Application of a simple point-of-care test to reduce UK healthcare costs and adverse events in outpatient acute respiratory infections

Abstract

Background: Acute respiratory infection (ARI) accounts for over two-thirds of total antibiotic prescriptions although most are caused by viruses that do not benefit from antibiotics. Most antibiotics are prescribed in the outpatients setting. Antibiotic overuse leads to antibiotic-related adverse events (AEs), inclusive of secondary infections, resistance, and increased costs. Point-of-care tests (POCT) may reduce unnecessary antibiotics. A cost analysis was performed to assess diagnostic POCT options to identify patients with an ARI that may benefit from antibiotics in a United Kingdom (UK) outpatient setting.

Methods: Healthcare savings were estimated using a budget impact analysis based on UK National Institute for Health and Care Excellence (NICE) data and direct costs (antibiotics, AEs, POCTs) derived from published literature. Otitis media, sinusitis, pharyngitis and bronchitis were considered the most common ARIs. Antibiotic-related AE costs were calculated using re-consultation costs for anaphylaxis, Stevens-Johnson syndrome, allergies/diarrhea/nausea, C. difficile infection (CDI). Potential cost-savings from POCTs was assessed by evaluating NICE guideline-referenced POCTs (CRP, FebriDx, Sarasota, FL) as well as a target product profile (TPP).

Results: Fifty-percent (7,718,283) of ARI consultations resulted in antibiotics while guideline-based prescribing suggest appropriate antibiotic prescriptions are warranted 9% (1,444,877) of ARI consultations. Direct antibiotic costs for actual ARI consultations associated with antibiotics was £24,003,866 vs. £4,493,568 for guideline-based, “appropriate” antibiotic prescriptions. Antibiotic-related AEs and re-consultations for actual vs. appropriate prescribing totaled £302,496,486 vs. £63,854,269. ARI prescribing plus AE costs totaled £326,729,943 annually without the use of delayed prescribing practices or POCT while the addition of delayed prescribing plus POCT totaled £60,114,564–£78,148,933 depending on the POCT.

Conclusions: Adding POCT to outpatient triage of ARI can reduce unnecessary antibiotics and antibiotic-related AEs, resulting in substantial cost savings. Further, near patient diagnostic testing can benefit health systems and patients by avoiding exposure to unnecessary drugs, side effects and antibiotic resistant pathogens.

Key points for decision makers

• Many patients are unnecessarily treated with antibiotics for respiratory infections.

• Antibiotic misuse leads to unnecessary adverse events, secondary infections, re-consultations, antimicrobial resistance and increased costs.

• Point-of-care diagnostic tests used to guide antibiotic prescriptions will avoid unnecessary adverse health effects and expenses.

Keywords:

• Acute respiratory infection
• point-of-care tests (POCT)
• host response biomarkers
• antibiotic-related adverse events
• FebriDx
• C-reactive protein (CRP)
• myxovirus resistance protein A (MxA)

JEL CLASSIFICATION CODES:

• C53
• I15
• I19

1. Introduction

Acute respiratory infection (ARI) comprised of otitis media, sinusitis, pharyngitis, acute cough and bronchitis account for more than two-thirds of the total antibiotic prescriptions for all medical conditions, even though most are caused by viruses for which antibiotics have no role in treatment. Most of these prescriptions occur in outpatient general practitioner (GP) or out of hours clinics. Antibiotic overuse leads to antibiotic-related adverse events (AEs), inclusive of antibiotic-associated secondary infections, antibiotic resistance, and increased healthcare costs¹.

Antibiotics are a common cause of medication-related severe AEs, ranging from mild (e.g. nausea, diarrhea or rash) to life-threatening (e.g. anaphylaxis, C. difficile infection (CDI))^2–5. Globally, it is estimated that nearly 700,000 people die annually, and 12,000 in the United Kingdom (UK) alone, as a direct result of resistant infections⁶ and the economic burden of antimicrobial resistance (AMR) is estimated to range from $20,000 to $2,15,000 United States Dollar (USD) per case, with a predicted $3 trillion USD in global gross domestic product (GDP) loss by 2050⁷. Due to the threat of rising antibiotic resistance, the UK government intends to reduce inappropriate antibiotic prescribing by 50% by 2020⁶. Efforts aimed at reducing unnecessary antibiotic prescriptions for uncomplicated ARIs could result in the most sizable contribution to achieving the goal^8,9.

