Clinical Microbiology and Infection xxx (2017) 1e6

Contents lists available at ScienceDirect

Clinical Microbiology and Infection

journal homepage: www.clinicalmicrobiologyandinfection.com

Original article

Initial empiric antibiotic therapy for community-acquired pneumonia

in Chinese hospitals
X.M. Nie 1, y, Y.S. Li 2, y, Z.W. Yang 3, y, H. Wang 3, S.Y. Jin 3, Y. Jiao 1, M.L. Metersky 4,
Y. Huang 1, *
1)
Department of Respiratory and Critical Care, Changhai Hospital, Second Military Medical University, Shanghai, China
2)
Department of Hepatobiliary Surgery, Shanghai Public Health Clinical Center, Shanghai, China
3)
Department of Pharmacy, Changhai Hospital, The Second Military Medical University, Shanghai, China
4)
Division of Pulmonary and Critical Care Medicine, University of Connecticut Health Center, Farmington, CT, USA

a r t i c l e i n f o a b s t r a c t

Article history: Objectives: Studies on treatment of community-acquired pneumonia (CAP) in China are scarce. We
Received 30 March 2017 performed a study to investigate empiric antibiotic practices for patients hospitalized with CAP in China
Received in revised form and the risk factors for treatment failure.
18 September 2017
Methods: Data were collected from a national Chinese hospitalization database. Adult patients who were
Accepted 21 September 2017
Available online xxx
diagnosed with CAP between 1 October 2014 and 30 September 2015 were identified. We studied initial
empiric antibiotic regimens, microbiologic sampling, treatment failure, in-hospital mortality and length
Editor: Dr. M. Leeflang of hospital stay.
Results: We included 18 043 adult patients from 185 hospitals who met all the study inclusion criteria.
Keywords: The most common initial antibiotic regimen for CAP was monotherapy with a fluoroquinolone (14.8%,
Antibiotic therapy 2671/18 043). The most common initial antibiotic (used alone or in combination with other antibiotics)
Atypical pathogen covering was levofloxacin (15.7%, 4597/29 278 (this denominator represents the total number of initial antibi-
Community-acquired pneumonia otics)). The microbiologic sampling rate was 26.9% (4851/18 043). A total of 4050 (22.4%) of 18 043
Length of stay
patients experienced treatment failure. Multivariate logistic regression demonstrated that older age,
Treatment failure
male sex, coexisting lung cancer and use of regimens not covering atypical pathogens were risk factors
for treatment failure. In-hospital mortality was 2.1% (380/18 043). The median hospital length of stay was
11 days (interquartile range, 8e15 days).
Conclusions: Patients receiving Chinese guideline-adherent regimens had better outcomes, and atypical
pathogen active regimens were associated with a lower treatment failure rate and shorter length of
hospital stay. X.M. Nie, Clin Microbiol Infect 2017;▪:1
© 2017 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All
rights reserved.

Introduction
Community-acquired pneumonia (CAP) is a common infectious
disease worldwide. Many countries have devoted efforts to
improve the care of CAP patients, and some organizations and
countries have developed CAP guidelines for adults [1,2]. China
* Corresponding author. Y. Huang, Department of Respiratory and Critical Care,
Changhai Hospital, the Second Military Medical University, 168 Changhai Road,
Shanghai, 200433, China.
E-mail address: huangliur@163.com (Y. Huang).
y patients in China and identify areas which require greater
X.M. Nie, Y.S. Li and Z.W. Yang contributed equally to this article.
developed its adult CAP guidelines in 2006 [3] and updated it in
2016 [4]. However, studies on treatment of CAP in China are scarce.
Although most CAP patients are treated successfully in outpa-
tient settings, about 25% require hospitalization in North America
[5]. Even in hospitalized patients, initial antibiotics are almost al-
ways chosen empirically, as the causative pathogen is not known.
The 2016 Chinese CAP guidelines [4] recommend several regimens
as initial empiric antibiotics, depending on patient characteristics.
To our knowledge, there have been no studies on initial empiric
antibiotics for hospitalized CAP patients in China.
The objective of our study was to examine antibiotic treat-
ment and microbiologic testing practices for hospitalized CAP

https://doi.org/10.1016/j.cmi.2017.09.014
1198-743X/© 2017 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All rights reserved.

