J Antimicrob Chemother 2015; 70: 382 – 395

doi:10.1093/jac/dku379 Advance Access publication 28 September 2014

Bacteriostatic versus bactericidal antibiotics for patients with serious

bacterial infections: systematic review and meta-analysis
Johannes Nemeth1*†, Gabriela Oesch2† and Stefan P. Kuster1†
1
Division of Infectious Diseases and Hospital Epidemiology, University Hospital and University of Zurich, Zurich, Switzerland;
2
University Children’s Hospital, Zurich, Switzerland

*Corresponding author. Tel: +41-44-255-21-73; Fax: +41-44-255-44-33; E-mail: [email protected]

†The authors contributed equally to this work.

Received 2 July 2014; returned 11 August 2014; revised 26 August 2014; accepted 27 August 2014

Objectives: Antibiotics are commonly classified into bactericidal and bacteriostatic agents based on their
antimicrobial action. We aimed to assess whether this distinction is clinically relevant.
Methods: OVID MEDLINE, EMBASE, The Cochrane Central Register of Controlled Trials (CENTRAL) and relevant
references and conference proceedings using the Web of Science and Scopus databases were searched for
randomized controlled trials comparing bactericidal with bacteriostatic antibiotics for treatment of severe infec-
tions. Main outcome measures were clinical cure rates and overall mortality. Abstracts of studies selected in the
database search were screened by one reviewer; full-text screening and data extraction were performed by three
independent reviewers.
Results: Thirty-three studies were included. Approximately half of patients were treated with bacteriostatic
monotherapy. Infections covered were pneumonia (n ¼ 13), skin and soft tissue infections (n ¼ 8), intra-
abdominal infections (n ¼ 4) and others (n ¼ 8). Neither clinical cure rates [risk ratio (RR), 0.99; 95% CI, 0.97 –
1.01; P ¼ 0.11] nor mortality rates (RR, 0.91; 95% CI, 0.76–1.08; P ¼ 0.28) were different between patients treated
with bactericidal drugs and those treated with bacteriostatic drugs. Subgroup analyses showed a benefit for clin-
ical cure rates associated with linezolid and increased mortality associated with tigecycline. In meta-regression,
clinical cure rates remained higher in patients treated with linezolid (P ¼ 0.01); tigecycline displayed a close to
significant association with increased mortality (P ¼0.05) if compared with other bacteriostatic agents.
Conclusions: The categorization of antibiotics into bacteriostatic and bactericidal is unlikely to be relevant in
clinical practice if used for abdominal infections, skin and soft tissue infections and pneumonia. Because
we were not able to include studies on meningitis, endocarditis or neutropenia, no conclusion regarding these
diseases can be drawn.

Keywords: drug classes, mechanism of action, generalizability

Introduction the recommendation that severely ill and immunosuppressed

patients with bacterial infections should be treated with bacteri-
The distinction between bactericidal and bacteriostatic antibiotics cidal antibiotics.2 – 4 Furthermore, some specific conditions are
is a successful concept to discriminate antibiotics that kill also thought to require bactericidal antibiotics, such as endocar-
bacteria—‘bactericidal’—from antibiotics that inhibit bacterial ditis.1 The cardiac valves are considered as focal, immunosup-
growth, i.e. ‘bacteriostatic’. This classification is applied in major pressed regions poorly accessible for phagocytic cells. Therefore,
textbooks of medicine and infectious diseases, clinical guidelines a phagocyte-independent killing by bactericidal drugs is generally
and advertisements of novel antibiotics.1 – 5 The intuitively under- recommended under such circumstances.
standable concept between the two groups of antibiotics sug- Unfortunately, there are no clinical data supporting the con-
gests that bactericidal drugs have more powerful antibacterial cept of bacteriostatic versus bactericidal antibiotics. This is quite
action and are able to kill bacteria. In contrast, bacteriostatic anti- remarkable given the major influence on recommendations for
biotics are assumed to require phagocytic cells to definitely clear treatment of severely ill patients. The reason may lie in the diffi-
bacteria and are therefore thought to be less effective without culties of assessing a drug class effect in a clinically meaning-
an efficient immune response. This theoretical model has led to ful way.

# The Author 2014. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved.
For Permissions, please e-mail: [email protected]

382
Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103
by guest
on 16 April 2018
 Systematic review JAC
The main challenge is that during bacterial infection and Table 1. Definition of bacteriostatic and bactericidal antibiotics (adapted
its treatment, relevant clinical outcomes such as cure rates and from Cohen et al.3)
mortality are influenced by three main factors: the host, the
pathogen and the drug. A drug class effect can only be secondary Bactericidal
to these three main factors and is therefore difficult to assess. Our aminoglycosides
methodological assumption was that the drug class effect ‘bac- b-lactams
tericidal versus bacteriostatic’ may be measurable if the three fluoroquinolones
main factors—host, drugs and pathogens—are as heterogeneous glycopeptides
as possible. If the common denominator is limited to the differ- lipopeptides
ence between bacteriostatic and bactericidal drugs amongst dif- nitroimidazoles and nitrofurans
ferent clinical trials, the results may be reduced to the difference
between bactericidal and bacteriostatic antibiotics. This approach Bacteriostatic
introduces a meta-level that is beyond the range of a single ran- glycylcyclines
domized trial. We therefore conducted a meta-analysis including lincosamides
a wide variety of prospective clinical trials using bacteriostatic ver- macrolides
sus bactericidal antibiotic drugs for the treatment of patients with oxazolidinones
severe bacterial infections. The main outcome measures were streptogramins
clinical outcome and overall mortality. sulphonamides

