Effect of Multiple Suggested Care Alternatives on Decision-Making in Primary Care Physicians

Physician participants were presented with two management plans for clinical scenarios commonly seen in primary care. Each scenario – one about a surgery referral for hip osteoarthritis, one on opioid prescribing for low-back pain – involved a decision about whether to remain with an existing management plan or to select an alternate plan. In the intervention condition physicians could receive two, three, or four appropriate alternatives. To further control for physician preferences, alternatives were randomly selected from a longer list of appropriate alternatives for each participant. Scenarios were about half a page in length, presented online in written format and no time limit was imposed. The clinical scenarios were co-designed during protocol development with the investigator team that included primary care physicians, behavioural economists, and investigators with expertise in research methods. Guidance on designing experimental vignette studies to identify drivers of healthcare variation was followed (Sheringham, 2021). Scenario 1 featured a patient with chronic hip pain and osteoarthritis and replicated the Redelmeier and Shafir (1995) study, with updates to ensure clinical relevance in 2024 (eg drug names, patient’s profession). In this scenario the patient had been seeing a physiotherapist and walked daily. They had tried one NSAID but stopped due to limited efficacy. The existing management plan (or status-quo option) in this scenario was to refer the patient to an orthopaedic surgeon, without trying a new NSAID. The alternative options presented were to continue with the referral but to also start the patient on an NSAID. Control physicians were presented with one NSAID alternatives and intervention physicians were presented with two NSAID alternatives. Scenario 2 featured a patient with chronic low-back pain. In this scenario the patient had been managing their back pain with physiotherapy and regular exercise over the last few months. Approximately 2 weeks ago the patient received a 3-day supply of an opioid analgesic (oxycodone) to help manage a flare up. In this scenario the patient had requested another 3-day supply of oxycodone but when the physician initiated the order an alert was triggered, suggesting that they consider an NSAID medicine instead. The status-quo option was to continue with the opioid analgesic and the alternative was to try an NSAID. Control physicians were presented with one NSAID alternative and intervention physicians were presented with two, three or four NSAID alternatives. Physicians were randomized (1:1) to control or intervention conditions using computerized randomization via the Qualtrics research platform. Physicians in the intervention groups were further randomized (1:1:1) to receive two, three or four alternatives for scenario 2. To ensure data integrity, the independent data manager removed responses that failed their data quality standards (eg rapid completion time or failed attention checks) before providing the final dataset for analysis.