Low necessity, high overuse: revisiting empirical vancomycin practices for children in low MRSA prevalence settings

Abstract

Background

The overuse of vancomycin as empirical therapy ranged from 62% to 66% worldwide, while the prevalence of proven infections requiring vancomycin was only 11% to 27%. Evidence from Thailand where the prevalence of methicillin-resistant Staphylococcus aureus (MRSA) infection is low and remains limited. This study aimed to determine the incidence of confirmed infections requiring vancomycin as definitive therapy among children aged 1 month-18 years receiving vancomycin as empirical therapy, and to identify factors associated with vancomycin necessity.

Methods

A single-centre retrospective cohort study was performed, which included 532 children receiving at least one dose of vancomycin as empirical therapy at a university hospital during 2014–2024. Demographic data, provisional diagnosis, ICU admission, haemodynamic status, presence of central venous catheters, and microbiological results were collected. Factors associated with proven infections requiring vancomycin were determined by multivariable log-binomial regression.

Results

Only 29 of 510 (5.68%) patients who had empirical treatment with vancomycin truly needed vancomycin as definitive treatment. Provisional diagnosis of device-associated infection, especially cerebrospinal fluid shunt (RR 6.97, 95%CI 2.92–16.63), and intra-abdominal infection following liver transplantation (RR 4.95, 95%CI 1.97–12.48), were significantly associated with proven infections requiring vancomycin in multivariable analysis. Febrile neutropenia, haemodynamic instability, and the presence of central venous catheters were not significantly associated with vancomycin necessity in univariate analysis.

Conclusions

In a setting with a low prevalence of MRSA infection, most patients who received empirical vancomycin were subsequently found not to require it, highlighting the potential for more targeted empirical vancomycin use, such as patients with cerebrospinal fluid shunts or post-liver transplantation intra-abdominal infections.

Introduction

Vancomycin has been widely used to treat drug-resistant Gram-positive bacterial infections. For patients with suspected infections due to these organisms, while awaiting confirmation of the causative pathogens, empirical treatment with intravenous vancomycin may be indicated based on the clinical scenario. In patients with febrile neutropenia, vancomycin is recommended if there are any signs of inflamed exit site of central venous catheters (CVC), suspicion of skin and soft tissue infection, pneumonia, haemodynamic instability, or if there is a high incidence of methicillin-resistant Staphylococcus aureus (MRSA) in the local healthcare setting [1, 2]. In paediatric patients with suspected bacterial meningitis, vancomycin should be initiated as empirical treatment if Gram staining of cerebrospinal fluid (CSF) reveals Gram-positive diplococci, Streptococcus pneumoniae is detected by latex agglutination, patients have received a beta-lactam antibiotic within the past two months, or if there is haemodynamic instability [3].

Despite published guidelines on appropriate vancomycin use and the growing concern over vancomycin-resistant bacteria, particularly vancomycin-resistant enterococci (VRE) [4–6], overuse of vancomycin, especially for empirical treatment, is commonly observed in clinical practice worldwide. A 2007 study in Brazil by Junior et al. evaluated vancomycin prescribing practices in a university hospital compared to U.S. Centers for Disease Control and Prevention (CDC) guidelines and found that 65.7% of empirical vancomycin prescriptions were inappropriate. Examples of unjustified vancomycin use included administration solely because a patient was critically ill without other supporting evidence (51.3%) or prescribing vancomycin for fever in patients with a central venous catheter (CVC) without signs of catheter-related infection (6.8%) [7]. Similarly, a 2014 study in the United States by Libuit et al. assessed vancomycin use in febrile neutropenic patients and found that 62% of prescriptions did not adhere to the Infectious Diseases Society of America (IDSA) guidelines [8]. In Thailand, a 2006 study by Rattanaumpawan et al. at Siriraj Hospital explored vancomycin prescribing in paediatric and adult patients and found that 59.5% of vancomycin prescriptions as definitive treatment were inappropriate [9]. However, the study did not analyze the appropriateness of vancomycin use as an empirical treatment.

The prevalence of colonization by MRSA strains seemed to be higher in Africa compared to Asia and South America [10]. In North America, the prevalence of MRSA was reaching 40% of S. aureus isolates [11]. However, during 2020 and 2022, the incidence of MRSA, a primary indication for vancomycin use, accounted for only 8.3% to 9.6% of all S. aureus isolates in Thailand [12]. Given this relatively low MRSA incidence, empirical use of vancomycin in Thailand based solely on concerns about MRSA may be less justified.

