The Pain App Study: A Novel Shared Decision Making Tool for People With Chronic Pain

Having completed our formative work, development and refinement of PainApp, the investigators will evaluate it in a small pilot randomized controlled trial (RCT) on a representative sample of patients with chronic pain and their HCPs. The investigators will conduct a small RCT, including up to 50 patients and 5-15 HCPs. Eligible patients will be enrolled online. After informed consent, they will be randomized to either Pain-APP or the control group, which will consist of online educational materials (TheACPA.org). Patients in both groups will be assessed before, immediately after viewing the tool, shortly after the index clinic visit, and 1 month later. Patient-reported outcomes include patient-provider communication, pain intensity and interference, physical function, and depression (See Table 3). The investigators also include qualitative patient evaluations on a sampling (\<20%) of patients to capture the full range of its impact and to explore potential mechanisms of action. Structured phone interviews, including open-ended questions, will assess patient perceptions of the platform, perceptions about how it affected communication and decisions making about treatments, trust in their HCP, and self-efficacy. HCPs will be surveyed shortly after the patient encounter to assess acceptability of the tool, its perceived utility, and its impact on clinical efficiency. Additionally, quality of the patient encounter, ease of assessing patients with pain and of making decisions regarding treatments to manage that patient's pain will be evaluated using previously developed instruments. The Study's Main Hypothesis: Compared to control, patients assigned to Pain-APP will have better patient-HCP communication. Phase I data analyses plan: Multidimensional scaling was used to analyze the Nominal Group Technique (NGT) and card sort data. Hierarchical cluster analysis was used to identify subgroups of homogeneous strategies. Cognitive maps reflecting different patterns of goals and shared decision making (SDM) strategies were developed for patients and HCPs. These analyses helped the investigators understand how goals/strategies are viewed and organized by our stakeholders, which informed the design of PainApp. Phase II data analysis plan: For the RCT, the primary outcomes analyzed will be patient-provider communication (COMM), pain (PAIN), and interference (FUNCTION), all of which will be treated as continuous. The investigators will compare Pain-APP and the control intervention. The investigators will adhere to accepted standards. Analysis will be done in 4 stages: The investigators will examine 1) whether variables meet expected distributional assumptions; 2) bivariate associations between treatment arms and each outcome, and between covariates and outcomes; 3) unadjusted measures of effect and potential confounders; 4) exploratory multivariable analyses. Patients who are lost to follow-up will be treated according to their assigned group (intent to treat analysis). The investigators will use analysis of covariance to assess differences in outcomes between assigned groups, adjusting for any baseline differences between the randomized comparison groups. Covariates will include baseline health status, moderators listed in Table 3, and prescriber characteristics. The investigators will use linear mixed-effects models, treating scores at different points in time as repeated measures to account for inter-correlation between these measurements. The investigators will control for patient- and HCP-level factors by including them as covariates. ii) Quantify the amount of change expected: The investigators expect to find, at minimum, a 20% improvement in communication, comparing the difference in communication pre- post- intervention in the Pain-APP group to that of the control group. Power analysis: The investigators expect a mean of 70 and a SD of 20. With the proposed sample size of 50 (25 in each arm), the investigators estimate having an 80% chance of detecting a 20% improvement on the primary outcome COMM at time=T1, with a type-I error rate of 0.05.