Diagnostic value of BinaxNOW antigen card for Severe Acute Respiratory Syndrome Coronavirus 2

ABSTRACT

Rapid laboratory detection is remarkably crucial to diagnosing coronavirus disease 2019 (COVID-19) infection, due to whose outbreak causes to the world pandemic. The BinaxNOW antigen card (BinaxNOW) is a simple, effective, and cheap tool to detect severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). The meta-analysis in this study was conducted to evaluate the diagnostic performance of BinaxNOW for SARS-CoV-2. The researchers independently retrieved the related databases (PubMed, Embase, Web of Science, Cochrane Library) before May 1st, 2021, and extracted the relevant data based on the early inclusion/exclusion criterion. Quality Assessment of Diagnostic Accuracy Study-2 was used to evaluate the quality of the enrolled studies. Stata 16.0, Meta-DiSc 1.4, and Review Manager 5.3 were used to generate analytical data for the statistical analysis. 59 sets of data were identified from the seven studies included in this meta-analysis. The combined sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, diagnostic odds ratio, and their 95% confidence intervals were 0.77 (0.76 to 0.79), 0.99 (0.99 to 0.99), 65.72 (48.23 to 89.56), 0.23 (0.19 to 0.28), and 461.10 (281.55 to 755.13), respectively. The area under curve was 0.9910 in the summary receiver operating characteristic curve. BinaxNOW is beneficial for symptomatic patients’ onset within 7 days. CT value and testing site may be the heterogeneity source of BinaxNOW accuracy. Moreover, this technology has an efficient performance for diagnosing COVID-19, especially in patients with heavy viral load. BinaxNOW may become a practical tool for large-scale or at-home use for COVID-19 in the post-pandemic era.Highlights● Pooled sensitivity with 0.77 and specificity with 0.99 in the BinaxNOW assay.● CT value and testing site may be the heterogeneity source of BinaxNOW accuracy.● BinaxNOW is beneficial for symptomatic patients’ onset within 7 days.● BinaxNOW may become a practical tool for large-scale or at-home use for COVID-19.

KEYWORDS:

• BinaxNOW antigen card
• SARS-CoV-2
• diagnosis
• meta-analysis
• systematic review

1 Introduction

The coronavirus disease 2019 (COVID-19) is a serious acute infectious respiratory disease caused by the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). It can be transmitted through respiratory droplets, aerosol, or intimate contact with infected individuals [1,2]. SARS-CoV-2 infection may be asymptomatic or cause a wide range of symptoms, such as dyspnea, pneumonia, respiratory failure, and life-threatening septic shock [3,4]. According to the latest report by the World Health Organization (WHO), the rapid worldwide spread of COVID-19 resulted in 218,946,836 cases and 4,539,723 deaths on 3 September 2021; this has had an extremely severe impact on global health [5]. Although a soaring number of people are being vaccinated, the emergence of several concerned variants that comprised Alpha, Beta, Delta, Gamma, and Omicron, led to a new wave of SARS-CoV-2 infection, which caused health crisis, tidy loss of lives, and economic stagnation around the world [6,7]. The rapid and accurate detection of SARS-CoV-2 still plays an essential role in slowing down the spread of the virus.

Recently, the diagnostic methods of SARS-CoV-2 have been mainly relying on the detection of samples from in vitro collections. These can be divided into three types: virus gene detection, human antibody detection, and SARS-CoV-2-related antigen detection [8]. Reverse transcription polymerase chain reaction (RT-PCR) is considered as the best method for detecting SARS-CoV-2 [9,10]. However, this method requires extensive and complicated equipment, specialized personnel, and kits [11]. Therefore, it may not be the optimal method for the large-scale early screening of COVID-19. Furthermore, the methods for human antibody detection are based on the lateral flow assay (LFA) and enzyme-associated immunoglobulin type assay, which detect specific antibody levels in human serum samples to determine the condition of the viral infection [8,10]. However, the production of the antibodies has taken 14 days after patients are infected with the SARS-CoV-2 and it can be easily contaminated when samples were performed to detect. Hence, it mainly is befitting for disease tracking and seroprevalence studies and is not suitable for patients with early infection [8,12]. Compared with the two aforementioned detection methods, the antigen detection of SARS-CoV-2 is cheaper and more convenient, and the results can be obtained immediately. Thus, basic clinical researches focusing on rapid antigen detection are being conducted.

