Impact of molecular point-of-care testing for respiratory pathogens on antibiotic use and clinical outcomes in acute respiratory tract infections: a systematic review and meta-analysis

Summary

Background

Molecular point-of-care testing (mPOCT) offers rapid identification of respiratory pathogens, but its impact on antibiotic use and patient outcomes remains uncertain. We aimed to comprehensively evaluate the effects of mPOCT on antibiotic use and major clinical outcomes in patients presenting with acute respiratory tract infections (ARTIs).

Methods

We searched MEDLINE, Embase, Web of Science, CENTRAL, CNKI, and Wanfang Data from inception to July 1, 2025, for randomised controlled trials (RCTs) evaluating mPOCT for patients presenting with ARTIs (PROSPERO CRD420251069333). The primary outcome was antibiotic use, assessed using pooled risk ratio (RR) with random-effects models. Risk of bias and certainty of evidence were assessed using the Risk Of Bias instrument for Use in SysTematic reviews-for Randomised Controlled Trials (ROBUST-RCT) and core Grading of Recommendations, Assessment, Development and Evaluation (GRADE), respectively.

Findings

We included 25 RCTs involving 12,638 patients, of whom 61.0% were adults. Overall, mPOCT probably had little to no important effect on antibiotic use (RR 0.95, 95% CI 0.90–1.00; moderate certainty) or treatment duration (mean difference −0.44 days, 95% CI −0.98 to 0.09; moderate certainty). In adults, high-certainty evidence showed no effect on antibiotic use (RR 1.00, 95% CI 0.98–1.02), whereas in children, low-certainty evidence suggested a potential reduction (RR 0.79, 95% CI 0.65–0.97). Although mPOCT increased appropriate antibiotic prescribing (RR 2.07, 95% CI 1.55–2.77; moderate certainty), it did not affect 30-day mortality (RR 0.97, 95% CI 0.82–1.15; high certainty) and intensive care unit admission (RR 0.90, 95% CI 0.65–1.25; high certainty).

Interpretation

Moderate to high certainty evidence suggests that mPOCT does not meaningfully reduce overall antibiotic use or improve patient outcomes, particularly in adults, despite enhancing prescribing appropriateness. Routine use of mPOCT for adults with ARTIs is therefore not supported.

Funding

National Natural Science Foundation of China, the Postdoctoral Science Foundation, the Chongqing Municipality Joint Science and Health Major Medical Research Project, Outstanding Youth in Science and Technology, the Chongqing Youth Talent Fund, and the Research Foundation Flanders.

Research in context.

Evidence before this study

We searched MEDLINE, Embase, Web of Science, CENTRAL, CNKI, and Wanfang Data from inception to July 1, 2025, without language restrictions, using terms for “point-of-care testing”, “molecular testing”, “respiratory tract infection”, “antimicrobial use”, and “randomised controlled trial”. We found that existing reviews have reached inconsistent conclusions. These reviews were often confined to rapid viral assays without accounting for contemporary multiplex platforms capable of detecting viruses, bacteria and atypical pathogens. Furthermore, they were constrained by small study numbers and the omission of several recent high-quality randomised controlled trials (RCTs). This fragmented and inconsistent evidence has resulted in divergent guideline recommendations, underscoring the need for a comprehensive, up-to-date evaluation of molecular point-of-care testing (mPOCT) and its impact on antimicrobial use and major clinical outcomes.

Added value of this study

To our knowledge, this systematic review and meta-analysis is the most comprehensive to date, integrating data from 25 RCTs encompassing 12,638 patients presenting with acute respiratory tract infections (ARTIs). We evaluated the effects of mPOCT on antibiotic use, as well as other major clinical outcomes (e.g., mortality, intensive care unit [ICU] admission), resource utilization, and safety. Our analysis provides moderate to high-certainty evidence that mPOCT is likely to have little to no clinically important effect on overall antibiotic use and duration. Crucially, we found high-certainty evidence demonstrating no meaningful impact in adults, whereas low-certainty evidence suggested a possible reduction in antibiotic use in children. Although mPOCT increased the rate of appropriate antibiotic prescriptions, this did not translate into improvements in major clinical outcomes or healthcare costs. We conducted subgroup analyses by infection type, healthcare setting, mPOCT type, specimen type, and study population, providing a more nuanced understanding of efficacy across diverse clinical contexts and addressing the limitations of earlier, narrower analyses.

Implications of all the available evidence

Based on moderate-to-high certainty evidence, the routine implementation of mPOCT is not warranted in adult patients presenting with ARTIs with the goal of reducing overall antibiotic consumption or improving short-term clinical outcomes. The potential benefit in paediatric patients remains uncertain and requires further investigation through adequately powered, high-quality RCTs. Our findings suggest that simply providing a rapid diagnostic result is insufficient to change prescribing behaviour or outcomes, highlighting the need to integrate mPOCT within a broader, multifaceted diagnostic and antimicrobial stewardship strategy to realise its potential benefits.

Introduction

Acute respiratory tract infections (ARTIs) represent a major global public health burden, causing substantial morbidity and mortality.^1,2 In 2021, lower respiratory tract infections (LRTIs) accounted for an estimated 344 million incident episodes and 2.18 million deaths worldwide, including more than half a million in young children. Upper respiratory tract infections (URTIs) were even more prevalent, with 17.2 billion incident cases in 2019, representing over 40% of all cases in the Global Burden of Disease study.^1,2 Identifying the microbiological cause of ARTIs remains difficult using traditional culture methods.³ Consequently, antibiotics are frequently prescribed empirically for suspected ARTIs, which can sometimes lead to the use of unnecessarily broad-spectrum agents or treatment that may not be required in certain cases.^3,4 Thus, rapid and accurate aetiological diagnosis is essential to guide targeted therapy, improve clinical outcomes, and mitigate the development of antimicrobial resistance.⁵

Molecular point-of-care testing (mPOCT), based on nucleic acid amplification techniques, has emerged as a promising diagnostic tool. This technology can detect specific pathogens in respiratory samples within a short timeframe (typically <2 h), fulfilling the clinical need for “sample in-result out” immediacy.⁶ Despite its potential to support timely triage and targeted treatment, robust evidence supporting the clinical utility of mPOCT in routine practice remains limited and inconsistent.⁶ Concerns about its high false positives and the influence of specimen quality on accuracy have contributed to conflicting evidence.⁷ Some studies reported significant reductions in antimicrobial use,^8,9 whereas others indicated limited benefits, with variations across age groups, healthcare settings, and infection types.^10,11 These conflicting findings have translated into inconsistent, and at times contradictory, recommendations across national clinical practice guidelines.12, 13, 14

