Statistical analysis

Participant characteristics were summarized using frequencies and proportions for categorical variables, and means and standard deviations or medians and interquartile ranges for continuous variables, as appropriate. Ineligible responders and incomplete responses (from unsubmitted surveys) were excluded. We measured the prevalence of self-reported mental health problems for patients with rheumatological disease, stratifying responses by mutually exclusive groups for clinician-diagnosed disorders versus self-diagnosed symptoms, and tabulated reported mental health treatments received by these groups. For participants indicating use and helpfulness of mental resources, qualitative text responses were analyzed using qualitative description to summarize positive or negative experiences with mental health resources, and illustrative text responses were selected. Qualitative description is a low-inference approach that presents the facts from exactly the informants’ point of view. It is the qualitative method of choice when only a description of a phenomenon is desired, and is particularly useful in health care research to focus on the patient experience, for the purpose of needs assessment and intervention development [26–28].

We calculated mean Likert scores for level of comfort with potential mental health providers, and frequency of barriers to seeking mental health treatment. Using multivariate linear regression to control for potential confounders (patient age, gender, disease duration, and patient/parent VAS), we used two separate models to compare patient and parent responses for ratings for comfort level with mental health providers, and frequency of barriers to mental health treatment. We also conducted two stratified regression analyses to compare responses by:1) patient age group (14–17 versus 18–24 years) to determine if the results differed between adolescents and young adults; and 2) presence or absence of reported mental health problem (either clinician- or self-diagnosed), to determine if actual experience with a mental problem impacted the results. Analysis was conducted using Stata^(R) Data Analysis and Statistical Software version 15.0. Alpha ≤0.05 denoted a statistically significant result.