The Resource Criterion: 28 Years of Asking the Same Question

1. The Email That Started Everything

In February 2026, a routine email landed in my inbox: "Your latest NICE monthly digest." The National Institute for Health and Care Excellence had published 11 resource impact templates for newly recommended technologies. Each contained an Excel planner that calculated nursing time in minutes, specified Agenda for Change bands, and let local NHS managers customise workforce projections down to the clinic level.

I am a former nurse. I spent years in understaffed healthcare services across Norway, from primary care to specialist hospitals, and from Oslo via Bergen to Tromsø and even Kautokeino. I know what it means when capacity constraints are the reason you skip your break to cover a colleague who called in sick. When I saw those planners, one question surfaced: why does Norway not have this?

I am now a health economics student, and my curiosity led me to investigate how Norway's HTA system handles the resource criterion. Norway gives MORE weight to resource scarcity in priority-setting decisions than almost any comparable country. Resources are a formal rejection criterion. Yet Norway's HTA outputs contain substantially less workforce-specific quantification than the mandatory resource impact templates NICE publishes for every recommended technology.

Last year I considered making this my thesis topic. I decided against it; the problem felt too close. But I could not stop researching. Over the weekend, I analysed 1,408 publicly available documents spanning 28 years of Norwegian health policy and HTA practice. I queried 11 NotebookLM research notebooks more than 368 times and classified 995 drug assessments by analytical depth.

This article is what I found.

2. 28 Years of Asking the Same Question

In 1997, the Lonning II committee published a question that still has no operational answer: "Har teknologien konsekvenser for personalbehovet?" Does the technology have consequences for personnel needs? (NOU 1997:18, p.112)

The same committee answered its own question: "Den viktigste knapphetsfaktoren er spesialister og fagfolk" (the most critical scarcity factor is specialists and professionals).

That was 28 years ago. A parallel government committee (Grundutvalget, NOU 1997:7) published narrower cost-effectiveness criteria. The narrow version became law. The broad version, with its personnel checklist, remained advisory. That fork shaped the next 17 years (1997-2014). When the Norwegian Medicines Agency (then called Statens legemiddelverk, SLV) began conducting HTAs around 2003, it followed the narrow law: financial cost-effectiveness only.

In 2014, Beslutningsforum reinvented Lonning II's personnel checklist as binary questions (Q6-Q8) without acknowledging the historical connection. In 2016, Meld. St. 34 expanded the resource criterion to include personnel, infrastructure, and equipment alongside budgets. In 2025, Meld. St. 21 mandated that personnel consequences of all measures shall be assessed and weighted: "Personellkonsekvenser av alle tiltak skal vurderes og vektlegges."

The rhetoric escalated. In 2016: "Det er ikke mulig å bevilge seg ut av denne utfordringen" (you cannot budget your way out of this challenge; Meld. St. 34, p.29). In 2025: "selv med gode helsebudsjetter vil særlig tilgangen på personell være en begrensende faktor" (even with good health budgets, access to personnel will be the limiting factor; Meld. St. 21, Kap. 3.3). And most starkly: "Norge kan ikke bemanne seg ut av utfordringene" (Norway cannot staff its way out of the challenges; Meld. St. 9 (2023-2024), cited in Meld. St. 21).

Five passages from four government reports spanning 28 years, asking the same question and giving the same answer with escalating urgency. Among the government documents reviewed, these five trace a single thread from diagnosis to mandate. Across the full 2003-2026 corpus reviewed for this analysis, zero assessments included a section distinguishing financial cost from physical capacity constraints.

As a former nurse, I knew this instinctively. You cannot treat a patient with a budget line. You treat them with hands, training, and time.

3. The Paradox: More Policy, Less Practice

As political recognition of workforce scarcity expanded, the actual methodology to measure it contracted. This is not a failure of intention. It is a structural outcome.

