Antimicrobial resistance patterns of WHO priority pathogens isolated in children from a tertiary care hospital in Southern India: A descriptive and time trend analysis

There is global consensus that Antimicrobial Resistance (AMR) poses an unprecedented challenge to modern medicine as we know it today; and the lack of new antibiotics in the pipeline is compounding the threat to contain emerging drug-resistant infections. In 2017, the World Health Organization (WHO) has articulated a priority pathogens list (PPL) to provide strategic direction to research and development of new anti-microbials. Anti-microbial resistance patterns of selected ‘drug-bug’ combinations based on the WHO-PPL in one tertiary health care facility in India are explored in this paper. The PPL denes priority of pathogens based on resistance to specic antimicrobials such as carbapenems, 3rd generation cephalosporins, vancomycin, methicillin, penicillins, or uoroquinolones. Pathogens other than those in the WHO PPL were excluded from the analysis. Time trend graphs were prepared for some key pathogens.


Background
There is global consensus that Antimicrobial Resistance (AMR) poses an unprecedented challenge to modern medicine as we know it today; and the lack of new antibiotics in the pipeline is compounding the threat to contain emerging drug-resistant infections. In 2017, the World Health Organization (WHO) has articulated a priority pathogens list (PPL) to provide strategic direction to research and development of new anti-microbials. Anti-microbial resistance patterns of selected 'drug-bug' combinations based on the WHO-PPL in one tertiary health care facility in India are explored in this paper.

Methods
Culture reports of laboratory specimens, collected between 1st January 2014 and 31st October 2019 from paediatric patients in a tertiary care hospital in India, were retrospectively extracted. The antimicrobial susceptibility patterns for selected antimicrobials based on the WHO-PPL are analysed and reported.

Results
Of 12,256 culture specimens screened, 2,335 (19%) showed culture positivity; of which 1,556 were organisms from the WHO-PPL. E. coli was the most common organism isolated (37%) followed by Staphylococcus aureus (16%). Total 72% of E. coli were extended-spectrum beta-lactamases producers, 55% of Enterobacteriaceae were resistant to 3rd generation cephalosporins, and 53% of Staphylococcus aureus were Methicillin resistant. Time-trend analysis of the data showed continued high resistance to carbapenem in E coli, Klebsiella pneumoniae and Enterobacter cloacae.

Conclusions
The AMR trends and prevalence patterns are likely to be different, across various local settings, than as de ned at the national level or the WHO-PPL. This difference needs to be recognised in decision and policy making. It is critical, that the evidence used at national and global levels, have reasonable geographical and population representation through standardised and more granular AMR surveillance, in order to improve the effectiveness of the overall national AMR response.

Background
Antimicrobial resistance (AMR) has been recognised as a major threat to global health. According to the World Health Organization (WHO), microorganisms mutate when exposed to antimicrobials resulting in AMR, which consequently render medicines ineffective and infections to persist in the body, increasing the risk of spread to others 1 . Because of AMR, our ability to treat common pathogens becomes challenging, resulting in increased duration of illness, and rise in the number of complications and deaths. AMR impacts the healthcare system by increasing the duration of hospitalization and treatment, and the overall costs of health care. By 2050, an estimated 10 million deaths are projected to occur due to AMR 2 , while another study projected AMR to cost the global economy US$100 trillion, in the same period 3 .
In 2015, the 68th World Health Assembly endorsed the Global Action Plan on Antimicrobial Resistance to tackle this global challenge 4 . This action plan has ve strategic actions, focusing on: (1)improving awareness and understanding of AMR; (2) strengthening AMR surveillance; (3) reducing incidence of infections; (4) optimizing antimicrobial use; and (5) developing the economic case for AMR control. To support the Global Action Plan, WHO has developed a priority pathogens list (PPL), through a consultative process in 2016-17 5 . The prioritization process involved multi-criteria decision analysis (MCDA) which used information from multiple sources in a systematic manner, and included: disease mortality, transmissibility, treatability, preventability in health care settings, health care burden, preventability in community settings, etc. Tuberculosis as a disease was given the highest priority for research and development of new antimicrobials. Twelve families of drug resistant bacteria, posing the greatest threat to human health, were categorized as critical, high and medium priority organisms, in terms of their resistance to speci c antimicrobials. Although this categorization was intended to prioritize and stimulate research and development of new antibiotics for speci c drug-resistant, it also makes a call for the prevention of infection and the rational use of antibiotics in both humans and animals.
India is considered to be the highest consumer of antibiotics for human in the world 6 . The AMR situation in India has raised grave public health concern 7 and an action plan for its control is crucial. 8,9 . Given its importance for human health, the Government of India has developed a National Action Plan on However, not many hospitals outside this network in India track AMR patterns among these pathogens. It is critical to understand the AMR patterns in tertiary care hospitals so that appropriate evidence is generated for decision making at various levels.
This study explores the AMR patterns for WHO priority pathogens identi ed in the Pediatrics Department of a tertiary care hospital in Mangalore, South India. The study site -the Department of Pediatrics, Yenepoya Medical College Hospital --refers around 2,000 culture tests of various clinical specimens annually, from outpatient and hospitalized cases. The results of the culture tests are mainly availed for treatment purposes, but are not systematically analyzed on a routine basis which could provide further evidence and guidance for the development of antimicrobial resistance control policy at the hospital and elsewhere. We conducted a retrospective review of laboratory culture reports to pro le antimicrobial susceptibility pattern of the WHO priority pathogens identi ed in clinical isolates from children visiting, or admitted to, the Paediatrics Department.