Studies assessing the proportion of patients who initially received empirical vancomycin and later required definitive therapy show variations in each geographic region. Only 11.2% of patients in the United States [13] required vancomycin as targeted treatment, compared to 27.3% in South Korea [14] and 14.3% in Jordan [15]. In Thailand, no studies have been conducted regarding these issues. This study aimed to determine the incidence of paediatric patients empirically treated with vancomycin who truly required vancomycin as definitive therapy and to identify factors associated with situations that required vancomycin as definitive therapy. Knowing these factors would provide rational guidance for more appropriate prescribing of empirical vancomycin, ultimately reducing the incidence of vancomycin-resistant bacteria and minimizing vancomycin-related side effects.

Methods

Study design and participants

This retrospective cohort study was conducted at Ramathibodi Hospital, a 1,200-bed university hospital and a centre for stem cell and organ transplantation in Bangkok, Thailand. Paediatric patients aged 1 month to 18 years who were admitted during 2014–2024 and were empirically treated with vancomycin were retrospectively reviewed. Only patients with at least one clinical specimen submitted for culture and who received at least one dose of intravenous vancomycin as empirical treatment were included.

The exclusion criteria comprised patients who received vancomycin as a culture-driven treatment, defined as cases in which Gram-positive bacteria had been recovered from clinical specimens before starting the medication. Patients who received oral or intravitreous vancomycin and those whose culture results were incomplete or unavailable were excluded.

Data collection

The clinical data of 532 eligible paediatric patients from electronic medical records were collected. Only the most recent episode was included in the study for patients with multiple episodes of having empirical treatment with vancomycin. The inclusion and exclusion processes are shown in Fig. 1.

Fig. 1.

CONSORT diagram of eligible patients

Demographic data, underlying diseases, hospital units (general ward, intensive care unit), presence of a CVC, concurrent neutropenia, haemodynamic instability, provisional diagnosis, microbiological results, therapeutic drug monitoring, and adverse drug effects were systematically recorded.

Operational definitions

Community-acquired infections were defined as infections with onset and specimen collection within 48 h of hospitalization, without any permanent indwelling catheters or medical devices. Patients not meeting these criteria were classified as having hospital-acquired infections.

Immunosuppressive status included patients with hematologic malignancies or solid tumors who had received chemotherapy within the previous 3 months; recipients of hematopoietic stem cell or solid organ transplants receiving ongoing immunosuppressive therapy; patients with connective tissue diseases, vasculitis, or inflammatory arthritis receiving immunosuppressive agents; those with inborn errors of immunity; and patients with HIV infection.

Haemodynamic instability was defined as the presence of the following characteristics: weak pulse, capillary refill time ≥ 3 s, cold extremities, or oliguria (even with normal blood pressure), or the requirement for inotropes or vasopressors.

Device-associated infection referred to infection in a patient with at least one medical device (e.g., prosthetic heart valve, joint prosthesis, peritoneal dialysis catheter, or CNS shunt) and clinical suspicion of device involvement. Surgical site infection was defined as infection at the surgical wound within 7 days post-operation.

Acute kidney injury (AKI) was defined as an increase in serum creatinine by ≥ 0.3 mg/dl within 48 h, ≥ 1.5 times baseline within 7 days, or urine output < 0.5 ml/kg/hour [16].

Empirical vancomycin use referred to vancomycin prescribed based solely on clinical suspicion and individualized risk factors before microbiological results were available. Culture-driven treatment referred to vancomycin prescribed after microbiological confirmation of Gram-positive bacteria.

Vancomycin necessity was defined as infection ultimately requiring vancomycin as definitive therapy, confirmed by microbiological results (usually within 5 days). Such infections included [1, 17]:

• Severe skin infections or other organ infections caused by methicillin-resistant Staphylococcus aureus (MRSA).

• Infections caused by Streptococcus pneumoniae or Enterococcus spp. resistant to beta-lactams.

• Infections caused by Bacillus spp., Corynebacterium spp.

• Infections caused by Gram-positive bacteria in patients with severe beta-lactam allergy.

• Infections caused by coagulase-negative staphylococci (CoNS) susceptible to vancomycin (except S. lugdunensis) were deemed to require vancomycin as definitive therapy, regardless of oxacillin susceptibility, owing to high oxacillin resistance rates [18] and concerns regarding heteroresistance [19].