Diagnostic uncertainty coupled with the fear of missing a serious infection and patient expectations frequently drive unnecessary antibiotic prescriptions^10–14. Diagnostic test offerings in the outpatient setting are limited and thus clinicians rely on history, physical exam and sometimes scoring criteria to evaluate severity of illness and need for antibiotics. Physical examination alone has a sensitivity ranging from 50 to 70% and specificity of 60–75% for diagnosing pneumonia¹⁵. The Centor criteria for identifying bacterial pharyngitis requiring antibiotics demonstrate 40–55% accuracy at identifying group A streptococcus compared to bacterial throat culture¹⁶. A FeverPAIN criteria score of 4 or 5 has a 62–65% probability of having a bacterial infection, which is slightly higher than people with a Centor score of 4 who have a 55% probability of a bacterial infection¹⁷. Further, antibiotic prescriptions are also overprescribed because they have no direct cost to the GP surgery itself and GPs often do not consider their own prescribing practices as contributing to increasing antibiotic resistance¹⁸. Of more than 1,000 UK GPs surveyed, 55% felt patient/parent pressure to prescribe antibiotics, even when unsure antibiotics were necessary, while 44% had prescribed antibiotics to get a patient to leave the clinic or surgery¹⁹. This misuse of antibiotics has been associated with antibiotic-related AEs, the emergence of resistant pathogens and rising healthcare costs.

Interventions which reduce diagnostic uncertainty would likely be effective in promoting prudent antibiotic use about appropriate ARI management in primary care. Several laboratory based tests are available to detect pathogen etiology of a suspected ARI but are not frequently used in the UK due to concern over suboptimal diagnostic performance, high costs, and turnaround times that exceed the decision window for initiating treatment often result in prescription of antibiotic therapy in the absence of a bacterial infection^20,21. Unfortunately, this has led to continued overprescribing of antibiotics that carries direct prescribing costs, increased re-consultations and the major threat of antibiotic resistance^20,21.

Novel point-of-care tests (POCT) have the potential to improve outcomes in primary care by optimizing prescribing decisions, reducing referrals, improving efficiency of care and decreasing costs²². The effectiveness of POCTs to limit antibiotic prescribing in lower respiratory tract infection (LRTI) by 21% using standalone CRP has been demonstrated in a recent study^23,24. GP surveys showed that at least 50–60% of respondents expressed interest in using POCTs (e.g. C-reactive protein (CRP), influenza, and Group A streptococcus) which suggests a need for a more widely accessible range of POCTs to aid clinicians with immediate decisions (urgent referrals, or immediate treatment decisions such as the decision to treat with antibiotics)²⁵.

Use of diagnostic tests has been recognized as a pathway to reduce the overuse of antibiotics, particularly simple, low-cost tests that can guide antibiotic prescription at the first point-of-contact^6,26. Host biomarkers, such as CRP and Myxovirus resistance protein A (MxA), are indicative of the body’s systemic immune response to a clinically significant infection, whereas the lack of a systemic immune response suggests a carrier state or colonization²⁷. CRP is nonspecific for inflammation while MxA is specific for viral infection and increases the specificity of standalone CRP tests. When paired together, host-response biomarkers, such as CRP and MxA, represent a systemic immune response to a clinically significant infection, whereas the lack of a systemic immune response in the presence of a detected pathogen suggests a carrier state or colonization²⁷. Although there are no diagnostic tests that demonstrate 100% accuracy, standalone CRP has a median sensitivity of 86%, specificity of 69%, and overall accuracy of 80%^19,28 for identifying a patient requiring antibiotic therapy, while the dual biomarker (CRP, MxA) FebriDx test has 95% sensitivity, 94% specificity, and overall accuracy of 92%²⁹.

A 1-year budget impact cost analysis model was developed using data from publicly available datasets and published literature to assess the costs incurred and cost savings associated with and without the use of POCTs in the diagnosis of ARI in the outpatient setting. Specifically, two POCTs included in National Institute for Health and Care Excellence (NICE) guidance for respiratory infection (standalone CRP; dual biomarker (CRP, MxA) FebriDx) as well as a hypothetical test which met a target product profile (TPP) defined by the global health community in 2015²⁶ were compared to no POCTs and to one another.

2. Materials and methods

2.1. Data collection

A structured approach for costing, using the NICE clinical guidelines, NHS websites, and peer reviewed literature, was used as the basis of this cost analysis. Published data related to ARI, including all upper respiratory infections including cough and bronchitis diagnosed in GP and urgent care settings and was applied to a derivation of the NICE costing model³⁰.

2.2. Patient population

According to the 2018 UK office of National Statistics, the UK population was assumed to be 66,400,000³¹. More patients consult GPs for ARI (30%) than for any other single disease³². In the UK, ARI accounts for 300–400 registered patients for GP consultations annually per 1,000 registered patients³³.