BinaxNOW antigen card (BinaxNOW) is a new category of in vitro rapid antigen diagnostic tests, which can qualitatively detect the nucleocapsid protein antigen of SARS-CoV-2 in a front nasal swab collected from patients based on lateral flow immunoassay [9]. With the convenient application, users can take the test cotton swabs collected from the patients’ nasal cavity to directly insert into the test card, close that, and wait for the results [13]. According to whether there were pink/purple lines, the test results can be visually interpreted and recorded by photos [14]. It only takes 15 minutes, costs $ 5, and requires no instruments [15,16]. BinaxNOW was granted the emergency use authorization (EUA) by the US Food and Drug Administration (FDA) on August 26th, 2020, with intended use in persons with suspected COVID-19 within 7 days of symptom onset [17]. Compared with other LFA kits, BinaxNOW’s high productive capacity and early approval by FDA lead it to have a large market share and is widely applied in the United States (US) and other regions [18,19]. 150 million test kits had been distributed by the US Department of Health and Human Services in 2021 [20] . Beginning January 15th, 2022, most people in America with a health plan are able to purchase an at-home test at no cost either through reimbursement or free of charge through their insurance [21]. Consequently, it can be readily available for users. Owing to its simplicity, short test time, relatively low cost, widespread use, readily availability, and easy-to-use, BinaxNOW is the ideal method for large-scale, early, and rapid screening of suspected COVID-19 patients [9,20]. To sum up, BinaxNOW has a promising clinical application. As a result, it is necessary to conduct a meta-analysis of BinaxNOW to clarify its diagnostic effectiveness for SARS-CoV-2 in a more objective view.

Currently, some studies regarding the diagnostic accuracy of BinaxNOW for SARS-CoV-2 have been conducted. However, the test results were different, no unified conclusion was determined, and no comprehensive meta-analysis of BinaxNOW for SARS-CoV-2 has been conducted. We hypothesized BinaxNOW had a considerable sensitivity and specificity. Hence, our study aimed to evaluate the accuracy, practicability, and applicability of BinaxNOW technology and provided an evidence-based reference for the rapid screening for COVID-19 in future clinical practice.

2 Materials and methods

2.1 Electronic searches

This study had made a systematic evaluation while following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines [22,23]. The relevant articles were independently retrieved by two researchers from four databases (PubMed, Web of Science, Cochrane Library, and Embase). The keywords were ‘BinaxNOW’, and ‘SARS-CoV-2’. And their synonymous extensions from the MeSH database and Emtree were utilized, combining them with Boolean operators AND or OR. The detailed search strategy was shown in Table S1. To avoid omissions, the bibliographies of most publications were scanned. Afterward, Endnote X9 was used to manage the articles. The researchers independently retrieved the related databases (PubMed, Embase, Web of Science, Cochrane Library) before May 1st, 2021.

2.2 Inclusion and exclusion criteria

We enrolled the articles that met the expected requirements, which comprised the following: (1) Studies were performed in English with full text, (2) RT-PCR was used as the gold standard in the studies, (3) 2*2 tables could be extracted, and (4) the data had the clear cycle threshold (CT) values. The exclusion criteria were as follows: (1) uncorrelated studies, abstract, duplicated publications, conference summaries, letters, and reviews, (2) studies using non-human samples, and (3) non-experimental studies.

