Trends in antibiotic resistance among major bacterial pathogens isolated from blood cultures tested at a large private laboratory network in India, 2008–2014☆

Highlights • This study examined long-term trends in antibiotic resistance on a national scale in India.• Colistin-resistant Klebsiella pneumoniae and Escherichia coli strains have emerged in India.• In 2014, the prevalence of carbapenem-resistant E. coli was11.5%, the highest reported to date globally.


Introduction
Antibiotic resistance (ABR) is a major public health problem worldwide and is associated with adverse health and economic consequences. 1 In India, a combination of mostly single-center studies and a few multicenter laboratory-based studies have shown a high prevalence of antibiotic resistance among common bacterial pathogens recovered from community-and hospitalacquired infections. [2][3][4][5][6][7] However, there have been no long-term studies on trends in ABR on a national scale in India.
Antimicrobial resistance surveillance on a national scale is critical because it provides information on the extent of established resistance rates, as well as emerging patterns of resistance. Understanding how resistance is changing is important for (1) establishing prescribing guidelines, (2) determining investment in new therapies, and (3) improving the targeting of campaigns to reduce antimicrobial resistance. It also provides a baseline for future analysis and comparison with other countries.
Rising rates of ABR in India are a significant concern because of the high burden of bacterial diseases 8 and the poor health system infrastructure, which relies on antibiotics in place of adequate vaccination coverage and other public health measures. In this study, data from a large private laboratory network were used to examine the resistance patterns of the organisms most commonly associated with bacteremia in patients across India for the period January 2008 to December 2014.

Methods
This was a retrospective study of patient blood cultures collected over a 7-year period (2008)(2009)(2010)(2011)(2012)(2013)(2014). Data on the microorganisms identified and their antimicrobial susceptibility were obtained from the SRL Limited laboratory network (Mumbai, India). The network includes 5700 collection centers spread across 26 of 29 states and two of seven Union Territories (UT). A collection center is a field site from where samples are collected based on physician orders. The collection centers include private hospitals (tertiary care, secondary care), diagnostic laboratories, and home collection agencies. Culture specimens collected are transported to the nearest of four reference laboratories located in four regions of the country ( Figure 1) for isolation, organism identification, and antimicrobial susceptibility testing. The BACTEC 9050 (Becton Dickinson Microbiology Systems, Sparks, MD, USA) automated system was used to process blood cultures at all four reference laboratories.
Data were retrieved electronically from the actual patient reports. The following information was obtained: (1) final blood culture result (positive growth or no growth); (2) organism identified if the culture was positive; (3) interpreted susceptibility results for tested antimicrobials (susceptible, intermediate resistance, or resistant); (4) patient identifier and demographic information (age, sex); (5) collection center information (name and geographical location); and (6) date of specimen collection. Organism identification and antimicrobial susceptibility testing were performed using the broth microdilution methodology (MicroScan panels, Siemens, Sacramento, CA, USA) in all reference laboratories. Categorical result interpretations (susceptible, intermediate, and resistant) were based on up-to-date Clinical and Laboratory Standards Institute (CLSI) criteria at the time of testing. 9 All culture-positive samples were considered without further interpretation of the results regarding clinical relevance.
The analysis considered all blood culture tests reported between January 1, 2008, and December 31, 2014. To avoid bias from duplicate cultures, data were filtered to retain only the first isolate from a patient. Identified microorganisms were stratified by year, age (<1, 1-17, 18-49, 50-64, and 65+ years), sex, and state or territory. Intermediate susceptible isolates were grouped with resistant isolates, as is now standard practice in the literature. 10 Antimicrobial susceptibility results for major bacterial pathogens to clinically important antimicrobial agents were examined. The organisms examined were coagulase-negative staphylococci (CoNS), Salmonella Typhi and Paratyphi A, Escherichia coli, Klebsiella pneumoniae, Acinetobacter species, Staphylococcus aureus, Pseudomonas aeruginosa, Enterococcus faecium, and Enterococcus faecalis. Resistance was defined at the antibiotic class level using data from at least one of the several agents within the same class. For third-generation cephalosporins, susceptibility results were reported for ceftriaxone, cefotaxime, or ceftazidime for Enterobacteriaceae. For all organisms, carbapenem resistance was defined as intermediate resistance or resistance to meropenem or imipenem. Fluoroquinolone resistance for all organisms was defined as intermediate resistance or resistance to ciprofloxacin or levofloxacin, except for Salmonella species, where nalidixic acid was considered. Aminoglycoside resistance for all organisms was defined as intermediate resistance or resistance to gentamicin, tobramycin, or amikacin. As minimum inhibitory concentration (MIC) values were not available in the database, the resistance percentages for pathogen-antibiotic combinations in the years prior to the change in MIC breakpoints are not displayed.

