Comparing survey data to qualitative themes in patient stories to help researchers better identify pressing community healthcare needs.

Background: Narrative medicine is a powerful approach for strengthening clinicians' therapeutic alliances with patients and helping them understand others' perspectives by giving patients a voice. MyPaTH Story Booth is a cross-sectional observational study that uses narrative medicine concepts to allow community participants to share their health and healthcare-related experiences through relatively unstructured interviews (i.e., stories). The archive forms a qualitative research 'commons' where researchers can learn about patient and caregiver perspectives and tailor research goals to better address community needs. Brief surveys allow storytellers to classify their stories so that the archive is searchable.

Methods: This study assessed sensitivity, specificity, positive predictive and negative predictive values of self-reported surveys for categorizing 83 stories, compared with a 'gold standard' (most commonly accepted method for interpreting narratives) of qualitative coding. Stories were classified as either having or not having each of 38 different classifications based on US National Library of Medicine (NLM) Health Topics (Type of disorder or condition and Part of the body discussed) or developed with stakeholder input (Type of health/healthcare experience).

Results: Survey data assessing 'Part of the body discussed' and 'Type of disorder or condition' typically was highly sensitive (64%-95%) and specific (73%-99%) suggesting high adequacy of surveys in identifying stories. Questions focused on 'Type of health/healthcare experience' showed inconsistent and relatively lower sensitivities (48%-93%) and specificities (30%-79%) suggesting lower adequacy of surveys in identifying stories.

Conclusion: Survey items based on NLM Health Topics allowed people to categorize their health narratives into relevant, searchable topics making the database more accessible.