A randomized controlled trial of self-directed learning with Twitter for ultrasound image interpretation

Participants are instructed to follow ten Twitter accounts that consistently post point-of-care ultrasound content for 3 months, using Twitter for at least 2 hours per week during the duration of the study.

2021-11-20

2024-07-31

Change in assessment score

Change from post- to pre-intervention assessment score, defined as percent correct of the 15 multiple choice question assessment.

Average post-intervention assessment score
Perceptions of Twitter as an educational tool
Hours using Twitter

Perceptions obtained by Likert-style questions on a 1-5 scale:
How do you view Twitter as a medical education resource?
How often do you learn concepts from Twitter that are applicable to clinical practice?
How likely are you to use Twitter as an educational resource?
What do you feel are the educational aspects of Twitter?

A parallel, unblinded, randomized controlled trial. Participants will include third- and fourth-year medical students, internal medicine residents, and emergency medicine residents meeting the below criteria.

Inclusion Criteria
• Age greater than or equal to 18
• Ability to access Twitter via personal device
• Active status as a third- or fourth-year medical student, or resident physician
Exclusion Criteria
• Formal training in point-of-care ultrasound outside of usual graduate medical education activities (i.e. certificate or fellowship in point-of-care ultrasound, previous work as ultrasound technician, etc.)
• Already following 5 or more accounts that consistently post content related to point-of-care ultrasound
• Already following more than 5 accounts listed in the intervention arm.
• Engaged, or planning to engage, in POCUS education outside of typical medical education curricula during the study period

Briefly, the intervention will consist of following at least 10 accounts that post consistent content in POCUS image interpretation, while the control arm will consist of following at least 10 medical journal accounts. Both arms will be given the recommendation to spend at least 2 hours per week on Twitter.

The study duration will be 12 weeks, with interval logs reported throughout. The logs will be sent out via email every three to four weeks and query how often participants are utilizing Twitter, as well as a few additional questions (to be developed). Participants will be recommended to obtain their Twitter use via the automatic app use logging on iPhones and most android devices.

At the conclusion of the study, participants will retake the POCUS image interpretation assessment, as well as a post-study survey.
The study will be unblinded due to feasibility.

Participant Timeline
Time 0 – Consent and enrollment. Initial survey and baseline POCUS image interpretation assessment
Week 3 – Interim Twitter log
Week 6 – Interim Twitter log
Week 9 – Interim Twitter log
Week 12 – Interim Twitter log, post-study survey, POCUS image interpretation assessment

All enrollment, randomization, and participation will be done remotely via digital communications.

Not available

Computer generated (via GraphPad) 4-block permuted block randomization method. Investigators enrolling participants are unaware of the randomization sequence until the time of a new enrollment, at which point a blinded investigator notifies the enrolling investigator.

Randomization is stratified by three educational stratifications (1. third- and fourth-year medical students 2. internal medicine residents 3. emergency medicine residents).

No

N/A

N/A

35 intervention, 35 control

A paucity of evidence in this field made an a priori sample size determination difficult. Our primary endpoint is improved POCUS image interpretation, determined by the percentage improved on the image interpretation assessment. We initially aimed to detect a 20% improvement in the intervention arm relative to the control arm, thus estimating a sample size of 32 would be required to detect a difference at the 95% confidence level with 80% power. However, it would be beneficial to the current literature to be able to assess this primary outcome in subgroup analyses of both medical students and residents, thus we aim to reach a final sample size of 70 to ensure at least 32 medical students and 32 residents.