2.3 Data extraction and quality assessment

The data were independently extracted by two researchers after full-text reading. The extracted contents of the studies consisted of the first author, country, gold standard, study design, published year, CT value, symptom, symptom onset within 7 days, sample type, age, testing site, and 2*2 data. Two researchers independently evaluated the quality of the enrolled studies according to the Quality Assessment of Diagnostic Accuracy Study-2 (QUADAS-2) guidelines, and the quality plot was generated by Review Manager 5.3 [24]. A third researcher would participate to solve any inconsistencies in the data they extracted and the quality of the studies they evaluated.

2.4 Statistical analysis

The Meta-Disc 1.4 software was used to generate the results that consist of the following items: sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR), and 95% confidence intervals (95%CI), which were expressed in the form of forest plots [25]. Additionally, the summary receiver operating characteristic curve (SROC) was plotted, and the area under curve (AUC) was estimated; this was used to evaluate the accuracy of the diagnostic BinaxNOW for COVID-19. The Q* test and the I² test were used to explore the heterogeneity of the enrolled studies [26]. The random-effects model was selected to combine the statistics when the heterogeneity was found (I² > 50%). The heterogeneity sources came from the threshold effect and non-threshold effect. The SROC and the Spearman correlation coefficient were used to verify whether there was a threshold effect during the enrolled studies. To explore the source of heterogeneity in the non-threshold effect, subgroup analysis and meta-regression analysis were conducted. Furthermore, to identify the publication bias of the enrolled studies, the Stata 16.0 software was used to draw a Deeks’ funnel plot and perform Egger’s test [27].

3 Results

3.1 Search results

The detailed retrieval and screening process of the literature was shown in Figure S1. Based on the previous search strategies, a total of 29 published articles were obtained from the above databases. Of these, 11 were from PubMed, nine from Embase, nine from Web of Science, and none from Cochrane Library. After removing 17 duplicates, 12 articles were further reviewed. Then two articles were excluded after reviewing the titles/abstracts. And one article was unable to form 2*2 tables, and two articles that had no clear CT value were excluded by full-text assessment. Finally, seven articles were identified for inclusion in this study [15,18,20,28–31].

3.2 Study characteristics

From the seven studies, 59 groups of data with 37,785 samples (range: 8 to 3302) were extracted. The positive RT-PCR patients were further stratified by different Ct values of the RT-PCR test. The data contents comprised the author’s name, published year, CT value, symptom, symptom onset within 7 days, sample type, age, testing site, 2*2 data, and other notes. The detailed characteristics of the included studies was displayed in Table S2. These studies were published from 2020 November until 2021 April, the index test was compared with RT-PCR under diverse CT values. The samples that contained anterior nasal swabs and nasopharyngeal swabs were collected from all ages including children, juveniles, and adults in the community, nursing homes, or hospitals. Both symptomatic and asymptomatic participants were recruited in these studies. Besides, 59 datasets all utilized RT-PCR as the gold standard, all of which were found in prospective studies in America.

3.3 Quality assessment

The QUADAS-2 tool was employed to appraise the quality of the individual studies. Review Manager 5.3 was employed to draw the graph of the quality, as presented in Figure 1. All studies were at low risk of bias in the patient selection area. Two studies were of high concern regarding applicability due to the presence of participants with asymptomatic states from the university or medical workers. As the index test results were explained without knowing the results of the reference standard and the threshold, the index test area was pre-designated in all studies. Thus, all studies were at low risk concerning both bias and applicability. In the field of the reference standard, the bias risk of all studies was low. The collected samples were not tested by the reference standard method twice to ensure the status of samples; hence, six studies were unclear concerning applicability. In the flow and timing area, the risk of bias was low in all studies.

Figure 1. Quality evaluation of the included studies.

