Cost-effectiveness of point-of-care diagnostics for AMR: a systematic review

Abstract

Background

Antimicrobial resistance (AMR) is a major threat to global health. By 2050, it is forecast that AMR will cause 10 million deaths and cost 100 trillion USD annually. Point-of-care tests (POCTs) may represent a cost-effective approach to reduce AMR.

Objectives

We systematically reviewed which POCTs addressing AMR have undergone economic evaluation in primary and secondary healthcare globally, how these POCTs have been economically evaluated, and which are cost-effective in reducing antimicrobial prescribing or the burden of AMR. Clinical cost-effectiveness was additionally addressed.

Methods

This systematic review, accordant with PRISMA guidelines, was pre-registered on PROSPERO (CRD42022315192). MEDLINE, PubMed, Embase, Cochrane Library, and Google Scholar were searched from 2000 to 2023 for relevant publications. Quality assessment was performed using the Consensus of Health Economic Criteria.

Results

The search strategy identified 1421 studies, of which 20 met the inclusion criteria. The most common POCTs assessed were for respiratory infections (n = 10), STIs (n = 3), and febrile patients in low- and middle-income countries (n = 3). All studies assessed costs from a healthcare provider perspective; five additionally considered the societal cost of AMR.

Eighteen studies identified POCT strategies that reduced antimicrobial prescribing. Of these, 10 identified POCTs that would be considered cost-effective at a willingness-to-pay (WTP) threshold of £33.80 per antibiotic prescription avoided. Most POCT strategies improved clinical outcomes (n = 14); the remainder were clinically neutral.

Conclusions

There is evidence that some POCTs are cost-effective in reducing antimicrobial prescribing, with potential concomitant clinical benefits. Such interventions—especially CRP POCTs in both high- and low-income settings—merit further, large-scale clinical evaluation.

Introduction

Antimicrobial resistance (AMR) is an urgent global health issue, associated with almost 5 million deaths in 2019,¹ disproportionately affecting low- and middle-income countries (LMICs).² The WHO lists AMR amongst the top 10 threats to global health and forecasts that by 2050 it will cause 10 million deaths and cost 100 trillion USD annually.^3,4 Bacterial AMR is driven in part by overuse of antibiotics,^5,6 due to a combination of diagnostic uncertainty and low risk tolerance.^7,8 Accurate diagnostic testing therefore plays a pivotal role in guiding appropriate antibiotic stewardship. However, the current gold standard is microbial culture, which is time-consuming and often unfeasible in primary care and resource-limited settings.^9,10

Point-of-care tests (POCTs) are bedside tests that provide rapid diagnostic information.¹¹ They represent quick, cheap, and effective potential diagnostic methods,^12–15 and have been proven efficacious in reducing prescribing in real-world scenarios.^16,17 Some POCTs have been demonstrated to be cost-effective in improving clinical outcomes, though the evidence is somewhat mixed.¹⁸ However, most previous economic analyses have not included the benefits of reducing inappropriate prescribing and AMR, and thus these evaluations may underestimate the advantages of POCTs.^19,20

Several reviews have sought to appraise the cost-effectiveness of interventions designed to reduce the burden of AMR. However, most address a wide range of educational, behavioural, policy, and testing strategies, and evaluate relatively few POCTs.^21–26 Those that have addressed POCTs directly have suggested POCTs likely represent cost-effective interventions both clinically and in reducing AMR, but these reviews are limited to specific scenarios, such as primary care or high- and middle-income countries.^26,27

Objectives

This systematic review sought to evaluate POCTs in both LMICs and high-income countries (HICs) to inform global policy and funding decisions.

Its three key questions were: (i) which POCTs addressing AMR have undergone economic evaluation in primary and secondary healthcare globally? (ii) how have these POCTs been economically evaluated? and (iii) which POCTs are cost-effective in reducing antimicrobial prescribing or the burden of AMR, and how does this relate to cost-effectiveness in improving clinical outcomes?

Methods

This study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.²⁸ A protocol for this systematic review was preregistered with PROSPERO (CRD42022315192).

Inclusion criteria

Articles were included if they: (i) were either trial- or model-based economic evaluations; (ii) analysed POCTs primarily aimed at reducing AMR (including rates of inappropriate antimicrobial prescription); (iii) were published from the year 2000 onwards, as articles published before this date are unlikely to reflect modern costs and clinical use of POCTs; and (iv) included a cost–benefit metric, such as cost per QALY, cost per disability-adjusted life year (DALY), incremental cost-effectiveness ratio (ICER), incremental cost-benefit ratio, net monetary benefit (NMB), or costs avoided.

Exclusion criteria

Articles were excluded if they: (i) did not specifically include a diagnostic POCT as their intervention (for instance, diagnostic algorithms or communication tools); (ii) used a point-of-care method, but it was not used as a POCT (such as C-reactive protein (CRP) testing on blood chemistry rather than as a POCT); (iii) did not aim to evaluate a metric related to AMR (including antibiotic prescription) in their study outcomes; (iv) did not include an economic outcome; (v) were systematic reviews, meta-analyses, editorials or commentaries; or (vi) were not available in English.

Data sources and search strategy

A systematic search of the literature was conducted in line with PRISMA guidelines. MEDLINE (Ovid), PubMed, Embase (embase.com), Cochrane Library, Tufts Cost-effectiveness Analysis Registry (CEA), Global Health CEA Registry, and Centre for Reviews and Dissemination’s National Health Service Economic Evaluation Database were searched for results from January 2000 to November 2022. Database search criteria consisted of concepts relating to AMR, POCTs and cost-effectiveness and analytic modelling, and were adapted for each database (see Table S1, available as Supplementary data at JAC Online). The search was re-run in August 2023, giving a total search period of January 2000 to August 2023. A Google Scholar search and both backward and forward referencing of included papers were also performed.

Article selection

After initial removal of duplicated titles and abstracts through manual inspection, a preliminary screening of articles was performed using the Systematic Review Accelerator tool by one reviewer (A.B.) to remove ineligible study types (systematic reviews, meta-analyses, editorials or commentaries) and further duplicates.²⁹ The remaining abstracts were independently screened by two reviewers (A.T. and R.M.). Full-text review was independently conducted by two reviewers (A.T. and R.M.). Discrepancies at both stages were resolved through discussions and the opinion of a third reviewer (A.B.).

Data extraction

Data extraction was performed independently by two reviewers (J.H.D. and R.M.) using a standardized data extraction template. Information regarding study context (country, setting, population, disease of interest), study design (study type, intervention, comparator, time horizon, methodology, included costs) and outcomes (clinical outcome measures, prescribing outcomes, incremental and total costs, ICERs, cost-effectiveness thresholds) were collected. Discrepancies were resolved through discussion with a third reviewer (A.T.). In studies assessing multiple strategies, only those involving a POCT that affected antibiotic prescribing are included. Only the three most cost-effective strategies are included where many strategies were assessed.

Quality assessment and critical appraisal

The quality assessment of both trial-based and modelling studies was conducted using the Consensus on Health Economic Criteria (CHEC) list.³⁰ The CHEC list contains 19 items addressing various economic evaluation domains. Study quality was graded as ‘excellent’ if 100% of the items were fulfilled, ‘good’ above 75%, ‘moderate’ between 50% and 75%, and ‘low’ below 50%.²⁷

Data analysis

Where ICERs were not explicitly reported, they were calculated where possible from available data to provide consistent outcome measures. Depending on data provided, ICER relative to usual care was calculated as (difference in total costs)/(difference in outcome of interest) or (adjusted incremental cost of POCT)/(percentage point reduction in outcome of interest). For ease of interpretation, all costs were converted to GBP (£) in line with exchange rates on 26 June 2023 (1 USD: £0.79; 1 EURO: £0.86). Inflation was not considered.

Given heterogeneity in study types and outcomes, meta-analysis was not conducted. A qualitative synthesis of included papers was undertaken.

Results

1,421 studies were identified through the databases searched (1,345 from January 2000 to November 2022, with 76 additional papers from November 2022 to August 2023). Following exclusion of duplicates (n = 554) and ineligible study types (n = 231), 636 abstracts were screened for eligibility. The full texts of 47 studies identified by the original search and 21 additional studies identified through citation searches and relevant protocol registrations were assessed against the inclusion and exclusion criteria. A total of 20 studies were included in this review (Figure 1).

