In January, we learned of an emerging threat in China, a variant of the frightening SARS virus, but worse—more infectious. Since then, we learned China, the first to both see and successfully manage its outbreak, seriously under-detected the disease. According to a widely circulated computer model study, true cases were probably around seven times greater than the official count. Calculations published a month later lent further support for this view.

Since a computer model relies on inputs and assumptions, however, it is worth questioning. But there are other paths to the same metric. For instance, the case fatality rate of COVID-19 is probably between 0.5% and 4%, and most estimates from populations that are contained (i.e., cruise ships) or widely tested (South Korea, Iceland) suggest the true number hovers around 1%, maybe less.

In my home state of Louisiana, the number of deaths officially reported due to COVID is now about 2,300. This means, if the case fatality rate is 1%, or one fatality for every hundred cases, total cases in Louisiana should number roughly 230,000. The confirmed case count? 33,837, about six times lower. And Louisiana has tested more of its people than any other state, suggesting these numbers may be closer to truth than in other states. But even this estimate, which depends on official COVID deaths, is guaranteed to be low since death surveys and post-mortems are routinely missing COVID.

More frightening still, a pre-publication study reported antibody responses in a California county where documented COVID-19 cases have numbered about 1,000. The prevalence of antibody findings among a sample of people in the county, however, suggests the true number of infections may be 50 times higher.

Bottom line: Official counts are way, way low.

And one more troubling twist: The test used for virtually all tallies is the polymerase chain reaction, or PCR, swab. To know true case numbers, we need to factor in the accuracy of the PCR, which misses COVID — a lot. Even the most rose-colored estimates suggest roughly a third or more of active infections are missed. One Chinese health official estimated 50 to 70% of cases there were being missed by the swab. PCR is so uniformly insensitive Mayo Clinic biostatisticians recently published a paper estimating the test’s potentially cataclysmic impact on the pandemic. Even their most conservative predictions are jaw-dropping and point to an urgent need for a new approach to diagnosing the disease.

So first, tack on 50-100% to every number of cases detected by the test. Then, multiply by six. Or seven. Or maybe 50.

Finally, keep in mind official counts are a political football, with incentives aligned to keep numbers low. On March 6, President Trump became the rare world leader willing to admit this publicly. Explaining why he preferred not to allow 21 people sick with COVID-19 to disembark from a cruise vessel, he said: “I don’t need to have the numbers double because of one ship.” Ironically, at that point, there were already thousands of unreported cases in the U.S.

Official case and death counts are wrong. Skewed by access to testing, accounting, and politics, at best, they reflect an iceberg tip. Accepting these numbers—or the headlines that tout them — would be like seeing a roach in your kitchen and believing it was the only one. Nobody knows the true numbers yet, but we know one thing for sure: What we see from the test doesn’t even begin to tell this story.

So to media of all stripes, it is time to fix your COVID language. In this pandemic, the reality is bigger than any headline.

Much bigger.

Daniel Hopkins is a physician.

Image credit: Shutterstock.com