Respiratory infections are a significant source of morbidity and mortality in hospitals and in the community, accounting for numerous hospital admissions every year (Beninca et al., 2017; Rijn et al., 2018). A major challenge is that respiratory infections, regardless of pathogen, often present with similar symptoms rendering an initial diagnosis difficult (Popowitch et al., 2013). A timely diagnosis is crucial for patient management, optimizing length of hospital admission, and preventing transmission. This is particularly important for the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic. In addition, a shorter length of stay (LOS) can reduce incidences of nosocomial infections and unnecessary antibiotic use (Dik et al., 2016; Wolkewitz et al., 2019).

The need for rapid syndromic testing has led to the rise of multiplex real-time PCR systems which are highly automated and perform all steps in a self-contained device. Systems such as the BioFire FilmArray (BioMérieux, Marcy-l’Étoile, France), Xpert Xpress (Cepheid, Sunnyvale, CA, United States), ePlex (GenMark Diagnostics, Roche, Basel, Switzerland), VERIGENE (Lumine DiaSorin, Saluggia, Italy), and the QIAstat-Dx (Qiagen, Hilden, Germany) are frequently used for patients exhibiting more than one respiratory symptom (Popowitch et al., 2013; Pinsky and Hayden, 2019). Screening for multiple pathogens simultaneously not only contributes to improving diagnostic stewardship and minimizing antibiotic use, but also enables seasonal tracking and identification of less common respiratory pathogens (Meyers et al., 2020).

Prior to the SARS-CoV-2 pandemic, the University Medical Center Groningen (UMCG) had incorporated the BioFire Respiratory Panel (RP) RP2.0 into routine diagnostics. Integration of point-of-care (POC) tests like the BioFire aid the “€hr” concept, where cost per sample is multiplied by the total turnaround time (Poelman et al., 2020). However, amplification data such as cycle threshold (Ct) values are not reported in the automatic output results (Poritz et al., 2011). Therefore, semi quantitative and qualitative information which may be useful for patient management is missing.

The QIAstat-Dx Respiratory SARS-CoV-2 panel RP2.0 (QIAstat-Dx RP2.0) (Qiagen, Germany) and the BioFire FilmArray Respiratory panel RP2.1 (BioFire RP2.1) have recently emerged in the market, generating results from 22 different respiratory targets, including SARS-CoV-2 in approximately 1 h (Rao et al., 2021). While the BioFire RP2.1 offers relatively quicker results (45 vs. 70 min), the QIAstat-Dx RP2.0 also offers direct introduction of the nasopharyngeal swab into the cartridge, reducing hands-on-time. Furthermore, the QIAstat-Dx RP2.0 also offers cycle threshold (Ct) values as an automatic output result, providing additional qualitative information and permitting further comparison with the laboratory developed test (LDT) (Boers et al., 2021). As the QIAstat-Dx and the BioFire are prominent platforms in the market, it is important to understand their strengths and weakness to ensure the quality of data collected. Continually evaluating and implementing new products for routine testing is crucial for diagnostics to move forward. The aim of the study was to compare the QIAstat-Dx RP2.0 with the BioFire RP2.1 in relation to analytical specificity, analytical sensitivity and clinical sensitivity for the detection of respiratory pathogens.

Point-of-care testing is becoming more readily implemented in routine diagnostics, with infectious etiologies simultaneous screened within an actionable timeframe (Basile et al., 2018). Initially the QIAstat-Dx RP2.0 and the BioFire RP2.1 was evaluated for the detection of SARS-CoV-2 from NPS, covering a wide range of Ct values. To determine a baseline prevalence and viral load in the SARS-CoV-2 samples at the UMCG, the distribution of Ct values over a period of 17 months were initially plotted (Figure 1). Measuring Ct values are important to monitor the patient and assist infection and control, particularly as Ct values can fluctuate over time, depending on the pre-symptomatic and symptomatic phases (Rabaan et al., 2021). The high Ct values observed from UMCG patients and workers could have resulted from the increased frequency of testing and follow-up testing, regardless if the patient or worker had a milder disease, which can correspond to a higher Ct value (Fuller et al., 2013; Trunfio et al., 2021). Alternatively, it could be that we are detecting the virus earlier, before the patient or worker was particularly ill and sought medical care. Patients tested through the Public Health Service appeared to have lower Ct values which could suggest they are detecting more acute infections at the time of sampling.

The Analytical Q panel for SARS-CoV-2 from Qnostics was used to determine the LOD for the QIAstat-Dx RP2.0 and the BioFire RP2.1 (Table 1). While the BioFire RP2.1 had 100% concordance with the LDT (50 dC/mL), the QIAstat-Dx RP2.0 could only detect SARS-CoV-2 at 500 dC/mL. The fact that SARS-CoV-2 was not found by the QIAstat-Dx RP2.0 at these loads could be potentially problematic, given that this patient population has been shown to have relatively high Ct values for this target (Figure 1). This does suggest that some acute infections with higher Ct values (lower viral loads) could be missed. Differences in sensitivity between the two platforms could be due to the fact that BioFire RP2.1 is based on a nested PCR, which typically has higher sensitivity (Shaffaf and Ghafar-Zadeh, 2021), or variation in gene targets or chemistries. Nevertheless, the results achieved for the QIAstat-Dx RP2.0 have concordance with the LOD described in the instruction manual. In addition, a previous study which evaluated the GeneFinder™ COVID-19 Plus RealAmp kit using the ELITe InGenius platform found a LOD of 500 dC/mL (RdRp and E genes), which was similarly observed in the QIAstat-Dx RP2.0 (Gard et al., 2022).