To address these inconsistencies, a few systematic reviews and meta-analyses have synthesised evidence on mPOCT; however important limitations remain.15, 16, 17 Two reviews focused only on viral nucleic acid testing and reached conflicting conclusions, without accounting for bacterial or atypical pathogens.^15,16 With the development of multiplex mPOCT assays that detect viruses, bacteria, and atypical pathogens simultaneously, evidence more representative of current practice is required. Another meta-analysis assessed multiplex pathogen detection but was limited by a small number of included studies, omission of recent high-quality randomised controlled trials (RCTs), and a narrow focus on antibiotic use, while overlooking key outcomes such as mortality, intensive care unit (ICU) admission, economic impact, and safety.¹⁷

Accordingly, we conducted a comprehensive systematic review and meta-analysis of all relevant RCTs to compare mPOCT with conventional aetiological testing in patients presenting with ARTIs. Our study assessed impacts of mPOCT on antibiotic use, major clinical outcomes, economic costs, and safety. We also performed detailed subgroup analyses to explore variations in effect across patient populations, infection types, and healthcare settings, aiming to identify contexts in which mPOCT may offer the greatest benefit. By integrating the highest-quality and most up-to-date evidence, our study provides robust, practice-relevant insights to guide the clinical adoption of mPOCT and to inform the development and updating of practice guidelines.

Methods

Search strategy and selection criteria

We systematically searched MEDLINE, Embase, Web of Science, the Cochrane Central Register of Controlled Trials (CENTRAL), China National Knowledge Infrastructure (CNKI), and Wanfang Data from their inception to July 1, 2025, without language restrictions. Additional studies were identified through reference list screening, citation tracking, and searching trial registries (ClinicalTrials.gov and the World Health Organization [WHO] International Clinical Trial Registry Platform) and Google searches. The search strategy combined MeSH terms and free-text words, including “point-of-care testing”, “molecular testing”, “respiratory tract infection”, “antimicrobial use”, and “randomised controlled trial” (Appendix pp 2–20).

We included RCTs that evaluated the use of mPOCT in adult or pediatric patients presenting with symptoms or signs suggestive of an ARTI. The encompassed patient with a presumptive diagnosis of pneumonia, bronchiolitis, bronchitis, influenza-like illness, pharyngitis, or an acute exacerbation of a chronic respiratory illness, as well as those presenting with key respiratory symptoms or signs such as fever/history of fever, cough, expectoration, sore throat, tachypnea, wheezing, difficulty breathing, chest pain, or abnormal auscultation/percussion findings. We defined mPOCT as any rapid nucleic acid amplification test for aetiological diagnosis with a turnaround time of less than 24 h. Eligible comparators included conventional microbiological testing (e.g., indirect immunofluorescence assay, direct fluorescent antibody staining followed by culture, pathogen culture, urinary antigen tests for Streptococcus pneumoniae and Legionella pneumophila), or no microbiological testing. We also included studies where the comparator was the same rapid molecular test performed with delayed results, when the treating physician was unaware of the results, or when testing was conducted in a central laboratory with a longer turnaround time rather than at the point of care. We excluded trials in which the target population was a non-infectious respiratory condition (e.g., allergic rhinitis, heart failure, pulmonary embolism), rather than an acute respiratory infection/ARTI-suspected population, non-RCT designs (e.g., quasi-RCTs, observational studies), head-to-head comparisons of two different mPOCT techniques, and diagnostic accuracy studies that did not assess clinical impact. Study screening was conducted independently by two investigator groups (group 1 was QYL, JBF, and XFF, group 2 was QZ and HHL). Disagreements were resolved through discussion or consultation with a third party (ZXL and LZ).

Ethics

This systematic review and meta-analysis are reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement¹⁸ and has been registered with PROSPERO (ID: CRD420251069333). Ethical approval was not required for this study because it is a systematic review and meta-analysis of data exclusively from published literature.

Data extraction and quality assessment

Two investigator groups used a standardised form to extract data independently from studies, including study characteristics (authors, year, country where the research was conducted, sample size, healthcare setting), participant details (age, sex, infection type), intervention specifics (test type, turnaround time of test, specimen type), comparators, and outcomes. For trials with multiple datasets (e.g., intention-to-treat [ITT] and per-protocol [PP] analyses), ITT data were prioritized, followed by PP data.⁴ When outcomes were reported at multiple time points, the longest available follow-up was used. Disagreements were resolved by discussion.

The risk of bias for each included study was independently assessed by two investigator groups using the Risk of Bias Instrument for Use in SysTematic Reviews—for RCTs (ROBUST-RCT).¹⁹ Discrepancies were resolved through discussion or adjudication by a third-party adjudication (YLC and LZ). We also graded the certainty of evidence using core Grading of Recommendations, Assessment, Development and Evaluation (GRADE), considering risk of bias, inconsistency, indirectness, imprecision, and publication bias.²⁰

Outcomes

The pre-specified primary outcome was antibiotic use, defined as the proportion of patients who received any antibiotic treatment within 30 days following randomisation, regardless of the healthcare setting (including inpatient, ICU, emergency department [ED], or outpatient). Antibiotic use was chosen as the primary outcome because it is a key indicator of antibiotic stewardship effectiveness, directly linked to antimicrobial resistance prevention, and aligns with established clinical guidelines.12, 13, 14 Additionally, the 30-day follow-up period enables the assessment of broader influence of mPOCT on overall antibiotic prescribing patterns beyond initial treatment decisions. Secondary outcomes included: (1) antimicrobial stewardship: duration of antibiotic therapy, proportion of patients with appropriate antibiotic use, and proportion of patients with antiviral drug use; (2) clinical outcomes and safety: 30-day all-cause mortality, need for ICU admission, length of ICU stay, length of hospital stay, length of ED stay, need for hospital readmission, time to clinical stability, and incidence of any adverse events and serious adverse events; (3) resource utilisation and others: number of laboratory tests ordered, number of chest radiography ordered, examination cost, treatment cost. In addition, we conducted post-hoc exploratory analyses for four outcomes: time to administration of appropriate antimicrobial treatment, proportion of patients undergoing antibiotic de-escalation and escalation, and need for hospitalization. These outcomes were included as they were frequently reported across studies and are clinically relevant to antibiotic stewardship and resource utilisation. All outcomes were evaluated within the study follow-up period, up to a maximum of 30 days after randomisation. Outcome definitions were detailed in Appendix p 21.