Statistical analysis
Unadjusted resistance rates were calculated for each year as the number of resistant isolates as a proportion of total isolates tested. The Chi-square test (Cochran-Armitage) for linear trend was used to test the significance of annual trends. A p-value of 0.05 was considered statistically significant. All statistical analyses were performed using Stata software, version 12 (StataCorp, TX, USA). This study was approved by the Institutional Ethics Committee of the Public Health Foundation of India.

Number and distribution of laboratories and cultures
A total of 135 268 blood cultures from unique persons were identified in the database for the period January 2008 to December 2014. Of these, 18 695 (14%) cultures were positive. Overall, the data came from 1820 unique collection centers spread across 425 cities and 27 states (including two UTs). The median number of blood cultures obtained from one collection center was 3 (interquartile range 1-99). Of the 1820 collection centers, 1409 (77.4%) contributed less than 11 blood cultures over 8 years ( Figure 2). Positive blood cultures were identified at 696 of the centers spread across 185 cities, 25 states, and two UTs. The geographic distribution of collection centers contributing positive culture data is illustrated in Figure 1. Of the positive cultures, 79% were contributed by 20 collection centers that are tertiary care hospitals located in seven major cities (Figure 1), and 92.1% of positive cultures were from one UT (Delhi, 27.4%) and five states:  (Table 1). Approximately 62% of the total and positive cultures were from males and about 35% of the total cultures and 30% of the positive cultures were from persons aged 18-49 years (Table 1). Data on the distribution of total and positive cultures by year, age, sex, and state are given in the Supplementary Material (Tables S1 and S2, respectively).
CoNS were the most common bacteria isolated in all years except 2008 and 2011 (Table 2). Among Salmonella species, 67% were Salmonella Typhi and 25% were Salmonella Paratyphi A. The database showed 66 polymicrobial cultures (cultures with more than one organism isolated). The three most common pathogens affecting infants (<1 year) were CoNS, K. pneumoniae, and Candida species. However, among pediatric individuals (aged 1-17 years) and young adults (aged 18-49 years), Salmonella species were the most common pathogens isolated. The three most common pathogens affecting older patients (>50 years) were CoNS, E. coli, and K. pneumoniae. More detailed information on the distribution of bacterial pathogens by age, sex, and state is given in the Supplementary Material (Table S4).
Among Gram-positive organisms, the average proportion of methicillin resistance and linezolid resistance in CoNS for all years was 73% (n = 2488) and 0.4% (n = 3579), respectively. Overall, three vancomycin-resistant CoNS isolates were observed during the study period (Table 3). Among S. aureus, the average proportion of methicillin-resistant S. aureus (MRSA) for all years was 44.2% (n = 608), with no significant change during the study period. Overall, two vancomycin-resistant S. aureus (VRSA), nine vancomycin-intermediate S. aureus (VISA), and 17 linezolidresistant S. aureus (LRSA) isolates were reported during the study period. The average proportion of vancomycin resistance in E. faecium and E. faecalis for all years was 16.6% (n = 235) and 2.4% (n = 169), respectively, with no significant change during the study period.