Methods Studies using inappropriate antibiotics in terms of microbiological

spectrum were excluded. Hence, possible differences are not attributable
Definitions to resistance of bacteria.
Severity
Outcomes
Requirement of hospitalization was used as a determinant to screen for
severity, unless the authors explicitly provided a definition for severity For the primary endpoint, clinical outcome as defined by the respective
in the study. Studies including patients with different levels of severity, study authors was used. Secondary endpoints were mortality and relapse
such as e.g. pneumonia and bronchitis, were excluded. Furthermore, rates. Relapse rates were defined as recurrence of signs and symptoms
studies performed with relatively benign infections were excluded associated with the primary infection during follow-up after initial clinical
(i.e. infections in the oral cavity, uncomplicated otolaryngeal infections, resolution or ‘relapse’ as defined by the authors. Subgroup analyses for dif-
uncomplicated gastrointestinal infections, sexually transmitted diseases, ferent diseases were performed.
Lyme disease, exacerbation of chronic bronchitis, pneumonia in the out-
patient setting etc.). Chronic infections such as diabetic foot infections
Data sources and searches
were excluded under the assumption that successful antibiotic therapy
in this setting does not rely on rapid killing but rather elimination of the bio- With the help of an experienced librarian with expertise in literature search
film, vascular supply and joint and bone involvement. for systematic reviews and meta-analyses, we performed an electronic
search of OVID MEDLINE, EMBASE and The Cochrane Central Register of
Antibiotics Controlled Trials (CENTRAL). We also searched relevant references of
included studies and conference proceedings using the Web of Science
In the bacteriostatic antibiotic patient group, bacteriostatic single therapy and Scopus databases. No time limits were applied. The search strategy
only without the concomitant use of any bactericidal drug was allowed, is displayed in detail in Figure S1 (available as Supplementary data at
with few exceptions: if all three authors agreed that the bactericidal com- JAC Online).
ponent of the combination therapy had no effect on bacteria, these
studies were included as well. Seven studies met these criteria.6 – 12
Six out of seven allowed the use of aztreonam in case of infection with Study selection
Gram-positive bacteria; one study allowed aminoglycosides in case of pro- We included all randomized controlled trials comparing bacteriostatic ver-
ven aminoglycoside resistance of Staphylococcus aureus infection.6 The sus bactericidal antibiotics in patients with serious bacterial infections
rate of resistance was similar between bacteriostatic and bactericidal anti- requiring hospitalization. There was no restriction by study site/country
biotics. For the seven studies with combination therapy, only the microbio- or follow-up period and there was no restriction by dose, frequency or
logically confirmed cases were extracted for analysis in the clinical cure method of drug administration. One author (J. N.) screened the title and
groups. In all other studies, the ITT population was assessed. If no ITT abstract of each reference identified by the search and applied the inclu-
population was available, results from PP analyses were extracted. sion criteria. For possibly relevant articles, the full-text article was obtained
In the bactericidal antibiotic treatment group, combination therapy and reviewed independently by two out of the three authors (J. N.,
was allowed. Antibiotics were defined as bactericidal and bacteriostatic G. O. and S. P. K.). Final inclusion of studies was determined by agreement
according to a classification obtained from a major textbook of infectious of both reviewers and involvement of a third author in case of discrepancy.
diseases, with one important exception (Table 1).3 Aminoglycosides act at After in-depth discussion of the different opinions, the authors agreed
the ribosome and are therefore classified as bacteriostatic by some unanimously on the final classification. Only studies published in English
authors. At higher concentrations, however, aminoglycosides display a were considered eligible.
very rapid bactericidal effect. Due to considerations regarding side effects,
aminoglycosides are recommended to be given once daily at high con-
centrations for most indications.13 Therefore, aminoglycosides are most
Data extraction and quality assessment
likely to be bactericidal in routine clinical practice and were classified Two out of the three authors (J. N., G. O. and S. P. K.) independently
accordingly. extracted data from included trials. Data extraction was performed

383
Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103
by guest
on 16 April 2018
 Systematic review

