We Keep Finding What We Decide to Look For

When the Centers for Disease Control and Prevention (CDC) released its SHIELD project data this summer, it did something routine surveillance never had. It quantified where HIV testing and prevention break down for people already in contact with the healthcare system. Among SHIELD participants who tested for other sexually transmitted infections in the year before their HIV diagnosis, 33% of those diagnosed early and 52% of those diagnosed late were not offered an HIV test at the same visit. That gap was invisible to case surveillance until CDC built an instrument to see it. SHIELD is what directed surveillance makes possible: point it at a stage of an epidemic that routine data leaves dark, and that stage lights up.

The same logic governs hepatitis C. We already have an instrument pointed at part of that epidemic, and it too shows precisely where a curable disease stops reaching people. Both instruments were the product of a choice to measure. Both also carry blind spots, and those blind spots are not accidents of design. They mark the stages no one has yet funded anyone to instrument.

The Cure Works. The System That Delivers It Doesn't.

Hepatitis C is curable. An 8-to-12-week course of direct-acting antiviral (DAA) treatment produces sustained viral clearance in at least 95% of cases. The medicine is not the problem, and stating its effectiveness plainly is what makes the delivery numbers land as an indictment rather than a lament.

CDC's state-specific HCV clearance cascades trace the sequence from a hepatitis C virus (HCV) diagnosis through to viral clearance. The diagnostic machinery works: across states, a median of 91% of people ever infected received RNA testing, reflecting the widespread adoption of reflex RNA testing on reactive antibody specimens. What happens after diagnosis is where the system fails. The median proportion of people with current, RNA-detectable infection who went on to clear the virus was 29%, with a national average of 35%, ranging from 10% in West Virginia to 51% in Connecticut. Every jurisdiction fell below the U.S. Department of Health and Human Services goal of 58% clearance by 2025, and all of them sit far below the 2030 target of 80%.

The national analysis using commercial laboratory data found the same shape: 88% of people ever infected received RNA testing, and among those with current infection, 34% showed evidence of being cured or cleared. Each figure is a share of the step before it, not of the whole population. What fails between a working diagnostic step and a curative treatment is the delivery apparatus, underfunded programs, coverage restrictions, and missing data infrastructure. The medicine already did its part.

What the Cascade Cannot See

The cascade is a real instrument, and like any instrument it has a field of view with edges. CDC named those edges itself. In its methodology guidance for building laboratory-based clearance cascades, the agency catalogued five limitations, and each one maps onto a stage of the epidemic that remains unmeasured.

Screening coverage is invisible. Because nonreactive HCV antibody results are not reportable in most jurisdictions, the cascade "is unable to assess the number of people who have been screened for HCV" and must begin at diagnosis. Treatment is invisible. Without treatment data, the cascade cannot separate spontaneous clearance from treatment-induced cure, which is why CDC deliberately titles it a clearance cascade rather than a cure cascade. Reinfection is indistinguishable from treatment failure; the agency cautions that its final step "should not be interpreted as the proportion of people with reinfection." Equity cannot be measured, because laboratory results "rarely include data on race and ethnicity," and people who move or die outside a jurisdiction are difficult to track. Finally, the instrument is incompletely built even where funding exists. Among CDC-funded health departments surveyed, only 26% received all undetectable HCV RNA results and 17% received none, while just 47% maintained a longitudinal registry for chronic HCV.

CDC's own position is that the cascade is fit for its purpose, tracking linkage to treatment, and the agency notes that the treatment-data gap "has little importance at a population level." The point here is narrower and precise. By CDC's own account, the cascade cannot yet measure screening coverage or disparities. Those are exactly the functions a directed instrument would add. The cascade's blind spots are the same category of failure that HIV case surveillance carried before SHIELD, blind to who was screened, blind to who fell out of care and why. SHIELD closed that gap for one stage by aiming a study at it. The cascade's limitations are a map of the work still undone.

The Instruments Exist, and So Does the Method

We are not theorizing about whether directed surveillance works. The Medical Monitoring Project (MMP) already does for HIV care what SHIELD did for diagnosis. MMP is a nationally representative survey of adults with diagnosed HIV, built specifically to measure the barriers to care and viral suppression that outcome data cannot explain. Its 2023 findings quantify the conditions shaping those outcomes: 20% of people with diagnosed HIV experienced homelessness or unstable housing, 20% faced food insecurity, 34% lived below the poverty line, and 24% of those who needed mental health services did not receive them.

SHIELD instrumented the Diagnose stage. MMP instruments the barriers behind treatment of HIV. The directed method has now proven itself on both the front and the middle of the continuum. What it has not been pointed at is the rest.

