A prospective, multi-centre US clinical trial to determine accuracy of FebriDx point-of-care testing for acute upper respiratory infections with and without a confirmed fever

Abstract

Background: FebriDx is a 10-minute disposable point-of-care test designed to identify clinically significant systemic host immune responses and aid in the differentiation of bacterial and viral respiratory infection by simultaneously detecting C-reactive protein (CRP) and myxovirus resistance protein A (MxA) from a fingerstick blood sample. FebriDx diagnostic accuracy was evaluated in the emergency room and urgent care setting.

Methods: A prospective, multicentre, observational cohort study of acute upper respiratory tract infections (URIs), with and without a confirmed fever at the time of enrolment, was performed to evaluate the diagnostic accuracy of FebriDx to identify clinically significant bacterial infection with host response and acute pathogenic viral infection. The reference method consisted of an algorithm with physician override that included bacterial cell culture, respiratory PCR panels for viral and atypical pathogens, procalcitonin, and white blood cell count.

Results: Among 220 patients enrolled, 100% reported fever 100.5°F within the last 72 hours while 55% had a measured hyperthermia (T > 100.4) at the time of enrolment. FebriDx demonstrated a sensitivity of 95% (95% CI: 77–100%), specificity of 94% (88–98%), PPV of 76% (59–87%), and a NPV of 99% (93–100%).

Conclusion: FebriDx may identify clinically significant bacterial URI’s and supports outpatient antibiotic decisions.

Key messages

• FebriDx is an outpatient POC test designed to identify a clinically significant systemic host immune response and aid in the differentiation of viral and bacterial infection through rapid measurement of MxA and CRP from a fingerstick blood sample. FebriDx test was determined to be an accurate test, with a 85% sensitivity, 93% specificity and 97% NPV to rule out bacterial infection for any patient presenting with symptoms and reported fever within the prior 3 days, and when confirming fever (hyperthermia) at the time of testing, the test was even more sensitive (95%) and specific (94%) with a 99% NPV. FebriDx may support antibiotic stewardship by rapidly identifying clinically significant bacterial URIs.

Keywords:

• FebriDx
• MxA
• CRP
• point-of-care
• URI
• diagnostics
• antibiotic stewardship

Introduction

Acute upper respiratory infection (URI), which includes rhinosinusitis, pharyngitis, the common cold, and acute uncomplicated bronchitis are the most common reasons for acute outpatient physician visits and antibiotic prescription in adults in the United States [1]. Forty-one million adult ambulatory care visits result in an antibiotic prescription, of which 50% may be unnecessary [1]. Approximately 5 to 25% of patients who use antibiotics have adverse events, and about 1 in 1000 has a serious adverse event [2]. Antibiotic adverse events are responsible for 1 of every 5 visits to emergency departments for drug-related events [3].

Antibiotic overuse leads to antibiotic resistance, antibiotic-associated infections, increased costs, and adverse events, ranging in severity from mild (e.g. diarrhoea and rash) to life-threatening (e.g. Stevens–Johnson syndrome and anaphylaxis) [4]. Clostridium difficile diarrhoea, usually a result of antibiotic treatment, is reported to cause 500,000 infections and 29,300 deaths in the United States annually, leading to an estimated $1 billion in extra medical costs [5]. Inpatient antibiotic stewardship programs focus on guiding appropriate antibiotic use when adequate evidence for bacterial infection exists [6]. The vast majority of patients with acute URI present to primary and urgent care clinics in the outpatient setting, where rapid diagnostic tests to assist clinicians in the identification of bacterial respiratory infections are lacking and has hindered antibiotic stewardship efforts [7].

FebriDx^® (RPS Diagnostics; Sarasota, FL) is a novel, single use disposable point-of-care (POC) diagnostic test designed to rapidly identify clinically significant host immune responses associated with bacterial and viral URIs and to assist with antibiotic prescribing decisions [8–10]. Within 10 minutes, FebriDx provides qualitative results for elevated serum levels of c-reactive protein (CRP) and myxovirus resistance protein A (MxA) [8,10]. CRP is an acute phase reactant that when elevated above 20 mg/L, is suggestive of a clinically significant infection, although it cannot differentiate viral or bacterial aetiology independently [11]. MxA is a derivative of interferon type I α/β and associated with the presence of a viral infection [12]. The purpose of this study was to evaluate the diagnostic accuracy of the FebriDx test for identifying immune responses associated with bacterial and viral URIs in patients with acute URI symptoms.

