Comments on “Stress, Anxiety, and Depression Before and Twelve Months After Bariatric Surgery: Repeated Cross-sectional Study”

Dear Editor,

ElBarazi presented a large Egyptian study that examined stress, anxiety, and depression in relation to bariatric surgery. ¹ The study found that one year after surgery, subjects had significantly increased levels of stress and anxiety and significantly lower levels of depression. The study was discussed in detail, and many reservations were expressed about its conclusions. We present our concerns in this letter.

The author drew far-reaching conclusions based on scores from a single instrument that, at follow-up, was emailed to the study participants. Readers are not told whether the version of the instrument employed had been standardized for use in Egypt. It is debatable whether a single, self-rated instrument can capture important clinical information about complex mental health constructs such as stress, anxiety, and depression. In particular, for more valid conclusions about anxiety and depression, patients would ideally need to be rated in person by an experienced interviewer and diagnosed using formal diagnostic criteria and not assessed remotely using a self-administered instrument. Notably, is that clinical interviews and diagnoses were not done even at baseline.

The baseline assessments could have been biased by anxieties related to the anticipation of surgery. ² The endpoint assessments are not comparable with the baseline assessments because they were administered differently (by email and hence vulnerable to responder bias) and in a different environment.

Regarding the follow-up assessment, patients who responded to the email may have differed systematically from those who did not. In this context, the author did not examine whether at least baseline characteristics were similar in those who responded and those who did not.

Given the study design, the descriptions for statistical methods and sample size estimation did not seem appropriate. Given the pre-post assessments, a paired t-test was all that was necessary, and for this, to detect a medium effect size of 0.5, a sample size of 34 would have sufficed. ³

The description of the results is incorrect. In table 2 in the published paper, the statistic of interest is the mean difference and not an incidence rate ratio. The findings in this table were probably statistically significant because of the large sample size for a paired t-test; the differences in pre- versus post-ratings were probably random (explaining why the depression scores moved in a different direction from the stress and anxiety scores) and too small to be clinically significant.

It is not possible to conclude from these study methods that bariatric surgery and not a host of other covariates explained the results. Data from a longitudinal study design such as this would need to be examined in the context of the success of the surgery, complications of the surgery, restrictions related to the surgery, comorbid conditions, presence of psychosocial support, interventions received, and a host of other confounds present at baseline or developing during follow-up.

As a final point, the correct description of the study design is that this was not a repeated cross-sectional study but a cohort study with patients assessed at baseline and at a 1-year follow-up.