Based on data from the UK GP Research Database, it was conservatively assumed that 300 patients/1,000 registered patients, or approximately 19,920,000 patients, visit a GP for an ARI annually³³. Consultations for lower respiratory infection inclusive of bronchitis was found to be 70 consultations per 1,000 registered patients each year, or approximately 4,648,000 patients³⁴. This suggests 77% or approximately 15,338,400 patients, would be defined as an ARI, inclusive of non-specific ARI (e.g. common cold, laryngitis, tracheitis, and flu like illness), acute otitis media, acute rhinosinusitis, and acute pharyngitis without acute bronchitis³².

The number and proportion of actual vs. appropriate antibiotic prescriptions per year was calculated for the following conditions: non-specified ARI (common cold, laryngitis and tracheitis), acute otitis media, acute rhinosinusitis, acute pharyngitis/sore throat and acute bronchitis/cough³⁵ (Table 1). Actual and appropriate antibiotic prescriptions were calculated based on antibiotic prescription rates established by a recent analysis of 3.7 million active patient cases, representing approximately 7% of the general UK population, that were extracted from The Health Improvement Network (THIN) database³⁵. The THIN database uses data from 550 GPs throughout the UK and is considered the reference standard for capturing GPs patient management patterns. Due to the broad database coverage and experience in modeling and healthcare economics, the 2018 analysis of ARI antibiotic prescribing practices in primary care was used as the basis for calculating actual antibiotic prescriptions. Analysis of GP prescribing demonstrates that approximately 50% of all ARI visits (7,718,283) patient visits lead to an antibiotic prescription. The proportion of consultations that resulted in an actual antibiotic prescription consisted of 25% for non-specified ARI (common cold, laryngitis, tracheitis and flu-like illness), 89% for acute otitis media, 88% for acute rhinosinusitis, 59% for acute pharyngitis/sore throat and 42% for acute bronchitis/cough³⁵. The number and proportion of consultations that warrant “appropriate” antibiotic prescriptions average of 9% (1,444,287) of ARI consultations per year derived from publications that evaluated appropriate antibiotic prescription rates based on current guidelines^{1, 35}. The consultations necessitating an appropriate antibiotic prescription consisted of 17% for acute otitis media, 11% for acute rhinosinusitis, 13% for acute sore throat and 10% for acute bronchitis/cough³⁵. Considering antibiotics are not indicated for non-specified ARI (common cold, laryngitis and tracheitis), it is assumed that the appropriate number of antibiotic prescriptions should be zero^1,35.

Table 1. Consultations resulting in antibiotic prescription: actual vs. appropriate prescribing assumptions calculated from published data (2014, 2018), combined for all age groups.

ARI conditions	Proportion of annual ARI consultations [Guilliford 2014]	Number of annual ARI consultations (proportion per condition × total ARI visits)	Proportion of ARI consultations receiving antibiotics (actual) [Pouwels 2018]	Number of annual ARI consultations receiving antibiotics (actual)	Proportion of ARI consultations receiving antibiotics (appropriate) [Pouwels 2018]	Number of ARI consultations receiving antibiotics (appropriate) [Pouwels 2018]	Number of ARI consultations receiving excess antibiotics (actual-appropriate)
Non-specified URI (common cold, laryngitis tracheitis and flu-like illness)	0.19	2,914,296	0.25	728,574	0	0	728,574
Acute otitis media (age 0 − 18 years)	0.06	920,304	0.89	819,071	0.17	156,451	662,618
Acute rhinosinusitis	0.09	1,380,456	0.88	1,214,801	0.11	151,850	1,062,951
Acute pharyngitis/sore throat/tonsillitis	0.27	4,141,368	0.59	2,443,407	0.13	538,377	1,905,029
Acute bronchitis/Cough	0.39	5,981,976	0.42	2,512,430	0.1	598,197	1,914,232
Total		15,338,400		7,718,283		1,444,877	6,273,406

A direct and indirect cost for antibiotics used in ARI (defined below) as well as the rate and costs associated with antibiotic associated AEs were determined. Within the ARI population, the number of actual patients treated with antibiotics and the presumed appropriate number of patients that should be treated with antibiotics were multiplied against the total number of patients receiving antibiotics to drive the base cost analysis. It was assumed that POCT would lead to an absolute reduction in antibiotics and proportionately reduce both their direct and indirect costs associated with re-consultations.