Methods

Design and URI study population

A prospective, cross-sectional, observational study using a convenience sample of patients reporting a history of fever within the preceding 72 hours and presenting with clinical signs and symptoms of a URI was performed. Children and adults were enrolled between November 2015 and July 2016 at 10 clinical sites in the United States, including seven academic emergency departments, two community urgent care centres, and 1 ambulatory clinical research site. Similar to a previous study by Self et al. inclusion criteria included: age >1 year, exhibited or reported fever ≥ 100.5 °F within the past 3 days, and new onset of cough or sore throat within the past 7 days. Major exclusion criteria included use of antibiotics, antiviral agents, interferon therapy, immunosuppressive therapy or a live viral immunization within the past 30 days [8]. The study was approved by the governing institutional review board for each enrolling centre. Written informed consent was obtained from each participant or their legal authorized representative, as appropriate.

FebriDx measurements

Each patient underwent testing with FebriDx, which is a rapid, qualitative, single-use, disposable, whole blood immunoassay with a turn-around time of 10 minutes [8]. FebriDx requires a fingerstick to obtain a 5 μL whole blood sample and provides a qualitative result for elevated levels of CRP (≥ 20 mg/L) and MxA (≥ 40 ng/ml) without any sample processing or accessory reader equipment. FebriDx results indicating elevated CRP without an associated elevated MxA were interpreted as a bacterial infection. FebriDx results indicating an elevated MxA with or without an associated elevated CRP were interpreted as a viral infection. Valid tests showing no elevated CRP or MxA were interpreted as negative.

Study personnel at the site obtained a whole blood sample by finger stick, performed FebriDx testing according to the manufacturer’s instructions [13], and interpreted the results as bacterial, viral or negative. FebriDx results were not used for clinical care and all personnel performing FebriDx testing were blinded to all reference testing outlined below.

Outcome

Joseph et al. describe key criteria for an acute respiratory infection comparator method designed to assist with the differentiation of clinically significant acute respiratory infections from probable colonization [11]. The current study incorporated a clinical algorithm that incorporates the Joseph criteria [11] and demonstrated a high level of expert physician agreement in a previous study [8]. Additionally, this study utilizes a respiratory viral and atypical bacterial PCR panel; oropharyngeal swabs sent for plating on routine blood agar, chocolate agar, and MacConkey agar; and laboratory analysis [8]. Compared to the previous algorithm described by Self et al, PCR testing for Fusobacterium necrophorum and Neisseria gonorrhoea were included and herpes simplex and cytomegalovirus PCR removed. The clinical algorithm results were reviewed on a patient by patient basis by a physician panel, with over-reading capability to ensure clinical accuracy. There were three potential outcomes: acute bacterial infection with host response, acute pathogenic viral infection, and a valid negative test consistent with a microbiologically unconfirmed respiratory illness (MURI) [8,10]. Procalcitonin (PCT) and white blood cell (WBC) count inclusive of lymphocytosis and bandaemia were used in the reference standard to identify and support the presence of a host response [11,14,15]. Viruses that are typically pathogenic when detected in the upper respiratory tract, such as influenza, respiratory syncytial virus, parainfluenza virus and metapneumovirus were classified as an acute viral infection [16–19]. Meanwhile, viruses that are commonly found in the respiratory tract without causing symptoms, such as rhinovirus or coranavirus [16,18,19], were only considered pathogenic in patients with negative bacterial cultures who demonstrated the presence of an immune response such as an elevated WBC, elevated PCT, lymphocytosis, or bandaemia.