Figure 1.

PRISMA 2020 flow diagram for new systematic reviews, which included searches of databases, registers and other sources.²⁸ This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Study characteristics

The characteristics of the studies that met the inclusion criteria are summarized in Table 1. Overall, 35 POCT strategies across 20 studies were included in this systematic review, as 11 studies assessed multiple strategies.

Table 1.

Characteristics of included studies

Study designation	POCT(s) assessed	Disease of interest	Population	Study design	Country	Sample size	Study setting
Fawsitt et al. (2022)31	CRP (threshold 20 mg/L)	Bacterial RTI	All adults presenting with acute RTI to primary care in Ireland	Modelling study	Ireland	1 334 677	Primary care
	CRP (threshold 20 mg/L) + communication training
Holmes et al. (2018)32	CRP (pragmatic testing, threshold 20 mg/L)	Bacterial RTI	Adults with symptoms of acute RTI for >12 h	Modelling study	Wales	70	Primary care
	CRP (testing according to guidelines, threshold 100 mg/L)		Adults with symptoms of lower RTI for >12 h			20
Hunter (2015)33	CRP (GP) (threshold not stated)	Bacterial RTI	Patients presenting to primary care with acute RTI	Modelling study	England	100	Primary care
	CRP (practice nurse) (threshold not stated)
	CRP + communication training (threshold not stated)
Boere et al. (2022)34	CRP (threshold not stated)	Bacterial LRTI	Residents of 11 nursing homes with suspected LRTIs	RCT-based economic evaluation	The Netherlands	241 (n = 162 in intervention group)	Nursing homes
Francis et al. (2020)35	CRP (antibiotic prescription at clinician’s discretion)	Acute bacterial exacerbation of COPD	Adults (age ≥40) with acute exacerbations of COPD	RCT-based economic evaluation	Wales and England	649 (n = 325 in intervention group)	Primary care
Lubell et al. (2018)36	CRP [threshold of 10 mg (≤5 years old), 20 mg (>5)]	Bacterial RTI	Adult and paediatric patients with non-severe acute RTI	RCT-based economic evaluation	Vietnam	2036	Primary care
Oppong et al. (2013)37	CRP (antibiotic prescription at clinician’s discretion)	Bacterial LRTI	Adults with acute cough or presentation suggesting LRTI	Observational study-based economic evaluation	Norway and Sweden	370 (281 received POCT CRP test)	Primary care
Oppong et al. (2018)38	CRP testing only (antibiotic prescription at clinician’s discretion)	Bacterial RTI	Adults with suspected RTI	Observational study-based economic evaluation	Belgium, The Netherlands, Poland, Spain, and the UK (England and Wales)	4264 participants (n = 660 CRP only)	Primary care
	CRP testing + communication skills training					4264 participants (n = 709 CRP plus communication)
Cals et al. (2011)13	CRP (antibiotic prescription at clinician’s discretion)	Bacterial LRTI	Adults with suspected LRTI	RCT-based economic evaluation	The Netherlands	431 (n = 110 in intervention group)	Primary care
	CRP + communication skills training					431 (n = 117 in intervention group)
Michaelidis et al. (2014)39	Procalcitonin (threshold of 0.25 ng/mL, though patients under the threshold could still be given antibiotics at the clinician’s discretion)	Bacterial RTI	Adults with acute RTIs judged by their physicians to require antibiotics	RCT-based economic evaluation	Germany, Switzerland	1710 (n = 781 in intervention group)b	Primary care
	Procalcitonin (threshold of 0.25 ng/mL, though patients under the threshold could still be given antibiotics at the clinician’s discretion)		All adults with acute RTIs
Little et al. (2014)40	IMI TestPack Plus Strep A (RADT)	Group A β-haemolytic Streptococcus	Patients aged ≥3 years with acute sore throat with a FeverPAIN score ≥3	RCT-based economic evaluation	UK	420a (n = 213 in intervention group)	Primary care
Harding-Esch et al. (2020)41	AMR POCT (azithromycin)	Antibiotic-resistant gonorrhoea	All confirmed cases of Neisseria gonorrhoeae in England, 2015	Modelling study	England	38 870	GUM clinic
	AMR POCT (ciprofloxacin)
Turner et al. (2017)42	AMR POCT (ciprofloxacin)	Antibiotic-resistant gonorrhoea	All adults who attended GUM clinics in England, 2014	Modelling study	England	1 400 000	GUM clinic
	AMR POCT (penicillin)
Turner et al. (2014)43	NAAT POCT test for chlamydia and gonorrhoea infections	Chlamydia and gonorrhoea infections	All adults who attended GUM clinics in England, 2011	Modelling study	England	1 258 706	GUM clinic
Dickinson et al. (2021)44	CRP (threshold 10 mg/L)	Bacterial infection	Febrile malaria-negative patients	Modelling study	Afghanistan	4391	Primary care
Zhang et al. (2020)45	PCR (leptospirosis)	Bacterial infection	Febrile malaria-negative patients	Modelling study	Thailand	NA	Hospital (emergency department)
	PCR (dengue)
	Multiplex PCR (dengue, leptospirosis and typhus)
Lubell et al. (2016)46	CRP (threshold 20 mg/L)	Bacterial infection	Febrile malaria-negative outpatients	Modelling study	Laos	1083	Outpatient clinic
	Dengue rapid tests (IgM/IgG + NS1 rapid test)
	Scrub typhus rapid test (IgM)
Manore et al. (2019)47	PCR (80% attend follow-up)	Antibiotic-resistant invasive NTS	Adults with gastroenteritis	Modelling study	Kenya	1 000 000 (n = 37 888 high-risk patients)	Outpatient clinic
	Bacterial culture (80% attend follow-up)
	Serology
Daniels et al. (2022)48	GeneXpert rapid test for GBS (rectal and vaginal swabs)	GBS	Women with risk factors for vertical GBS transmission in established labour	RCT-based economic evaluation	England	1628 (n = 722 in intervention group)	Hospital (maternity units)
Butler et al. (2018)49	Flexicult™ SSI-Urinary Kit [culture and antibiotic susceptibility testing (results in 18–24 h)]	Bacterial UTI	Adult women presenting with symptoms of uncomplicated UTI	RCT-based economic evaluation	England, Wales, The Netherlands, Spain	643 (n = 324 in intervention group)	Primary care

GBS, Group B Streptococcus; Ig, immunoglobulin; LRTI, lower RTI; NTS, non-typhoidal Salmonella; RADT, rapid antigen detection tests; RTI, respiratory tract infections; UTI, urinary tract infections.

^aDelayed group (control) + rapid test group.

^bData collated from two cohorts [Briel et al., 2008 (Germany)⁵⁰ and Burkhardt et al., 2010 (Switzerland)⁵¹].

Ten studies were model-based evaluations, eight were economic evaluations based on randomized controlled trial (RCT) data, and two were economic evaluations based on cross-sectional observational studies. Most assessed European populations (n = 15), with the remainder assessing interventions in Asia (n = 4) and Africa (n = 1). The majority were primary care-based (n = 15), including GUM clinics (n = 3), with the remainder in secondary care (n = 4) and nursing homes (n = 1).

Ten studies assessed POCTs for patients presenting to primary care with respiratory tract infections (RTIs), of which nine assessed CRP POCTs and one assessed procalcitonin POCTs. One study assessed rapid antigen detection tests (RADTs) for acute sore throat. Three studies evaluated nucleic acid amplification test (NAAT) POCTs to guide prescribing for sexually transmitted infections (STIs). Three studies assessed multiple POCTs—including CRP, PCR and serology—to distinguish viral from bacterial infections in febrile patients in Asian populations. The final three studies assessed intrapartum POCTs for Group B Streptococcus (GBS), detection and susceptibilities of bacterial urinary tract infections (UTIs), and POCTs for non-typhoidal Salmonella (NTS).