Methods
Administrative permission to access laboratory culture records through the computer backbone system was obtained from Yenepoya Medical College. Retrospective culture reports between 1st January 2014 to 31st October 2019 from various clinical specimens were extracted from the computer backbone system. The sources and types of specimens collected for culture included blood, urine, stool, pus, cerebrospinal uid (CSF), sputum and any other bodily uids or clinical specimens such as catheter, umbilical, and central line tips. Culture reports of all paediatric cases were included, irrespective of the location of sample collection, namely: outpatients, inpatient wards, and neonatal and paediatric intensive care units (ICUs). Tests for antimicrobial susceptibility patterns were performed and the results uploaded to the computer backbone system by Microbiology Department. The Kirby-Bauer disk-diffusion method and/or by BD phoenix automated system were used for performing antimicrobial susceptibility patterns and reported according to standard (Clinical Laboratory Standards Institute -CLSI) guidelines.
The culture access numbers, of specimens for which any positive results were reported for any bacterial pathogens, were listed. The antibiogram (i.e., which antibiotics were tested, and which were susceptible or resistant) was extracted for all culture isolates. Culture access numbers were used to link culture source and the date of sample collection to the antibiograms. The laboratory reports indicating contamination were excluded at data entry. The reports containing duplicate or repeat samples, from same source and subject, were also excluded.
The data entry and analysis were performed on Microsoft Excel. The data from the computer backbone was entered directly on a master Excel sheet followed by removal of duplicates. De-identi ed data were organised by specimen types (such as blood, urine etc.) in chronological order. Number of bacterial pathogens isolated was summated by their species (such as E coli) and resistance patterns. The pathogens were classi ed into three main groups (Critical Priority, High Priority and Medium Priority) for selected antimicrobials, based on WHO priority pathogens list (Fig. 1). The PPL de nes priority of pathogens based on resistance to speci c antimicrobials such as carbapenems, 3rd generation cephalosporins, vancomycin, methicillin, penicillins, or uoroquinolones. Pathogens other than those in the WHO PPL were excluded from the analysis. Time trend graphs were prepared for some key pathogens.

Ethical issues
This study did not involve human subjects directly. An approval from the scienti c and ethics committees of Yenepoya Medical College was obtained for this study. To maintain con dentiality, no identi able information such as names, addresses, or phone numbers of subjects were collected. The data set, once nalised, was delinked from culture access numbers before analysis, to retain con dentiality.

Results
A total of 12,256 culture specimens were screened for bacteriological results, of which 2,335 (19%) showed culture positivity. Of these, 1,556 were from the set of WHO PPL organisms. The largest number of bacterial isolation was seen in urine specimens (50%) followed by blood (16%) ( Table 1). E. coli was the most common organism isolated (576), followed by Staphylococcus aureus (252).       (Fig. 2).
Similarly, E coli, Klebsiella pneumoniae continue to show trends of high ESBL rates (Fig. 3). Among the WHO 'High priority' pathogens, Staphylococcus aureus continues to show high methicillin-resistance in recent years (Fig. 4).
Three of the WHO 'High priority' pathogens namely, Helicobacter pylori (clarithromycin resistant), Campylobacter species ( uoroquinolone-resistant) and Neisseria gonorrhoeae (3rd generation cephalosporin-resistant, uoroquinolone-resistant), were not detected in our study specimens. Similarly, one of the WHO 'Medium priority' pathogen, Haemophilus in uenzae (ampicillin-resistant), was not observed in this study.

