Study suggests many symptoms are more common after COVID-19 infection, but only a few are commonly associated with a long COVID diagnosis

By Pooja Toshniwal PahariaReviewed by Danielle Ellis, B.Sc.Jan 11 2023

In a recent study posted to the medRxiv* preprint server, researchers compared symptoms in patients with and without prior coronavirus disease 2019 (COVID-19) history and investigated symptoms associated with long COVID diagnosis in the United Kingdom (UK).

*Important notice: medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.

Background

Long COVID is a well-established consequence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections. However, data on the symptomatic burden of COVID-19 presented by patients at prime care centers (general practices), based on which long COVID diagnosis is established, are limited since the symptoms are usually documented by general practitioners (GPs) in unstructured clinical notes only.

About the study

In the present study, researchers described the long-lasting COVID symptomatic profiles as documented by GPs, and compared symptoms reported by individuals reporting positive or negative prior COVID-19 history. They also identified the symptoms related to long COVID diagnosis by GPs and long COVID risk factors.

The team utilized prime care electronic medical records data obtained from the THIN (the health improvement network) Cegedim prime care database, extracted by natural language processing using the FMA (free-text matching algorithm).

Adult individuals registered with GPs for ≥1.0 years in England, Wales, or Scotland, having a Read clinical terminology for suspected SARS-CoV-2 infection or confirmed SARS-CoV-2 infection, or clinical Read terminology for unspecific respiratory or viral infections with a mention of SARS-CoV-2 infection in unstructured data, or SARS-CoV-2-positive reports, were included.

For comparison, individuals without prior SARS-CoV-2 exposure or prior unspecific viral disease history were included as controls in a 1:1 ratio with case individuals. Data related to 89 COVID-19 symptoms and the diagnosis of long COVID were extracted. The data comprised structured information (comprising symptoms and diagnoses coded based on the Read terms) and unstructured information (comprising free-text clinical information in GP records, were analyzed.

For the initial evaluation of the symptomatic burden, odds ratios (ORs) were calculated for symptom documentation based on patient categories in four-week periods post-index date in comparison to the pre-index date reference period of eight weeks to 12.0 weeks.

Cox proportional hazard modeling was performed, and the adjusted hazard ratios (aHRs) were calculated with data adjustments for sex, age, ethnicity, medical history, body mass index, smoking habits, medical visits within one year prior to the index dates, and symptoms recorded 1.0 month to 3.0 months prior to the index dates).

The data were adjusted for every symptom reported after 12.0 weeks of COVID-19 diagnosis, representing the period exceeding which symptoms observed could contribute to the long COVID condition diagnosis, based on the World Health Organization (WHO) diagnostic criteria. Analyses were weighted based on inverse probability weighting and stratified based on general practice. LCA (latent class analysis) was performed to identify the symptoms and risk factors associated with the diagnosis of long COVID.

Results

The team assessed data from 11,015 individuals with confirmed SARS-CoV-2 infection and 15,841 individuals with suspected SARS-CoV-2 infection, compared to 18,098 controls. Among 60,800 study participants, 63% (n=38,407) were women, with a mean participant age of 52.0 years, with equivalent participation from individuals residing in Wales, Scotland, and England. The majority (90%) of participants were Whites.

Twenty percent and 80% of symptoms were coded and freely available, respectively. A persistently higher symptom documentation rate was observed 9.0 months beyond the index dates for COVID-19 patients and was greater among individuals with confirmed COVID-19. Associations inferred from structured data were identical to associations derived from unstructured data, and similar associations were observed in the subgroup analyses based on sex, age, and nationality.

Several symptoms were observed in association with SARS-CoV-2 infections ≥12.0 weeks post-SARS-CoV-2 infection, with the most robust associations in relation to the symptoms of fatigue, breathlessness, phlegm, and palpitations with aHR values of 4.0, 3.1, 2.9, and 2.8, respectively. However, only a few symptoms were documented within one week before the long COVID diagnosis in >20% of cases. They included breathlessness, fatigue, pain in the chest pain, cough, fatigue, and depression/anxiety.

The variables associated with significantly greater long COVID risks included the female gender (HR 1.2), age (HR 1.0 for each year elder), number of days with symptoms documented within one to three months before the index dates (HR 1.1 for each day of symptom documentation), annual medical visit frequency (HR 1.0), and hospitalization in the acute COVID-19 period (HR 3.6). LCA analysis findings indicated that the 2.0-class model was most appropriate for the study data.

Conclusion

The study findings showed that several symptoms are documented at prime care settings ≥12 weeks post-SARS-CoV-2 infection. However, only a few are commonly associated with long COVID diagnoses by general practitioners. In addition, a higher incidence of several symptoms was observed among individuals with prior COVID-19 history.

The wide spectrum of clinical symptoms of long COVID is indicative of multiple distinctive long COVID subtypes, with probably different immune pathways and other molecular aspects of COVID-19 pathophysiology.

*Important notice: medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.