Model parameters

The model parameters of included studies are presented in Table 2. The unit cost of the POCTs varied considerably. The estimated cost of CRP POCTs ranged between £0.73 and £11.31 per test.^31,36 The GBS rapid test was the most expensive POCT assessed, at a unit cost of £94.69.⁴⁸ Calculation of adjusted costs was heterogeneous. Most evaluations included staff time taken to administer the test (n = 14) and the cost of prescribing antibiotics (n = 16). Twelve studies estimated the impact of the testing strategy on utilization of healthcare resources and included these costs.

Table 2.

Model parameters

Study designation	POCT(s) assessed	Standard care cost (per patient)	Unit cost of POCT	Adjusted incremental cost of POCT	Included costs for adjusted incremental POCT	Time horizon
Fawsitt et al. (2022)31	CRP (threshold 20 mg/L)	£218.86a	£3.00a	£26.61a	POCT unit cost, equipment, GP training, staff time, external quality assurance scheme	5 years
	CRP (threshold 20 mg/L) + communication training		£3.00a	£22.31a
Holmes et al. (2018)32	CRP (pragmatic testing, threshold 20 mg/L)	NA	£9.85	£11.92	POCT unit cost, staff time, healthcare costs (reconsultation cost, adverse drug reactions), quality assurance, antibiotic prescription	28 days
	CRP (testing according to guidelines, threshold 100 mg/L)			£10.05
Hunter (2015)33	CRP (GP) (threshold not stated)	£18.08	£4.19	£0.04 less costly than usual care	POCT unit cost, (test material, device depreciation, GP training, staff time), antibiotics, hospital admissions	3 years
	CRP (practice nurse) (threshold not stated)		£4.19	£0.68 less costly than usual care
	CRP + communication training (threshold not stated)		£4.19	£0.53
Boere et al. (2022)34	CRP (threshold not stated)	NA	£8.58a	£27.46a	POCT unit cost (test materials, transport costs, device depreciation), healthcare resource use, antibiotic costs	3 weeks
Francis et al. (2020)35	CRP (antibiotic prescription at clinician’s discretion)	£116.68	£11.31	£45.09	POCT unit cost (materials, consumables, staff time, training), medication, healthcare resource use, cost of work lost as a result of acute exacerbations of COPD	6 months
Lubell et al. (2018)36	CRP (threshold of 10 mg (≤5 years old), 20 mg (>5))	£1.03a	£0.73a	£0.73a	POCT unit cost (test materials, transport, device, training), staff time, routine diagnostic tests, and medications	14 days
Oppong et al. (2013)37	CRP (antibiotic prescription at clinician’s discretion)	NA	£5.32a (Sweden), £3.52a (Norway)	£9.67a	POCT unit cost, antibiotics, healthcare resources (investigations, hospitalizations, GP and nurse time)	14 days
Oppong et al. (2018)38	CRP (antibiotic prescription at clinician’s discretion)	£24.23a	Indexed by country, varies between £3.25 and £5.21a	£17.76a	POCT unit cost, antibiotics, over-the-counter medications, healthcare resources (investigations, hospitalizations, GP, out-of-hours GP visits and nurse time)	28 days
	CRP + communication skills training			£27.44a
Cals et al. (2011)13	CRP (antibiotic prescription at clinician’s discretion)	£30.85a	£4.05a	£1.39a	POCT unit cost (device depreciation, test material, staff time), communication skills training, antibiotics, additional diagnostic tests, reconsultations, out-of-hours consultations, days off work	28 days
	CRP + communication skills training		£8.64a	£1.56a
Michaelidis et al. (2014)39	Procalcitonin for patients judged to require antibiotics (threshold of 0.25 ng/mL)	NA	£31.02a	£17.30a	POCT unit cost (materials), staff time, antibiotics	Acute RTI treatment episode
	Procalcitonin for all patients with RTIs (threshold of 0.25 ng/mL)			£26.73a
Little et al. (2014)40	IMI TestPack Plus Strep A (RADT)	£51.29	£3.25	£2.03 less costly than usual care	POCT unit cost, clinician time, antibiotic costs, healthcare resource use (GP visits, treatment of complications, hospital admissions related to infections)	4 weeks
Harding-Esch et al. (2020)41	AMR POCT (azithromycin)	£73.48	£29	£10.69	POCT unit cost, staff time, antibiotics, additional follow-up costs	Initial patient treatment episode
	AMR POCT (ciprofloxacin)		£29	£15.47
Turner et al. (2017)42	AMR POCT (ciprofloxacin)	£140	£25	£25	POCT unit cost only (hypothetical)—does not consider reduction in treatment costs	NA
	AMR POCT (penicillin)		£25	£25
Turner et al. (2014)43	NAAT POCT test for chlamydia and gonorrhoea infections	£79.77	£19.71	£4.27 less costly than usual care	POCT unit cost (test + sample collection kit), laboratory staff time, clinic staff time	28 days
Dickinson et al. (2021)44	CRP (threshold 10 mg/L)	£0.30a	£0.79a	£1.13a	POCT unit cost (including tariffs and shipment), training to administer/interpret tests, outpatient services delivery cost, dispensary costs, patient out-of-pocket expenditure, patient and carer opportunity costs, and cost of antibiotics	28 days
Zhang et al. (2020)45	PCR (leptospirosis)	£108a	£11.01a	£9.21a (bacterial-predominant), £11.54a (viral-predominant)	POCT unit cost, antibiotics, baseline daily healthcare expenditure of untreated disease	45 days
	PCR (dengue)		£11.01a	£11.09a less costly than usual care (bacterial-predominant), £1.64a less costly than usual care (viral-predominant)
	Multiplex PCR (dengue, leptospirosis and typhus)		£39.31a	£37.33a (bacterial-predominant), £45.42a (viral-predominant)
Lubell et al. (2016)46	CRP (threshold 20 mg/L)	£0.13a	£1.18a	£1.18a	POCT unit cost only	NA
	Dengue rapid tests (IgM/IgG + NS1 rapid test)		£1.18a	£1.18a
	Scrub typhus rapid test (IgM)		£1.18a	£1.18a
Manore et al. (2019)47	PCR (80% attend follow-up)	NA	£7.86a	NA	Cost of diagnostics, cost of antibiotics, cost of resistant strain treatment	1000 days
	Bacterial culture (80% attend follow-up)		£3.93a	NA
	Serology		£0.79a	NA
Daniels et al. (2022)48	GeneXpert rapid test for GBS (rectal and vaginal swabs)	£4003	£94.69	£125	POCT unit cost (materials, machine running time, staff time), antibiotics, delivery-associated costs	NA
Butler et al. (2018)49	Flexicult™ SSI-Urinary Kit [culture and antibiotic susceptibility testing (results in 18–24 h)]	NA	NA	£48 (England), £48.05a (The Netherlands), £27.45a (Spain)	POCT unit cost (materials, clinician time), antibiotics, healthcare resource use	3 months

^aFor ease of interpretation, all costs have been converted to GBP (£) in line with exchange rates on 26 June 2023 (1 USD: £0.79; 1 EURO: £0.86). Inflation has not been taken into account.

All studies performed cost-effectiveness analysis from the healthcare provider perspective. Three studies included costs from the patients’ perspective, including out-of-pocket expenditure, days of work lost, and opportunity cost of attending the clinic.^13,35,44 Five studies performed additional cost-effectiveness analysis including the estimated societal cost of AMR.^{32,36,38,39,44} Although Manore et al.⁴⁷ also incorporated the costs of AMR by ‘inflat[ing] the cost of resistant antibiotic treatment’, their methods were not made clear in the text.

Clinical effectiveness

Eighteen studies (32 strategies) reported clinical outcomes, as summarized in Table 3. Nine studies calculated QALYs, two calculated DALYs and the remainder reported heterogeneous clinical outcomes.

Table 3.

