Use of the FebriDx® host-response point-of-care test may reduce antibiotic use for respiratory tract infections in primary care: a mixed-methods feasibility study

Abstract Introduction FebriDx® is a CE-marked, single-use point-of-care test with markers for bacterial [C-reactive protein (CRP)] and viral [myxovirus resistance protein A (MxA)] infection, using finger-prick blood samples. Results are available after 10–12 min. We explored the usability and potential impact of FebriDx® in reducing antibiotic prescriptions for lower respiratory tract infection (LRTI) in primary care, and the feasibility of conducting a randomized controlled trial (RCT). Methods Patients (aged ≥1 year) with LRTI deemed likely to receive antibiotic prescription were recruited at nine general practices and underwent FebriDx® testing. Data collection included FebriDx® results, antibiotic prescribing plan (before and after testing) and re-consultation rates. Staff completed System Usability Scale questionnaires. Results From 31 January 2023 to 9 June 2023, 162 participants participated (median age 57 years), with a median symptom duration of 7 days (IQR 5–14). A valid FebriDx® result was obtained in 97% (157/162). Of 155 patients with available results, 103 (66%) had no detectable CRP or MxA, 28 (18%) had CRP only, 5 (3%) had MxA only, and 19 (12%) had both CRP and MxA. The clinicians’ stated management plan was to prescribe antibiotics for 86% (134/155) before testing and 45% (69/155) after testing, meaning a 41% (95% CI: 31%, 51%) difference after testing, without evidence of increased re-consultation rates. Ease-of-use questionnaires showed ‘good’ user-friendliness. Conclusions Use of FebriDx® to guide antibiotic prescribing for LRTI in primary care was associated with a substantial reduction in prescribing intentions. These results support a fully powered RCT to confirm its impact and safety.


Introduction
Clinically differentiating bacterial from viral lower respiratory tract infection (LRTI) is challenging, with LRTI having the highest inappropriate antibiotic prescribing rates of conditions seen in primary care. 1 Inappropriate antibiotic use risks side effects and drives antimicrobial resistance. 2 Rapid diagnostic testing ['point-of-care testing' (POCT)] has potential to reduce antibiotic use, [3][4][5][6] but its adoption into UK primary care remains limited. 7,8briDx ® (Lumos Diagnostics, USA) 9 is a single-use, hand-held, lateral flow POCT device designed to help distinguish bacterial from viral infections.It detects two host-response proteins, creactive protein (CRP) and myxovirus resistance protein A (MxA), in finger-prick blood, with results available after 10-12 min.CRP is an acute phase reactant that generally increases to higher levels with bacterial compared to viral infection, and MxA is a derivative of interferon α/β associated with viral infection. 10,11s a dual-marker test, FebriDx ® may be more clinically useful than POCT devices detecting a single biomarker (typically CRP

Eligibility criteria
Patients (aged ≥1 year) presenting to their GP surgery remotely or inperson with symptoms suggestive of a LRTI were eligible following clinical assessment if a prescribing clinician deemed that they would be likely to prescribe antibiotics in the absence of further diagnostic testing.We defined suspected LRTI as a cough, lasting <21 days, judged to be infective in origin, with other symptoms or signs localizing to the lower respiratory tract (shortness of breath, sputum, chest pain). 23Antibiotic prescriptions could be immediate or delayed (advised to wait for a specified period before taking them, and only if necessary).Patients were ineligible if they had taken antibiotics in the last 30 days or were unwilling/unable to provide informed consent.

Intervention
FebriDx ® (Lumos Diagnostics, USA) 9 is a CE-marked, FDA-approved, single-use POCT device with a turnaround time of 10-12 min (Figure 1).Capillary blood obtained by finger-prick (5 μL) of is drawn into a sample tube, transferred to a lateral flow strip, and test reagents released with a button.Results are generated in the form of three lines: a grey line indicating elevated CRP (lower limit of detection = 20 mg/L), a red line indicating elevated MxA (lower limit of detection = 40 ng/ml), and a blue control line indicating a valid test.An elevated MxA, with or without elevated CRP, is suggestive of viral infection.The presence of elevated CRP alone is suggestive of a bacterial infection.Presence of a control line only indicates a negative test result for both markers.
Verbal and written guidance was provided during study training.Practices were given flexibility over how to integrate the FebriDx ® into their clinics.In the case of a failed test, participants were offered repeat testing.Once results were available, these were interpreted by the recruiting clinician and communicated to the patient before proceeding with any clinical management deemed appropriate.Clinicians were advised to provide clear safety-netting advice regarding the need to seek medical attention in the event of persistent or worsening symptoms.
Optional nasopharyngeal swabbing was introduced part-way through the study.Those who consented to this aspect were asked to provide a swab (taken by the staff or patient) that was then posted to Southampton General Hospital microbiology laboratory.Nasopharyngeal swabs were frozen on arrival, and later underwent viral analysis by multiplex PCR.

