Towards a reliable diagnosis of long Covid

Objectives

There are multiple definitions of long Covid but none have been externally validated. Reliability analysis can be used as an alternative to identify clinical features that make a reliable diagnosis.

Methods

Between October, 2022, and November, 2023, 25 patients were seen consecutively at a general medicine outpatient clinic and diagnosed with long Covid. Demographic details and clinical features were taken from the patients’ notes. The patients’ referring primary care physician was contacted to confirm if the diagnosis of long Covid remained the principal diagnosis or if another diagnosis had replaced it. Unreliability of the diagnosis of long Covid was defined as whether an alternative diagnosis was offered to explain the patient’s symptoms. Bayesian regression analysis was used to estimate hazard ratios for clinical variables and Weibull analysis was used to quantify the “failure” rate and the time needed for 50% of diagnosis to become unreliable.

Results

The total follow up time for the cohort was 25.3 patient years. Seventeen of the 25 patients diagnosed with long Covid (72%) were female sex, the median age was 45 years (range, 19 to 78 years), and the median duration of symptoms before the diagnosis of long Covid was 9.9 months (range, 3.4 to 25.4 months). Six patients received an unreliable diagnosis (24%, 95% CrI 11 to 43%). Variables predictive of an unreliable diagnosis were “fatigue”, “headache”; other differential diagnoses plausible after assessment; reduced exercise tolerance; male sex; and age. Variables that were predictive of a reliable diagnosis were "dizziness", palpitations; dyspnoea; myalgia; "memory difficulties" or "memory problems"; "poor concentration" or "brain fog"; migraine; post exercise malaise; patients with more comorbidities; and investigative tests performed after the first assessment. The predicted time needed for 50% of the diagnoses to be unreliable was 892 days.

Conclusions

An active diagnostic process, which excluded the existence of alternative diagnoses which could explain a patient’s symptoms, was more likely to produce a reliable diagnosis of long Covid. Reliability analysis can provide clinicians with useful information to improve diagnostic reliability even in circumstances where scientific knowledge of a disease entity is lacking.

Competing interest statement

No competing interests

Funding statement

No external funding was used

Author declarations

I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained.

Quality Improvement Medical, General Medicine and Acute Medical Unit Committee of the Fiona Stanley Fremantle Hospital Group gave ethical approval for this work.