Background/Purpose:

Systemic autoimmune diseases (SAD) are characterized by a wide spectrum of demographic patterns with respect to the age at diagnosis, gender distribution and ethnic differences. Studying the distribution of these diseases across geographic regions and ethnic groups using a big data-driven approach may facilitate understanding of the corresponding genetic and environmental underpinnings, and help obtain a more “high-definition resolution” of these complex diseases.

Methods:

We explored the potential of the Google search engine to collect and merge cohorts (>100 patients) of patients with systemic lupus erythematosus (SLE) reported in the Pubmed library. We made a text-word search in Google (www.google.com) between 8th and 15th May 2015 using the following text algorithm: “systemic lupus erythematosus” and “100…100000000 patients” and “site:http://www.ncbi.nlm.nih.gov/pubmed”. We analyzed potential links between epidemiological features (age, gender ratio), ethnic distribution and geographical variables (total population, nominal gross domestic product –GPD-, the pollution index –PI- and the exponential pollution index –expPI- of the country of origin of each cohort of SLE patients).

Results:

We merged the data of 133 SLE cohorts including 171,000 patients; gender was detailed in 130 cohorts: 150,937 (88%) women and 17,958 (12%) men (female: male ratio, 8,4). mean age at onset (29.89 ± 3.48), at diagnosis (32.33 ± 2.99), and at protocol (40.57 ± 5.01). SLE was diagnosed according to  ACR criteria in 89% of cohorts, and according to ICD codes in 11%. The countries contributing the most cohorts were the USA (31), Japan (8) and Spain (5). A higher female: male ratio was found in cohorts with a higher frequency of Asian patients (r = 0.386, p=0.015), a higher number of participant countries (ratio of 14.5 in cohorts from more than one country vs. 10.7 in cohorts from one country), cohorts from countries with a greater population (r=0.189, p=0.043), cohorts that recruited patients in the 21^st  century (ratio of 11.7 vs. 8.9 in cohorts that recruited patients in the 20^th century) and cohorts in which the time of recruitment was < 10 years (11.7 vs. 8.9 in cohorts with a time of recruitment > 10 years). The PI (r=-0.416, p=0.035) and the expPI (r=-0.414, p=0.036) were inversely associated with the mean age at diagnosis of SLE.

Conclusion:

This analysis is the first to show the potential benefits of a combined search using Google and Pubmed to seek geographical factors that influence the phenotypic expression of SLE. This approach is an example of multidisciplinary collaboration that could involve clinical researchers, epidemiologists, mathematicians, and health informatics experts. The approach used here is a first, exploratory step which will require significant improvement and refinement. Using a “big data” approach enabled hitherto unseen connections between the environment and SAD to emerge.

« Back to 2015 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/google-driven-search-for-autoimmune-big-data-links-air-pollution-and-younger-age-at-diagnosis-of-systemic-lupus-erythematosus-geoepidemiological-analysis-of-171000-adult-patients/