Data collection
Data were collected by participating staff via an online case-report form completed in a sequential fashion before and after FebriDx ® testing.Initial pre-test data included baseline patient characteristics, clinical features of the presenting illness, clinicians' perception of the likelihood of bacterial aetiology (graded 1-10 on a Likert scale, with 10 = 'very likely'), and clinicians' antibiotic prescription plan had no further testing been available (immediate, delayed, or no antibiotics).
Post-test data included FebriDx ® test result, the time of collection/result/when the patient was informed, clinicians' post-test perception of the likelihood of bacterial aetiology on the same 10-point Likert scale, clinicians' post-test antibiotic prescription plan, and clinicians' post-test confidence in the need for antibiotics (graded 1-5 on a Likert scale, with 5 = 'very confident that antibiotics ARE needed').Follow-up data (after 28 days) included subsequent healthcare contacts, antibiotic prescriptions, and serious complications (including sepsis or death).Practice-level data collected included socioeconomic status (Index of Multiple Deprivation).
At the end of the study, practice staff were invited to complete an anonymous online ease-of-use questionnaire regarding the use FebriDx ® .This contained the System Usability Scale (SUS), a well-established usability score that involved grading FebriDx ® on a 5-point Likert scale across 10 usability criteria 24 (Figure S1, available as Supplementary data at JAC Online).

Sample size
As a feasibility study, a formal sample size calculation was not required. 22ith regards to antibiotic use, we calculated that 156 participants would allow us to describe feasibility or outcome rates of 50% to within a 95% confidence interval of ±7.8%.Rates higher or lower than 50% would be described with a greater precision.
Wilcox et al.

Data analysis
Statistical analysis was performed using STATA v.18 (StataCorp, USA, 2023).As this was a feasibility study, descriptive statistics are reported.Comparison of FebriDx ® with viral PCR was used to assess diagnostic accuracy (sensitivity and specificity).The analysis was conducted by C.W. with oversight from N.F./T.B./N.I.

Test results and time to result
A valid test result was obtained in 97% (157/162) of participants: on the first attempt in 86% (139/162), on the second attempt in 10% (15/162), on the third or fourth attempt in 2% (3/162), and was abandoned in 3% (5/162).Reasons for initial test failure are displayed in Figure 2, and were most commonly due to difficulty obtaining sufficient blood from the fingerpick, or insufficient filling of the blood transfer tube.For two participants, the clinician documented that they obtained a valid test result, but did not document the result.Therefore, test results were available for 96% (155/162).
Clinicians' stated management plan was to prescribe immediate or delayed antibiotics for 86% (134/155) of participants before FebriDx ® testing and 45% (69/155) after testing, meaning there was a 41% (95% CI: 31%, 51%) difference before and after FebriDx® in primary care: a feasibility study testing (Table 3).Following testing, 47% (73/155) had an antibiotic treatment plan that was likely to reduce antibiotic use (change from immediate antibiotics to none or delayed), 45% (70/155) had no change in their treatment plan, and 8% (12/155) had a change that would likely result in increased use (Table S2).
Only those with a CRP-positive only result were more likely to receive antibiotics after testing, with all other results being associated with a reduction in antibiotic use (Figure 3).Clinicians indicated that they planned to prescribe antibiotics to 34% (35/103) of participants with a negative test result, 100% (28/28) with a CRP-only positive result, 0% (0/5) with a MxA-only result, and 32% (6/19) with a combined CRP/MxA positive result.

Follow-up data on antibiotic use and re-attendance
Follow-up data were obtained via clinical notes review (after 28 days).The clinical records differed from the study case-report form on three occasions: one where a patient to be prescribed immediate antibiotics was admitted directly to hospital; and two where the initial management plan following a telephone review was changed after a planned face-to-face review with a GP later the same day (one prescribed immediate antibiotics rather than delayed antibiotics, and one prescribed no antibiotics rather than delayed antibiotics).
Twenty-three percent (35/155) sought additional medical attention for the same illness within 28 days of their initial consultation (Table S3).No serious adverse events were recorded.The highest re-consultation rate was seen among those prescribed immediate antibiotics (33%, 17/52).Furthermore, re-attendance rates were higher among patients for whom the clinician kept to their pre-test decision to prescribe immediate antibiotics (32%, 13/41), compared with patients for whom the clinician changed their decision from immediate antibiotics to delayed (0%, 0/6) or no antibiotics (28%, 8/29) following testing (Table S4).Antibiotics were prescribed after re-consultation in 15% (23/155) of cases, of whom 43% (10/23) had not been prescribed antibiotics initially, meaning the overall antibiotic prescription rate within 28 days was 51% (79/155).