Effect on clinical outcomes

Study designation	POCT(s) assessed	Effect on clinical outcomes compared with standard care	ICER (for relevant clinical outcomes)	Cost-effectiveness (does not include costs of AMR)	Total costs
Fawsitt et al. (2022)31	CRP (threshold 20 mg/L).	Clinical outcomes not reported.	Clinical outcomes not reported.	Clinical outcomes not reported.	Likely net cost of £15.6 milliona over 5 years. Budget impact of £20.6 milliona over 5 years.
	CRP (threshold 20 mg/L) + communication training.				May be budget-saving over 5 years, with potential savings of £0.9 milliona. Budget impact of £3.9 milliona over 5 years.
Holmes et al. (2018)32	CRP (pragmatic testing, threshold 20 mg/L).	0.99% increase in QALYs.	£19 705 per QALY gained.	At a threshold of £20 000 per QALY, the probabilities of CRP testing being cost-effective was 0.49.	More costly than usual care. Scenario analysis incorporating the estimated cost of AMR reduced the ICER to £19 525 (European)/£13 854 (U.S.) per QALY. If global costs of AMR are included, the intervention dominates usual care.
	CRP (testing according to guidelines, threshold 100 mg/L).	3.78% increase in QALYs.	£4390 per QALY gained.	At a threshold of £20 000 per QALY, the probabilities of CRP testing being cost-effective was 0.84.	More costly than usual care. Scenario analysis incorporating the estimated cost of AMR reduced the ICER to £4321 (European)/£2140 (U.S.) per QALY. If global costs of AMR are included, the intervention dominates usual care.
Hunter (2015)33	CRP (GP) (threshold not stated)	0.05% increase in QALYs.	Dominates.	More cost-effective than usual care. 0.05% increase in NMB (£20 000 per QALY). Dominant compared with current practice in 50% of simulations.	Less costly than usual care.
	CRP (practice nurse) (threshold not stated).	0.05% increase in QALYs	Dominates.	More cost-effective than usual care. 0.06% increase in NMB (£20 000 per QALY). Dominant (compared with current practice in 65% of simulations.	Less costly than usual care.
	CRP + communication training (threshold not stated).	0.02% decrease in QALYs.	There was no clinical benefit to this intervention.	Less cost-effective than usual care. 0.02% decrease in NMB (£20 000 per QALY). Dominant compared with current practice in 19% of simulations.	More costly than usual care.
Boere et al. (2022)34	CRP (threshold not stated).	5% reduction in full recovery rate (not significant).	There was no clinical benefit to this intervention.	There was no clinical benefit to this intervention.	Point-of-care CRP testing is associated with a statistically non-significant minor net profit for the nursing homes when antibiotic costs are taken into account. Net benefits = 0.88, benefit–cost ratio = 1.09, return on investment = 8.54 per patient.
Francis et al. (2020)35	CRP (antibiotic prescription at clinician’s discretion).	2.91% increase in QALYs. 7% reduction in Clinical COPD Questionnaire scores (non-significant). Reduced cost of productivity loss (£510.42).	£15 251 per QALY gained at 6 months.	The probability of CRP POCT being cost-effective at a WTP threshold of £20 000 per QALY is 56%.	More costly than usual care. Extrapolated to the UK population, the estimated budget impact over 5 years is £534 million.
Lubell et al. (2018)36	CRP [threshold of 10 mg (≤5 years old), 20 mg (>5)]	No difference in clinical outcomes compared with standard care.	There was no clinical benefit to this intervention.	There was no clinical benefit to this intervention.	More costly than usual care.
Oppong et al. (2013)37	CRP (antibiotic prescription at clinician’s discretion).	0.12% increase in QALYs (not significant). No significant difference in symptom severity or recovery time. Significant reduction in hospitalizations and increase in medical investigations.	£8 057a per QALY gained.	At £25 792a per QALY, the incremental NMB was £21.67a.	Non-significant increase in mean healthcare costs for patients receiving point-of-care CRP.
Oppong et al. (2018)38	CRP testing only (antibiotic prescription at clinician’s discretion).	Non-significant increase in EQ-5D score, 0.02% increase in QALYs.	£177 600a per QALY gained.c	Not cost-effective at threshold of between £20 000 and £30 000 per QALY.	More costly than usual care. However, dominates usual care when costs of AMR are included.
	CRP testing + communication skills training.	Non-significant increase in EQ-5D score, 0.03% decrease in QALYs.	There was no clinical benefit to this intervention, dominated by usual care.	There was no clinical benefit to this intervention.	More costly than usual care, dominated by usual care when costs of AMR are included
Cals et al. (2011)13	CRP (antibiotic prescription at clinician’s discretion).	No significant difference in reconsultation rates, symptom scores, or patient-reported time to recovery.	There was no clinical benefit to this intervention.	There was no clinical benefit to this intervention.	More costly than usual care.
	CRP + communication skills training		There was no clinical benefit to this intervention.	There was no clinical benefit to this intervention.	More costly than usual care.
Michaelidis et al. (2014)39	Procalcitonin for patients judged to require antibiotics (threshold of 0.25 ng/mL).	2.5% reduction in QALYs lost.	£93 400a per QALY gained.	No WTP threshold given, but not cost-effective using NICE’s threshold of £20 000 per QALY.	More costly than usual care in the base case and across all one-way sensitivity analyses, except when the POCT costs <£13.36a or the antibiotic cost >£47.95.a
	Procalcitonin for all patients with RTIs (threshold of 0.25 ng/mL).	0.8% reduction in QALYs lost.	£452 151a per QALY gained.	No WTP threshold given, but not cost-effective using NICE’s threshold of £20 000 per QALY.	More costly than usual care in the base case and across all one-way sensitivity analyses, except when the POCT costs <£11.79.a
Little et al. (2014)40	IMI TestPack Plus Strep A (RADT).	11% faster symptom resolution (not significant). 2.3% increase in QALYs.	Dominates.	Cost-effective; improved symptom scores and QALYs at a lower cost than standard care.	Dominates usual care (delayed prescription). However, this intervention is dominated by FeverPAIN scoring alone for symptom score. The RADT has an ICER of £24 528 per QALY gained versus FeverPAIN alone.
Harding-Esch et al. (2020)41	AMR POCT (azithromycin).	58.7% fewer suboptimal treatments, treatment failure rate increased from 0% to 0.09%.	£415 per optimal treatment gained.	No WTP threshold given for optimal treatment gained. No other clinical benefits were demonstrated.	More costly than usual care. Would become cost-saving if AMR POCT cost was ≤£18.
	AMR POCT (ciprofloxacin).	52.4% fewer suboptimal treatments, treatment failure rate increased from 0% to 0.64%.	£672 per optimal treatment gained.	No WTP threshold given for optimal treatment gained. No other clinical benefits were demonstrated.	More costly than usual care. Would become cost-saving if AMR POCT cost was ≤£16.
Turner et al. (2017)42	AMR POCT (ciprofloxacin).	Mean treatment time reduced by 2.2 days. 100% reduction in patients lost to follow-up.	£43 481 per patient lost to follow-up.	No WTP threshold given for patients lost to follow-up.	More costly than usual care, although cost savings from AMR POCT, such as reduced antibiotic prescribing and treatment costs were not taken into account.
	AMR POCT (penicillin).		£43 481 per patient lost to follow-up.	No WTP threshold given for patients lost to follow-up.	More costly than usual care, although cost savings from AMR POCT, such as reduced antibiotic prescribing and treatment costs, were not taken into account.
Turner et al. (2014)43	NAAT POCT test for chlamydia and gonorrhoea infections.	0.026% increase in QALYs. 100% reduction in onward transmissions. 85% reduction in cases of pelvic inflammatory disease.	Dominates.	Cost-effective; increased QALYs at a lower cost than standard care.	POCT NAAT testing is 10% (£11.7 million) less costly than standard care.
Dickinson et al. (2021)44	CRP (threshold 10 mg/L).	20% increase in patients treated correctly.	Costs per additional correct treatment: healthcare perspective = £11.32a, societal perspective = £8.86a, scenario analysis including the economic cost of AMR = £7.58.a	At a WTP of £9.87a per additional correct treatment, the probability of the intervention being cost-effective is 50% from a healthcare perspective, 61% from a societal perspective, and 66% when including the estimated costs of AMR.	For healthcare delivery alone, there is a 0% probability of the intervention being cost-effective. From a societal cost perspective, 6% of simulations estimated POC CRP testing to be less costly compared with usual care. Including the cost of AMR, this rose to 13% of simulations.
Zhang et al. (2020)45	PCR (leptospirosis)	17% increase in DALYs (bacterial-predominant), 28% increase in DALYs (viral-predominant).	£36.09a,b per additional DALY (bacterial-endemic), £48.48a,b per additional DALY (viral-endemic).	See cost-effectiveness outcome (antibiotic prescribing).	More costly than usual care.
	PCR (dengue)	13% reduction in DALYs (bacterial-predominant), 6% reduction in DALYs (viral-predominant).	Reduction in DALYs in both scenarios.	See cost-effectiveness outcome (antibiotic prescribing).	Less costly than usual care.
	Multiplex PCR (dengue, leptospirosis and typhus)	6% increase in DALYs (bacterial-predominant), 18% increase in DALYs (viral-predominant).	£446a,b per additional DALY (bacterial-endemic), £239a,b per additional DALY (viral-endemic).	See cost-effectiveness outcome (antibiotic prescribing).	More costly than usual care.
Lubell et al. (2016)46	CRP (threshold 20 mg/L).	54% increase in correct diagnoses. 73% Decrease in untreated bacterial infections. 21.3% increase in DALYs.	£73.90a per additional DALY.	80% probability of being cost-effective compared with usual care (WTP per DALY averted = Laos GDP per capita (£1110a).	More costly than usual care.
	Dengue rapid tests (IgM/IgG + NS1 rapid test).	12% decrease in correct diagnoses. 4% Increase in untreated bacterial infections. 7.5% decrease in DALYs.	Dominated by usual care.	Not cost-effective; more costly than usual care with little or no advantage in terms of health outcomes.	More costly than usual care.
	Scrub typhus rapid test (IgM).	13% increase in correct diagnoses. 42% Decrease in untreated bacterial infections. 38.8% increase in DALYs.	£37.74a per DALY averted	90% probability of being cost-effective compared with usual care (WTP per DALY averted = Laos GDP per capita (£1 110a).	More costly than usual care.
Manore et al. (2019)47	PCR (80% attend follow-up)	84.6% (test all)/36.7% (test only mildly symptomatic) fewer deaths than usual care. 97.3% (test all)/32.3% (test only mildly symptomatic) fewer improperly treated than usual care.	Dominates (investment of < £79a (test all), £158–237a (test only symptomatic) per life savedd).	Cost-effective—less costly than usual care and reduced number of deaths. Test all more cost-effective than test only symptomatic.	84.2% less costly than usual care (test all strategy). 0.6% more costly than usual care (test only mildly symptomatic patients strategy).
	Bacterial culture (80% attend follow-up)	80.5% (test all)/33.6% (test only mildly symptomatic) fewer deaths than usual care. 92.8% (test all)/28.7% (test only mildly symptomatic) fewer improperly treated than usual care.	Dominates (investment of <£79a (test all), £158–237a (test only symptomatic) per life savedd)	Cost-effective—less costly than usual care and reduced number of deaths. Test all more cost-effective than test only symptomatic.	83% less costly than usual care (test all strategy). 1% less costly than usual care (test only mildly symptomatic patients strategy).
	Serology	95.5% (test all)/45.5% (test only mildly symptomatic) fewer deaths than usual care. 95.5% (test all)/36.3% (test only mildly symptomatic) fewer improperly treated than usual care.	Dominates (investment of <£79a (test all), £158–237a (test only symptomatic) per life savedd).	Testing all patients with antibody testing was the most cost-effective strategy assessed—less costly than usual care and reduced number of deaths.	93% less costly than usual care (test all strategy). 12% less costly than usual care (test only mildly symptomatic patients strategy).
Daniels et al. (2022)48	GeneXpert rapid test for GBS (rectal and vaginal swabs).	Clinical outcomes not reported.	Clinical outcomes not reported.	Clinical outcomes not reported.	More costly than usual care.
Butler et al. (2018)49	Flexicult™ SSI-Urinary Kit [culture and antibiotic susceptibility testing (results in 18–24 h)].	No significant differences in recovery, patient enablement, UTI recurrences, reconsultation, antibiotic resistance, and hospitalizations at follow-up.	Dominated by usual care.	This intervention is only cost-effective in limited cases, and against a high WTP.	This intervention is never cost-saving.