Please cite this article in press as: Nie XM, et al., Initial empiric antibiotic therapy for community-acquired pneumonia in Chinese hospitals,
Clinical Microbiology and Infection (2017), https://doi.org/10.1016/j.cmi.2017.09.014
2 X.M. Nie et al. / Clinical Microbiology and Infection xxx (2017) 1e6

attention. We focused on several processes and outcomes of care,

including the most common initial empiric antibiotic regimens
and drugs, rate of microbiologic sampling, risk factors, frequency
of treatment failure, in-hospital mortality and length of hospital
stay (LOS).

Methods

Data for this study were obtained retrospectively from a na-

tional Chinese hospitalization database. This database includes all
the hospitals belonging to the Chinese People's Liberation Army,
and all of these hospitals are open to the public. A detailed
description of this database is available online in Supplementary
Appendix 1. The database contains deidentified comprehensive
clinical records from hospitalized patients. Information available
for each patient includes hospital characteristics, patient de-
mographics, admission and discharge diagnoses (in International
Classification of Diseases, Tenth Revision (ICD-10), format), all
parenteral antibiotics administered, microbiologic testing and LOS.
We included patients 18 years old who were both admitted
and discharged with a principal diagnosis of pneumonia (ICD-10
diagnosis codes J13eJ16, J18) between 1 October 2014 and 30
September 2015. The inclusion criteria were receipt of parenteral
antibiotics within 24 hours after admission; and receipt of paren- Fig. 1. Enrollment of community-acquired pneumonia patients.
teral antibiotics for at least 72 hours. We excluded patients with
probable fungal pneumonia, based on usage of only antifungal
agents; and patients who were diagnosed as having lung inflam-
patients are listed in Table 1. Geographic regions of China are shown
matory pseudotumor (ICD-10 diagnosis code J18.802) and post-
in Fig. 2. Among the study patients, approximately two thirds were
obstructive pneumonia (ICD-10 diagnosis code J18.805).
men. The mean (±SD) age was 59 ± 23 (range, 18e106) years. Most
We captured patient characteristics, including age, gender and
of the patients were non-ICU patients.
presence of selected comorbidities (as secondary discharge di-
Among initial empiric antibiotic regimens, monotherapy was
agnoses); we also listed the teaching status and geographic region
used more often (50.5%, 9117/18 043) than combination therapy
of the hospitals. We defined initial empiric antibiotic therapy as all
(49.5%, 8926/18 043). The three most common initial antibiotic
parenteral antibiotics received within 24 hours after admission. We
regimens and drugs are shown in Tables 2 and 3, respectively.
defined intensive care unit (ICU) patients as all patients admitted to
Geographic regionespecific antibiotic usage patterns are shown in
the ICU within 24 hours of admission.
Supplementary Appendix 2. Non-ICU patients received more
On the basis of previous literature [6] and the Chinese guidelines
guideline-adherent regimens (53.3%, 8414/15 795) than ICU
[4], we defined treatment failure as any of the following occurring
more than 72 hours after the first antibiotic dose: change of initial
empiric antibiotic therapy, transfer of patient to the ICU or death.
Table 1
We defined time until first antibiotic dose (TFAD) as the time from
Demographic and clinical characteristics of patients
admission to the time of the first dose of antibiotics. The chi-square
and Wilcoxon rank sum tests were used to analyse categorical and Characteristic n (%)
continuous data, respectively. Logistic regression was used to Gender
analyse factors associated with failure of the initial antibiotic Male 11 124 (61.7)
regimen. The multivariable logistic regression model contained Female 6919 (38.3)
Age group
several variables, including age, gender, hospital type, ICU admis- 18e44 years 5110 (28.3)
sion within 24 hours, comorbidities, TFAD, regimen covering 45e64 years 4336 (24.0)
atypical pathogen and adherence to Chinese guidelines. All vari- 65e79 years 4186 (23.2)
ables were included in the logistic regression analysis without se- 80 years 4411 (24.4)
Geographic region of hospital
lection. In addition, we also considered the interaction of ICU
Northeast 3329 (18.5)
admission within 24 hours and adherence to Chinese guidelines in North China region 4627 (25.6)
the logistic regression model. Differences were considered statis- East China region 4425 (24.5)
tically significant at a value of p < 0.05. All analyses were conducted Central and South China region 2385 (13.2)
by PC-SAS 9.3 software (SAS Institute, Cary, NC, USA). The Shanghai Northwest region 2131 (11.8)
Southwest region 1146 (6.4)
Changhai Hospital ethics committee waived approval for this study. ICU or non-ICU
ICU 2248 (12.5)
Results Non-ICU 15 795 (87.5)
Teaching or nonteaching
Teaching 1157 (6.4%)
Adult patients who were admitted and discharged with a
Nonteaching 16 886 (93.6%)
principal diagnosis of pneumonia from 192 hospitals between 1 Comorbidity
October 2014 and 30 September 2015 were identified. Of these Chronic obstructive pulmonary disease 1833 (10.2%)
patients, 18 043 from 185 hospitals met all the study inclusion Diabetes mellitus 2406 (13.3%)
criteria (Fig. 1). There were 177 nonteaching hospitals and eight Heart failure 1871 (10.4%)
Lung cancer 358 (2.0%)
teaching hospitals. Demographic and clinical characteristics of the

