Urea breath test for Helicobacter pylori infection in adult dyspeptic patients: A meta-analysis of diagnostic test accuracy

Abstract

BACKGROUND

Helicobacter pylori (H. pylori) infection has been well-established as a significant risk factor for several gastrointestinal disorders. The urea breath test (UBT) has emerged as a leading non-invasive method for detecting H. pylori. Despite numerous studies confirming its substantial accuracy, the reliability of UBT results is often compromised by inherent limitations. These findings underscore the need for a rigorous statistical synthesis to clarify and reconcile the diagnostic accuracy of the UBT for the diagnosis of H. pylori infection.

AIM

To determine and compare the diagnostic accuracy of ¹³C-UBT and ¹⁴C-UBT for H. pylori infection in adult patients with dyspepsia.

METHODS

We conducted an independent search of the PubMed/MEDLINE, EMBASE, and Cochrane Central databases until April 2022. Our search included diagnostic accuracy studies that evaluated at least one of the index tests (¹³C-UBT or ¹⁴C-UBT) against a reference standard. We used the QUADAS-2 tool to assess the methodological quality of the studies. We utilized the bivariate random-effects model to calculate sensitivity, specificity, positive and negative test likelihood ratios (LR+ and LR-), as well as the diagnostic odds ratio (DOR), and their 95% confidence intervals. We conducted subgroup analyses based on urea dosing, time after urea administration, and assessment technique. To investigate a possible threshold effect, we conducted Spearman correlation analysis, and we generated summary receiver operating characteristic (SROC) curves to assess heterogeneity. Finally, we visually inspected a funnel plot and used Egger’s test to evaluate publication bias.

RESULTS

The titles and abstracts of 4621 studies were screened; 79 articles were retrieved and selected for full-text reading. Finally, 60 studies were included in the diagnostic test accuracy meta-analysis. Our analysis demonstrates superior diagnostic accuracy of ¹³C-UBT over ¹⁴C-UBT, indicated by higher sensitivity (96.60% vs 96.15%), specificity (96.93% vs 89.84%), likelihood ratios (LR+ 22.00 vs 10.10; LR- 0.05 vs 0.06), and area under the curve (AUC; 0.979 vs 0.968). Notably, ¹³C-UBT's DOR (586.47) significantly outperforms ¹⁴C-UBT (DOR 226.50), making it the preferred diagnostic tool for dyspeptic individuals with H. pylori infection. Correlation analysis revealed no threshold effect (¹³C-UBT: r = 0.48; ¹⁴C-UBT: r = -0.01), and SROC curves showed consistent accuracy. Both ¹³C-UBT and ¹⁴C-UBT showed high AUC values (¹³C-UBT 0.979; ¹⁴C-UBT 0.968) near 1.00, reinforcing their excellent accuracy and endorsing both as reliable diagnostic tools in clinical practice.

CONCLUSION

In summary, our study has demonstrated that ¹³C-UBT has been found to outperform the ¹⁴C-UBT, making it the preferred diagnostic approach. Additionally, our results emphasize the significance of carefully considering urea dosage, assessment timing, and measurement techniques for both tests to enhance diagnostic precision. Nevertheless, it is crucial for researchers and clinicians to evaluate the strengths and limitations of our findings before implementing them in practice.

Key Words: Helicobacter pylori, Urea breath test, Diagnosis, Diagnostic test accuracy, Meta-analysis

Core Tip: The urea breath test (UBT) is a pivotal noninvasive method for detecting Helicobacter pylori (H. pylori); however, its reliability is challenging. This meta-analysis aimed to compare the precision of the ¹³C-UBT and ¹⁴C-UBT in diagnosing H. pylori among adults with dyspepsia, providing insights to enhance clinical strategies.

INTRODUCTION

Helicobacter pylori (H. pylori) is a spiral-shaped, gram-negative microaerophilic bacterium that infects approximately 43% of the global population[1]. While the majority of infected individuals remain asymptomatic, chronic gastritis inevitably ensues, leading to a significant burden of morbidity and mortality[2,3]. Adults who are infected with H. pylori are at increased risk of developing peptic ulcer disease, gastric cancer, and mucosa-associated lymphoid tissue lymphoma[4-6]. To address this, current guidelines advocate for either a test-and-treat or a scope-and-treat approach in managing uninvestigated dyspepsia, underscoring the importance of timely diagnosis and intervention[7,8].

Diagnostic testing for H. pylori infection typically involves two primary categories: Invasive (endoscopic) and non-invasive testing, depending on the application of upper endoscopy[9]. For individuals aged 50 years or older or those with alarm features, the recommended standard diagnostic approach involves upper endoscopy, followed by histopathological examination (HE) or rapid urease test (RUT), and occasionally, culture[8]. In contrast, in dyspeptic patients under 50 years without specific risk factors or alarm symptoms, non-invasive methods such as urea breath testing (UBT), stool antigen testing, and serology are preferred[8,10].

Among non-invasive diagnostic techniques, the UBT has emerged as a prominent method. This approach capitalizes on the urease activity of H. pylori, initiating the hydrolysis of ingested urea and consequent release of labeled carbon dioxide[11]. Two commonly utilized isotopic variants, ¹³C-UBT and ¹⁴C-UBT, offer distinctive features. In ¹³C-UBT, a stable isotope (carbon-13) is employed, and breath samples are collected and analyzed for labeled carbon dioxide using various methods such as mass spectrometry and infrared spectrometry[12]. This method presents important advantages, notably the absence of ionizing radiation, rendering it suitable for repeated application and applicable in vulnerable populations, including pregnant women and children[11,13]. In contrast, ¹⁴C-UBT utilizes a radioactive isotope (carbon-14) and primarily relies on scintillation counting for detection[14,15]. Despite its historical use, concerns regarding radiation exposure have diminished its preference in contemporary clinical practice.

In a prior meta-analysis, Ferwana et al[16] assessed the diagnostic accuracy of the UBT, encompassing both ¹³C-UBT and ¹⁴C-UBT, for detecting H. pylori infection in adult dyspeptic patients. Despite its high accuracy, the reliability of UBT results was constrained by significant unexplained heterogeneity, persisting even after subgroup analysis[16]. This pattern persisted in subsequent studies, with Zhou et al[17] finding analogous challenges in calculating pooled estimates of diagnostic accuracy for ¹⁴C-UBT. Moreover, a subsequent systematic review emphasized that the variability in thresholds and reference standards across studies limited the data available for pooling accuracy measures at specific UBT thresholds[18].

These findings underscore the need for a rigorous statistical synthesis to clarify and reconcile the diagnostic accuracy of the UBT for the diagnosis of H. pylori infection, addressing challenges identified in prior research. To address this gap in the evidence, we conducted a systematic review and meta-analysis to determine the diagnostic accuracy of the UBT for H. pylori infection in adult patients with dyspepsia.

MATERIALS AND METHODS

This study adhered to the guidelines outlined in the PRISMA-DTA[19]. These guidelines encompass a 27-item checklist and a 3-phase flowchart, both designed to enhance the transparency of systematic review reporting. Accordingly, our study protocol has been officially registered in the PROSPERO database under the registration number CRD42023449854.

