Skip site navigation

All major indicators of COVID-19 transmission in the United States continue to fall this week. Nationally, cases have been falling for six weeks, hospitalizations have been dropping sharply for five weeks, and deaths have been declining for four weeks. The average number of people in the hospital with COVID-19 this week is just under the average we saw during the high points of the April and July hospitalization surges. States and territories reported 14,463 COVID-19 deaths in the past seven days, the first weekly total under 15,000 we’ve seen so far in 2021.

The decline in cases and deaths appears to have slowed this week. After so many weeks of very rapid improvements in reported cases and deaths, a slowing of those declines is inevitable. But we don’t think that a true plateau is the only or even likeliest explanation for what we’re seeing this week. Instead, we think we’re once again seeing the equivalent of a holiday reporting effect

Two confounding events that occurred last week—President’s Day and the beginning of a major winter storm that knocked out power for millions of Americans—resulted in slowdowns in case reporting. As we’ve seen with other holidays and storms, these kinds of disruptions produce a predictable series of reporting artifacts: first an artificial drop, then an artificial rise. We should always look for confounding factors before interpreting the data as suggesting a change in the direction of the pandemic—and in this week’s data, we found them. Artificially low numbers from last week’s disruptions make this week’s case and death declines look smaller by contrast, and make the daily numbers look as though they’re reversing direction, especially as backlogs roll in.

The death reporting process was also affected by the federal holiday and the storm, producing similar artifacts in the data. Even hospitalization numbers appear to have wobbled in a similar way, probably as a result of both small reductions in reporting and small reductions in the number of people who sought care at hospitals over the three-day weekend and during major regional storms.

We’ve also seen large numbers of deaths from much earlier periods reported this month, including a backlog of more than 1,500 deaths in Indiana reported February 4, nearly 4,500 older but previously unreported deaths in Ohio reported between February 11 and 13, plus a smaller addition of deaths in Virginia (total size still unknown), which the state notes is due to processing death certificates from the post-holiday (January) surge.

This brings us to a crucial point that news summaries frequently get wrong: the deaths states and territories report on a given day do not represent people who died on that day. Reported deaths lag behind cases by two to three weeks on average, and many reported deaths actually took place substantially earlier. When reported cases rose during previous surges, deaths lagged weeks behind. The same is true now, as cases decline. 

Why this matters: We have every reason to believe that far fewer people actually died of COVID-19 this week than in previous weeks, because cases and hospitalizations continue to drop. But we won’t see those smaller death numbers for weeks to come—and probably for more than two or three weeks, as previously overwhelmed public health officials are able to catch up on processing death certificates.

Deaths and cases are down in LTC facilities, but state data dumps are increasingly muddying the data

The backlogs in reported deaths have also affected our numbers for nursing homes and other long-term-care facilities: Indiana added 659 historical resident deaths and one staff death to its cumulative total for the week ending February 17, and Ohio added 1,150 historical resident deaths. We can chart a national trajectory of deaths in long-term-care facilities despite these and other recent large additions by working with the data from the 52 jurisdictions that have not included major backlogs and reassignments in recent months. 

The resulting visualization is an incomplete representation of deaths in LTC facilities in absolute numbers, but it allows us to understand the national trend: weekly deaths in long-term-care facilities continue to decline and are down 69 percent from their January 14 peak.

If we view deaths in long-term-care facilities in these 52 jurisdictions as a share of COVID-19 deaths in the US, we see that the percentage of the country’s COVID-19 deaths that are occurring in these facilities also continues to decline.

The New York Times also recently analyzed trends in nursing home cases and deaths in relation to national case and death figures.

When our project ceases data compilation on March 7, the only comparable public federal dataset will be the CDC/Centers for Medicare and Medicaid Services dataset on COVID-19—which is only partially comparable, as it includes only nursing homes and not other long-term-care facilities, such as assisted living or independent living facilities. This nursing home data has been reported weekly by facilities to CMS since May 17, 2020 through the CDC’s National Healthcare Safety Network. Cases and deaths are reported as cumulative and weekly totals, and some facilities include cumulative data dating back to January 1, 2020. We’ll be writing more about that in next week’s additions to our series of trainings on federal COVID-19 data.

Inequities persist in vaccination efforts, and vaccination data remains inadequate 

Although COVID-19 has disproportionately harmed Black, Latinx, and Indigenous people in many areas of the US, it appears that vaccinations are not being proportionately administered to many of these communities. The Verge visualized vaccine administration in Chicago, New York City, and Washington, DC, and found substantial mismatches between areas with the highest case or death rates and areas with highest vaccine administration. A Los Angeles Times investigation likewise found that the areas hardest hit by COVID-19 in and around Los Angeles currently have the lowest vaccination rates. Inequities in vaccine distribution and administration result from many factors, including inaccessible systems, state-level decisions, fixed age cutoffs, and problems with people actively circumventing the processes established. 

Federal vaccine data includes race and ethnicity for only slightly more than half of all reported doses, isn’t broken down by state, and doesn’t allow for crucial age-race cross-comparison. Data from the 40 states that publicly report race and ethnicity data for vaccinations, compiled by KFF and others, tracks vaccinations by race and ethnicity, but most states do not include age breakdowns, and some are missing race and ethnicity data for more than 30 percent of vaccinations. Reporting at the metro level and rough analysis within states has given us plentiful warning that vaccine rollouts are currently missing many of the communities at the greatest risk from COVID-19. Without better data about who has been vaccinated across all US jurisdictions, we will not be able to identify and address the full range of inequities in the rollout.

Learn about using federal COVID-19 data

We have now published user guides for federal COVID-19 hospitalization data, case data and death data, as well as a deep dive on inconsistencies between state and federal testing data. Our final free training session on federal data will cover long-term care and race and ethnicity datasets, and will take place next Thursday, March 4, at 2pm ET/11am PT. You can sign up here to receive an invitation.


Mandy Brown is editorial lead at The COVID Tracking Project. She previously served as VP of Product at Vox Media.


Artis Curiskis is outreach & reporting co-lead at the COVID Tracking Project and collaboratively runs the CTP special projects Long-Term Care COVID Tracker and City Data.


Alice Goldfarb leads The COVID Tracking Project’s part in The COVID Racial Data Tracker, and is a Nieman Visiting Fellow.


Erin Kissane is a co-founder of the COVID Tracking Project, and the project’s managing editor.


Kara Oehler is outreach & reporting co-lead at the COVID Tracking Project and collaboratively runs the CTP special projects Long-Term Care COVID Tracker and City Data.

Headshot_JMR 0420.png

Jessica Malaty Rivera has an MS in Emerging Infectious Diseases and is the Science Communication Lead at The COVID Tracking Project.


Peter Walker is Head of Marketing & Growth at PublicRelay and Data Viz Co-Lead at The COVID Tracking Project.


Related posts

More Good News: This Week in COVID-19 Data, Feb 18

Case numbers are dropping rapidly, and deaths are down 20 percent from last week. Vaccines are finally showing up in the data with dramatic declines in deaths from long-term-care facilities. We urge caution about interpreting wobbles in the data in the next week or two, given storm disruptions affecting Texas and elsewhere.

The Third Surge Continues to Ease: This Week in COVID-19 Data, Feb 11

Another week of good news: Cases and hospitalizations continue to drop nationally, and deaths are down for the second week in a row. We’re concerned about ambiguous indicators in the Northeast, and about testing declines.

The Numbers Are High but Improving: This Week in COVID-19 Data, Feb 4

Cases and hospitalizations continue to drop, and now COVID-19 deaths finally appear to be declining.