Please cite this article in press as: Nie XM, et al., Initial empiric antibiotic therapy for community-acquired pneumonia in Chinese hospitals,
Clinical Microbiology and Infection (2017), https://doi.org/10.1016/j.cmi.2017.09.014
 X.M. Nie et al. / Clinical Microbiology and Infection xxx (2017) 1e6 3

Fig. 2. Relevant geographic regions of China.

Table 2
Three most common initial empiric antibiotic regimens for community-acquired pneumonia in China

Description Regimen 1 Regimen 2 Regimen 3

Total Fluoroquinolone 14.8% (2671/18 043) Second-generation cephalosporin Cephalosporin/b-lactamase inhibitor 6.7% (1217/18 043)
10.6% (1909/18 043)
ICU Carbapenem 10.2% (229/2248) Cephalosporin/b-lactamase inhibitor Fluoroquinolone 7.0% (157/2248)
9.7% (218/2248)
Non-ICU Fluoroquinolone 15.9% (2514/15 795) Second-generation cephalosporin Third-generation cephalosporin 6.7% (1054/15 795)
11.6% (1831/15 795)
Teaching hospital Fluoroquinolone 26.2% (303/1157) Penicillin/b-lactamase inhibitor Cephalosporin/b-lactamase 7.5% (87/1157)
8.6% (100/1157)
Nonteaching hospital Fluoroquinolone 14.0% (2368/16 886) Second-generation cephalosporin Cephalosporin/b-lactamase inhibitor 6.7% (1130/16 886)
11.2% (1884/16 886)

ICU, intensive care unit.

Table 3
Three most common initial empiric antibiotic drugs for community-acquired pneumonia in China

Description Regimen 1 Regimen 2 Regimen 3

Total Levofloxacin 15.7% (4597/29 278) Moxifloxacin 11.5% (3355/29 278) Piperacillin/tazobactam 5.1% (1485/29 278)
ICU Moxifloxacin 13.3% (528/3972) Imipenem/cilastatin 10.0% (398/3972) Levofloxacin 7.7% (306/3972)
Non-ICU Levofloxacin 17.0% (4291/25 306) Moxifloxacin 11.2% (2827/25 306) Azithromycin 5.1% (1298/25 306)
Teaching hospital Moxifloxacin 24.8% (441/1775) Cefoperazone/sulbactam 8.3% (147/1775) Piperacillin/tazobactam 6.4% (114/1775)
Nonteaching hospital Levofloxacin 16.4% (4509/27 503) Moxifloxacin 10.6% (2914/27 503) Piperacillin/tazobactam 5.0% (1371/27 503)

ICU, intensive care unit.

patients (25.8%, 580/2248). In our series, 20.3% (3660/18 043) of
patients received a regimen covering atypical pathogens.
A total of 4851 (26.9%) of 18 043 patients had at least one
microbiologic sample collected (Table 4). The sample collection rate
was higher in teaching hospitals (49.5%, 573/1157) than in
nonteaching hospitals (25.3%, 4278/16 886) (p < 0.0001). ICU pa-
tients also had a higher collection rate (37.2%, 837/2248) than non-
ICU patients (25.4%, 4014/15 795) (p < 0.0001).
A total of 4050 (22.4%) of 18 043 patients experienced treatment
failure. Univariate analysis of treatment failure is shown in Table 5.
Logistic regression was used to analyse factors associated with
treatment failure (Table 6). After adjustment for confounders, older
age, male sex, coexisting lung cancer and use of a regimen not
covering atypical pathogens were independent predictors of
treatment failure. There was an interaction between ICU admission
within 24 hours and adherence to Chinese guidelines. We further
analysed the individual effects of these two factors. Among non-ICU
patients, compliance with the Chinese guidelines was associated
with a lower rate of treatment failure (p < 0.0001). However,
among ICU patients, compliance with the Chinese guidelines did
not predict treatment success (p ¼ 0.0529). Some literature sug-
gests that the ratio of outcome events to independent variables in
logistic regression should be 10:1 or higher. In our analysis, this
ratio was 4050:12, which means the sample size of the logistic
regression analysis was sufficient. The ability of a logistic regression
model to separate patients with different outcomes is known as