using a standardized data collection form. When missing data were

16 490 references
encountered, the corresponding authors were contacted to retrieve them.
For the assessment of clinical cure, data from ITT populations were
extracted if possible. Some studies did not report the clinical cure rates
of the ITT population or—in case of older studies—did not discern between Title and abstract
different study populations at all. In such a case, results from modified 674 references
ITT or PP populations were extracted. In case of the seven studies with
combination therapy, only the microbiologically confirmed cases were
extracted.6 – 12
For mortality analysis, the ITT (‘safety’) population was used through- Full paper screen
out. The primary outcome measures were (i) clinical cure rate as defined 81 references
Excluded
by the investigators in the study and (ii) mortality rate. Data were extracted
- Bactericidal antibiotics allowed in
as proportions if results were only reported as probability of events or
bacteriostatic group (n = 24)
mortality and if the number of events was not explicitly provided.
Analysed - Outpatients included (n = 9)
Unfortunately, follow-up data and periods of follow-up after end of treat-
33 studies - No randomized controlled trial (n = 6)
ment were heterogeneous. - Poor quality of reporting (n = 6)
Subgroup analysis was performed for different diseases and different - Double publication (n = 3)
classes of antibiotics. Meta-analysis was performed only if at least three
studies included the same disease or antibiotic.
Figure 1. Study selection.
To assess methodological quality and risk of bias, included articles were
examined for (i) randomization process, (ii) blinding, (iii) incomplete out-
come data and (iv) reporting bias/sponsorship as described in the
a bacteriostatic agent. Meta-analysis indicated that clinical
Cochrane Handbook for Systematic Reviews of Interventions.14
cure rates were not different between bactericidal and bacterio-
static drugs [risk ratio (RR), 0.99; 95% CI, 0.97 – 1.01; P ¼ 0.11;
Data synthesis and analysis Figure 2], with an overall heterogeneity of I 2 ¼ 24%. Analysis of
subgroups was performed on studies using tigecycline, linezolid,
Data synthesis was performed using Review Manager (version 5.2, The
Cochrane Collaboration, The Nordic Cochrane Centre, Copenhagen, macrolides, b-lactam antibiotics, glycopeptides and fluoroquino-
Denmark). P values ,0.05 were considered statistically significant. lones. For the other drug classes, an insufficient number of studies
Subgroup analyses for differences between different classes of antibiotics were found, precluding statistical analysis. Subgroup analyses
and differences between different diseases were performed. Because het- revealed that treatment with linezolid appeared to be associated
erogeneity was anticipated between studies, a random-effects model was with better clinical cure rates compared with its bactericidal
used for all analyses. Statistical heterogeneity was inspected graphically comparator (RR, 0.93; 95% CI, 0.87 – 0.99; P ¼ 0.04; Figure 2). In
(forest plot) and the degree of heterogeneity quantified using the I 2 stat- meta-regression, combined cure rates of trials using linezolid
istic. Publication bias was investigated using a funnel plot in which the were better than those from tigecycline trials (RR, 1.07; 95% CI,
standard error of the effect estimate of each study was plotted against 1.02– 1.12; P¼ 0.01). There was no benefit detectable for clinical
the estimate. An asymmetric plot suggested possible publication bias. cure rates among subgroups of bactericidal antibiotics (Figure 3).
We used STATA 13 (StataCorp 2013, College Station, TX, USA) for
meta-regression assessing differences between bacteriostatic drug
classes. Mortality
The population for the assessment of mortality consisted of
Results 13 098 patients in total. There was no significant increase in over-
all mortality in the patient group treated with bacteriostatic anti-
Study selection biotics compared with patients treated with bactericidal
The literature search yielded 16 490 references. After screening antibiotics (RR, 0.91; 95% CI, 0.76 – 1.08; P ¼ 0.28; Figure 4).
titles and abstracts, 81 articles were selected for full-text screen- Heterogeneity was low (I 2 ¼ 0%). An increased mortality was
ing (Figure 1). Thirty-three studies met the inclusion criteria found in studies using tigecycline in the bacteriostatic regimen
(Table 2). The bacteriostatic agents included tigecycline, linezolid, group (RR, 0.66; 95% CI, 0.46 – 0.94; P ¼ 0.02; Figure 4). In
macrolides, sulphonamides, tetracyclines and streptogramins. meta-regression, there was a strong trend towards reduced mor-
The bactericidal agents included b-lactam antibiotics, glycopep- tality in studies using linezolid compared with tigecycline trials
tide antibiotics, fluoroquinolones and aminoglycosides. Diseases (RR, 0.64; 95% CI, 0.41 – 1.00; P ¼ 0.05). There was no mortality
under study included pneumonia (13 studies6,11,15 – 19,21 – 26 ), benefit detectable for subgroups of bactericidal antibiotics
skin and soft tissue infections (8 studies7,8,27 – 32), intra-abdominal (Figure 5).
infections (4 studies33 – 36) and others (8 studies10,20,37 – 42). No
trials in neutropenic patients or patients with endocarditis or men-
ingitis met inclusion criteria. All studies except one study on chil- Relapse
dren with typhoid included adults only.37 In 12 studies, relapse rates were reported.7,8,17,18,20,34,35,37,39 – 42
Five studies included relapse rates in the clinical cure rate end-
point.17,18,35,39,42 In the remaining seven studies, differences in
Clinical cure rates relapse rates between bacteriostatic and bactericidal antibiotics
Data from 9597 patients were available for meta-analysis of clin- were not significant (RR, 0.76; 95% CI, 0.43 – 1.35; P ¼ 0.51; data
ical cure rates. Of these patients, 4717 (49.2%) were treated with not shown).

384
Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103
by guest
on 16 April 2018
 Systematic review
Table 2. Study characteristics

Antibiotics,
Author Year Type of study Disease Antibiotics, bactericidal bacteriostatic

