Characterizing barriers to CPR training attainment using Twitter

Accepted Manuscript Title: Characterizing barriers to CPR training attainment using Twitter Authors: Shaun K. McGovern, Audrey L. Blewer, Andrew Murray, Marion Leary, Benjamin S. Abella, Raina M. Merchant PII: DOI: Reference:

S0300-9572(18)30118-7 https://doi.org/10.1016/j.resuscitation.2018.03.010 RESUS 7528

To appear in:

Resuscitation

Received date: Revised date: Accepted date:

21-8-2017 28-2-2018 5-3-2018

Please cite this article as: McGovern Shaun K, Blewer Audrey L, Murray Andrew, Leary Marion, Abella Benjamin S, Merchant Raina M.Characterizing barriers to CPR training attainment using Twitter.Resuscitation https://doi.org/10.1016/j.resuscitation.2018.03.010 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Characterizing barriers to CPR training attainment using Twitter

SC RI PT

Running title: Characterizing barriers to CPR training via Twitter

Shaun K. McGoverna, Audrey L. Blewera,b,*, Andrew Murraya, Marion Learya, Benjamin S. Abellaa, Raina M. Merchanta,b,c

a

Center for Resuscitation Science and Department of Emergency Medicine, University of

Penn Medicine Center for Digital Health, Philadelphia, PA, USA

M

A

c

Leonard Davis Institute of Health Economics, Philadelphia, PA, USA

N

b

U

Pennsylvania, Philadelphia, PA, USA

*

Corresponding author (Audrey L. Blewer)

CC

D

EP

Email: Mobile: Fax:

Department of Emergency Medicine University of Pennsylvania 3400 Spruce Street, Ground Ravdin Philadelphia, PA 19104 [email protected] 267-239-1765 215-573-2701

TE

Mail:

Article word count: 1,489 words

References: 20

Tables/Figures: 3 (3 supplementary)

A

Abstract word count: 284 words

McGovern et al – Characterizing barriers to CPR training via Twitter

Page 2 of 12

Abstract Study Aim Recent investigations have suggested that CPR training rates are low within the U.S and barriers

SC RI PT

to CPR training are poorly understood. Social media holds great potential for large scale capture of the public’s CPR training experiences and may illuminate barriers to CPR training. While studies have examined Twitter data for behaviors associated with cardiovascular health, no investigation has evaluated Twitter data to understand public perception of CPR training. We characterized Tweet content about CPR training and associated sentiment to better understand

U

barriers associated with CPR training. We hypothesized that negative CPR training impressions

A

N

would be identifiable as barriers to CPR training attainment.

M

Methods

We extracted Tweets from 2011-2015 originating in Pennsylvania including the keyword CPR

D

(n=8,419). A random subset of 1000 tweets were independently coded by two authors using

TE

grounded theory (mean kappa=0.74). CPR training Tweets were analyzed for subtopic and

EP

sentiment (“positive” or “negative”). Descriptive statistics were used; a chi squared test was used to examine differences in positive and negative responses.

CC

Results

A

Of 8,419 Tweets, CPR training was the most frequent queried result (16%). Within the coded 1000 subset, 18% referenced a CPR training experience. Upcoming CPR training (22%), CPR training curriculum (17%), job-related training (12%), and duration of training (10%) were the most discussed topics regarding CPR training experiences. Of those, the majority of CPR training

McGovern et al – Characterizing barriers to CPR training via Twitter

Page 3 of 12

experiences were negative (53% vs. 47%, p<0.01) and barriers to CPR training emerged as the primary source of negative experiences. Conclusions

SC RI PT

CPR training is the most referenced theme in CPR Tweets from Pennsylvania, and tweets were predominately negative, particularly referencing barriers such as time, location, and duration. Social media is useful for tracking barriers to CPR training attainment and future CPR education

U

modalities.