Please cite this article in press as: Nie XM, et al., Initial empiric antibiotic therapy for community-acquired pneumonia in Chinese hospitals,
Clinical Microbiology and Infection (2017), https://doi.org/10.1016/j.cmi.2017.09.014
4 X.M. Nie et al. / Clinical Microbiology and Infection xxx (2017) 1e6

Table 4 Table 6
Microbiologic sampling Multivariable logistic regression for treatment failure

Sample type Collection rate Factor p OR (95% CI)

Sputum culture 11.7% (2119/18 043) Age (vs. 80 years)

Blood culture 9.4% (1701/18 043) 18e44 years <0.0001 0.46 (0.41, 0.51)
Mycoplasma antibody 5.9% (1061/18 043) 45e64 years <0.0001 0.58 (0.52, 0.65)
Sputum smear 3.7% (671/18 043) 65e79 years 0.0008 0.75 (0.68, 0.83)
Nasopharyngeal swab 0.2% (40/18 043) Male vs. female 0.0018 1.13 (1.05, 1.22)
Nonteaching vs. teaching hospital 0.7597 0.98 (0.84, 1.14)
Non-ICU vs. ICUa <0.0001 0.14 (0.12, 0.17)
Comorbidities
discrimination, which is often measured via a concordance index Non-COPD vs. COPD 0.5117 1.04 (0.92, 1.17)
(C-index). A C-index of 1.0 indicates perfect predictions, whereas Non-DM vs. DM 0.2971 0.94 (0.85, 1.05)
0.5 is equivalent to a coin toss. In our study, the C-index of the lo- Noneheart failure vs. heart failure 0.2591 0.93 (0.83, 1.05)
Nonelung cancer vs. lung cancer 0.0156 0.74 (0.58, 0.94)
gistic regression was 0.727, suggesting reasonable model
Time from admission until initial 0.3098 1.06 (0.95, 1.19)
performance. antibiotics >4 vs. 4 hours
In our series, 2.1% (380/18 043) of CAP patients died. ICU pa- Regimens not covering vs. covering 0.0013 1.21 (1.08, 1.35)
tients had a higher mortality rate (10.1%, 226/2248) than non-ICU atypical pathogens
Not adherent to guidelines vs. adherent <0.0001 1.05 (0.85, 1.29)
patients (1.0%, 154/15 795) (p < 0.0001). Patients who received
to guidelinesa
Chinese guidelineeadherent initial antibiotic therapy had lower in- Interaction between ICU and guideline <0.0001 3.03 (2.43, 3.76)
hospital mortality (1.25%, 112/8994) than patients who did not adherence
(2.96%, 268/9049) (p < 0.0001).
CI, confidence interval; COPD, chronic obstructive pulmonary disease; DM, diabetes
The median LOS was 11 days (interquartile range, 8e15 days). mellitus; ICU, intensive care unit; OR, odds ratio.
Patients experiencing treatment failure had longer LOS than those a
Because interaction of ICU and adherence to Chinese guidelines is included in
not experiencing treatment failure (median, 16 vs. 10 days, logistic regression model, OR of ICU indicated that risk of treatment failure of pa-
tients from non-ICU is 0.14 times of those from ICU when Chinese guidelines were
p < 0.0001). Regimens adherent to the Chinese guidelines and
followed. Similarly, OR of guideline adherence indicated that risk of treatment
covering atypical pathogens were associated with a 2-day (median, failure of patients treated without adherence to guidelines is 1.05 times of those
treated according to guidelines in ICU.

Table 5
Univariate analysis of factors associated with treatment failure 10 vs. 12 days) and 1-day (median, 10 vs. 11 days) shorter LOS,
respectively (p < 0.0001).
Factor Failure rate p

