In the April 9, 2026, issue of SCIENCE, see an article by Phillips, Horn, & Califf on the value of diagnostics and good diagnostics policy. It's titled:
- Diagnostics investments and disease burden
- Regulatory and payment policies could facilitate investment and innovation in diagnostics
Here's the abstract:
- Screening and testing for disease using “diagnostics” (see the box) is increasingly essential across global health care systems to identify and target individuals who will most benefit from health care interventions and to address unmet disease burden. However, access to diagnostics is uneven and often inadequate, particularly in lower- and middle-income countries, with an estimated 47% of the world’s population having limited or no access to diagnostics (1).
- Despite these needs, diagnostics historically have received lower investment, insurance coverage, and payment rates than drugs, creating barriers to innovation (2).
- We leverage a recent, in-depth assessment of the mismatch between drug development and disease burden in the US (2) to address two critical gaps:
- (i) policies specifically relevant to diagnostics and integrated diagnostic–drug combinations, and
- (ii) how findings for the US extrapolate globally.
- We focus on underinvestment in diagnostics and explore how regulatory and payment policies could facilitate diagnostic innovation.
- ###
Science paper argues the real bottleneck in precision medicine is diagnostics, not drugs
A Policy Forum published Thursday in Science delivers a familiar complaint in unusually blunt terms: health systems are celebrating targeted drugs while underfunding, underpaying, and under-prioritizing the diagnostic tests that determine whether those drugs should be used at all. The paper, by Kathryn A. Phillips and Danea M. Horn of UCSF and former FDA commissioner Robert Califf, argues that this mismatch is now large enough to impede both innovation and patient care.
For expert readers, the striking point is not that diagnostics matter. That is old news. The sharper claim is that diagnostics are being systematically disadvantaged by the structure of regulation and reimbursement. The authors say that tests are typically evaluated and paid for separately from the therapies they inform, even when the test is clinically necessary for the appropriate use of the drug. In their telling, the result is a fragmented market in which the treatment captures the value and the test absorbs the uncertainty.
The paper is a policy analysis rather than a new clinical or claims-based study, but it arrives with a few numbers designed to crystallize the imbalance. It cites an estimate that 47% of the world’s population has limited or no access to diagnostics. It also reports that from 2015 to 2024, 26.0% of drugs received accelerated approval, versus 4.6% of diagnostics receiving Breakthrough Device Designation, the roughly analogous expedited pathway for devices. The implication is not simply that tests move more slowly, but that the system offers manufacturers weaker reasons to pursue ambitious evidence generation in the first place.
Phillips, Horn, and Califf frame the problem as one of misaligned incentives. Drug developers can often justify large investments because the downstream commercial path is relatively legible. Diagnostics developers, by contrast, face a harder evidentiary problem, since a test provides information rather than direct treatment, and must often prove not just analytic performance but some link to clinical utility. That burden is compounded by the fact that coverage and payment frequently remain unclear even after regulatory progress. The authors describe a feedback loop in which uncertain reimbursement suppresses investment, which in turn suppresses the sort of evidence payers say they want.
Two examples anchor the article. One is GLP-1 therapy for obesity and diabetes. The paper notes that although the class is clinically and commercially transformative, a meaningful share of patients do not respond or discontinue treatment within a year. Yet there is no validated diagnostic to predict likely response before treatment begins. For the authors, that is precisely the kind of missed opportunity created by a system that rewards therapeutics more reliably than the tests that could target them.
The second example is Alzheimer’s disease, where the authors see a particularly conspicuous policy asymmetry. New anti-amyloid drugs can cost roughly $30,000 annually in the United States, while blood-based biomarker tests that could help identify appropriate patients are priced around $1,000. Even so, the paper says reimbursement for those tests remains unclear or absent. It points to Lumipulse as a case in point: a blood-based test with European regulatory approval, FDA Breakthrough Device Designation, and US availability through LabCorp, but no published evidence of insurer coverage as of March 2026. For a field now trying to operationalize treatment selection at scale, that gap is more than symbolic.
What the authors want is not simply higher payment for tests. They are asking for a more integrated policy architecture. Their recommendations include broader use of expedited pathways for diagnostics, more joint review of tests and therapies for both regulatory and payment purposes, better incentives for collecting real-world evidence on clinical utility, and value assessments that examine diagnostic-guided care as a combined intervention rather than treating the test as a narrow cost input. They also point to CMS’s Coverage with Evidence Development as one mechanism that could, at least in principle, support this kind of evidence generation, though they note it has been used far more often for devices than for diagnostics.
The global angle is also worth noting. The authors do not treat this as a purely American coding or Medicare problem, even if many of the examples are US-facing. They point to international harmonization efforts such as the International Medical Device Regulators Forum and to Europe’s Health Technology Assessment Regulation as signs that at least some jurisdictions are beginning to think more systematically about how diagnostics fit into innovation policy. Still, the paper suggests that most health technology assessments remain far more comfortable pricing the drug than valuing the information that makes the drug usable.
For policy insiders, the paper’s real significance may lie in how directly it tries to reposition diagnostics. This is not an argument that tests are ancillary tools deserving modest technical cleanup. It is an argument that diagnostics should often be treated as essential components of therapeutic strategy, with regulatory, evidentiary, and reimbursement pathways built accordingly. That is a bigger claim, and a more disruptive one, because it challenges the longstanding habit of letting therapeutics define value while diagnostics justify themselves piecemeal.
In that sense, the paper reads as both a diagnosis and a warning. Precision medicine can keep producing increasingly sophisticated drugs, but if the tests needed to identify the right patients remain stuck in lower-status regulatory and payment channels, the field risks building a treatment paradigm whose most expensive components are better supported than its decision tools. For a system that prides itself on targeted care, that is an awkward contradiction.
