There was no bias in demographic variables like gender, age, socioeconomic status, and time from diagnosis among online forum users with gastrointestinal stromal tumors (GISTs), found a new study published in the Journal of Medical Internet Research. However, there was a statistically significant difference between forum users and nonusers in terms of the self-reported treatment phase.
“The consequent overrepresentation and underrepresentation of certain types of patients should be considered when sourcing patient forums for patient-generated health data,” the study authors concluded.
Read more about GIST
Social media and online forums are increasingly being recognized as potential resources for health data generated by patients.
Here, a team of researchers led by Wessel Kraaij, PhD, from Leiden Institute of Advanced Computer Science at Leiden University and The Netherlands Organisation for Applied Scientific Research investigated whether there was sample bias in patient-centered social media among patients with GIST in the Netherlands.
The team conducted a population-based survey of 328 patients with GIST who were diagnosed between 2 and 13 years previously to understand their use of digital channels to communicate with other patients.
The results showed that 17.9% of patients used social media to contact another patient with GIST. The majority of these patients (78%) also reposted using forums for patients with GIST.
There were no statistically significant differences in terms of age, gender, socioeconomic status, and time from diagnosis between patients who did and do not use forums.
When they analyzed the answers of the respondents in terms of the treatment phase, the researchers found that only 4% of forum users “were cured and not being monitored,” and only 7% were receiving adjuvant curative treatment. However, almost half of them (41%) were being monitored following adjuvant treatment and 48% were receiving palliative care.
On the contrary, a much larger proportion (20.5%) of non-forum users “were cured and not being monitored,” while 11.3% were receiving curative treatment. Like forum users, half (50.9%) of non-forum users were being monitored after treatment but only 15.3% were receiving palliative care.
The researchers calculated the likelihood of being on a patient forum to be 2.8 times higher for a patient who was being monitored than a patient was who cured. Similarly, the likelihood of being on a patient forum was 1.9 times high for a patient on curative treatment and 10 times higher for a patient receiving palliative care than a patient who was cured.
“Sample bias is inherent to any information source, and only through awareness of these biases can these resources be used as a source for complementary real-world evidence in the future,” the authors concluded.
Dirkson A, den Hollander D, Verberne S, et al. Sample bias in web-based patient-generated health data of Dutch patients with gastrointestinal stromal tumor: survey study. JMIR Form Res. Published online December 15, 2022. doi:10.2196/36755