N

Keywords: cardiopulmonary resuscitation, education, basic life support, cardiac arrest, social

M

A

media

TE

D

Introduction Prompt delivery of bystander cardiopulmonary resuscitation (CPR) can double, or even triple, the odds of survival from out-of-hospital cardiac arrest (OHCA), yet less than one-third of victims

EP

receive bystander CPR in some neighborhoods.1-4 Investigations have demonstrated that CPR

CC

training may increase the likelihood of a person performing bystander CPR5, yet national bystander CPR training rates remain markedly low.6,7 Analyzing social media, an influential form of modern

A

communication, may help characterize and describe barriers to CPR training. Recent statements from the American Heart Association and National Academy of Medicine (formerly Institute of Medicine) call for improving bystander CPR rates by increasing CPR training using digital strategies such as social media.8,9 Many investigations have focused on barriers for provision of bystander CPR, yet few have focused on barriers for CPR training

McGovern et al – Characterizing barriers to CPR training via Twitter

Page 4 of 12

attainment.10-12 Prior work by Sasson et. al. used focus groups and interviews to assess barriers to bystander CPR training in a Latino community, citing cost and language concerns as primary prohibitory factors for CPR training attaintment.13 Innovative training modalities such as video-

SC RI PT

only training have sought to address these barriers, yet low CPR training rates persist. 14-16 To the best of our knowledge, no prior studies have used publicly available social media content to assess negative CPR training experiences which is important for the development of novel strategies for reduction of barriers for CPR training attainment.

To address this knowledge gap we utilized Twitter, a popular online news and social media service

U

where users post and interact with messages referred to as Tweets. A growing body of literature

N

has identified Twitter as a useful tool for public health research.17-19 This may provide the

A

resuscitation community with a novel method to assess barriers to CPR training attainment and the

D

M

associated negative CPR training experiences faced by the public.

EP

Data Source Twitter

TE

Methods This investigation was approved by the University of Pennsylvania Institutional Review Board.

Twitter is a popular news and social media service with over 300 million monthly active users

CC

(over 60 million in the United States alone)20. Tweet data (n=301,000,000) was collected from July 23, 2009 to February 5, 2015 and comprised of the “Twitter decahose”, a 10% sample of

A

Tweets covering 52 months of Tweets, and the “Twitter spritzer”, a 1% sample of Tweets covering the remaining 15 months. From this group, Tweets geocoded to Pennsylvania containing “CPR” were extracted (n=8,419). Tweet data included unique message id, message, friend count, followers count, county location, and city location.

McGovern et al – Characterizing barriers to CPR training via Twitter

Page 5 of 12

Variable Definitions Users share their thoughts, ideas, and experiences in messages of 140 characters or less known as Tweets.. Followers are individuals or organizations who subscribe to the users Twitter feed, while

SC RI PT

a Friend is an account that the user follows, more colloquially referred to as Following. Users have the option to share other Tweets to their Followers; this is known as a Retweet and is defined as a Tweet message prefaced with “RT”. Tweet link content was defined as a Tweet containing “http” in the message. Potential exposure of the tweet was calculated by quantifying the amount of Tweets from a user multiplied by the median amount of Followers. Other variables in the dataset included

N

U

location, county, and city.

A

Tweet Analysis Full dataset. Word frequency query was performed on the dataset (n=8,419)to ascertain common

M

language and themes to inform possible domains. A random subset of 1000 Tweets was generated

D

for a more detailed analysis.

TE

Sample of 1000. After sampling the messages of the smaller dataset (n=1,000), domains were arrived at through iterative discussion until agreement was reached. Tweets were independently

EP

coded by two authors (S.M. and A.B.) using grounded theory (mean kappa = 0.74) for each of the

CC

domains. See supplement Table 1 for domains and descriptions. CPR Training Tweets. Tweets within the “CPR Training” domain were additionally coded for

A

content and sentiment (“positive, “negative”, or “neutral”). Differing codes were iteratively discussed until agreement was reached. CPR Training subdomains are described in the online supplemental.

McGovern et al – Characterizing barriers to CPR training via Twitter

Page 6 of 12

Data Analysis Word frequency query was produced using qualitative data analysis software (Nvivo 11, QSR International, Melbourne, Australia). All data were analyzed using a statistical software package (STATA 14, Statacorp, College Station, TX). Descriptive statistics were used to describe message

SC RI PT

frequency; a chi squared test was used to examine differences in CPR training Tweet sentiment, which are positive or negative responses.