Reference testing algorithm

Each patient underwent the following diagnostic tests: (1) throat swab bacterial culture was performed consisting of an Eswab™ (Copan) that was inoculated on routine blood agar, chocolate agar, and MacConkey agar, incubated at 35–37 °C for 18–24 h before reading, and allowed for an additional 24-h incubation overnight at room temperature for plates that yield negative results at 24 hours; (2) multiplex polymerase chain reaction (PCR) of a combined nasopharyngeal and oropharyngeal (NP/OP) sample using the FilmArray® Respiratory Panel (BioMerieux, Inc.; Marcy-l’Etoile, France) [20] underwent validation to accommodate oropharyngeal swabs; (3) real-time reverse transcriptase PCR of an NP/OP sample for Epstein–Barr virus (EBV), a validated PCR-based LDT using ASR (Tampa General Hospital, Tampa, FL) was used to test for Fusobacterium necrophorum, and a PCR-based target amplification nucleic acid probe test called the APTIMA Combo #2 Assay on the PANTHER System was used to test for Neisseria Gonorrhea/Chlamydia Trachomatis after completing a validation for respiratory samples; (4) EBV IgM serum antibody with the Immunosimplicity^® IS-EBV-VCA IgM Test Kit (Diamedix Co; Miami Lakes, FL) [21]; (5) serum PCT concentration measurement with the BRAHMS PCT KRYPTOR Analyzer (Thermo Fisher; Waltham, MA) [22]; and (6) WBC count with band differential percentage. A central laboratory performed all reference testing that was blinded to patients, treating clinicians, and study personnel who performed FebriDx testing.

The reference testing algorithm classified patients as having a bacterial infection if any of the following 5 criteria were met: (1) throat culture positive for a bacteria that commonly causes pharyngitis (Group A, C, G beta hemolytic Streptococci, N gonorrhoeae, C. diphtheria, A. haemolyticum) with a Centor Criteria ≥ 2 plus PCT ≥ 0.1 ng/ml; (2) throat culture positive for any other bacteria plus PCT ≥ 0.25 ng/ml or PCT ≥ 0.15 ng/ml plus WBC ≥ 12,000 cells/mcL; (3) NP/OP sample PCR positive for atypical bacteria (M. pneumoniae, C. pneumoniae, B. pertussis) or F. necrophorum plus PCT ≥ 0.1 ng/ml; (4) PCT ≥ 0.25 ng/ml plus no identified pathogen; (5) PCT ≥ 0.15 ng/ml plus WBC ≥ 15,000 cells/mcL or the presence of WBC bands plus no identified pathogen [23–27]. Meanwhile, the reference testing algorithm classified patients as having a viral infection if any of the following four criteria were met: (1) NP/OP sample PCR positive for influenza A or B, adenovirus, respiratory syncytial virus (RSV), human metapneumovirus, or parainfluenza viruses 1–4; (2) NP/OP sample PCR positive for rhinovirus or coronavirus plus PCT ≥ 0.1ng/ml, WBC ≥ 10,000 cells/mcL, lymphocytosis ≥ 4000, or the presence of bands (3) NP/OP sample PCR positive for EBV plus serum IgM positive for EBV; (4) PCT between 0.15 ng/ml and 0.25 ng/ml plus WBC <15,000 cells/mcL plus no WBC bands plus no identified pathogen [8,11]. If a patient met criteria for both a viral and bacterial infection, the patient was classified as bacterial. If both a viral and a common bacterial pathogen (see point (1) above) were microbiologically confirmed in a single subject, the subject was provisionally characterized as a bacterial infection if the subject was also associated with 0.1 ng/ml ≤ PCT ≤0.25 ng/ml and moderate to significant bacterial growth. Alternatively, if both a viral and a common bacterial pathogen (see point (1) above) were microbiologically confirmed in a single subject and were associated with rare to low bacterial growth, the bacteria was deemed colonizing, and the patient characterized as viral. If the PCT ≥ 0.25 ng/ml, it was deemed bacterial regardless of the amount of growth on routine blood agar, chocolate agar and MacConkey agar. If both a viral and any bacterial pathogen were microbiologically confirmed in a single subject, the subject was provisionally characterized as bacterial if the PCT ≥ 0.75 ng/ml regardless of WBC. Otherwise, the patient was characterized as a viral infection. Patients who did not meet any of these criteria for bacterial or viral infection were classified as negative by the reference testing algorithm.

Physician panel over-read

Two expert physicians in respiratory infections reviewed each case in detail and were blinded to the FebriDx results. The reviews included verification of the reference testing algorithm classification (bacterial, viral or negative) as a guideline, but also included a review of all clinical and laboratory information available from the study case report form including the demographics, patient history, presenting symptoms and signs, comorbidities, results from throat culture, molecular respiratory pathogen panels, and the following blood tests: CBC including WBC and a differential with bands and lymphocytes, procalcitonin, Epstein–Barr virus IgM/IgM as well as additional standard of care tests performed including rapid flu and rapid strep tests. Following the review, the physicians developed a classification of bacterial, viral or negative. This physician over-read could lead to reclassification compared to the initial classification from the algorithm. Thus, the final reference standard classification (bacterial, viral or negative) consisted of an algorithm-guided consensus decision.