Literature search

This search strategy was designed following the Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy (Version 2.0, 2022)[20]. We performed independent, computer-assisted searches of the: (1) PubMed/MEDLINE; (2) EMBASE; and (3) Cochrane Library databases. MeSH (Medical Subject Headings) and Emtree (EMBASE Subject Headings) index terms and free-text words were combined. Search terms included "urea breath test," "breath test," "13C-urea breath test," "14C-urea breath test," "13C-UBT," "14C-UBT," "Helicobacter pylori," "H. pylori," and "dyspepsia." Boolean operators (AND, OR) were also used to narrow or broaden the search as required. No language restriction was applied. To identify additional studies, reference lists were also scanned. Finally, we conducted a “citing reference” search (by searching articles which cited the included studies) in PubMed/MEDLINE and EMBASE. Following the search, all identified citations were collated and uploaded into the Rayyan (https://www.rayyan.ai/) tool, and all duplicates were removed.

Selection of studies

Two independent reviewers, Lemos FFB and Calmon MS, screened the references against predefined eligibility criteria. In the case of disagreement, a 3^rd researcher, Silva Luz M, was consulted. Full-text papers were obtained for references considered relevant. If any study was not retrieved, the authors were contacted. Two authors, Lemos FFB and Correa Santos GL, independently screened the full-text papers against the eligibility criteria. In the case of disagreement, consensus was reached.

We included diagnostic accuracy studies that evaluated at least one of the index tests (¹³C- or ¹⁴C-UBTs) against a reference standard (biopsy fragments followed by culture or HE or RUT and/or not serology/stool antigen-based tests in adult dyspeptic patients. Exclusion criteria were as follows: (1) Studies that enrolled children or adolescents under 18 years of age; (2) studies that included only patients with acute upper gastrointestinal bleeding; (3) studies that enrolled subjects who presented for reasons other than dyspeptic symptoms, complicated dyspeptic cases that need surgery, those who received previous therapy for H. pylori within the last 3 months, or long-term use of corticosteroids and immunosuppressant drugs; (4) screening studies; (5) studies that did not report true positive, false positive, false negative, and true negative data and the threshold used for the index tests; (6) case-control studies because these are prone to bias[21]; and (7) full-text articles not available or articles not available in English, Spanish, or Portuguese.

Data extraction and management

Two review authors, Rocha GR and Correa Santos GL, independently extracted data from each included study using a pre-piloted data extraction form. In case of discrepancies, a 3^rd researcher, Lemos FFB, was consulted. The extracted data included: (1) Information about the studies, such as the first author, publication year, and country; (2) details about the study design, including the type of study (prospective and retrospective cohort studies, cross-sectional studies, or randomized clinical trials), the reference standards used, blinding of the index test and reference standard, and the flow and timing (retrospective/prospective); (3) participant information, i.e., the total number of participants and population characteristics (age, mean ± SD, sex, and disease prevalence); (4) reference standard details, including the time interval between the index test and the reference standard; index test information, including the model (¹³C- or ¹⁴C-UBT), cut-off values, urea dosing, time for measurement after urea administration (min), and measurement technique; and (5) diagnostic accuracy data, including the number of true positives, false positives, false negatives, and true negatives.

Assessment of methodological quality

Two independent reviewers, Silva Luz M and de Oliveira Silva LG, conducted critical appraisal using the QUADAS-2 tool. In cases of disagreement, they consulted a 3^rd researcher, Lemos FFB. The QUADAS-2 tool is applied in four phases[21]: Summarizing the review question, tailoring the tool and producing review-specific guidance, constructing a flow diagram for the primary study, and evaluating bias and applicability. This tool comprises four domains: patient selection, index test, reference standard, and flow and timing. Each domain is assessed for the risk of bias, and the first three domains are also evaluated for concerns regarding applicability. It's important to note that "risk of bias" refers to internal validity, i.e., whether there are systematic errors in conducting the study with respect to the specific domain, while "applicability concern" pertains to external validity, i.e., whether there are concerns that the population, index test, or reference standard used in the studies align with the review question. Signaling questions were also included to assist in assessing the risk of bias.

Statistical analysis and data synthesis

Eligible studies were subjected to data extraction, and we organized the data into 2 × 2 tables. In our analysis, we selected only the optimal threshold values for H. pylori positivity in cases where multiple thresholds were presented. We added 0.5 to values equal to zero to ensure computational stability and prevent potential issues[22].

To address the anticipated diversity in meta-analyses of diagnostic accuracy studies, we utilized the random-effects model to calculate sensitivity, specificity, positive and negative test likelihood ratios (LR+ and LR-), as well as the diagnostic odds ratio (DOR)[23]. We also determined the corresponding 95% confidence intervals (95%CIs). The results of the ¹³C- and ¹⁴C-UBT are presented separately. Subgroup analyses were conducted with a focus on urea dosing, time for measurement after urea administration (in minutes), and the assessment technique employed. To investigate the possibility of a threshold effect in the analysis, we conducted Spearman correlation analysis. A substantial threshold effect was recognized when the correlation coefficient reached or exceeded 0.6[24].

We performed a bivariate random-effects meta-analysis and generated summary receiver operating characteristic (SROC) curves to visually assess heterogeneity. Furthermore, these curves allowed us to predict accuracy by summarizing diagnostic performance as the area under the curve (AUC)[25]. We categorized accuracy levels as follows: fail (0.50-0.60), poor (0.61-0.70), fair (0.71-0.80), good (0.81-0.90), and excellent (0.91-1.00)[22].

To evaluate publication bias, we conducted a visual inspection of a funnel plot and employed Egger’s tests for statistical assessment. The creation of this plot and the assessment of the risk of data due to missing data required a minimum of ten studies.

All analyses were performed using R version 4.2.1, an environment for statistical computing in Vienna, Austria, utilizing the “meta” package (version 6.5-0), “dmetar” package (version 0.1.0), and the “mada” package (version 0.5.11).

RESULTS

Study selection

Database searches initially yielded 10902 reports, from which duplicates were removed. No additional references were discovered through alternative search methods. Subsequently, the titles and abstracts of 4621 studies underwent screening, resulting in the retrieval and selection of 79 articles for full-text examination. Ultimately, 60 studies fulfilled the inclusion criteria. The reasons for exclusion were as follows: Incorrect population (n = 8), unsuitable reference standard (n = 4), insufficient accuracy measures (n = 2), missing threshold information (n = 2), failure to compare against the reference standard (n = 1), writing in a foreign language (n = 1), and involvement of the same sample (n = 1). Figure 1 illustrates the flow of information through various phases of the systematic review.

Figure 1 PRISMA 2020 flow diagram. This flowchart delineates the progression of information throughout various phases of the systematic review, illustrating the quantities of records identified, included, and excluded, along with the rationales for study exclusion.

