As the coronavirus continues to sweep across the United States, one question is on everyone’s mind: Where are different communities in the cycle of the pandemic, and how bad will it get? To find answers, state governments and research groups are now collecting detailed testing and hospitalization data that can help assess how each state is performing when it comes to detecting and addressing COVID-19. Which states are doing well? Where might we see future hot spots? In this series, we interpret the data in more detail, one state at a time.
By Liana Woskie and Tynan Friend, HGHI
Estimates for the arc of COVID-19 in California are optimistic: The state is projected to get through the pandemic relatively unscathed. Accounts from clinicians and hospital leadership have also been positive. CA is not observing patient overloads like those seen in other large states. And last week, the Surgeon General heralded the state a success; an uplifting example of “bending the curve” amidst other, much more somber, press briefings.
The Bay Area’s efforts to curb the disease especially, such as ordering people to avoid crowds and shelter in place, took place early and appear to be paying off. But as we take well-founded comfort in this news, a look past state-level averages sheds light on a more nuanced story: while San Francisco is faring remarkably well, Los Angeles may still be struggling.
If we look at the state-level numbers, how does California compare? According to the latest data from The COVID Tracking Project, California has nearly 17k cases, behind New York, New Jersey, Michigan, and Louisiana in terms of overall reported cases. Given the state’s size, however, this isn’t cause for alarm – by population, CA has approximately 429 cases per million, which is lower than most. These summary numbers, however, mask wide variation at the county level. If you look at within-state variation you can see that reported cases in LA are over 6,500 (and increasing) while San Francisco has just over 2,500 overall with less change from day-to-day. Deaths due to COVID-19 are also higher in LA than San Francisco at 16.7 v 13.5 deaths per million respectively.
Why is the Bay Area doing well? As noted above, one reason may be that California has been a leader in promoting social distancing measures. San Francisco enacted a shelter in place order on the 16th of March, before many other states. LA also enacted a stay at home order on the 19th of March. How closely these guidelines have been followed, however, is another question and may explain some of the within-state variation. When you look at the different counties within CA, you can see that social distancing has varied significantly in the state’s cities over time. San Francisco is doing well, with low encounters per-person and locations visited, but these numbers have been more concerning surrounding LA. Fortunately, LA Mayor Eric Garcetti made strong public statements last week, highlighting the need to stay home, to wear (non-medical) masks if going out, and to refrain from visiting non-essential businesses. We are beginning to see safer social distancing practices around LA, which will hopefully continue to improve in the coming weeks.
What is concerning for the state as a whole is that California has the fifth lowest number of reported tests per population (behind Georgia, Kansas, Oklahoma, and Texas). A recent Atlantic article shed light on the issue: as of last week, thousands of Californians have been swabbed for the virus, but their samples had not yet been examined. As a result, there is likely a significant undercount of cases in California. An undercount impacts not only how prevalent we think COVID is in CA, but also how believable other key metrics are.
Some newer projections, for example, base their estimates of total COVID cases in the state on observed death rates. Due to patchy testing data and differences in testing protocol, this is definitely a safer bet than using testing numbers; patients who die are more likely to have been tested. If we know what share of infected people are likely to die from the disease, using deaths to project overall cases should provide an accurate total headcount. However, given that CA has one of the lowest per-capita testing rates in the country, it is plausible that even COVID-related mortality estimates are unreliable. For example if patients were never tested before they died, COVID would not be listed in the chain of events (diseases, injuries, or complications) that caused their death. Because the first confirmed case of COVID in LA was in January – well before San Francisco (March) or many other parts of the country – the potential that spread was occurring through February is high, but but deaths due to COVID were not being coded as such because it was not yet considered a plausible cause of death.
One strategy to better understand if inadequate testing is leading to an underestimate of COVID-related deaths is to reassess the number of recent deaths amongst patients who may be considered presumptive positives i.e. have “adjacent” clinical diagnoses. For example, this may involve quantifying rates of excess pneumonia-related death in the state, which would ensure models that currently base their estimates on mortality rates are more accurate.
Another indicator to look at is the percent of COVID-19 tests that come back positive. A high share of people testing positive may be indicative that there is a high burden of cases in the community that have not yet been identified. For California, this number is relatively high – but not an outlier – at 11% of all tests that have been conducted to date (20th highest nationally when compared to other states). Taking an international perspective, the WHO Health Emergencies Program Director, Micael Ryan, noted that in countries with wider community testing fewer than 12% of their tests are generally positive. Sitting just below this benchmark, it is clear that California has been expanding access to testing, setting an example for high-positive-rate states. 11% also represents a significant decrease; CA’s positive rate was as high as 26% just last week. The drop likely reflects resolution of the diagnostic backlog mentioned above. However, we do not yet have this data at the county level, so it is difficult to tell how well the expansion in testing has been distributed.
Overall it looks like California is on track to prove the Surgeon General right: the state appears to be bending the curve of COVID-19. But ensuring this stays true it is a matter of doubling down and disaggregating data to reflect the realities faced by all Californians. In LA, for example, early data suggests 17% of positive cases (for which we have data) are black, yet black residents make up only 9% of the city’s population.
We need a positive example right now. But before we claim CA as an unequivocal success, we require a better understanding of who in the state is still getting COVID-19 and what factors impede their safety. Amidst federal cutbacks on testing support, this requires more testing, not less – and correspondingly granular data – data that reflect the diverse realities across the state. This is fundamental to mitigating the spread of the disease, but also to understanding if the curve is truly being flattened – for everyone – in California.
|Indicator||US State Average||CA
|Cumulative cases, absolute number||9,188||16,957||2,602||8,546|
|Cumulative cases per million||1,110||429.2||550.2||663.8|
|Cumulative deaths per million||36.0||11.2||13.5||16.7|
|Cumulative tests per million||8,011||3,916.1||–||–|
|Positive test rate (all tests conducted to date)||12.8%||11.8%||–||–|
*U.S. State Averages are averages of the values for the 50 U.S. states and the District of Columbia, they do not include protectorates or territories.