Learning points from a study on cyberchondria

To the Editor, In an online survey of cyberchondria and its correlates in medical and nonmedical students, Agrawal et al.[1] found that cyberchondria was associated with smartphone addiction, that it was more severe in nonmedical students, and that it was associated with subjective well-being in nonmedical students but not in medical students. The study was discussed in eJournal Club India,[2] where several important points were raised about the design and analysis of the study and the interpretation of the findings. We believe that there are learning points from this study that would benefit the authors, readers, future researchers in the field, and reviewers. First, a proper method of sampling is one of many cornerstones of good research; it allows researchers to generalize to the population the findings obtained from a sample. In online surveys, because there is no understanding of the nature of the population from which the sample was drawn and because of the risk of selection bias during the sampling process, the findings of an online survey merely describe the sample and cannot be generalized to any real-world population.[3] Online surveys should therefore be discouraged as a method of sampling and should be resorted to only when standard methods of sampling, for whatever reason, are not possible. Second, unless there is a justifiable reason to do so, continuous data should not be categorized as Agrawal et al.[1] did with regard to the representation of the duration of smartphone use. This is because, inter alia, categorization results in loss of precision and decreased statistical power.[4] Third, when a sample is sufficiently large, even small effect sizes may be statistically significant; this does not mean that they are clinically significant. There are two illustrations of this precept in the findings of Agrawal et al.,[1] whose study had a large sample of 1033 respondents. The statistically significant differences in cyberchondria total scale and subscale scores were very small (mostly lying in a decimal point range) and appeared unlikely to represent a meaningful difference between medical and nonmedical students, and the significant association between cyberchondria and well-being in nonmedical students explained <2% of the variance in the relationship, which is again unlikely to represent a meaningful relationship in the real world. We hope that our observations will be useful to readers, researchers, and reviewers. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.