Evidence Brief September 18, 2026

This week’s strongest evidence raises a deceptively important question across patient preferences, evidence generation, and health economics:

When stakeholders agree on what matters, do they also agree on how much it matters?

Two new discrete choice experiments suggest the answer is often no. A third study shows how the apparent economic value of an intervention can change when the decision is expanded from an isolated technology to the surrounding care pathway.

Peer-Reviewed Evidence

1. Jones et al. Preferences for the Use of Artificial Intelligence Technologies to Help Detect Skin Cancer in Primary Care Settings: A UK-Wide Discrete Choice Experiment. British Journal of Cancer.

This large UK discrete choice experiment included general practitioners, patients, and members of the public and examined preferences for the use of artificial intelligence in skin-cancer detection.

Attributes included false-negative and false-positive rates, implementation setting, cost, performance across different skin tones, and guideline recommendations.

All three stakeholder groups placed greatest importance on avoiding false-negative results and on evidence that the technology performed appropriately across different skin tones. General practitioners were particularly averse to false negatives. Participants also generally preferred clinician-mediated use in primary care over direct patient use at home.

Why it matters

This is a useful example of preference research conducted before widespread technology implementation.

The study moves beyond asking whether people “like” AI. It identifies the conditions under which an AI-enabled diagnostic pathway may be acceptable and therefore helps define the implementation strategies that subsequent clinical and economic evaluations should compare.

The broader lesson is important for evidence-generation planning: technical performance alone does not determine value. Where a technology is used, who mediates it, and whether its evidence base represents the population can all influence acceptability and adoption.

Preference research can therefore help shape the decision problem before an economic model is built.

2. Wu et al. Treatment Preferences for Acute Myeloid Leukemia in Harbin, China: A Discrete Choice Experiment Among Patients, Caregivers, and Hematologists. The Patient.

This study separately elicited treatment preferences from patients with acute myeloid leukemia, caregivers, and hematologists.

Attributes included remission, quality of life, long-term toxicity, route of administration, and monthly treatment cost.

All three groups identified remission as the most important attribute. However, the similarity largely ended there.

Hematologists were willing to accept substantially greater trade-offs to obtain higher remission rates, while patients and caregivers placed relatively greater emphasis on cost, quality of life, and long-term toxicity.

Why it matters

This study demonstrates why agreement on attribute rankings should not be mistaken for agreement on preferences.

Patients, caregivers, and physicians can all identify efficacy as their highest priority while assigning very different value to incremental improvements in efficacy relative to toxicity, quality of life, or cost.

That distinction is easily lost when preference studies report only relative attribute importance.

For evidence intended to inform product development, benefit-risk assessment, or shared decision-making, measures of actual trade-offs — such as willingness to pay, marginal rates of substitution, or predicted choice probabilities — may be considerably more informative than rankings alone.

Clinician preferences should also not be assumed to proxy patient preferences simply because both groups identify the same outcome as important.

3. Cao et al. Cost-Effectiveness of Smoking Cessation Interventions Integrated Into Lung Cancer Screening. JAMA Network Open.

The investigators used participant data from randomized studies combined with lifetime microsimulation to evaluate smoking-cessation interventions delivered within US lung-cancer screening programs.

Smoking-cessation strategies integrated with screening were economically favorable compared with screening alone. The most intensive strategy, combining pharmacotherapy and counseling, generated the greatest projected health benefit while remaining highly cost-effective relative to less intensive alternatives.

Why it matters

The most important contribution is not the specific ICER.

It is the way the decision problem was defined.

Evaluating lung-cancer screening and smoking cessation as independent interventions risks missing the interaction between two services delivered through the same clinical infrastructure.

For payers and health systems, this is a useful example of why economic evaluation should sometimes focus on care pathways rather than isolated technologies.

An intervention that appears to add cost when viewed alone may represent substantially greater value when downstream consequences and shared implementation infrastructure are incorporated into the analysis.

4. Becking et al. Treatment Preferences in Waldenström Macroglobulinemia: An International Discrete Choice Experiment in 1,455 Patients. Blood Neoplasia.

This large international DCE examined treatment preferences among patients in Australia, Canada, the Netherlands, the United Kingdom, and the United States.

Duration of response and avoidance of temporary toxicity were the strongest determinants of treatment choice, followed by the risk of secondary malignancy and persistent toxicity. Oral administration and fixed-duration treatment were preferred, but convenience attributes carried less weight than clinically meaningful differences in efficacy and toxicity.

Preferences were also relatively consistent across participating countries.

Why it matters

The study provides unusually robust preference evidence in a rare malignancy and offers a useful product-development message.

Convenience matters, but improvements in administration or treatment duration may create less patient value than improvements in treatment durability or toxicity when those clinical differences are substantial.

The result should nevertheless be interpreted within the experimental design. Relative attribute importance depends on the ranges investigators choose to present.

A finding that efficacy dominates convenience does not mean convenience has little intrinsic value. It means that, within the trade-offs represented in this particular experiment, respondents were willing to sacrifice convenience for sufficiently meaningful clinical benefit.

Payer / Evidence-Strategy Watch

NICE recommended trastuzumab deruxtecan for HER2-low advanced breast cancer following an extended pricing and value assessment process.

The clinical value proposition was not fundamentally new. What changed was the set of conditions under which that evidence represented acceptable value to the health system.

Why it matters

An unfavorable HTA decision does not necessarily end the value story.

Price, commercial arrangements, comparator assumptions, evidence maturity, and the framing of uncertainty can all change the reimbursement decision even when the underlying clinical evidence changes relatively little.

For evidence teams, this reinforces the importance of treating HTA as an iterative decision process rather than a single submission event.

Preprints and Emerging Evidence

No new oncology DCE, DCE-TTO, patient-preference, or health-economic preprint published this week met our threshold for inclusion.

The Evidence Brief favors selectivity over completeness. Preprints are included when they introduce methods or findings with a plausible near-term impact on study design, interpretation, evidence generation, or payer strategy.

What This Means for Evidence Strategy

The strongest theme this week is decision perspective.

Patients, clinicians, caregivers, and the public may agree on the outcomes that matter while placing very different value on incremental changes in those outcomes.

Likewise, the economic value of an intervention can change when the analytical boundary expands from a single technology to the surrounding care pathway.

The broader implication is straightforward:

Agreement on what matters does not imply agreement on how much it matters.

Rankings, statistical significance, and isolated ICERs can therefore conceal important differences in the underlying decision problem.

Decision-grade evidence requires explicit attention to trade-offs, stakeholder perspective, comparator structure, and the boundaries placed around the analysis.

Worth Reading First

For preference methodology, Wu et al. is the strongest paper this week. The divergence between patients and clinicians despite agreement on the highest-ranked attribute is a particularly clear demonstration of why preference research should emphasize trade-offs rather than rankings.

For evidence-generation strategy, Jones et al. shows how preference research can be used prospectively to shape technology implementation rather than merely evaluate acceptance after launch.

For health economics, Cao et al. provides the most useful strategic lesson: changing the boundary of an economic evaluation from an isolated intervention to the broader care pathway can materially change the value proposition.