Characterizing Tweet Volume and Content About Common Health Conditions Across Pennsylvania: Retrospective Analysis

Christopher Tufts; Daniel Polsky; Kevin G Volpp; Peter W Groeneveld; Lyle Ungar; Raina M Merchant; Arthur P Pelullo

doi:10.2196/10834

Characterizing Tweet Volume and Content About Common Health Conditions Across Pennsylvania: Retrospective Analysis

JMIR Public Health Surveill. 2018 Dec 6;4(4):e10834. doi: 10.2196/10834.

Authors

Christopher Tufts¹, Daniel Polsky^{2

3

4}, Kevin G Volpp^{2

5

6}, Peter W Groeneveld^{2

6

7}, Lyle Ungar^{1

8}, Raina M Merchant^{1

9}, Arthur P Pelullo¹

Affiliations

¹ Center for Digital Health, Penn Medicine, Philadelphia, PA, United States.
² Leonard Davis Institute of Health Economics, University of Pennsylvania, Philadelphia, PA, United States.
³ Center for Health Equity Research and Promotion, University of Pennsylvania, Philadelphia, PA, United States.
⁴ Population Studies Center, School of Arts and Science, University of Pennsylvania, Philadelphia, PA, United States.
⁵ Department of Medical Ethics and Policy, University of Pennsylvania, Philadelphia, PA, United States.
⁶ Philadelphia VA Medical Center, Philadelphia, PA, United States.
⁷ Department of General Internal Medicine, University of Pennsylvania, Philadelphia, PA, United States.
⁸ Department of Computer and Information Science, School of Engineering and Applied Science, University of Pennsylvania, Philadelphia, PA, United States.
⁹ Department of Emergency Medicine, University of Pennsylvania, Philadelphia, PA, United States.

PMID: 30522989
PMCID: PMC6302232
DOI: 10.2196/10834

Abstract

Background: Tweets can provide broad, real-time perspectives about health and medical diagnoses that can inform disease surveillance in geographic regions. Less is known, however, about how much individuals post about common health conditions or what they post about.

Objective: We sought to collect and analyze tweets from 1 state about high prevalence health conditions and characterize the tweet volume and content.

Methods: We collected 408,296,620 tweets originating in Pennsylvania from 2012-2015 and compared the prevalence of 14 common diseases to the frequency of disease mentions on Twitter. We identified and corrected bias induced due to variance in disease term specificity and used the machine learning approach of differential language analysis to determine the content (words and themes) most highly correlated with each disease.

Results: Common disease terms were included in 226,802 tweets (174,381 tweets after disease term correction). Posts about breast cancer (39,156/174,381 messages, 22.45%; 306,127/12,702,379 prevalence, 2.41%) and diabetes (40,217/174,381 messages, 23.06%; 2,189,890/12,702,379 prevalence, 17.24%) were overrepresented on Twitter relative to disease prevalence, whereas hypertension (17,245/174,381 messages, 9.89%; 4,614,776/12,702,379 prevalence, 36.33%), chronic obstructive pulmonary disease (1648/174,381 messages, 0.95%; 1,083,627/12,702,379 prevalence, 8.53%), and heart disease (13,669/174,381 messages, 7.84%; 2,461,721/12,702,379 prevalence, 19.38%) were underrepresented. The content of messages also varied by disease. Personal experience messages accounted for 12.88% (578/4487) of prostate cancer tweets and 24.17% (4046/16,742) of asthma tweets. Awareness-themed tweets were more often about breast cancer (9139/39,156 messages, 23.34%) than asthma (1040/16,742 messages, 6.21%). Tweets about risk factors were more often about heart disease (1375/13,669 messages, 10.06%) than lymphoma (105/4927 messages, 2.13%).

Conclusions: Twitter provides a window into the Web-based visibility of diseases and how the volume of Web-based content about diseases varies by condition. Further, the potential value in tweets is in the rich content they provide about individuals' perspectives about diseases (eg, personal experiences, awareness, and risk factors) that are not otherwise easily captured through traditional surveys or administrative data.

Keywords: Twitter messaging; disease; prevalence; public health surveillance; social media.

©Christopher Tufts, Daniel Polsky, Kevin G Volpp, Peter W Groeneveld, Lyle Ungar, Raina M Merchant, Arthur P Pelullo. Originally published in JMIR Public Health and Surveillance (http://publichealth.jmir.org), 06.12.2018.