Correct sepsis classification–A must for antimicrobial stewardship: A longitudinal observational study

Abstract

BACKGROUND

Sepsis is a critical medical condition, and poses a substantial global health burden, with significant morbidity, mortality, and economic costs, particularly pronounced in low- and middle-income countries. Effective management of sepsis relies on early recognition and appropriate intervention, underscoring the importance of accurate classification to guide treatment decisions. The correct diagnosis will lead to effective antimicrobial stewardship practices.

AIM

To assess the distribution of sepsis categories and the use of empirical antibiotics classified by the World Health Organization (WHO) Access, Watch, and Reserve (AWaRe) system in a tertiary care hospital in Northern India and to correlate antibiotic usage with sepsis classifications.

METHODS

This longitudinal observational study in the Department of General Medicine, in a tertiary care hospital in Northern India, from 2023 to 2024, aimed to assess the use of empirical antibiotics classified by the WHO AWaRe system. The study also aimed to correlate antibiotic usage. Patients were categorized into sepsis classes (Asepsis, Possible Sepsis, Probable Sepsis, Confirmed Sepsis) and followed until discharge or Day-28. Descriptive and inferential statistical analyses were employed to assess sepsis categories and empirical antibiotic usage classified by the WHO AWaRe system.

RESULTS

A total of 1867 patients admitted with suspected sepsis were screened, with 230 meeting the inclusion criteria. Among the study cohort (mean age 40.70 ± 14.49 years, 50.9% female), initial sepsis classification predominantly included probable sepsis (51.3%) and possible sepsis (35.7%), evolving to asepsis (57.8%) upon final classification, but all received antibiotics. Empirical antibiotic use showed a predominance of Watch group antibiotics (72.2%), with ceftriaxone and piperacillin-tazobactam being the most commonly prescribed; however, no statistical association could be established among the different classes of sepsis with the AWaRe groups.

CONCLUSION

Accurate sepsis classification is pivotal for clinical decision-making, optimizing antibiotic use, and combating antimicrobial resistance. The majority of the asepsis category was labelled as probable or possible sepsis and given antibiotics at initial hospitalization. The high reliance on Watch group antibiotics in empirical therapy signals a need for enhanced diagnostic strategies to refine treatment initiation, potentially reducing unnecessary antibiotic exposure. Future efforts should focus on establishing sepsis classification checklists as in this study and promoting adherence to antimicrobial stewardship principles to mitigate the global threat of antimicrobial resistance.

Key Words: Antibiotic overuse; Antimicrobial resistance; AWaRe classification; Empirical therapy; New sepsis definition

Core Tip: Accurate sepsis classification is pivotal for clinical decision-making, optimizing antibiotic use, and combating antimicrobial resistance. The majority of the asepsis category was labelled as probable or possible sepsis and given antibiotics at initial hospitalization. The high reliance on Watch group antibiotics in empirical therapy signals a need for enhanced diagnostic strategies to refine treatment initiation, potentially reducing unnecessary antibiotic exposure.

INTRODUCTION

Sepsis, a life-threatening condition caused by a dysregulated host response to infection[1], remains a leading cause of global morbidity and mortality, affecting over 49 million people annually and causing approximately 11 million deaths[1,2]. It is an extremely heterogeneous entity, both at the clinical and molecular level, can have various subtypes and there are even conditions which can mimic sepsis, i.e., sepsis mimics[3]. The clinical variability of sepsis, ranging from mild organ dysfunction to multi-organ failure, complicates treatment and underscores the need for rapid diagnosis and intervention. However, the growing prevalence of antimicrobial resistance (AMR) significantly hinders effective treatment[4].

As per The Global Burden of Disease study, it was estimated that there were around 48.9 million sepsis cases worldwide in the year 2017, with around 11 million deaths related to sepsis in the same year, which accounts for around 19.7% of all global deaths[5]. The majority of the sepsis-related burden is found in low- and middle-income countries in the world. In many developing countries, such as India, with a population of more than 1.4 billion people, the data regarding prevalence, apart from other epidemiological data, are missing and poorly understood, despite high rates of mortality related to sepsis. In 2017, the estimated number of cases of sepsis in India was around 11.3 million, and among them, 2.9 million deaths were noted[6]. To determine the latest epidemiological data regarding sepsis, a study named “Sepsis in India Prevalence Study” is ongoing.