Statistical analysis

Diagnostic accuracy was calculated by comparing the FebriDx result to the reference standard diagnosis. Primary analysis of diagnostic accuracy was based on Sensitivity (TP/P) and Specificity (TN/N), where P, N, TP and TN correspond to positives, negatives, true positives and true negatives, respectively. Additional measures of accuracy were calculated including Positive predictive value (PPV = TP/[TP + FP]) and Negative predictive value (NPV = TN/[TN + FN]. Initially, FebriDx and reference standard results were analysed using a three-category classification scheme (bacterial vs. viral vs. negative). FebriDx accuracy for bacterial and viral detections were also evaluated in two-category classification schemes (bacterial vs. not bacterial, and viral vs. not viral). Sensitivity, specificity and positive and negative predictive values were separately calculated for bacterial and viral detection. Confidence intervals for sensitivity, specificity and predictive values were calculated using the binomial exact method.

Results

URI study population

During the study period, 223 febrile URI patients were enrolled. All enrolled URI patients underwent FebriDx testing and had a valid FebriDx result. Two patients (0.9%) did not have adequate testing for the reference standard and were excluded, and one patient had an invalid FebriDx test resulting in a final population of 220 URI patients for analysis. These patients were excluded prior to analysis. This population included 13 (5%) children <18 years old and 207 (95%) adults (Table 1). Among 220 patients enrolled that reported fever ≥ 100.5 within the last 72 hours, 121 (55%) had a confirmed fever (observed hyperthermia) at the time of enrolment.

Table 1. Demographics.

			Gender		Race				Ethnicity
Presence of fever (observed hyperthermia)	Sample size	Age (years) (range; mean)	Females	Males	Asian	Black	White	Other	Hispanic	NonHispanic
Reported within last 3 days or exhibited on enrollment	220	2–86;	134	86	5	65	114	36	56	164
		36.8
Exhibited upon enrolment	121	2–86;	68	53	3	45	48	25	31	90
		36.9

Reference standard results

Application of the reference standard, including the reference testing algorithm followed by physician panel over-read, was used to classify patients. Of the total enrolled patients, 15% (34/220) were classified as bacterial, 56% (124/220) were classified as viral and 28% (62/220) negative by the reference standard. The physician panel agreed with classification assigned by the algorithm in 99% (219/220) of the cases. One patient that was rhinovirus positive by PCR with an associated PCT of 2.34 ng/ml and 6% bands, was reclassified by the physician panel as more likely a bacterial infection.

Of 124 patients with a microbiologically confirmed viral infection, 29 (23%) had PCT >0.1 ng/ml, 11 (9%) had PCT ≥ 0.25 ng/ml and 56% had a CRP ≥ 20 mg/L. Overall, the most common pathogens were influenza A (n = 60), Influenza B (n = 20) and GABHS (n = 13) (Table 2).

Table 2. Reference standard results.