Age <0.0001 Discussion

18e44 years 14.9% (760/5110)
45e64 years 18.3% (795/4336)
65e79 years 24.7% (1035/4186) This study describes initial empiric antibiotic use practices for
80 years 33.1% (1460/4411) adult CAP patients in China and associated patient outcomes. To our
Gender <0.0001 knowledge, this is the first study of this topic in China.
Male 23.5% (2618/11 124) The microbiologic sampling rate was relatively low. Although
Female 20.7% (1432/6919)
Hospital type <0.0001
ICU patients had higher rates of sampling than non-ICU patients,
Nonteaching 22.1% (3729/16 886) only 37.2% of ICU patients had samples collected. The Infectious
Teaching 27.7% (321/1157) Diseases Society of America/American Thoracic Society (IDSA/ATS)
ICU admission within 24 hours <0.0001 guidelines [2] recommend that all ICU patients undergo culture of
Non-ICU 18.3% (2887/15 795)
blood and sputum samples (in patients with a productive cough),
ICU 51.7% (1163/2248)
Comorbidities and Chinese guidelines [4] emphasize the importance of appro-
Non-COPD 22.0% (3558/16 210) <0.0001 priate aetiologic testing when antimicrobial adjustment is neces-
COPD 26.8% (492/1833) sary as a result of insufficient efficacy of empiric anti-infective
Non-DM 21.5% (3367/15 637) <0.0001 treatment. Our results demonstrated that these recommendations
DM 28.4% (683/2406)
Noneheart failure 21.2% (3433/16 172) <0.0001
have not yet been widely incorporated into Chinese practice.
Heart failure 33.0% (617/1871) Educational efforts that include dissemination of the Chinese CAP
Nonelung cancer 22.3% (3936/17 685) <0.0001 guidelines may be of benefit.
Lung cancer 31.8% (114/358) A total of 22.4% of patients experienced treatment failure. This is
Time from admission until 0.0907
consistent with prior studies, in which the failure rate of initial
initial antibiotics
> 4 hours 23.9% (520/2179) antibiotic regimens ranged between 6% and 33% [7e13]. The effect
4 hours 22.3% (3530/15 864) of antibiotic coverage of atypical organisms on patient outcomes in
Antibiotics active against <0.0001 CAP remains unclear. Zhou et al. [14] found initial empiric b-lactam
atypical pathogens monotherapy resulted in a higher treatment failure rate (20.5%)
No 24.1% (3465/14 383)
Yes 16.0% (585/3660)
than fluoroquinolone monotherapy (5.5%) in low-risk CAP patients.
Adherence to Chinese <0.0001 However, Mills et al. [15] found no advantage of atypical pathogen
guidelines in non-ICU patients active antibiotics over b-lactam antibiotics (0.97, 95% confidence
Not adherent 27.2% (2004/7381) interval 0.87 to 1.07) in a meta-analysis. The 2016 Chinese CAP
Adherent 10.5% (883/8414)
guidelines recommend monotherapy with a b-lactam for non-ICU
Adherence to Chinese guidelines 0.0529
in ICU patients patients [4], but the British Thoracic Society [1] and IDSA/ATS [2]
Not adherent 53.0% (883/1668) guidelines do not recommend this regimen for non-ICU patients.
Adherent 48.3% (280/580) However, we found a lower rate of treatment failure in patients
COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; ICU, intensive who received antibiotics active against atypical organisms. More
care unit. research, especially prospective research, is needed on this issue. If

Please cite this article in press as: Nie XM, et al., Initial empiric antibiotic therapy for community-acquired pneumonia in Chinese hospitals,
Clinical Microbiology and Infection (2017), https://doi.org/10.1016/j.cmi.2017.09.014
 X.M. Nie et al. / Clinical Microbiology and Infection xxx (2017) 1e6 5