There is also a growing incidence of AMR around the globe, which is emerging as a public health concern, and has led to the emergence of many multidrug resistance microbes, also called “superbugs”. In 2017, the World Health Organization (WHO) Expert Committee on Selection and Use of Essential Medicine developed the Access, Watch, and Reserve (AWaRe) classification of antibiotics as a tool to support antibiotic stewardship at various levels[7]. The antibiotics were categorized into 3 groups- Access, Watch and Reserve on the basis of their impact on AMR. This action is a step forward in reducing the burden of AMR by following the targets set by the WHO. To understand the importance of this, it should be noted that in 2019, it was estimated that AMR was responsible for around 1.3 million deaths worldwide[8]. Inappropriate use of antibiotics in humans is a major well-known and proven factor for AMR, in addition to other causes[9]. The misuse of antibiotics, including using them unnecessarily or selecting the incorrect antibiotic at the wrong dose, for the wrong duration, and by the wrong method, is a prevalent issue affecting 30% to 50% of all antibiotic prescriptions[10,11]. The decision to initiate empiric antibiotics must be based on some parameters and a ‘wait and watch’ strategy can be followed in some cases[12]. Hence, to quantify antibiotic use using the AWaRe classification, we also need an appropriate sepsis classification/definition. Both classifications are complementary to each other in righteousness.

This study was carried out at a tertiary care teaching hospital in northern India with the aim of assessing the distribution of various sepsis categories (asepsis, possible sepsis, probable sepsis, and confirmed sepsis) in admitted patients and the use of empirical antibiotics as per the WHO AWaRe classification. The study also aimed to highlight implications for antimicrobial stewardship by examining the use of AWaRe class antibiotics and their correlation with sepsis classifications.

MATERIALS AND METHODS

Study design

This longitudinal study was designed to assess the distribution of various sepsis categories in a real-world clinical setting from admission to discharge and to observe the AWaRe classification.

Study setting

The study was conducted at a tertiary care teaching hospital in northern India. Data were collected from the Department of General Medicine from January 1, 2023 to December 31, 2024. The study was approved by the Institute Ethics Committee, ensuring patient confidentiality, data protection, and adherence to ethical standards in research involving human participants.

Participants

Participants included adult patients aged 18 years or older admitted to the Department of General Medicine with suspected sepsis (need for antibiotics, evidence of infection, organ dysfunction not explained by non-infective causes, or improvement following antibiotic treatment) during the study period. Patients diagnosed with an alternative condition within 5 days of admission were excluded, including patients with incomplete data.

Sample size

The study employed universal sampling due to the absence of prior reference studies for precise sample size calculation. All eligible patients during the study period were included.

Data collection

Patients without evidence of dysregulated host response (step-1) were subjected to step-2 for evaluation of risk factors for infection (step-2) and patients without dysregulated host response and no evident risk factors for infection were categorized into the asepsis group. Patients with evidence of dysregulated host response (step-1) were directly subjected to step-3 for sepsis classification; however, risk factors for infection were also evaluated (Table 1).

Table 1 Three-step model used for sepsis classification.

Step	Description
Step 1: Evidence of dysregulated host response	Assessed using the National Early Warning Score-2 ≥ 6
Step 2: Risk factors for infection	Evaluated based on the presence of risk factors such as chronic illnesses, malnutrition, unhygienic living conditions, immunosuppressive states, age, trauma, structural diseases, recent surgery, travel history, animal bites, and previous hospitalizations
Step 3: Evidence of infection	Determined through:
3 (A) Clinical evidence	Syndromic diagnosis including pyelonephritis, infective endocarditis, intra-abdominal infections, skin and soft tissue infections, meningitis, cerebrospinal fluid shunt infections, catheter-related infections, osteomyelitis, abscesses, and pneumonia
3 (B) Supportive/suggestive evidence	Imaging (X-ray, ultrasonography, computed tomography, magnetic resonance imaging, positron emission tomography) and biomarkers (blood, urine, other fluids)
3 (C) Confirmatory evidence	Direct visualization, endoscopic evidence, microscopy and culture growth, PCR/gene detection, and immunological methods
Interpretation	Sepsis categories/classification
(1) Step-1 = negative; (2) Step-1 = positive with step-2 and 3 = negative	Asepsis
Step-1, 2 and 3 (A) = positive	Possible sepsis
Step-1, 2 and 3 (B) = positive	Probable sepsis
Step-1, 2 and 3 (C) = positive	Confirmed sepsis

Step description

Step 1: Evidence of dysregulated host response.