Meld. St. 21 captures the paradox in a single sentence: "Det foreslås ingen vesentlige endringer i prioriteringskriteriene... men konsekvenser for behov for helsepersonell skal tydeliggjøres" (no significant changes to the priority-setting criteria, but consequences for healthcare personnel needs shall be made visible).

Across the pre-reform period, 11 instances reached what I classify as Depth Level 5: specific FTE quantification, including Lonning II's 1997 macro-level workforce data and 10 individual drug HTAs from 2004-2014. After 2014, that number dropped to zero. For 11 consecutive years (2014-2024), not a single DMP assessment in the corpus classified for this article (n=995) produced a dedicated workforce quantification. The dominant depth levels shifted to L1-L2: no mention or boilerplate, accounting for roughly 70-85% of assessments in most post-reform periods (with one outlier period, 2020-2021, where monetised personnel costs at L3 accounted for 85% of assessments).

How does a system that formally recognises workforce scarcity produce less analysis over time? Through a three-step mechanism I call neutralisation.

Step 1: Monetise. Convert all personnel inputs to financial variables using the Enhetskostnadsdatabase (unit cost database). A nurse hour becomes a line item, ranging from 323 to 600 NOK across seven drugs identified in this analysis from a single period (2020-2021). The rate is identical whether the nurse exists or not.

Step 2: Adjust down. Override company time estimates with DMP's "realistic" rates.

Step 3: Dismiss. Conclude that personnel costs are negligible compared to drug price.

The dismissal is explicit. For an immunotherapy assessment in 2016: "Blant kostnadene er det kun legemiddelpris som har betydelig påvirkning på IKER" (only the drug price significantly affects the ICER). For a targeted oncology therapy in 2017: "De andre kostnadene betyr forholdsvis lite. Vi har derfor ikke gjort en inngående vurdering" (other costs matter relatively little; we did not conduct a thorough assessment).

This neutralisation extends to language. At the political level, the resource criterion is "ressursbruk" (resource use). At DMP's operational level, it becomes "merkostnad" (additional cost). In Beslutningsforum's decisions, it narrows to "pris" (price). Across the publicly available Beslutningsforum rejection texts reviewed for this analysis (2014-2026), no rejection invoked capacity. All cited price.

Nine drugs assessed between 2024 and 2026 use identical exclusion formulations. Even for advanced cell therapies requiring hospital certification and trained personnel, the same template dismissing non-drug costs as negligible is used.

I recognise this pattern from clinical practice. When the system cannot measure something, it prices it. When the price is small relative to the drug, it disappears. The nurses do not disappear.

4. What CGM Actually Did (and What It Did Not)

In December 2025, DMP published its assessment of continuous glucose monitoring (CGM) for Type 2 diabetes, containing a remarkable sentence: "for the first time, this assessment includes a separate analysis of the impact of introducing a new health technology on national healthcare workforce requirements in Norway."

CGM is the first DMP assessment with a dedicated FTE methodology section in the classified corpus spanning 2003-2026. It used the Dutch Zorginstituut Nederland three-step method (no Norwegian method exists): measure incremental workload per patient, convert to FTE fractions, scale to national totals.

Two implementation models: a specialist-only pathway requiring 21-32 specialist nurse FTEs, and a collaborative model (specialist plus primary care) requiring 91-128 FTE across five workforce categories (specialist nurse, specialist physician, GP, healthcare worker, secretary/admin). One FTE equals 1,663 hours.

The assessment articulated a chain that should have been obvious: "Resource availability determines the extent to which patients can access the technology and, consequently, the additional health benefits measured in QALYs." Resources constrain access. Access constrains benefit. It took DMP 22 years to write it down.

Having a number is not the same as having a method. A 2010 assessment for an oral anticoagulant calculated 21.53 FTE, produced by an employers' organisation (NHO Service), embedded in budget impact analysis, measuring savings. CGM had two models, five workforce categories, scenario modelling, and a transferable methodology, conducted by DMP itself, measuring demand. The difference is between a fact and a framework.

