Demonstrating transportability of real-world evidence: what do HTA bodies and health technology developers need?

Attention-deficit hyperactivity disorder (ADHD) is a neurodevelopmental disorder characterized by high levels of hyperactivity, impulsivity, and/or inattention (American Psychiatric Association, ). The precise etiology of ADHD has not been identified, but prevailing neurobiological ADHD theories assume that cortico-striatal dysfunction of the catecholamines dopamine (DA) and norepinephrine (NE) alters cognitive functioning and reward processing (Frank, Santamaria, O’Reilly, & Willcutt, ; Sagvolden, Johansen, Aase, & Russell, ; Sonuga-Barke, ; Tripp & Wickens, ), which in turn may cause ADHD symptoms. The central role of catecholamines in theories of ADHD is driven by the fact that ADHD symptoms and associated cognitive impairments are ameliorated with methylphenidate (MPH) (Coghill et al., ; Fredriksen & Peleikis, ), a central stimulant that increases synaptic availability of DA and NE (Del Campo, Chamberlain, Sahakian, & Robbins, ; Volkow, Wang, Fowler, & Ding, ). ADHD is associated with impaired real-life decisions, including reckless driving, substance abuse, and risky sexual behavior (Barkley, Murphy, Dupaul, & Bush, ; Faregh & Derevensky, ; Flory et al., ; Lee et al., ; Molina & Pelham, ; Sarver, McCart, Sheidow, & Letourneau, ). An important feature of most real-life decisions is the need to balance costs and benefits to determine the net value of actions. Upgrading to the newest phone is rewarding but comes with the cost of having less money for other expenses. It is unclear how participants with ADHD deviate from the normal population in balancing costs and benefits. Here we set out to investigate the mechanisms underlying cost-benefit decision-making in ADHD, and how these processes are influenced by MPH. There are several ways in which alterations in underlying cognitive processes could cause alterations in observed cost-benefit decision-making, including altered sensitivity to costs and/or benefits, or in the tradeoff between making fast but more erratic versus slow and more accurate decisions. Computational models can be used to identify and disentangle the cognitive processes underlying behavior (Kriegeskorte & Douglas, ), and further explore how these processes are affected in clinical groups (Huys, Michael Browning, Paulus, & Frank, ; Maia, Quentin, & Frank, ; Pedersen et al., ). Here, we apply the drift diffusion model (DDM) (Ratcliff, ; Ratcliff & McKoon, ), a computational model of decision-making, to capture the dynamic processes underlying choosing to accept or reject offers based on the associated costs and benefits of alternatives. The drift diffusion model decomposes choice and response times into components that make up a process model of decision-making supported by behavioral (Ratcliff & Smith, ) and neurobiological data (Forstmann, Ratcliff, & Wagenmakers, ; Gold & Shadlen, ; Steinemann et al., ). The model explains fast choosing between two alternatives as a process of continuous information accumulation until reaching a response threshold in favor of one of the alternatives (Ratcliff & McKoon, ) (). The drift rate captures the speed of evidence accumulation dependent on the quality of available evidence and the decision-makers’ ability, level of attention, and effort. The amount of evidence required to reach a decision is modeled by the decision threshold parameter, representing a tradeoff between speed and accuracy. The intercept of the drift process captures preference for a decision alternative, while time spent on sensory encoding and motor execution is accounted for by the nondecision time parameter. In the current task, participants decided whether to accept or reject a stimulus based on its overall value (), which was determined by learned associations between shapes and benefits, and colors and costs (or vice versa, counterbalanced across participants). For this cost-benefit decision-making task, the drift rate on each trial was estimated as a weighted combination of benefit and cost values, while decision thresholds represented choosing to either accept or reject the stimulus. The starting point parameter represented whether participants displayed tendencies to favor accepting or rejecting before seeing the stimulus on the current trial.