prospective research also supports atypical pathogen coverage, the
Chinese guidelines may need to be changed to recommend atypical
coverage for all patients.
The importance of TFAD for CAP has been debated. Two retro-
spective studies demonstrated statistically significantly lower
mortality among patients who received early antibiotic therapy
[16,17]. One study suggested that antibiotics be used within 8 hours
[16], whereas the other suggested 4 hours [17]. However, pro-
spective trials have not demonstrated a survival benefit of receiving
antibiotics within 4 to 8 hours [18,19]. Early antibiotic administra-
tion does not appear to shorten the time to clinical improvement or
improve clinical outcomes [20,21]. On the basis of these prospec-
tive trials, the IDSA/ATS and Chinese guidelines do not recommend
a specific time window for delivery of the first antibiotic dose. Our
study, which demonstrated no association between antibiotic
timing and patient outcomes, provides support for this recom-
mendation in Chinese patients.
Monotherapy with a fluoroquinolone was the most commonly
used initial regimen overall, and monotherapy with a carbapenem
was the most commonly used regimen in ICU CAP patients in our
study (Table 2). Imipenem/cilastatin was the most commonly used
carbapenem (Table 3). Several epidemiologic surveys of CAP con-
ducted in Chinese adults have shown that Pseudomonas aeruginosa
is infrequently isolated [22e24]. On the basis of these surveys, the
2016 Chinese CAP guidelines recommended ertapenem, as a rela-
tively narrow-spectrum carbapenem, be used in empiric therapy
for some ICU patients [4]. Frequent empiric use of wide-spectrum
carbapenems such as imipenem/cilastatin in CAP is problematic,
as such a practice likely increases the prevalence of antibiotic-
resistant bacteria [25].
Our study has some limitations. Like all observational studies,
our results might be subject to unmeasured biases. Furthermore, a
change of initial antibiotic therapy more than 72 hours after the
start of antibiotics was one of our definitions of treatment failure.
By this definition, we include de-escalation to a narrower-spectrum
intravenous antibiotic, thereby potentially overestimating the
treatment failure rate. On the basis of our knowledge of clinical
practice in China, we believe that this is uncommon. To investigate
this, we further looked at LOS and mortality of patients with
changes in antibiotics, and found that both were worse than in
patients without changes. Results are shown in Supplementary
Appendix 3. Also, we did a repeat analysis in which we redefined
treatment failure as transfer of patient to the ICU or death more
than 72 hours after the first antibiotic dose, and we found similar
results with respect to relationship with atypical pathogen
coverage and guideline adherence (Supplementary Appendix 3).
On the basis of these findings, we believe changes in antibiotics
were mainly related to treatment failure.
Our study only included parenteral antibiotics. To our knowl-
edge, most patients in China receive only parenteral antibiotics
while in hospital. Nonetheless, we likely have misclassified the
initial empiric antibiotic regimens of some patients. In this retro-
spective study, patients admitted with hospital-acquired pneu-
monia may not be totally excluded. Considering that empiric
coverage for methicillin-resistant Staphylococcus aureus (MRSA)
and Pseudomonas spp. is recommended in the IDSA/ATS hospital-
acquired pneumonia guidelines [26], the use of antibiotics active
against MRSA and Pseudomonas spp. may be overestimated in our
study. Unfortunately, given the inherent limitations of the study
database, we cannot calculate the overestimation.
In conclusion, in this novel study of initial empiric antibiotic
therapy and patient outcomes of CAP in China, monotherapy with a
fluoroquinolone was the most commonly used initial antibiotic
regimen, and levofloxacin was the most commonly used initial
antibiotic. The finding that patients receiving regimens adherent to
the Chinese guidelines had a lower treatment failure rate in non-
ICU patients, lower mortality and shorter LOS supports the need
to disseminate the 2016 Chinese guidelines. Atypical pathogen
active regimens were associated with lower treatment failure rate
and shorter LOS, so future guidelines may emphasize the impor-
tance of atypical pathogen active regimens. Further prospective
research is needed to study other aspects of antibiotic usage for CAP
in China.
Transparency declaration
This work was supported by the National Natural Science
Foundation of China (81370135) and the National Key Research and
Development Plan of China (2017YFC1309704). All authors report
no conflicts of interest relevant to this article.
Appendix 1
Description of database
This database includes all of the hospitals belonging to the
Chinese People's Liberation Army, and all of these hospitals are
open to the public. All the data from these hospitals were collected
by the database automatically. There are 192 hospitals included.
These hospitals are from grade 1 to 3 by Chinese grading system
(grade 1 hospitals are community hospitals, and grade 3 hospitals
are large general hospitals), and all of them agreed to share their
data. The purpose of this database is to monitor the appropriate-
ness of drug use, including antibiotics. The database is not open to
the public yet.
Appendix 2

Appendix Table 1
The three most common initial empiric antibiotic regimens for CAP in different regions of China (following digitals are the percentages of each regimen, and numerators and
denominators are in the brackets)

Regions 1 2 3
nd
Northeast 2 -generation Cephalosporin Fluoroquinolone 20.1% (668/3329) Cephalosporin/beta-lactamase inhibitor
22.5% (749/3329) 12.7% (422/3329)
North China region Fluoroquinolone 16.0% (741/4627) Cephalosporin/beta-lactamase inhibitor 2nd-generation Cephalosporin 6.1% (282/4627)
6.2% (286/4627)
nd
East China region 2 -generation Cephalosporin Fluoroquinolone 8.1% (360/4425) 3rd-generation Cephalosporin 5.7% (252/4425)
8.7% (384/4425)
Central and South Fluoroquinolone 10.4% (249/2385) 3rd-generation Cephalosporin 10.4% (249/2385) 2nd-generation Cephalosporin 7.1% (169/2385)
China region
Northwest region Fluoroquinolone 15.3% (325/2131) 3rd-generation Cephalosporin 14.9% (317/2131) 2nd-generation Cephalosporin 12.9% (275/2131)
Southwest region Fluoroquinolone 28.6% (328/1146) Penicillin/beta-lactamase inhibitor 9.0% (103/1146) 3rd-generation Cephalosporin 5.1% (59/1146)

