Developing a long covid phenotype for post-acute COVID-19 in a national primary care sentinel cohort:, 2022, Meyer, Greenhalgh et al

Full title: Developing a long covid phenotype for post-acute COVID-19 in a national primary care sentinel cohort: an observational retrospective database analysis

ABSTRACT

Background:

Following COVID-19 up to 40% of people have ongoing health problems, referred to as “post-acute COVID-19” or long covid (LC). LC varies from a single persisting symptom to a complex multi-system disease. Research has flagged that this condition is under recorded in primary care record; and seeks to better define its clinical characteristics and management. Phenotypes provide a standard method for case definition and identification from routine data and are usually machine processable. A LC phenotype can underpin research into this condition.

Objective:

To develop a phenotype for long COVID-19 to inform the epidemiology and future research into this condition. We will compare clinical symptoms in people with long COVID-19 before and after their index infection. We will also compare people recoded as having acute infection with those with LC who have been hospitalised with those who are not.

Methods:

We used data from the Primary Care Sentinel Cohort (PCSC) of the Oxford-Royal College of General Practitioners (RCGP) Research and Surveillance Centre (RSC) database. The PCSC is recruited to be nationally representative of the English population. We developed a long LC phenotype using our established three-step ontological method: (1) Ontological step: Defining the reasoning process underpinning the phenotype; (2) Coding step: Exploring what clinical terms are available; (3) Logical extract model: testing performance. We created version of this phenotype using Protégé in the ontology web language (OWL) for Bioportal and using Phenoflow. We used the phenotype to compare people with long COVID-19 with: (1) Their symptoms in the year prior to acquiring COVID-19; and (2) People with acute COVID-19. We also compared hospitalised people with long COVID-19 with those not hospitalised. We compared socio-demographic details (age, gender, ethnicity, socioeconomic-status, obesity and smoking), comorbidities (cardiometabolic, respiratory and mental health) and Office of National Statistics (ONS) defined long covid symptoms between groups. We used descriptive statistics and logistic regression.

Results:

The long covid phenotype, available in Bioportal and Phenoflow, formats that differentiates people hospitalised with LC from people who are not, and where there no index infection is identified. The PCSC (N=7.4 million) includes 428,479 patients with an acute COVID-19 diagnosis confirmed by a laboratory test and 10,772 patients with clinically diagnosed COVID-19. We have before and after symptoms related to LC for hospitalized and community strata. 7,471 (1.74%;95CI=1.70-1.78) people were coded as having LC, 1,009 (13.5%;95CI=12.7-14.3) had a hospital admission related to acute COVID-19, 6,462 (86.5%;95CI=85.7-87.3) were not admitted to hospital of whom 2,728 had no COVID-19 index date recorded. 15.6% (95CI=14.7-16.5) of people with LC were hospitalised compared to 4.9% (95CI=4.8-5.0, p<0.0001) with uncomplicated COVID-19.

Conclusions:

Our long covid phenotype identifies LC cases and enables us to conduct a comparison between LC and defined comparator groups.

Open access, https://preprints.jmir.org/preprint/36989/accepted

 

I can't really make sense of what this paper is about. The abstract is written so poorly I genuinely don't know what they're saying, it seems written explicitly to hide the fact that it's not saying anything. As best as I can tell they defined phenotypes for hospitalized and non-hospitalized?

They are relying on healthcare records for this, which are very poor and misleading, and essentially admit that it's not possible:

They are limited by what symptoms are recorded and most aren't recorded because healthcare records usually don't bother with that. So in the end they make it about differentiating between hospitalized vs. non-hospitalized based on whether... they were hospitalized.

This as useless a study as it gets. EBM is dead, long live EBM+, which is the exact same, except more subjective and judgmental.