Follow-up on Flu vs. Covid
Follow-up on Flu vs. Covid
A deep-dive into a published study of hospitalization data
I know it’s been a few weeks since my last post. But I am back with a deep dive into a study on pediatric hospitalizations. The study is actually from last summer, but I just came across it again recently and decided to look into it in more detail.
After the Covid Twitter files came out, I launched my Substack a little earlier than I had planned, but I’m planning to publish more frequently going forward. I’ve also imported my previous articles from covid-georgia.com into my Substack archive for easy reference.
A couple of weeks ago, Congressman Ted Lieu tweeted about kids and Covid, and linked to an outdated and incorrect article Bloomberg from June 2022, Covid Is Way More Lethal to Kids Than The Flu. That article was based on a flawed analysis by Jeremy Faust, Delta and Omicron killed far more children than flu ever does. At the time it was written (and immediately promoted by White House Covid Czar Dr. Ashish Jha), several people, including myself, criticized that analysis. (Here’s the post I wrote detailing the flaws.) After almost a week, Dr. Faust made a few (inadequate) revisions to the original article, and wrote a second article, Pediatric Covid-19 versus influenza: further thoughts and statistics.
In his second article, Dr. Faust talked about pediatric hospitalization data and shared a tweet from the CDC’s influenza group, who he said “clearly went out of their way to agree” with him:
I criticized the Bloomberg article in a reply to Congressman Ted Lieu on Twitter, and Dr. Faust once again defended his analysis in replies to me, and he also replied directly to the Congressman Lieu with a screenshot of the CDC tweet and a link to his “further thoughts” article.
As a result, I thought I’d take another look at the study he referenced, and I’m glad I did. It turns out that the study doesn’t really support the claim that flu is much worse than Covid for children. It was an interesting read, so I thought I’d break down some of the findings…
Hospitalizations with Influenza vs. Covid
While the CDC’s tweets refer to the Omicron period, the study linked to in the tweet is primarily a detailed comparison of hospital data from COVID-NET for a year of Covid (October 2020-September 2021) and from FluSurv-NET for prior flu seasons (October-April 2017-2021). This includes the Delta wave, but not the Omicron wave.
However, before they published in May 2022, they did a minimal post hoc analysis of the period October 2021 through April 9, 2022, which includes the latter part of the Delta wave in the northern US plus the January Omicron wave. Unfortunately, the post hoc analysis includes very little data. Most of the interesting data is available only for the original study period.
For the original study period, keep in mind that the study compared a year of Covid hospitalizations to flu hospitalizations during prior flu seasons, which are only 7 months long. They used the flu numbers as an annual rate with the reasoning that flu detection is low during May-September, so the burden of flu during flu season is comparable to an annual rate (essentially assuming zero out-of-season flu hospitalizations).
Here’s a table showing some of the top line results, comparing hospitalization rates for past flu seasons with Covid hospitalization rates during the original one year study period:
Based on these rates during the original study period, they concluded the following:
For ages 0-4, Covid hospitalizations were LOWER than flu hospitalizations.
For ages 5-11, Covid hospitalizations were SIMILAR to flu hospitalizations.
For ages 12-17, Covid hospitalizations were HIGHER than flu hospitalizations.
It’s worth noting that the higher rate of hospitalization for older children 12-17 (59.94) is still less than the flu hospitalization rate for children under 4 (77.79) — in other words, teens were hospitalized with Covid less than infants and toddlers are typically hospitalized with flu. This is even more noteworthy when you consider two major limitations of the data shown above: incidental Covid hospitalizations and missed flu hospitalizations.
Incidental Covid Hospitalizations
As most people realize, historically we have not done widespread influenza testing, certainly for asymptomatic patients. However, universal Covid testing in hospitals is very common, regardless of symptoms or reason for admission. As a result, many Covid hospitalizations are incidental to the admission due to required Covid testing, which was never an issue with flu in the past. Previous studies have found about 40-45% of pediatric Covid hospitalizations were incidental to the Covid diagnosis (or to be more precise, incidental to the SARS-CoV-2 positive test result).
This study provides data from the hospital surveillance networks about the percentages of the reported flu and Covid hospitalizations with one or more symptoms at admission. Across ages 0-17, 20.3% of Covid hospitalizations during the original study period had no symptoms, while only 3.1% of flu hospitalizations had no symptoms. Additionally, 35.2% of Covid hospitalizations had no respiratory or febrile symptoms, versus only 4.4% of flu hospitalizations. Clearly, the ascertainment of Covid hospitalizations is much higher than for influenza. The study also reported that 25.1% of Covid hospitalizations did not list Covid as the primary reason for admission. These incidental hospitalization numbers are actually lower than many other studies have found during similar time periods — these are very conservative rates.
The percent of symptom-free and incidental hospitalizations also varied by age group, with older children and teens much more likely to have incidental admissions during the study period, which include reasons such as “obstetrics/labor and delivery, inpatient surgery or procedures, psychiatric admission needing acute medical care, trauma….”
