As one of the largest studies of long COVID, it aims to inform public health and clinical care by identifying who gets long COVID, and more specifically to better understand community-dwelling populations where large-scale data, longitudinal follow-up, presence of comparison groups, and a consensus definition of long COVID are often found lacking.
Defining long COVID has differed among studies. In this study, Song and Giuriato compared administrative claims of two groups of “long haulers,” a narrow definition (based on a recorded diagnostic code) and broad definition (based on symptoms).
They measured the days between initial COVID-19 diagnosis and long Covid diagnosis for the narrow definition long haulers. They also compared both long hauler groups to a similar group of non-long haulers, who were initially diagnosed with COVID-19 but had no ensuing diagnosis or symptoms of long COVID.
Using data on over 800,000 people who have had COVID-19 from the OptumLabs Data Warehouse, the study identified 8,329 long haulers based on the diagnosis code and 207,537 long haulers using the symptom-based definition. They were each compared with 600,161 non-long haulers.
Considering demographic variables such as age, sex, health insurance plan, and characteristics obtained by county (race, ethnicity, urbanicity, and geography), and using pre-pandemic diagnoses from January 2019-February 2020, the investigators compared long haulers to non-long haulers on 31 categories of comorbidities. They used statistical analyses to assess the associations between demographic and pre-pandemic clinical factors with later having long COVID.
There were several limitations of the study. The narrow definition of long COVID likely had lower sensitivity capturing long haulers compared to the broad definition. However, the broad definition likely included patients whose symptoms were attributed to pre-existing conditions, making it difficult to differentiate pre-existing symptoms from those of long COVID.
In addition, because many long hauler cases were likely unreported and undocumented, the comparison group of non-long haulers likely contained some people who had long COVID.
It was also noted that access to care is unevenly distributed by race, ethnicity, and other social factors. Additionally, the study did not include people who disenrolled or died during the study period. Moreover, results might not generalize to populations outside of the commercial and Medicare Advantage enrollees in the OptumLabs Data Warehouse.
And finally, the study was not able to consider how virus adaptation and evolving immunity might affect the risk of developing long COVID.
The overall findings of the study revealed that long haulers were most likely to be older, female, and have existing health conditions such as hypertension, chronic lung disease, obesity, diabetes, and depression. Both narrow definition and broad definition long haulers displayed a higher incidence of pre-pandemic comorbidities than non-long haulers, with narrow definition long haulers having a higher pre-pandemic incidence of cardiovascular, pulmonary, endocrine, and other conditions such as depression, liver disease, and arthritis.
Also, among narrow definition long haulers, the average time between COVID-19 and long COVID diagnoses was 250 days, with the distribution mirroring surges of the pandemic: 43% were within the first 6 months of initial COVID-19 diagnosis (likely involving the delta variant), 27% within the next 6 months (encompassing the winter 20/21 surge), and 29% more than one year (reflecting the initial surge of the pandemic).
As long COVID continues to loom over the health care system and society, understanding its risk factors is important for patients, clinicians, and policymakers. Further study is necessary to reach a better understanding of long COVID among commonly occurring comorbidities and will be key in assessing the causes and consequences of long COVID.