3.4 Merged analysis

A meta-analysis of the BinaxNOW assay for COVID-19 was performed in this study using the random effect model. The pooled results are shown in Figures 2–3. The combined sensitivity, specificity, PLR, NLR, DOR, and their 95% CI were 0.77 (0.76 to 0.79), 0.99 (0.99 to 0.99), 65.72 (48.23 to 89.56), 0.23 (0.19 to 0.28), and 461.10 (281.55 to 755.13), respectively. Moreover, the overall result can be seen in Figure 3(b). The AUC was 0.9910, and Q* was 0.9611 (SE = 0.0092) in the SROC curve.

Figure 2. Forest plots for the combined sensitivity(a), specificity(b), positive LR(c), negative LR(d) of BinaxNOW.

Figure 3. Forest plots for the combined diagnostic OR of BinaxNOW(a) and summary receiver operating characteristic curves of COVID-19 infections detected by BinaxNOW(b).

3.5 Threshold effect analysis

If there was a typical ‘shoulder-arm’ pattern or if the Spearman correlation coefficient was greater than 0.6, it would indicate that a threshold effect may exist [27]. The Spearman correlation coefficient was 0.113, and the P-value was 0.395 in our study. Moreover, a ‘shoulder-arm’ model was not present in Figure 3(b). As mentioned above, these results suggest that there was no threshold effect in the included studies.

3.6 Non-threshold effect analysis

The quantitative index to judge heterogeneity was calculated by observing the I² in the forest plots. The heterogeneity grading was interpreted in the Cochrane handbook by checking the inconsistency index. As we can see (Figure 3(a)), the results – Cochran-Q = 302.53, P < 0.01, and Inconsistency = 80.8% (>75%) – meant that there was considerable heterogeneity in this study. High heterogeneity can also be seen in other forest plots, such as the I² of sensitivity (91.6%), specificity (77.9%), PLR (78.7%), NLR (88.7%), and DOR (80.8%).

3.7 Subgroup analysis and meta regression analysis

We found that considerable heterogeneity existed in this study. Thus, a subgroup analysis was conducted to probe the differences and main sources of heterogeneity, such as symptom, days of onset, CT value, age stage, sample type, and testing site. In the aspect of sensitivity, the lower CT-value group had higher sensitivity; the asymptomatic group’s sensitivity (0.60) was lower than the symptomatic group (0.84); the symptom onset within 7 days had higher sensitivity (0.92 vs 0.68); the pooled sensitivity of age >18 years old group was 0.70 and <18 years old group was 0.65; in terms of testing sites, the sensitivity of community, nursing homes, and hospitals were 0.81, 0.67, 0.57, respectively. In the aspect of specificity, all groups were high (0.98 to 1.00). More details were shown in Table 1.

Table 1. Subgroup analyses for the exploration of factors influencing heterogeneity in the BinaxNOW assay.

Subgroup analysis		Size	Sensitivity(95%CI)	I^2	Specificity(95%CI)	I^2
CT value	Low CT	11	1.00(0.98–1.00)	5.00%	0.99(0.99–0.99)	82.80%
	Median CT	14	0.84(0.83–0.86)	91.00%	1.00(0.99–1.00)	0.00%
	High CT	34	0.67(0.65–0.69)	81.60%	0.99(0.99–0.99)	78.50%
Testing site	Community	37	0.81(0.80–0.83)	93.20%	0.99(0.99–0.99)	73.80%
	Nursing Home	19	0.67(0.63–0.70)	0.00%	0.99(0.98–0.99)	77.30%
	Hospital	3	0.57(0.51–0.62)	78.10%	1.00(1.00–1.00)	0.00%
Symptom	Asymptomatic	9	0.60(0.56–0.63)	49.80%	0.99(0.99–1.00)	79.70%
	Symptomatic	18	0.84(0.82–0.87)	87.20%	0.99(0.99–0.99)	66.80%
Symptom onset within 7 days	NO	17	0.68(0.64–0.71)	89.10%	0.99(0.99–0.99)	73.30%
	YES	9	0.92(0.89–0.94)	88.80%	0.99(0.98–0.99)	74.60%
Age stage	>18-year	22	0.70(0.67–0.73)	63.70%	0.99(0.99–0.99)	77.30%
	<18-year	18	0.65(0.62–0.69)	83.80%	0.98(0.98–0.99)	14.20%

CT: cycle threshold; Low CT group: CT ≤ 30 or 33; Median CT group: CT ≤ 35 or 37; High CT group: CT ≤ 40 or 42.