General characteristics of included studies

Among all the studies, 39 (comprising 65%) employed the ¹³C-UBT as their primary diagnostic test, featuring a median population size of 200 individuals (lower to upper quartile: 84.5-254). Cross-sectional study design was predominant, making up 97.5% of the total, while only one study (2.5%) adopted a randomized controlled trial approach. For the ¹³C-UBT, the median pre-test probability was 51.2% (lower to upper quartile: 47.8-67.6). Various reference standards were used, with the most common being "H. pylori culture (HpC) or (HE and RUT)," accounting for 22.5% of cases. Other reference standards included "HE and RUT" (17.5%), "HE" (12.5%), "HE or HpC" (12.5%), "HE or (RUT and serology)" (5%), "RUT" (5%), "HE, HpC, and RUT" (5%), and "HpC" (2.5%). Some studies also combined reference standards, such as "RUT or HE" and “(HE, HpC, RUT) at least two positives," each constituting 2.5% of the sample, as shown in Table 1.

Table 1 Characteristics of the included studies that assessed the diagnostic test accuracy of the ¹³C-urea breath test.

Ref.	Country	Design	Population, n	Prevalence (%)	Reference standard	Index test (model)	Optimal cut-off	Urea dosing (mg)	Time after administration (min)	Measurement tecnique	TP	FP	TN	FN
Wang et al[29], 2021	China	Cross-sectional	217	65.9	HE	13C-UBT	10.4‰ DOB	50	30	IS	120	14	60	23
Alzoubi et al[33], 2020	Jordan	Cross-sectional	30	56.7	RUT or HE	13C-UBT	4‰ DOB	75	30	IS	16	3	10	1
Nawacki et al[34], 2018	Poland	Cross-sectional	50	36.0	RUT	13C-UBT	9.5‰ DOB	NR	30	IS	16	0	32	2
Som et al[27], 2014	India	Cross-sectional	83	59.0	RUT	13C-UBT	1.47‰	75	Multiple times	ICOS	49	0	34	0
Bruden et al[35], 2011	United States	Cross-sectional	280	53.2	HE or (HpC and RUT)	13C-UBT	7 DOB	NR	NR	NR	139	16	115	10
Peng et al [36], 2009	Taiwan	Cross-sectional	100	53.0	HpC or (HE and RUT)	13C-UBT	4.8‰ DOB	100	15	IS	53	7	40	0
Jordaan et al[37], 2008	South Africa	Cross-sectional	103	58.3	HE	13C-UBT	4.5‰ DOB	75	NR	GCMS	55	3	40	5
Gatta et al[26], 2006	Italy	RCT	100	43.0	HE and RUT	13C-UBT	4.40‰-6.26‰ DOB	25	30	IRMS	43	0	57	0
Peng et al[38], 2005	Taiwan	Cross-sectional	50	36.0	HpC or (HE and RUT)	13C-UBT	5‰ DOB	100	15	IRMS	18	0	32	0
Kato et al[39], 2004	Japan	Cross-sectional	254	51.1	HpC or (HE and RUT)	13C-UBT	2.5‰ DOB	100	20	IRMS	252	5	242	6
Ohara et al[40], 2004	Japan	Cross-sectional	254	51.2	HpC or (HE and RUT)	13C-UBT	2.5‰ DOB	100	Multiple times	IRMS	127	2	122	3
Chen et al[41], 2003	Taiwan	Cross-sectional	554	66.6	HpC or (HE and RUT)	13C-UBT	3.5‰ DOB	100	20	IS	361	6	179	8
Valdepérez et al[42], 2003	Spain	Cross-sectional	85	76.8	HE and RUT	13C-UBT	NR	100	30	NR	61	0	19	2
Gatta et al[43], 2003	Italy	Cross-sectional	200	56.5	HpC or (HE and RUT)	13C-UBT	3.11‰-6.84‰ DOB	75	30	IRMS	113	0	87	0
Wong et al[44], 2003	China	Cross-sectional	200	49.5	HE and RUT	13C-UBT	2.1‰ DOB	50	20	IRMS	99	0	101	0
Ng et al[45], 2002	China	Cross-sectional	213	54.9	HE and RUT	13C-UBT	4.0‰-6.5‰ DOB	75	30	IRMS	112	2	94	5
Wong et al[46], 2001	China	Cross-sectional	101	48.1	HE and RUT	13C-UBT	7.0-8.0‰ DOB	50	20	IRMS	99	4	103	0
Wong et al[47], 2001	China	Cross-sectional	294	55.4	HE, HpC, CLO (RUT), in-house RUT, PCR, UBT (at least four positive)	13C-UBT	5‰ DOB	75	30	IRMS	151	4	127	12
Shirin et al[48], 2001	United States	Cross-sectional	97	47.4	HE and RUT	13C-UBT	Positive: > 6‰ DOB (> 2 points)/negative: < 3‰ DOB (> 2 points)	75	5	MCS	45	2	49	1
Pilotto et al[49], 2000	Italy	Cross-sectional	96	51.0	HE, HpC, and RUT	13C-UBT	5‰ DOB	100	30	IRMS	49	2	45	0
Sheu et al[50], 2000	Taiwan	Cross-sectional	177	47.5	HE or HpC	13C-UBT	3.5 DOB	50	15	IS	81	1	92	3
Wong et al[51], 2000	China	Cross-sectional	202	56.4	HE and RUT	13C-UBT	4.5‰ DOB	75	30	IRMS	108	2	86	6
Hahn et al[52], 2000	United States	Cross-sectional	67	6.0	HE and at least two positives of (definitive presence of H. pylori organisms in HE, UBT, Serology)	13C-UBT	2.4‰ DOB	125	30	IRMS	4	9	54	0
Chen et al[53], 2000	Japan	Cross-sectional	162	83.3	HE and Serology	13C-UBT	2.5‰ DOB	100	20	IRMS	135	1	26	0
Peng et al[54], 2000	Taiwan	Cross-sectional	136	59.6	HpC or (HE and RUT)	13C-UBT	4.8‰ DOB	100	15	IRMS	76	6	49	5
Riepl et al[55], 2000	Germany	Cross-sectional	84	35.7	HE, HpC, and RUT	13C-UBT	6.5‰ DOB	75	15	IS	30	0	54	0
D'Elios et al[56], 2000	Italy	Cross-sectional	256	45.3	HE	13C-UBT	4‰ DOB	75	30	IRMS	113	2	138	3
van der Hulst et al[57], 1999	Italy	Cross-sectional	544	52.2	HE or HpC	13C-UBT	7.5‰ DOB ± 0.8	100	30	LOGES	260	21	239	24
Leodolter et al[58], 1999	Germany	Cross-sectional	320	48.1	HpC or (HE and RUT)	13C-UBT	4‰ DOB	75	30	IRMS	142	2	164	12
Mock et al[59], 1999	Canada	Cross-sectional	98	19.8	HE or (RUT and Serology)	13C-UBT	3‰ DOB	75	30	IRMS	17	2	75	2
Mock et al[59], 1999	Korea	Cross-sectional	107	68.2	HE or (RUT and Serology)	13C-UBT	3‰ DOB	75	30	IRMS	69	1	33	4
Perri et al[60], 1998	Belgium	Cross-sectional	172	73.3	HE or HpC	13C-UBT	1.15‰ DOB	75	60	IRMS	121	1	45	5
Ohara et al[61], 1998	Japan	Cross-sectional	213	77.5	HpC or at least two positives of (HE, RUT, Serology)	13C-UBT	2.5‰ DOB	100	20	IRMS	162	1	47	3
Leodolter et al[62], 1998	Germany	Cross-sectional	40	50.0	HpC or (HE and RUT)	13C-UBT	4‰ DOB	75	10	IRMS	20	0	20	0
Andersen et al[63], 1998	Denmark	Cross-sectional	97	54.6	HE or HpC	13C-UBT	5‰ DOB	100	Multiple times	IRMS	46	4	40	7
Ellenrieder et al[64], 1997	Germany	Cross-sectional	132	43.2	(HE, HpC, RUT) at least two positives	13C-UBT	3.5‰ DOB	NR	30	IS	52	8	67	5
Epple et al[65], 1997	Germany	Cross-sectional	126	61.1	HE	13C-UBT	1.3‰ DOB	75	30	IRMS	74	7	42	3
Labenz et al[66], 1996	Germany	Cross-sectional	70	67.1	HE or HpC	13C-UBT	4‰ DOB	75	30	IRMS	46	0	23	1
Logan et al[67], 1991	England	Cross-sectional	56	68.0	HE	13C-UBT	4.5‰ DOB	125	Multiple times	IRMS	32	1	15	2
Dill et al[68], 1990	Scotland	Cross-sectional	69	49.3	HpC	13C-UBT	3‰ c-PDR	250	20	IRMS	33	0	35	1