GDP, gross domestic product.

^aFor ease of interpretation, all costs have been converted to GBP (£) in line with exchange rates on 26 June 2023 (1 USD: £0.79; 1 EURO: £0.86). Inflation has not been taken into account.

^bICER not explicitly stated; calculated as (difference in total costs)/(total difference in clinical outcome of interest).

^cICER not explicitly stated; calculated as (adjusted incremental cost of POCT)/(percentage point reduction in clinical outcome of interest/100).

^dValues are approximated from Figure 4b, as not explicitly stated in the text.

Of the 32 strategies for which clinical outcomes were reported, 75% (n = 24) reported an improvement in the clinical outcome assessed and 12.5% (n = 4) of strategies resulted in no change in clinical outcomes. Four (12.5%) resulted in worse clinical outcomes, though notably several studies did not conduct significance tests on the effect of the intervention.

Nine studies (15 strategies) assessing POCT for RTI reported clinical outcomes. Six CRP POCT strategies were cost-effective at a willingness-to-pay (WTP) threshold of £20 000 per QALY,^32,33,35,37 of which two dominated usual care.³³ Two procalcitonin testing strategies improved clinical outcomes but were not cost-effective at a WTP threshold of £20 000 per QALY.³⁹ Seven POCT strategies resulted in no significant improvement in clinical outcomes.^13,33,36,38

The three studies (five strategies) assessing POCT in genitourinary medicine (GUM) clinics demonstrated clinical benefit,^41–43 although AMR POCT for gonorrhoea slightly increased treatment failure rates.⁴¹ Switching NAAT testing for STIs from laboratory testing to POCT dominated usual care.⁴³ No WTP threshold was provided for the other two studies.

The three studies (nine strategies) assessing POCTs to determine antibiotic requirement for febrile malaria-negative patients found most strategies were clinically cost-effective.^44–46 CRP testing was cost-effective at a relatively low WTP threshold of £9.87 per patient treated correctly,⁴⁴ or £1110 per DALY.⁴⁶ Scrub typhus rapid testing was clinically cost-effective at the same threshold per DALY.⁴⁶ However, dengue rapid testing led to worse clinical outcomes.^44,45

Regarding clinical outcomes for the remaining three studies, RADTs for acute sore throat and all POCT strategies for NTS patients dominated usual care,^40,47 but urinary culture for suspected UTI had no effect on clinical outcomes and only had a chance of being cost-effective at a high WTP threshold.⁴⁹

Effect on antibiotic prescribing

The impact of POCT on antibiotic prescribing is summarized in Table 4. Sixteen studies reported that all strategies assessed (n = 27) decreased antibiotic prescribing, with a range of 6%–100% decrease relative to standard care. Two of the remaining four studies found differential effects on prescribing depending on the POCT used for febrile malaria-negative patients.^45,46 CRP testing and scrub typhus serology testing strategies increased total prescriptions, but this largely resulted from an increase in correctly treated cases; the other POCTs reduced prescribing.⁴⁶ The final two studies found point-of-care urine culture and susceptibility testing did not change overall antibiotic prescribing,⁴⁹ and GBS POCTs led to a non-significant increase in intrapartum antibiotic prescribing.⁴⁸

Table 4.