Discussion
The 2017 guidance document of WHO indicated the highest carbapenem resistance worldwide in Acinetobacter baumannii (91%) and Pseudomonas aeruginosa (82%), which is one of the reasons for classifying them as Critical Priority 5 pathogens. The same study reported > 50% carbapenem resistance in Acinetobacter baumannii, and 31-50% carbapenem resistance in Pseudomonas aeruginosa in the Indian sub-continent. Early results from the surveillance data from up to 22 ICMR-AMRSN sites in India showed around 80% carbapenem resistance in Acinetobacter baumannii and around 30% in Pseudomonas aeruginosa 11 . However, another study in children from Mumbai, India, identi ed only 15% of Pseudomonas aeruginosa are resistant to carbapenem 12 which is comparable to our study. This observation of low AMR in our study is based on less than 60 specimens in one tertiary hospital, and need not be considered representative of the situation in India. On the other hand, the variable incidence across study sites indicate carbapenem-resistant Acinetobacter baumannii and Pseudomonas aeruginosa, may perhaps be not that high across India.
The WHO report 5 identi ed high carbapenem resistance in E. coli (55%), Klebsiella (70%) and Enterobacter 'spp' (59%), which is comparable to the ndings in this study. The ICMR-AMRSN data also showed similar carbapenem resistance prevalence in Klebsiella pneumoniae (40%-50%) but a signi cantly lower level in E.coli (15-25%). The ESBL trends in E. coli and Klebsiella 'spp'(70-80%) as well as methicillin-resistance in Staphylococcus aureus (53%), were high in this study and comparable to the WHO report. However, due to the low sample size of Salmonella spp, Shigella spp and Streptococcus pneumoniae, it may not be appropriate to compare the results from this study to others.
Several studies on AMR have been published in India in recent years. However, the data on AMR patterns in systematically collected clinical specimens from multiple sources is limited. Nevertheless, it shows high variability across study sites and settings. An observational study from a tertiary care hospital in Puducherry showed 12.5% of isolated organisms exhibited resistance to one or more drugs, but did not focus the results on WHO PPL pathogens 13 . Another retrospective ve year follow-up study in a tertiary care hospital in North India showed increasing trends of AMR in urinary tract infection-causing isolates 14 . This study reported prevalence of ESBL increased from 22% in 2009 to 33% in 2014, which is a lot lower than our study. Increasing trends of AMR was observed among gram negative samples collected across seven hospitals in India over a four-year period, but the reported carbapenem resistance prevalence in Klebsiella 'spp' (39%) and E. coli (12%) were lower than our study 15 . Another observational study among intensive care unit (ICU) patients in a tertiary care hospital in Delhi showed 80% of Klebsiella 'spp' were resistant to beta-lactam group of antimicrobials, which is similar to our study 16 . A retrospective data collected for one year in the ICU of a hospital in Southern India found that 40.6% of Staphylococcus aureus were Methicillin-resistant (MRSA) and 11.9% of Enterococci were Vancomycinresistant which is comparable to the current study 17 . Among Enterobacteriaceae isolated from a paediatric tertiary care hospital in Mumbai, 24% were extended spectrum beta-lactamase (ESBL) producers and 27% were carbapenem-resistant isolates showing a lower resistance level than our study 12 .
As our study has deployed retrospective data collection based on computerised laboratory records, it has several limitations, but which may have been partially compensated due to the large sample size and standardised quality checks within the hospital. The sample collection, microbiological analysis, and report update on computerised records were done on a routine basis alongside the provision of healthcare services. Although the health facility maintains good quality of clinical and laboratory services along with proper documentation, one cannot ensure that the quality checks in retrospective data is fully compatible with the highest quality standards of a well-conducted prospective study. Although the sample collection, microbiological analysis, and report updates uses standard procedures, it has likely been conducted by different people over the 5 year-period, which may have had inter-personnel variations on the quality of laboratory procedures.
There were no standard inclusion criteria for sample collection; as it was generally left to the discretion of the treating physician. Furthermore, culture methodology and practices may have changed over time during the (period under study due to various operational reasons. Occasional shortfall of laboratory reagents and other essential supplies for conducting tests may have resulted in the omission of some tests during certain periods, resulting in some bias, the direction of which is di cult to assess. Some of the WHO PPL pathogens were not included in this study as culture specimens were collected from children and which do not generally include genitourinary swabs or gastric biopsy specimens suitable for the culture of Neisseria gonorrhoeae or Helicobacter pylori. Similarly, Campylobacter 'spp' was also not identi ed in the specimens either because of low incidence or because samples collected may not be best suited for its isolation. However, the large sample size was an important strength of this study.

Conclusion
Among the WHO PPL pathogens, E. coli, Klebsiella species, Enterobacteriaceae and Staphylococcus aureus (methicillin-resistant) seem to have high AMR patterns in our study site. On the other hand, AMR patterns for Acinetobacter baumanni, Pseudomonas aeruginosa and Staphylococcus aureus (vancomycin resistant) seem to be lower than the WHO global estimates. This study highlights the wide variability of AMR prevalence patterns across sites within countries, leave alone across countries. National and global AMR trends, using surveillance in tertiary hospitals as the basis for trend analysis, may not accurately re ect the trends and prevalence patterns within local milieus. It is imperative for policy decisions to factor in geographical and population representativeness within AMR surveillance networks so that national or global action plans are implemented based on a reasonable evidence base if not a robust one. While AMR policies, action plans and control activities should be aligned with global and national trends, they should be adaptive to the need and variability of local situations. It is essential that large health facilities carefully monitor and review emerging AMR patterns and trends periodically so that they can prioritise, plan, and implement health facility level polices and guidelines for the optimal use of antimicrobials.

Declarations
Ethics approval and consent to participate Described under manuscript Consent for publication The list of organisms and antimicrobial resistance patterns included in the analysis based on World Health Organization priority pathogens list (PPL) Figure 3 Third generation cephalosporin-resistant trends (extended-spectrum beta-lactamases (ESBL) producers) among selected WHO Critical priority pathogens