Reference standard classification criterion	n	Organisms detected
Bacterial	34
Throat culture positive for bacteria commonly causing pharyngitis and PCT ≥ 0.10 ng/ml	16	11 Group A Streptococcus
		2 Group A Streptococcus + rhinovirus
		2 Group C Streptococcus
		1 Group G Streptococcus
Throat culture positive for other bacteria and PCT ≥ 0.15 ng/ml	3	1 S. aureus
		2 S. aureus + F necrophorum
NP/OP PCR positive for atypical bacteria and PCT ≥ 0.10 ng/ml	8	5 F necrophorum
		2 Group A Streptococcus + F necrophorum
		1 F necrophorum + Group A
		Streptocoocus non A, C, G
Throat culture negative or contaminant, PCR negative, and PCT PCT ≥ 0.25 ng/ml	4	4 none
Throat culture negative or contaminant, PCR negative, and PCT between 0.15 ng/ml and 0.25 ng/ml, and WBC >15k or bands present	2	2 none
Physician panel over-read classified as bacterial after algorithm suggested negative	1	rhinovirus
Viral	124
NP/OP positive PCR for pathogenic virus	96	54 influenza A
		2 influenza A + rhinovirus
		3 influenza A + coronavirus
		1 influenza A + adenovirus
		19 influenza B
		1 influenza B + rhinovirus
		4 parainfluenza virus
		4 RSV
		5 hMPV
		3 adenovirus
NP/OP sample PCR positive for rhinovirus or coronavirus plus PCT ≥ 0.1ng/ml, WBC ≥ 10,000 cells/mcL, lymphocytosis ≥ 4000, or the presence of bands	14	6 rhinovirus
		7 coronavirus
		1 rhinovirus + coronavirus
Throat culture negative or contaminant, PCR negative or contaminant, and PCT between 0.15 ng/ml and 0.25 ng/ml and WBC < 15 k, and no bands present	11	6 none
		1 S. aureus
		4 Group B Streptococcus
Negative	62
Throat culture negative or contaminant, NP/OP PCR negative, and PCT < 0.15 ng/ml	33	33 none
Throat culture positive for bacteria commonly causing pharyngitis and PCT < 0.10 ng/ml	6	6 Group A Streptococcus
NP/OP sample PCR positive for rhinovirus or coronavirus plus PCT < 0.1ng/ml, WBC < 10,000 cells/mcL, lymphocytosis < 4000, and no presence of bands	10	6 rhinovirus
		1 rhinovirus + S. pneumococcus
		1 rhinovirus +1 H. parainfluenzae
		1 coronavirus
		1 coronavirus + H. parainfluenzae + S. aureus
Throat culture positive for other bacteria and PCT <  0.15 ng/ml	13	5 S. aureus
		2 Group B Streptococcus
		2 F. necrophorum
		1 R. planticola
		1 M. catarrhalis
		1 H. parainfluenzae
		1 Serratia Marcescens

Diagnostic accuracy of FebriDx

For patients reporting fever in the last 72 hours and representing the full study patient population, when classifying results as bacterial versus not bacterial, overall agreement was 92% (95% CI: 88–95%). FebriDx demonstrated a sensitivity of 85% (69–95%), specificity of 93% (89–96%), positive predictive value (PPV) of 69% (56–79%), and a negative predictive value of 97% (94–99%). For viral detection in patients reporting fever in the last 72 hours, when classifying results as viral versus not viral, overall agreement was 87% (82–91). FebriDx showed a sensitivity of 90% (83–94%), specificity of 76% (66–84%), PPV of 83% (77–87%), and a NPV of 85% (77–90%).

When patients exhibiting an elevated temperature (measured hyperthermia − t > 100.4) at the time of enrolment are examined, classifying results as bacterial versus not bacterial, overall agreement was 94% (88–98). FebriDx demonstrated a sensitivity of 95% (77–100%), specificity of 94% (88–98%), PPV of 76% (59–87%), and a NPV of 99% (93–100%). For viral detection in patients exhibiting fever, when classifying results as viral versus not viral, overall agreement was 88%. FebriDx showed a sensitivity of 90% (81–96%), specificity of 78% (62–89%), PPV of 89% (82–93%) and a NPV of 80% (67–89%). In a secondary analysis, when an MxA ELISA was used to examine microbiologically unconfirmed cases, the FebriDx specificity for viral infection increases to 84% (77/92) with a 95% CI: 75–91%.

Of the 34 patients deemed to have a clinical bacterial infection, five patients had false negative FebriDx results for bacterial detection. Two of these patients had a throat cultures positive for GABHS with PCT of ≥ 0.1 ng/ml. The other three patients had negative bacterial cultures and PCR tests but were classified as bacterial by the reference algorithm due to PCT concentrations between 0.15 ng/ml and 0.25 ng/ml with a concurrent white blood cell count > 12 thousand cells/mcL (1 patient) or the presence of bands (2 patients) (Table 3).

Table 3. FebriDx performance values with and without hyperthermia at time of enrolment.