CLO: Campylobacter-like organism; GCMS: Gas chromatography-mass spectrometry; HE: Histopathological examination; HpC: Helicobacter pylori culture; ICOS: Integrated Cavity Output Spectrometry; IRMS: Isotope ratio mass spectrometry; IS: Infrared spectrometry; LOGES: Laser opto-galvanic Effect Spectroscopy; MCS: Molecular correlation spectrometry; NR: Not reported; RUT: Rapid urease test; UBT: Urea breath test; DOB: Delta over baseline; RCT: Randomized clinical trial.

On the other hand, the ¹⁴C-UBT accounted for 35% of the total (21 studies) with a median population size of 108.5 (lower to upper quartile: 63.5-125.5). For the ¹⁴C-UBT, the median pre-test probability was 64.9% (lower to upper quartile: 43.6-73.1). Various reference standards were employed in these studies, with "HE" being the most prevalent, accounting for 38.1% of cases. Other reference standards included "HE and RUT" (14.3%) and "HpC or HE" (9.5%). Some studies also used combinations of reference standards, such as "HE, RUT, Serology (at least two positive)" and "HpC or [HE and (RUT or Gram staining)]," each comprising 4.8% of the sample, as detailed in Table 2.

Table 2 Characteristics of the included studies that assessed the diagnostic test accuracy of the ¹⁴C-urea breath test.

Ref.	Country	Design	Population, n	Prevalence (%)	Reference standard	Index test (model)	Optimal cut-off	Urea dosing (μCi)	Time after administration (min)	Measurement tecnique	TP	TN	FP	FN
Han et al[30], 2023	China	Cross-sectional	205	42.4	HE and RUT	14C-UBT	100 dpm	0.75	20	SC	83	3	115	4
Wang et al[29], 2021	China	Cross-sectional	267	71.9	HE	14C-UBT	238 dpm	0.75	25	NR	158	12	63	34
Miftahussurur[69], 2021	Indonesia	Cross-sectional	55	23.6	HE	14C-UBT	57 cpm	1	10	HA	12	1	41	1
Cosgun et al[70], 2016	Turkey	Cross-sectional	126	92.1	HpC or HE	14C-UBT	NR	1	10	HA	112	7	3	4
Atli et al[71], 2012	Turkey	Cross-sectional	100	35.0	HE	14C-UBT	Positive: > 50 cpm/suspicious: 25-50 cpm /negative: < 25 dpm	1	10	HA	32	4	61	3
Alarcón-Rivera et al[72], 2011	Mexico	Cross-sectional	84	70.2	HE	14C-UBT	Positive: > 50 ppm/indeterminate: 25-50 ppm/negative: < 25 ppm	1	10-15	HA	56	1	24	3
Mansour-Ghanaei et al[73], 2011	Iran	Cross-sectional	125	56.8	HE, RUT, Serology (at least two positive)	14C-UBT	50 cpm	1	15	HA	67	0	54	4
Ozdemir et al[74], 2008	Turkey	Cross-sectional	89	66.3	HE, RUT, PCR (at least two positive)	14C-UBT	Positive: > 50 cpm/equivocal: 25-50 cpm; negative: < 25 dpm	1	10	HA	57	0	30	2
Rasool et al[75], 2007	Pakistan	Cross-sectional	94	64.9	RUT	14C-UBT	50 cpm	1	10	β-SC	60	3	30	1
Gurbuz et al[76], 2005	Turkey	Cross-sectional	65	44.6	HE	14C-UBT	Positive: > 50 cpm/suspicious: 25-50 cpm /negative: < 25 dpm	1	10	HA	26	8	28	3
Gatta et al[31], 2003	Italy	Cross-sectional	117	65.0	HpC or (HE and RUT)	14C-UBT	130-136 dpm (dpm at sample-dpm at T0)	1	12.5	LSC	73	2	39	3
González et al[77], 2003	Chile	Cross-sectional	NR	71.9	Two or more positives	14C-UBT	200 dpm	1	10	LSC	61	14	11	3
Oztürk et al[78], 2009	Turkey	Cross-sectional	75	65.8	HE	14C-UBT	100 dpm	1	10	LSC	48	5	20	0
Gomes et al[79], 2002	Brazil	Cross-sectional	137	83.9	HE and RUT	14C-UBT	1000 cpm	5	15	LSC	114	1	21	1
Desroches et al[80], 1997	Canada	Cross-sectional	56	80.4	HE or HpC	14C-UBT	0.33‰ AS (14CO2 specific activity)	5	20	LSC	44	0	11	1
Allardyce et al[81], 1997	New Zealand	Cross-sectional	63	34.9	HE and RUT	14C-UBT	49 dpm	1	30	β-SC	22	2	39	0
Faigel et al[82], 1996	United States	Cross-sectional	50	42.6	HE or RUT	14C-UBT	Positive: > 200 dpm in any sample/borderline: 100-200 dpm (as the peak count)/negative: < 100 dpm (in all samples)	1	Multiple times	LSC	18	1	26	2
Goh et al[83], 1995	Malaysia	Cross-sectional	63	50.8	HpC or [HE and (RUT or Gram staining)]	14C-UBT	1275 dpm	5	15	LSC	32	0	31	0
Kao et al[84], 1993	China	Cross-sectional	184	53.8	HpC or RUT	14C-UBT	150‰	5	10	LSC	99	14	71	0
Vivas et al[85], 1993	Venezuela	Cross-sectional	15	53.3	HE	14C-UBT	100 dpm	1	20	β-SC	8	1	6	0
Novis et al[28], 1991	Israel	Cross-sectional	64	80.3	HE	14C-UBT	4,7‰	10	Multiple times	LSC	59	3	12	2

CLO: Campylobacter-like organism; HA: Heliprobe Analyser; HE: histopathological examination; HpC: Helicobacter pylori culture; LSC: Liquid scintillation counting; NR: Not reported; RUT: Rapid urease test; SC: Solid scintillation counting; UBT: Urea breath test; β-SC: Beta-scintillation counting.

