A new study puts a number on that gap.

Researchers, including Stanford epidemiologist Mathew Kiang and UCSF physician Kirsten Bibbins-Domingo, estimate that more than 150,000 COVID-19 deaths in the United States likely went unrecognized in 2020 and 2021 — never identified as deaths caused by the coronavirus on death certificates

The study, published Wednesday in Science Advances, the study puts the true death toll for that period at 995,787 — well above the 840,251 deaths officially recorded.

That’s roughly one missed death for every five that made it into the official count.

California accounts for an estimated 11,613 of those likely unrecognized deaths, one of the highest totals in the country, behind only Texas and New York.

Many of those deaths happened out of sight of the health system. The researchers estimate more than 111,000 occurred at home — where testing was inconsistent and, early on, often unavailable.

In California, limited testing and long delays meant many families never received a diagnosis. Instead, death certificates often listed heart disease, diabetes or dementia — conditions COVID can worsen or set off — rather than the virus itself.

The undercount was not evenly distributed.

Unrecognized deaths were more likely among people identified on death certificates as Hispanic, Black, Asian, American Indian and Alaska Native, and among those with less education and lower incomes. The study found these patterns at the county level but did not identify specific cities.

“This underreporting that we found wasn’t random,” Kiang said in comments to Scientific American. “Pretty systematically, what we found was that communities in areas that were most impacted by the pandemic were also the ones with the most unrecognized COVID-19 mortality.”

California officials saw pieces of this in real time — outbreaks in essential-worker communities, higher death rates in poorer neighborhoods — even as the official numbers lagged behind what hospitals and families were experiencing.

The authors caution that their model cannot definitively prove every missed case was COVID-19. But the findings suggest the official toll likely undercounted the pandemic’s true scale.