Statistics

We performed meta-analyses using a random-effects model to account for anticipated between-study heterogeneity. We calculated risk ratios (RRs) with 95% CIs for dichotomous outcomes and mean differences (MDs) with 95% CIs for continuous outcomes. To facilitate interpretability, we also present risk differences (RDs) according to the probability of achieving the minimal important difference (MID). Based on a review of the literature and expert panel discussions, we defined the MID as follows: 10% for antibiotic use and appropriate antibiotic use; 5% for 30-day mortality and ICU admission; 1 day for duration of antibiotic therapy, time to appropriate antibiotic use, and length of hospital stay.21, 22, 23 Statistical heterogeneity was quantified using the I² statistic, with values of less than 50%, 50–75%, and more than 75% interpreted as low, moderate, and high heterogeneity, respectively.²⁴ When data were missing or the reporting format was inappropriate for meta-analysis, we derived the required data from other reported information, following the methods recommended by the Cochrane Handbook.²⁵ Subgroup analyses explored sources of heterogeneity, including age group (children vs adults), infection type (pneumonia vs bronchitis vs upper respiratory tract infection [URTI] vs others), healthcare setting (emergency or outpatient departments vs inpatient ward vs ICU), mPOCT type (viral detection vs bacterial detection vs multiplex pathogen detection [viral, bacterial, and/or atypical]), specimen type (nasopharyngeal [NA] swab or aspirates vs sputum vs bronchoalveolar lavage fluid [BALF], and test result (virus positive vs negative). Because the suspected infection group might inadvertently include a small proportion of non-infected individuals, it could potentially diminish the observed effectiveness of mPOCT and thereby introduce bias into the overall results. Given that this meta-analysis relied on trial-level data rather than individual patient data, we were precluded from excluding non-infected individuals or isolating them for separate analysis. Therefore, to evaluate the impact of this potential confounding factor, we conducted a post-hoc subgroup analysis based on the study population (suspected vs confirmed infection). Interaction tests were performed to determine whether the differences between subgroups were statistically significant. A p value of less than 0.05 was considered indicative of a subgroup effect.²⁶ We also performed a sensitivity analysis to assess the robustness of our findings, including excluding one trial at a time from each analysis and analysing studies that excluded poor-quality respiratory sampling. Additionally, in response to reviewers' suggestions, we examined the impact of mPOCT on antibiotic use at different time point (4 h, 24 h, and 48 h after randomisation). If ten or more studies were included, publication bias was evaluated using Egger's test.²⁷ Multiple testing poses a prevalent challenge in systematic reviews, and clear guidelines for its effective statistical management are currently lacking.^28,29 Therefore, the results presented in this manuscript have not been adjusted for multiple testing. All analyses were performed using RevMan 5.4 software and STATA15.0 (StataCorp, College Station, TX).

Role of the funding source

The funders of the study had no involvement in the study design, data collection, data analysis, interpretation, or writing of report.

Results

Our search identified 11,756 records. After removing 2310 duplicates, 9446 titles and abstracts were screened, of which 9383 were excluded. 63 full-text articles were assessed for eligibility, and 25 RCTs involving 12,638 participants were included (Fig. 1).^10,11,30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52 Reasons for excluding the 38 studies at full-text review are provided in the Appendix (pp 22–25). Across all included studies, 54.9% (6938/12,638) of participants were male and 61.0% (7709/12,638) were adults. 20 (80.0%) of 25 studies reported a turnaround time of less than 2 h for mPOCT results.^11,30, 31, 32, 33^,35, 36, 37, 38, 39, 40^,42, 43, 44, 45, 46, 47^,49,50,52 One study (4.0%) enrolled only patients with hospital-acquired pneumonia and ventilator-acquired pneumonia,³⁶ six studies (24.0%) enrolled only community-acquired pneumonia patients.^{10,31,32,37,38,41} 16 studies (64.0%) were conducted in emergency or outpatient departments^11,30, 31, 32, 33^,35,37, 38, 39, 40, 41, 42, 43^,47, 48, 49; 16 studies (64.0%) used NA swabs or aspirates as specimens^{11,30,31,33,35,}37, 38, 39, 40^,42, 43, 44^,47, 48, 49, 50; and 18 studies (72.0%) used multiplex mPOCT panels capable of detecting viruses, bacteria, or atypical pathogens.^10,31, 32, 33^,35, 36, 37, 38^,40, 41, 42, 43, 44^,46,47,50, 51, 52 A detailed list of the mPOCT panel targets used in each study can be found in the Appendix (pp 26–29). Basic characteristics of the included RCTs are summarised in Table 1.

Fig. 1.

Study flow diagram. ARI = acute respiratory infection; CKNI = China National Knowledge Infrastructure; mPOCT = molecular point-of-care testing; RCT = randomised controlled trial.

Table 1.

Characteristics of included RCTs.