Please cite this article in press as: Nie XM, et al., Initial empiric antibiotic therapy for community-acquired pneumonia in Chinese hospitals,
Clinical Microbiology and Infection (2017), https://doi.org/10.1016/j.cmi.2017.09.014
6 X.M. Nie et al. / Clinical Microbiology and Infection xxx (2017) 1e6

Appendix Table 2
The three most common initial empiric antibiotic drugs for CAP in different regions of China (following digitals are the percentages of each drug, and numerators and de-
nominators are in the brackets)

Regions 1 2 3

Northeast Moxifloxacin 15.2% (681/4476) Cefminox 9.7% (435/4476) Cefoperazone-tazobactam 9.7% (432/4476)
North China region Moxifloxacin 16.3% (1298/7965) Levofloxacin 12.1% (961/7965) Piperacillin-tazobactam 8.5% (681/7965)
East China region Levofloxacin 21.7% (1779/8215) Penicillin 7.9% (650/8215) Moxifloxacin 7.0% (578/8215)
Central and South China region Levofloxacin 20.0% (793/3965) Cephazolin 7.3% (289/3965) Cefmenoxime 6.9% (275/3965)
Northwest region Levofloxacin 15.8% (465/2941) Moxifloxacin 7.4% (217/2941) Piperacillin- sulbactam 7.2% (211/2941)
Southwest region Moxifloxacin 21.2% (364/1716) Levofloxacin 10.8% (186/1716) Azithromycin 7.5% (129/1716)