Bergallo et al. 15 2009 Phase 3, multicentre, double-blind study community-acquired pneumonia levofloxacin tigecycline
Bernard et al. 27 1992 prospective, randomized, complicated skin and soft tissue penicillin roxithromycin
multicentre trial infections
Bohte et al. 16 1995 open-label, randomized, community-acquired pneumonia penicillin azithromycin
multicentre study
Breedt et al. 28 2005 randomized, double-blind, controlled, complicated skin and soft tissue vancomycin tigecycline
multicentre trial infections
Chen et al. 33 2010 Phase 3, multicentre, open-label study complicated intra-abdominal infection imipenem/cilastatin tigecycline
Chuang et al. 29 2011 two Phase 3, multicentre, randomized, complicated skin and soft tissue vancomycin/aztreonam tigecycline
double-blind studies infections
Dartois et al. 17 2008 randomized, Phase 3, multicentre trial community-acquired pneumonia levofloxacin tigecycline
Ellis-Grosse et al. 30 2005 two Phase 3, randomized, double-blind, complicated skin and soft tissue vancomycin/aztreonam tigecycline
multicentre studies infections
Frenck et al. 37 2000 randomized, open-label, controlled, typhoid fever ceftriaxone azithromycin
single-centre trial
Genne et al. 18 1997 open-label, prospective, randomized, community-acquired pneumonia amoxicillin clarithromycin
single-centre study
Itani et al. 7 2010 prospective, randomized, open-label, complicated skin and soft tissue vancomycin linezolid
controlled, multicentre, Phase 4 study infections
Jauregui et al. 8 2005 randomized, double-blind, controlled, complicated skin and soft tissue dalbavancin linezolid
multicentre, Phase 3 trial infections
Kohno et al. 38 2007 randomized, open-label, nosocomial pneumonia, complicated vancomycin linezolid
comparator-controlled, skin and soft tissue infections or sepsis
multicentre study caused by MRSA
Kuzman et al. 19 2005 randomized, open-label, community-acquired pneumonia cefuroxime azithromycin
multicentre study
Lin et al. 10 2008 randomized, double-blind, pneumonia or complicated skin and soft vancomycin linezolid
comparator-controlled, tissue infection due to suspected or
multicentre study known Gram-positive pathogens
Markowitz et al. 39 1992 randomized, double-blind, comparative, infection with S. aureus vancomycin trimethoprim/
single-centre study sulfamethoxazole
Mehtar et al. 20 1982 open-label, randomized, severe respiratory tract infections cefuroxime trimethoprim/
single-centre study sulfamethoxazole
Mwengee et al. 40 2006 randomized, controlled, comparative, plague gentamicin doxycycline

JAC
open-label trial

Continued
385

Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103

by guest
on 16 April 2018
 Table 2. Continued
386

Systematic review
Antibiotics,
Author Year Type of study Disease Antibiotics, bactericidal bacteriostatic

Mokabberi et al. 21 2010 prospective, randomized, double-blind, community-acquired pneumonia levofloxacin doxycycline
single-centre trial
Ode et al. 41 1983 open-label, randomized, prospective, acute pyelonephritis ampicillin/mecillinam trimethoprim/
single-centre, Phase 3 trial sulfamethoxazole
Oliva et al. 34 2005 prospective, randomized, double-blind, complicated intra-abdominal infection imipenem tigecycline
multicentre trial
Plouffe et al. 22 2000 two multicentre, open-label Phase 3 community-acquired pneumonia cefuroxime azithromycin
trials, one parallel-group, randomized,
one non-comparative with sequential
inclusion
Qvist et al. 35 2012 randomized, open-label, multicentre, complicated intra-abdominal infections ceftriaxone/metronidazole tigecycline
Phase 3b/4 trial
Raad et al. 42 1999 evaluator-blind, prospective, catheter-related infections vancomycin quinupristin/dalfopristin
randomized, Phase 2, multicentre trial
Sacchidanand et al. 31 2005 Phase 3, randomized, double-blind, complicated skin and soft tissue vancomycin/aztreonam tigecycline
multicentre study infections
San Pedro et al. 11 2002 multicentre, randomized, open-label trial pneumonia with Streptococcus ceftriaxone/cefpodoxime linezolid
pneumoniae
Shanson et al. 23 1984 prospective, randomized, community-acquired pneumonia ampicillin/flucloxacillin erythromycin
single-centre trial
Sterner et al. 24 1967 randomized, open-label, community-acquired pneumonia cefaloridine erythromycin
single-centre trial
Stevens et al. 32 2000 randomized, double-blind, complicated skin and soft tissue infection oxacillin/dicloxacillin linezolid
multicentre trial
Tanaseanu et al. 26 2009 prospective, double-blind, non-inferiority, community-acquired pneumonia levofloxacin tigecycline
multicentre, Phase 3 trial
Tanaseanu et al. 25 2008 two Phase 3, multicentre, randomized, community-acquired pneumonia levofloxacin tigecycline
double-blind studies
Towfigh et al. 36 2010 multicentre, open-label, randomized, complicated intra-abdominal infection ceftriaxone/metronidazole tigecycline
Phase 3b/4 study
Wunderink et al. 6 2012 prospective, double-blind, controlled, confirmed MRSA pneumonia vancomycin linezolid
multicentre trial

Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103

by guest
on 16 April 2018
 Systematic review JAC

Figure 2. RRs for clinical cure rates stratified by use of different bacteriostatic antibiotics. Data markers indicate RRs and error bars indicate 95% CIs.

387
Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103
by guest
on 16 April 2018
 Systematic review

Figure 3. RRs for clinical cure rates stratified by use of different bactericidal antibiotics. Data markers indicate RRs and error bars indicate 95% CIs.

388
Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103
by guest
on 16 April 2018
 Systematic review JAC

Figure 4. RRs for mortality rates stratified by use of different bacteriostatic antibiotics. Data markers indicate RRs and error bars indicate 95% CIs.

389
Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103
by guest
on 16 April 2018
 Systematic review

Figure 5. RRs for mortality rates stratified by use of different bactericidal antibiotics. Data markers indicate RRs and error bars indicate 95% CIs.