Regression analysis was also used to investigate the source of heterogeneity. The results were outlined in Table 2. From the P-value (P < 0.05 of CT value and Testing site), different CT values and testing sites may affect the accuracy of the BinaxNOW assays for COVID-19.

Table 2. Weighted meta-regression to assess the effects of various factors on the diagnostic accuracy of the BinaxNOW assay.

Covariate		Coefficient	RDOR(95% CI)	P
CT value	Low CT (11)	−1.90	0.15(0.08–0.30)	<0.001
	Median CT (14)
	High CT (34)
Testing site	Community (37)	−0.96	0.38(0.15–0.98)	0.046
	Nursing Home (19)
	Hospital (3)
Symptom	Asymptomatic (9)	0.97	2.63(0.57–12.11)	0.202
	Symptomatic (18)
Symptom onset within 7 days	No (17)	1.06	2.90(0.47–17.86)	0.239
	Yes (9)
Age stage	>18-year (22)	0.65	1.92(0.67–5.53)	0.217
	<18-year (18)

CT: cycle threshold; Low CT group: CT ≤ 30 or 33; Median CT group: CT ≤ 35 or 37; High CT group: CT ≤ 40 or 42.

3.8 Publication bias

Deek’s funnel chart asymmetry test was utilized to evaluate the potential publication bias in the enrolled studies, and the resulting P-value of 0.10 (P > 0.05) indicates that a low publication bias was found in this study (Figure 4).

Figure 4. Deeks’ funnel plot asymmetry test to assess publication bias in estimates of diagnostic OR for BinaxNOW detection of COVID-19 infections.

3.9 Comparison of the diagnostic performance of various LFA kits

In this study, we compared BinaxNOW with other LFA kits, including Panbio, Innova, STANDARD Q, CORIS, Bioeasy, COVID-VIRO, BD Veritor, BIOCREDIT [32]. Overall results were shown in Table S3. In terms of sensitivity, only Bioeasy, COVID-VIRO, and BD Veritor had advantages over BinaxNOW. In terms of specificity, all types of LFA tests were similar.

4 Discussion

BinaxNOW is a simple, effective, and low-cost emerging rapid lateral flow immunoassay that can directly detect the novel coronavirus nucleocapsid antigen from nasal swabs [15,33]. Up to now, the mainstream gold standard is RT-PCR for detecting SARS-CoV-2 [34]. However, some shortcomings of RT-PCR, such as being expensive, time-consuming, and complicated, have led to its limited application in some developing countries with large population bases, poor resources, and insufficient laboratories [35]. BinaxNOW can circumvent these deficiencies, therefore, there is considerable interest in it for researchers and has had a promising clinical application [15]. As a result, it is necessary to conduct the pooled analysis of BinaxNOW to clarify its diagnostic effectiveness for SARS-CoV-2.

In this study, the pooled sensitivity, specificity, PLR, NLR, and DOR were 0.77, 0.99, 65.72, 0.23, and 461.10, respectively. Medium sensitivity and high specificity indicated that the missed rate and misdiagnosis rate of BinaxNOW were both low, especially in the misdiagnosis rate. Moreover, the PLR (>10) and NLR (>0.1) revealed that the advantage of BinaxNOW was in terms of confirmation rather than exclusion. Hence, a negative result cannot be used completely to exclude COVID-19. A high DOR value indicated an optimum ability to distinguish the target condition of the participants. Furthermore, the SROC was very close to the upper left corner. At the same time, the AUC was 0.9910 (close to 1), indicating that this test’s diagnostic performance was extraordinarily excellent. From the overall results, it can be concluded that the diagnostic capacity of BinaxNOW for SARS-CoV-2 is considerably high.

