Inselspital
Neurology

Author Of 1 Presentation

COVID-19 Late Breaking Abstracts

LB1257 - Nowcasting Covid-19 in patients with Multiple Sclerosis via social media (ID 2153)

Speakers
Presentation Number
LB1257
Presentation Topic
COVID-19

Abstract

Background

SARS-Cov2 pandemic, starting in December 2019, demonstrated the need to rapidly gather data. Results of multicenter registries were published online first on 30-Apr-2020, 90 days after the 1st report of Covid-19 in Italy. Nevertheless, social media channels like Twitter® reported more promptly, but unstructured about Covid-19 cases.

Objectives

First, we aim to investigate whether an automated big data Twitter® search for reported MS cases with SARS-Cov2 infection can be used to set up a database and fill the timely gap between start of pandemic and publication of registry results. Second, we aim to compare the anonymized Twitter generated registry with the published information on Covid-19 taken from two reports of MS cohorts.

Methods

Twitter® was searched using over fifty hashtags from February 29th until publication of first registry data (30-Apr-2020). The resulting Twitter® data was compared with the Italian and French registry data published in Lancet Neurology and JAMA Neurology, respectively: for both cohorts: age, and sex, for French cohort only: hospitalization due to Covid-19. Chi2 test and a one-sample t-test (for age comparison) with mean age of Italian and French cohort as control values were used.

Results

Until 30-Apr-2020, 40 MS Covid-19 cases were identified and most cases have been published until 31-Mar-2020 (36/40); one month before first registry data was available. All cases were anonymized. MS patients were treated with anti-CD20 13/40, fingolimod 9/40, alemtuzumab 5/40, dimethyl fumarate 4/40, cladribine 3/40, and natalizumab/glatiramer acetate/interferon each 2/40. Information on severity was present in 38/40 cases and 7/38 (18.4%) were hospitalized compared to 73/347 in the French cohort (21.0%, p-value ≥0.05). Age was reported in 21/40 and sex in 20/40 cases with patients being in mean 41.4 years (SD 12.7) old and in 12/20 female. These rates did not differ from reported rates in the Italian or French cohort (each p-value ≥0.05).

Conclusions

Social media demonstrated to rapidly provide limited data on SARS-Cov2 pandemic in patients with MS faster than classical patient registries. Within the reported data, severity, age and sex appeared to be comparable to previous reports, however, for age and sex limited by the amount of missing values. Further limitations are the single point of observation lacking complete follow-up data, and the unstructured reporting within the given limitations of word counts. However, if in the future structured reporting of anonymized data sets could be established and flagged via common hash tags, quality of a social media reporting might serve as complementary strategy to common registry work. The main positive impact of such an approach is the rapid distribution of crucial information in a given emergency setting like the SARS-Cov2 pandemic and the complete availability to all persons without any access restrictions.

Collapse