Effect on antibiotic prescribing

Study designation	POCT(s) assessed	Standard care	Effect of POCT on total antibiotics prescribed (relative to % prescribed antibiotics under standard care)	ICER (cost per antibiotic prescription avoided, healthcare provider only)	Cost-effectiveness (including costs of AMR where explicitly stated)
Fawsitt et al. (2022)31	CRP (threshold 20 mg/L)	Clinical diagnosis; 56% of patients prescribed antibiotics	27% reduction	£199.97a	Significantly reduced antibiotic prescribing. No WTP threshold provided.
	CRP (threshold 20 mg/L) + communication training		48% reduction	£95.27a	Significantly reduced antibiotic prescribing. No WTP threshold provided.
Holmes et al. (2018)32	CRP (pragmatic testing, threshold 20 mg/L)	Clinical diagnosis: 100% of patients prescribed antibiotics	74% reduction	£16.07	Significantly reduced antibiotic prescribing. No WTP threshold provided.
	CRP (testing according to guidelines, threshold 100 mg/L)		100% reduction	£9.31	Significantly reduced antibiotic prescribing. No WTP threshold provided.
Hunter (2015)33	CRP (GP) (threshold not stated)	Clinical diagnosis: 59% of patients prescribed antibiotics	26% reduction	Dominatesb	Significantly reduced antibiotic prescribing. No WTP threshold provided.
	CRP (practice nurse) (threshold not stated)		9% reduction	Dominatesb	Significantly reduced antibiotic prescribing. No WTP threshold provided.
	CRP + communication training (threshold not stated)		26% reduction	£7.45b	Significantly reduced antibiotic prescribing. No WTP threshold provided.
Boere et al. (2022)34	CRP (threshold not stated)	Clinical diagnosis: 82% of patients prescribed antibiotics	36% reduction	£117.58a	The probability of CRP POCT being cost-effective at a WTP threshold of £0 per antibiotic prescription avoided was 40% and this increased to 80% at a threshold of £557.36a per antibiotic prescription avoided.
Francis et al. (2020)35	CRP (antibiotic prescription at clinician’s discretion)	Clinical diagnosis: 77.4% of patients consumed antibiotics	32% reduction (26% reduction in antibiotics consumed)	£222.00	Significantly reduced antibiotic prescribing. No WTP threshold provided.
Lubell et al. (2018)36	CRP [threshold of 10 mg (≤5 years old), 20 mg (>5)]	Clinical diagnosis: 63% of patients prescribed antibiotics	32% reduction	£3.65a,c	With an estimated societal cost of AMR of £3.22a per course of broad-spectrum β-lactams, this intervention is not cost-effective. However, the test has a positive NMB if adherence exceeds 70% and the cost of the test is halved, or with adherence at 80% and no reduction in unit cost.
Oppong et al. (2013)37	CRP (antibiotic prescription at clinician’s discretion)	Clinical diagnosis ± POCT CRP	10% reduction (not significant)	£96.69a	Did not significantly reduce antibiotic prescribing. No WTP threshold provided.
Oppong et al. (2018)38	CRP testing only (antibiotic prescription at clinician’s discretion)	Clinical diagnosis ± POCT CRP	43.6% reduction	£82.09a	Significantly reduced antibiotic prescribing, but not cost-effective when costs of AMR are not taken into account. WTP threshold at which intervention cost-effective ranged from £47.69a (Poland) to £175.27a (Belgium) per prescription avoided. However, this intervention was less cost-effective than communication skills training alone.
	CRP testing + communication skills training		43% reduction	£108.53a	Significantly reduced antibiotic prescribing, but not cost-effective when costs of AMR are not taken into account. WTP threshold at which intervention cost-effective ranged from £70.48a (Poland) to £201.52a (Belgium) per prescription avoided. This intervention was dominated by communication skills training alone
Cals et al. (2011)13	CRP (antibiotic prescription at clinician’s discretion)	Clinical diagnosis: 68% of patients prescribed antibiotics	43% reduction	£4.97a	CRP POCT is cost-effective at a WTP of <£8.60a per antibiotic prescription avoided. However, CRP POCT was dominated by communication training alone.
	CRP + communication skills training		66% reduction	£3.56a	CRP POCT + communication training is cost-effective at a WTP of <£4.60a per % reduction in antibiotic prescribing. However, the addition of POCT CRP to communication skills training alone gives an ICER of £104.63a per antibiotic prescription avoided.
Michaelidis et al. (2014)39	Procalcitonin for patients judged to require antibiotics (threshold of 0.25 ng/mL)	Clinical diagnosis: 97% of patients prescribed antibiotics	76% reduction	£24.37a	This intervention is cost-effective at an estimated WTP of £33.80a per antibiotic prescription safely avoided (reflecting the estimated cost of AMR). The likelihood of this strategy being preferred to usual care was 58.4% at the same WTP threshold.
	Procalcitonin for all patients with RTIs (threshold of 0.25 ng/mL)	Clinical diagnosis: 37% of patients prescribed antibiotics	62% reduction	£117.15a	This intervention is not cost-effective at an estimated WTP of £33.80a per antibiotic prescription safely avoided (reflecting the estimated cost of AMR). The likelihood of this strategy being preferred to usual care was 2.8% at the same WTP threshold.
Little et al. (2014)40	IMI TestPack Plus Strep A (RADT)	Delayed prescription to be collected after 3–5 days if symptoms not settling, or soon if symptoms significantly deteriorate; 46% of patients use antibiotics	24% reduction	Dominatesb	Cost-effective; reduced antibiotic prescribing at lower cost than standard care. ICER versus FeverPAIN of £255c per prescription avoided.
Harding-Esch et al. (2020)41	AMR POCT (azithromycin)	NAAT + microscopy; 100% of patients prescribed IM ceftriaxone + oral azithromycin	95% reduction in ceftriaxone prescriptions	£11.29 per ceftriaxone prescription avoided	Significantly reduced ceftriaxone prescribing. No WTP threshold provided.
	AMR POCT (ciprofloxacin)		67% reduction in ceftriaxone prescriptions	£22.94 per ceftriaxone prescription avoided	Significantly reduced ceftriaxone prescribing. No WTP threshold provided.
Turner et al. (2017)42	AMR POCT (ciprofloxacin)	NAAT + microscopy; 2.8% of GUM attendees prescribed ceftriaxone + azithromycin for confirmed gonorrhoea	66% reduction in ceftriaxone prescriptions	£1 561b	Significantly reduced ceftriaxone prescribing. No WTP threshold provided.
	AMR POCT (penicillin)		79% reduction in ceftriaxone prescriptions	£1 300b	Significantly reduced ceftriaxone prescribing. No WTP threshold provided.
Turner et al. (2014)43	NAAT POCT test for chlamydia and gonorrhoea infections	Laboratory NAAT testing (7 day turnaround); 17.2% of GUM attendees prescribed antibiotics for suspected or confirmed gonorrhoea/chlamydiad	100% reduction in inappropriate antibiotic prescriptions	Dominates	Cost-effective; reduced antibiotic prescribing at lower cost than standard care.
Dickinson et al. (2021)44	CRP (threshold 10 mg/L)	Clinical diagnosis: 56% of patients prescribed antibiotics	6% reduction; 12% reduction in inappropriate antibiotic prescriptions	£34.79a,b (cost per inappropriate prescription avoided = £17.08a,b)	No WTP threshold provided. Including societal costs gives an ICER of £13.38a,b per inappropriate antibiotic prescription avoided. Including costs of AMR gives an ICER of £11.44a,b.
Zhang et al. (2020)45	PCR (leptospirosis)	Clinical diagnosis; empirical antibiotics for patients with severe disease	87% decrease in probability of overprescribing antibiotics (bacterial-predominant) 87% decrease in probability of overprescribing antibiotics (viral-predominant)	NA	Significantly reduced antibiotic overprescribing. Dominated by strategies below.
	PCR (dengue)		85% increase in probability of overprescribing antibiotics (bacterial-predominant) 20% decrease in probability of overprescribing antibiotics (viral-predominant)	NA	When NMB is calculated as (DALYs averted × WTP of Thailand’s GDP per capita) − incremental costs − penalty for antibiotic overuse, dengue PCR had the highest NMB when penalty for antibiotic use is £10 265a in bacterial-endemic scenario, and between £790a and £10 265 in the viral-endemic scenario. This is due to reduced costs despite increased antibiotic overprescribing.
	Multiplex PCR (dengue, leptospirosis and typhus)		87% decrease in probability of overprescribing antibiotics (bacterial-predominant) 90% decrease in probability of overprescribing antibiotics (viral-predominant)	NA	Significantly reduced antibiotic overprescribing. When NMB is calculated as above, this strategy had the highest NMB when penalty for antibiotic overuse was >£10 265a for both scenarios.
Lubell et al. (2016)46	CRP (threshold 20 mg/L)	Clinical diagnosis: 38% of patients prescribed antibiotics	26% increase (41% decrease in inappropriate antibiotic prescriptions) (adapted from Figure 3)	Dominated for overall prescriptions £12.71a,c (cost per inappropriate antibiotic prescription avoided)	Significantly reduced antibiotic prescribing. No WTP threshold provided.
	Dengue rapid tests (IgM/IgG + NS1 rapid test)		16% reduction (23% reduction in inappropriate antibiotic prescriptions) (adapted from Figure 3)	Not possible to calculate ICER for overall prescriptions. £23.60a,c (cost per inappropriate antibiotic prescription avoided)	Significantly reduced antibiotic prescribing. No WTP threshold provided.
	Scrub typhus rapid test (IgM)		32% increase (9% increase in inappropriate antibiotic prescriptions) (adapted from Figure 3)	Dominated for overall and inappropriate prescriptionsc	Antibiotic prescribing increased.
Manore et al. (2019)47	PCR (80% attend follow-up)	Clinical diagnosis; empirical broad-spectrum antibiotics for patients with severe symptoms (Scenario 4 in the model)	97% reduction (test all), 32% reduction (test only mildly symptomatic) in percentage improperly treated	Dominates	Cost-effective—less costly than usual care and reduced number improperly treated. Test all more cost-effective than test only symptomatic.
	Bacterial culture (80% attend follow-up)		93% reduction (test all), 29% reduction (test only mildly symptomatic) in percentage improperly treated	Dominates	Cost-effective—less costly than usual care and reduced number improperly treated. Test all more cost-effective than test only symptomatic.
	Serology		95% reduction (test all), 36% reduction (test mildly symptomatic) in percentage improperly treated	Dominates	Cost-effective—less costly than usual care and reduced number improperly treated. Test all more cost-effective than test only symptomatic.
Daniels et al. (2022)48	GeneXpert rapid test for GBS (rectal and vaginal swabs)	Treatment based on risk factors; 36% of patients received intrapartum antibiotics	14% increase in intrapartum antibiotics for GBS prophylaxis (non-significant), 29% reduction in antibiotics given to babies	Dominated	Antibiotic prescribing increased. At a WTP of £0 per course of antibiotics, there is a 40% chance of the intervention being cost-effective. Cost-effectiveness decreases as the WTP increases.
Butler et al. (2018)49	Flexicult™ SSI-Urinary Kit [culture and antibiotic susceptibility testing (results in 18–24 h)]	Clinical diagnosis + urine dipstick; 88.4% of patients prescribed antibiotics	No differences in overall antibiotic prescribing, 6.8% reduction in patients receiving antibiotics at initial consultation.	Dominated	No significant difference in overall antibiotic prescribing. This intervention is only cost-effective in limited cases, and against a high WTP.

