|Year : 2019 | Volume
| Issue : 1 | Page : 4-9
Antimicrobial stewardship programme – from policies to practices: A survey of antimicrobial stewardship programme practices from 25 centres in India
Sanjay Bhattacharya1, Vinitha Mary Joy1, Gaurav Goel1, Swapna R Nath2, Shabina Santosh3, Kalpana George4, Ranganathan Iyer5, Kavita Raja6, A Uma7, Aarti Gupta8, Anitha Madhavan9, Arindam Chakraborty10, Barsha Sen11, Beena Philomina3, Dhruv K Mamtora12, KP Dinoop1, J Lancy13, Mahua Dasgupta14, Mannu R Jain15, Mohd Tajuddin16, Neethu Kishor4, Prithi Nair17, K Rejitha18, Rekha N Nair19, Sarada Devi20, TS Shailaja21, A Shilpa22, Shoba Kurian23, KV Suseela24, SG Sagila2, Syed Mustaq Ahmed25, Yogesh Kumar Gupta26
1 Department of Microbiology, Tata Medical Centre, Kolkata, West Bengal, India
2 Department of Microbiology, Regional Cancer Centre, Thiruvananthapuram, Kerala, India
3 Department of Microbiology, Government Medical College, Kozhikode, Kerala, India
4 Department of Microbiology, Government Medical College, Kollam, Kerala, India
5 Department of Microbiology, Global Hospital, Hyderabad, Telangana, India
6 Department of Microbiology, Sree Chitra Tirunal Institute for Medical Sciences and Technology, Thiruvananthapuram, Kerala, India
7 Department of Microbiology, SRM Medical College Hospital, Trichy, Tamil Nadu, India
8 Department of Microbiology, FMRI, Gurugram, Haryana, India
9 Department of Microbiology, Government Tirumala Devaswam Medical College, Alappuzha, Kerala, India
10 Department of Microbiology, Fortis Hospital, Kolkata, West Bengal, India
11 Department of Microbiology, Sri Aurobindo Seva Kendra, Kolkata, West Bengal, India
12 Department of Microbiology, S. L. Raheja Hospital, Mumbai, Maharashtra, India
13 Department of Microbiology, Government Medical College, Kochi, Kerala, India
14 Department of Microbiology, Bhagirathi Neotia Women and Child Care Center, New Town, Kolkata, West Bengal, India
15 Department of Microbiology, Surat Muncipal Institute of Medical Education and Research, Surat, Gujarat, India
16 Department of Microbiology, Delhi State Cancer Institute, New Delhi, India
17 Department of Microbiology, Government Medical College, Thrissur, Kerala, India
18 Department of Microbiology, Government Medical College, Manjeri, Kerala, India
19 Department of Microbiology, Fortis Hirananani Hospital, Navi Mumbai, Maharashtra, India
20 Department of Microbiology, Government Medical College, Thiruvananthapuram, Kerala, India
21 Department of Microbiology, PK comDas Institute of Medical Sciences, Ottappalam, Kerala, India
22 Department of Microbiology, Fortis Hospital, Bengaluru, Karnataka, India
23 Department of Microbiology, Government Medical College, Kottayam, Kerala, India
24 Department of Microbiology, Amala Institute of Medical Sciences, Thrissur, Kerala, India
25 Department of Microbiology, Muslim Educational Society Medical College, Perinthalmanna, Kerala, India
26 Department of Microbiology, Fortis Escorts Hospital, Jaipur, Rajasthan, India
|Date of Web Publication||12-Aug-2019|
Dr. Sanjay Bhattacharya
Department of Microbiology, Tata Medical Center, Kolkata, West Bengal
Source of Support: None, Conflict of Interest: None
|How to cite this article:|
Bhattacharya S, Joy VM, Goel G, Nath SR, Santosh S, George K, Iyer R, Raja K, Uma A, Gupta A, Madhavan A, Chakraborty A, Sen B, Philomina B, Mamtora DK, Dinoop K P, Lancy J, Dasgupta M, Jain MR, Tajuddin M, Kishor N, Nair P, Rejitha K, Nair RN, Devi S, Shailaja T S, Shilpa A, Kurian S, Suseela K V, Sagila S G, Ahmed SM, Gupta YK. Antimicrobial stewardship programme – from policies to practices: A survey of antimicrobial stewardship programme practices from 25 centres in India. J Acad Clin Microbiol 2019;21:4-9
|How to cite this URL:|