	Country	Sample size	Male (%)	Age, years, mean (SD)	Disease	Study setting	Specimen type	Rapid testa	Comparator testb
Abelenda-Alonso et al., 202410	Spain	242	147/242 (60.7)	72.8 (14.7)	CAP	Inpatients	Sputum or nasopharyngeal swab	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Bibby et al., 202211	Canada	1128	620/1128 (55.0)	NR	Patients presenting with ARTI, with no specific diagnosis reported	ED and inpatients	Nasopharyngeal swab or aspirate	Xpert Xpress; viral detection	Conventional microbiological testing
Brendish et al., 201730	UK	714	346/714 (48.5)	59.7 (23.9)	ARTI or fever (≤7 days)	ED and acute medical unit	Nose and throat swabs	BioFire FilmArray; viral detection	Conventional microbiological testing
Cantais et al., 202531	France	499	277/499 (55.5)	3.7 (3.1)	CAP	ED	Nasopharyngeal aspirate	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Cartuliares et al., 202332	Denmark	291	148/291 (51.0)	71.3 (15.7)	Suspected CAP	ED	Tracheal secretion or expectorated sputum	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Clark et al., 202133	UK	613	280/613 (45.7)	60.4 (23.2)	Suspected ARTI	ED and acute medical unit	Nose and throat swab	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Darie et al., 202234	Switzerland	208	135/208 (65.0)	65.9 (14.0)	Suspected Pneumonia (CAP and HAP)	Inpatients	BALF	Unyvero HPN Cartridge; bacterial detection	Conventional microbiological testing
Echavarría et al., 201835	Argentina	432	213/432 (49.3)	30.3 (28.0)	Suspected ALRI	ED	Nasopharyngeal swab	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Enne et al., 202536	UK	545	373/545 (68.4)	50.3 (26.7)	Suspected pneumonia (HAP and VAP)	ICU	Lower respiratory tract specimen (sputum, ETA, BALF)	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Gelfer et al., 201537	USA	59	24/59 (41.3)	63.7 (16.2)	CAP	ED	Nasal swabs	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Gilbert et al., 201638	USA	127	62/127 (48.8)	70.4 (17.7)	CAP	ED	Nasal swabs	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Gill et al., 202439	Canada	227	124/227 (54.6)	7.9 (3.7)	Suspected GAS pharyngitis	ED	Throat swab	ID NOW Strep A 2 assay; bacterial detection	Conventional microbiological testing
Li et al., 202540	China	1000	537/1000 (53.7)	5.1 (3.0)	Suspected ARTI	Outpatients	Nasopharynx or oropharynx swab	Sansure Biotech; multiplex detection	Conventional microbiological testing
Markussen et al., 202441	Norway	374	221/374 (59.1)	70.6 (14.2)	Suspected CAP	ED	Lower respiratory tract specimen (sputum, ETA)	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Mattila et al., 202242	Finland	1243	696/1243 (56.0)	3.0 (3.6)	Patients presenting with ARTI; most common diagnoses were viral wheezing or bronchiolitis	ED	Nasopharyngeal specimens	QIAstat-Dx; multiplex detection	Conventional microbiological testing
May et al., 201943	USA	191	81/191 (42.4)	29.0 (24.0)	Suspected AUTI or influenza-like illness	ED	Nasopharyngeal swab	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Meltzer et al., 202444	USA	360	128/360 (35.6)	35.8 (17.2)	Suspected ARTI	Urgent care center	Nasopharyngeal aspirate	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Paonessa et al., 201945	USA	45	17/45 (37.8)	65.1 (15.9)	Suspected pneumonia (CAP, HAP and VAP)	ICU	BALF	Xpert Xpress; bacterial detection	Conventional microbiological testing
Poole et al., 202246	UK	200	140/200 (70.0)	62.8 (14.7)	Pneumonia (CAP, HAP and VAP)	ICU	Lower respiratory tract specimen (sputum, ETA, BALF)	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Rao et al., 202147	USA	908	508/908 (56.0)	2.92 (3.5)	Influenza-like illness	ED	Nasopharyngeal swab	BioFire FilmArray; multiplex detection	Same test, clinician unaware of results
Saarela et al., 202048	Finland	998	526/998 (52.7)	61.5 (17.3)	Patients presenting with ARTI, with no specific diagnosis reported	ED	Nasal swab	Anyplex™ II RV16 Detection; viral detection	Same test, results delayed to the clinician
Schechter-Perkins et al., 201849	USA	197	90/197 (45.7)	35.7 (20.9)	Suspected influenza-like illness	ED	Nasopharyngeal swab	Cobas Liat Influenza A/B assay; viral detection	Conventional microbiological testing
Shengchen et al., 201950	China	800	456/800 (57.0)	60.9 (17.9)	CAP, AECOPD or acute exacerbation of bronchiectasis	Inpatients	Nasopharyngeal swab	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Verroken et al., 202451	Belgium	85	54/85 (63.5)	62.6 (15.2)	Suspected severe pneumonia (CAP, HAP, VAP)	ICU	Lower respiratory tract specimen (sputum, ETA, BALF)	BioFire FilmArray; multiplex detection	Conventional microbiological testing
Virk et al., 202452	USA	1152	733/1152 (63.6)	63.5 (14.8)	Suspected pneumonia (CAP, HAP, VAP)	Inpatients	Lower respiratory tract specimen (sputum, ETA, BALF)	BioFire FilmArray; multiplex detection	Conventional microbiological testing

AECOPD = acute exacerbation of chronic obstructive pulmonary disease. ALRI = acute lower respiratory tract infection. ARTI = acute respiratory tract infection. AURI = acute upper respiratory tract infection. BALF = bronchoalveolar lavage fluid. CAP = community acquired pneumonia. ED = emergency department. ETA = endotracheal aspirate. HAP = hospital acquired pneumonia. ICU = intensive care unit. GAS = group A streptococcus. NR = not report. UK=The United Kingdom. USA=The United States of America. VAP = ventilator associated pneumonia.

^aMultiplex detection referred to the simultaneous detection of viruses, bacteria, and/or atypical pathogens.

^bConventional microbiological testing included bacterial cultures and antimicrobial susceptibility testing, urine antigen testing, immunofluorescence, and conventional PCR for respiratory pathogens.

We evaluated the risk of bias in 25 RCTs using the ROBUST-RCT tool. The assessment revealed that only six core items required consideration, while the eight optional items could be disregarded. Among the 19 RCTs reporting the primary outcome (antibiotic use), 18 (94.7%) were rated as “probably or definitely low” risk of bias for random sequence generation,30, 31, 32, 33, 34^,36,39, 40, 41, 42, 43, 44^,46, 47, 48, 49, 50^,52 16 (84.2%) for allocation concealment,^11,30, 31, 32, 33, 34^,36,39, 40, 41, 42, 43, 44^,48, 49, 50 and 18 (94.7%) for missing outcome data.^11,30, 31, 32, 33, 34^,36,39, 40, 41, 42, 43^,46, 47, 48, 49, 50^,52 Although blinding of participants, clinicians, and outcome assessors was absent in more than half of the trials, this was judged unlikely to substantially bias estimates of antibiotic use due to inherent differences in turnaround times between mPOCT and standard testing. Detailed assessments for other outcomes are provided in Appendix (pp 30–39).