Sample size (n)	Confirmation of fever (Hyperthermia)	Diagnosis	Sensitivity % and [95% CI]	Specificity % and [95% CI]	PPV % and [95% CI]	NPV % and [95% CI]
220	Reported within last 3 days or exhibited on enrolment	Bacterial	85 (29/34) [69–95]	93 (183/196) [89–96]	69 ( 29/42 ) [56–79]	97 (183/188) [94–99]
		Viral	90 (111/124) [83–94]	76 (73/96) [66–84]	83 (111/134) [77–87]	85 (73/86) [77–90]
121	Exhibited on enrolment	Bacterial	95 (19/20) [77–100]	94 (95/101) [88–98]	76 (19/25) [59–87]	99 (95/96) [93–100]
		Viral	90 (72/80) [81–96]	78 (32/41) [62–89]	89 (72/81) [82–93]	80 (32/40) [67–89]

Of note, among 60 patients with influenza A and 20 influenza B detected by RT-PCR, 57 95% (57/60) of influenza A and 80% (16/20) of influenza B had a viral result by FebriDx.

Discussion

In the current clinical study, we sought to evaluate the accuracy of FebriDx for identifying clinically-significant acute bacterial and viral URIs in patients presenting with and without confirmed fever (measured hyperthermia) at the time of enrolment. Of the 220 outpatients enrolled, 55% presented with confirmed hyperthermia at the time of testing while 45% presented with a reported history of fever within the prior 72 hours but no documented hyperthermia at the enrolling healthcare facility. FebriDx sensitivity for bacterial pathogens was higher in the population with confirmed fever (95%) compared with the full population (85%). In a previous prospective multicentre clinical trial, FebriDx revealed a bacterial sensitivity of 80%, and specificity of 93%; however, only 13% of the patients had a fever at the time of enrolment, possibly accounting for the lower sensitivity in the prior study [8]. Most importantly, the FebriDx test was found to have a 97–99% NPV that supports a delayed antibiotic prescribing strategy for FebriDx-negative or viral-positive patients.

FebriDx showed an overall viral agreement of 86%, sensitivity of 90% and specificity of 78%. However, when an elevated MxA ELISA results were added to the criterion standard for viral detection in patients with no pathogens detected by the reference testing, FebriDx specificity for viral infection increases to 84% which may represent a tautology with the MxA in FebriDx or the presence of other viruses such as bocavirus, herpes simplex virus (HSV), cytomegalovirus (CMV) or human immunodeficiency virus (HIV).

In the previous study where the majority of patients were afebrile at the time of enrolment, 62% of the patients were found to have a microbiologically unconfirmed respiratory illness (MURI) [8]. In this study, where more than half the patients were febrile, the rate of MURI was reduced to 28%.

The FebriDx test is a 10-minute, single-use, disposable POC diagnostic test designed to rapidly provide clinicians with actionable results for the management of outpatient URIs. In one small study, FebriDx was shown to impact outpatient clinical management decisions in 48% of cases and reduced antibiotic use by 80% [9]. FebriDx simultaneously and qualitatively identifies elevated levels of CRP and MxA, biomarkers that are associated with bacterial and viral infections, respectively [12,28]. CRP is a non-specific acute phase protein, when elevated above 20 mg/L suggests a clinically significant immune response but cannot reliably differentiate between viral and bacterial aetiology [11,29,30], whereas a CRP below this 20 mg/L threshold are likely to have non-bacterial or self-limited infections [11,29,30]. Meta-analyses and systematic reviews have shown that a standalone CRP measurement at cut-off of 20 mg/ml can identify clinically significant acute respiratory infection requiring antibiotic therapy without impacting morbidity [11,29,30] but cannot differentiate viral from bacterial infections [8,9]. Using CRP to guide therapeutic decisions can help to prevent missing a serious bacterial infection but will lead to nearly 40% overtreatment of viral infections [8]. A CRP concentration ≥ 20 mg/L triggers a positive CRP reading on the FebriDx test.

MxA is a viral biomarker with broad antiviral activity against both RNA and DNA viruses [31] and is upregulated by interferon-α and interferon–β (type-I interferons) in the presence of viral infection [12,32]. MxA concentrations ≥ 40 ng/ml, a level that has been identified as a sensitive threshold for identifying viral infections, triggers a positive reading on FebriDx [10,12,33]. MxA provides the required specificity to the CRP interpretation to allow for guiding therapeutic decisions. By combining CRP and MxA into a single test, FebriDx includes sensitive markers of both bacterial and viral infection.

The majority of URI’s are self-limiting, including confirmed bacterial infections and uncommonly lead to complications [1,34]. Coinfections are rare and occur in less than 2% of cases [1]. Melbye et al measured daily CRP in the course of untreated URI and showed a moderately elevated CRP value (10–60 mg/l) is a common finding in viral URI, with a peak during days 2–4 of illness, while persistent CRP elevations or new elevations in CRP may indicate a bacterial infection [35].

