Validating a single-item screener for low digital health literacy among surgical oncology patients.

Abstract

1634 Background: Digital health literacy (DHL; the ability to find and understand health information using electronic media) may affect surgical oncology patients’ ability to navigate patient education and discharge coordination activities that rely on technology. Screening for low DHL can identify patients who may benefit from alternate modalities or additional support. In our prior work, a single item from the validated electronic Health Literacy Scale (eHEALS) had strong sensitivity and specificity for identifying low DHL in the outpatient setting. In this cross sectional, secondary analysis, we explored which sociodemographic and clinical factors were related to low DHL and tested whether the single-item screener could identify patients with low DHL in the surgical oncology setting. Methods: Surgical oncology patients were recruited during their post-operative hospitalization for a parent study on health literacy and clinical outcomes. The 8-item eHEALS assessed DHL; total scores ≤20 indicate low DHL. Chi-squared, Fisher’s exact, or Mann-Whitney tests were used to examine whether DHL varied by patients’ sociodemographic (age, sex, race/ethnicity, education, marital status, general health literacy) or clinical characteristics (cancer type, ECOG performance status, comorbidities). Receiver operating characteristic (ROC) curve analysis assessed sensitivity, specificity, and accuracy of eHEALs item 4 ("I know where to find helpful health resources on the Internet") to identify low DHL. Results: Participants (N=224) were primarily non-Hispanic White (87%), older (median=68, IQR 61-75 yrs), married (69%), men (59%) with gastrointestinal cancer (60%) who completed at least some college education (76%). On the 8-item eHEALS, 23 patients (10.3%) reported low DHL. Patients with lower general health literacy (p<0.001), lower education level (p=0.007), and unpartnered marital status (X 2 = 4.359, p=0.039) were more likely to have low DHL. Other factors were not associated with DHL (p-values>0.299). Older patients were marginally more likely to report low DHL than younger patients (U=2847, p=0.069). In ROC curve analysis, eHEALS item 4 strongly distinguished low DHL vs. high DHL (AUC=0.989; 95% CI 0.977-0.997). At the optimal threshold (item scores <3, on a 1-5 Likert scale), item 4 predicted low DHL with an accuracy of 96.9%, sensitivity of 97.0%, and specificity of 95.7%. Conclusions: These data suggest a single-item screener (eHEALS item 4) can accurately identify patients with low DHL in the surgical oncology setting, adding support for its ability to detect low DHL in patients with cancer. Among surgical oncology patients, those with lower education, lower general health literacy, and older age may be more likely to experience lower DHL. Future prospective studies with larger samples are warranted to assess utility, acceptability, and feasibility of the screener in clinical practice.