Methodological quality assessment

Supplementary Figure 1 and Figure 2 provide a visual representation of the comprehensive methodological quality assessment of the included studies. In the patient selection domain, 22 studies (35.5%) were categorized as having a low risk of bias, 36 studies (58.1%) were associated with a high risk of bias, and 2 studies (3.2%) were considered to have an unclear risk of bias. In terms of patient selection applicability, 40 studies (64.5%) exhibited low concern, 17 studies (27.4%) showed high concern, and 3 studies (4.8%) had unclear concern.

Figure 2 QUADAS-2 methodological quality graph. The QUADAS-2 methodological quality graph consists of four sections, each representing one of the key domains assessed.

Within the index test selection domain, 33 studies (53.2%) were rated as having a low risk of bias, 26 studies (41.9%) were identified with a high risk, and 1 study (1.6%) had an unclear risk of bias. Concerning index test applicability, 57 studies (91.9%) displayed low concern, while 3 studies (4.8%) raised high concern.

In the reference standard domain, 47 studies (75.8%) demonstrated a low risk of bias, 12 studies (19.4%) showed a high risk of bias, and 1 study (1.6%) had an unclear risk of bias. Notably, none of the studies raised concerns about the applicability of the reference standard.

Lastly, in the flow and timing domain, 23 studies (37.1%) were associated with a low risk of bias, 23 studies (37.1%) exhibited a high risk of bias, and 7 studies (11.3%) had an unclear risk of bias.

Overall accuracy of the ¹³C-UBT

The ¹³C-UBT test was evaluated for its diagnostic accuracy in 39 studies via a comprehensive meta-analysis The results demonstrated a high sensitivity of 96.60% (95%CI: 95.64-97.56; P value < 0.01; I² = 65.0%) and an equally impressive specificity of 96.93% (95%CI: 96.04-97.82; P value < 0.01; I² = 58.0%) for this test (Figure 3). Additionally, the DOR was calculated at 586.47 (95%CI: 340.03-1011.51), with a positive likelihood ratio (LR+) of 22.00 (95%CI: 15.60-30.10) and a negative likelihood ratio (LR-) of 0.05 (95%CI: 0.04-0.06) as presented in Supplementary Table 1.

Figure 3 Forest plot for studies based on the ¹³C-urea breath test for Helicobacter pylori infection in dyspeptic patients. A: Forest plot for overall sensitivity; B: Forest plot for overall specificity. 95%CI: 95% confidence interval.

Subgroup analysis of the ¹³C-UBT - Urea dosing

Among the thirty-six studies that documented the urea dosage, a 25 mg urea dose demonstrated notably high sensitivity (98.85%; 95%CI: 95.68-100.00) and specificity (99.13%; 95%CI: 96.73-100.00), as illustrated in Supplementary Figure 2. Increasing the urea dose to 50 mg across four studies resulted in a sensitivity of 95.28% (95%CI: 88.51-100.00) and a specificity of 94.91% (95%CI: 87.67-100.00). Seventeen studies explored the use of 75 mg of urea in the ¹³C-UBT, revealing a sensitivity of 96.47% (95%CI: 95.14-97.79) and a specificity of 98.33% (95%CI: 97.59-99.07). In cases where 100 mg of urea was used (in 12 studies), the ¹³C-UBT demonstrated a sensitivity of 97.31% (95%CI: 95.92-98.70) and a specificity of 96.08% (95%CI: 94.34-97.82). Two studies employing 125 mg of urea showed a sensitivity of 93.76% (95%CI: 86.13-100.00) and a specificity of 88.66% (95%CI: 81.07-96.25). Lastly, in a single study using 250 mg of urea, the ¹³C-UBT exhibited a sensitivity of 97.06% (95%CI: 91.38-100.00) and a specificity of 98.59% (95%CI: 94.72-100.00).

Subgroup analysis of the ¹³C-UBT - Time for assessment after urea administration

Among the 36 studies that provided information on the time after urea administration, optimal sensitivity (98.87%; 95%CI: 98.14-99.60) and specificity (98.14%; 95%CI: 96.98-99.30) were achieved when the assessment was conducted 20 min after urea administration [in 7 studies (Supplementary Figure 3)]. Notably, there were variations in sensitivity and specificity for different time intervals following urea administration.

For tests conducted 5 min post-urea administration (in one study), sensitivity was 97.83% (95%CI: 93.61-100.0), and specificity was 96.08% (95%CI: 90.75-100.00). Tests performed 10 min after urea administration (based on one study) yielded a sensitivity of 97.56% (95%CI: 90.88-100.0) and a specificity of 97.56% (95%CI: 90.88-100.00).

Similarly, in the case of tests carried out at 15 min post-urea administration (as reported in five studies), sensitivity averaged at 97.61% (95%CI: 95.68-99.55), with specificity at 95.85% (95%CI: 91.33-100.00). Longer intervals, such as 30 min and 60 min, as well as tests conducted at multiple time points after urea administration, displayed some variability. For instance, tests performed 30 min after urea administration (in 19 studies) had a sensitivity of 95.15% (95%CI: 93.30-96.92) and a specificity of 96.18% (95%CI: 94.48-97.87). A single study conducting tests 60 min post-urea administration reported a sensitivity of 96.03% (95%CI: 92.62-99.44) and a specificity of 97.83% (95%CI: 93.61-100.00). In the case of four studies investigating multiple time points after urea administration, the sensitivity was 96.13% (95%CI: 92.13-100.0), and the specificity was 97.95% (95%CI: 96.08-99.81).

Subgroup analysis of the ¹³C-UBT - Assessment technique

In our analysis of 38 studies that included data on the ¹³C-UBT assessment technique, Integrated Cavity Output Spectrometry (ICOS) for measuring CO₂ Isotope Ratios exhibited exceptional performance. ICOS demonstrated a sensitivity of 98.99% (95%CI: 96.20-100.00) and a specificity of 98.55% (95%CI: 94.56-100.00), as visualized in Supplementary Figure 4. In contrast, Infrared spectrometry, assessed in 8 studies, displayed a sensitivity of 94.72% (95%CI: 90.91-98.54) and a specificity of 98.55% (95%CI: 88.17-98.22).

Gas chromatography-mass spectrometry, investigated in a single study, yielded a sensitivity of 91.67% (95%CI: 84.67-98.66) and a specificity of 93.02% (95%CI: 85.41-100.00). Isotope-ratio mass spectrometry, scrutinized in 17 studies, demonstrated a sensitivity of 97.37% (95%CI: 96.45-98.28) and a specificity of 98.38% (95%CI: 84.67-98.66). Molecular correlation spectrometry, examined in a solitary study, exhibited a sensitivity of 97.83% (95%CI: 93.61-100.00) and a specificity of 96.08% (95%CI: 90.75-100.00). Similarly, Laser opto-galvanic effect spectroscopy, reported in one study, recorded a sensitivity of 91.65% (95%CI: 88.31-94.78) and a specificity of 91.92% (95%CI: 88.61-98.21).