A difference in clinical sensitivity in the QIAstat-Dx RP2.0 and the BioFire RP2.1 for SARS-CoV-2 was observed, with 85% and 97%, respectively. Clinical sensitivity of SARS-CoV-2 in the QIAstat-Dx RP2.0 has been investigated previously, with one study determining 100% concordance with the LDT (n = 17 samples), while other studies reported a PPA of 90% (n = 120 samples) and 94.32% (n = 88 samples) (Visseaux et al., 2020; Gupta et al., 2021; Lebourgeois et al., 2021). The latter study also reported a failure rate of 7.5% (n = 13) in the QIAstat-Dx RP2.0 (Lebourgeois et al., 2021), which was similarly observed in this study with 6.1% (2 samples). Additionally, four samples previously positive for SARS-CoV-2 in the LDT, were negative in the QIAstat-Dx RP2.0, suggesting poorer sensitivity for samples with higher Ct values. A possible explanation for the false negative results in QIAstat-Dx RP2.0 (Ct 22) and in the BioFire RP2.1 (Ct 23) could be RNA degradation, too many freeze-thaw cycles or mutations at the primer binding sites (Kanji et al., 2021; Mouliou and Gourgoulianis, 2021). Other studies investigating SARS-CoV-2 in the BioFire RP2.1 have found a PPA of 100% (n = 25 samples) and 98% (n = 49 samples) in clinical samples (Creager et al., 2020; Eckbo et al., 2021). In this small cohort, the BioFire RP2.1 did not report any instrument failures, however a previous study has reported a failure rate of 2% (n = 2 samples) (Creager et al., 2020).

In comparison to the BioFire RP2.1, the QIAstat-Dx RP2.0 also reports Ct values for test interpretation. This can help toward estimating the viral burden and association with disease severity. Although Ct values are generally not comparable between platforms, the Ct values in the QIAstat-Dx RP2.0 differed substantially from the LDT and were approximately five Ct values higher (Figure 2).

False positive results were not observed for SARS-CoV-2; therefore, this study can report a negative percent agreement (NPA) of 100% for both platforms. False positives can lead to unnecessary isolation for the patient and delay a true diagnosis. Previous studies investigating the QIAstat-Dx RP2.0 have reported a NPA of 90.48% (n = 189 samples) and 93% (n = 29 samples) (Visseaux et al., 2020; Lebourgeois et al., 2021), while studies investigating the BioFire RP2.1 have reported a NPA of 100% (n = 5 samples) (Trunfio et al., 2021), (n = 49 samples) (Creager et al., 2020).

Multiplex panels from Qnostics were tested on both the QIAstat-Dx RP2.0 and the BioFire RP2.1 and exhibited a similar performance (Tables 2, 3). However, in pool 1, the QIAstat-Dx RP2.0 only detected PIV1. This could be accredited to competition, as single-infections for PIV1, PIV2, and PIV4 were detected by the QIAstat-Dx RP2.0 (Table 4). As no other respiratory pathogen was detected in pools 1–4 other than what was indicated, it suggests there was no cross-reaction.

The QIAstat-Dx RP2.0 detected 113/137 targets for single infections at first attempt (Table 4). The device presented a 2.81% (n = 4 samples) failure rate for various reasons (IC, cartridge or test failure). The QIAstat-Dx RP2.0 performed well in most of the single infections, obtaining an overall PPA of 82% (Table 4). However, for the HRV/EV target, the lowest PPA (67%) was obtained. This could be a problem as HRV/EV is our most frequently identified target (Supplementary Figure 2). Discrepant results among these targets on the QIAstat-Dx RP2.0 and other syndromic panels have been described previously (Leber et al., 2020; Boers et al., 2021). Although some of these targets are considered a low public health risk (CE-IVD) or linked to self-limited disease (Hollestelle and de Boer, 2008), the high prevalence of positive specimens could render these viruses relevant (Supplementary Figure 2). Similar results were found among the multiple infections, with HRV/EV being the most frequently found target with discrepancies (Table 5). Although a PPA of 91.2% for the HRV/EV target has been described in the QIAstat-Dx RP2.0 manual, this was not observed in this study.

This study had some limitations. Firstly, as our data set is small for some respiratory targets, definitive conclusions are not necessarily able to be formed at this point. Secondly, as some samples did not contain enough volume, further comparisons or repeat testing, both on the QIAstat-Dx RP2.0 and BioFire RP2.1 could not always be performed. Thirdly, additional sample freezing or thawing could have impacted the results. Finally, non-SARS-CoV-2 viruses were not re-evaluated in the BioFire RP2.1, due to constraints in samples volume and time. However, according to the manufacturer, as the other respiratory targets were not altered in the update, it should not impact the results.

In conclusion, this study has shown that the QIAstat-Dx RP2.0 and BioFire RP2.1 offer comparative performances; however, differ slightly in analytical and clinical sensitivity for SARS-CoV-2. A rapid and accurate differential diagnosis will ensure the most appropriate patient management decision. Point-of-care platforms are the way to move forward, however improvements are necessary such as the LOD and quantitative information.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

HC: study design, laboratory and data analysis, writing—original draft preparation, writing—reviewing, and editing. MG: laboratory and data analysis, writing—reviewing, and editing. EL-F: study design, laboratory and data analysis, writing—reviewing, and editing. HN and LG: conceptualization, editing, and supervision. All authors contributed to the article and approved the submitted version.

Qiagen provided the QIAstat-Dx respiratory SARS-CoV-2 panel V2 test cartridges used in this study but did not take part on the study design, data collection, analysis, or manuscript preparation. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.