b-lactams/b-lactamase inhibitors
(68.3% versus 23.7%, P " 0.001). Additionally, the rCDI group was
exposed to chemotherapy more often following iCDI (82.9% ver-
sus 62.7%, P " 0.028) than the non-rCDI group, including exposure
to salvage lymphoma regimens (19.5% versus 3.4%, P " 0.014)
(Table 2).
Multivariable analysis of longitudinal data was performed and
an ROC curve with a high AUC was created. On multivariable ana-
lysis, exposure to fluoroquinolones (OR 3.96, 95% CI 1.04–15.15,
P " 0.044), ceftriaxone (OR 18.93, 95% CI 1.27–281.95, P " 0.033)
and piperacillin/tazobactam (OR 10.4, 95% CI 1.81–59.64,
P " 0.009) were all associated with a significantly increased risk of
rCDI (Table 2). The duration of neutropenia was also identified as a
statistically significant risk factor for rCDI on multivariable analysis
(OR 0.94, 95% CI 0.89–0.98, P " 0.010), although in this case an
increased duration of neutropenia was protective for rCDI.
Additional possible haematology-specific risk factors identified on
univariate analysis did not remain significant on multivariable ana-
lysis. The ROC AUC of the multivariable model was 0.870 (95% CI
0.802–0.938).
The impact of CDI on chemotherapy was also compared be-
tween the two groups. CDI impacted chemotherapy significantly
more in the rCDI group compared with the non-rCDI group (53.7%
versus 11.9%, P
,0.001). Among these patients, CDI primarily
caused delays in initiation of subsequent cycles of chemotherapy.
Delays in chemotherapy were statistically higher in the rCDI group
(46.3% versus 10.3%, P
,0.001). Finally, chemotherapy was dis-
continued in one patient in both groups, and two patients in the
rCDI group received a reduced dose due to the impact of CDI
(Figure 2).
Discussion
The burden of CDI has significantly increased over the past dec-
ade.
1–3
Perhaps more concerning is the propensity for recurrence
further impacting the patient and contributing to poor outcomes.
A recent study by Olsen et al.
22
found that rCDI was associated
with a significantly increased risk of death within 6 months of iCDI
(36.3% versus 26.0%, P
,0.001). In our study, the rate of rCDI in
haematology patients was remarkably high, at 41%. The higher re-
currence rate of 41% may partially be explained by the 6 month
time frame used to define recurrence in our analysis; however,
similar studies examining the incidence of rCDI as far out as
6–12 months had significantly lower CDI recurrence rates of
23.2% and 22%, respectively.
2,21
At 60 days, the recurrence rate in
our study was 33.2%, which is significantly higher than the rate
seen in the general hospitalized patients at our institution (9%)
Patients with CDI and malignancy
June 2010–December 2014
N =649
CDI in haematology patients
n =100
Excluded (n =549)
• Solid tumour (n =436)
• Inadequate f/u (n =56)
• Transplant (n =35)
• <18 years (n =22)
non-rCDI patients
n =59
rCDI patients
n =41
Figure 1. Patient enrolment and exclusion information. f/u, follow-up.
Scappaticci et al.
1490
Downloaded from https://academic.oup.com/jac/article/72/5/1488/2982029 by guest on 21 August 2022
Table 1. Baseline demographics and traditional and haematological risk factors at iCDI
Variable
a
Univariate analysis Multivariable LR at index
rCDI (N " 41) Non-rCDI (N " 59) P OR (95% CI) P
Age (years) 63 (49–70) 63 (44–73) 0.702
Male, n (%) 24 (58.5) 32 (54.2) 0.670
Caucasian, n (%) 34 (82.9) 52 (88.1) 0.460
Severe CDI, n (%) 9 (22.0) 4 (6.8) 0.027 4.82 (1.31–17.66) 0.018
Haematological disease
b
, n (%)
AML 7 (17.1) 22 (37.3) 0.028
ALL 2 (4.9) 7 (11.9) 0.302
CLL 3 (7.3) 3 (5.1) 0.687
NHL 18 (43.9) 18 (30.5) 0.170
MM 6 (14.6) 6 (10.2) 0.543
Traditional risk factors
age
.65 years, n (%) 19 (46.3) 22 (37.3) 0.365
history of CDI, n (%) 2 (4.9) 1 (1.7) 0.359
hospitalized (days) 8.5 (5–17) 13 (5–28) 0.100
PPI, n (%) 27 (65.9) 36 (61.0) 0.622
H2RA, n (%) 3 (7.3) 4 (6.8) 1.000
PPI ! H2RA, n (%) 1 (3.2) 9 (18.0) 0.044
iCDI treatment
oral vancomycin, n (%) 8 (19.5) 17 (28.8) 0.291
oral metronidazole, n (%) 25 (61.0) 27 (45.8) 0.134
duration of treatment (days) 14 (13–22) 20 (14–29) 0.160
Antibiotic exposure
c
, n (%)
aminoglycosides 2 (4.9) 6 (10.2) 0.466
carbapenems 2 (4.9) 2 (3.4) 1.000
cephalosporins 11 (26.8) 14 (23.7) 0.725
fluoroquinolones 5 (12.2) 9 (15.3) 0.665
b-lactams/b-lactamase inhibitors 8 (19.5) 22 (37.3) 0.056
piperacillin/tazobactam 7 (17.1) 19 (32.2) 0.090
other (SXT) 0 (0) 5 (8.5) 0.056
Haematological risk factors
d
neutropenia (days) 9 (5.5–25) 25 (11–37) 0.142
chemotherapy at iCDI, n (%) 28 (68.3) 44 (74.6) 0.491
chemotherapy (days before) 9.5 (5–14) 7.5 (4–13) 0.399
3 ! 7, n (%) 2 (4.9) 7 (11.9) 0.302
FLAG, n (%) 1 (2.4) 2 (3.4) 1.000
clofarabine based, n (%) 0 (0) 6 (10.2) 0.079
HiDAC, n (%) 1 (2.4) 4 (6.8) 0.646
decitabine, n (%) 3 (7.3) 0 (0) 0.066
Hyper-CVAD A, n (%) 1 (2.3) 3 (5.1) 0.642
Hyper-CVAD B, n (%) 3 (7.3) 3 (5.1) 0.687
CHOP, n (%) 4 (9.8) 1 (1.7) 0.156
salvage lymphoma
e
, n (%) 4 (9.8) 1 (1.7) 0.156 9.64 (1.02–91.15) 0.048
oral chemotherapy, n (%) 4 (9.8) 6 (10.2) 1.000
other chemotherapy, n (%) 3 (7.3) 6 (10.2) 0.733
LR, logistic regression; SXT, trimethoprim/sulfamethoxazole.