19 studies (11,544 patients) reported antibiotic use.^11,30, 31, 32, 33, 34^,36,39, 40, 41, 42, 43, 44^,46, 47, 48, 49, 50^,52 Pooled estimates indicated that mPOCT probably had little to no important effect on antibiotic use overall (RR 0.95, 95% CI 0.90–1.00, Fig. 2, Table 2; RD −0.04, 95% CI −0.07 to 0.00; Appendix p 40; I² 85%; moderate certainty). Subgroup analysis revealed that mPOCT may slightly decrease antibiotic use among paediatric patients (RR 0.79, 95% CI 0.65–0.97; I² 88%; low certainty); however, mPOCT had little to no important effect on antibiotic use in adult patients (RR 1.00, 95% CI 0.98–1.02; I² 0%; high certainty) (p_interaction = 0.03, Table 3). No statistically significant differences were found between subgroups defined by confirmed or suspected infection, disease type, healthcare setting, specimen type, mPOCT platform, or positive/negative test result (Table 3).

Fig. 2.

Random-effects meta-analysis of mPOCT vs control. (A) Antibiotic use. (B) Antibiotic duration. mPOCT = molecular point-of-care testing.

Table 2.

Primary and secondary outcomes.

	Number of trials	Sample size	Overall effect (95% CI)	Heterogeneity (I2)	p value
Primary outcome
Antibiotic use	19	11,544	RR 0.95 (0.90–1.00)	85%	0.04
Secondary outcomes
Duration of antibiotic therapy, days	9	3999	MD −0.44 (−0.98 to 0.09)	85%	0.10
Appropriate antibiotic use	7	3083	RR 2.07 (1.55–2.77)	83%	<0.0001
Time to administration of appropriate antimicrobial treatment, h	3	1611	MD −8.35 (−11.58 to −5.12)	99%	<0.0001
Antibiotic de-escalation	5	2532	RR 1.05 (0.74–1.50)	71%	0.77
Antibiotic escalation	3	1550	RR 2.43 (0.60–9.84)	83%	0.21
Antiviral drug use	6	4555	RR 1.01 (0.82–1.25)	52%	0.92
30-day all-cause mortality	18	8690	RR 0.97 (0.82–1.15)	8%	0.72
ICU admission	11	6224	RR 0.90 (0.65–1.25)	29%	0.52
Length of ICU stay, days	4	1892	MD −0.01 (−1.18 to 1.16)	18%	0.99
Hospitalisation	7	4979	RR 1.03 (0.95–1.12)	25%	0.48
Length of hospital stay, days	14	6811	MD −0.56 (−0.99 to −0.13)	73%	0.01
Length of emergency department stay, days	4	2569	MD 0.10 (−0.05 to 0.25)	27%	0.19
Hospital readmission	13	7035	RR 0.89 (0.80–1.00)	0%	0.05
Time to clinical stability, days	4	1265	MD −0.07 (−0.19 to 0.05)	0%	0.24
Any adverse events	13	4892	RD −0.00 (−0.01 to 0.01)	73%	0.71
Any serious adverse events	8	3221	RD 0.00 (−0.00 to 0.00)	0%	0.97
Number of laboratory tests ordered	3	2354	RR 0.99 (0.90–1.10)	0%	0.90
Number of chest radiography ordered	4	3561	RR 0.94 (0.80–1.10)	67%	0.45

ICU = intensive care unit. MD = mean difference. RD = risk difference. RR = risk ratio.

Table 3.

Subgroup analysis of antibiotic use.

	Number of trials	Number of patients	Risk ratio (95% CI)	Risk difference (95% CI)	pinteraction
Age
Children	7	4652	0.79 (0.65–0.97); I2 = 88%	−0.14 (−0.25 to −0.02); I2 = 94%	0.03
Adults	11	5891	1.00 (0.98–1.02); I2 = 0%	−0.00 (−0.02 to 0.01); I2 = 0%
Infection type
AURI	3	928	0.58 (0.41–0.82); I2 = 74%	−0.23 (−0.43 to −0.03); I2 = 90%	0.09
Bronchitis	1	318	0.91 (0.80–1.04); NA	−0.07 (−0.16 to 0.03); NA
CAP and HAP	5	1924	1.00 (0.98–1.02); I2 = 0%	−0.00 (−0.02 to 0.02); I2 = 0%
CAP	4	1443	0.92 (0.78–1.09); I2 = 96%	−0.06 (−0.20 to 0.08); I2 = 96%
HAP	1	545	1.00 (0.97–1.03); NA	0.00 (−0.03 to 0.03); NA
Influenza-like illness	3	1250	1.09 (0.82–1.45); I2 = 62%	0.03 (−0.03 to 0.09); I2 = 23%
Exacerbation of COPD/bronchiectasis/asthma	2	590	0.98 (0.89–1.09); I2 = 54%	−0.02 (−0.10 to 0.07); I2 = 56%
Healthcare setting
Emergency or outpatient department	12	6459	0.90 (0.78–1.03); I2 = 86%	−0.06 (−0.13 to 0.01); I2 = 91%	0.32
Inpatient ward	4	2492	1.00 (0.97–1.02); I2 = 0%	−0.00 (−0.02 to 0.02); I2 = 0%
Intensive care unit	2	745	1.00 (0.97–1.03); I2 = 0%	0.00 (−0.03 to 0.03); I2 = 0%
Specimen type
Nasopharyngeal aspirate or swab	13	8775	0.90 (0.83–0.99); I2 = 87%	−0.06 (−0.11 to −0.01); I2 = 90%	0.15
Sputum	1	290	1.02 (0.90–1.16); NA	0.02 (−0.08 to 0.12); NA
BALF	1	208	1.01 (0.93–1.10); NA	0.01 (−0.06 to 0.09); NA
Mixed sputum and BALF	4	2271	1.00 (0.98–1.03); I2 = 0%	0.00 (−0.02 to 0.02); I2 = 0%
mPOCT type
Viral detection	4	3037	0.99 (0.94–1.05); I2 = 0%	−0.01 (−0.04 to 0.02); I2 = 40%	0.51
Bacterial detection	2	435	0.67 (0.20–2.22); I2 = 99%	−0.21 (−0.71 to 0.28); I2 = 98%
Multiplex pathogen detection	13	8072	0.96 (0.91–1.01); I2 = 85%	−0.03 (−0.07 to 0.01); I2 = 86%
Test result
Viral positive	4	600	0.80 (0.59–1.08); I2 = 77%	−0.08 (−0.15 to −0.01); I2 = 31%	0.07
Viral negative	1	221	1.07 (0.95–1.22); NA	0.06 (−0.04 to 0.16); NA
Study population
Confirmed infection	4	2405	0.95 (0.81–1.12); I2 = 94%	−0.06 (−0.18 to 0.07); I2 = 81%	0.97
Suspected infection	16	9139	0.96 (0.91–1.01); I2 = 79%	−0.03 (−0.07 to 0.00); I2 = 87%

AURI = acute upper respiratory tract infection. BALF = bronchoalveolar lavage fluid. CAP = community acquired pneumonia. COPD = chronic obstructive pulmonary disease. HAP = hospital acquired pneumonia. mPOCT = molecular point-of-care testing. NA=Not applicable.