The limitations are real. FTE figures rest on expert consensus from four endocrinologists and one specialist nurse, not time-motion studies. National aggregates only; no hospital-level data. Displacement absent: 128 nurses deployed to CGM with no calculation of what services they leave. Training excluded. Education pipeline (a 9-year specialist training lag) ignored. No probabilistic sensitivity analysis on the FTE estimates.

DMP added the Dutch methodology mid-project in response to Meld. St. 21. This was policy-forced innovation, not organic evolution.

The Diabetesforbundet's consultation response captures the ground-level reality: "pressede fastleger med varierende innsikt i diabetes" (stressed GPs with varying insight into diabetes; CGM consultation, p.195). At least six analytical tools identified in this review already exist in Norwegian government practice that could be applied to HTAs. Five have never been used in HTA. One, the Dutch method, was used once: for CGM.

5. The Aggregate Problem Nobody Is Counting

In 2023, Beslutningsforum considered 93 drugs and approved 59. Each assessed in isolation. Each assumed its workforce impact was small. Nobody calculated the aggregate.

The expert group advising on Meld. St. 21 named this: "Dette kan skape et 'mange bekker små'-problem..." (a "many small streams" problem, where each assessment assumes the whole does not change significantly, while collectively the measures have consequences for demand for certain professional groups; Meld. St. 21, Kap. 4.3.6).

Every assessment includes a 5-year budget impact table. No cumulative register of workforce demand exists across Nye metoder decisions.

The expert group advising on Meld. St. 21: "Analysene som gjøres, tar sjelden hensyn til bemanningsmangler og påfølgende vridningseffekter" (analyses rarely account for staffing shortages and resulting distortion effects; Meld. St. 21, p.104). "Vridningseffekter" appears in Meld. St. 21 but in zero DMP reports.

Two government white papers quantify the challenge through completely different projections. Meld. St. 34 (2016), citing SSB projections for the specialist health service: "i underkant av 30 pst. flere årsverk i spesialisthelsetjenesten fram mot 2030 og 40 pst. fram mot 2040" (nearly 30% more FTEs in specialist healthcare by 2030 and 40% by 2040). Meld. St. 9 (2023-2024), projecting across the entire health and care service: "et økt årsverksbehov... på ytterligere 30 prosent de neste 15 årene" (an additional 30% FTE need over the next 15 years). Different scopes, different timeframes, same direction.

Consider a recent assessment for a complex CAR-T therapy. DMP's budget model assumed treating fewer than 10% of eligible patients per year, a costing scenario base case that implied over 90% of eligible patients would not receive treatment. This assumption reduced the projected budget impact by more than 90%. The patient limit was not a formal Beslutningsforum decision but a budget model assumption. Only a single national centre qualified for delivery. Across the publicly available Beslutningsforum decision texts reviewed for this analysis (2014-2026), no rejection invoked capacity. All cited price.

In my nursing years, we called this "task stacking." Each new protocol was "just five minutes." Thirty protocols later, you have lost your lunch break and your ability to think clearly. The Norwegian HTA system is doing this at national scale, except nobody is counting the protocols.

6. What NICE Does That Norway Does Not

The Norwegian government acknowledges the stakes: "Kompetent helsepersonell er helsetjenestens viktigste ressurs, men denne ressursen er et knapt gode" (competent health personnel are the health service's most important resource, but this resource is a scarce good; HOD, 2025).

In February 2026, a single monthly NICE newsletter contained 11 resource impact templates for newly recommended technologies. Each specified staff time in minutes by Agenda for Change band (13 grades, 3 costing points each), modelled adoption across roughly 1,300 geographic areas, and included blue-cell inputs for local NHS managers to customise projections.

I am comparing Norway's 28-year history against one monthly newsletter from NICE. Even that single month demonstrates workforce impact measurement Norway has never achieved.