Episodes in which clinical specimens were not obtained directly from the infection site were excluded, as vancomycin necessity could not be assessed.

Statistical analysis

Data were analyzed using Stata version 17.0 (StataCorp, College Station, TX, USA). In univariate analysis, the chi-square or Fisher’s exact test was used to test whether each factor was associated with vancomycin necessity. Factors with P-value < 0.2 in univariate analysis were included in multivariable log-binomial regression.

Results

Among the 532 eligible patients, 22 patients could not be assessed for the necessity of vancomycin use as a definitive treatment due to the inability to obtain specimens directly from the suspected sites of infection for bacterial culture. Most of these patients had a provisional diagnosis of skin and soft tissue infection, such as non-purulent cellulitis. Therefore, only 510 patients (episodes) were included in the analyses (Fig. 1).

Among the 510 patients with culture results, the majority (359 patients, 71%) had no detected pathogens. For those with positive cultures, the most common isolated organisms were Gram-negative bacteria (73 patients, 14%), followed by Gram-positive bacteria (57 patients, 11%), fungi (15 patients, 3%) and multiple pathogens (6 patients, 1%). In this study, 19 patients were found to have S. aureus as the causative pathogen, and only one was identified as MRSA, accounting for 5.3% of all S. aureus infections.

Only 29 of 510 (5.7%) patients who had empirical treatment with vancomycin truly needed vancomycin as a definitive treatment. Among the 29 cases requiring vancomycin, the most commonly identified pathogen was coagulase-negative staphylococci (CoNS), found in 14 cases (48.3%), followed by Enterococcus spp. in 8 cases (27.6%) and Bacillus spp. in 5 cases (17.2%). Meanwhile, MRSA and Corynebacterium spp. were each detected only in 1 case, accounting for 3.4% each. Demographic data and clinical characteristics of all 510 patients were shown in Table 1.

Table 1.

Demographic data and clinical characteristics

Factors	All (N = 510)	Vancomycin necessity (N = 29)	No vancomycin necessity (N = 481)	P-value
Age (years)
- Median (IQR)	5 (10)	3 (10)	5 (10)	0.281
Sex, N (%)
- Male	269 (52.7)	13 (44.8)	256 (53.2)	0.379
Underlying disease(s), N (%)
- None - ≥1 underlying disease with immunocompetent status - Immunocompromised status	63 (12.4) 205 (40.2) 242 (47.4)	0 (0) 12 (41.4) 17 (58.6)	63 (13.1) 193 (40.1) 225 (46.8)	0.099
Drug allergy, N (%)
- None - Beta-lactam allergy - Allergy to other medications	410 (80.4) 20 (3.9) 80 (15.7)	22 (75.9) 1 (3.4) 6 (20.7)	388 (80.6) 19 (4.0) 74 (15.4)	0.664
Hospital units, N (%)
- General wards - Intensive care units	177 (34.7) 333 (65.3)	12 (41.4) 17 (58.6)	165 (34.3) 316 (65.7)	0.437
Presence of CVCs, N (%)
- None - Short-term catheters - Long-term catheters	167 (32.7) 269 (52.8) 74 (14.5)	15 (51.7) 11 (37.9) 3 (10.4)	152 (31.6) 258 (53.6) 71 (14.8)	0.078
Febrile neutropenia, N (%)
- Yes	75 (14.7)	5 (17.2)	70 (14.6)	0.599
Haemodynamic instability, N(%)
- Yes	202 (39.6)	8 (27.6)	194 (40.3)	0.173
Recent influenza/chickenpox,N (%)
- Yes	10 (2)	0 (0)	10 (2.1)	1.000
Type of infection, N (%)
- Community-acquired - Hospital-acquired	35 (6.9) 475 (93.1)	0 (0) 29 (100.0)	35 (7.3) 446 (92.7)	0.248

IQR: interquartile range; CVC: central venous catheter

When eligible patients were classified by provisional diagnosis, the three diagnoses with the highest incidence of vancomycin necessity were device-associated infection (8 out of 30, 26.7%), endocarditis (1 out of 7, 14.3%), and intra-abdominal infection (5 out of 47, 10.6%), respectively.

