Exploring patient-centered perspectives on suicidal ideation: A mixed methods investigation in gastrointestinal cancer care

Many patients diagnosed with gastrointestinal (GI) cancer experience
emotional struggles, including feelings of depression and suicidal
thoughts, yet these psychological needs often remain overlooked. This
study was designed to better understand the mental health challenges faced
by GI cancer patients—especially after treatment ends—by listening
directly to their experiences through surveys and interviews. The data for
this study were collected through a validated web-based survey. Any
individual in the US with GI cancer (undergoing or completed treatment)
was eligible to participate in this study. In phase 1, a web-based survey
was distributed to cancer patients through social media and open cancer
forums via an audience paneling service, Centiment. The survey contained
validated questions adapted from the Columbia Suicide Severity Rating
Scale (C-SSRS) and the Patient Health Questionnaire (PHQ-4). A control
question was included to maintain response integrity: “To confirm
attentive reading and thoughtful responses, please select Green as your
answer.” Responses failing this criterion were excluded from the analysis.
The descriptive statistics of all the variables were calculated in R (JASP
Version 0.19.3). To ensure validity and reliability of the survey
measures, the questions adapted from PHQ-4 and C-SSRS were converted into
latent constructs, namely mental health (Factor 1) and suicidal ideation
(Factor 2), respectively. The constructs were validated using multigroup
confirmatory factor analysis across both groups (participants from rural
Appalachia and others). Model fit was assessed by the comparative fit
index (CFI), Tucker Lewis index (TLI), and Bentler Bonnet Normed fit index
(NFI). Additional measures such as the parsimony normed fit index (PNFI),
Bollen’s relative fit index (RFI), Bollen’s incremental fit index (IFI),
and relative non-centrality index (RNI) were also observed. Reliability
and validity were assessed using Cronbach’s alpha, which was required to
be greater than 0.70; (b) outer loadings greater than 0.50 Fornell-Larcker
criterion, and the Heterotrait–Monotrait (HTMT) ratio required to be less
than 0.85. Upon validating the latent constructs, the scores of each
construct were extracted, and the dataset was scaled. Given the non-normal
distribution of the data (based on the Shapiro–Wilk test), we used
Bayesian Pearson’s Rho correlation analysis as it is more robust to
non-normality than the classical Pearson correlation. Two hundred
and two individuals participated, with 76 from the rural Appalachian
region and 78 undergoing treatments. Quantitative analysis showed a higher
prevalence of passive suicidal ideation than active planning. The
post-treatment recovery period was the most emotionally challenging.