a
Results that are not a number and percentage are the median (IQR).
b
Included diseases in four or more patients.
c
Antibiotics reported individually if P ,0.200; others reported by class.
d
Included chemotherapy if three or more patients exposed.
e
Included R-ICE, ICE, R-ESHAP and R-CODOX.
Recurrent CDI in haematology patients JAC
1491
Downloaded from https://academic.oup.com/jac/article/72/5/1488/2982029 by guest on 21 August 2022
Table 2. Longitudinal risk factors
Variable
a
Univariate analysis Longitudinal multivariable LR
rCDI (N " 41) Non-rCDI (N " 59) P OR (95% CI) P
Hospitalized (days) 22 (17–33) 13 (8–30) 0.071 1.05 (0.99–1.09) 0.059
PPI, n (%) 26 (63.4) 39 (66.1) 0.782
H2RA, n (%) 3 (7.3) 4 (6.8) 1.000
PPI ! H2RA, n (%) 6 (14.6) 5 (8.5) 0.351
Antibiotic exposure
b
, n (%)
aminoglycosides 6 (14.6) 5 (8.5) 0.351
carbapenems 4 (9.8) 3 (5.1) 0.440
cephalosporins 15 (36.6) 15 (25.4) 0.231
ceftriaxone 5 (12.2) 2 (3.4) 0.119 18.93 (1.27–281.95) 0.033
fluoroquinolones 24 (58.5) 17 (28.8) 0.004 3.96 (1.04–15.15) 0.044
b-lactams/b-lactamase inhibitors 28 (68.3) 14 (23.7)
,0.001
piperacillin/tazobactam 26 (63.4) 11 (18.6)
,0.001 10.40 (1.81–59.64) 0.009
other (daptomycin) 6 (14.6) 2 (3.4) 0.061 6.14 (0.58–65.25) 0.133
other (vancomycin IV) 28 (68.3) 18 (30.5)
,0.001 0.90 (0.17–4.84) 0.903
number of antibiotic exposures 2 (1–2.25) 1 (1–2) 0.013 0.61 (0.29–1.29) 0.199
Haematological risk factors
c
neutropenia (days) 11 (5–24) 19 (12–57) 0.053 0.94 (0.89–0.98) 0.010
chemotherapy exposure, n (%) 34 (82.9) 37 (62.7) 0.028 1.42 (0.28–7.15) 0.671
chemotherapy (cycles), n (%) 2 (1–4) 2 (1–4) 0.938
clofarabine based, n (%) 2 (4.9) 5 (8.5) 0.697
HiDAC, n (%) 4 (9.8) 8 (13.6) 0.757
decitabine 5 (12.2) 2 (3.4) 0.119 2.15 (0.29–16.25) 0.457
Hyper-CVAD A, n (%) 3 (7.3) 2 (3.4) 0.398
Hyper-CVAD B, n (%) 2 (4.9) 2 (3.4) 0.963
CHOP, n (%) 3 (7.3) 1 (1.7) 0.158 2.35 (0.12–45.46) 0.572
salvage lymphoma
d
, n (%) 8 (19.5) 2 (3.4) 0.014 2.57 (0.33–20.32) 0.371
oral chemotherapy, n (%) 6 (14.6) 10 (16.9) 0.756
other chemotherapy, n (%) 7 (17.1) 7 (11.9) 0.460
LR, logistic regression.
a
Results that are not a number and percentage are median (IQR).
b
Antibiotics reported individually if P ,0.200; others reported by class.
c
Included chemotherapy if three or more patients exposed.
d
Included R-ICE, ICE, R-ESHAP and R-CODOX.
Chemotherapy
impacted
(
P
< 0.001)
n
= 7
(11.9%)
n
= 22
(53.7%)
rCDI
Non-rCDI
n
= 6
(10.3%)
n
= 19
(46.3%)
Chemotherapy
delayed
(
P
< 0.001)
0 10% 20% 30%
Percentage of patients
40% 50% 60%
Figure 2. Impact of CDI on chemotherapy.
Scappaticci et al.
1492
Downloaded from https://academic.oup.com/jac/article/72/5/1488/2982029 by guest on 21 August 2022