Bhattacharya S, Joy VM, Goel G, Nath SR, Santosh S, George K, Iyer R, Raja K, Uma A, Gupta A, Madhavan A, Chakraborty A, Sen B, Philomina B, Mamtora DK, Dinoop K P, Lancy J, Dasgupta M, Jain MR, Tajuddin M, Kishor N, Nair P, Rejitha K, Nair RN, Devi S, Shailaja T S, Shilpa A, Kurian S, Suseela K V, Sagila S G, Ahmed SM, Gupta YK. Antimicrobial stewardship programme – from policies to practices: A survey of antimicrobial stewardship programme practices from 25 centres in India. J Acad Clin Microbiol [serial online] 2019 [cited 2020 Jan 18];21:4-9. Available from: http://www.jacmjournal.org/text.asp?2019/21/1/4/264247
| Introduction|| |
The rise of antimicrobial resistance (AMR) in several groups of pathogens is a complex public health challenge. Infectious diseases (ID) are a major contribution to global mortality and AMR, an all-encompassing roadblock to its solution. One of the major contributing factors in India is the rampant inappropriate use of anti-microbial agents (AMA) with a recent surveillance study indicating 40% of patients in community on antibiotics and even higher rate of AMA abuse. It is our duty as health care professionals to curb this emerging threat, as the development and research into new AMA is a slow process and a shallow pool of opportunities at present. One of the strategies to counter this would be to initiate antibiotic stewardship programmes.
Anti-microbial stewardship programme (AMSP) is an important intervention to control antibiotic usage, promote rational prescribing, optimise clinical outcome and reduce the incidence of drug resistant infections. However, the practice of AMSP is relatively new in the Indian context and in most centres in the subcontinent; it is yet to develop as a routine practice in clinical settings due to lack of infrastructure.
| Materials and Methods|| |
An AMSP questionnaire [Figure 1] was developed by the editorial board of the Journal of Academy of Clinical Microbiologists (ACM), which was published and circulated via both hard copy and paperless format through the 2018 July–December issue of the journal. The questionnaire had two parts, of which the first part consisted of the description of the structure and process of AMSP, while the second part pertained to centre-specific antibiograms during the year 2018. The circulation of the questionnaire was reinforced by targeted communications to ACM members and their known associates and colleagues.
|Figure 1: Breakdown of survey respondents in terms of various anti-microbial stewardship programme systems|
Click here to view
The responses provided by the clinical microbiologists from multiple centres were collated in an excel format and subsequently analysed for cumulative results. Any ambiguous data from participating centres were clarified through E-mail and telephonic communications whenever necessary. The information was anonymous and confidential.
| Results|| |
A total of 25 centres responded to the invitation to participate in this AMSP survey. Of these, 11 (44%) were attached to government medical colleges and fourteen to private hospitals (56%). Of them, 8 (33%) centres had attained accreditations, with either National Accreditation Board of Hospitals and Healthcare Providers (NABH) or National Accreditation Board for Testing and Calibration Laboratories (NABL) status, while the rest had none. More than half of the participating centres (13/25) hailed from Kerala (53%), while West Bengal was represented by 16% (4/25), Maharashtra by 8% (2/25) and one each from Tamil Nadu, Karnataka, Gujarat, New Delhi, Rajasthan and Haryana states [Figure 2].
The bed strength of these hospitals ranged from 55 to 3695; with 14 (56%) having bed strength up to 500 and 11 (44%) had >500 beds (including many of the government hospitals). Only nine centres had a separate ID department with one or more ID consultants (66% of these institutions were privately affiliated). The total number of MD microbiologists from each centre ranged from 1 to 17 with the government centres having the maximum number (median 3).