When factors commonly influencing physicians’ decisions to initiate empirical vancomycin were analyzed (Table 2), febrile neutropenia, haemodynamic instability, immune status, and hospital units (general wards or intensive care units) showed no statistically significant association with the vancomycin necessity as a definitive therapy in univariate analysis. Similarly, provisional diagnoses of skin and soft tissue infection, surgical site infection, intra-abdominal infection, and endocarditis were not significantly associated with the vancomycin necessity. In contrast, device-associated infection was the only provisional diagnosis demonstrating a statistically significant association with the vancomycin necessity (crude RR 6.10, 95% CI 2.95–12.60). Interestingly, the presence of CVCs was found in univariate analysis to be a protective factor, significantly associated with a lower vancomycin necessity (crude RR 0.45, 95% CI 0.22–0.92).

Table 2.

Univariate analysis of factors associated with vancomycin necessity

Risk factors	Vancomycin necessity (N = 29)	No vancomycin necessity (N = 481)	Crude RR (95% CI)	P-value
Febrile Neutropenia
- Yes - No	5 24	70 411	1.21 (0.48–3.07)	0.599
Haemodynamic instability
- Yes - No	8 21	194 287	0.58 (0.26–1.29)	0.173
Type of infection
- Hospital-acquired - Community-acquired	29 0	446 35	N/A	N/A
Immunocompromised
- Yes - No	17 12	225 256	1.57 (0.77–3.21)	0.215
Presence of CVCs
- Yes - No	14 15	329 152	0.45 (0.22–0.92)	0.025*
Hospital units
- Intensive care units - General wards	17 12	316 165	0.75 (0.37–1.54)	0.437
Provisional diagnosis Pneumonia
- Yes - No	0 29	53 428	N/A	N/A
Skin & soft tissue infection
- Yes - No	1 28	32 449	0.52 (0.07–3.68)	1.000
Device-associated infection
- Yes - No	8 21	22 459	6.10 (2.95–12.60)	< 0.001*
Surgical site infection
- Yes - No	2 27	18 463	1.82 (0.46–7.12)	0.317
CNS infection
- Yes - No	0 29	16 465	N/A	N/A
Intraabdominal infection
- Yes - No	5 24	42 439	2.05 (0.82–5.13)	0.173
Endocarditis
- Yes - No	1 28	6 475	2.57 (0.40-16.32)	0.338
Bone and joint infection
- Yes - No	0 29	2 479	N/A	N/A

RR: relative risk; CVC: central venous catheter; CNS: central nervous system; N/A: not applicable

*Statistically significant difference (P-value < 0.05)

In multivariable analysis (Table 3), the presence of CVCs and haemodynamic instability were not statistically associated with the vancomycin necessity. On the other hand, device-associated infection and intra-abdominal infection were found to be statistically significant factors associated with the vancomycin necessity (adjusted RR 5.88, 95%CI 2.49–13.95, and adjusted RR 2.87, 95%CI 1.11–7.45, respectively).

Table 3.

Multivariable log-binomial regression of factors associated with Vancomycin necessity

Risk factors	Crude RR (95% CI)	P-value	Adjusted RR (95% CI)	P-value
Presence of CVCs	0.45 (0.22–0.92)	0.025*	0.65 (0.30–1.43)	0.287
Device-associated infection	6.10 (2.95–12.60)	< 0.001*	5.88 (2.49–13.95)	< 0.001*
Haemodynamic instability	0.58 (0.26–1.29)	0.173	0.99 (0.41–2.27)	0.924
Intraabdominal infection	2.05 (0.82–5.13)	0.173	2.87 (1.11–7.45)	0.030*
CNS shunt infection	6.56 (3.19–13.47)	< 0.001*	6.97 (2.92–16.63)	< 0.001*
Intra-abdominal infection following liver transplantation	3.59 (1.48–8.69)	0.016*	4.95 (1.97–12.48)	0.001*

RR: risk ratio; CVC: central venous catheter; CNS: central nervous system

*Statistically significant difference (P-value < 0.05)

When specifically analyzing patients with a provisional diagnosis of device-associated infection who had complete culture results (30 cases), the majority of patients had CNS shunt infections (28 cases, 93.3%), while the remaining two cases were peritoneal dialysis shunt infection and prosthetic valve infection. When focusing on suspected CNS shunt infections, both univariate and multivariable analysis demonstrated that this provisional diagnosis was significantly associated with the vancomycin necessity (Table 3). In patients suspected of having a CNS shunt infection who required vancomycin, out of 8 cases, 7 (87.5%) were found to have coagulase-negative staphylococci (CoNS) as the causative pathogen and the remaining 1 case was caused by methicillin-resistant Staphylococcus aureus (MRSA).

