TheraDoc

The engine evaluates alert logic that the hospital's own pharmacists and infection preventionists write, continuously against incoming clinical data. Premier makes no artificial intelligence or machine learning claim for the product anywhere on its own pages, and that restraint is credited: a rules engine described as a rules engine is more useful to a buyer than one dressed up as something else.

The only genuine model in the shipped product is licensed, since Premier's own material states that the integrated precision vancomycin dosing capability inside TheraDoc is InsightRX. This record was a borderline call and the reasoning is recorded so it can be reversed cleanly. On the Orion Health ground, a platform with no model and no claim would be rejected.

It is included because the product a hospital actually buys does contain a model, and because the incumbent's role as a distribution channel for a startup's algorithm is a defining fact about this category rather than a footnote.

Alerts reach a clinical pharmacist or an infection preventionist who investigates and acts, and the platform never touches an order. The structural difference from a model driven product is that the customer authors the logic: alert rules are written by the hospital's own clinicians rather than supplied by the vendor, so the institution controls both what fires and what threshold fires it.

That is the strongest form of local control available in this category and it inverts the usual oversight problem, since there is no vendor decision to defer to. The corresponding weakness is also local. A rule written badly, or left unrevised as practice changes, produces exactly the alert fatigue this segment is notorious for, and the vendor publishes no alert burden figure, no guidance on rule review cadence and no tooling described for identifying rules that fire without ever changing care.

Rules engines win this axis by construction and the point is worth making plainly for the category. When the hospital's pharmacists author the alert logic themselves, every criterion behind every alert is readable by the institution that will act on it, which is a level of inspectability no machine learning product in this index offers. A pharmacist can answer why an alert fired by opening the rule. Held at B rather than higher for two reasons.

The embedded precision dosing component is a third party's model and neither its mechanism nor its parameters are described in Premier's material, so the one part of the product that does learn is the part a buyer cannot see. And the vendor publishes no library documentation, no description of the shipped starter rule set and no account of how supplied rules were validated before distribution.

The mechanism is inspectable and the parties are not. Because the alerting layer is a rules engine configured by the hospital, the institution can read what the system evaluates without needing the vendor to describe it, which answers more of this axis by construction than most disclosure statements do.

One third party is acknowledged to exist without being characterised: an embedded precision dosing component belongs to another company, and neither its mechanism nor its parameters appear in the vendor's material, so a buyer knows a supplier is in the chain and nothing else about it. No hosting arrangement or sub processor list was located. The parent relationship is where the substantive question sits and it deserves a direct answer rather than an assumption.

The platform consumes continuous live feeds of laboratory, microbiology, pharmacy, admission and medication data across an entire hospital, which is one of the broader clinical footprints in this category, and the parent separately holds data on a large share of United States health system discharges and operates as a healthcare data business. Whether surveillance data stays scoped to the hospital's own alerting or contributes to that wider analytics operation is the material term. Ask it in writing, and ask for the dosing component's supplier and terms.

The market position is long established and the published evidence is promotional rather than scientific. Premier publishes customer results including a claimed forty percent reduction in daily surveillance time through automated infection tracking and standardised workflows, illustrated with a named health system case study.

Retrieval located no peer reviewed outcome study, no controlled evaluation and no published data on alert validity, override rate or the proportion of alerts that changed care, which are the measures this category turns on. Twenty five years of continuous use across a large installed base is a meaningful signal about workflow fit and reliability, and this index consistently declines to treat deployment scale as a substitute for evidence of benefit. A vendor with this much longitudinal data across this many hospitals is unusually well placed to publish something better.

The platform consumes continuous live feeds of laboratory, microbiology, pharmacy, admission and medication data across an entire hospital, which is one of the broader clinical data footprints in this category, and the parent company separately states that it holds data on a large share of United States health system discharges.

Against that breadth, the product page makes specific compliance commitments rather than general ones, naming health information technology certification, third party vulnerability assessments and privacy rule compliance together.

Held at B rather than higher because no retention schedule, de identification practice or statement about whether surveillance data contributes to the parent's wider analytics business was located, and that last question deserves a direct answer given the parent's scale as a healthcare data company and its recent transition to private equity ownership.

Compliance is asserted on the product page as part of a named set alongside health information technology certification and third party vulnerability assessment, which is a more substantial posture than the bare claim most vendors in this lane offer and considerably more than the several that publish nothing at all.

The parent is a large established supplier to health systems with correspondingly mature contracting, so the underlying instrument is certain to exist and to have been negotiated many times. Held at B for the reason that applies across this index: the agreement itself is not published, so breach notification timelines, subcontractor flow down and data return obligations on termination remain invisible until contact. Buyers should also confirm that contracting entities and terms are unaffected by the change of ownership at the parent.

