The Antibiotic Resistance Patterns of Klebsiella pneumoniae Clinic Isolates: A Comprehensive Meta-Analysis

Background/Purpose: Recently, Klebsiella pneumoniae has become a health care concern due to its production of extended-spectrum beta-lactamase (ESBL) and its resistance to carbapenem. In Turkey, systematic meta-analyses investigating antibiotic resistance in K. pneumoniae are lacking. Methods: Consequently, we performed a systematic review of the literature followed by a meta-analysis to investigate antibiotic resistance in K. pneumoniae. This study was designed and conducted in accordance with the PRISMA guidelines. We identi ﬁ ed observational studies published from 2000 to 2015 using the various search engines. In total, 2,225 articles were published during this study period, but we only included 25 in our meta-analysis because of eligibility criteria. Results: We observed a signi ﬁ cant increase in antibiotic resistance (>40%) to the following antibiotics: cefazolin, amoxicillin-clavulanic acid, cefuroxime, cefepime, ceftriaxone, and ceftazidime. Unfortunately, the majority of these antibiotics were commonly prescribed for the treatment of K. pneumoniae infections. The rate of bacterial ESBL production has been steadily increasing and in this study was calculated at 39.66% ± 12.46%. In this study, we observed >30% resistance to cipro ﬂ oxacin. Furthermore, the rates of resistance to imipenem and meropenem were 5.1% and 3.4%, respectively. Conclusion: The data obtained from this study may be bene ﬁ cial for prescribing appropriate antibiotics and preventing their unnecessary use. The frequent checks of resistance rates with new detailed report may suggest to the development of National Antibiotic Stewardship.


Introduction
After the extended-spectrum beta-lactamase (ESBL) enzymes were shown, the number of infections caused by ESBL-producing organisms in hospitalized patients increased remarkably [1]. In the following process, the phrase multidrug resistance (MDR) started to be used commonly and MDR gram-negative bacilli (MDR-GNB) caused by nosocomial outbreaks were reported [2]. The most common view held about MDR is resistance to at least three antibiotics among cephalosporin (ceftazidime or cefepime only), aminoglycoside, fl uoroquinolone, carbapenem and piperacillin groups [3]. It is stated in the literature that one of the biggest factors of outbreaks caused by MDR-GNB is Klebsiella pneumoniae. In K.
pneumoniae, MDR develops due to modifi cation in the outer membrane permeability, activation of effl ux system and AmpC beta-lactamase, ESBL and carbapenemase enzymes [4].
It is stated in the studies that resistance to antibiotics especially the ones used in nosocomial outbreaks caused by K. pneumoniae has increased gradually and that antibiotics that are effective on isolates in different regions have become limited [5]. Moreover, it is emphasized that patients treated in intensive care units are at risk in terms of MDR K. pneumoniae infections and this situation is associated with the frequent use of carbapenems in intensive care units [6,7]. For, imipenem (IPM) and meropenem (MEM) are the antibiotics that are most commonly used for patients infected with K. pneumoniae in intensive care units [8].
Another problem that needs to be focused on is the colonization of MDR K. pneumoniae isolates. For, it is emphasized that the colonization time of resistant isolates may be longer than a year and it is stated that resistance problem can be experienced more intensely in the near future [9]. More importantly, there are no data regarding Turkey in the study of K. pneumoniae antibiotic resistance covering a number of countries conducted in 2015 by the World Health Organization [10]. "Klebsiella pneumoniae antibiotic resistance" and "Klebisella pneumoniae antibiotic resistance Turkey".

Qualitative analysis of the studies
Our meta-analysis study was prepared based on quinolone resistance within the framework of PRISMA statement guidelines [12]. The methodological evaluation of research was done with the criteria are shown in Table 1. However, the quality of the studies was not regarded as an exclusion criterion.
Qualitative examination of the studies included in the metaanalysis was subjected to scoring with a checklist designed with a critical evaluation by two independent reviewers. They were assessed in terms of the age groups of the patients, scope of the studies, identifi cation methods of strains, methods for determination of antibiotic activity, number of strains, the time period of conduction of the study and healthcare provider steps are shown in [Table 1].

Eligibility criteria
The eligibility criteria for the meta-analysis were as follows: it had to be a scientifi c study in Turkish or English; it had to examine the state of resistance/susceptibility at least 30 K. pneumoniae isolates based on NCCLS CLSI or EUCAST criteria between 2000 and 2015 in Turkey and the data had to be consistent. On the other hand, the multicenter studies that were made before 2000, whose full texts could not be accessed, that provided data for Klebsiella spp., that provided data for that Enterobacteriaceae family without making distinction between species, that assessed fewer than 30 isolates, whose ESBL results were not reported and that did not have detailed descriptions were excluded.

