Provisional case definitions for COVID-19-associated neurological disease - Authors' reply.

We read with interest the Correspondence by Hai-Feng Li and colleagues on our proposed definitions for COVID-19-associated neurological disease. We thank the authors for recognising the importance of collecting cases together with accurate diagnostic evidence to elucidate disease mechanisms.

Any case criterion for a neurological syndrome associated with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection must incorporate a definition of acute SARS-CoV-2 infection, criteria for the diagnosis of the neurological syndrome itself, and an attempt to link the two in a temporal relationship, excluding other potential causes. The definition of acute SARS-CoV-2 infection must also reflect rapidly evolving diagnostic approaches.

In our proposed definition for SARS-CoV-2-associated Guillain-Barré syndrome, we selected a pragmatic definition of acute COVID-19 infection, reflecting the WHO definition of confirmed infection. However, we accept that the timing of the infection onset is a challenge. A 6-week interval between viral symptoms onset and neurological disease is somewhat arbitrary, but from our knowledge of other infections triggering Guillain-Barré syndrome, a longer delay than this would cast the association into doubt. In patients without symptoms of SARS-CoV-2 infection but with positive RT-PCR or antibody testing, the true date of infection is even more difficult to elucidate.

We agree that it is important to exclude influenza as a potential trigger of Guillain-Barré syndrome, and viral symptoms might be difficult to distinguish. Epidemiological data can be informative, especially as the incidence of respiratory pathogens changes with the seasons around the world. RT-PCR testing for influenza and other respiratory viruses could be done alongside SARS-CoV-2 testing when possible. We advise caution in interpreting the results of studies using positive serum antibody testing for the diagnosis of influenza, which can be vulnerable to cross-reactivity and poor inherent test accuracy. Additionally, the study by Kong and colleagues cited by Li and colleagues' Correspondence did not report co-infection, but rather early cases of COVID-19 in Wuhan, China, that were detected through the national influenza surveillance programme; existing influenza surveillance networks have been used for sentinel testing and to look for potential signs of community transmission worldwide, as supported by the WHO Global Influenza Surveillance and Response System. Furthermore, influenza-like illness is a syndromic definition and does not imply influenza to be the causative illness; its description aligns closely with the “acute respiratory infection” definition used to prompt testing for COVID-19 in earlier WHO and national guidelines.

Our group has shown previously that, in patients with new neurological disease and evidence of more than one infection, there are additional challenges in thinking about causality, particularly when the results are from specimens collected outside the CNS.5, 6