When analyzing only patients with a provisional diagnosis of intra-abdominal infection requiring vancomycin (5 cases), all five patients had undergone liver transplantation and were suspected of having an intra-abdominal infection within the first 60 days following liver transplantation. Further analysis was conducted on post-liver transplant patients with a provisional diagnosis of intra-abdominal infection. Multivariable analysis demonstrated a statistically significant association with the vancomycin necessity (Table 3). Additionally, 4 out of 5 patients (80%) had Enterococcus spp. as the causative pathogen, including Enterococcus faecium and ampicillin-resistant Enterococcus faecalis.

In a subgroup analysis of patients with febrile neutropenia, the presence of CVCs, haemodynamic instability, and the combination of both were not associated with the vancomycin necessity. In a subgroup analysis of immunocompromised patients, these factors were not associated with the vancomycin necessity.

Among the 481 patients for whom vancomycin was not required as a definitive treatment, the majority (75%) had vancomycin discontinued within 72 h. However, 20% of them had vancomycin continued without a proper indication, while the other 5% remained on the drug for more than 72 h due to certain indications, such as awaiting investigation results. Regarding therapeutic drug monitoring (TDM), the majority (84%) of patients who received vancomycin for more than 48 h underwent TDM.

Most patients receiving vancomycin (89.5%) did not experience adverse effects. Among those who did, 7.7% developed acute kidney injury (AKI). However, none required renal replacement therapy and all of them fully recovered. Additionally, 2.6% experienced rash or red man syndrome, while 0.2% had anaphylaxis. In addition, 51.2% of patients who experienced AKI had received the drug for no more than 48 h.

Discussion

The incidence of patients who required vancomycin as definitive therapy in this study was only 5.7%, and the incidence of MRSA infection was only 5.3% among S. aureus infection. These findings can be explained by low prevalence of MRSA infections in Thailand, which ranges from 8% to 10% [12], significantly lower than in other regions like Europe or the United States, where MRSA can account for 25% to 50% of infections [11]. Therefore, in countries with low MRSA prevalence, if S. aureus is the sole suspected pathogen, routine empirical vancomycin for all cases may not be appropriate.

The study’s findings indicate that suspected CNS shunt infections significantly necessitate vancomycin as a targeted treatment. This is reasonable and aligns with several standard clinical practice guidelines, such as the 2017 IDSA clinical practice guideline for healthcare-associated ventriculitis and meningitis [20]. This guideline recommends anti-pseudomonal beta-lactam and vancomycin as the first-line treatment to cover Gram-positive bacteria such as CoNS and S. aureus. This aligns with the findings of this study, which identified CoNS as the causative pathogen in over 85% of cases. Therefore, despite the prevalence of MRSA infections in Thailand is low, vancomycin is necessary to cover CoNS until an alternative causative pathogen is identified.

According to this study, liver transplant recipients with suspected intra-abdominal infections within the first 60 days post-transplantation were found to have a significant association with the vancomycin necessity. Among these patients, 80% had infections caused by E. faecium or ampicillin-resistant E. faecalis. This finding is consistent with the 2010 guideline by the Surgical Infection Society and the IDSA on the diagnosis and management of complicated intra-abdominal infections in adults and children [21], which recommends empiric anti-enterococcal therapy in cases of suspected post-surgical intra-abdominal infection, in patients with prior exposure to cephalosporins or other antibiotics lacking enterococcal coverage, and in immunocompromised individuals. In Ramathibodi Hospital, all liver transplant recipients received ampicillin-sulbactam and/or a third-generation cephalosporin as pre-operative prophylaxis. This may explain the increased risk for infections caused by E. faecium and drug-resistant Enterococcus faecalis, necessitating subsequent vancomycin therapy.

