Fig 1.
Conceptual model for how the mCME intervention was hypothesized to work to improve medical knowledge.
The figure outlines our model for the mCME intervention. The figure outlines two pathways that could lead to our desired output: improvement on the endline exam score. The first is learning from the information within the SMS messages themselves (the weak pathway). However, these are very brief, cover each topic superficially, and so would not be expected to have much effect per se. Rather, the SMS was hoped to serve as a stimulus promoting increased self-study on the same thematic areas as addressed in the messages through lateral learning (the strong pathway).
Fig 2.
Seating diagram for baseline and endline evaluations.
Each cell represents an individual sitting a desk taking a specified version and sub-version of the examination. Each 4x4 block of sixteen seats can be repeated ad infinitum, and will never result in two individuals who are taking the same exam versions sitting adjacent to each other in any direction. This was intended to minimize potential that participants could work together or share answers during the baseline and endline evaluations. It can flexibly be adapted to suit different sized/shaped rooms. E1V1 = Exam version 1, sub-version 1; E1V2 = Exam version 1, sub-version 2. E2V1 = Exam version 2, sub-version 1; E2V2 = Exam version 2, sub-version 2.
Fig 3.
Study flow diagram.
Table 1.
Characteristics of study participants.
Fig 4.
Weekly response rates (Groups 2 and 3).
Group 2 and 3 participants were required to respond to the daily SMS messages. The figure summarizes the weekly average response rates for the two groups separately. As can be seen, participation was high for both groups, but statistically significantly higher for Group 3 participants. Group 1 participants were not asked to respond to the messages.
Fig 5.
Weekly non-conforming answer rates (Group 3 only).
This summarizes the week-by-week average rates of non-conforming answers sent back by Group 3 participants. Responses were interpretable only if in the formats ‘a’, ‘b’, ‘c’, ‘d’, or ‘A’, ‘B’, ‘C’, or ‘D’. All other responses were ‘non-conforming’. Non-conforming answers were most common in the initial few weeks of the study, and then fell to < 10%.
Fig 6.
Weekly correct answer rates (Group 3 only).
The figure summarizes the proportion of Group 3 participants who provided a correct response to the daily SMS question over the 6 months of the study. While there was considerable week-to-week variation, there was no trend to improvement over time.
Table 2.
Comparison of baseline and endline medical knowledge exam mean test scores, and mean changes in test scores between baseline and endline.
Table 3.
Impact of mCME intervention on key secondary endpoints: self-efficacy and job satisfaction.
Table 4.
Impact of mCME intervention on endline self-education behaviors.