Data analysis
The data were divided into three groups as 2000-2004, 2005-2009 and 2010-2015; they were examined by two independent investigators; disputes between writers were decided by unanimous votes for eligibility criteria. The names of the fi rst researchers of the articles, their regions, the places where the samples were taken from, the total number of isolates, the methods used in the study, the properties of the patient populations, especially antibiotic resistance data were collected; and tables were formed. In the tables, the states of antibiotic resistance were shown in numbers; thus, all of the studies were enabled to be assessed by a common unit. The resistance data reported in the articles were converted to digital resistance data by ratio-proportion method. The antibiotic resistance or susceptibility data of K. pneumoniae isolates were analyzed by taking CLSI M100-S25 and M100-S-20-U into consideration. Besides, notifi cations related to resistance were grouped based on National Antibiotic Stewardship (NAS) [13].

Statistical analysis
In the meta-analysis, "forest plot" analysis was performed by using the program Comprehensive Meta-Analysis (CMA) (Biostat, USA). The study was assessed through Cochran Q test to fi nd out whether studies showed similar effects or not. The homogeneity of the data was examined by the value I 2 . The value I 2 ≥50 was accepted as limit value for homogeneity. The changes of K. pneumoniae isolates in the state of antibiotic resistance over the years were assessed statistically by using the program SPSS 20.0 version (SPSS, Inc., USA) through One Way ANOVA method. In the calculations, the value of p≤0.05 was expressed as the difference at signifi cant level statistically.

Results
During the literature review conducted based on specifi ed criteria, a total of 2225 articles between the years 2000 and 2015 that could potentially be used related to our topic were determined. 25 scientifi c studies having the eligibility criteria were included in the meta-analysis. In the qualitative examinations, studies were graded between 8 and 15. The qualitative average score of the studies was calculated as 11.24 ±1.98 are shown in Table 2.
In the calculations made with the ESBL data presented in the study, the rate was determined as 39.66% ±12.46. In the "forest plot" analysis made in 95% confi dence intervals (CI) with CMA, the ESBL rate that can be encountered at any time based on the fi xed effects model [14], was found to be 37.9% (95% CI, 37.9 to 40.8%) and the ESBL rate that can be encountered at any time based on the random effects model was found to be 39.8% (95% CI, 32.8 to 47.2%) (Q = 579.8, P <0.0001, I 2 = 95.5) are shown in Table 2. In the assessment made within the framework of NAS, all isolates were reported to be resistant to ampicillin (AM), which is among group A antibiotics to be tested primarily for K. pneumoniae. The average resistance rates to other antibiotics in this group were found to be 58.79%±22.47 and 27.93%±17.68 for cefazolin (CZ) and gentamicin (GM) respectively.
The average rates of resistance for amikacin (AK), amoxicillin-clavulanate (AMX), piperacillin-tazobactam (PIPTAZ), cefuroxime (CXM), cefepime (FEP), cefoxitin (FOX) and ceftriaxone (CRO), which are in Group B, were calculated as 7.59%±14. 25 Calculation was not made for ampicillin-sulbactam (SAM) because suffi cient data were not presented in the study for it. It was observed that a rate of 49,59%±13,07 was reported for ceftazidime (CAZ), which is in group C. No notifi cation was made for the antibiotics in group U in any study.
In the "forest plot" analysis made in 95% CI with CMA for IPM resistance, the rate that can be encountered at any time based on the fi xed effects model was found as 5.1% (95% CI; 3,9-6,7%) and the rate that can be encountered at any time based on the random effects model was found as 1.8% (95% CI, 0,9-3,7%) (Q =103,7, P < 0.0001, I 2 =79,8). In the "forest plot" analysis made in 95% CI with CMA for MEM resistance, the rate that can be encountered at any time based on the random effects model was found as 1.5% (95% CI, 0.6-3.4%) (Q=32.3, P <0.0001, I 2 =69.5).
In the analysis made based on years, an increase in the rates of resistance to all antibiotics except AK was determined. This was not found to be statistically signifi cant (p=0.094).
Change in resistance over the years could not be calculated for FOX and CRO since there were not suffi cient data about them.
Changes in the resistance rates based on years are shown in It was determined that the CIP resistance passed the 30% band  reported recently in many countries [42]. In multicenter studies, the rate of ESBL was found to be 48.7% in K. pneumoniae isolates [43]. In our meta-analysis, the rate of ESBL determined in the last 5 year period is 49.08%; as determined as factors in the analysis the rate of ESBL that can be encountered at any time determined as a factor in K. pneumoniae isolates was found to be 37.9% based on the fi xed effects model in our "forest plot" analysis. Under these circumstances, we think it is necessary to add the notifi cation of ESBL positivity to NAS statements so that the success of antibiotic treatment based on the personal experience of clinicians will increase, the ESBL rate will be brought under control and awareness will be raised.