Step 2: Risk factors for infection.

Evaluated based on the presence of risk factors such as chronic illnesses, malnutrition, unhygienic living conditions, immunosuppressive states, age, trauma, structural diseases, recent surgery, travel history, animal bites, and previous hospitalizations.

Step 3: Evidence of infection. Determined through: 3 (A) Clinical evidence: Syndromic diagnosis including pyelonephritis, infective endocarditis, intra-abdominal infections, skin and soft tissue infections, meningitis, cerebrospinal fluid shunt infections, catheter-related infections, osteomyelitis, abscesses, and pneumonia; 3 (B) Supportive/suggestive evidence: Imaging (X-ray, ultrasonography, computed tomography, magnetic resonance imaging, positron emission tomography) and biomarkers (blood, urine, other fluids); and 3 (C) Confirmatory evidence: Direct visualization, endoscopic evidence, microscopy and culture growth, PCR/gene detection, and immunological methods.

Interpretation

Sepsis categories/classification: Asepsis: (1) Step-1 = negative; Or (2) Step-1 = positive with step-2 and 3 = negative.

Possible sepsis: Step-1, 2 and 3 (A) = positive.

Probable sepsis: Step-1, 2 and 3 (B) = positive.

Confirmed sepsis: Step-1, 2 and 3 (C) = positive.

Data were collected from patient records and medical charts using RedCap software (AIIMS Rishikesh version) and Microsoft Excel.

Statistical analysis

In this longitudinal study, statistical analysis was performed and the data were entered into an MS Excel sheet and RedCap software (AIIMS version). The data were summarized using descriptive and inferential statistics. Continuous variables were presented as mean ± SD. Categorical variables were presented as frequencies and percentages. A two-sided significance level of 0.05 was used for all statistical tests. Inferential statistics were performed using the χ² test. Results with a P value < 0.05 were considered statistically significant. All statistical analyses were conducted using the statistical software SPSS-25.

RESULTS

A total of 1867 patients were screened, and a cohort of 230 patients was included in the analysis. The results revealed that the mean age (in years) was 40.70 ± 14.49 years, and 113 (49.1%) were male and the remainder were females. Age was further subcategorized into 3 major groups, with 49.13% of patients aged 18-40 years, 40.87% aged 41-60 years, and the remainder aged > 60 years. Patients were classified into different sepsis categories based on a novel sepsis classification system, as shown in Table 1. There was a significant shift in sepsis classification from the day of hospital admission to discharge (Figure 1).

Figure 1  Graph showing the percentage of patients in the initial and final sepsis category during hospitalization.

Empirical antibiotic usage was studied and classified as per the WHO AWaRe classification, as shown in Figure 2, which shows that the major antibiotic usage was from the Watch group. Individual antibiotic usage was also estimated, as shown in Figure 3, revealing the most frequently used antibiotics were ceftriaxone, piperacillin-tazobactam, vancomycin, azithromycin, meropenem, doxycycline, levofloxacin, and linezolid, followed by others. Figure 4 shows the relationship between the antibiotics used per AWaRe classification and the different sepsis categories at different time points. These commonly used antibiotics were then assessed for their usage as per the sepsis classification as shown in Figure 5. No statistical association (P > 0.05) was found among the different sepsis classes with the AWaRe groups or even different types of antibiotics when χ² tests were performed.

Figure 2  Pie chart showing the distribution of antibiotics as per the AWaRe classification.

Figure 3  Graph showing the proportion of various antibiotics used.

Figure 4 Graphs showing usage of AWaRe antibiotics in different sepsis categories. A: During initial time period; B: During final hospitalization time period.

Figure 5 Graphs showing common antibiotics used in different sepsis categories. A: During initial time period; B: During final hospitalization time period.

