The three coronavirus metrics that most jurisdictions are publishing are cases, hospitalizations and deaths. The number of cases depends on the extent of testing, but the number of hospitalizations and deaths do not, at least not to the same extent, because while there are a lot of people who may have an infection and not get tested, testing is much more common for people who are hospitalized, and hospitalization most frequently precedes a death. Hospitalizations and deaths do, however, depend to some extent on the guidelines that are used to determine what qualifies as a coronavirus case.
There is another interesting fact about hospitalizations and deaths — they lag each other. Because a person will be hospitalized some time after being infected, and most frequently be hospitalized before dying, any peak in deaths should come after a peak in hospitalizations, and a peak in hospitalizations should come after a peak in cases. That is what we see in the data for San Diego County.
Here is a plot for San Diego County of cases, hospitalizations and deaths, with both data points and a smoothed line ( lowess, 20% ) versus days since March 13.
A peak in deaths is hard to discern in this plot, but there is a clear peak for cases and hospitalizations, with hospitalizations lagging cases, exactly what we should expect. What I find interesting is that all of the lines shoot up after day 38, and they all shoot up at the same time, with no lag.
If there had been a change in behavior that increased the number of infections, then the hospitalization line should start moving up about 8 days later, and the deaths line 5 days or so after that. But all three metrics move up together, with no lag.
The only explanation that I can think of is that there has been a change in the accounting system: the County is using a different method for counting what qualifies as a coronavirus-related death, hospitalization or death. Determining if that is true will require some investigation with the County, but if it is true, then the recent increase in the number of cases is not as concerning as it would be without the change in policy.