A wide range was observed in the total number of blood cultures received per year by each centre (384–14495; median 3746) with the most numbers from tertiary care government centres. Similarly, the total positive blood cultures per year ranged from 21 to 619 (median 241). Each laboratory processed approximately 4–202 clinical specimens per day, of which blood cultures constituted 25%–50% with a positivity rate of 1–10 blood cultures/day. Automated blood culture systems were utilised in 84% of participating centres, of which 15 (60%) used only automation [Figure 3]. Of the nine centres (36%) which relied on non-automated methods, one centre outsourced their samples while the rest utilised an in-house broth for processing. It was noted that 16 (64%) of the involved centres used automated systems (namely VITEK 2 COMPACT) for bacterial identification and antimicrobial susceptibility testing (AST); however, six of them (24%) used automation for selected isolates (for example, isolates from intensive care unit [ICU] and/or blood cultures only). Ten centres (40%) had no access to any type of automated ID/AST system and utilised conventional means like disc diffusion and/or broth dilution for AST.
|Figure 3: Automation versus conventional means in contemporary microbiology laboratories|
Click here to view
Yeast identification was done in-house by all but one of the 25 centres, 8 (33%) centres using only automation. One of these eight centres confirmed the result by sequencing. Four centres (16%) outsourced their yeast isolates, of which three depended on reference centres for final confirmation of identification. Sixteen centres (64%) performed anti-fungal susceptibility for yeast with a majority of 13 (52%) depending on VITEK 2 COMPACT, and the rest using disc-diffusion technique. The remaining ten centres conducted no in-house testing, while three of them outsourced the isolates for sensitivity reports.
Sixteen centres (64%) used an information technology (IT) system for laboratory and hospital communications, with twelve of them having both hospital information system (HIS) and laboratory information system (LIS).
Anti-microbial stewardship programme practices
Fifteen centres (60%) reported the implementation of AMSP, with two centres continuing the programme without the insistence of administration. However, three centres admitted to having no AMSP even though administration had insisted on such a plan. Majority of the AMSPs (40%) were led by clinical microbiologists (6/15). Other leaders included clinical pharmacologist (one), ID physician (one), intensivist (one) and other physicians (six) [Figure 4].
|Figure 4: Breakdown of survey participants in terms of anti-microbial stewardship programme team leadership|
Click here to view
The number of AMSP meetings per year ranged from 1 to 12 in the participating centres, with seven centres (47%) having approximately four to six meetings/year. IT software was utilised by fourteen centres, of which 11 (%) used WHONET and others used Excel or Siemens HIS system. Monitoring of prescription and/or drug use was done in 14 (36%) centres either by a clinical pharmacologist (8/14) or clinical pharmacist (12/14) with six having both. Of these 14 centres, periodic audits of antibiotic use (including dose, route, infusion time and other parameters) and prescriptions were conducted in 92% (13/14), while feedback to the clinical and AMSP team was done by 12 centres (86%); ten of the fourteen centres (40%) performed all four functions (prescription audit, feedback to clinical team, feedback to AMSP team and antimicrobial consumption audit). Nine centres (36%) performed regular review of all antibiotic prescriptions as part of the systematic evaluation of on-going treatment plans, while 19 (76%) centres only followed up on those with positive blood cultures and 12 (48%) centres added ICU drug prescriptions to their review. Defined daily dose (DDD) monitoring was done in 9/25 (38%) centres, by the clinical microbiologist (three) or clinical pharmacologist (one) or pharmacist (five). Monitoring of drug consumption was reported from 14 centres (58%), with all centres monitoring Colistin, followed by Carbapenems (13), Tigecycline (11), Polymyxin B and Glycopeptides (10), Amphotericin B and Linezolid (9), Echinocandins (7) and Fosfomycin (6) in decreasing order of monitoring. Only one centre monitored use of Daptomycin. Frequency of monitoring varied between daily to annually, while five centres did random checking.
Antibiogram was released periodically by all centres, with annual circulation in six centres, half-yearly or quarterly among seven, and monthly from three centres, with 6/25 (24%) centres giving out antibiograms at random intervals.
Ward rounds were conducted regularly for monitoring of cases among 11 (44%) centres, while random visits were done by 12 (%) centres to see 'interesting' cases, ten centres reporting both reasons. Most commonly misused antibiotics reported from the participating centres in decreasing order of frequency: Carbapenems (10), Linezolid (9), Glycopeptides (8), Beta lactam + beta lactam inhibitors (6), Fluoroquinolones (4) and Polymyxins (2). Majority of misuse was reported in cases of surgical prophylaxis (11), followed by use of Carbapenems and Linezolid in community-acquired infections.