Nine studies, encompassing 3999 patients, reported on antibiotic duration.^{10,30,33,34,41,45,46,48,50} Similarly, mPOCT probably had little to no important effect on the duration of antibiotic therapy (MD −0.44 days, 95% CI −0.98 to 0.09; I² 85%; moderate certainty; Fig. 2, Table 2). Regarding other pathogen-directed therapy outcomes, mPOCT probably resulted in an increase in appropriate antibiotic use (RR 2.07, 95% CI 1.55–2.77; Table 2, Appendix p 41; I² 83%; moderate certainty). Post-hoc exploratory analyses showed that mPOCT probably shortened the time to appropriate antibiotic initiation (MD −8.35 h, 95% CI −11.58 to −5.12; Table 2, Appendix p 42; I² 99%; moderate certainty). However, exploratory analyses of antibiotic de-escalation (RR 1.05, 95% CI 0.74–1.50; Table 2, Appendix p 43; I² 71%) and antibiotic escalation (RR 2.43, 95% CI 0.60–9.84; Table 2, Appendix p 44; I² 83%) revealed no significant differences between the mPOCT and control groups. Similarly, there were no significant differences between mPOCT and control groups in antiviral use (RR 1.01, 95% CI 0.82–1.25; Table 2, Appendix p 45; I² 52%).

For patient prognosis and safety-related outcomes, mPOCT had little to no effect on 30-day mortality (RR 0.97, 95% CI 0.82–1.15; Table 2, Appendix p 46; I² 8%; high certainty) and ICU admission (RR 0.90, 95% CI 0.65–1.25; Table 2, Appendix p 47; I² 29%; high certainty). It also probably had little to no effect on length of hospital stay (MD −0.56 days, 95% CI −0.99 to −0.13; Table 2, Appendix p 48; I² 73%; moderate certainty). No statistically significant differences were found between mPOCT and control groups for the need for hospitalisation (RR 1.03, 95% CI 0.95–1.12; Table 2, Appendix p 49; I² 25%; post-hoc exploratory analysis finding), ICU length of stay (MD −0.01 days, 95% CI −1.18 to 1.16; Table 2, Appendix p 50; I² 18%), ED length of stay (MD 0.10 days, 95% CI −0.05 to 0.25; Table 2, Appendix p 51; I² 27%), time to clinical stability (MD −0.07 days, 95% CI −0.19 to 0.05; Table 2, Appendix p 52; I² 0%), need for hospital readmission (RR 0.89, 95% CI 0.80–1.00; Table 2, Appendix p 53; I² 0%), any adverse events (RD -0.00, 95% CI −0.01 to 0.01; Table 2, Appendix p 54; I² 73%), and serious adverse events (RD 0.00, 95% CI −0.00 to 0.00; Table 2, Appendix p 55; I² 0%).

Regarding healthcare expenditure, the numbers of laboratory tests ordered (RR 0.99, 95% CI 0.90–1.10; Table 2, Appendix p 56; I² 0%) and chest imaging ordered (RR 0.94, 95% CI 0.80–1.10; Table 2, Appendix p 57; I² 67%) were similar between mPOCT and control groups. Correspondingly, there were no significant differences in costs associated with laboratory tests (MD $13.57, 95% CI −96.69 to 123.84; I² 72%) or chest imaging (MD $45.18, 95% CI −11.56 to 101.92). For treatment costs, Shengchen et al.⁵⁰ reported lower intravenous antibiotic costs in the mPOCT group (median $189.9 [IQR 103.5–316.5] vs $245.8 [138.1–397.8]; p < 0.001), and Gilbert et al.³⁸ found a lower median cost of therapy per 1000 patient days in the mPOCT group ($3037 vs $7952; p = 0.02). However, Saarela et al. found no significant difference in hospital treatment between the groups (MD €200, 95% CI −669 to 1069).⁴⁸

Sensitivity analysis confirmed the robustness of our findings (Appendix pp 58–59). No evidence of publication bias was detected for the primary outcome of antibiotic use (Egger's p = 0.17, Appendix p 60). The certainty of evidence, assessed using core GRADE, was moderate for antibiotic use and antibiotic duration, and ranged from moderate to high for secondary outcomes. Detailed GRADE assessments are available in Table 4.

Table 4.

Summary of findings table comparing effects of mPOCT vs control for patients with acute respiratory tract infection.