Overall accuracy of the ¹⁴C-UBT

A total of 21 studies investigated the diagnostic accuracy of the ¹⁴C-UBT, revealing a combined sensitivity of 96.15% (95%CI: 94.47-97.82; P value < 0.01; I² = 62.0%) and specificity of 89.84% (95%CI: 84.90-94.77; P value < 0.01; I² = 78.0%), as depicted in Figure 4. Within this dataset, a DOR of 226.50 (95%CI: 102.57-500.15), a positive likelihood ratio (LR+) of 10.10 (95%CI: 5.74-16.90), and a negative likelihood ratio (LR-) of 0.06 (95%CI: 0.04-0.08) were observed, as summarized in Supplementary Table 1.

Figure 4 Forest plot for studies based on the ¹⁴C-urea breath test for Helicobacter pylori infection in dyspeptic patients by time after urea administration. A: Forest plot for overall sensitivity; B: Forest plot for overall specificity. 95%CI: 95% confidence interval.

Subgroup analysis of the ¹⁴C-UBT - Urea dosing

Twenty-one studies investigated varying urea dosages in the context of the ¹⁴C-UBT. Among these, the use of a 5 µCi marked urea dose, as examined in four studies, demonstrated exceptional sensitivity (99.21%; 95%CI: 98.20-100.00) and specificity (93.43%; 95%CI 86.45-100.00), as depicted in Supplementary Figure 5. Elevating the urea dose to 10 µCi, as explored in a single study, resulted in a sensitivity of 96.72% (95%CI: 92.15-100.00) and a specificity of 80.00% (95%CI: 56.76-100.00). Conversely, when employing 1 µCi of marked urea (in 14 studies), the ¹⁴C-UBT exhibited a sensitivity of 96.78% (95%CI: 95.46-98.09) and a specificity of 87.19% (95%CI: 59.76-95.81). Lastly, two studies using 0.75 µCi of urea reported a sensitivity of 88.94% (95%CI: 76.10-100.00) and a specificity of 91.32% (95%CI: 78.18-100.00).

Subgroup analysis of the ¹⁴C-UBT - Time for measurement after marked urea administration

When considering the time for measurement after urea administration, an analysis of all included studies consistently revealed the highest sensitivity (98.39%; 95%CI: 96.36-100.00) and specificity (98.71%; 95%CI: 96.58-100.00) when the tests were conducted 15 minutes after urea administration, as illustrated in Supplementary Figure 6.

In studies conducted shortly after urea administration (within 10 minutes, n = 9), the sensitivity was consistently high at 97.83% (95%CI: 96.34-99.33), while specificity was somewhat lower at 79.90% (95%CI: 66.15-93.65). A single study, conducted at 12.5 minutes post-administration, reported a sensitivity of 96.05% (95%CI: 91.67-100.00) and a specificity of 95.12% (95%CI: 88.53-100.00). Studies conducted between 10- and 15-minutes post-urea administration (n = 3) showed a sensitivity of 94.92% (95%CI: 89.31-100.00) and a specificity of 96.00% (95%CI: 88.32-100.00).

However, longer intervals (20, 25, and 30 min), as well as tests conducted at various time points after urea administration, exhibited more variability. For instance, studies conducted at 20 min post-administration (n = 3) showed a sensitivity of 96.52% (95%CI: 93.50-97.55) and a specificity of 97.23% (95%CI: 94.48-99.97). A single study conducted at 25 min post-urea administration reported a sensitivity of 82.29% (95%CI: 76.89-87.69) and a specificity of 84.00% (95%CI: 75.70-92.30). A study conducted at 30 minutes post-administration yielded a sensitivity of 97.78% (95%CI: 91.69-100.00) and a specificity of 95.12% (95%CI: 88.53-100.00). In the case of two studies that investigated multiple time points after urea administration, the sensitivity was 96.03% (95%CI: 91.79-100.00), and the specificity was 91.02% (95%CI: 76.07-100.00).

Subgroup analysis of the ¹⁴C-UBT - Assessment technique

In the assessment of 20 studies with available data on the assessment technique, it was observed that liquid scintillation counting yielded a higher sensitivity of 98.79% (95%CI: 97.90-99.69) while maintaining a specificity of 87.24% (95%CI: 77.69-96.79). Conversely, Solid Scintillation UBT (scintillation counting) demonstrated higher specificity, reaching 97.46% (95%CI: 94.62-100.00), with a sensitivity of 95.40% (95%CI: 91.00-99.80), as illustrated in Supplementary Figure 7.

In contrast, the Heliprobe Analyser, assessed in 7 studies, displayed a sensitivity of 95.41% (95%CI: 93.32-97.50) and a specificity of 88.10% (95%CI: 74.43-100.00). Ultimately, the use of Beta-scintillation counter for the assessment of ¹⁴C-UBT resulted in a sensitivity of 98.11% (95%CI: 95.33-100.00) and a specificity of 93.47% (95%CI: 88.11-98.82).

Threshold effect and SROC curve

Spearman’s correlation analysis for studies evaluating ¹³C-UBT revealed a correlation coefficient (r) of 0.48, indicating the absence of a threshold effect. Similarly, ¹⁴C-UBT studies exhibited a negligible correlation (r = -0.01), also suggesting the absence of a threshold effect. Visual inspection of the SROC curves did not reveal any significant heterogeneity. Both the ¹³C-UBT (AUC = 0.979; Figure 5A) and the ¹⁴C-UBT (AUC = 0.968; Figure 5B) displayed excellent diagnostic accuracy.

Figure 5 Summary operating characteristics curve curves for studies based on the ¹³C-urea breath test and the ¹⁴C-urea breath test for Helicobacter pylori infection in dyspeptic patients. The summary operating characteristics curve (SROC) curve is a graphical representation that combines sensitivity and specificity data from multiple studies or diagnostic tests. It illustrates how these measures change with different threshold settings or study parameters. The curve is accompanied by the area under the curve (AUC), which provides a quantitative assessment of the test's overall performance. A higher AUC value indicates better discriminatory ability across tested thresholds. Furthermore, the diagnostic odds ratio (DOR) is derived from the ROC curve and offers an evaluation of the test's diagnostic precision. A higher DOR signifies stronger discriminatory power, reflecting the odds of a positive test result in individuals with the condition compared to those without it. A: SROC curve for studies based on the ¹³C-urea breath test (UBT) for Helicobacter pylori (H. pylori) infection in dyspeptic patients; B: SROC curve for studies based on the ¹⁴C-UBT for H. pylori infection in dyspeptic patients. SROC: Summary operating characteristics curve; AUC: Area under the curve; DOR: Diagnostic odds ratio.

Publication bias

The funnel plot visualization exposed asymmetry in both the ¹³C-UBT (Figure 6A) and ¹⁴C-UBT (Figure 6B) models. Additionally, Egger's test confirmed the presence of publication bias in both tests. The intercept was 2.54 with a P value < 0.001 for ¹³C-UBT and 3.04 with a P value < 0.001 for ¹⁴C-UBT.

