2.5 Data Analysis
The researchers reviewed verbatim transcripts and identified broad themes drawing on the interview guide topics to serve as top-level codes. Examples of top-level codes included factors influencing provider’s rationale for current treatments as well as those factors influencing treatments that were avoided. Using inductive and deductive analysis with constant comparison methods that Parker and colleagues developed for similar qualitative studies, 3 members of the interview team (EB, NB, AJ) each coded one-third of the transcripts and reviewed 100% of each other’s work.16 We created sub-codes by reviewing the quotes associated with each top-level code. Each interviewer took one-third of the quotes and applied the sub-codes. A secondary coder reviewed 25% of the sub-codes and discrepancies were discussed and resolved. We used Atlas.ti V6.2 qualitative data management software to assist with coding and analysis.
Results from the closed-ended questions were included for descriptive purposes only.
RESULTS
Demographic information for the participants is presented in Table 2. Several key themes emerged regarding factors that contributed to either new prescriptions of benzodiazepines or maintenance of existing prescriptions in veterans with PTSD. We categorized prominent barriers and facilitators as organizational, provider, and patient factors (See Table 3). Beliefs around these factors were similar between PC and MH providers. Where they are not, we describe these differences. Identified facilitators contributed to both decreased incidence and prevalence of benzodiazepine prescriptions, suggesting that such factors could help de-implement the use of benzodiazepines by reducing new prescriptions as well as renewal prescriptions.