A falling CRP, is an indicator of clinical improvement and/or therapeutic response [35]. PCR may still identify DNA associated with a pathogen long after an infection has cleared [36–38]. In addition, rare to low culture growth is not uncommon during the first few days of a healing process, even after the initiation antibiotics [36]. Evidence of an elevated CRP along with bacterial cell culture growth on routine blood agar, chocolate agar, and MacConkey agar or bacterial PCR positivity is indicative of an ongoing clinically significant infection [39]; however, lack of elevated CRP would suggest clinical improvement [35] and this may explain the lower sensitivities and the apparent disconnect with the performance values when testing after the resolution of fever.

Infection is defined by the establishment of a microorganism within a host, while infectious disease applies when the interaction causes damage or an altered physiology resulting in clinical signs and symptoms [40]. Viral infections are frequently present in the oropharyngeal and nasopharyngeal, asymptomatically and symptomatically and may lead to bacterial colonization, however this does not necessarily lead to clinically significant infection [41]. Pathogen-specific testing, whether antigen, culture or molecular, cannot discriminate between colonization or the carrier state [42,43]. Differentiation of infectious disease from colonization or a clinically insignificant local infection requires the demonstration of a systemic antibody response, though not practically measured in the outpatient setting [44]. Host biomarkers, such as CRP and MxA, may serve as surrogates for the antibody response [8,10,11].

Historically, there has been a lack of rapid and accurate diagnostic tests to help clinicians distinguish between viral and bacterial infection [1,6,7,45,46]. The ability to rapidly differentiate viral from bacterial infection is essential for successful outpatient antibiotic stewardship. Biomarkers such as standalone CRP and PCT, as well as CRP in combination with either MxA or both TRAIL and IP-10, or gene expression studies are the leading candidates for assisting clinicians to achieve this goal [8,47–50]. There is no definitive reference standard for distinguishing between viral and bacterial respiratory infections, which is especially problematic in the afebrile state; we used an algorithm-based approach to maximize reproducibility and validity of our reference standard, but some misclassification is possible. When comparing these biomarkers, alone or in combination, it is important to normalize the analysis to those studies testing patients with a confirmed fever at the time of enrolment from those reporting a fever within several days of testing (Table 4). It is also important to recognize that testing hospital inpatients, especially those with pneumonia and sepsis, are more likely to have more robust immune responses that are not necessarily seen in lower acuity outpatient infections. Examining outpatient pharyngitis, the median (range) values of the PCT was shown to be 0.374 ng/ml (0.11–6.5) [26] and CRP has been shown to be CRP 44 mg/L (38–60) [51] to 50 mg/L (22–71) [26]. When comparing clinical performance of diagnostic tests, defining test performance in afebrile patients at the time of enrolment poses challenges because of the difficulties in establishing an appropriate reference method, resulting from persistent antigen or nucleic acids during an initial clinical cure.

Table 4. FebriDx compared to other host protein and gene expression tests.

Device	Population	Sample size	Confirmed fever at time of testing (hyperthermia)	Non-febrile – reported fever within past 3 days	Bacterial sensitivity	Bacterial specificity	Bacterial positive predictive value	Bacterial negative predictive value	Time to results (minutes)	Citation
FebriDx Qualitative	Outpatient/ER	220	55% (121)	45% (99)	85%	93%	69%	97%	10	Current
MxA + CRP	Outpatient/ER	121	100% (121)	None	95%	94%	76%	99%	10	study
Standalone PCT 0.25 ng/ml	Outpatient	220	55% (121)	45% (99)	41%	94%	54%	90%	20
Standalone CRP 20mg/L	Outpatient	209	55% (115)	45% (94)	81%	51%	23%%	94%	5 to 30
FebriDx Qualitative MxA + CRP	Outpatient/ER	205	13% (26)	87% (179)	80%	93%	63%	97%	10	[8]
CRP + TRAIL + IP-10	Inpatient	124	96% (119)	4% (5)	91%	93%			120	[39]
Standalone CRP 20 mg/L	Inpatient	124	96% (119)	4% (5)	96%	58%			5 to 30	[39]
CRP + TRAIL + IP-10	Inpatient	443	100% (443)	0%	86.70%	91.10%	60.50%	97.80%	120	[40]
Standalone Procalcitonin 0.5 ng/ml	Inpatient	506	100% (506)	0%	85%	87.10%	51.70%	97.20%	5 to 30	[40]
CRP + TRAIL + IP-10	Inpatient	307	100% (361)	0%	93.80%	89.80%			120	[41]
Standalone PCT 0.5 ng/ml	Inpatient/ER	307	100% (307)	0%	63.10%	82.30%			20	[41]
Standalone CRP 40 mg/L	Inpatient/ER	307	100% (307)	0%	88.20%	73.20%			5 to 30	[41]
Gene expression	Inpatient	1057	100% (1057)	0%	94%	59.8	98.30%	29.20%	240 to 360	[42]