2.3. Antibiotic cost assumptions

Cost assumptions were categorized as direct costs (i) antibiotics plus dispensing fee at the primary encounter and (ii) indirect costs (a) re-consultations and subsequent treatments due to mild or non-severe antibiotic-related AEs (allergies, nausea, diarrhea) as well as (b) severe antibiotic-related AEs (anaphylaxis, Stevens-Johnson syndrome (SJS), CDI, and (c) diagnostic tests (throat culture and POCTs).

To determine the direct costs of antibiotics, average antibiotic costs per episode for otitis media, sore throat, and cough were extracted from the NICE guidelines which were also used to determine the appropriate antibiotic treatment duration by condition, pharyngitis a 10-day course, otitis media a seven day course, and cough a five day course^36–39. The 2019 Framlington Place Newcastle upon Tyne Regional Drug and Therapeutics Centre therapeutic costs were used to determine the specific antibiotic cost of treatment for each ARI condition (Table 2)⁴⁰. Antibiotic prescriptions for sinusitis and pharyngitis were assumed to be similar and were treated the same in the cost assumption (Table 2). Since published data indicate that patients older than 16 years of age represent the majority of patients receiving antibiotics, the most frequently prescribed antibiotics for patients 16 years and older was used to generate an average antibiotic cost of £1.21 per ARI episode⁴⁰. A dispensing fee of £1.90 was added to the average antibiotic cost and was assumed to be included for every prescription, based on the mean dispensing rate per item for pharmacists⁴¹. Out-of-pocket payments for prescription charges applicable to these medicines were not included.

Table 2. Antibiotic cost of treatment for each ARI condition.

Condition	Antibiotic	% Utilization	Price (£)/5–10 day course	References
Acute bronchitis/acute cough	Amoxicillin	79.0	1.12	Excellence NIFHaC^38
	Other (cefaclor, azithromycin, cefalexin, trimethoprim, flucloxacillin)	4.5	1.62
	Erythromycin	7.9	0.91
	Co-amoxiclav	2.9	1.37
	Clarithromycin	4.8	0.86
	Doxycycline	0.9	0.44
Otitis media	Amoxicillin	80.3	1.34	Excellence NIoHaC^39
	Other (cefaclor, doxycycline, azithromycin, cefalexin, trimethoprim, flucloxacillin)	4.0	1.73
	Erythromycin	7.5	1.09
	Co-amoxiclav	4.5	1.64
	Penicillin-V	1.2	1.44
	Clarithromycin	2.5	1.03
Acute rhinosinusitis Acute sore throat Non-specified URI^a (common cold, laryngitis and tracheitis, Flu-like Illness)^a	Penicillin -V	56.0	1.20	Excellence NIFHaC^38
	Amoxicillin	26.8	1.12
	Other (cefaclor, doxycycline, azithromycin, cefalexin, trimethoprim, flucloxacillin)	4.5	1.44
	Erythromycin	8.4	0.91
	Co-amoxiclav	1.5	1.37
	Clarithromycin	2.8	0.86

^a Antibiotics are not indicated for non-specified URI.

2.4. Adverse event and secondary infection assumptions

Four categories of antibiotic-related AEs were assumed to lead to re-consultations: anaphylaxis, SJS, allergies consisting of skin rashes, and gastrointestinal upset inclusive of nausea and diarrhea. Based on a literature analysis, an antibiotic-related AE rates were assumed to be 0.01% for anaphylaxis⁴², 0.04% for SJS⁴³, 10% for allergies^44–46, 25% for diarrhea and nausea^47,48. These rates were multiplied by the total amount of antibiotics prescribed to determine estimates for each subsequent AE. Secondary infections, such as CDI also contribute to significant costs. To estimate the frequency of CDI cases, only the community onset CDI rate of 12.1/100,000, or 7,865 annual cases was used and multiplied by the annual percentage of antibiotics prescribed for ARI (7,865 cases × 63.6%) to determine the incidence of 5,002 cases or (0.1%) serve as the 2019 base case⁴⁹.

2.5. Diagnostic POCT assumptions

The cost of each POCT was derived from literature, guidelines, or directly from the test manufacturer (Table 4). The TPP is reported to cost between £1 and 5²⁶. The cost calculations assumed a cost of £5 based on the availability of current, similar diagnostics that cost nearly 2–10 times this price range. The cost of each CRP test was determined to be £13.50, inclusive of the testing device, reagents, disposables, calibrators and controls¹⁷. The cost of each FebriDx test was determined to be £11.25 (direct correspondence with manufacturer).

Annual costs of diagnostic tests were estimated based on the direct costs of the tests multiplied by the prevalence of ARI, and the indirect costs from post-diagnostic antibiotic prescriptions, re-consultations due to AEs. For the purposes of illustrating a complete picture of the potential costs incurred and saved, POCT utilization was presumed to be 100%.