The antibiogram was analysed to report the prevalence of AMR in the major pathogenic bacteria from our country. Among patients with blood stream infections (BSI), prevalence of methicillin resistant Staphylococcus aureus ranged between 17% and 40% (interquartile range) with median 29%, while Vancomycin-Resistant Enterococci BSI was reported to be between 0% and 40%. Ampicillin resistance in Enterococcus spp. ranged between 28% and 80% (interquartile range) and median 63%, and high level aminoglycoside resistance was between 33% and 67% (interquartile range) with median 61%.
The prevalence of extended spectrum beta lactamase in Gram-negative bacilli isolated from BSI ranged from 50% to 87% (interquartile range), with 40%–80% seen in Acinetobacter spp., 19%–34% in Pseudomonas spp., 56%–78% in Klebsiella spp. and 50%–75% in Escherichia More Details coli. Carbapenem resistance of 10%–50% (interquartile range) with median of 30% was reported from all centres, with 30%–75% resistance seen in Acinetobacter spp., 19%–50% in Pseudomonas spp., 35%–65% in Klebsiella spp. and 16%–50% in E. coli. Colistin resistance was reported in Acinetobacter spp. (12%–38%) and Klebsiella spp. (9%–18%) and Fosfomycin resistance in Klebsiella spp. ranging between 9% and 12%. Tigecycline resistance was reported in E. coli (23%–37%), Klebsiella spp. (38%–50%) and Acinetobacter spp. (24%–45%).
Incongruous reports of Vancomycin and/or Teicoplanin resistance in S. aureus was reported from two centres without further clarification. Ampicillin sensitivity among Staphylococcus spp. isolates ranging between 25% and 57% was also reported from 15 centres. One centre also reported Enterococcus spp. resistance to Linezolid of 7.6%. These unusual resistances without reference laboratory confirmation have not been included in the results.
| Discussion|| |
In the present study, we found approximately one-third of the involved centres (9/25%–36%) have attained either NABL or NABH status, with only one centre being from the government sector. This calls for a renewed effort towards laboratories and hospitals across health care providers. Accreditation of NABL is for investigations rather than for the laboratory, whereas NABH accreditation is for the hospital. Such qualifications help to put in place optimal quality management systems, quality system procedures, standard operating procedures and external quality assurance systems. These help to streamline the structure and process that are necessary for AMSP, for example, institution-based antibiotic usage policies based on hospital antibiogram, which in turn is dependent on quality-assured antimicrobial susceptibility results.
Among the participating centres, about two-third (doubled from earlier - 30%) reported the use of IT systems to communicate laboratory information such as AST results and antibiograms. This is an encouraging development as the increasing use of IT, namely LIS and/or HIS leads to less post-analytic errors, enables more efficient transfer of patient information and facilitates better understanding of antibiograms and AMSP among clinicians and clinical information gathering for microbiologists.,
A significant factor in implementing AMSP across all areas of the hospital is the total bed strength and occupancy rate. In our study, eleven of the participating centres had bed strength >500 with the median number of MD-qualified microbiologists being only three. This skewed ratio provides a major challenge to conducting ward rounds as part of AMSP throughout the hospital even though our study shows that 68% of involved centres regularly follow up patients, mainly for reasons of personal interest. It is also noteworthy that all qualified microbiologists may not be adequately trained to take up AMSP leadership roles.
This is evident from our study where of the 15 centres who implemented AMSP (60%), only 6 were led by clinical microbiologists (40%). Others included the clinical pharmacologist, ID physician or intensivist who have previously been recognised as the most common element of the team [Figure 4].
A key role of the clinical microbiologist as part of AMSP team is to develop and circulate the hospital antibiogram to all stakeholders. From the current survey, we found that all of the involved centres participated in this aspect, which is a promising development from prior surveys (80%) as reported by Walia et al., with almost half of them publishing it annually or half-yearly. However, nine centres (36%) reported random or unclear timings in their antibiogram development. The above fact implies that further work to be done for the regulation and proper implementation of AMSP, and clinical microbiologists are likely to play a major role in fulfilling this objective.