Viral PCR analysis and diagnostic accuracy
As a result of the late introduction of this voluntary aspect of the study we only obtained nasopharyngeal swab results for 18% (28/155) of participants and did not have sufficient data to report reliable test characteristics (Tables S5 and S6).Wilcox et al.

Ease-of-use questionnaires
the study, and at least one member from all sites.The mean SUS of 72.2 suggests a 'good' level of user-friendliness on a proposed adjective rating scale based on previous usability studies using the SUS. 24urther comments (Table S7) were provided by 38% (6/16) of respondents.Users were generally positive about the device, but acknowledged there was a 'learning curve' to its use.Additional specific points included practical difficulties with transferring blood from the collection tube onto the lateral flow test strip, difficulties interpreting results due to the faintness of result lines and the need for an even quicker turnaround time for it to be practical to integrate into a routine GP consultation.

Discussion
This is the largest study evaluating the potential clinical impact of FebriDx ® in primary care, and demonstrates that FebriDx ® testing may reduce unnecessary antibiotic use in patients with LRTI.Following FebriDx ® testing, clinicians felt more confident in their antibiotic prescribing decisions more than 80% of the time, with more confidence that antibiotics were not required in over 50%.In keeping with this, their plan to prescribe antibiotics reduced from 86% of participants before testing to 45% following testing.Ease-of-use assessment demonstrated good userfriendliness, but identified some technical challenges and the need for operators to become skilled in using the device.
Strengths of the study were that it was a multi-centre prospective study that collected data on clinician intention, confidence and actual behaviour, as well as participant re- FebriDx® in primary care: a feasibility study consultation, subsequent antibiotic use and outcomes.Exceeding our recruitment target meant an adequate sample size to address our feasibility outcomes.The main limitation is the lack of randomization, which limits our ability to conclude that the reduction in antibiotic use was caused by use of the FebriDx ® test, as clinicians may not in reality have acted according to their stated pre-test prescribing plan.Additionally, the feasibility nature of the study meant that we were limited to One of the 155 patients with a documented FebriDx ® result did not have data recorded on clinician confidence.
Wilcox et al.
descriptive statistics.Nevertheless, the strength and consistency of signal seen is highly suggestive of an important effect.Other limitations include the relatively short run-in period, a low number of children and an uneven distribution of participants.These increase the potential for selection bias and reduce the generalisability of our results, however our qualitative interview substudy (reported separately) does explore aspects of the usability/feasibility in more depth.The study had low ethnic minority representation and nearly all GP practices were in areas of high socioeconomic status (Index of Multiple Deprivation decile 9 or 10).The low number of viral swabs prevents us from providing data on diagnostic accuracy.6][27] A recent meta-analysis of CRP POCT devices for LRTI in primary care demonstrated a reduction in immediate antibiotic prescribing of 20% (without affecting symptom resolution or hospital admissions), however, this reduction was not maintained at 28 day follow-up, and there was a significant increase in re-attendance. 28In this study, re-attendance rates were similar to that seen in previous studies of LRTI, 29 and it was encouraging to see that re-attendance was actually lower among patients who were not prescribed immediate antibiotics following FebriDx ® testing.Only one small retrospective study at a single GP practice has previously studied the use of FebriDx ® in UK primary care, involving 21 patients (mean age 46 years). 12Of the 12 patients presenting with suspected bacterial aetiology, clinical management was reportedly altered in 67% (8/12) who were not subsequently prescribed antibiotics.No data was reported on re-attendance rates, test failure rate, diagnostic accuracy or ease-of-use. 12he low rate of MxA detection in our study was similar to that seen in recent studies of FebriDx ® for LRTI in secondary care (14%-16%).A lower rate of CRP detection meant that the rate of negative results was higher in our study compared with these studies (20%-49%), 13,16,[18][19][20] which may be due to differences in our primary care patient cohort (including lower disease severity).It is also worth noting that nearly half of our participants presented with over a week of symptoms, and given that MxA is known to rise very early in viral infection (with a half-life of 2 days), this may have also contributed to the low MxA detection rate. 11In our study, 46% (24/52) of those with positive test results had detectable MxA, either alone or combined with CRP.When considering the beneficial effect of MxA testing, if only CRP testing were available, we can estimate that all 19 (an additional 13) participants with a combined MxA/CRP result would have been prescribed antibiotics and 32% of those with a MxA-only positive result (an additional two participants).Therefore, MxA testing is likely to have led to an extra 10% (15/ 155) reduction in antibiotic prescribing over CRP testing alone.A joint MxA-CRP result may indicate a viral infection with an associated inflammatory response, or a 'dual' infection/bacterial superinfection.Thorough clinical assessment and safety-netting is therefore key, but unless pneumonia is suspected, such patients can usually be managed safely without antibiotics. 