People with heart disease or diabetes should be extra careful about Covid19

Many people may have seen the all-cause mortality statistics in #NYC. They are pretty grim. If you haven't, I downloaded the data from the the National Center for Health Statistics (#NCHS) and plot them here. You can learn a lot if you dive into these data.

(Warning: You will see a lot of graphs comparing causes of death. The subtitle will indicate the cause. Also, I am making causal inference based on associations, which also should have a Surgeon General's warning.)

Why no turndown in NYC? Week 1 is the first week of January. The gray lines plot the last five years, and the cranberry line is this year (to date). You can see the excess mortality in NYC relative to past years clearly. You may have seen it in the media as well. Because of NCHS lags in reported deaths, I have excluded the last month. If I included the provisional counts for the last month, you would see a turn down in deaths. At the peak, however, it looks like NYC had about 4000 excess deaths.

What does this mean for social distancing? What the media hasn't reported is the experience in other states. I like to contrast New York's experience with California, which isolated earlier. Here is California in the plot below.

Note that California looks 'bumpier' just because of the scale on the axis. New York actually shows a similar bumpiness if you take out this year. Everywhere, we see a bowl shape to mortality because more people die in the winter months. This likely has to do with flu but also stress from the holidays. To see the impact of flu, I plot flu and pneumonia deaths in California.

California was having a good season for flu and pneumonia starting in January, but likely Covid19 has kept it elevated relative to past years. Here is what flu and pneumonia looks like for NYC.

The problem is that the flu and pneumonia are only coding as excess deaths of about 200 weekly. Put another way, mortality is about 5x higher due to flu and pneumonia at the height of the epidemic. The bottom line is that social distancing appeared to work in California.

What about people with heart disease? It turns out much of the excess mortality is being coded as heart disease. The key question we need to solve is whether this is causal, but to my mind the fact that in NYC mortality is three times higher.

What about people with diabetes? Diabetes deaths show a similar trend. (Of course, deaths among people with diabetes could be coded as many other causes, but this is a specific code for diabetes.)

Are the spikes in heart disease and diabetes because NYC health system was overwhelmed? This gets to the heart of whether covid19 has direct effects on the chronically-ill, or was it another victim of NYC health system being overwhelmed. If it was the system, then we should see it in other causes of death. Below I graph cancer deaths in NYC.

Cancer mortality didn't change in NYC over this period. So, if the NYC system was overwhelmed, it didn't affect cancer patients in March. (It may down the line though due to disruptions in treatment.) As more clinical data come in, I think we will find that covid19 has additional risks for people with heart disease and diabetes--even beyond age. Please be careful!