Figure 6 Funnel plots for studies based on ¹³C-urea breath test and ¹⁴C-urea breath test for Helicobacter pylori infection in dyspeptic patients. A: Funnel plot for studies based on ¹³C-urea breath test (UBT); B: Funnel plot for studies based on ¹⁴C-UBT.

DISCUSSION

Insights from ¹³C- and ¹⁴C-UBT performance analysis

Our analysis has revealed that the ¹³C-UBT outperforms the ¹⁴C-UBT in terms of diagnostic accuracy, as evidenced by the following values: DOR, Likelihood Ratios (LR+ and LR-), and AUC values. Specifically, the ¹³C-UBT has sensitivity and specificity values of 96.60% (95%CI: 95.64-97.56; P value < 0.01; I² = 65.0%) and 96.93% (95%CI: 96.04-97.82; P value < 0.01; I² = 58.0%), respectively. In contrast, the ¹⁴C-UBT has sensitivity and specificity values of 96.15% (95%CI: 94.47-97.82; P value < 0.01; I² = 62.0%) and 89.84% (95%CI: 84.90-94.77; P value < 0.01; I² = 78.0%). The LR+ values for the ¹³C-UBT and ¹⁴C-UBT are 22.00 and 10.10, respectively, indicating the likelihood of positive results in individuals with H. pylori infection. Conversely, the LR- values, suggesting a reduced likelihood of negative test results for individuals with the infection, are 0.05 for the ¹³C-UBT and 0.06 for the ¹⁴C-UBT.

Furthermore, the DOR values show a substantial difference between the two tests. The ¹³C-UBT yields a significantly higher DOR of 586.47 compared to the ¹⁴C-UBT's DOR of 226.50. These results indicate that the ¹³C-UBT is statistically superior at distinguishing dyspeptic individuals with and without H. pylori infection, making it the preferred diagnostic tool in this clinical context.

Finally, it is essential to emphasize that our correlation analysis, utilizing both the ¹³C-UBT (r = 0.48) and the ¹⁴C-UBT (r = -0.01), yielded no evidence of a threshold effect. Visual examination of the SROC curves revealed no heterogeneity, indicating consistent accuracy assessments across the studies. Additionally, both the ¹³C-UBT and the ¹⁴C-UBT displayed remarkably high AUC values: 0.979 for the ¹³C-UBT and 0.968 for the ¹⁴C-UBT, which approaching 1.00 reinforces the excellent accuracy of these tests in detecting H. pylori infection in individuals with dyspepsia. These findings strongly support the reliability of the ¹³C-UBT and the ¹⁴C-UBT as valuable diagnostic tools in clinical practice.

¹³C-UBT performance: Urea dose, assessment timing, and measurement technique selection

Our analysis highlights the critical importance of selecting the appropriate urea dose when conducting the ¹³C-UBT for diagnosing H. pylori infection. While the 25 mg urea dose displays the highest sensitivity (98.85%) and specificity (99.13%), concerns regarding the generalizability of these results arise due to the fact that these findings are primarily based on a single study[26]. In contrast, the use of 75 mg and 100 mg doses is supported by a larger body of evidence, maintaining excellent diagnostic accuracy with sensitivity and specificity exceeding 96%. Conversely, higher doses, such as 125 mg or 250 mg, exhibit a modest reduction in accuracy, particularly in terms of specificity. These findings strongly advocate for the consideration of 75 mg and 100 mg doses when aiming to optimize both sensitivity and specificity.

A crucial factor affecting the performance of the ¹³C-UBT is the timing of the assessment following urea administration. Our observations reveal that the optimal sensitivity and specificity, both exceeding 98%, are achieved at the 20-minute mark post-urea administration. Tests conducted at shorter intervals, such as 5 min and 10 min, also demonstrate high sensitivity and specificity, albeit slightly lower than the 20-min assessment. Conversely, assessments at 15 min maintain excellent accuracy, with sensitivity close to 98% and specificity around 95%. However, assessments at longer intervals, such as 30 min, 60 min, and multiple time points, exhibit some variability, with sensitivity and specificity values slightly lower than the 20-min assessment. These results highlight the 20-min assessment as the most reliable time point, offering a balance between high sensitivity and specificity. Nevertheless, the test remains accurate when conducted at shorter intervals.

The choice of assessment technique is also crucial for test accuracy. ICOS is the most accurate technique, with a sensitivity of 98.99% and a specificity of 98.55%. However, it is important to note that ICOS was evaluated in a single study[27], potentially limiting the generalizability of these results. To address this limitation, Isotope-ratio mass spectrometry is a more advisable option. In contrast, Infrared spectrometry, gas chromatography-mass spectrometry, isotope-ratio mass spectrometry, molecular correlation spectrometry, and Laser opto-galvanic effect spectroscopy yield varying levels of sensitivity and specificity. These findings underscore the significance of selecting the right assessment technique. While ICOS may be preferred when available due to its exceptional accuracy, other factors such as cost, availability, and local expertise should also be considered when making this choice.

¹⁴C-UBT performance: Urea dose, assessment timing, and measurement technique selection

Our research indicates that the urea dosage utilized in the ¹⁴C-UBT can also impact test accuracy. Specifically, a urea dose of 5 µCi was examined in four studies and was found to possess exceptional sensitivity (99.21%) and specificity (93.43%). These findings underscore the potential benefits of employing a 5 µCi dose for the ¹⁴C-UBT, as it offers a high level of accuracy in detecting H. pylori infection. However, increasing the urea dose to 10 µCi, as investigated in a single study[28], resulted in a slightly lower sensitivity (96.72%) and a specificity of 80.00%. This suggests that while higher urea dosages may still provide reliable results, they may be associated with a decrease in specificity, which could lead to more false-positive results.

On the other hand, the use of 1 µCi of marked urea, which was the most commonly used dosage in 14 studies, resulted in a sensitivity of 96.78% and a specificity of 87.19%. This indicates that a 1 µCi dose remains a viable option for the ¹⁴C-UBT, offering a good balance between sensitivity and specificity. Two recent studies using 0.75 µCi of urea reported a sensitivity of 88.94% and a specificity of 91.32%, suggesting that even lower urea doses can provide reasonable diagnostic accuracy[29,30].

Regarding the time for measurement, tests conducted 15 min after urea administration consistently exhibited the highest sensitivity (98.39%) and specificity (98.71%). This indicates that the 15-min time point is optimal for maximizing test accuracy. Tests conducted within 10 min post-administration maintained high sensitivity (97.83%) but had a somewhat lower specificity (79.90%). A single study conducted at 12.5 min post-administration reported favorable sensitivity (96.05%) and specificity (95.12%)[31]. In contrast, longer intervals (20, 25, and 30 min) showed more variability, with varying levels of sensitivity and specificity. This suggests that measurements taken beyond 15 min may not be as reliable for H. pylori detection. Clinicians should carefully consider the timing of the ¹⁴C-UBT to ensure accurate results, with a preference for the 15-min mark when feasible.