The average number of AMSP meetings per year was about five, but these may only include operational meetings with major stakeholders rather than targeted training in antibiotic prescribing. The ID department is crucial for the efficient functioning of AMSP, diagnosis and treatment of difficult infections, and it continues to be a pressing need in our country with only 36% of participating centres having a separate ID consultant though only marginal improvement has been noted from earlier surveys.,
Other inadequately recognised members of the AMSP team include the clinical pharmacologist and pharmacist, who need to collaborate with faculty in other disciplines such as microbiology to achieve good outcomes for optimal patient care in the hospital setting., Our study revealed a low rate of involvement, with only 8 (32%) centres having a clinical pharmacologist in their AMSP team, while 12 (48%) having a clinical pharmacist.
The above specialists performed a myriad group of functions which are an essential supplement to AMSP education,,, namely prescription audits in fourteen of the involved centres (56%), prospective feedback in 12 (48%) and random review of antibiotic prescriptions in 19 centres (76%). While most centres only followed up only the positive blood culture patients, nine centres (36%) performed regular review of all antibiotic prescriptions and 12 (48%) conducted evaluations of ICU antibiotic use which showed a marked increase from previous surveys. Earlier studies have proven that such interventions have led to documented reduction in the use of third-generation cephalosporins and a stable rate of use of fluoroquinolones, along with decreasing incidence of antibiotic-associated diarrhoea.
The above study showed that only nine centres (36%) conducted regular monitoring of drug consumption by DDD calculations, with the clinical pharmacist being the responsible party in five of them which shows no significant change from prior reports. An intermittent monitoring of antibiotic use was conducted by 15 centres (60%) with Colistin 93% (14), Carbapenems 86% (13), Tigecycline 73% (11), Polymyxins and Glycopeptides 66% (10), Amphotericin B and Linezolid 60% (9), Echinocandins 46% (7) and Fosfomycin 40% (6) in the descending order of frequency in monitoring.
All participating centres reported inappropriate use of antibiotics (either indication or duration) in various situations, the most common being surgical prophylaxis and community-acquired infections; incidence of misuse being highest among Carbapenems (40%), Linezolid (36%) Glycopeptides (32%), Beta lactam + beta lactam inhibitors (24%), Fluoroquinolones (26%) and Polymyxins (8%). Private health care centres continue to demonstrate superior standards in AMSP implementation than government centres, which may be attributed to their higher levels of accreditation, illuminating the need for mandatory accreditation for government health care institutions [Table 1].
|Table 1: Anti-microbial stewardship programme features in government and private health care institutions (government health care institutions and private health care institutions)|
Click here to view
Blood cultures are perhaps the most important samples to help in developing the hospital antibiogram because the cultured bacteria or yeast (apart from skin commensals) are likely to represent true pathogens. In our study, it was found that the median number of blood cultures received by all participating centres was about 3746, implying an average of 10 per day. Given the fact that many hospitals are having bed strength above 500, this indicates a significant deficiency in sending blood cultures from eligible patients among participating centres. Therefore, there is a sharp need to increase awareness amongst clinicians, microbiologists and the hospital administration to facilitate better means of sending blood cultures from appropriate patients.
Automated blood culture systems (e.g. BacT/Alert/BACTEC) are becoming more common in the clinical microbiology laboratories [Figure 3]. However, the present study shows that the number of centres (9) having conventional means of blood culture are significant (36%), especially among government centres. Major challenges to maintain a fully automated system for blood cultures in a government hospital include high procurement and consumable costs. This obstacle may be overcome by better R&D investments in biomedical equipment and manufacturing processes within the country to develop a made-in-India automated system for use in our country. Approximately two-third of the participating centres had VITEK 2 COMPACT system for automated identification and AST; nonetheless, similar issues relating to expense is emphasised by concurrent use of manual methods for the same. Although manual methods like disc diffusion testing (DDT) may be used for AST, the increasing reliability on MIC-based interpretative guidelines for certain pathogens according to CLSI (e.g. Colistin in Pseudomonas aeruginosa and Acinetobacter spp., Oxacillin and Vancomycin in S. aureus) has made DDT redundant in some cases. Alternatives to DDT are E-test (Epsilometer test) and broth micro dilution (BMD); but these techniques are both labour- and cost-intensive in actual practice.
Robust identification of yeast is challenging, especially if the number of isolates is insufficient to justify investment in this sector. Automated means of yeast identification may be an easier alternative to manual means based on germ tube test and/or biochemical; but, cost is a major constraint seen in its use. Other non-conventional means include CHROMagar (standardised differential media) and API Auxacolor (single use strip for biochemicals); their use is limited by the lack of accuracy in the results.
