Why do long COVID estimates seem to vary so much from study to study?

Data and Metrics Data Literacy Long COVID

A: Different case definitions, study participants, data collection methods, and study timing across studies.

TL:DR: Current estimates of the proportion of COVID-19 cases that develop long COVID range from 3-20% and have been as high as 50%! Considering how studies differ can help us understand why estimates vary so much across studies.

Things to keep in mind when comparing long COVID studies:

1. Case definition. Take note of *what* definition of long COVID was used! To count cases of a disease, you have to set criteria for what counts as a case (known as a case definition). The World Health Organization defines long COVID as symptoms lasting >2 months among those who were usually infected within the last 3 months that cannot be explained by an alternative diagnosis. In contrast, a recent Centers for Disease Control and Prevention study defined it as symptoms lasting ≥ 3 months that people did not have prior to having COVID-19.

Tip: Ask yourself whether the definition of long COVID was the same across studies.

2. Study participants. Pay attention to *who* participated in the study! In order to get an estimate of how many people in the entire population have long COVID, you either need to ask everyone in the entire population about their symptoms, or ask a representative slice of the population. Doing this is challenging, so more often studies include people who volunteer to participate. People who say, “Yes! I want to be in a study about long COVID!” may be different from those that say, “No way!”.

Tip: Consider whether the people who participated in each study might be more or less likely than the general population to have long COVID.

3. Data collection. Think about *how* the information in the study was collected! Studies that rely on medical records to identify people with long COVID, for example, might only be representative of those that have good access to health care or sought medical help for their symptoms. On the other hand, studies that ask people to respond to a survey on a smartphone may exclude those that only have a landline or aren’t web-savvy and also rely on people’s ability to recall their COVID-19 symptoms accurately.

Tip: Assess whether the data collection methods could lead to the proportion of people with long COVID being over- or under-estimated in each study.

4. Study timing. Watch out for *when* the study was carried out! Two recent studies identified that long COVID risk was lower during the omicron surge compared to the delta surge, but that risk of long COVID was higher with the BA.2 than BA.1 variant and we don’t yet know what risk will be with BA.4, BA.5, or other variants. In addition, different proportions of the population have also been vaccinated over the course of the pandemic, which may also have an impact on the burden of long COVID at different time points.

Tip: Pay attention to the period in which each study occurred.

All these considerations aside, whether 20% or only 3% of infected people develop long COVID, given the large number of COVID-19 cases that have happened worldwide, many people are and will be dealing with long-term effects of this disease.

To those who have experienced long COVID symptoms-we see you and we will keep sharing out any new information that sheds light on the proportion of people that are affected by this condition.

———————————–
Additional Links:

For more discussion on why long COVID estimates may vary across studies, see a recent STAT News article here, or Nature article here.

For a more in-depth discussion on why estimates vary and info on some recent long COVID studies, tune into our recent FB live at ~8 minute mark here.

For more on studies showing that risk for long COVID differs across variants, see our recent post here.