The examples below illustrate NICE's broader system capability, not the February newsletter specifically. For an implantable cardiac monitor: changing the monitoring nurse from Band 5 to Band 7 altered the ICER by 66%; the workforce assumption changed the cost-effectiveness conclusion. In Norwegian practice, "nurse" is a single generic category. For an intravenous oncology biologic: 210 minutes of Band 7 nursing per cycle, totalling 80,597 additional nursing hours per year nationally. For a targeted eye-impacting therapy: 3,306 new ophthalmology appointments by Year 3, cross-specialty demand a Norwegian oncology assessment would not flag. For a biologic treating chronic inflammation: an annual steady-state reduction of 491 endoscopic sinus surgeries by Year 3 at 50% uptake, a workforce saving Norway's system cannot capture.

NICE uses resource impact for implementation planning, not rejection. Norway uses it as a rejection criterion. NICE's floor (Level 4 workforce analysis for all recommended technologies) is Norway's ceiling. Measure like NICE, decide like Norway.

A 2025 Lancet study used NHS programme budgeting data to estimate that 183 NICE-recommended drugs consumed resources worth 5 million QALYs while gaining only 3.75 million QALYs: a net loss of 1.25 million QALYs. Norway cites this study in Meld. St. 21 as justification for tightening the resource criterion, yet has no equivalent data infrastructure to produce such an estimate for its own system.

7. The Mandate and What Comes Next

In 2025, Meld. St. 21 established five new requirements, the most important being that personnel consequences shall be specifically assessed for all priorities. Assignment letter TB2025-110 tasked Helsedirektoratet with developing the methodology.

The mandate is explicit: "Personellkonsekvenser skal utredes spesifikt..." (personnel consequences shall be specifically assessed; TB2025-110).

The timeline is not: "Departementet kommer tilbake med nærmere frister" (the department will return with more specific deadlines; TB2025-110). No deadline has been set.

This creates a circular dependency. Industry has no requirement to submit workforce data. DMP has no method to assess it. No standardised data exists to analyse. CGM broke this cycle by having DMP generate the FTE analysis itself, using a borrowed Dutch methodology.

Twelve evidence gaps remain. No Norwegian empirical opportunity cost estimate. No cumulative FTE register. No hospital-level capacity data. No time-motion studies. No displacement analysis. No education pipeline analysis. No waiting list modelling. No regional capacity breakdowns. No training cost assessment. No cross-sector municipal capacity assessment. No modern equivalent to Lonning II's macro workforce data. No post-implementation monitoring. Seven new policy terms (personellkonsekvenser, vridningseffekter, bemanningskonsekvenser, and others) appeared in 2025 government documents. None are yet reflected in DMP practice.

If resource costs of new drugs consistently exceed their health gains, measuring those resources is the difference between informed priority-setting and flying blind.

I decided not to write my thesis on operationalising the resource criterion because the problem felt too large. A year later, the government mandated what I was considering researching. The question is no longer whether to do it, but how.

8. What I Wish I Had Known as a Nurse

When I was a nurse, I never thought about health technology assessment. I thought about the patient in front of me and the colleague who had not shown up. Capacity was not a policy concept. It was whether I could take a bathroom break before the next medication round.

It took leaving clinical practice, pursuing the European Master in Health Economics and Management (EU-HEM), and spending a whole weekend inside 1,408 documents to see what I could not see from the bedside: the system that decides which treatments patients receive does not count the people who deliver them. Twenty-eight years of asking the same question, 995 assessments classified, 368 research queries later, the answer is simple. Norway knows that funding is not capacity. It has said so since 1997. It just has not built the tools to act on it.

CGM changed something. Whether it changed enough, we will find out.

You have just gone through 1,408 documents with me. The complete list is available for download below. This is an invitation to a conversation, not a verdict.

If you are a health economist, a clinician, a policymaker, or anyone who has lived the gap between policy rhetoric and ward reality, I would like to hear from you.