Diagnostic performance of each POCT was determined by review of published literature describing minimal and optimal test characteristics of TPP as well as currently available, regulatory approved POCTs mentioned in NICE guidance²⁶. Diagnostic performance characteristics of standalone CRP and FebriDx consisting of the sensitivity (true positive rate) and specificity (true negative rate) of each test was determined based on published evidence^29,50. The hypothetical tests based on the TPP used a sensitivity of 90% and specificity of 85%²⁶ (Table 3).

Table 3. Assumptions for the cost and accuracy of diagnostic tests.

POCT	Sensitivity (%) (median)	Specificity (%) (median)	Accuracy (%)	Cost (GBP)	References
Biomarker triage (TPP)	90–95	80–90	88	1 − 5^a	Dittrich et al.^26
C-reactive protein (CRP)	64–95 (86)	49–76 (69)	80^b	13.5^c	Simon et al.^50
Multiplex (FebriDx)	95	94	92	11.25	Shapiro et al.^29

Abbreviations. TPP, Target product profile; AUC, Area under the curve; POCT, Point-of-care-test.

^a Costs in the original publication were estimated in US dollars and for illustration purposes converted using a 1:1 conversion factor.

^b The accuracy value was averaged from a recent systematic review which found a sensitivity range of 0.59–95 in febrile populations.

^c Inclusive of analyzer reagent costs as per NICE 2017.

The number of post-diagnostic antibiotic prescriptions was estimated based on diagnostic performance of each POCT. It was assumed that 100%-true sensitivity for bacterial infection represents patients that should have received an antibiotic and were more likely to return to clinic for disease progression and required antibiotic therapy. Similarly, it was also assumed that 100%-true specificity represents patients that were falsely prescribed unnecessary antibiotics when an antibiotic was not necessitated.

2.6. Sensitivity analysis

A one-way sensitivity analysis was performed on major cost drivers such as the number/cost of antibiotics, number of antibiotic related AEs the cost of the POCTs and total ARI population. Total annual cost to the health system with and without POCT were calculated based on the aforementioned assumptions included a 25% increase and decrease in major cost drivers (antibiotics, AEs, total ARI population). POCT costs were varied by 20% as it would not be feasible to cover the cost of goods for most POCTs if costs were reduced by 25%. Results are presented in Table 6.

To further test the validity of the model, the total costs of antibiotics, AEs, and re-consultations related to antibiotic use was determined. It was assumed that all patients with anaphylaxis, SJS and CDI would reconsult to the health facility for care. Thus, the number and cost of anaphylaxis, SJS and CDI were considered to be constant when considering different prescribing strategies (i.e. immediate, delayed, and delayed plus POCT). Additionally, it was assumed that although the incidence of non-severe AEs (allergies, diarrhea/nausea) ranges upwards of 25%, only a small proportion (111.98/100,000 = 5.6%) of patients would actually return to a healthcare practitioner for care⁵¹. This was chosen as it is in line with published data⁵¹. Moreover, the GRACE study showed that the baseline re-consultation rate without antibiotics was 19.7% for worsening symptoms and increased to a re-consultation rate of 25.3% with immediate antibiotic use⁵¹. Thus, a 5.6% re-consultation rate was considered to be the base case for mild AEs (allergies, nausea, diarrhea) and was the only variable adjusted when calculating the impact of different prescribing strategies on the cost of AE re-consultation.

3. Results

Over 15 million patients visit a GP for management of ARI conditions annually. Despite NICE recommendations advocating for delayed antibiotic prescribing, 7,718,283 (50.3%) consultations were associated with an actual antibiotic prescription. When considering guideline-based prescribing practices, 1,444,877 (9%) consultations would have resulted in an appropriate antibiotic prescription (Figure 1). A difference of 6,273,405 (81%) consultations, which resulted in antibiotic prescriptions where not needed.

Figure 1. Budget impact analysis.

The corresponding antibiotic costs for actual ARI consultations resulting in antibiotic prescriptions amounted to £2,40,03,860 vs. £44,93,568 for guideline-based, appropriate antibiotic prescriptions.

Based on actual antibiotic prescriptions and the assumed rate of antibiotic-related AEs, it was calculated that 450,748 antibiotic-related AEs would occur annually under “status quo” prescribing practices. The annual cost of AEs resulting from actual antibiotic use was calculated to be £302,496,486. This was combined with the costs for antibiotics as well as the costs for throat cultures which amounted to a cumulative cost of £32,67,29,943 per year without the use of delayed prescribing practices or POCT (Figure 1).