In addition, neither publication bias nor a threshold effect exists in the study, indicating that the pooled results of this study are credible and stable. Thus, the heterogeneity in the research may come from the non-threshold effect. The I² of pooled sensitivity (91.6%), specificity (77.9%), PLR (78.7%), and NLR (88.7%) indicate that a high heterogeneity exists in the meta-analysis. The heterogeneity may influence the accuracy of the meta-analysis results. To further explore the source of heterogeneity, we conducted a subgroup analysis and meta-regression analysis. According to the results of the meta-regression, the P-value of CT-value and testing site were significant. Therefore, the CT value and testing site were likely important sources of heterogeneity.

In the subgroup analysis, BinaxNOW showed high specificity in all states. However, regarding sensitivity, different subgroups displayed different diagnostic effects. The sensitivity values of the low, medium, and high CT-value groups were 1.00 (0.98–1.00), 0.84 (0.83–0.86), and 0.67 (0.65–0.69), indicating that the CT-value and sensitivity of BinaxNOW for detecting SARS-CoV-2 are negatively correlated and that false-negative results more likely occurred in the subgroup with a high CT-value. It is known that the low CT-value indicated high viral load [36]. According to the results, the sensitivity of antigen detection depends on the viral load to a great extent, and the BinaxNOW antigen card is more suitable for diagnosing the COVID-19 patients with a high viral load. In the aspect of epidemiological prevention and treatment, large-scale detection with BinaxNOW can quickly find people with high viral load and strong infectivity, thus achieving the purpose of slowing down the spread of SARS-CoV-2 [29].

In terms of symptoms, the sensitivity of the asymptomatic group was 0.60 (95%CI: 0.56–0.63) while the symptomatic group was 0.84 (95%CI: 0.82–0.87). In the aspect of the symptom onset days, the symptom onset within 7 days group with a sensitivity of 0.68 (95%CI:0.64–0.71), compared to the symptom not within 7 days 0.92 (95%CI:0.89–0.94). The 95%CI of the symptomatic group and the asymptomatic group did not overlap, and the 95%CI of the group within 7 days of onset did not exist overlap between the group within 7 days of onset, which indicated that there are significant differences between the above groups [37]. When sensitivity is ≥ 80% and specificity is ≥ 97% in a diagnostic assay, this assay can meet the appropriate criteria of WHO’s priority target product profiles [38,39]. Therefore, in terms of clinical significance, with the same specificities (0.99), the sensitivity of symptoms (0.84) versus non-symptoms (0.60), and symptoms within 7 days (0.92) versus symptoms not within 7 days (0.68) illustrated that BinaxNOW is beneficial for detecting SARS-CoV-2 in symptomatic patients’ onset within 7 days. This result supports the decision of the FDA and is consistent with the relevant observation of Pollock’s and Pilarowski’s study [30,31,40]. However, there is no significant P-value found in the meta-regression of the symptom group and symptom onset within 7 days group. It indicated that the BinaxNOW still may have applicable value in symptoms not onset within 7 days patients. As asymptomatic patients have low viral loads, they need to be tested again by other diagnostic methods to confirm the diagnosis [18,41].

Interestingly, in the testing site, the subgroup analysis results showed that the sensitivity of community, nursing homes, and hospitals were 0.81, 0.67, and 0.57, respectively, and the P-value was 0.046 (P < 0.05) in meta-regression analysis. In order to explore some possible influencing factors, we reviewed these articles again. In Pilarowski’s study, the testing results were retrospectively quantified from computerized image data, each assay was read by 2 independent observers, and a site supervisor served as a tiebreaker [20]. However, the results about how to read in other studies are seldom described or only read by one operator. Multi-person evaluation and interpretation may help reduce subjective bias and improve accuracy. In addition, combined with the understanding of BinaxNOW’s instructions (uploading data by scanning Quick Response code), we suggest that manufacturers can develop related applications to avoid subjective deviation through image pixel color difference analysis. In terms of age stage, there are no significant differences between >18 years old group and <18 years old group, which means that BinaxNOW had a similar diagnostic effect on children and adults.