The paper acknowledges these incidental hospitalizations in the limitations section, and notes that they did not exclude them from their rate calculations:
“FluSurv-NET and COVID-NET cases may have been hospitalized for reasons other than influenza or COVID-19. This may have been more common for COVID-19 due to SARS-CoV-2 screening practices, which were universal among hospitalized patients at some facilities during certain time periods. While influenza or COVID-19 may not have been the primary reason for admission for all hospitalizations, such cases were included in rate calculations….” [emphasis added]
Although the study authors did not adjust their hospitalization rates to remove incidental Covid admissions, they did provide the necessary data for me to adjust the hospitalization rates from above, which results in this:
Based on the adjusted rates excluding incidental hospitalizations…
For ages <12, symptomatic Covid hospitalizations were LOWER than symptomatic flu hospitalizations.
For ages 12-17, symptomatic Covid hospitalizations were HIGHER than symptomatic flu hospitalizations.
For children overall (ages 0-17), symptomatic Covid hospitalizations were SIMILAR to symptomatic flu hospitalizations.
Missed Flu Hospitalizations
An additional issue related to the differences in rates of historical flu testing versus current Covid testing is the increased likelihood of missed flu hospitalizations in prior years due to limited flu testing. The paper’s limitations section also acknowledges this issue: “Underdetection of influenza was likely greater than COVID-19 due to underutilization of seasonal influenza testing.” As a result, the adjusted numbers in the table above above are still biased to undercount flu hospitalizations as compared to Covid hospitalizations.
Post Hoc Analysis During Omicron
The original study used data through September 2021, which includes the summer Delta wave in the southern US. As I mentioned, the authors also did a brief post hoc analysis of hospitalization data from October 2021 through April 9, 2022, which includes the latter part of the Delta wave in the northern US plus the January Omicron wave. The paper states simply:
“…preliminary COVID-19–associated hospitalization rates among children during October 2021–April 2022 were higher than COVID-19–associated hospitalization rates observed during October 2020–September 2021 and influenza-associated hospitalization rates observed during the 2017–2018 through 2021–2022 seasons.”
That finding is based on combined rates for ages 0-17 that were not adjusted to remove incidental Covid hospitalizations. Unfortunately, the post hoc analysis doesn’t provide rates for asymptomatic or incidental Covid hospitalizations during this time period. However, we know from other sources that across all ages, hospitals reported many more incidental hospitalizations during Omicron than during previous periods. LA county reports incidental hospitalization rates by month. For the months of Aug - Nov 2021, the rate was about 25%. In Dec 2021 - Jan 2022, at the peak of the Omicron wave, it was ~40%, rising to 60% in March 2022.
There was also a big variation in the age groups of children hospitalized during the Omicron wave. The paper doesn’t provide any specific numbers for the post hoc analysis period, only the graphs below. These graphs show that Omicron hospitalizations were only noticeably higher than previous flu or Covid seasons for two age groups: 0-6 months and 12-17 years. The higher rate for teens is consistent with the pre-Omicron findings (and no doubt highly affected by incidental hospitalizations).
Young babies are hospitalized at much higher rates than older children for a variety of illnesses, so I plan to do a separate article where I can review in more detail the data we have on infants and Covid.
UPDATE: Here’s my follow-up article on infant Covid hospitalizations.
Comorbidities
A final interesting tidbit in the paper was the difference in comorbidities between Covid and influenza. For flu, asthma was more common (23.6% for flu versus 16.3% for Covid). However, these numbers are for all hospitalizations with a positive test, so they include the incidental Covid hospitalizations. The asthma rates may have been closer if they compared to children hospitalized with Covid who had 1+ symptom or where Covid was the primary reason for admission
For Covid, obesity was more common (35% for Covid versus 17.5% for flu). This also varied by age. For children 12-17 years old, 41.4% of Covid hospitalizations listed obesity versus 21.4% for flu. From the study:
“The prevalence of obesity was approximately double among children with COVID-19 compared with influenza. Notably, the proportion of children with COVID-19 who were obese (35%) was similar to findings from another study of hospitalized children with COVID-19 (32%) and higher than the national obesity prevalence among persons 2–19 years old (22% in 2020).”
Again, this includes incidental hospitalizations with Covid. It’s likely that obesity was even more common in children and teens who were hospitalized primarily for Covid.
The Bottom Line
I compared pediatric deaths from flu and Covid in an earlier post, and addressed how incidental Covid detections, missed flu detections, and different case definitions for flu deaths versus Covid death makes it impossible to compare equivalent metrics.
This study allowed me to dig deeper into hospitalization differences between Covid and flu, and quantify how incidental infections inflate Covid numbers compared to flu during the study period. For children 6 months to under 12, the data show that Covid resulted in fewer hospitalizations than a typical flu season, even during the Omicron wave. Yet many areas in the U.S. closed elementary schools for a year or more, at a crucial time when children are learning to read and developing other fundamental skills.
This study did show that Covid led to more hospitalizations in older children, aged 12-17, during the Delta and Omicron waves than flu normally would. However, that’s in part because flu is more serious for young children than it is for teens. The symptomatic Covid hospitalization rate in teens was still less than young children are typically hospitalized for flu.
And now, the situation is even better than at any time covered in this study. After the large winter Omicron wave in January 2022, many more Covid hospitalizations are incidental than before, and MIS-C has all but disappeared. So unless something changes significantly, the risk of pediatric Covid hospitalization is greatly reduced from what it was during Delta and the initial Omicron wave in early 2022.
This study is one of those great examples where it really pays to go beyond the headline. The CDC parroted the combined rate of hospitalization for ages 0-17 years, without adjusting for incidentals or breaking down further by age subgroups, in order to exaggerate the risk of Covid to children, but when you look at the data more closely, a much more nuanced story unfolds.