IM, intramuscular.

^aFor ease of interpretation, all costs have been converted to GBP (£) in line with exchange rates on 26 June 2023 (1 USD: £0.79; 1 EURO: £0.86). Inflation has not been taken into account.

^bICER not explicitly stated; calculated as (difference in total costs)/(difference in total number of antibiotics prescribed).

^cICER not explicitly stated; calculated as (adjusted incremental cost of POCT)/(percentage point reduction in antibiotic prescribing/100).

^dValue not explicitly stated; calculated from (total number of cases of chlamydia + total cases of gonorrhoea + number overtreated under standard care)/total GUM visits

Cost-effectiveness for reducing antibiotic prescribing

Thirteen studies explicitly reported an ICER for antibiotic prescriptions avoided from a healthcare perspective. Sufficient data were provided to calculate ICERs for either antibiotic prescription avoided or inappropriate antibiotic prescription avoided for an additional six studies. In total, ICERs were calculated for 32 point-of-care strategies across 19 studies, as shown in Table 4.

Twenty-eight strategies (across 17 studies) reduced antibiotic prescribing—7 strategies dominated usual care,^33,43,47,52 and the ICER of the remaining 21 ranged from £3.56 to £1561 per antibiotic prescription avoided. Of the final four strategies (across three studies), three were dominated by usual care as they cost more and did not reduce antibiotic prescribing.^46,48,49 One increased overall prescribing but was cost-effective in reducing inappropriate prescriptions.⁴⁶

Only two studies used a set WTP threshold per prescription avoided.^39,52 Lubell et al.⁵² and Michaelidis et al.³⁹ used a WTP threshold of the estimated societal cost of AMR at £3.22 and £33.80 per antibiotic prescription avoided, respectively. At the first threshold, only the seven strategies that dominated usual care would be considered cost-effective. However, at the second, a further 7 strategies would be considered cost-effective,^{13,32,33,36,39} rising to 12 if ICERs for inappropriate prescriptions avoided are included.^41,44,46

Cost-effectiveness by setting

In HICs, 8 of the 16 scenarios evaluating POCT for RTIs were cost-effective at the threshold of £33.80 per prescription avoided. The maximum ICER was £222 per prescription avoided.³⁵ The benefit of the additional of communication skills training to CRP POCT for RTIs was mixed; two studies found that it reduced cost-effectiveness,^33,38 and two studies found it increased cost-effectiveness.^13,31 However, it should be noted that one study found communication skills training alone was more cost-effective than POCT alone.¹³ Similarly, RADTs for acute sore throat dominated usual care, though were less cost-effective than FeverPAIN score alone.⁴⁰ Urine culture and intrapartum GBS POCTs were dominated by usual care.⁴⁹

Regarding STI POCTs in GUM clinics, NAAT POCTs for gonorrhoea and chlamydia dominated laboratory NAAT testing.⁴³ The cost-effectiveness of testing for antibiotic-resistant gonorrhoea depended on the proposed implementation strategy—when AMR testing was targeted to patients with confirmed gonorrhoea, both tests (azithromycin and ciprofloxacin) would be cost-effective at the threshold of £33.80 per inappropriate prescription avoided.⁴¹ However, testing all attendees at a GUM clinic regardless of their presenting symptoms resulted in significantly lower cost efficacy.⁴²

Cost-effectiveness of POCTs in LMICs varied by test and context. For febrile malaria-negative patients, Dickinson and colleagues⁴⁴ found CRP POCT to have a relatively low ICER of £34.79 per prescription avoided, which dropped to £11.44 when the estimated societal costs of AMR were included in adjusted incremental costs. Similarly, Lubell et al.⁴⁶ found that CRP testing and dengue rapid testing were likely to be cost-effective at a WTP threshold per prescription avoided of £33.80, though scrub typhus rapid testing was dominated by usual care. ICERs could not be calculated for Zhang et al.⁴⁵ However, they found a positive NMB for multiplex PCR testing, but only at a high penalty of over £10 000 for antibiotic overuse.⁴⁵ CRP POCT for RTIs in Vietnamese primary care were cost-effective at a low WTP threshold.³⁶ Finally, culture, PCR and serology for antibiotic-resistant NTS all dominated usual care, largely as a result of significant decreases in modelled outbreak sizes.⁴⁷

Quality assessment

No studies attained an ‘excellent’ quality mark using CHEC list criteria. Three studies were graded as ‘good’,^35,44,46 15 were graded as ‘moderate’^{13,31–34,36–43,47,48} and 2 were graded as ‘low’ quality.^45,49 The items that were fulfilled by the fewest studies were time horizons, inclusion of all relevant costs and outcomes, and future discounting.

Discussion

This review systematically evaluated economic evaluations of POCTs designed to reduce antimicrobial prescribing globally and appraise the evidence for their clinical and cost efficacy to inform national and global health policy. There has been limited research on the subject given its importance—we identified only 20 studies despite broad eligibility criteria. The quality of reporting and conduct of studies was variable when assessed against the CHEC criteria. Overall, we found that POCTs are effective interventions at reducing antibiotic prescribing—four studies identified POCTs that dominated usual care, and a further six identified POCTs that would be cost-effective if AMR was accounted for at a relatively low WTP threshold of £33.80 per prescription avoided. The majority had concomitant clinical benefits.

