Can Providers be Nudged into Expanding their Skill Set? A State-wide, Two-stage Experiment in North Carolina

Background: Response to public health emergencies often requires a cadre of healthcare providers who have received specialized training. An open question is how to efficiently recruit and train providers in a way that ensures affected communities have sufficient access to appropriate care. Opioid overdoses currently claim the lives of 185 Americans per day. This epidemic is alarming, not just in its effects on the lives of individuals and families, but also on the light it has shined on the disturbing lack of treatment options in many areas. A number of effective treatments are available for opioid use disorder, but barriers to entry may exist. Aims of the Study: We examine approaches to recruit and encourage providers to undergo workforce training to help address the opioid overdose epidemic. Methods: We conducted a state‐wide experiment of nudges and incentives to recruit providers to a learning collaborative on medications for opioid use disorder (MOUD) in North Carolina. We designed the experiment as a two‐stage factorial trial. In the first stage, we randomized 15,835 eligible primary care providers across the state to one of four recruitment arms in a cross‐cutting design. Providers were first randomized to either a ''pro‐social nudges'' messaging arm, in which recruitment materials incorporated specific information on the opioid epidemic in a provider's community, or to a standard recruitment approach which did not include pro‐social language. Orthogonally crossing these two arms, we further randomized providers to a ''financial incentive'' messaging arm indicating that financial reimbursement for participation was available or to an arm that omitted mention of any financial support for participation. In the second stage of the experiment, recruited providers (n=72) were randomized into one of three levels of performance pay tied to participation in learning collaborative sessions. This experimental design allowed us to estimate the effects of each combination of recruitment nudge and performance pay and, importantly, to separate the selection effects of different recruitment nudges from the incentive effects of performance pay. Results: We highlight several main findings. First, we find that prosocial messaging combined with mention of financial reimbursement yielded a significantly higher number of responses than the control group, while each nudge was ineffective in isolation. Second, there was substantial heterogeneity in provider response by type: nurse practitioners were substantially more likely to respond than other types of primary care providers. Third, recruited providers offered performance pay participated in significantly more sessions, with participation approximately linearly increasing in reimbursement size. Discussion: We use our experimental results to analyze how recruitment nudges and partition‐based incentives can be optimized to cost‐effectively recruit and train providers to address the opioid crisis. Beyond this, our results contribute to several strands of academic literature in health, labor, and behavioral economics. Implications for Health Policies: These recruitment methods could be implemented in a low‐cost manner for other regions to enhance the provider workforce.