DISCUSSION

This single-center observational study at a tertiary care teaching hospital aimed to classify patients into different sepsis categories with the help of a novel approach and to study their antibiotics usage[13]. Initial sepsis classification was dominated by probable and possible sepsis groups, which were finally changed to the majority being in the asepsis group over the duration of the hospital stay. The patients with suspected sepsis were prescribed antibiotics based on the treating clinician's decision and Indian Council of Medical Research Treatment guidelines for antimicrobial use in common syndromes[14]. Antibiotics used were classified as per the WHO AWaRe classification, showing major usage from the Watch group (72.2%), followed by the Access (20.6%) and Reserve (7.2%) groups. Commonly prescribed empirical antibiotics were ceftriaxone, piperacillin-tazobactam, vancomycin, azithromycin, meropenem, doxycycline, levofloxacin and linezolid, whose utilization was approximately ≥ 5%. These commonly used antibiotics were then studied in terms of their usage and corresponding sepsis category, revealing shocking results. It was noted that antibiotics were commonly prescribed to asepsis patients on initial encounters at healthcare facilities. However, initiation of antibiotics in suspected sepsis patients is a risky situation, and delay can be associated with adverse outcomes and should be taken into account. The treating clinician should decide following local or standard antibiotic guidelines, and any structured approach as in this study is always productive.

The antibiotic usage per the AWaRe classification was studied; however, due to lack of available sepsis classification criteria/model, direct comparison could not be conducted. However, antibiotic prescribing patterns were studied by Negi et al[15] at the same institute, revealing 57.61% of antimicrobials in the Access category, 38.27% in the Watch and 4.11% in the Reserve category, which is significantly different to our study results. The study by Negi et al included the whole institute, hence, eccentric inference can be found in a specific department. The overzealous use of antibiotics from the Watch category poses a risk of disruption of the local microbiome or microecology, which maintains homeostasis and promotes the co-existence of multiple flora, which in turn leads to the emergence of drug resistant organisms and repeated infection, making it difficult to treat and increases patient morbidity and mortality.

Sepsis has evolved in its definition and understanding over time to its latest definition in Sepsis-3 guidelines reflecting a deficit in its proper understanding and diagnosis, which was addressed in this study using a novel approach to classify sepsis (1). Using this approach (Table 1), patients were classified as shown in Figure 1, demonstrating the dynamic nature of sepsis[13,16]. Various novel tools such as antibiotic bundle care (ABC-bundle) can improve the delivery of appropriate antibiotic therapy and antimicrobial stewardship[17]. AMR is exacerbated by the misuse and overuse of antibiotics, accelerating the development of resistant bacterial strains. A study by Ventola[18] highlighted the role of broad-spectrum antibiotic misuse in promoting multidrug-resistant infections, which are associated with poorer outcomes in septic patients[18]. The WHO has recognized AMR as a top global health threat, predicting that drug-resistant infections could cause 10 million deaths annually by 2050 without immediate intervention[19]. To address this and for optimization of antibiotic use, the WHO introduced a new classification system AWaRe in 2017, categorizing antimicrobials into three groups—AWaRe—based on their spectrum, the anticipated risk of resistance development, toxicity risk, and clinical utility, which was revised twice, in 2019 and 2021, respectively[20-22]. In this study, we also utilized the WHO AWaRe classification for the classification of used antibiotics, which revealed that the majority of patients received antibiotics from the Watch group (Figure 2), which is not in line with the WHO’s proposed target of using at least 60% of all consumed antibiotics to be from the “Access” group for a country[23]. Antibiotics utilized from different groups were also evaluated, revealing that a large portion of patients received antibiotics from the Watch group based on initial sepsis categorization, which was later classified as asepsis (Figure 4), highlighting the improper utilization of antibiotics contributing to AMR, considered a global pandemic. The AWaRe system of classification, introduced by WHO in 2017 as a part of its antimicrobial stewardship, is considered a milestone step in the fight against increasing AMR, as it was more objective in nature and a user-friendly tool for better organization of antibiotics[24,25]. Some intensive care unit (ICU) settings have better antibiotic consumption than others, as evaluated by various studies with many parameters[26,27]. There is evidence of irrational antibiotic prescriptions in hospitals, which may be due to inadequate available methods for the diagnosis of sepsis[28]. For antimicrobial stewardship, the most important thing is to make the right diagnosis and the right use of antibiotics for its management. The discontinuation of prophylactic antibiotics, giving antibiotics courses for the appropriate duration and their timely de-escalation are essential to improve antibiotic practices in ICU settings to meet international recommendations[29]. This study highlights the importance of the correct sepsis diagnosis, which will be followed by the appropriate treatment in the form of antibiotics, as many patients have received antibiotics when they were wrongly classified as sepsis, and later classified as asepsis, showing misuse rather than overuse, of antibiotics leading to ongoing AMR. The lack of availability of a gold standard test for sepsis, associated with very high mortality and morbidity, highlights the need for an appropriate diagnostic model/algorithm, guiding clinicians and treating physicians for better utilization of resources[30]. The misuse of antibiotics is a widespread issue, affecting 30% to 50% of all prescriptions. This includes instances where antibiotics are used unnecessarily, the incorrect antibiotic is chosen, or the dosage, duration, and method of administration are incorrect[10,11]. Molecular testing methods such as Biofire (a polymerase chain reaction-based test), when incorporated into conventional testing methods, are associated with improved clinical outcomes and effective antimicrobial stewardship programs and procalcitonin for reduction of antibiotic use[31,32].