POCT for RTIs in primary care

Distinguishing bacterial from viral RTIs is a common issue in primary care; accurate POCTs therefore have the potential to significantly reduce antimicrobial prescribing.^14,17 We found that CRP POCTs have mixed clinical efficacy depending on the scenario in which they were deployed but may be cost-effective at reducing antibiotic prescribing at low WTP thresholds. This is in keeping with existing literature, which has shown CRP POCTs lead to significant reductions in antibiotic prescribing without clear benefits to clinical outcomes.^14,16,53

The range of cost-efficacy estimates resulted in part from heterogeneity in study design. Only one study evaluated POCT CRP in an LMIC, but found a low ICER in this scenario.³⁶ Higher ICERs were seen in specific populations such as nursing homes or patients with COPD.^34,35 Studies that did not include cost savings from reduced antibiotic prescribing also made higher estimates.³¹ It is also noteworthy that clinical and cost efficacy were generally higher in modelling studies, based on hypothesized parameters, relative to evaluations based on RCT data.^32,33 Real-world factors, such as physician and patient acceptance of POCT, may limit their uptake and efficacy,^37,54,55 as suggested by other systematic reviews concluding that cost efficacy improved when POCTs were used in combination with communication training.²⁷

Although several reports and guidelines—including NICE—currently recommend the use of CRP POCTs in RTI presentations on clinical grounds,^4,56 they are not widely used in primary care.^57,58 This review, including both modelling studies and real-world data, finds moderate evidence that such tests may also be cost-effective interventions at reducing antibiotic use, supporting best practice guidelines recommending their clinical use.

POCT procalcitonin was a cost-effective strategy for reducing antibiotic prescribing at a low WTP threshold, but only with targeted use in cases where bacterial RTI was clinically suspected.³⁹ Use of procalcitonin for RTIs generally has been found to reduce antibiotic exposure and improve survival.⁵⁹ However, mixed sensitivity and specificity represent important caveats for its use as a lone clinical biomarker.⁶⁰ Procalcitonin may represent a promising adjuvant test to CRP but further studies are required to evaluate the clinical and cost efficacy of combined POCT strategies.^61,62

POCTs for STIs

Three modelling studies in HIC settings examined the use of POCTs in the diagnosis and treatment of STIs and suggested them to be cost-effective interventions in clinically suitable scenarios. This is in keeping with evidence showing them to be generally clinically cost-effective interventions, though without evaluating their effects on antimicrobial prescribing.^63,64

POCTs for AMR in gonorrhoea and chlamydia are of particular importance given the rise of multidrug-resistant (MDR) strains.⁶⁵ Previous modelling studies have shown theoretical POCTs for AMR can reduce resistance.^66,67 The economic evaluations identified by this review further suggest POCTs for resistant strains may be cost-effective at a low WTP threshold.^41–43 However, at the time of writing, POCTs for gonorrhoea AMR are not commercially available and development is challenging due to variety in the genetic determinants of resistance.⁶⁵ Molecular tests are currently being developed with promising results, and treatment guidelines now encourage the use of older antibiotics when susceptibility has been confirmed by laboratory testing.^68–70 However, such molecular tests may be prohibitively expensive for LMICs, foregrounding the need for development of affordable POCTs.^69,71,72

POCTs in LMICs

A range of POCTs have been evaluated in LMICs for reducing AMR. Of the five studies identified in this review, only one utilized RCT data, highlighting a particular lack of real-world evidence in these settings.³⁶

Due to the growing use of malaria POCTs, there is a significant cohort of febrile malaria-negative patients in LMICs who are often given antibiotics empirically.^44,46 Using POCTs to inform prescribing in this cohort has been hypothesized as a cost-effective intervention to improve stewardship.⁴⁴ Although POCTs for specific conditions, such as dengue, typhoid and leptospirosis, are available, they are limited by mixed efficacy and the requirement for multiple separate tests. Zhang et al.⁴⁵ found that the use of a combined multiplex PCR most significantly reduced antibiotic overprescribing in this cohort, with additional clinical benefits. However, given the costs of such strategies, they found that empirical antibiotic treatment was the most cost-effective option if the reduction in antibiotic use was not a priority; other work in LMICs has come to similar conclusions.^45,73

However, testing for inflammatory markers, such as CRP, may be a more realistic and cost-effective strategy than pathogen-specific tests given issues around laboratory access and variations in underlying disease prevalence across LMICs. CRP POCTs in LMICs significantly reduced inappropriate antibiotic prescribing for febrile non-malaria patients and acute RTI.^36,44,46 They represent potentially cost-effective interventions to reduce AMR and improve clinical outcomes at a low WTP threshold and warrant further clinical evaluation.

In another LMIC scenario, Manore and colleagues⁴⁷ found a variety of POCTs for NTS dominated usual care. Unlike other studies included in this review, they modelled the impact of POCT on outbreak size and found that POCTs were cost-effective as they led to a large reduction in NTS cases. Future evaluations of POCTs should try to account for the benefits of reduced transmission that result from prompt diagnosis and treatment using POCTs in their analyses, though data from real-world implementation trials are also required.

POCTs in other scenarios

Other scenarios evaluated are not promising avenues for the adoption of POCTs to reduce AMR: an RADT for acute sore throat was cost-effective, but less so than a clinical scoring tool;⁴⁰ a urinary culture POCT was not cost-effective as clinicians failed to alter prescriptions based on test results;⁴⁹ a GBS POCT was dominated by usual care.⁴⁸

Expanding the scope of costs incorporated

Despite the importance of AMR and the case for accounting for its projected costs in economic evaluations, only five studies performed additional analyses incorporating the costs of AMR.^{32,36,37,39,44} Their analytical approach to incorporating the cost of AMR varied. Two studies used analysis performed by Shrestha et al. who estimated costs based on correlation coefficients between consumption and resistance, the costs of resistance for five common organisms, and consumption rates of the relevant antibiotics, stratified by national income.^36,44,52 This analysis suggested the cost of AMR ranged from $0.30 to $35.10 per course avoided.⁵² Two studies used the cost calculated by Oppong et al., which assumed all antibiotic classes contribute equally to AMR and did not account for geographical variation.^33,38,57 One study used their own WTP threshold for the costs of AMR, based on the proportion of estimated excess costs of AMR attributable to antimicrobial prescribing for RTIs in humans.³⁹

Overall, the societal cost of AMR is often not incorporated into economic analyses. Although this has been argued to result from difficulties in quantifying it, there is a clear need to include such costs to appropriately appreciate and compare the impact of POCTs and other AMR interventions.^19,20 In line with other reviews,²⁴ we recommend that future economic evaluations incorporate up-to-date estimates of the costs of AMR. More POCTs would dominate usual care were these wider societal costs included.

Strengths and limitations

This systematic review, accordant with PRISMA guidelines, comprehensively assessed the cost-effectiveness of POCTs in reducing AMR in a range of settings, adding to existing reviews through identifying several additional studies and including LMICs.^24,26

One important limitation was the relative paucity of manuscripts that met the inclusion criteria, which makes it difficult to draw robust conclusions. Potential economic evaluations performed by biotechnology and pharmaceutical companies may not be published in the peer-reviewed literature and thus could not be included. Studies identified were heterogeneous in setting and their included costs, making comparison challenging and prohibiting meta-analysis. Finally, the impact of delayed healthcare costs, such as reconsultation and hospitalization, as well as AMR,¹⁷ may have been underestimated by short time horizons utilized by studies.

Conclusions and future directions

Despite the considerable estimated economic, societal and healthcare cost of AMR, the issue is often de-prioritized by policymakers.^1,4 In particular, POCTs represent a neglected area within AMR.⁵³ This systematic review has several recommendations for future research and policy: (i) future studies evaluating interventions that may reduce antimicrobial prescribing should include the cost of AMR in their modelling; (ii) up-to-date estimates for the societal, healthcare and economic cost of AMR are required to inform decisions on the cost efficacy of interventions; (iii) policymakers focused on HICs should consider large-scale evaluation and implementation of CRP POCTs for diagnosis of RTIs in primary care to reduce antibiotic prescribing with concomitant clinical benefits; (iv) funders and researchers should consider further development and evaluation of POCTs that can determine the resistance profile of MDR pathogens such as gonorrhoea, as modelling shows they are likely to be cost-effective; and (v) policymakers focused on LMICs should consider real-world evaluation of POC CRPs for febrile patients given cost efficacy and clinical efficacy in modelling studies.

Funding

This study was carried out as part of our routine work.

Transparency declarations

A.B. is the founder of AMR Funding Circle, an independent non-profit conducting research evaluations in the field. AMR Funding Circle did not support this work financially or otherwise.

Supplementary data

Table S1 is available as Supplementary data at JAC Online.

Author notes

Abraham Tolley and Akhil Bansal contributed equally to this work and should be regarded as joint first authors.