Strengths of the current study include (1) the prospective, multicentre design; (2) standardized procedures and definitions used across all sites; (3) appropriate blinding of FebriDx and reference standard results to personnel who performed FebriDx and reference testing measurements; and (4) use of a clinical algorithm to determine infection [8,11]. With an overall agreement 94%, sensitivity of 85–95% and specificity of 93–94%, and NPV of 97–99%, data are supportive of using FebriDx in a clinical setting to assist with decisions about prescribing antibiotics. Potential advantages of FebriDx include the convenience of simultaneously screening for pathogenic bacteria and viruses with a single test and the rapidity of the results. Although FebriDx cannot differentiate a rare coinfection, the 97–99% NPV suggests that the FebriDx test may support an initial watchful waiting antibiotic strategy [8,11].

Our study had limitations. First, since children only represented 5% of the overall enrolment, this study better reflects an adult population. Second, our population is best described as a convenience sample because of the constraints of shipping biological samples and availability of research staff. Third, blood and sputum cultures were not performed due to low anticipated yield for these tests in outpatients with URIs. Fourth, we did not test for other possible viral pathogens known to cause URI including HSV, CMV, bocavirus or HIV. Fifth, we only captured information during a patient’s index visit; including follow-up information from later visits may have provided additional data useful for classifying URIs as viral or bacterial. Finally, it is possible that patients had a coexisting infection outside the respiratory tract.

Conclusions

FebriDx is an outpatient POC test designed to identify a clinically significant systemic host immune response and aid in the differentiation of viral and bacterial infection through rapid measurement of MxA and CRP from a fingerstick blood sample. In this study of diagnostic accuracy, the FebriDx test was determined to be an accurate test, with a 85% sensitivity, 93% specificity and 97% NPV to rule out bacterial infection for any patient presenting with symptoms and reported fever within the prior 3 days. In confirmed febrile patients at the time of testing, the test was even more sensitive (95%) and specific (94%) with a 99% NPV. Comparatively, FebriDx reveals greater accuracy than a standalone CRP or PCT test for identifying clinically-important viral and bacterial URI and similar accuracy to other combined protein biomarker panels; however, FebriDx is easier to use and more rapidly reports. FebriDx may support antibiotic stewardship by rapidly identifying clinically significant bacterial URIs.

Acknowledgements

We would like to warmly thank then following people for their tremendous help during this study and manuscript preparation: Jennifer Kasper, Adrienne Baughman, Itegbemie Obaitan, Jasmine Gale, Guruprasad Jambaulikar, Nivedita Patkar, and Paige Farely.

Disclosure statement

The authors received research funding from RPS Diagnostics to conduct this study. Dr. Shapiro reports serving as a paid consultant for Baxter and Cytovale and has received funding from Siemen’s diagnostics and LAjolla pharmaceuticals. Dr. Self reports serving as a paid consultant for Ferring Pharmaceuticals, Cempra Pharmaceuticals, BioTest AG, and Abbott Point of Care. Dr. Hou has served as a paid consultant for Cheetah Medical. Dr. Kurz has received honoraria from Zoll Medical Corporation, and has received research funding from Zoll Medical Corporation and Boehringer-Ingelheim. Dr. Sambursky is an executive at RPS Diagnostics that receives a salary and owns stock options.

Additional information

Funding

This work was supported by RPS Diagnostics, the maker of FebriDx. None of the clinical investigators have any financial interest in RPS Diagnostics. Dr. Sambursky has a financial interest in the company as described above.