Economic perspective
The viewpoint that determines which costs and consequences enter the business case. It should be declared together with the comparator, time horizon and volume scenario [1].

Three pathway times
Analytical time, operational TAT and time to decision describe different intervals. Explicit start and end points are needed to compare them [2].

Transferred cost
A cost that changes location, budget or owner without disappearing. It becomes a saving only when the item is genuinely avoidable and is not counted elsewhere.

Unused capacity
The share of installed capacity not absorbed by ordinary demand. It may be inefficiency or a necessary reserve for peaks and emergencies, provided it is quantified and financed transparently.

Expansion gate
Criteria established before the pilot to decide whether to scale, stop or redesign on the basis of diagnostic performance, operational continuity and clinical-economic consequences.

Decentralisation does not make PCR intrinsically cheaper or more expensive. It changes the geography of the work: it may reduce some waiting and logistical activities, but it distributes instruments, controls, expertise, stock and responsibilities. Assessing it requires following the pathway from sample to decision, rather than stopping at the price of the test or the instrument run time.

This Insight builds the business case around that chain. It distinguishes the different TATs, separates avoided and transferred costs, measures utilisation and reserve capacity, considers error and waiting, and turns the comparison into a pilot with criteria for expansion, termination or redesign. The hybrid network is not assumed to be a universal solution: it is a configuration to test when the value of reduced latency justifies distribution and the centre continues to provide scale, specialisation and governance.

Unit cost is appealing because it appears to offer an immediate comparison. It combines the cartridge or reagents, a share of the instrument and a few minutes of work to produce a precise number. Yet that number rarely represents the real pathway.

In the centralised laboratory, volumes, automation, specialist staff and consolidated purchasing lower the analytical cost. Around that price, however, there remain activities that are often charged to other budgets: collection and packaging, courier transport, accessioning, batch waiting, repeats caused by sample quality, communication of the result and clinical management while waiting. In a distributed network, some items decline while others emerge: more instruments, operator training, controls across multiple sites, connectivity, support, fragmented inventories and unused capacity outside peak periods.

To choose between them, it is not enough to ask which test costs less. A readable business case states the comparator, the perspective (laboratory, hospital, network or health system), the time horizon and the volume scenario [1]. If a cost moves from the laboratory to a ward or peripheral site, the system has not necessarily achieved a saving. Genuinely avoidable costs must be separated from fixed or already incurred costs, and logistics, staff or infrastructure must not be counted twice. It then becomes possible to estimate the cost of a reliable decision within the window in which it can still be useful.

Turnaround time does not have uniform boundaries: it may begin when the test is requested, the sample is collected or the specimen arrives, and end at validation, availability or viewing of the result. The literature shows that, even within the same clinical domain, different definitions make benchmarking difficult [2].

Three intervals help make the comparison readable, provided that the timestamps are stated. Analytical time here means the interval from the start of instrument analysis to the availability of the technical result. Operational TAT includes, according to the chosen starting point, transport, queueing, preparation, amplification, validation and report availability. Time to decision ends when the result is actually used. A rapid platform may shorten the first interval while leaving the last unchanged: this happens, for example, if the result arrives after the clinical round, does not enter the information system or there is no protocol authorising de-escalation.

This gap explains why apparently similar studies produce different results. In a randomised trial of adults admitted with lower respiratory tract infection, adding a rapid molecular panel, described by the authors as point-of-care, to routine PCR was associated with one fewer day of intravenous antibiotics, one fewer day in hospital and lower hospital costs. However, the economic estimate did not include the cost of the panel, the study was conducted at a single centre, and the supplier provided instruments and consumables free of charge [3]. In the pragmatic ResPOC trial, rapid molecular testing improved viral detection, appropriate antiviral use and isolation management, but did not reduce the overall proportion of patients treated with antibiotics or the total duration of therapy [4].

The two studies concern specific platforms, populations and hospital workflows. On their own, they do not demonstrate a TCO advantage for every distributed qPCR network, still less for a specific product. In both cases the technology was fast; its value was determined by the population, protocols, clinical behaviour and the decision window.

An earlier result has value when it meets a decision that is still open: starting or stopping an antimicrobial, assigning an isolation room, admitting or discharging a patient, or activating an infection-control measure. If management does not depend on the result, or if the decision has already been made and is unlikely to be reconsidered, that time margin matters far less.

In the Norwegian trial on community-acquired pneumonia, a syndromic PCR panel used in the emergency department increased the proportion of patients receiving pathogen-directed treatment and reduced the time to that treatment by 9.4 hours [5]. The trial demonstrates an effect on the timing and direction of treatment in that pathway, not a general improvement in outcomes: there were no significant differences in length of stay, readmission or mortality, and the authors call for further cost-effectiveness and implementation assessments. The outcome cannot automatically be transferred to an outpatient clinic, a residential facility or a population with a different prevalence.

Economic evaluation of diagnostics therefore begins with the decision, not the panel menu. Expanding the number of targets may increase detections, but it may also produce findings that are difficult to interpret, colonisation or results unable to change treatment. A target creates value only when its presence or absence leads to a proportionate action.

When instruments are distributed, responsibilities are distributed with them. Every peripheral site must be able to receive and store materials, identify the patient, prepare the sample, perform controls, manage invalid results, prevent contamination, dispose of waste, transmit the result and recognise when to call the laboratory. These activities are clinically necessary, and their cost belongs in the model.

ISO 15189:2022 specifies quality and competence requirements for medical laboratories and is also applicable to testing performed near the point of care [6]. This does not automatically turn every distributed site into an autonomous laboratory, nor does it alone define the regulatory scope of an individual workflow. It does, however, confirm that operator qualification, traceability, quality control and supervision must be placed within the model rather than among the extras. A network with twenty instruments is not equivalent to twenty independent small laboratories: it needs central oversight capable of authorising operators, tracking lots and versions, comparing performance, maintaining competence and intervening when deviations arise.