Anti-fungal susceptibility testing (AFST) in yeasts has become very important in view of emerging candida infections, most importantly Candida auris. Many of the C. auris strains are developing increasing resistance not only to Fluconazole but also to Voriconazole, Amphotericin B and even to Caspofungin. The recommended method for AFST according to CLSI is BMD. However, this may not be feasible in most centres due to cost constraints, lack of expertise and inadequate quality control systems. In this scenario, it is vital to check and/or confirm the AFST results from reference labs like PGIMER, Chandigarh. As a single laboratory cannot cater to the needs of the entire country, this further implies the acute need to develop more mycology reference centres across the country.
Some centres have reported inconsistent resistant strains, namely decreased Vancomycin susceptibility in S. aureus and Ampicillin susceptible strains of Klebsiella and Pseudomonas spp. These instances reveal a critical want of understanding of the existing AMR patterns and must be addressed by continuing medical education and workshops periodically.
| Conclusion|| |
The current study is a multicentric attempt to cumulatively analyse the structure and process of AMSP within India. These data should hopefully be used by stakeholders and policy makers to make further investment to develop AMSP as a routine practice.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Kumari SI, Chandy SJ, Jeyaseelan L, Kumar R, Suresh S. Antimicrobial prescription patterns for common acute infections in some rural & urban health facilities of India. Indian J Med Res 2008;128:165-71.
] [Full text]
Walia K, Ohri VC, Mathai D; Antimicrobial Stewardship Programme of ICMR. Antimicrobial stewardship programme (AMSP) practices in India. Indian J Med Res 2015;142:130-8.
] [Full text]
MacDougall C, Polk RE. Antimicrobial stewardship programs in health care systems. Clin Microbiol Rev 2005;18:638-56.
Griffith M, Postelnick M, Scheetz M. Antimicrobial stewardship programs: Methods of operation and suggested outcomes. Expert Rev Anti Infect Ther 2012;10:63-73.
Centres for Disease Control and Prevention. Core elements of Hospital Antibiotic Stewardship Programs. U.S. Dept of Health and Human Services. Atlanta, Georgia: Centres for Disease Control and Prevention; 2014.
Laxminarayan R, Duse A, Wattal C, Zaidi AK, Wertheim HF, Sumpradit N, et al.
Antibiotic resistance-the need for global solutions. Lancet Infect Dis 2013;13:1057-98.
Rimawi RH, Mazer MA, Siraj DS, Gooch M, Cook PP. Impact of regular collaboration between infectious diseases and critical care practitioners on antimicrobial utilization and patient outcome. Crit Care Med 2013;41:2099-107.
Chandy SJ, Michael JS, Veeraraghavan B, Abraham OC, Bachhav SS, Kshirsagar NA. ICMR programme on antibiotic stewardship, prevention of infection & control (ASPIC). Indian J Med Res 2014;139:226-30.
] [Full text]
Carling P, Fung T, Killion A, Terrin N, Barza M. Favorable impact of a multidisciplinary antibiotic management program conducted during 7 years. Infect Control Hosp Epidemiol 2003;24:699-706.
Filice G, Drekonja D, Greer N, Butler M, Wagner B, MacDonald R, et al
. Antimicrobial Stewardship Programs in Inpatient Settings: A Systematic Review. Washington (DC): Department of Veterans Affairs (US); 2013.
Reller LB, Weinstein M, Jorgensen JH, Ferraro MJ. Antimicrobial susceptibility testing: A review of general principles and contemporary practices. Clin Infect Dis 2009;49:1749-55.
Posteraro B, Efremov L, Leoncini E, Amore R, Posteraro P, Ricciardi W, et al.
Are the conventional commercial yeast identification methods still helpful in the era of new clinical microbiology diagnostics? A meta-analysis of their accuracy. J Clin Microbiol 2015;53:2439-50.
Lockhart SR, Jackson BR, Vallabhaneni S, Ostrosky-Zeichner L, Pappas PG, Chiller T, et al.
Thinking beyond the common candida species: Need for species-level identification of candida due to the emergence of multidrug-resistant Candida auris
. J Clin Microbiol 2017;55:3324-7.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]