Good morning Mission. And welcome to Virus Village, your daily data dump.
Despite mounting deaths and problematic reporting, Gavin Newsom announced California “is turning the corner”, (though which corner not exactly clear).
Meanwhile in the Mission, the virus, like everybody else, may have been paying more attention to Kamala Harris than infections.
The City is reporting contact tracing numbers again, and the numbers are not inspiring. Only 74 percent of positive cases, and 77 percent of named contacts have been reached over the two weeks ending August 8.
The numbers today give us a mixed picture today of where we’re at. Positivity rates are going down, while hospitalizations rise.
HiGeorge, a data visualization startup, developed some new visualizations for Mission Local, which we will be using and fine-tuning in the days to come.
Another 11 positive cases were added to the Mission, bringing us to 1104 total cases. Yesterday I incorrectly reported that the area surrounding Dolores Park has the lowest number of cases per 10,000 residents. Wrong. That area is in the Castro.
The seven-day average of daily cases for the week ending August 6 was 89. This is the first time in a month that the average daily case number has fallen below 90. Still far too high (the red zone), and that figure will probably rise given the late reporting of test results.
A reader points out that although San Francisco has a lower number of deaths than the adjacent counties, the case anddeath figures are based on residency, not where the virus was contracted. Given the number of “essential workers” who come to SF daily for work, he raises a good point.
For the week ending August 6, the seven-day rolling average positivity rate fell below 3 percent for the first time since July 4. Definitely a good sign.
In the four days from August 7 to August 11, Covid positive patients (confirmed and suspected) went from 91 to 117. As the chart indicates, most of the increase comes from Acute Care admissions. DPH reports a diminished hospital capacity, with 74 ICU beds, and 341 Acute Care beds available as of August 11. Definitely not a good sign.
With our local R number still hovering close to 1, the virus continues its spread.
These reports are nice, but the sarcastic comments and editorializing create confusion and really muddy the waters. You might save them for the end, after you’ve presented the facts that people care about. If you choose to share your subjective opinion on whether something is a “good sign”, or “too high”, great, but at least cite some specific evidence to back it up.
Again thanks for this. You are my go to source for local well presented information on Covid data.
I just looked at the number of positive cases reported, and saw that the two most recent days that were over 100, were Wednesday and Friday. Can’t help but wonder if that is because there is low barrier testing at 24th Street on Wednesday and Friday.
Hi Mark,
RE: couple of points brought up in today’s article.
On area with lowest number of cases per 10,000 population, you referred to area around Dolores Park, then in the Castro. Going by Map of Confirmed Cases and Deaths on https://data.sfgov.org/stories/s/adm5-wq8i/ the area with lowest case rate per 10,000 population is census tract 06075015700 (Lone Mountain / USF). The map page does not calculate case rate for areas with fewer than 20 cases and does not show the actual number of cases if less than 10 so can’t know which area actually has lowest case rate. The actual lowest case rate area is academic. The important point is areas with disadvantaged population have much higher case rates.
Which brings up an annoying issue on the data.sfgov.org map page. That is the color coding for cases per 10,000 residents has not changed since the map page was put up around mid-April when the number of cases was around 1300. The number of cases today is close to 8000 or more than 6 times higher. So the old color coding has the effect of hiding the relative difference in case rates of different areas in SF. E.g., the first 3 color codes (5 to 10, 10 to 15, and 15 to 20) do not appear anywhere in today’s map and are useless. The old color coding also has effect of causing areas to jump from the “Case count too small (<20)" color code to the "Greater than 50" color code when then number of cases reaches 20. E.g., census tract 06075025100 (Castro) currently has 19 cases which falls in "Case count too small (<20). If one more case occurs there, the map page will calculate the case rate as (20 / 3573) * 10,000 = 56 which is color coded "Greater than 50".
The color coding could simply be scaled, say by a factor of 6, since the number of cases is about 6 times higher than in April when the map first started. However, this has disadvantage of requiring more scaling in the future as the number of cases continues to increase. A better method would be to normalize to the area with the highest case rate being 100. Keep the threshold for case counts fewer than some number (today it is 20) to account for the Relative Standard Error as mentioned on the map page. Once case count for an area reaches the threshold, the case rate is calculated and normalized to the case rate of the area with the highest case rate and a scaled color code applied. E.g,, could have 6 color codes — Case count too small (<20), <20% of highest case rate, 20% to 40% of highest case rate, 41% to 60% of highest case rate, 61% to 80% of highest case rate, 81% to 100% of highest case rate.
Current color coding shows areas in SF with 50 to 555 cases per 10,000 with same color code. Area is Mission with case rate more than 6X higher than area in Marina have same color code. The normalized color code proposed above would break out the same areas into 5 color codes. This gives viewers a more understandable picture of how the virus is affecting different areas in SF.
The other point is regarding number of deaths in SF vs adjacent counties and the cases and deaths based on residency. The point on essential workers living outside SF is misleading here because BOTH cases and deaths are based on residency. There would be a point if the cases are not based on residency and the deaths are or vice versa. What we want to look at here is the CFR (Case Fatality Rate), number of deaths divided by number of cases. Since BOTH cases and deaths are based on same residency, the CFR for an area is only dependent on the area and is NOT dependent on cases or deaths from outside of the area.
San Francisco does consistently have a lower CFR vs adjacent counties over the course of pandemic. In May, SF CFR was 1.7%, other Bay Area counties CFR were 3.0% to 5.6%. Today, SF CFR is 0.8%, other Bay Area counties are 1.1% to 1.8%. CFR is dependent on testing (among other factors). If testing does not find cases, then CFR will be higher. Testing increased substantially in all Bay Area counties since May, so more cases found lead to lower CFR. However, I don't know that testing differences explain why SF CFR is consistently lower. Other factors are quality of hospital care (higher CFR for overwhelmed ICUs) and age distribution of population (higher CFR for older populations). SF is similar to other Bay Area counties in these factors so also don't know how they would cause lower SF CFR. Would be interesting to read a research paper on this.
SF CFR of 0.8% is close to the 0.6% to 0.7% Infection Fatality Rate that I have read in Bob Wachter Twitter and research papers. A positive spin would be that SF is conducting enough tests to find most of the cases that are out there.
Ironically, Trump complained about too much testing causing higher number of cases. Then in the Axios interview, he brags about how US is doing great because of low CFR. If it were not for testing finding the cases, he would not be able to brag about low CFR.
Great point about SF workers being grouped with residents of other counties.
Thanks for this roundup, it is such a valuable service!