Three Pillars, One Proven Method

The Ending the HIV Epidemic (EHE) initiative rests on four strategies: Diagnose, Treat, Prevent, and Respond. Reading them against what we can currently measure reveals where the method has been used and where it has not.

Diagnose has its instrument; SHIELD mapped the front-end barriers. Treat is well covered. The HIV care continuumshows that among the more than 1.1 million people with diagnosed HIV at year-end 2024, 77% received some care, 56% were retained in care, and 69% achieved viral suppression; MMP names the reasons behind those gaps, and the HCV cascade documents the parallel treatment failure for that disease.

Prevent is where the method is only half applied. A directed instrument already measures part of it: National HIV Behavioral Surveillance (NHBS) surveys the populations at highest risk for HIV in rotating cycles, men who have sex with men, people who inject drugs, and heterosexuals at increased risk, with a separate cycle covering transgender women, and it tracks PrEP awareness and use along with receipt of prevention services. Its 2024 data on people who inject drugs show the shape of the failure: among those without HIV, 43% were aware of PrEP and 2% were using it. AIDSVu's 2025 data map the same inequity by race and sex, with Black people making up 39% of new HIV diagnoses in 2024 but only 15% of PrEP users in 2025, and women accounting for 20% of diagnoses but 10% of users. What no standing national instrument captures is the reasons people who need PrEP never start it. SHIELD asked people who went undiagnosed why they had not tested; no equivalent surveillance asks people at risk why they have not begun prevention. That question is left to individual academic studies rather than a directed system built to answer it at scale.

Respond is thinner still. Coinfection analysis shows where HIV and hepatitis C converge among people who inject drugs, and CDC has operated pieces of an answer, NHBS and an injection drug use surveillance pilot among them. It also funded a national survey of syringe services program capacity through a five-year cooperative agreement, the kind of instrument that maps where prevention services reach people who inject drugs and where they do not. That agreement's funding page now sits archived, and it runs against a federal retreat from harm reduction that has already narrowed what harm reduction supplies federal grants may purchase and prompted modeling of the additional overdose deaths that funding cuts would produce.

What has never existed is integrated surveillance linking HIV and HCV data for the coinfected population. A five-state assessment of data capacity found no state able to build a combined care cascade across these conditions. The instruments run in parallel, rarely connect, and the ones we built are not guaranteed to survive the budget cycle.

The Constraint Is a Choice

What stops us is money, and the way we allocate it reveals what we have decided to see. A seven-jurisdiction study of HIV and HCV surveillance capacity found that HIV programs ran a median of 4 full-time-equivalent staff while HCV programs ran 1.25, more than three times fewer. Standardized by disease burden, the disparity widens to sixteen-fold: 0.37 versus 0.023 staff per 1,000 people affected. The study is descriptive and covers a small number of jurisdictions during the COVID-19 pandemic, which diverted surveillance staff, so it cannot establish that staffing caused outcomes. Its own authors present the link between the best-staffed program and its stronger results as suggestive. The structural point survives those caveats. Most jurisdictions already build and post HIV care cascades and could add HCV cascades to routine reporting; the enhanced HIV/AIDS Reporting System already anchors HIV surveillance. The instruments are shared. What differs is investment. The distance between what we can see and what we actually measure tracks who we fund to look.

What We Should Push For

The method is validated. Converting it into a funded mandate requires four specific actions.

First, fund directed barriers surveillance across every pillar. SHIELD and MMP have proven the approach; Congress and CDC should sustain both and extend the model to Prevent and Respond rather than treating each effort as a one-time pilot. Second, make the negatives reportable. State and federal policy should require reporting of nonreactive antibody and undetectable RNA results, so the cascade can finally measure screening coverage and clearance rather than beginning at diagnosis. Third, integrate and staff HIV and HCV surveillance together. Closing the program staffing gap and dismantling the silos between them would let shared infrastructure do double duty. Fourth, protect demographic and community-led data collection. Equity cannot be measured on data that omits race, ethnicity, identity, and residence, a point CANN has made before.

We Cannot Close Gaps We Refuse to Measure

The method works, the blind spots are documented, and the only remaining barrier is the will to fund and protect the act of measurement. That will is now in question. On September 2, 2026, the President signed H.R. 6500, the Continuing Appropriations and Extensions Act, 2027, a short-term continuing resolution that funds federal agencies only through December 11, 2026. Decisions about prevention and surveillance funding are deferred, not resolved, and the December deadline will force them back into the open.

We built the map. We know where it goes dark, and we know how to light those places, because we have already done it. Whether the map stays legible is a choice being made right now, in a budget process that treats the capacity to see as optional. We should insist otherwise.