Page 287 - SAHCS HIVMed Journal Vol 20 No 1 2019
P. 287
Page 3 of 9 Original Research
Objectives non-responders possibly skewing the results. Forty-five
17
per cent (74/166) of the participants did, in fact, not submit
The specific objectives are:
responses. The investigator attempted to minimise this
• to analyse clinicians’ use of the WhatsApp chat group as threat as much as possible by regular email and WhatsApp
a learning tool reminders. The questionnaire was kept as short as possible
17
• to assess clinicians’ confidence in managing complicated and attempts were made to simplify access to it with an easy
HIV/TB patients after participating in the WhatsApp to use Internet link being sent to the participants – all this to
case discussion group minimise non-responses. 17
• to describe the perceived usefulness of the chat group as
a learning tool A link to the questionnaire in the Google form was initially
• to understand clinicians’ knowledge and use of informed sent to each clinician in the WhatsApp discussion chat group
consent when sharing patient case details on a public via WhatsApp. When the clinicians clicked on the link, they
platform such as WhatsApp.
were taken to the electronic Google form. Google saved each
Research methods and design completely filled questionnaire in the investigator’s Google
drive. This form was completed by the respondent by a click
Study design on the most appropriate response. There were no open-ended
or continuing questions, making the questionnaire simple
An observational, descriptive cross-sectional design was
used, with an anonymous Internet questionnaire, distributed and fairly quick to answer; the investigator estimated around
to the clinicians who formed part of the WhatsApp group, as 5 to 10 min per form. Participants were able to answer the
the data source. A quantitative approach was chosen for the questions within their own time frame, enabling them to
study as the responses from the questionnaire were graded have privacy or choice of space.
and therefore easily quantifiable.
All the completed forms were available to view on the drive,
Study population, setting and sampling which was password protected, and could be downloaded
when needed for analysis. The clinicians were also emailed
The study population that was used in this study were 166 the link as well. Emailing helped to collect data from the
doctors from the EC province that accepted the organiser’s clinicians that may have at any stage left the group during
invitation to be part of the WhatsApp clinical discussion the period under investigation.
group from January 2016 to July 2017. The inclusion criteria
for the study included doctors from the EC Department
of Health, as well as clinicians from collaborating non- Data management and analysis
governmental organisations (NGOs). All the 166 doctors in The individual responses saved on Google drive were
the group were included to minimise any non-response, and collected and transferred to an Excel spreadsheet, where data
to improve representation of the clinicians in the group. 17 cleaning occurred. Any incompletely answered questionnaire
was removed as a data source. Text responses were also
Data collection tools and collection allocated a numerical key for easier analysis. The data were
then imported into Epi info statistical programme for analysis
Data were collected using a structured, anonymous Internet and were initially explored using basic frequencies for the
questionnaire. This comprised 17 statements or questions, categorical data.
each with a corresponding answer or choice of answers. The
main themes for the questionnaire centred on access to the Summary statistics were presented to give a general
WhatsApp or Internet; usage of the group; aid in improving description of the above responses using analysis tables and
clinical confidence; usefulness as a learning tool and the
confidentiality of cases posted (doctors’ perceptions). graphs. These categorical variables were summarised as the
number and percentage of responses in each category or
To reduce information bias, the investigator used a exposure variable. Further analysis was done by looking
standardised tool, and each doctor received the same at other possible associations between clinical confidence
questionnaire. The questionnaire had been reviewed by a to group engagement and clinical confidence because of
group of three colleagues to ensure clarity and the exclusion perceived usefulness of the group as a learning tool. In
of external bias. Once ethical approval was received and the confidence variable, like–like response options were
before distribution to the participants, the questionnaire in its recorded for ease of analysis. Other associations included the
electronic format was piloted with the same colleagues who recommendation of the group based on the perceived
are a part of the WhatsApp group to further improve question usefulness of the group as a learning tool. For all the above
clarity and ease of participation. The questionnaire was self- associations, frequency distributions and cross-tabulations of
administered, so no measurement bias was introduced by a the above-mentioned variables were generated. Bivariate
third party. 18 analysis was done to determine significant associations
between the differing variables using p-values, odds ratios
There was a threat to validity in terms of sampling bias (ORs) and 95% CIs. The assessment of any significant
when administrating the questionnaire, with the potential of differences was conducted using a Mid-P Exact test. This was
http://www.sajhivmed.org.za 280 Open Access