390
Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103
by guest
on 16 April 2018
 Systematic review JAC
Different diseases drugs. A possible difference from a previously published meta-
analysis on tigecycline is the strict exclusion of concomitant
We performed subgroup analyses of the most frequent diseases:
bactericidal medication.43
abdominal infections, skin and soft tissue infections and pneumo-
It is important to note that the analysis of clinical cure rates and
nia. No differences in clinical outcome between diseases could be
mortality includes two different populations. Clinical cure rates
found (Figure 6). However, increased mortality was detected in
included all patients who were followed up adequately; overall
skin and soft tissue infections treated with bacteriostatic agents
and including all the older studies, they represent a sample that
(RR, 0.43; 95% CI, 0.23 –0.84; P ¼ 0.01; Figure 7). If studies using
is closer to PP. The mortality analysis consists of safety data,
tigecycline were excluded, the difference was no longer significant
resembling to a great extent an ITT population. These two popula-
(RR, 0.57; 95% CI, 0.26 –1.27; P ¼0.17), suggesting that this effect
tions cannot be distinguished by contemporary definitions,
was again due to the increased overall mortality attributed to
because we included a substantial number of older studies.
tigecycline. Notably, all four studies with abdominal infections
Furthermore, it is important to note that the subgroup analyses
had tigecycline in the bacteriostatic regimen arm without differ-
are supportive of the main research question, i.e. whether a class
ence in mortality.
effect between bacteriostatic and bactericidal drugs is detectable.
They should not be regarded as ‘independent’ investigations and
Study quality their results should not be overemphasized.
Renowned textbooks suggest that immunocompromised
The quality of included studies was variable (Figures S2 and S3). patients should be treated with bactericidal antibiotics.3,4 Our
More than half of studies (17/33) were unblinded, resulting in sig- analyses include a wide variety of serious bacterial infections,
nificant risk of performance bias. More than 50% of studies were including respiratory tract infections, complicated abdominal
sponsored by the pharmaceutical company that manufactured infections and skin infections. It is widely accepted that a central
the respective drug and were thus regarded to be at high risk of pathogenetic feature of most bacterial infections is either local
reporting bias. Funnel plots did not suggest publication bias (e.g. disruption of the membrane integrity, such as injury of the
(Figures S4 –S7). skin serving for severe skin infection, or obstruction, such as
appendicitis caused by luminal obstruction) or systemic immuno-
suppression (e.g. neutropenia).46 As it is generally assumed that
Discussion
bacteriostatic antibiotics require a fully functioning immune sys-
In our meta-analysis comparing bactericidal with bacteriostatic tem to kill bacteria, one would expect differences in clinical cure
antibiotics, no significant differences in clinical cure rates were rates in the patient group receiving bacteriostatic drugs. Our
found in the diseases under study. Subgroup analysis even sug- meta-analysis suggests that this is not the case. Therefore, it
gested that linezolid may have better clinical cure rates if compared may be hypothesized that either the local immunosuppression
with its bactericidal comparators. This subgroup included severely ill in respiratory tract infections, complicated abdominal infections
patients with confirmed MRSA infections, amongst others.6,38 and skin infections is not clinically relevant enough to require
Sufficient antibacterial action of antibiotics in such a patient group the presence of a bactericidal drug or that the biology of bacterial
is therefore crucial. As a note of caution, involvement of the sponsor killing is much more complex than the simplistic concept of bac-
in the trials using linezolid was substantial and meta-analysis teriostatic versus bactericidal drug activity suggests, even in local
showed some heterogeneity. Furthermore, vancomycin, which is immunosuppression. In any case, our data suggest class indiffer-
difficult to dose appropriately, was the comparator drug in the ence for infection where the immune system is intact or at least
majority of studies. Therefore, it is premature to conclude that not impaired in a major way. Based on the data analysed, no con-
linezolid is more efficient than bactericidal antibiotics. However, clusions regarding severely immunosuppressed patients, such as
our findings suggest that linezolid monotherapy is on par with the patients with neutropenia, can be drawn.
standard bactericidal therapy in this patient group. Because of our strict inclusion criteria, we were unable to
Our meta-analysis did not detect differences in overall mortal- include studies assessing neutropenic patients or patients with
ity between the patient groups. However, there was an increased endocarditis. These are the ‘classical’ indications for a bactericidal
mortality associated with the use of tigecycline, an observation in drug regimen. Only one randomized controlled trial using a bac-
line with data published recently.43 Meta-regression showed teriostatic drug—linezolid—in patients with neutropenia was
a trend towards increased mortality of patients treated with tige- found.47 Unfortunately, a bactericidal add-on medication was
cycline compared with linezolid. The association between allowed, precluding inclusion in our meta-analysis.47 Thus, the
increased mortality and tigecycline was predominantly found findings from our work can only be applied to the diseases
amongst patients with severe skin infections. Remarkably, mortal- under study in our meta-analysis and may not be extended to
ity was not increased in 931 patients treated with tigecycline for endocarditis and neutropenia.
complicated intra-abdominal infections. An increased mortality in A major theoretical complication of bacteriostatic agents is
specific patient populations without differences in clinical cure relapse of infection after treatment because of the failure to
rates may have various reasons, such as toxicity of the drug, phar- clear the infection. In the current investigation, there was no
macokinetic/pharmacodynamic issues or other reasons.43 Some detectable difference in relapse rates between bactericidal and
authors suggest that the drug may be inadequately dosed, bacteriostatic antibiotics.
at least for ventilator-associated pneumonia.44 This may well Our study has limitations. To our knowledge, this is the first
explain observed breakthrough bacteraemias as well.45 In study to undertake a meta-analysis on a class effect of drugs.
conclusion, increased mortality is probably a specific feature of Therefore, we have no means to measure if the studies included
tigecycline and not a class effect inherent to all bacteriostatic are heterogeneous enough regarding the three main factors

391
Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103
by guest
on 16 April 2018
 Systematic review

Figure 6. RRs for clinical cure rates stratified by different diseases. Data markers indicate RRs and error bars indicate 95% CIs.