Lastly, our analysis of assessment techniques uncovered differences in sensitivity and specificity. Liquid scintillation counting demonstrated the highest sensitivity (98.79%) but had a specificity of 87.24%. In contrast, Solid Scintillation UBT (scintillation counting) showed higher specificity (97.46%) at the expense of sensitivity (95.40%). The Heliprobe Analyser and Beta-scintillation counter also demonstrated moderate sensitivity and specificity. When choosing the assessment technique, the trade-off between sensitivity and specificity should be considered in relation to the clinical context. For instance, if high sensitivity is paramount to avoid missing positive cases, liquid scintillation counting may be the preferred method. Conversely, if high specificity is crucial to minimize false positives, solid scintillation counting could be a better choice.

Strengths and limitations

This meta-analysis adhered to established guidelines and rigorous methodological principles, enhancing the validity and reliability of our findings. We used a bivariate random-effects model to calculate sensitivity, specificity, likelihood ratios, and the DOR, alongside generating SROC curves for a comprehensive statistical analysis of the included studies. Subgroup analyses based on urea dosing, measurement timing, and assessment technique were conducted to explore potential sources of variation, while Spearman correlation analysis was used to assess the threshold effect's impact on diagnostic accuracy. Additionally, we assessed publication bias through visual inspections of funnel plots and Egger's tests.

However, it's important to acknowledge inherent limitations in our analysis. These include potential language bias, reliance on available data, and challenges associated with the inherent heterogeneity in diagnostic accuracy studies. Although we did not impose language restrictions in our search, the inclusion of studies conducted in English, Spanish, or Portuguese may introduce language bias[32]. The exclusion of studies due to unavailability of full-text articles or articles not in these specified languages could potentially lead to the omission of essential data.

Furthermore, the quality of our meta-analysis is closely tied to the quality of the primary studies we included. Biases within these primary studies can affect our analysis outcomes. In particular, we have concerns regarding the inclusion of patients, as there was no reported consecutive patient inclusion in some studies, and the index test was not always performed using a pre-specified threshold. Moreover, the diversity in diagnostic accuracy studies can present challenges when consolidating results, and despite subgroup analyses, residual heterogeneity may impact the broad applicability of our findings. Encouragingly, the visual examination of the SROC curves indicates consistent accuracy assessments across the included studies. Nevertheless, it is imperative to underscore that the reliability of our meta-analysis hinges on the data provided in these included studies. The absence or inconsistency of critical data points can significantly affect the precision of our analysis. Researchers and clinicians should consider these strengths and limitations when applying our findings in their practice.

CONCLUSION

In summary, our study offers crucial insights for selecting optimal diagnostic methods to detect H. pylori infection in clinical settings. We found that the ¹³C-UBT outperforms the ¹⁴C-UBT in terms of diagnostic accuracy, making it the preferred diagnostic approach. Furthermore, our findings highlight the significance of precise considerations when choosing urea dosage, assessment timing, and measurement techniques for both the ¹³C-UBT and ¹⁴C-UBT, thus enhancing diagnostic precision. These insights provide practical guidance to healthcare practitioners when choosing the most suitable diagnostic method for H. pylori infection, tailored to their specific clinical context. Factors like diagnostic accuracy, cost, and availability should be carefully weighed in this decision-making process. Our findings also have the potential to contribute significantly to the standardization of testing procedures, ensuring consistent and reliable results, especially for patients with dyspepsia or suspected H. pylori infection. Nevertheless, it's essential for researchers and clinicians to consider the strengths and limitations when applying our findings in their practice.

ARTICLE HIGHLIGHTS

Research background

The urea breath test (UBT) has become a widely accepted non-invasive method for detecting Helicobacter pylori (H. pylori). While numerous studies have confirmed its high accuracy, its reliability is often hindered by inherent limitations.

Research motivation

In a previous investigation, the diagnostic accuracy of the UBT, which encompasses both ¹³C-UBT and ¹⁴C-UBT, was evaluated in adult patients with dyspepsia to determine the presence of H. pylori infection. Although the test demonstrated a high degree of precision, its reliability was compromised by significant and unexplained heterogeneity, which persisted even after conducting subgroup analyses. This trend continued in subsequent studies, with similar challenges encountered in determining pooled estimates of diagnostic accuracy for ¹⁴C-UBT. Furthermore, a subsequent systematic review revealed that the variability in thresholds and reference standards across studies limited the available data for pooling accuracy measures at specific UBT thresholds. These findings underscore the need for a rigorous statistical synthesis to clarify and reconcile the diagnostic accuracy of the UBT for the diagnosis of H. pylori infection, addressing challenges identified in prior research.

Research objectives

To evaluate and contrast the diagnostic accuracy of ¹³C-UBT and ¹⁴C-UBT for H. pylori infection in adult patients with dyspepsia.

Research methods

We conducted independent searches of PubMed/MEDLINE, EMBASE, and Cochrane Central databases until April 2022, focusing on diagnostic accuracy studies that evaluated at least one of the index tests (¹³C-UBT or ¹⁴C-UBT) against a reference standard. We utilized the QUADAS-2 tool to assess the methodological quality of the studies, and we calculated sensitivity, specificity, positive and negative test likelihood ratios (LR+ and LR-), as well as the diagnostic odds ratio (DOR) and their 95% confidence intervals using the bivariate random-effects model. We conducted subgroup analyses based on urea dosing, time after urea administration, and assessment technique. To investigate a possible threshold effect, we conducted Spearman correlation analysis, and we generated summary receiver operating characteristic (SROC) curves to assess heterogeneity. Lastly, we visually inspected a funnel plot and used Egger’s test to evaluate publication bias.

Research results

A screening of 4621 studies led to the selection of 60 articles for inclusion in a diagnostic test accuracy meta-analysis after full-text reading. Our analysis highlights the superior diagnostic accuracy of ¹³C-UBT compared to ¹⁴C-UBT, as evidenced by higher sensitivity (96.60% vs 96.15%), specificity (96.93% vs 89.84%), likelihood ratios (LR+ 22.00 vs 10.10; LR- 0.05 vs 0.06), and AUC values (0.979 vs 0.968). Particularly noteworthy is the significantly higher DOR of ¹³C-UBT (586.47) compared to ¹⁴C-UBT (DOR 226.50), establishing ¹³C-UBT as the preferred diagnostic tool for individuals with dyspepsia and H. pylori infection. Correlation analysis indicated no threshold effect for both ¹³C-UBT (r = 0.48) and ¹⁴C-UBT (r = -0.01), and the SROC curves consistently demonstrated accurate performance for both tests. The high AUC values (¹³C-UBT: 0.979; ¹⁴C-UBT: 0.968), nearing 1.00, further affirm the excellent accuracy of both UBT variants, solidifying their reliability as diagnostic tools in clinical practice.

Research conclusions

Our study establishes ¹³C-UBT as the superior diagnostic approach over ¹⁴C-UBT. Furthermore, our findings underscore the critical importance of meticulously considering factors such as urea dosage, assessment timing, and measurement techniques for both tests to optimize diagnostic accuracy. However, it is paramount for researchers and clinicians to thoroughly evaluate the strengths and limitations of our conclusions before integrating them into clinical practice.

Research perspectives

Future research should focus on improving the comprehension, practicality, and dependability of UBTs for H. pylori infection. This endeavor involves refining techniques, examining sources of variability, exploring threshold effects, conducting longitudinal and comparative investigations, addressing biases, and assessing cost-effectiveness.