Based on guideline-based, appropriate prescribing practices the number of antibiotic-related AEs was estimated to be 115,131, a 75% reduction when compared to status quo prescribing practices. The annual cost of AEs resulting from appropriate antibiotic prescriptions was calculated to be £6,38,54,269 which amounted to an 79% reduction in AEs when compared to status quo prescribing practices. The annual cost of AEs resulting from appropriate antibiotic prescribing was combined with costs for antibiotics amounted to a cumulative cost of £68,347,837. Total annual costs for AEs were also calculated for delayed prescribing plus the addition of each POCT (CRP, TPP and FebriDx). The total cost for AEs re-consultation inclusive of antibiotics was £7,81,48,933 for delayed prescription plus CRP, £6,01,14,564 for delayed prescription plus FebriDx, and £6,61,26,020 for delayed prescription plus TPP (Figure 1).

Total annual costs for each POCT prescription strategy was calculated by factoring in the cost of performing the POCT at each ARI consultation, influence of diagnostic performance on antibiotic prescription as well as antibiotic-related AEs resulting from delayed prescribing plus the POCT. FebriDx may avoid antibiotic prescription in 38–56% of ARI cases associated with a CRP ≥ 20 mg/L^29,52.

The cumulative annual costs for CRP, FebriDx, and TPP were £29,02,95,065, £23,76,14,489, and £14,78,05,881, respectively, while status quo prescribing practices amounted to £32,65,00,346 in annual costs (Table 4). Use of the theoretical TPP resulted in the largest potential cost reduction when compared to status quo prescribing with an annual cost saving of £17,86,94,465 (54% reduction) followed by FebriDx with an annual cost saving of £8,88,85,857 (27% reduction). CRP POCT also resulted in annual cost savings of £3,62,05,281 (11% reduction) (Table 5). The sensitivity analysis also confirmed antibiotic prescribing without a POCT to be the highest cost prescription strategy followed by prescription guided by CRP testing. FebriDx and TPP were consistently found to be the lowest cost prescription strategy.

Table 4. Estimated total costs for ARI Consultation (2018) with and without POCT.

Total annual cost to health system without POCT (AB + AE)	Total annual cost to health system with CRP (POCT + AB + AE)	Total annual cost to health system with TPP (POCT + AB + AE)	Total annual cost to health system with FebriDx (POCT + AB + AE)
£32,65,00,346	£29,02,95,065	£14,78,05,881	£23,76,14,489

Table 5. Estimated annual cost for ARI consultation (2018) savings with POCT.

Without POCT-CRP	Without POCT-TPP	Without POCT-FebriDx
£3,62,05,281	£17,86,94,465	£8,88,85,857

The sensitivity analysis showed that variation in the total ARI population assumptions, POCT cost and antibiotic-related AEs had the most impact on total cost to the healthcare system whereas antibiotic costs influenced total cost to a lesser extent (Table 6).

Table 6. Sensitivity analysis.

	Variable impacted			Impact on total cost
	CRP	FebriDx	TPP	CRP	FebriDx	TPP
Cost of the POCT
Base cost	£13.50	£11.25	£5.00	£29,02,95,065	£23,76,14,489	£14,78,05,881
Test cost + 20%	£16.20	£13.50	£6.00	£33,17,08,745	£27,21,25,889	£16,31,44,281
Test cost – 20%	£10.80	£9.00	£4.00	£24,88,81,385	£20,31,03,089	£13,24,67,481
Target population (ARI) tested
Base target population	15,338,400	15,338,400	15,338,400	£29,02,95,065	£23,76,14,489	£14,78,05,881
Target population + 25%	19,173,000	19,173,000	19,173,000	£34,20,62,165	£28,07,53,739	£16,69,78,881
Target population – 25%	11,503,800	11,503,800	11,503,800	£23,85,27,965	£19,48,70,673	£12,86,32,881
Antibiotic related AEs
Base antibiotic related AEs	116,449	89,576	98,534	£29,02,95,065	£23,76,14,489	£14,78,05,881
Antibiotic related AEs + 25%	145,561	111,970	123,167	£30,98,32,298	£25,26,43,130	£16,43,37,386
Antibiotic related AEs − 25%	86,440	67,182	73,900	£26,17,84,868	£22,25,85,848	£13,12,74,376
Average antibiotic prescriptions/cost
Base antibiotic prescriptions	1,993,992	1,533,840	1,687,224
Antibiotic prescriptions + 25%	2,492,490	1,917,300	2,109,030
Antibiotic prescriptions − 25%	1,495,494	1,150,380	1,265,418
Base antibiotic costs	£50,77,732	£49,42,925	£49,87,860	£29,02,95,065	£23,76,14,489	£14,78,05,881
Antibiotic costs + 25%	£55,76,230	£53,26,385	£54,09,666	£29,07,93,563	£23,79,97,949	£14,82,27,687
Antibiotic costs – 25%	£45,79,234	£45,59,465	£45,66,054	£28,97,96,567	£23,72,31,029	£14,73,84,075

Antibiotic-related AE costs per episode were determined to be £1,345 for anaphylaxis⁵³, £36,718 for SJS⁵⁴, £235 for allergies and diarrhea/nausea (inclusive of ED, out of hours, and GP costs)⁵³, and £10,000 for CDI^17,55.

