Millions of people are likely to be suffering from long COVID but an analysis has revealed a low level of SNOMED coding for the condition.
A recent systematic review of the available evidence, suggested that a substantial proportion of individuals reported a variety of symptoms four or more weeks after a COVID-19 diagnosis. The term long COVID has been very loosely described by NICE in their guideline on the subject, as signs and symptoms that continue or develop after acute COVID-19. To help support clinical care and implementation of NICE guidance, SNOMED coding have been developed for use in electronic health records and were released in November 2020. While these codes have been developed for UK-based clinicians, SNOMED is also designed for use in the European and international clinical arenas.
The SNOMED coding define “on-going symptomatic COVID-19” where symptoms persist for between 4 and 12 weeks and “post-COVID-19 syndrome” where symptoms still continue beyond 12 weeks. In total there are 15 SNOMED codes, diagnostic (2), referral (3) and assessment (10). With a predicted high number of patients likely to experience long COVID, a team from the OpenSAFELY Collaborative, which represents an open-source software platform for analysis of electronic health records, decided to examine the usage of the SNOMED coding in primary care in an effort to understand the distribution of long COVID cases throughout England. The rational for the study was the importance of ensuring appropriate coding for ongoing patient care, research into disease prevalence as well as public health resource management and planning. The OpenSAFELY platform creates pseudonymised data and coded diagnoses, medications, age, gender, geographic regions and an index of deprivation, were all extracted from patient records. The outcome of interest was any record of long COVID based on use of one or more of the SNOMED codes. The period of data collection was January 2020 through to the end of April 2021. While this covers a period before the codes were introduced, it is possible for practices to backdate a diagnostic code and the team calculated the proportion of patients with long COVID codes over the whole study period per 100,000 patients.
Findings
There were 58 million people in the data cohort with an equal gender mix. Ages were categorised with the highest proportion (20%) aged 0 to 17 years, followed by 55–69 years (17.1%) with an equal proportion (13%) from those aged 25–54. The main ethnicity was white (55.6%). There were 23,273 (0.04%) patients with a SNOMED coding for long COVID, giving an overall prevalence of 40.1 per 100,000 population and the diagnostic code “Post-COVID-19 syndrome” accounted for 64.3% of all codes. When stratified by age, there was an increasing prevalence, ranging from 18/100,000 in those age 18 to 24 and peaking at 84.5/100,000 in those between the ages of 45 and 54. However, a sizeable number of patients (62.9/100,000) in the 55 to 69 age range also had documented long COVID. Furthermore, long COVID was more common in females (52.1 vs 28.1/100,000, female vs male). Coding also varied considerably with geographic location, peaking at 55.6/100,00 in London and lowest at 20.3 in the East of England.
Discussing their findings, the authors main comment was the prevalence of long COVID appeared to be much lower than might be expected from other current survey data. They concluded that possible reasons for low reporting might include, a lack of patient presentation, that clinicians were simply not diagnosing long COVID among their patients or even a lack of awareness of the diagnostic SNOMED coding.
Citation
Walker AJ et al. Clinical coding of long COVID in English primary care: a federated analysis of 58 million patient records in situ using OpenSAFELY. Br J Gen Pract 2021
