WADA 2026: The Testing Curve Is Slowing, and 45% of Failures Sit in Sample Collection

On my desk in Nha Trang, the WADA annual report opens at page seventeen. The bar chart runs from 2026 to 2026. The 2026 column jumps 12.5% over the previous year. The 2026 column drops to 3.2%. Same series, same unit of measurement, but the slope has reversed. Two days after publication, almost every news item I read repeated exactly one headline: the world anti-doping body is expanding sample testing. Nobody wrote about the slope.

I make my living reading slopes. Numbers never lie, but they know how to hide. An absolute total can conceal a cooling trend. A rate that holds flat for three straight years can conceal a process cracking underneath. This year's WADA report carries both kinds of camouflage, and they sit side by side on the same page.

Context: a governance report, not an athlete file

WADA coordinates global anti-doping rules, comprising the World Anti-Doping Code and its accompanying International Standards. Its annual report is not a story about an athlete, a specific violation, or a lane in a pool. It is an operations ledger: sample volume, adverse analytical finding rates, compliance audit results, finances, and the progress of investigation campaigns. In other words, it reports on the machinery, not on the swimmer.

For Vietnamese readers, the distance between such a document and a domestic swim meet sounds large. It is not. Every Vietnamese swimmer competing in events governed by World Aquatics sits inside this system: subject to out-of-competition testing, required to file quarterly whereabouts, and overseen by a national anti-doping organization. If that machinery is weak at the collection stage, the person who ultimately carries the risk is not an official. It is the athlete.

Based on my experience tracking matches, major championships and domestic league datasets, I always read a governance report in two passes. First pass, I record the numbers. Second pass, I look for where the numbers fail to match the wording. I rebuild the series in my own spreadsheet, set a significance threshold before seeing results, and only then allow myself to speak. This report has at least four places where the numbers and the wording do not travel together.

One timing note before the detail, because it matters for every citation that follows. The document is called the 2026 report, but the testing data inside is largely 2026 and 2026 activity, while the financial section is stated by 2026 fiscal year. The chart series runs 2026 to 2026. This may be a one-year publication-lag convention, or loose labelling. For anyone working with data, that is a check-before-citing flag, not a trivial detail.

The core: four evidence chains

Testing volume is still growing, but the growth rate has fallen nearly fourfold in two years. In 2026 the system recorded 288,865 samples. In 2026 the figure was 297,965, an increase of roughly 9,100 samples, or 3.2%. But the prior leg, 2026 into 2026, grew 12.5%. Same growth series, but the pace dropped from double digits to single digits. Read only the total of 297,965 and the story is more. Read the slope and the story is more, but steadily less more.

For anyone who works markets, a decelerating slope inside a budget-driven system rarely comes from running out of work to do. It comes from running out of resources to do it with. And the financial section of this very report confirms that suspicion, as I set out below.

The adverse analytical finding rate has stood essentially still for three years. The figures are 0.77%, 0.80% and 0.78%. The band of variation is 0.03 percentage points. This is the most robust series in the entire document, and also the easiest to misread.

I have to be explicit here, because many anti-doping write-ups confuse two concepts. The adverse analytical finding rate is a base rate of the testing process. It measures the share of samples returning an adverse result against all samples analysed. It does not measure the share of athletes using prohibited substances. The two quantities coincide only when samples are drawn fully at random from a homogeneous population, and that does not exist in practice. A better-targeted system can collect fewer samples and hold the positivity rate steady. A scattershot collection system can push the rate artificially low.

So when I read that positivity held near 0.78% for three years, I do not write that world sport is stably clean. I write that the system is sampling with a near-constant targeting structure. A rate flat for three years is evidence about process stability, not evidence about the cleanliness of the population.

Compliance audits show a clean picture at the top and a crack at the bottom. The report records average non-conformities per audit falling from 25.5 to 20. That sounds like clear progress. But read it with the sample size attached: that figure is drawn from nine audits. Nine. In a system with more than 60 countries in the audit pool and hundreds of national anti-doping organizations, nine audits is far too thin a sample to establish a trend. Statistically, the improvement may be sampling noise.

Meanwhile, the composition of findings is far more troubling. Testing-related non-conformities account for 45% of all findings, up 11% year on year, and 23% of those are classified as critical. In other words: total findings fell, but failures in the single most important stage, sample collection and sample handling, rose in both share and count.

This is the kind of anomaly I hunt for in any dataset. An aggregate indicator moves down while a component indicator moves up. Look only at the aggregate and you conclude the system is improving. Look at the component and you find the leak sitting exactly where a lost sample or a breached procedure can render a result unusable as evidence.

For swimming, the consequence is indirect but real. A urine sample collected under the wrong procedure, a broken chain of custody, a mis-filed whereabouts declaration: any of these can turn a blameless athlete into a complicated file, or let a genuine cheating case drift past. Collection quality is not paperwork. It is the foundation of fairness in the water.

Finance: an $8.3 million shortfall and two suspended contributions. The report shows revenue $8.3 million below budget. Of that, the United States has not paid $3.8 million and Russia owes $3.9 million. Offsetting that is a $1.2 million net surplus, indicating tightened spending or deferred plans.