Results Descriptive Tweet Data

U

Of 8,419 Tweets, attending a CPR training was the most frequent theme; 12% of the Tweets

N

referenced “class”, while 4% of the Tweets referenced “time”. Additionally, 27% (2,280) of the

A

Tweets were classified as Retweets and 33.9% (2,856) contained a link to an external website. The

M

average user had a median of 1567 Followers and 952 Friends. The potential exposure for those

TE

Sample of 1000

D

Tweets containing CPR extends to 13,194,140 users. (Table 1)

EP

Eighteen percent referenced a layperson describing CPR training, such as a class or training. Eighteen percent referenced news media. Sixteen percent referenced a CPR used as a

CC

metaphor.Additionally, 11% were miscellaneous references, 9% were advertisements, and 7% referenced a CPR certification. Six percent stated soley CPR information, 5% were irrelevant, and

A

3% referenced CPR on an animal. Two percent referenced entertainment media, another 2% referenced an actual CPR event, and an additional 2% referenced CPR concerns. Only 1% were humorous references to CPR. (Table 2) CPR training Tweets

McGovern et al – Characterizing barriers to CPR training via Twitter

Page 7 of 12

On balance, more Tweets were characterized as “negative” than “positive” (53% vs 47%, p<0.01) for example, “The last thing I want to do is spend three hours in a classroom to get CPR certified, but that is precisely what I'm doing with my night.” Several barriers emerged as particularly

SC RI PT

causing negative experiences such as curriculum (60% vs. 40%, p=0.013), duration of training (94% vs 6% p<0.01), and time of training (83% vs. 17%, p<0.01). The most discussed topics surrounding CPR training experiences were curriculum (20%), upcoming training (18%), time of training (12%), and duration of training (12%).Positive CPR training experiences included in class impression (60% vs. 40%, p=0.09), job-related training (61% vs 39% p=0.04), passing training

U

test (100% positive). Twenty percent of CPR training domain Tweets were neutral and excluded

N

from sentiment analysis (Figure 1). CPR Training subdomain definitions, associated barriers or

M

A

facilitators, and example Tweets are located in Supplemental Table 2a and 2b.

D

Discussion This investigation has three main findings concerning barriers to CPR training attainment and the

TE

use of Twitter for resuscitation research. Above all, it is evident that the public discusses negative CPR training experiences and associated barriers to CPR training attainment on social media.

EP

Furthermore, these data allow for prioritization of needed changes to current CPR education

CC

modalities to reduce such barriers. Additionally, as a popular social media platform with publically accessible data, Twitter may be useful for future surveillance of CPR education modalities.

A

Social media provides the public with a platform to share their daily experiences. These data may be used to illuminate negative CPR training experiences and barriers to CPR training that may cause poor CPR training rates. Twitter is particulary viable for research, when compared to other social media platforms (e.g. Facebook), as privacy settings are defaulted to public and Twitter offers publically available samples of Tweet data. Similar studies have used more traditional

McGovern et al – Characterizing barriers to CPR training via Twitter

Page 8 of 12

means of measuring public perception of barriers to CPR training, such as post-training surveys, yet none have used social media.13 Evidence from this investigation suggests that Twitter is a useful data source; the public does indeed discuss their negative CPR training experiences on

SC RI PT

Twitter in a way that is useful for the resuscitation community. The volume of Tweets, along with the corresponding sentiment, has allowed us to identify barriers to CPR training faced by a portion of the U.S. population that uses social media. For example, many references to the time of day or duration of CPR trainings were significantly associated with negative sentiment suggesting that the public considers these to be considerable barriers to CPR

N

U

training. These findings may inform future development of CPR education modalities.

A

Limitations to studies using Twitter are similar to those of any observational study, but may also

M

include a non-representative population, un-validated methodology, and as Twitter is a form of social media, it does not exist for health purposes. Furthermore, demographic data pertaining to

D

Twitter users were unknown and may have impacted their sentiment towards CPR training. Future

TE

research in the area of barriers to CPR training may address if these barriers are unique to the population that utilizes social media. Furthermore, future research in developing novel CPR

EP

training modalities that utilize social media may help move the needle of public CPR training rates.