Connectivity reduces part of the burden when it automates demographics, logs and result transfer. In return, the model must absorb LIS/EMR integration, cybersecurity, update management and continuity during a network outage. Likewise, a reagent kit that is stable at room temperature may simplify the supply chain without eliminating transport conditions, protection from humidity, expiry dates, sample preparation or requirements for biological material. The claim 'cold-chain-free' applies to a validated configuration and limits; by definition, it does not describe the entire workflow.

A distributed instrument creates capacity, but the effective cost depends on how much that capacity is used. At low volumes, each test absorbs a large share of depreciation, maintenance, controls and training. At peak times, queues, staff shortages and a need for backup may instead arise. The balance must be sought in actual demand, not nominal capacity.

Seasonality further complicates the calculation. A respiratory network may operate intensively for a few months and remain underused during the rest of the year. A multi-application menu helps increase utilisation only if each assay has adequate volumes, expertise and governance. Adding tests merely to keep the instrument busy risks encouraging inappropriate use.

An annual average alone conceals what the business case needs: the number of tests on ordinary days, the scale of peaks, the sites with sufficient demand and the cost of reserve capacity. This capacity is not necessarily inefficient. In an emergency, it may function as insurance, provided that it is recognised and funded as such.

False negatives, false positives, invalid results and late results belong in the economic assessment. An inexpensive test that must often be repeated may cost more than expected. A false or uninterpretable result may trigger treatment, isolation, confirmation and further sampling; a correct result available only after discharge may have little value for that episode.

Studies of rapid tests show that savings, when they emerge, often come from downstream items: length of stay, isolation, antimicrobials or the use of other investigations. A modelling study of influenza and RSV estimated possible reductions in hospital costs in scenarios where a faster result shortened isolation and, in the most favourable assumption, brought discharge forward [7]. This is a conditional estimate, not an observed saving: if clinical behaviour or length of stay does not change, much of the advantage disappears.

Economic evaluations of rapid tests in antimicrobial resistance are heterogeneous and are not equivalent to evidence on distributed qPCR. A 2024 review included 20 studies covering different tests, settings and models: all adopted the healthcare-provider perspective, and only five added the societal cost of AMR. Respiratory tests, particularly CRP, were among the most frequent strategies; the body of evidence does not prove the value for money of a distributed qPCR network. Cost-effectiveness conclusions depended on the test, setting and chosen threshold [8]. There is therefore no universal ROI for decentralisation.

The central laboratory retains advantages that are difficult to replicate: economies of scale, breadth of menu, expertise in complex cases, confirmatory testing, sequencing, cultures and exception management. The periphery intervenes where it can reduce critical latency and improve access. An efficient network arises from the combination of these two levels.

In the hybrid design, peripheral sites receive tests with high time value, a simplified workflow and a defined action; rare or complex analyses, high-multiplex tests and those requiring specialist interpretation remain at the centre. An escalation pathway is needed between the two levels because a doubtful or discordant result cannot remain isolated at the site.

The placement decision can be framed as a question: how costly is it to move the sample, and how costly is it to move the expertise? In some settings it is preferable to transport the sample; in others it is preferable to bring the test closer to the patient while maintaining remote supervision. The boundary may change over time with volumes, menu and infrastructure.

CAPEX, leasing, reagent rental and pay-per-use models distribute risk and cash flow differently. Purchasing may be advantageous with predictable volumes and adequate internal capacity. A service model may reduce the initial investment and may include maintenance, but it can tie costs to minimum consumption, contract duration and data portability. The financing structure alone does not strengthen a weak use case.

The comparison should include the total cost over the contract term, service levels, replacement times, responsibility for updates, training, interfaces, data ownership and exportability, management of expired reagents and exit conditions. 'Diagnostics-as-a-Service' is a procurement model, not evidence of value for money.

The stated cost per test must also be questioned: does it include controls and invalid results? How does it change with volume? What happens if a lot is recalled or the instrument is out of service? An offer is economically readable when it also makes these events visible.

The current flow should be observed before constructing the distributed scenario. Explicit timestamps are needed for collection, departure, arrival, accessioning, start of testing, validation, availability in the system, clinical viewing and decision; then volumes by hour and day, the percentage of urgent samples, invalid results, repeats, staff time and logistics costs. The new pathway is simulated on this basis, separating the base scenario, peaks and reserve capacity.

The pilot should measure indicators at three levels. At the diagnostic level: agreement, invalid results, repeats, controls, deviations between sites and performance by operator. At the operational level: time to decision, availability, staff time, consumption, stock and continuity during downtime. At the clinical-economic level: changes in treatment, isolation, admission, additional investigations and cost per episode, stating the adopted perspective. It is important to define in advance which results justify expansion and which require termination or redesign.

A trial of rapid molecular blood-culture identification showed that coupling antimicrobial stewardship with the result accelerated de-escalation more than the rapid result alone, without differences in mortality, length of stay or cost [9]. This is a concrete example of how technology and pathway produce value together.

Decentralising PCR means reallocating capacity, expertise and risk, not reproducing the central laboratory on a smaller scale. Even with a higher analytical cost, the pathway may cost less when the time gained enables a timely decision that changes the workflow in a documented way. Without adequate volumes, governance and defined actions, the same architecture may instead add costs and complexity.

A credible business case makes explicit what disappears from the pathway, what moves, what remains fixed and what must be built. It does not promise the end of logistics or call every transferred cost a saving. In the hybrid network, the centre remains the coordinating authority, reference point and home of exceptions; peripheral sites cover the steps where latency has measurable clinical and operational value.