(host, pathogen and drug) to conclude that the only difference sound arguments in favour of such a conclusion. Regarding the
between the two patient groups really is the difference between pathogen, we included studies on a wide array of diseases, ran-
bacteriostatic and bactericidal antibiotics. There are, however, ging from Yersinia pestis to methicillin-resistant staphylococci.6,40

392
Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103
by guest
on 16 April 2018
 Systematic review JAC

Figure 7. RRs for mortality rates stratified by different diseases. Data markers indicate RRs and error bars indicate 95% CIs.

393
Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103
by guest
on 16 April 2018
 Systematic review

Regarding the host, patients with severe local infections such as S. P. K. Critical revision of the manuscript for important intellectual content:
severe skin infections or severe abdominal infections were ana- J. N., G. O. and S. P. K.
lysed as well as patients with systemic infections such as salmon-
ellosis or central line-associated infections. Regarding the drugs, a
total of 10 different antibiotic classes were assessed. These differ- Supplementary data
ent drugs may vary significantly in tissue distribution and may Figures S1 to S7 are available as Supplementary data at JAC Online (http://
have different effects according to their dosing schedules. jac.oxfordjournals.org/).
Amongst the bacteriostatic drugs, the newer drugs tigecycline
and linezolid are over-represented. The numbers of patients
included in clinical trials increases over time, probably due to a ris- References
ing standard in research practice. This may explain the observed 1 Habib G, Hoen B, Tornos P et al. Guidelines on the prevention, diagnosis,
over-representation of newer bacteriostatic drugs. and treatment of infective endocarditis (new version 2009): the Task Force
The included studies have a significant risk for performance on the Prevention, Diagnosis, and Treatment of Infective Endocarditis of
and publication bias. In general, there may be a publication bias the European Society of Cardiology (ESC). Endorsed by the European
in favour of bacteriostatic drugs—i.e. linezolid and tigecycline. Society of Clinical Microbiology and Infectious Diseases (ESCMID) and
However, some studies may have a publication bias in favour of the International Society of Chemotherapy (ISC) for Infection and
the bactericidal component as well, as in the case of dalbavancin Cancer. Eur Heart J 2009; 30: 2369– 413.
compared with linezolid.8 2 American Thoracic Society. Guidelines for the management of adults
The microbiological definition of bacteriostatic versus bacteri- with hospital-acquired, ventilator-associated, and healthcare-associated
cidal suggests a degree of clarity that is not supported by the pneumonia. Am J Respir Crit Care Med 2005; 171: 388–416.
evidence. Some drugs labelled as ‘bacteriostatic’ do have ‘bacteri- 3 Cohen J, Powderly WG, Opal SM et al. Infectious Diseases, 2 vols.
cidal’ effects under some in vitro conditions and vice versa.48 For Mosby, 2003.
example, the exchange of culture media—media used for grow- 4 Fauci A, Dennis L, Longo D et al. Harrison’s Principles of Internal Medicine.
ing of eukaryotic cells instead of ‘classical’ Mueller – Hinton McGraw-Hill Professional, 2008.
broth—may significantly affect drug susceptibility in vitro. 49 5 Cubist Pharmaceuticals Inc. Lexington, MA, USA. http://www.cubicin.
Thus, it is probably very difficult to predict the action of a given com/about.html.
antibiotic drug in terms of bacteriostatic or bactericidal in an 6 Wunderink RG, Niederman MS, Kollef MH et al. Linezolid in methicillin-
actual patient with an ongoing bacterial infection. Furthermore, resistant Staphylococcus aureus nosocomial pneumonia: a randomized,
the action of a given drug may be dependent on the bacterial controlled study. Clin Infect Dis 2012; 54: 621– 9.
load and the interaction with the immune system at the site of 7 Itani KM, Dryden MS, Bhattacharyya H et al. Efficacy and safety of
infection. For example, the capacity of neutrophils to phagocytose linezolid versus vancomycin for the treatment of complicated skin and
bacteria is limited by the sheer quantity of bacteria.50 soft-tissue infections proven to be caused by methicillin-resistant
To our knowledge, this is the first study using clinical data to Staphylococcus aureus. Am J Surg 2010; 199: 804–16.
investigate the difference between bactericidal and bacteriostatic 8 Jauregui LE, Babazadeh S, Seltzer E et al. Randomized, double-blind
antibiotics. Summarized, the data at hand suggest that this clas- comparison of once-weekly dalbavancin versus twice-daily linezolid ther-
sification is clinically irrelevant if used for abdominal infections, apy for the treatment of complicated skin and skin structure infections.
skin and soft tissue infections and pneumonia. These findings Clin Infect Dis 2005; 41: 1407 –15.
cannot be applied directly to patients with meningitis, neutro- 9 Kohno S, Yamaguchi K, Aikawa N et al. Linezolid versus vancomycin for
penia or endocarditis, because we were unable to include these the treatment of infections caused by methicillin-resistant Staphylococcus
types of infection in our meta-analysis. aureus in Japan. J Antimicrob Chemother 2007; 60: 1361– 9.
10 Lin DF, Zhang YY, Wu JF et al. Linezolid for the treatment of infections
caused by Gram-positive pathogens in China. Int J Antimicrob Agents
Acknowledgements 2008; 32: 241–9.
We are thankful for fruitful discussions with Rainer Weber, Nicolas Müller, 11 San Pedro GS, Cammarata SK, Oliphant TH et al. Linezolid versus ceftri-
Jan S. Fehr, Dominique L. Braun, Huldrych F. Guenthard, Ulrich Matt, axone/cefpodoxime in patients hospitalized for the treatment of
Elisabeth and Titus Nemeth. Streptococcus pneumoniae pneumonia. Scand J Infect Dis 2002; 34:
720–8.
12 Stevens DL, Smith LG, Bruss JB et al. Randomized comparison of line-
Funding zolid (PNU-100766) versus oxacillin-dicloxacillin for treatment of compli-
This study was carried out as part of our routine work. cated skin and soft tissue infections. Antimicrob Agents Chemother
2000; 44: 3408– 13.
13 Hatala R, Dinh T, Cook DJ. Once-daily aminoglycoside dosing in
Transparency declarations immunocompetent adults: a meta-analysis. Ann Intern Med 1996; 124:
None to declare. 717–25.
14 Cochrane Collaboration. Cochrane Handbook for Systematic Reviews of
Interventions. http://handbook.cochrane.org/.
Author contributions 15 Bergallo C, Jasovich A, Teglia O et al. Safety and efficacy of intravenous
Study concept and design: J. N., G. O. and S. P. K. Acquisition of data: J. N., tigecycline in treatment of community-acquired pneumonia: results from
G. O. and S. P. K. Statistical analysis: J. N. and S. P. K. Analysis and interpret- a double-blind randomized phase 3 comparison study with levofloxacin.
ation of data: J. N., G. O. and S. P. K. Drafting of the manuscript: J. N. and Diagn Microbiol Infect Dis 2009; 63: 52– 61.

