Discover more from Vinay Prasad's Observations and Thoughts
Does COVID19 cause diabetes in kids?
The CDC's publishes an embarrassing study
The CDC is back with a new piece of propagan—- I mean, a new publication. It claims that kids (<18) with COVID are more likely to be diagnosed with diabetes in the next 30 days than kids without COVID or kids with other pre-pandemic respiratory viruses. It asserts this is a causal effect. COVID causes diabetes in kids.
To make this claim, the CDC examines 2 databases: IQVIA and HealthVerity. From IQVIA, they pull out kids <18 with COVID19 diagnosis, and age and sex matched kids without the diagnosis, as well as those with a prior non-covid respiratory infection. From HealthVerity, they pull out kids <18 with COVID19 diagnosis and kids <18 who got tested for COVID19 and were negative*. (we shall return to this).
In all cases, administrative/billing codes for COVID19 and diabetes were used to see who had COVID19 and who developed diabetes within 30 days of the index encounter. Of course, many, many more kids had COVID who are not in the database. Some may not have even had symptoms, and others may not have sought testing.
In IQVIA, among kids with COVID19, a whopping 68 out of 80,000+ or 0.08% ended with diabetes; among kids without COVID19, it was 132 out of 400,000+ or 0.03% ended up with diabetes, and among kids with prior respiratory infection it was ~0.06%
The absolute risk of diabetes due to COVID (if you believed this is causal) appears to be an increase on par with a swiftly eaten bag of skittles.
Now in the HealthVerity database, the risk of diabetes post COVID19 was 0.25% (a quarter of 1 percent), if you were tested for COVID19 but negative, it was 0.19% (one fifth of 1 percent). Here, COVID appears as risky as a McDonalds supersized soda.
Is that a fair take away or a fear-mongering distortion?
First, the whole analysis hinges on the idea that age-sex matched kids without covid should be comparable to the kids who got covid in terms of risk of diabetes. The only difference between the kids should be that some, unfortunately, had covid. But COVID may be more likely to affect kids of lower socioeconomic status, of certain races, and kids who were already overweight or suffering from medical problems.
Does the CDC attempt to correct for any of these confounders? Not at all. They surely have height and weight, and could adjust for BMI, but do not. I am truly puzzled as to why.
Second, they don’t have the true denominator. This is only kids who present and have a COVID19 diagnosis. Seroprevalence is needed to find the real denominator of kids with COVID19. This will lower the absolute risks. COVID19 may be downgraded from a whole bag of skittles to a single, red skittle.
Does the CDC adjust for this? Nope
Third, kids who seek medical care for COVID19 may get more blood tests than those without COVID19, and perhaps more than those with other respiratory viruses in yesteryear. This too may capture more diabetes.
Does the CDC correct for ascertainment? Not at all
Finally, the HealthVerity database is particularly odd, as the kids who get tested but test negative might be very dissimilar from those with COVID19. It may include some who are planning a trip to Maui (Hawaii used to require testing to visit), and compare them to kids who come in feeling very sick.
Does the CDC correct for this? No
What about the absolute risk?
Even if these results are true, the risk of diabetes due to COVID19 within 30 days was less than 1 in a thousand. Hospitalization is an order of magnitude more frequent. Of course, no kid wants COVID19. You don’t need the boogieman of diabetes to know you didn’t want it for your kid; and far more kids are hospitalized from it, than develop diabetes from it.
Lastly, the CDC uses this study to push vaccination, the findings “underscore the importance of COVID-19 prevention among all age groups, including vaccination for all eligible children and adolescents…”. But, the study provides no useful information to further weigh the risk benefit balance for vaccination beyond what was presented at the VRBAC meeting for 5 to 11, and a prior one on 12-17, a topic of wide global debate, with differing recommendations by nation (US vs UK).
As Marty Makary says, the CDC’s study would not pass a 7th grade science fair competition. It is certainly not serious scholarship. I have no idea if COVID19 causes more diabetes in kids than not having it; and the study gives me no better information to guide vaccination policies.
But that doesn’t make for a good tweet