Anxiety and Depression Among Cancer Survivors: The Role of Engagement With Sources of Emotional Support Information

Abstract

This study explores cancer survivors' engagement with information about emotional support from doctors, interpersonal sources, and the media and examines to what extent such engagement affects subsequent self-reported anxiety and depression. Patients with colorectal, breast, or prostate cancer (n = 1,128) were surveyed over 3 years following diagnosis. Using lagged logistic regression, we predicted the odds of experiencing anxiety or depression based on earlier engagement with sources of emotional support, adjusting for prior symptoms and confounders. Among those reporting anxiety or depression (n = 476), we also asked whether information engagement affected the severity of those symptoms. Participants obtained information about emotional support from multiple sources, but most often from physicians. Discussions with physicians about emotional support increased the odds of cancer survivors subsequently reporting anxiety or depression by 1.58 times (95% CI: 1.06 to 2.35; p = 0.025), adjusted for prior symptoms and confounders. Scanning from media sources was also significantly associated with increased odds of reporting emotional symptoms (OR=1.72; 95% CI: 1.03 to 2.87; p = 0.039). However, among those who reported symptoms, doctor–patient engagement predicted slightly reduced interference of these symptoms with daily activities (B = −0.198; 95% CI: −0.393 to −0.003; p = 0.047). Important implications for health communication research and practice are discussed.

Notes

¹We used available demographic and disease status information at baseline as independent variables for a regression model predicting whether participants completed the round 3 survey. Analysis showed that participants who were older (vs. younger), black (vs. white), diagnosed with stages III–IV cancer (vs. stages 0–II), female colon cancer patients (vs. prostate cancer patients), or who had a high school and below education (vs. some college and above) were less likely to complete the survey at round 3. However, this set of 12 predictors accounted for a small amount of the variation in the likelihood of participation (Nagelkerke pseudo r-squared = .114). All of these variables were included in the final regression models subsequently presented and thus their influence was taken into account.

²Because of the presence of gender-specific cancer types, we combined gender and cancer type into a single covariate such that four categories were adjusted for in the analysis (female colorectal, male colorectal, breast, and prostate cancers). This did not influence the substantive findings in the analysis.

³The time period for the item capturing physician discussions was longer than that of the seeking and scanning items (an average of 27 months vs. the past 12 months). Despite the longer time frame, we argue that the observed difference between the proportion of patients seeking, scanning, and discussion with physicians is real, given that people, in general, have more opportunity for interaction with interpersonal and media sources. This opportunity, in a sense, counterbalances the greater length of time.

⁴The observed difference may be attributable to how physician engagement was measured. This particular survey item did not differentiate between physician discussions that were actively sought versus scanned. Accordingly, comparisons should be interpreted with caution.

⁵We conducted an additional sensitivity analysis to ensure that the observed effects could not be attributed to an oversampling of stage IV and metastatic patients. In this study, these two indicators of disease severity were not significant predictors; however, both coefficients tended toward the positive direction, suggesting that this group may have a higher tendency to experience emotional symptoms. Therefore, we filtered out these patients (n = 184) and reran the same set of analyses on non-stage-IV, non-metastatic patients. As expected, the effects of physician discussions and media scanning on anxiety/depression for this subpopulation were slightly reduced in strength, but remained positive and statistically significant (results not shown), which reinforces our original results.