Discussion
This study examined the ABR prevalence of bloodstream isolates obtained from patients across India. It greatly expands on prior studies of antimicrobial resistance in India, 2-7 providing detailed long-term descriptions of the profile and ABR patterns of organisms isolated in blood cultures from various regions of India.  Overall, it was observed that Gram-negative organisms are frequently isolated in blood cultures. Salmonella species associated with enteric fever were the most frequently isolated Gramnegative organisms, followed by E. coli and Klebsiella species. Although CoNS represented the organisms most frequently isolated, this is likely because CoNS are a common contaminant in clinical specimens. 11 This study provides evidence that enteric fever is the major cause of bacteremia primarily affecting children and young adults, justifying the need for improvements in sanitation and indicating the urgent need for a vaccine conferring long-term immunity. The low percentage of polymicrobial cultures (66 of 18 695 positive blood cultures) may indicate a low percentage of surgical patients in this database. High resistance rates to both frontline antibiotics and those of last-resort were observed for all Gram-negative organisms isolated from blood cultures, but resistance was not static over the period of the study. For Salmonella Typhi and Salmonella Paratyphi A, nalidixic acid resistance rates remained extremely high (>95%), consistent with other studies performed in India, 3,12 while resistance rates to older antibiotics, ampicillin and trimethoprim-sulfamethoxazole, decreased over time. Similar findings were observed in single-center studies in India, with increasing susceptibility to older antibiotics like ampicillin, chloramphenicol, and trimethoprim-sulfamethoxazole. 13,14 These changes are likely due to the replacement of these drugs as an empiric treatment option for enteric fever with newer drugs, such as the fluoroquinolones. These findings suggest that fluoroquinolones can no longer be considered an empiric treatment option for suspected enteric fever; rather, physicians may be able to use older drugs again or third-generation cephalosporins (resistance to cephalosporins was minimal). Unfortunately, third-generation cephalosporin-resistant Salmonella Typhi and Salmonella Paratyphi A strains have emerged; these constituted about 0.8% of the isolates in this study, consistent with other studies in India. 3,12 The resistance rates of E. coli to fluoroquinolones and thirdgeneration cephalosporins remained high throughout the study period; both were above 80% in 2014. High resistance rates were also observed for other antibiotics frequently used as empiric treatment options for E. coli, such as aminoglycosides (61.1%) and piperacillin-tazobactam (37.7%). Most alarming was the high carbapenem resistance, which was greater than 10%, a rate that is significantly higher than in other countries from which surveillance data are available. Of 41 countries with data from 2013 or 2014, only 12 reported detecting carbapenem resistance in E. coli and only three recorded a rate greater than 3%: Bulgaria (3.5%), Turkey (5%), and Vietnam (9%). 15 Carbapenem resistance rates for K. pneumoniae also increased significantly over the study period, reaching approximately 60% by the end of the study, which again is higher than any other country except Greece, which had a similar percentage of carbapenem resistance in 2013 (60%). 16 The rising      carbapenem resistance among E. coli and K. pneumoniae isolates is a cause for concern, given the frequency of infections caused by these and the associated mortality, which for carbapenemresistant K. pneumoniae bacteremia is estimated to be about 50%. 17 The primary drug for treating carbapenem-resistant strains of K. pneumoniae and E. coli is colistin; however, worryingly, colistin resistance has already emerged. Colistin-resistant strains of P. aeruginosa and Acinetobacter species have also emerged. This is significant because these pathogens are intrinsically resistant to several antibiotics, leaving physicians with few options to treat infections. Among Gram-positive organisms, S. aureus and Enterococcus species were the most frequently isolated organisms after CoNS. Methicillin resistance in CoNS was very high at 73%, consistent with other studies in India; [18][19][20] however CoNS remain highly susceptible to vancomycin and linezolid. Similar to a multicenter study in India, 2 the proportion of S. aureus isolates that were resistant to methicillin was high in this study (42% in the multicenter study vs. 44% in this study); however, of more significance were the isolates that were resistant to vancomycin and linezolid. Although S. aureus remains highly susceptible to both drugs, 3% of the isolates were linezolid-resistant in 2014. Several other studies in India have reported similar frequencies of LRSA, consistent with this finding. [21][22][23] In the present study, two cases of VRSA were observed; however, vancomycin resistance in Enterococcus species was much higher, with 17% of the E. faecium isolates being vancomycinresistant. Efforts have been made in the USA to reduce the threat posed by VRSA through reporting standards for infections caused by these pathogens. 24 Similar efforts may be necessary in India, because genes conferring resistance, once they have evolved, can spread rapidly both within a country and around the world.
In this study, the resistance rates were also examined by age and sex. No significant differences were found (data not shown).
As with most large data surveillance studies, this study has limitations. First, although the analysis was confined to blood isolates, which likely portend infection, no clinical information was included. Second, information on variables differentiating community-acquired from healthcare-acquired infections was not available. However, enteric fever caused by Salmonella species (Salmonella Typhi and Paratyphi A) is a common cause of community-acquired bloodstream infection. 25 Thus, it is likely that most if not all Salmonella cultures identified had a community origin. Third, although the data are national in scale, they may not be nationally representative of all Indian states or types of healthcare facility. Finally, MIC values were not available to reinterpret the resistance percentage for years prior to the change in MIC breakpoints. However, for the majority of pathogen and antibiotic combinations (28 of 37 combinations), the MIC values did not change during the study period.
In conclusion, increased antibiotic use has long been directly linked to higher rates of antibiotic resistance. 26,27 With the highest volume of antibiotic sales in 2010, 28 it is not surprising that India has a simmering public health crisis related to antibiotic resistance. The increasing consumption of the two antibiotics of last-resort, carbapenems and polymyxins, between 2000 and 2010, portends a likely rise in the proportion of Gram-negative organisms resistant to these two antibiotic classes. As has been demonstrated before, resistance in India can spread rapidly to other parts of the world, 29 making these results important not just for India. These results also indicate the urgent need for new antibiotics against Gram-negative organisms, as well as the necessity of continued surveillance of resistance patterns, especially in the Gram-negative organisms. Finally, the implementation  of standard infection control practices and antimicrobial stewardship programs in healthcare facilities should be a priority.