Federal Data Is Disappearing: What the Tracker Reveals

At least 28 federal datasets have been deleted and another 338 modified over the past 18 months, according to an online tracker released by dataindex.us, a group of data policy experts, researchers, and developers formed to monitor changes to government data. The Guardian published its latest investigative interactive on the findings on August 18, 2026, detailing deletions and alterations spanning more than 60 federal agencies across the environment, labor, law enforcement, and health domains. (The Guardian)

Health datasets are the most affected, making up roughly 40% of all terminated or altered data collections tracked by dataindex.us. The changes follow a pattern. Many align directly with executive actions, particularly the order titled "Defending Women from Gender Ideology Extremism and Restoring Biological Truth to the Federal Government," which led multiple federal surveys to stop collecting information on queer and trans populations. The CDC removed gender identity questions from at least five instruments: the National HIV Surveillance System, the National Violent Death Reporting System, the CDC Diabetes Prevention Recognition Program, the National Intimate Partner and Sexual Violence Survey, and surveys evaluating PrEP (a medication that reduces the risk of HIV infection) uptake among gay men and in communities of color. (The Guardian)

At least four health datasets are no longer being updated at all. Among them is the Substance Abuse and Mental Health Administration's Drug Abuse Warning Network (DAWN), a surveillance system that tracked drug-related emergency department visits in near real-time. Its discontinuation leaves a gap in the ability of public health authorities to monitor overdose trends as they unfold. (The Guardian)

The tracker catalogues a striking breadth of affected resources beyond health. Examples include a database of federal law enforcement officer misconduct records, a survey measuring physical, sexual, and emotional violence against children and youth, and real-time pollution readings from US embassy air quality monitors worldwide. The State Department's China travel advisory page, for instance, notes the use of AI, historical data, and global pollution modeling to generate three-day air quality forecasts, a function that depends on the kind of monitoring infrastructure now disrupted. (The Guardian; State Department)

Denice Ross, who served as US chief data scientist under the Biden administration, is a cofounder of dataindex.us. The organization was formed the previous year specifically to fill the monitoring vacuum left when federal datasets began disappearing. (The Guardian)

The data attrition began almost immediately after Trump's second inauguration in January 2025. By February 2025, a federal judge ordered US agencies to restore public access to health-related webpages and datasets removed to comply with administration directives. Agencies restored several pages the following day. The New York Times reported that more than 8,000 government web pages had been removed in the aftermath of the presidential transition, while nearly 3,400 datasets had been pulled from Data.gov alone as of February 22, 2025. (AP; Reuters; University of Maryland Libraries)

Marketplace characterized the removals as a "targeted, surgical removal" of datasets not aligned with the administration's priorities. AP reported that datasets were altered or disappeared from government websites in unprecedented ways after Trump began his second term. The pattern extended to data connected to an executive order rolling back protections for transgender people, with administration officials deleting material from government websites and electronic publications. (Marketplace; AP; AP)

The Guardian has been building its "Deleted data" investigative series since at least May 2026. Its initial interactive on the subject, published May 7, 2026, laid out five ways data deletion harms Americans, from infant mortality to hunger. The series formally launched a week later. In June, the Guardian reported that the National Data Service (NDS) removed website visitor-tracking software after the newspaper contacted the White House with a detailed series of questions about the practice. (The Guardian; The Guardian; The Guardian)

Civil society responses have mounted in parallel. The Environmental Data and Governance Initiative (EDGI), which uses volunteer analysts to track changes to roughly 25,000 government webpages across multiple agencies, has been monitoring federal data integrity since Trump's first term. The Harvard Law School library took steps to preserve government data amid the early 2025 purges. Academic institutions including Occidental College and the University of Maryland have maintained data rescue guides cataloguing affected resources. (The Guardian; Reuters; Occidental College; University of Maryland Libraries)

The broader context here matters for anyone who relies on federal data for operational, research, or policy purposes. When a near-real-time surveillance system like DAWN goes dark, the loss is not merely archival. It degrades the signal that emergency rooms, epidemiologists, and municipal health departments depend on for situational awareness during overdose surges. Think of it like a weather radar going offline right before a storm: the data was needed in the moment, not after the fact. When CDC surveys stop collecting gender identity data, researchers lose the ability to stratify health outcomes for transgender populations at all, not just for the duration of this administration. The gap in the longitudinal record (data collected over time to track trends) is permanent. Similarly, when embassy air quality monitors stop reporting, countries and global health organizations that rely on that data, particularly in regions lacking their own monitoring infrastructure, lose a comparative baseline that cannot be retroactively reconstructed.

What sets the current moment apart from earlier episodes of federal data modification is the combination of scale, systematic targeting of specific demographic data, and the apparent coordination with executive orders. Prior administrations altered or removed web content, but the dataindex.us tracker, by documenting 28 deletions and 338 modifications across more than 60 agencies, provides a level of detail that earlier monitoring efforts could not. The judicial pushback in early 2025 secured the restoration of specific health datasets, but the trend since then has been toward more, not fewer, alterations. Researchers, journalists, and data stewards who depend on federal datasets as primary sources now face an environment in which availability, content, and collection methodology can shift without notice, and in which independent tracking infrastructure has become essential rather than supplementary.