4. Discussion

AMR is a global threat that requires a multifaceted approach to combat spread of resistant pathogens⁶. Patient concern for their symptoms and diagnostic uncertainty are the most common reasons that lead to an antibiotic prescription for ARI despite lack of evidence of clear benefit²⁴. POCTs capable of triaging patients that may not benefit from antibiotics have been proposed as an intervention to reduce over prescription in uncomplicated ARIs that are caused by a non-bacterial or self-limiting bacterial etiology⁶. This cost analysis confirmed that POCT can have a significant impact on healthcare savings. The primary drivers of the cost savings, and hence the sensitivity of the analysis, include direct antibiotic costs, GP re-consultations, and AEs inclusive of rashes, gastrointestinal complaints, anaphylaxis, SJS, and secondary infections such as CDI.

POCTs with high accuracy (sensitivity/specificity) and negative predictive value (NPV) can help drive antibiotic stewardship in the outpatient GP setting by helping to guide antibiotic therapeutic decisions, such as watchful waiting strategies, where the vast majority of unnecessary antibiotics are prescribed. Without POCT, the UK health system spends over £326 million annually and with POCT this can lead to over £36 million to over £178 million in annual savings depending on the type of POCT deployed.

Two review studies, including a Cochrane Review, have noted that delayed prescribing, in place of immediate antibiotic prescription may be a safe approach for uncomplicated ARI presentation to significantly reduce unnecessary antibiotics and antibiotic resistance while preserving patient safety and satisfaction^27,28. This cost analysis showed that delayed prescription alone resulted in over £68 million cumulative cost savings and an 75% reduction in antibiotic-related AEs when compared to immediate prescription while delayed prescription plus POCT resulted in approximately £36–£178 million in cumulative cost savings.

When using the UK as an example, the findings suggest that the use of diagnostics to inform antibiotic prescription is associated with significant overall cost savings, but also the avoidance of antibiotic-related AEs which offer individual patient benefit. Thus, not only can point-of-care diagnostic testing help to prevent the misuse of antibiotics that leads to AMR, which in turn benefits public and global health, but is also beneficial to reduce healthcare spending and AEs-associated morbidity carried by the individual. Qualitative research exploring GPs attitudes toward POCTs found enthusiasm for a POCT finger-prick blood test that could distinguish viral from bacterial infection⁵⁶. GPs emphasized that such a test would be most valuable in supporting the decision not to prescribe antibiotics in situations where patients are not likely to benefit⁵⁶.

Although the vast majority of antibiotics are prescribed in the outpatient setting while antibiotic stewardship is largely limited to the inpatient setting. A major benefit of POCT is the ability to improve antibiotic stewardship in the primary and urgent care setting by providing tangible results that both increase diagnostic certainty and confidence to delay or withhold antibiotics when bacterial infection is unlikely. Despite obvious benefits of POCTs, only a small number of POCTs, such as glucose testing and D-dimer testing, have broad utilization in the UK The National Health Service (NHS) model for primary care, which currently does not include POCT items of service payments, results in additional work and costs for initial purchase and subsequent maintenance and consumables²⁴. This lack of reimbursement is the major limiting factor to accessing POCTs in the outpatient setting; however, other considerations include space to accommodate a range of technologies, time to train staff, regulatory requirements and uncertainty about test accuracy^{22, 57–59}. However, these costs may be partially or completely offset by enhanced workflow and payment strategies. An accurate, inexpensive test that facilitates high throughput to support patient flow in the clinic could be used as a triage tool to guide treatment of uncomplicated patients with respiratory symptoms. This scenario would avoid cumulative costs and morbidity associated with AEs that are significantly higher in the scenario without POCT. Additionally, Johnson et al. suggest that increased utilization of POCT could be operationalized by the introduction of an opt-in program that provides a separate fee for the delivery of additional “locally enhanced” services (LES), such as POCT, in management of ARI. An LES reimbursement plan may provide the needed financial incentive to uptake POCT for the management of ARI in an NHS primary care setting²⁴.