These two suspended payments are not merely accounting. They come from the nations with the greatest political and sporting weight in the system. A global anti-doping body whose revenue depends on government goodwill will always risk being dragged into negotiations that have nothing to do with a urine sample.

Set side by side, two data points, the testing growth rate falling from 12.5% to 3.2% and the $8.3 million shortfall, form a hypothesis with real weight: the cooling slope is not because the system ran out of work, but because it hit a resource ceiling. That is my inference, not the report's conclusion. But it is built on two independent data clusters pointing the same way.

The genuine bright spot: the pivot from testing to investigations. The report records more than 250 investigation operations, 88 laboratories dismantled and over 90 tonnes of substances seized. This is the strongest data in the document, and it represents a strategic shift: from primarily testing athletes to investigating supply and distribution networks.

Technically, it is the right direction. One positive sample catches one athlete at one moment. One dismantled laboratory cuts supply to dozens or hundreds of athletes across years. Forced to spend the same budget dollar one way or the other, I lean to the second, but only on the condition that baseline sample numbers do not fall below the threshold needed to catch isolated cases. A dense investigations net with a sparse testing net still lets the most sophisticated actors through.

And the most important calendar marker: the 2027 Code. WADA adopted the revised World Anti-Doping Code and Standards in Busan, South Korea. The text takes effect on January 1, 2027. Ahead of that, the organization and its Athlete Council ran a consultation process with more than 600 athletes and commissions across more than 60 countries and more than 70 sports.

Any federation can plan around this. The effective date is known. The requirements are known. The period from now to January 1, 2027 is a compliance gap, during which national anti-doping organizations can keep operating on old habits while waiting for new standards. In a system where the main driver of behaviour is audit risk, that gap is not neutral.

One more point about the 600-plus athletes and 70-plus sports. A consultation of that scale signals an effort to build legitimacy. But the report offers no evidence that athlete input changed the substance of the Code. A broad consultation with no measurable outcome is good communications, not necessarily good reform. That is where I place a question mark, and I leave it standing.

The contrarian angle: stable does not mean clean

If there is one mistake I see repeated in commentary, it is turning a stable rate into a moral proof. Positivity sits at 0.78% for three years, and two opposite readings appear instantly: optimists say the system has the situation under control; sceptics say the number proves the system is not catching anyone. Both readings fail in the same place. Both treat a procedural base rate as a measure of cheating.

I learned this lesson long ago, and I learned it in a different sport. Germany 2026 did not collapse through bad luck. PPDA told us in the group stage. When Germany controlled 74% of possession against South Korea and lost, most viewers called it an accident. But a PPDA of 13.2 showed they allowed opponents 13 passes before pressing, a signature of sluggish movement, not of misfortune. Possession share is a seductive number about a feeling of dominance, but it does not measure danger. Likewise, a positivity rate measures the activity of the testing net, not the prevalence of substance use.

The real concern in this report is not the positivity rate. It is the 45% testing non-conformity share, up 11%, with 23% critical. A collection system failing at the gathering stage generates two error types at once: catching the innocent and releasing the guilty. Both destroy the system's legitimacy, and neither appears in the headline.

At the same time, the 3.2% growth rate should itself be read as a capacity signal, not a demand signal. A system that expanded at 12.5% and fell to 3.2% in two years has usually hit some kind of ceiling: budget, personnel, or laboratory infrastructure. And with budgets $8.3 million short, the budget ceiling is the leading candidate.

People look at the price table, I look at the curve. Many transfers die before they are announced. Here too: a system can announce 9,100 more samples in a year while its real capacity began contracting earlier.

Finally, there is one detail I consider the biggest blind spot: this report does not break data down by sport. It does not tell me how many of those 297,965 samples were swimming, how many were out-of-competition, and how many came from thinly resourced federations. For anyone working with data, an aggregate report without sport-level breakdown is not sufficient to draw sport-level conclusions. I refuse to draw conclusions about swimming from unsegmented data. That is a professional rule, not caution.

What to watch in the next cycle

A team does not collapse in one night. It collapses when the indicators stop connecting to each other. Here, the indicators have not snapped, but they are drifting apart: volume still rising, positivity still flat, collection compliance worsening, and resources suspended at the two largest line items. Those four curves have not crossed yet. They only need to cross once.

I will track five signals over the next 12 to 24 months. First, the testing volume growth rate in the next report: if it reaches 0% or turns negative, that is evidence of capacity erosion, not of running out of work. Second, the positivity rate across the 2026 and 2026 series: if it breaks above 0.85%, the question becomes a change in targeting, not a new doping wave. Third, the testing non-conformity share: if it stays above 45%, it is a systemic collection problem. Fourth, the contribution status of the United States and Russia: every deferral or settlement directly changes operating capacity. Fifth, national anti-doping organizations' readiness for the 2027 Code.

Once again I return to my old story. COVID closed the stadiums, so I reopened the V-League directory. No league is meaningless. Vietnamese swimming may not appear on the front page of the WADA report, but it is in there, on a line of unsegmented data. And as long as a young athlete has to believe the scales of fairness are working, these seemingly dry numbers remain our business.

Luck is something I do not have. I have probability and data thick enough to stand on. And this time, the data thick enough to stand on is saying something different from the headline.