Many studies across the globe have shown that 20%–50% of all prescribed antimicrobials are either unnecessary or inappropriate[33,34]. From the studies conducted in China and Bangladesh, the antimicrobials selected to treat proven bacterial infections were deemed inappropriate in 63% and 50%, respectively[35,36]. Antibiotics utilized as per the WHO AWaRe classification in this study had nearly similar utilization patterns as observed previously from the same institute[28,29], necessitating the urgent need for the correct sepsis diagnosis, and making the right selection and usage of antibiotics in different groups. Setting up a multidisciplinary team such as an endocarditis team and incorporating its suggestions and recommendations in treating and diagnosing patients is associated with better clinical outcomes, halts misuse of antibiotics and helps in antimicrobial stewardship[37,38]. In this study, we noted the undue usage of antibiotics in patients not rightly classified as sepsis, such as ceftriaxone, piperacillin-tazobactam, and vancomycin, who were later classified as asepsis, and may not have required antibiotics in the first place (Figure 5). This needs to be addressed with utmost importance to curb the upcoming global pandemic of AMR, and this novel approach could be helpful; with the initial correct sepsis diagnosis, the appropriate antibiotic becomes an easy task, except in some circumstances.

Sepsis still does not have a proper definition, is evolving, and diagnostic tests pose a challenge for doctors globally in terms of the correct initial sepsis diagnosis amounting to misdiagnosis, irrational usage of antibiotics, increasing morbidity and mortality of patients and increasing financial burden on the healthcare system with improper utilization of available resources, especially in developing countries, all contributing to increasing AMR. The modern world requires the correct sepsis diagnosis on initial contact with healthcare personnel or treating doctors, which, in turn, allows for better implementation of the 4Ds of antimicrobial stewardship practices in this study[39].

As with most research studies, this study has some limitations. These included the following: this was a single-center and single department observational study, limiting its generalizability to a broader population with different characteristics, for example, head and neck surgery patients, using a novel method to classify sepsis, not using routinely available methodology; unavailability of prior reference studies, making it much more difficult and challenging, and lastly external validity of the study, all of which needs to be considered when interpreting the study results. Other potential confounders include differences in clinical judgement, patient demographics, and local antibiograms, which should also be kept in mind before extrapolating the results. However, these limitations can be overcome with future relevant studies.

CONCLUSION

The correct diagnosis of sepsis and its classification are crucial steps in antimicrobial stewardship. High reliance on antibiotics can be overcome by using a structured approach to diagnose sepsis, giving more confidence to treating doctors in the right decision-making, ultimately improving patient outcomes and morbidity and mortality associated with sepsis. Global targets of antimicrobial utilization can be achieved with the correct sepsis diagnosis. Overutilization of antimicrobials from the Watch group of the AWaRe classification is concerning, especially in asepsis, and control of this situation can limit the emergence of resistant microorganisms. However, its validation involving diverse patient groups, including surgical fields and healthcare settings, may be needed before generalizing to the global population.

ACKNOWLEDGEMENTS

We acknowledge the contributions of our mentors and colleagues whose insights and guidance significantly enriched this research. Our heartfelt thanks go to the study participants for their cooperation and contribution. Finally, we recognize the role of anonymous reviewers and previous researchers whose work laid the foundation for this study. Their collective efforts were instrumental in the successful completion of this work.