CC

In conclusion, Twitter holds large potential for understanding barriers to CPR training attainment faced by the public. Our findings show that time of day, duration, and curriculum of current CPR

A

training modalities are discussed in a primarily negative context by the public on social media. The volume of information that social media provides about public experiences surrounding CPR training is a viable area of research for the resuscitation community.

McGovern et al – Characterizing barriers to CPR training via Twitter

Page 9 of 12

Conflict of Interest Statement

Ms. Blewer received funding from the American Heart Association (Mentored Clinical Research Program Award). Ms. Leary has received research support from the Medtronic Foundation, the

SC RI PT

Laerdal Foundation the American Heart Association, as well as in-kind support from Laerdal

Medical and Physio-Control. Ms. Leary has ownership in ImmERge Labs, LLC. Dr. Abella has received research funding from the NIH, PCORI, the Medtronic Foundation, the American Heart Association and CR Bard. He has received honoraria from Philips Healthcare and CR Bard, as well as in-kind research support from Laerdal Medical Corporation. No other authors have

A

CC

EP

TE

D

M

A

N

U

conflicts to report.

McGovern et al – Characterizing barriers to CPR training via Twitter

Page 10 of 12

References Malta Hansen C, Kragholm K, Pearson DA, et al. Association of Bystander and FirstResponder Intervention With Survival After Out-of-Hospital Cardiac Arrest in North Carolina, 2010-2013. JAMA. 2015;314(3):255-264.

2.

Wissenberg M, Lippert FK, Folke F, et al. Association of national initiatives to improve cardiac arrest management with rates of bystander intervention and patient survival after out-of-hospital cardiac arrest. JAMA. 2013;310(13):1377-1384.

3.

Hasselqvist-Ax I, Riva G, Herlitz J, et al. Early cardiopulmonary resuscitation in out-ofhospital cardiac arrest. N Engl J Med. 2015;372(24):2307-2315.

4.

Chan PS, McNally B, Tang F, Kellermann A. Recent trends in survival from out-ofhospital cardiac arrest in the United States. Circulation. 2014;130(21):1876-1882.

5.

Tanigawa K, Iwami T, Nishiyama C, Nonogi H, Kawamura T. Are trained individuals more likely to perform bystander CPR? An observational study. Resuscitation. 2011;82(5):523-528.

6.

Anderson ML, Cox M, Al-Khatib SM, et al. Rates of cardiopulmonary resuscitation training in the United States. JAMA Intern Med. 2014;174(2):194-201.

7.

Blewer AL, Ibrahim SA, Leary M, et al. Cardiopulmonary Resuscitation Training Disparities in the United States. Journal of the American Heart Association. 2017;6(5).

8.

Committee on the Treatment of Cardiac Arrest: Current S, Future D, Board on Health Sciences P, Institute of M. The National Academies Collection: Reports funded by National Institutes of Health. In: Graham R, McCoy MA, Schultz AM, eds. Strategies to Improve Cardiac Arrest Survival: A Time to Act. Washington (DC): National Academies Press (US) Copyright 2015 by the National Academy of Sciences. All rights reserved.; 2015.

CC

EP

TE

D

M

A

N

U

SC RI PT

1.

A

9.

Sayre MR, Berg RA, Cave DM, Page RL, Potts J, White RD. Hands-only (compressiononly) cardiopulmonary resuscitation: a call to action for bystander response to adults who experience out-of-hospital sudden cardiac arrest: a science advisory for the public from the American Heart Association Emergency Cardiovascular Care Committee. Circulation. 2008;117(16):2162-2167.

McGovern et al – Characterizing barriers to CPR training via Twitter

Page 11 of 12

Linderoth G, Hallas P, Lippert FK, et al. Challenges in out-of-hospital cardiac arrest - A study combining closed-circuit television (CCTV) and medical emergency calls. Resuscitation. 2015;96:317-322.

11.

Sasson C, Magid DJ, Chan P, et al. Association of neighborhood characteristics with bystander-initiated CPR. N Engl J Med. 2012;367(17):1607-1615.