In addition to the above factors, we found that there are still more factors that may influence the accuracy of the BinaxNOW. First, the proper environmental conditions are needed for accurate detection. The manufacturer recommends that tests be run at temperatures > 59°F. In Pollock et al. ’s study, when the BinaxNOW tests were conducted at low temperature (46 to 58.5°F), the sensitivity was 66.7% and specificity was 95.2%, compared to testing performed at >59°F with a sensitivity of 93.7% and specificity of 100% [30]. It reminded us that we must pay attention to the temperature in testing. At the same time, humidity may also affect the test results, but correlative experimental data is still not clear enough, so more relevant research should be carried out [30]. As a result, maintaining the proper temperature and humidity helps to keep stable test results. Second, the sample type may contribute to the test results. To probe into whether sample type would affect the accuracy, we conducted a pooled analysis and the results showed that using anterior nasal as the testing sample only, the sensitivity was 0.81, while in the sample types that were mixed anterior nasal and nasopharyngeal, the sensitivity was 0.67 (Figure S2) [28]. Therefore, the antigen should detect in the anterior nasal, while using the other sample types to test (nasopharyngeal, oral et al.) the accuracy may be influenced. Besides, nasal swabs also should not be displaced by cheek or throat swabs in using BinaxNOW [42]. Recent studies pointed out that, under compared with RT-PCR as a gold standard, nasopharyngeal and AN swabs have similar performance in the antigen test [28,43]. These are consistent with the BinaxNOW instruction’s recommendation of an anterior nasal swab as the test sample [13]. Third, the operation procedure may influence the accuracy too. Moving the antigen card during testing, reading the result too late or early, and placing the sample wrong would lead to errors.

However, some limitations remain in this study. First of all, although we have conducted a detailed search, it is challenging to ensure that no articles and datasets were missed. Besides, two studies provided more than half of the datasets (37/59 = 62.7%) in our meta-analysis, which may create a bias in pooled results, such as sensitivity and specificity [28,31]. Second, we attempted to conduct a comprehensive analysis of study quality using a quality assessment method several times; however, studies of slightly poor quality may be biased due to the investigators’ subjectivity. Third, the sources of heterogeneity in the meta-analysis have not yet been fully identified. Hence, we need more data to prove our conclusion.

5 Conclusion

Through a series of meta-analysis, the BinaxNOW has a high diagnostic performance for SARS-CoV-2, especially in patients with a high viral load. Although various factors could impact the accuracy to some degree, the BinaxNOW has great clinical significance in symptomatic patients’ onset within 7 days and may become a useful diagnostic tool for large-scale or at-home use. However, more future research should be explored to corroborate our findings.

Author contributions

Xu-Guang Guo conceived the study. Yue-Xue Mai, Xi-Feng Qian, Jia-Hao Zhang, and Zhan-Fu Mo collected the data, analyzed the data, and wrote the manuscript. Ke-Fan Wu, Zi-Yuan Yu, Qi-Qing Ye, Xin Yin, and Xiao-Ying Zhong participated the data collection and amended the manuscript. The final manuscript was read and approved by all the authors.

Statement of ethics

All analyses were based on previously published studies; thus, no ethical approval and patient consent are required. The Third Affiliated Hospital of Guangzhou Medical University Institutional Review Board provided the ethics exemption and support our study.

Acknowledgements

We show our great gratitude to all members of our work term for their cooperation. Thanks to my good friend Lin Keer for her support.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All data generated or analyzed during this study are included in this published article.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21655979.2023.2180221

Additional information

Funding

This study was supported by The National College Students' Innovation And Entrepreneurship Training Program (number:202210570046).