The product page names three specific things rather than gesturing at security: certification under the national health information technology programme, third party vulnerability assessments, and privacy rule compliance. Third party vulnerability assessment is the item worth highlighting, because testing cadence is the question most vendors in this category leave entirely unanswered and naming it at all puts this product ahead of most of the lane.

Health information technology certification is a genuine external review, though of interoperability and functionality rather than of security specifically, and the distinction should not be blurred. Held at B because no service organisation controls report, no HITRUST certification, no ISO 27001, no trust centre and no vulnerability disclosure programme were located, so the claims are stated without documents a security team can read.

No device clearance was located and none would be expected, since the product surfaces alerts for a clinician to investigate rather than computing a therapeutic quantity, which is squarely within the statutory exclusion for non device clinical decision support.

One distinction deserves stating because the product page invites confusion: health information technology certification is administered under an entirely different programme from device regulation, and assesses interoperability, data standards and functionality rather than clinical safety or efficacy. A hospital reading certification on the page should not infer that any regulator has assessed whether the alerts are clinically correct.

The embedded precision dosing component raises the more interesting question, since dose computation is the function that made other products in this category regulated devices, and which entity carries responsibility for that component is not addressed publicly.

Retrieval located no performance reporting by any patient characteristic, no bias assessment, no rule validation methodology and no post deployment monitoring statement. The exposure in a rules driven system is different from a model driven one and is not smaller.

Rules encode the assumptions of whoever wrote them, including thresholds drawn from populations that may not resemble the patients being screened, and unlike a model there is no drift monitoring concept in this segment at all, so a rule that has quietly become wrong as practice or resistance patterns changed continues firing indefinitely with institutional authority behind it. That is the pathway staleness exposure this index recorded for content led vendors, applied to alerting. The standing ask is what detects a rule that has gone stale, and who owns that review.

The design allocates responsibility more honestly than most products in this index, and that is worth stating as the reason for the grade. The hospital's own pharmacists author the alert logic, so every criterion behind every alert is readable by the institution that will act on it, and a pharmacist asked why an alert fired can answer by opening the rule.

That is contestability of a kind no learned model here offers, and it also places authorship where the clinical accountability already sits rather than obscuring it behind a vendor's model. Two things hold it at C. The shipped starter rule set is undocumented, with no account of how supplied rules were validated before distribution, so the rules a hospital inherits on day one carry the vendor's judgement without the vendor's evidence, and most institutions will run substantial parts of that set unchanged.

And the embedded precision dosing component is a third party's model whose mechanism and parameters are not described anywhere in the vendor's material, so the one part of the product that actually learns is the part a buyer cannot inspect, which inverts the transparency the rest of the design provides. No accuracy figure for that component, and no warranty, indemnity or remediation commitment, was located. Ask for the validation behind the shipped rules and for the dosing engine's performance and parameters.

The platform is designed to sit across whatever record system a hospital runs, consuming live laboratory, microbiology, pharmacy and admission feeds and returning alerts and dashboards, with certification under the national health information technology programme providing external confirmation of its data standards conformance. Reach is substantial, spanning a large installed base of hospitals and health systems including federal medical centres reached through a government reseller.

Held at B because the integration is described functionally rather than technically, with no FHIR conformance statement, no marketplace listing and no public interface documentation located, and because alert delivery back into the clinician's workflow is not described with the specificity a competitor offers.

The product is delivered as software as a service on the parent's integrated technology platform, which is the model it has used since the acquisition. Retrieval located no named hosting provider, no cloud region, no data residency commitment and no on premise option, and no description of what happens to surveillance coverage during an outage, which for a continuous alerting system is a clinical question rather than an availability one.

The ownership change adds a reason to ask now rather than at renewal, since a take private transaction is often followed by infrastructure consolidation, and a hospital relying on continuous surveillance should establish whether its hosting arrangements are contractually fixed.

No price, unit, pricing basis, contract shape or implementation fee was located. Two commercial dynamics are worth raising before a renewal. The parent completed a take private transaction valued at 2.6 billion dollars after reporting an eleven percent revenue decline in fiscal 2025 and a multiyear contraction, and private equity ownership typically brings a review of product pricing and portfolio scope, so a customer should seek contractual protection on renewal terms rather than assume continuity.

And the same parent owns a separate point of care decision support product, which creates an obvious bundling path: establish what is priced separately today and what is likely to be packaged tomorrow.

Functional coverage is broad within acute care, spanning infection prevention with mandatory reporting to the national surveillance network, antimicrobial stewardship including de escalation and organism to therapy mismatch alerting, pharmacy surveillance for adverse drug events, anticoagulation monitoring and antibiogram reporting, so a single platform serves the pharmacy department, the infection prevention team and quality reporting.

Institutional coverage extends across a large installed base of hospitals and health systems and into federal medical centres through a government reseller, which is a demanding procurement environment and a meaningful signal. Held at B because the footprint is essentially inpatient acute care in the United States, with no evidence of ambulatory or post acute coverage of the kind its direct competitor describes.