This study found that febrile neutropenia was not associated with vancomycin necessity, both in univariate and multivariable analysis, supporting the 2010 guideline for the use of antimicrobial agents in neutropenic patients with cancer by the IDSA [22], which states that vancomycin should not be routinely administered empirically in patients with fever and neutropenia. When a subgroup analysis specifically for patients with febrile neutropenia was performed, having CVCs, haemodynamic instability, or both factors were not associated with the vancomycin necessity. This suggested that empirical vancomycin should not be routinely given simply because patients have CVCs. Vancomycin may be considered when patients known to have MRSA colonization and catheter-related infection or exit site infection is suspected [1]. However, the findings of this study did not fully support the IDSA guideline that recommends empirical vancomycin in febrile neutropenic patients with haemodynamic instability. Based on epidemiological data in Thailand, Gram-negative bacteria have consistently been identified as the predominant pathogens (77–89%) in febrile neutropenic patients [23, 24], and the prevalence of MRSA remains below 5% compared to all cultured pathogens. Moreover, the majority of Gram-positive organisms in this study were CoNS, which, except for S. lugdunensis and S. schleiferi, are typically indolent pathogens that do not cause haemodynamic instability, as they usually lack virulence factors capable of stimulating a vigorous immune response [25, 26].

The fact that about half of the patients who experienced acute kidney injury following vancomycin use had received the drug for no more than 48 h underscores the cautious consideration when initiating empirical vancomycin therapy. It should be reserved for cases where there is a clear clinical benefit. Patients in critical illness are already at risk for developing multi-organ failure, particularly kidney failure. The use of empirical vancomycin without appropriate indications may further exacerbate renal dysfunction and contribute to worsening outcomes.

This study is the first study in Thailand to determine the incidence of infections requiring vancomycin as the specific treatment in Thai children. A sufficiently large sample size and a data collection period of up to 10 years enhance the reliability of the obtained incidence. Factors associated with this necessity were analysed. The findings from this study should be further explored through multicentre collaborations in regions with a low incidence of MRSA. Such efforts may contribute to developing more practical guidelines for empiric vancomycin use in these settings, ensuring patient safety while promoting antibiotic stewardship.

Due to low incidence of vancomycin necessity, this study had some limitations. The analysis of some factors previously believed to be associated with vancomycin necessity is limited due to this low incidence. In addition, this study is a single-centre study conducted at a university hospital with a higher proportion of complicated and immunocompromised patients. The results might be different if the study was conducted in general hospitals. False-negative culture results represent an inherent limitation of this retrospective study. They may have arisen from two factors: incomplete data on the blood volume submitted for culture, and specimens that could only be collected intraoperatively, typically after vancomycin initiation, leaving only pre-treatment blood cultures available. Nevertheless, in this study, none of the patients classified as not requiring vancomycin as a specific treatment subsequently needed to restart vancomycin due to clinical deterioration or the detection of drug-resistant Gram-positive bacteria in any newly obtained clinical specimens.

Another limitation is that the inclusion of Bacillus spp. and Corynebacterium spp. as vancomycin-indicated pathogens may have led to a slight overestimation of the true necessity of vancomycin use. Although these cases were considered true infections based on clinical and microbiological criteria, some uncertainty remains, particularly in a retrospective design where clinicians’ real-time judgments in distinguishing true infection from contamination could not be fully verified.

Conclusion

In the setting of low prevalence of MRSA infection, the incidence of infections truly requiring vancomycin was relatively low. Factors such as febrile neutropenia, haemodynamic instability, and central venous catheters, especially when only one factor was present, were not indicators of vancomycin necessity. This study suggested empirical vancomycin treatment be considered when suspected of CNS shunt infection or intra-abdominal infection following liver transplantation.

Abbreviations

CVC

Central venous catheters

MRSA

Methicillin-resistant Staphylococcus aureus

CSF

Cerebrospinal fluid

VRE

Vancomycin-resistant enterococci

CDC

Centers for Disease Control and Prevention

IDSA

Infectious Diseases Society of America

HIV

Human immunodeficiency virus

CNS

Central nervous system

AKI

Acute kidney injury

CoNS

Coagulase-negative staphylococci

Author contributions

PN was responsible for data collection, statistical analysis, interpretation of results, and manuscript writing. NA was responsible for study concept and design, interpretation of results, writing and critical review of the manuscript. CT, SB, and SC were responsible for data acquisition and critical review of the manuscript.

Funding

Open access funding provided by Mahidol University. None.

Data availability

The datasets used and analysed during the current study are available from the corresponding author on the reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Human Research Ethics Committee, Faculty of Medicine Ramathibodi Hospital, Mahidol University (MURA2023/843). As this was a retrospective study using anonymized patient data, the ethics committee waived the requirement for informed consent and assent in accordance with the institutional guidelines. The study was conducted in accordance with the principles of the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.