Re-consultations may indicate persistent symptoms or that the main concerns were not adequately addressed during the first consultation and often increase the pressure to prescribe. POCTs offer a strategy to address patient concerns during the initial visit and therefore may reduce unnecessary re-consultations. Moreover, antibiotic misuse puts individual patients at risk for preventable side effects, and exposes both patients and society to pathogens that develop resistance to antibiotics¹⁸.

Significant changes in facilities and infrastructure are not needed to incorporate POCTs in the outpatient setting, particularly for the simple lateral flow technologies. POCTs in GP clinics could reduce unnecessary antibiotic use and provide an objective measure to reassure patients and providers that antibiotics are not needed. CRP is produced in response to inflammation and high (>20 mg/L) levels can be indicative of a bacterial infection or potentially other causes of inflammation⁶⁰, but CRP has low specificity to differentiate a viral from bacterial infection at lower concentrations²⁷. Despite these limitations, standalone CRP testing has been accepted by patients and has also been shown to reduce antibiotic prescribing in primary care²⁷. The effectiveness of CRP testing to reduce antibiotic prescribing has been demonstrated in multiple settings, including primary care, ranging from 36 to 43% reduction in the UK⁶¹, 25% decrease in a meta-analysis from European countries and the United States⁶², and a 10–20% decrease in South East Asia⁶³.

The UK data presented here serve as an example for other industrialized health systems, particularly those that are moving toward universal healthcare coverage where health spending is covered by the government rather than the individual. In addition to cost savings and implications for public health, the avoided ∼5.7–6 million antibiotics prescriptions improved the outcome and immediate health of ∼1 million patients a year, assuming each AEs episode is related to an individual patient. Outside of the UK, similar antibiotic pressures are felt in lower- and middle-income countries and access to inexpensive testing may have an even greater impact on resource limited regions⁶⁴. While host immune biomarker responses have high NPVs for bacterial infection and may directly guide therapeutic management in settings without malaria, these tests would likely need to be performed subsequent to a rapid diagnostic malaria negative test confirmation prior recommending antibiotic withholding. Even the cost of performing two rapid tests pales in comparison to the threat of increasing rates of antibiotic resistant infections⁶⁵.

This study was designed as a simple benefit analysis focusing on the direct costs associated with antibiotic overuse in order to make it more easily comparable between higher income countries.

This study has a number of limitations as it significantly simplified both the antibiotic prescribing environment and the associated costs. Primary limitations that may lead to an under-estimation of the testing benefits include lack of long-term cost implications of antibiotic overuse on global health, higher costs associated with novel therapies required to treat resistant infections, time associated with phone consultations, inpatient or hospital emergency diagnostic testing, emergency department costs, radiographic or laboratory tests, and inclusion of only direct costs. The costs for AMR are estimated to range from £3 to £45,000 ($5 to $55,000) per patient treated⁶⁶ and annual costs ranging from $2.2 billion^7,67 to upwards of $55 billion inclusive of direct healthcare service costs ($35 billion) and indirect loss of productivity ($20 billion)^66,68. Considering that 81% of antibiotics are prescribed for ARI inappropriately, POCT-guided antibiotic prescribing has the potential to significantly reduce both the incidence and associated cost of AMR in in the outpatient setting.

Additionally, prescribing practices and hospitalization costs change depending on age groups³⁶ with higher prescriptions in children and elderly populations. Age stratification was not performed due to a lack of age stratified data for all the different symptom groups and follow up studies to address this limitation is needed. Further, the cost analysis did not account for potentially reduced test usage in higher priced diagnostics. Bearing the limitations in mind, the potential savings highlighted by this work makes a fiscal as well as patient-centered case for the value of diagnostics to inform antibiotic prescribing.

Conclusions

The majority of antibiotics are prescribed to ARI patients in the outpatient setting. Use of POCTs may reduce antibiotic misuse and reduce the spread of AMR while preserving healthcare funding and ensuring the best individual patient outcome.

Transparency

Declaration of funding

Support for FIND’s work on improved fever triage diagnostic is provided by the Government of the Netherlands, UK aid from the UK government, and the Australian Government.

Declaration of financial/other relationships

The authors declare no competing interest. JME peer reviewers on this manuscript have received an honorarium from JME for their review work, but have no other relevant financial relationships to disclose.

Acknowledgements

The authors thank Dr. Rob Sambursky (Lumos Diagnostics) for support with data collection and critical discussions of the manuscript and data. Support for FIND’s work on improved fever triage diagnostic is provided by the Government of the Netherlands, UK aid from the UK government, and the Australian Government.