	Patients (studies)	Risk ratio (95% CI)	Absolute effects (95% CI)			Certainty of evidence	Plain language summary
			Control	mPOCT	Difference
Rate of antibiotic use (all populations)	11,544 (19 RCTs)	0.95 (0.90–1.00)	654 per 1000	622 per 1000 (589–654)	−33 per 1000 (−65 to 0)	Moderatea	mPOCT probably has little to no effect on the rate of antibiotic use.
Rate of antibiotic use (adults)	5891 (11 RCTs)	1.00 (0.98–1.02)	802 per 1000	802 per 1000 (786–818)	0 per 1000 (−16 to 16)	High	mPOCT does not reduce the rate of antibiotic use in adult patients.
Rate of antibiotic use (children)	4652 (7 RCTs)	0.79 (0.65–0.97)	453 per 1000	358 per 1000 (295–440)	−95 per 1000 (−159 to −14)	Lowa,b	mPOCT may slightly decrease the rate of antibiotic use in paediatric patients.
Rate of antibiotic use (Confirmed infection)	2405 (4 RCTs)	0.95 (0.81–1.12)	657 per 1000	624 per 1000 (736–532)	−33 per 1000 (−125 to 79)	Lowa,b	mPOCT may not reduce the rate of antibiotic use in confirmed ARTI patients.
Rate of antibiotic use (Suspected infection)	9139 (16 RCTs)	0.96 (0.91–1.01)	654 per 1000	628 per 1000 (660–595)	−26 per 1000 (−59 to 7)	Moderatea	mPOCT probably not reduce the rate of antibiotic use in suspected ARTI patients.
Duration of antibiotic therapy	3999 (9 RCTs)	–	8.11 days	7.67 days	−0.44 (−0.98 to 0.09)	Moderatea	mPOCT probably has little to no effect on duration of antibiotic therapy.
Rate of appropriate antibiotic use	3083 (7 RCTs)	2.07 (1.55–2.77)	252 per 1000	522 per 1000 (391–698)	270 per 1000 (139–446)	Moderatea	mPOCT probably increases the rate of appropriate antibiotic use.
Time to administration of ATA	1611 (3 RCTs)	–	29.80 h	21.45 h	−8.35 h (−11.58 to −5.12)	Moderatea	mPOCT probably reduces the time to appropriate antimicrobial treatment.
30-day mortality	8690 (18 RCTs)	0.97 (0.82–1.15)	74 per 1000	72 per 1000 (61–85)	−2 per 1000 (−13 to 11)	High	mPOCT has little to no effect on 30-day mortality.
Rate of ICU admission	6224 (11 RCTs)	0.90 (0.65–1.25)	44 per 1000	39 per 1000 (28–55)	−4 per 1000 (−15 to 11)	High	mPOCT has little to no effect on the rate of ICU admission.
Length of hospital stay	6811 (14 RCTs)	–	8.09 days	7.39 days	−0.7 days (−1.13 to −0.27)	Moderatec	mPOCT probably has little to no effect on length of hospital stay.

AAT = appropriate antimicrobial treatment. ICU = intensive care unit. MD = mean difference. mPOCT = molecular point-of-care testing. RCTs = randomised controlled trials.

^aRated down one level for inconsistency because of the heterogeneity among the included studies was large (I² > 75%).

^bRated down one level for imprecision because of the 95% CI crossing the minimally important differences decision threshold (MID = 10%).

^cRated down one level for inconsistency because of the heterogeneity among the included studies was substantial (I² = 73%), and there were some differences in definitions of length of hospital stay across the included studies.

Discussion

In this comprehensive systematic review and meta-analysis of 25 RCTs involving 12,638 patients presenting with ARTIs, we found that mPOCT for respiratory pathogens probably had little to no meaningful effect on overall antibiotic use and treatment duration. While mPOCT modestly increased appropriate antibiotic prescribing and expedited targeted therapy, it did not improve short-term clinical outcomes such as 30-day mortality or ICU admission, nor did it lead to measurable reductions in healthcare costs.

Our findings that mPOCT has a limited impact on antibiotic use are consistent with prior meta-analyses.15, 16, 17 To explore potential factors influencing mPOCT efficacy, we performed comprehensive subgroup analyses based on study population, pathogen detection range, specimen source, clinical setting, and test result. Previous meta-analyses primarily focused on rapid viral tests and did not fully consider the advantages of multiplex assays in identifying bacterial co-infections, especially in critically ill patients.^15,16 Although modern multiplex PCR can detect a broader range of pathogens, our results showed that mPOCT had no significant impact on overall antibiotic use, regardless of whether virus-only or multiplex assays were used. Similarly, despite the superior diagnostic accuracy of lower respiratory tract specimens (e.g., BALF, tracheal aspirates),⁵³ specimen source did not significantly influence antibiotic prescribing patterns. Furthermore, mPOCT failed to improve antibiotic utilisation in either outpatient or ICU settings, despite substantial differences in patient complexity between these environments. Crucially, even when stratifying by test result (positive vs negative), we observed no significant effect on antibiotic use, underscoring a lack of clinical translation despite previous evidence suggesting potential benefits in virus-positive patients.³⁰

This consistent lack of significant effect likely stems from shared challenges in ARTI management. First, suboptimal physician adherence to mPOCT protocols may limit its effectiveness. Successful implementation depends on reliable results, yet many studies fail to report physician compliance. One study found only 30.5% adherence to the mPOCT algorithm.³⁶ In outpatient settings, empirical prescribing persists due to time pressures, patient expectations, or fears of complications, even when viral results are positive. Conversely, in ICUs, illness severity often drives broad-spectrum antibiotic use irrespective of initial findings, with de-escalation hampered by persistent concerns over clinical deterioration, co-infections, or resistance. Our test-result subgroup analysis further illuminates this dilemma. Positive viral test results modestly reduced antibiotic initiation but rarely shortened treatment durations, as empirical coverage continued amid ongoing worries about subsequent infections. Negative results, which should reassure against bacterial infection, similarly failed to prompt discontinuation. Clinicians often disregarded these negative findings when faced with diagnostic uncertainty or low pretest probability thresholds. These patterns reveal a broader trust deficit in mPOCT, where clinical judgment and cautious guidelines frequently override test results, emphasizing the need to boost adherence and build trust for optimal benefits. Second, integrating mPOCT results into clinical decision-making is challenging, particularly for less experienced practitioners. Clinicians often combine mPOCT with biomarkers like C-reactive protein and procalcitonin to distinguish between infection types.⁵⁴ Recent research found that utilising mPOCT in conjunction with procalcitonin can reduce antibiotic duration.⁵⁵ Therefore, simply introducing mPOCT is insufficient to curb antibiotic overuse. A comprehensive diagnostic and antimicrobial stewardship approach is essential.

While prior subgroup analyses failed to identify a clearly defined beneficiary population for mPOCT, our age-stratified subgroup analysis revealed differential findings. High-quality evidence indicates that mPOCT does not reduce antibiotic use in adults, whereas low-quality evidence suggests a potential reduction in paediatric patients. Contradictorily, earlier meta-analyses found no significant difference in mPOCT's effect on antibiotic use between adults and children.^15,17 This discrepancy likely arises from the limited number of paediatric studies included in those analyses and the exclusion of recent paediatric-specific evidence.