394
Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103
by guest
on 16 April 2018
 Systematic review JAC
16 Bohte R, van’t Wout JW, Lobatto S et al. Efficacy and safety of azithro- 33 Chen Z, Wu J, Zhang Y et al. Efficacy and safety of tigecycline mono-
mycin versus benzylpenicillin or erythromycin in community-acquired therapy vs. imipenem/cilastatin in Chinese patients with complicated
pneumonia. Eur J Clin Microbiol Infect Dis 1995; 14: 182– 7. intra-abdominal infections: a randomized controlled trial. BMC Infect Dis
17 Dartois N, Castaing N, Gandjini H et al. Tigecycline versus levofloxacin 2010; 10: 217.
for the treatment of community-acquired pneumonia: European experi- 34 Oliva ME, Rekha A, Yellin A et al. A multicenter trial of the efficacy
ence. J Chemother 2008; 20 Suppl 1: 28– 35. and safety of tigecycline versus imipenem/cilastatin in patients with compli-
18 Genne D, Siegrist HH, Humair L et al. Clarithromycin versus amoxicillin- cated intra-abdominal infections [Study ID Numbers: 3074A1-301-WW;
clavulanic acid in the treatment of community-acquired pneumonia. Eur J ClinicalTrials.gov Identifier: NCT00081744]. BMC Infect Dis 2005; 5: 88.
Clin Microbiol Infect Dis 1997; 16: 783– 8. 35 Qvist N, Warren B, Leister-Tebbe H et al. Efficacy of tigecycline versus
19 Kuzman I, Dakovic-Rode O, Oremus M et al. Clinical efficacy and safety ceftriaxone plus metronidazole for the treatment of complicated
of a short regimen of azithromycin sequential therapy vs standard cefur- intra-abdominal infections: results from a randomized, controlled trial.
oxime sequential therapy in the treatment of community-acquired pneu- Surg Infect 2012; 13: 102–9.
monia: an international, randomized, open-label study. J Chemother 36 Towfigh S, Pasternak J, Poirier A et al. A multicentre, open-label, rando-
2005; 17: 636–42. mized comparative study of tigecycline versus ceftriaxone sodium plus
20 Mehtar S, Parr JH, Morgan DJ. A comparison of cefuroxime and metronidazole for the treatment of hospitalized subjects with complicated
co-trimoxazole in severe respiratory tract infections. J Antimicrob intra-abdominal infections. Clin Microbiol Infect 2010; 16: 1274 –81.
Chemother 1982; 9: 479–84. 37 Frenck RW Jr, Nakhla I, Sultan Y et al. Azithromycin versus ceftriaxone
21 Mokabberi R, Haftbaradaran A, Ravakhah K. Doxycycline vs. levofloxa- for the treatment of uncomplicated typhoid fever in children. Clin Infect Dis
cin in the treatment of community-acquired pneumonia. J Clin Pharm Ther 2000; 31: 1134– 8.
2010; 35: 195–200. 38 Kohno S, Yamaguchi K, Aikawa N et al. Linezolid versus vancomycin for
22 Plouffe J, Schwartz DB, Kolokathis A et al. Clinical efficacy of intraven- the treatment of infections caused by methicillin-resistant Staphylococcus
ous followed by oral azithromycin monotherapy in hospitalized patients aureus in Japan. J Antimicrob Chemother 2007; 60: 1361– 9.
with community-acquired pneumonia. The Azithromycin Intravenous 39 Markowitz N, Quinn EL, Saravolatz LD. Trimethoprim-sulfamethoxazole
Clinical Trials Group. Antimicrob Agents Chemother 2000; 44: 1796 –802. compared with vancomycin for the treatment of Staphylococcus aureus
23 Shanson DC, McNabb WR, Williams TD et al. Erythromycin compared infection. Ann Intern Med 1992; 117: 390– 8.
with a combination of ampicillin plus flucloxacillin for the treatment of 40 Mwengee W, Butler T, Mgema S et al. Treatment of plague with genta-
community acquired pneumonia in adults. J Antimicrob Chemother micin or doxycycline in a randomized clinical trial in Tanzania. Clin Infect
1984; 14: 75– 9. Dis 2006; 42: 614– 21.
24 Sterner G, Fransen H, Tunevall G et al. A comparative trial of cephalor- 41 Ode B, Cronberg S, Flamholc L et al. Intravenous mecillinam, trimethoprim
idine and erythromycin in hospitalized pneumonic patients. Postgrad Med J and ampicillin in acute pyelonephritis. Drugs Exp Clin Res 1983; 9: 337–43.