Appendix 3 [7] Arancibia F, Ewig S, Martinez JA, Ruiz M, Bauer T, Marcos MA, et al. Antimi-
crobial treatment failures in patients with community-acquired pneumonia:
causes and prognostic implications. Am J Respir Crit Care Med 2000;162:
Appendix Table 3 154e60.
Hospital length of stay in patients with and without a change in antibiotics after 72 [8] Mene ndez R, Torres A, Zalacaín R, Aspa J, Martín Villasclaras JJ, Borderías L,
hours et al. Risk factors of treatment failure in community acquired pneumonia:
implications for disease outcome. Thorax 2004;59:960e5.
Median interquartile P [9] Genne  D, Sommer R, Kaiser L, Saaïdia A, Pasche A, Unger PF, et al. Analysis of
(days) range (days) factors that contribute to treatment failure in patients with community-
acquired pneumonia. Eur J Clin Microbiol Infect Dis 2006;25:159e66.
With changs in antibiotics 16 12, 24 <0.0001 [10] Hoogewerf M, Oosterheert JJ, Hak E, Hoepelman IM, Bonten MJ. Prognostic
Without change in antibiotics 10 8, 13 factors for early clinical failure in patients with severe community-acquired
pneumonia. Clin Microbiol Infect 2006;12:1097e104.
[11] Roson B, Carratala
 J, Fernandez-Sabe  N, Tubau F, Manresa F, Gudiol F. Causes
and factors associated with early failure in hospitalized patients with
community-acquired pneumonia. Arch Intern Med 2004;164:502e8.
Appendix Table 4 [12] Bruns AH, Oosterheert JJ, Hustinx WN, Gaillard CA, Hak E, Hoepelman AI. Time
Mortality in patients with and without a change in antibiotics after 72 hours for first antibiotic dose is not predictive for the early clinical failure of
moderate-severe community-acquired pneumonia. Eur J Clin Microbiol Infect
Mortality P
Dis 2009;28:913e9.
With change in antibiotics 6.4% (248/3845) <0.0001 [13] Aliberti S, Amir A, Peyrani P, Mirsaeidi M, Allen M, Moffett BK, et al. Incidence,
Without changes in antibiotics 0.93% (132/14198) etiology, timing, and risk factors for clinical failure in hospitalized patients
with community-acquired pneumonia. Chest 2008;134:955e62.
[14] Zhou QT, He B, Zhu H. Potential for cost-savings in the care of hospitalized
low-risk community-acquired pneumonia patients in China. Value Health
2009;12:40e6.
[15] Mills GD, Oehley MR, Arrol B. Effectiveness of beta lactam antibiotics
Appendix Table 5
compared with antibiotics active against atypical pathogens in non-severe
Percentage of patients with treatment failure, redefining treatment failure to include
community acquired pneumonia: meta-analysis. BMJ 2005;330:456.
only patients who died or were transferred to ICU after 72 hours and univariate
[16] Meehan TP, Fine MJ, Krumholz HM, Scinto JD, Galusha DH, Mockalis JT, et al.
analysis of associated factors Quality of care, process, and outcomes in elderly patients with pneumonia.
JAMA 1997;278:2080e4.
Antibiotics active against atypical pathogens 0.0004
[17] Houck PM, Bratzler DW, Nsa W, Ma A, Bartlett JG. Timing of antibiotic
No 3.7% (534/14383)
administration and outcomes for Medicare patients hospitalized with
Yes 2.5% (92/3660)
community-acquired pneumonia. Arch Intern Med 2004;164:637e44.
Chinese guideline adherence in Non-ICU patients 0.0128 [18] Benenson R, Magalski A, Cavanaugh S, Williams E. Effects of a pneumonia
Not adherent 2.7% (197/7381) clinical pathway on time to antibiotic treatment, length of stay, and mortality.
Adherent 2.1% (174/8414) Acad Emerg Med 1999;6:1243e8.
Chinese guideline adherence in ICU patients 0.1009 [19] Marrie TJ, Wu L. Factors influencing in-hospital mortality in community-
Not adherent 12.0% (200/1668) acquired pneumonia: a prospective study of patients not initially admitted
Adherent 9.5% (55/580) to the ICU. Chest 2005;127:1260e70.
[20] Silber SH, Garrett C, Singh R, Sweeney A, Rosenberg C, Parachiv D, et al. Early
administration of antibiotics does not shorten time to clinical stability in
References patients with moderate-to-severe community-acquired pneumonia. Chest
2003;124:1798e804.
[1] Lim WS, Baudouin SV, George RC, Hill AT, Jamieson C, Le Jeune I, et al. BTS [21] Wachter RM, Flanders SA, Fee C, Pronovost PJ. Public reporting of antibiotic
guidelines for the management of community acquired pneumonia in adults: timing in patients with pneumonia: lessons from a flawed performance
update, 2009. Thorax 2009;64(Suppl. 3):iii1e55. measure. Ann Intern Med 2008;149:29e32.
[2] Mandell LA, Wunderink RG, Anzueto A, Bartlett JG, Campbell GD, Dean NC, [22] Tao LL, Hu BJ, He LX, Wei L, Xie HM, Wang BQ, et al. Etiology and antimicrobial
et al. Infectious Diseases Society of America/American Thoracic Society resistance of community-acquired pneumonia in adult patients in China. Chin
consensus guidelines on the management of community-acquired pneumonia Med J (Engl) 2012;125:2967e72.
in adults. Clin Infect Dis 2007;44(Suppl. 2):S27e72. [23] Bao Z, Yuan X, Wang L, Sun Y, Dong X. The incidence and etiology of
[3] Diagnosis and treatment of community-acquired pneumonia. Chin J Tuberc community-acquired pneumonia in fever outpatients. Exp Biol Med (May-
Respir Dis 2006;29:651e5. wood) 2012;237:1256e61.
[4] Cao B, Huang Y, She DY, Cheng QJ, Fan H, Tian XL, et al. Diagnosis and treat- [24] Liu YN, Chen MJ, Zhao TM, Wang H, Wang R, Liu QF, et al. A multicentre study
ment of community-acquired pneumonia in adults: 2016 clinical practice on the pathogenic agents in 665 adult patients with community-acquired
guidelines by the Chinese Thoracic Society, Chinese Medical Association. Chin pneumonia in cities of China. Zhonghua Jie He He Hu Xi Za Zhi 2006;29:3e8.
J Tuberc Respir Dis 2016;39:253e9. Available in English at: Clin Respir J. In [25] Bassetti M, Righi E. Multidrug-resistant bacteria: what is the threat? Hematol
press. Am Soc Hematol Educ Program 2013;2013:428e32.
[5] File TM, Marrie TJ. Burden of community-acquired pneumonia in North [26] Guidelines for the management of adults with hospital-acquired, ventilator-
American adults. Postgrad Med 2010;122:130e41. associated, and healthcare-associated pneumonia. Am J Respir Crit Care Med
[6] Oster G, Berger A, Edelsberg J, Weber DJ. Initial treatment failure in non-ICU 2005;171:388e416.
community-acquired pneumonia: risk factors and association with length of
stay, total hospital charges, and mortality. J Med Econ 2013;16:809e19.

Please cite this article in press as: Nie XM, et al., Initial empiric antibiotic therapy for community-acquired pneumonia in Chinese hospitals,
Clinical Microbiology and Infection (2017), https://doi.org/10.1016/j.cmi.2017.09.014