Recent paediatric studies have shown positive findings^31,39,40; and our updated analysis incorporating this evidence suggests that mPOCT may indeed reduce antibiotic use in children. The potential benefit in paediatric ARTIs may reflect the substantially higher prevalence of viral infections compared to adults.⁵⁶ A positive mPOCT result for viral infection may increase clinician confidence in safely withholding or discontinuing antibiotics. Furthermore, paediatric cases are often less clinically complex than adult cases, which frequently involve comorbidities or polymicrobial infections, thereby facilitating more straightforward treatment decisions based on pathogen identification.⁵⁶ Parental concerns about antibiotic use in children may further encourage clinicians to avoid unnecessary prescriptions following a definitive viral diagnosis. Although these subgroup findings are biologically plausible, cautious interpretation is warranted given the limited and heterogeneous supporting evidence. Well-designed, adequately powered RCTs are needed to validate the value of mPOCT in paediatric populations.

Our study reveals a paradox: while mPOCT improved appropriate antibiotic use and shortened time to appropriate therapy, it did not reduce overall antibiotic consumption or duration, improve short-term clinical outcomes, nor decrease costs. This improvement in “appropriateness” may not translate into broader benefits because modest time savings in initiating appropriate therapy might not affect hard endpoints such as 30-day mortality. The advantages of mPOCT, including potential reductions in antimicrobial resistance, may be long-term and therefore not captured by short-term follow-up. Furthermore, the substantial cost of mPOCT itself might offset any potential economic savings derived from more targeted prescribing, making it difficult to show a clear cost-effectiveness advantage within a short timeframe.

Our study has several limitations. First, significant statistical heterogeneity existed among the included studies, likely reflecting differences in disease severity and type, mPOCT type, and clinical practices across countries. To address this, we employed a random-effects model and conducted subgroup analyses to explore and mitigate the impact of heterogeneity. Second, due to the nature of mPOCT interventions requiring active administration and result communication, blinding of clinicians and patients was not feasible, potentially introducing bias. However, as the primary outcome is an objective measure, such bias is unlikely to have significantly affected the overall findings. Third, data for certain secondary outcomes and subgroup analyses were derived from a small number of studies, limiting the robustness of these findings. Forth, although mPOCT may influence long-term prognosis and potentially benefit patients with high compliance with mPOCT-guided treatment recommendations or recent pre-randomisation antibiotic use, our meta-analysis could not investigate these effects due to limited primary data. Fifth, a potential limitation is the inclusion of non-infected individuals within suspected populations, which theoretically poses a dilution bias. However, our subgroup analysis showed no significant differences between suspected and confirmed cases. This aligns with clinical reality, as it is challenging to differentiate infectious from non-infectious causes based solely on initial symptoms, hence most RCTs enroll suspected patients. Therefore, mPOCT's core value lies in triaging these uncertain cases to guide antibiotic decisions. Restricting analysis to confirmed infections would limit its utility to mere pathogen typing. Thus, despite the inability to exclude non-infected individuals due to the trial-level nature of our data, our results reflect the pragmatic effectiveness of mPOCT in real-world diagnostic scenarios. Future individual patient data meta-analyses could offer further precision. Moreover, we did not adjust for multiple testing, which could increase false positives. However, most results are non-significant, and applying multiplicity corrections would not change this, reinforcing our conclusions. Finally, the inclusion of post-hoc exploratory outcomes may introduce a risk of outcome reporting bias and false positives. However, their results are consistent with the primary findings and indicate a negative effect. Nevertheless, these findings should be interpreted with caution and will require validation through high-quality prospective studies.

From a global health perspective, the high incidence of ARTIs means that even small reductions in antibiotic use through mPOCT can have significant benefits. Our findings suggest limited overall benefit in the general population, but promising results in specific groups like children. While guidelines recommend considering mPOCT for severe pneumonia and immunocompromised patients, research in these groups is limited.¹⁴ Future studies should focus on identifying the best patient populations and optimal timing to clarify its clinical value. Moreover, most current research emphasizes short-term outcomes; the long-term impact on antimicrobial resistance and infection control remains unclear. Future work should explore how mPOCT can help prevent cross-infections, detect outbreaks early, and support contact tracing. Ultimately, the real-world success of mPOCT depends on proper implementation, including standard procedures, accurate interpretation, integration with other diagnostics, and strong antimicrobial stewardship.^57,58 Education for clinicians, public awareness, and involvement of labs and infection control teams are also essential.^57,58 Ensuring equitable access, especially in low-resource settings, is crucial. However, mPOCT alone is not enough; it should be part of a comprehensive infection control strategy.

Moderate-quality evidence indicates that mPOCT has limited clinically meaningful effect on antibiotic use and treatment duration overall. High-quality evidence confirms this lack of impact in adults, suggesting routine mPOCT implementation in adult patients presenting with ARTIs is currently unwarranted. Although low-quality evidence suggests a potential reduction in antibiotic use among paediatric patients, this finding remains uncertain and requires confirmation through dedicated, high-quality trials. Our results highlight that the true value of mPOCT lies in its integration within broader diagnostic and antimicrobial stewardship strategies, supported by clinician education and biomarker-guided approaches, rather than as a standalone intervention. Future research should prioritise assessing the long-term impact of mPOCT and its cost-effectiveness to inform sustainable health policy decisions.

Contributors

QYL, QZ, ZXL, and LZ designed the systematic review and meta-analysis project. QYL, QZ, JBF, XFF, and HHL performed the literature search, screening, data extraction, and quality assessment. LZ, ZXL, and YLC participated in the resolution of discrepancies. QYL, QZ, and ZKY conducted data analysis. QYL and QZ drafted the manuscript. YLC, ZKY, FJS, JL, DCC, RK, DLT, JLT, JFT, AT, JDW, JC, JXJ, ZXL, and LZ participated in data interpretation and manuscript editing. All authors read and approved the final version of the manuscript. QYL, QZ, ZXL, and LZ directly accessed and verified the underlying data reported in the manuscript. QYL, ZXL, LZ, and JDW acquired funding. All authors had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Data sharing statement

All data utilised or synthesised to support the findings in this study are available within this manuscript and its Supplementary Information Files.

Declaration of interests

Jan J. De Waele has served as a consultant for Biomerieux, Grifols, Menarini, MSD, Pfizer, Roche Diagnostics, and Viatris, with all honoraria paid to his institution. Jean-François Timsit has served as a research consultant for Biomerieux and has lectured at symposia for Biomerieux and Qiagen; his University Hospital received a research grant from Biomerieux for an investigator-driven study.