1967; 43 Suppl 43: 53–4. 42 Raad I, Bompart F, Hachem R. Prospective, randomized dose-ranging
25 Tanaseanu C, Bergallo C, Teglia O et al. Integrated results of 2 phase 3 open phase II pilot study of quinupristin/dalfopristin versus vancomycin
studies comparing tigecycline and levofloxacin in community-acquired in the treatment of catheter-related staphylococcal bacteremia. Eur J
pneumonia. Diagn Microbiol Infect Dis 2008; 61: 329–38. Clin Microbiol Infect Dis 1999; 18: 199– 202.
26 Tanaseanu C, Milutinovic S, Calistru PI et al. Efficacy and safety of tige- 43 Tasina E, Haidich AB, Kokkali S et al. Efficacy and safety of tigecycline for
cycline versus levofloxacin for community-acquired pneumonia. BMC Pulm the treatment of infectious diseases: a meta-analysis. Lancet Infect Dis
Med 2009; 9: 44. 2011; 11: 834–44.
27 Bernard P, Plantin P, Roger H et al. Roxithromycin versus penicillin in the 44 Ramirez J, Dartois N, Gandjini H et al. Randomized phase 2 trial to
treatment of erysipelas in adults: a comparative study. Br J Dermatol evaluate the clinical efficacy of two high-dosage tigecycline regimens ver-
1992; 127: 155–9. sus imipenem-cilastatin for treatment of hospital-acquired pneumonia.
Antimicrob Agents Chemother 2013; 57: 1756– 62.
28 Breedt J, Teras J, Gardovskis J et al. Safety and efficacy of tigecycline in
treatment of skin and skin structure infections: results of a double-blind 45 Cho SY, Kang CI, Chung DR et al. Breakthrough bacteremia due to
phase 3 comparison study with vancomycin-aztreonam. Antimicrob extended-spectrum-b-lactamase-producing Klebsiella pneumoniae dur-
Agents Chemother 2005; 49: 4658 –66. ing combination therapy with colistin and tigecycline. Antimicrob Agents
Chemother 2012; 56: 4994– 5.
29 Chuang YC, Chang CM, Aradhya S et al. Efficacy and safety of tigecyc-
line monotherapy compared with vancomycin-aztreonam in the treat- 46 Mandell G, Bennett J, Dolin R. Principles and Practice of Infectious
ment of complicated skin and skin structure infections in patients from Diseases. Philadelphia: Churchill Livingstone, 2010.
India and Taiwan. J Microbiol Immunol Infect 2011; 44: 116–24. 47 Jaksic B, Martinelli G, Perez-Oteyza J et al. Efficacy and safety of linezolid
30 Ellis-Grosse EJ, Babinchak T, Dartois N et al. The efficacy and safety of compared with vancomycin in a randomized, double-blind study of febrile
tigecycline in the treatment of skin and skin-structure infections: results of neutropenic patients with cancer. Clin Infect Dis 2006; 42: 597–607.
2 double-blind phase 3 comparison studies with vancomycin-aztreonam. 48 Pankey GA, Sabath LD. Clinical relevance of bacteriostatic versus bac-
Clin Infect Dis 2005; 41 Suppl 5: S341–53. tericidal mechanisms of action in the treatment of Gram-positive bacterial
31 Sacchidanand S, Penn RL, Embil JM et al. Efficacy and safety of tigecyc- infections. Clin Infect Dis 2004; 38: 864–70.
line monotherapy compared with vancomycin plus aztreonam in patients 49 Buyck JM, Plesiat P, Traore H et al. Increased susceptibility of
with complicated skin and skin structure infections: results from a phase 3, Pseudomonas aeruginosa to macrolides and ketolides in eukaryotic cell cul-
randomized, double-blind trial. Int J Infect Dis 2005; 9: 251– 61. ture media and biological fluids due to decreased expression of oprM and
32 Stevens DL, Smith LG, Bruss JB et al. Randomized comparison of line- increased outer-membrane permeability. Clin Infect Dis 2012; 55: 534–42.
zolid (PNU-100766) versus oxacillin-dicloxacillin for treatment of compli- 50 Drusano GL, Liu W, Fikes S et al. Interaction of drug- and granulocyte-
cated skin and soft tissue infections. Antimicrob Agents Chemother mediated killing of Pseudomonas aeruginosa in a murine pneumonia
2000; 44: 3408– 13. model. J Infect Dis 2014; doi:10.1093/infdis/jiu237.

395
Downloaded from https://academic.oup.com/jac/article-abstract/70/2/382/2911103
by guest
on 16 April 2018