23his is the first study to evaluate FebriDx ® ease-of-use.As a single-use, hand-held test, FebriDx ® offers advantages over many current POCT devices that require an additional desktop analyser, especially in the primary care setting where clinicians usually work in single rooms. 3,5,6,12,30There are likely to be technical challenges initially, and users need experience before they can use test reliably.Understanding usability issues is important as they will impact on adoption into routine care, 3,5,6,31 and we have conducted a qualitative process analysis alongside this study, which we report separately.The test failure rate was higher than the 0%-5% rate reported in recent UK studies of FebriDx ® as an emergency department triage tool, 15,16,18 possibly due to a higher degree of operator error (at least initially) compared with users in emergency departments who perform a higher number of tests.Longer run-in periods in those studies may have allowed users to gain confidence before data collection. 15,16,18hese results support a funding application for a fully powered trial to assess the impact of using FebriDx ® to guide antibiotic prescribing for LRTI in primary care.A future trial should also assess impact on symptoms and safety (including re-attendance) and cost-effectiveness, particularly as costs of implementation are a key barrier to routine adoption of POCT. 3,5,26At approximately £12.75 per FebriDx ® test (shelf life of 18 months), the overall cost is similar to CRP POCT cartridges, but without any additional up-front or maintenance costs. 30,32It is also important to assess clinician and patient views on FebriDx ® to explore feasibility and usability in more depth.This includes experience of reading/interpreting results and communicating these to patients, as well as overall patient satisfaction and the feasibility of integrating FebriDx ® into real-life practice.We will explore these in our qualitative interview sub-study (reported separately).
Future studies should also assess the role of FebriDx ® for upper respiratory tract infections for which antibiotics are commonly prescribed (such as sinusitis), as well as the impact on antiviral prescriptions.It is also important to consider the implementation of FebriDx ® and other POCT devices within the wider primary care system.Delivery of primary care in the UK is evolving, and involves a diverse range of allied health care professionals, including dedicated LRTI clinics at primary care network level.POCT testing in such clinics may be more effective and sustainable than opportunistic use in a traditional clinic setting.Future research should also consider assessing the use of FebriDx ® in other settings, such as nursing homes and out-of-hours urgent care (settings associated with the highest rates of inappropriate antibiotic prescribing 33,34 ), as well as community pharmacies, considering the expanding role of POCT and antibiotic prescribing in this setting. 35inally, the 'real world' diagnostic accuracy of FebriDx ® in the primary care setting should be assessed, as data from secondary care cannot necessarily be extrapolated as the sensitivity of a test may vary by disease severity (spectrum bias).Future analyses should also explore differences in those presenting in the first week of illness (for which FebriDx ® is formally marketed) compared with those presenting after 7 days.Assessment of MxA diagnostic accuracy may be confounded by a low viral load (i.e.low level viral RNA can be detected for prolonged periods after the host immunological response has resolved), as well as certain viruses (including Rhinovirus) that are largely confined to the respiratory tract and may not be associated with a detectable MxA response. 36Assessing accuracy for bacterial detection is also challenging due to the lack of reference standard and inability to distinguish colonizing organisms from pathogens, and FebriDx® in primary care: a feasibility study so would likely rely on clinical adjudication alongside laboratory biomarkers and pathogen detection. 20

Conclusions
Use of FebriDx ® may reduce unnecessary antibiotic use in patients with LRTI.These findings need confirming in an adequately powered RCT, and our study has found good evidence for the feasibility of conducting such a trial.

Figure 1 .
Figure 1.The FebriDx ® device and its possible results.(a) The FebriDx ® device.(b) Negative result with control line.(c) CRP-only positive.(d) MxA-only positive.(e) CRP and MxA positive.Figures 1b-e supplied by Lumos Diagnostics, USA.This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Figure 2 .
Figure 2. Flow diagram for the study.
System usability score (SUS) questionnaires were returned from 89% (16/18) of GP practice staff who used FebriDx ® devices in

Table 1 .
Baseline demographics and clinical characteristics of patients

Table 2 .
FebriDx ® results in all patients, and those with a symptom duration of ≤7 and >7 days a Ten of the 155 patients with a documented FebriDx ® result did not have data recorded on symptom duration.

Table 3 .
Antibiotic prescription plan before and after FebriDx ® testing Antibiotic prescribing plan before and after testing split by FebriDx ® result.This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Table 4 .
Clinician's confidence in the need for antibiotics following FebriDx ® testing, split by FebriDx ® result a