12.

Moon S, Bobrow BJ, Vadeboncoeur TF, et al. Disparities in bystander CPR provision and survival from out-of-hospital cardiac arrest according to neighborhood ethnicity. Am J Emerg Med. 2014;32(9):1041-1045.

13.

Sasson C, Haukoos JS, Ben-Youssef L, et al. Barriers to calling 911 and learning and performing cardiopulmonary resuscitation for residents of primarily Latino, high-risk neighborhoods in Denver, Colorado. Ann Emerg Med. 2015;65(5):545-552 e542.

14.

Blewer AL, Putt ME, Becker LB, et al. Video-Only Cardiopulmonary Resuscitation Education for High-Risk Families Before Hospital Discharge: A Multicenter Pragmatic Trial. Circ Cardiovasc Qual Outcomes. 2016.

15.

Einspruch EL, Lynch B, Aufderheide TP, Nichol G, Becker L. Retention of CPR skills learned in a traditional AHA Heartsaver course versus 30-min video self-training: a controlled randomized study. Resuscitation. 2007;74(3):476-486.

16.

Ikeda DJ, Buckler DG, Li J, et al. Dissemination of CPR video self-instruction materials to secondary trainees: Results from a hospital-based CPR education trial. Resuscitation. 2016.

17.

Bosley JC, Zhao NW, Hill S, et al. Decoding twitter: Surveillance and trends for cardiac arrest and resuscitation communication. Resuscitation. 2013;84(2):206-212.

CC

EP

TE

D

M

A

N

U

SC RI PT

10.

A

18.

Leary M, McGovern S, Dainty KN, et al. The Latest in Resuscitation Science Research: Highlights from the American Heart Association's 2017 Resuscitation Science Symposium. Journal of the American Heart Association. 2018;7(2).

19.

Sinnenberg L, DiSilvestro CL, Mancheno C, et al. Twitter as a Potential Data Source for Cardiovascular Disease Research. JAMA Cardiol. 2016.

20.

Twitter Usage. 2017; https://about.twitter.com/company.

McGovern et al – Characterizing barriers to CPR training via Twitter

Page 12 of 12

Figure Legend Figure 1. Vertical bar chart displaying positive and negative Tweet sentiment by CPR training subdomain.

A

CC

EP

TE

D

M

A

N

U

SC RI PT

Figure 1. CPR Training Tweet sentiment by subdomain (n=150)

McGovern et al – Characterizing barriers to CPR training via Twitter

Page 13 of 12

Table 1. Descriptive Tweet Data

Sample of 1000 n=1000

Retweets, n (%)

2,280 (27.1)

296 (28.1)

CPR Training Domain n=187 9 (4.8)

Links, n (%)

2,856 (33.9)

386 (36.7)

26 (13.9)

Followers, median [IQR]

1567 [18068]

1179 [6364]

433 [421]

User Friends, median [IQR]

952 [13514]

682 [1565]

422 [516]

Potential Exposure

13,194,140

1,240,308

80,971

SC RI PT

Full Dataset n=8419

Tweet Sentiment Positive

71 (38.0)

Neutral

37 (19.8) 79 (42.2)

N

U

Negative

News Media

186 (18)

Metaphor Miscellaneous

Link, n (%)

9 (5)

26 (14)

90 (48)

149 (80)

169 (16)

68 (40)

11 (6)

116 (11)

29 (25)

16 (14)

97 (9)

25 (28)

72 (74)

69 (7)

8 (12)

13 (19)

Informational

67 (6)

29 (43)

42 (63)

Irrelevant

55 (5)

12 (22)

28 (51)

CPR on an animal

34 (3)

6 (18)

14 (42)

Entertainment Media

25 (2)

7 (28)

3 (12)

CPR Event

18 (2)

1 (6)

1 (5)

CPR Concerns

17 (2)

5 (29)

6 (35)

Humorous

12 (1)

7 (58)

5 (42)

A

CC

EP

Certification

TE

Advertisement

M

187 (18)

D

CPR Training

A

Table 2. Domain of Tweets arrived by coding Domain n (%) Retweet, n (%)