In Multiple Sclerosis, Behavioral Model Predicts Patients Likelihood of Using DMT

February 24, 2019

The findings suggested that a patient’s likelihood of taking medications systematically decreased as the probability of potential AEs occurring increased or the efficacy of treatment decreased.

David Jarmolowicz, PhD

A novel behavioral model has been shown to predict when patients with multiple sclerosis (MS) are likely to refuse treatment with disease-modifying therapy (DMT) due to delayed and fickle efficacy and the probability of potential adverse events (AEs).

Despite the fact that treatment with a DMT can help prevent the symptoms, and disease-progression and activity in patients, only 50% of patients agree to initiate treatment with DMTs, and between 25% and 50% halt treatment against their physician’s advice, according to lead author David Jarmolowicz, PhD, an associate professor of Applied Behavioral Science and the director of undergraduate studies at the University of Kansas.

"In our prior research, we found that both side effect probability and [adverse] effect severity impacted choice," Jarmolowicz told NeurologyLive’s sister publication MD Mag. "In this study, we wanted to know how these behaviorally active factors interacted."

In total, the study included 299 patients with MS at a single center specialty clinic. The patients were asked to complete questionnaires during normal clinic visits, or remotely by connecting to a dedicated server post-visit. Of the almost 300 participants, 73.3% (n = 219) had relapsing MS, 17.5% (n = 52) had secondary progressive MS, 7.9% (n = 23) had primary progressive MS, and 1.4% (n = 5) had relapsing progressive MS.

The behavioral model provided the investigators with the ability to model both the different influences on patients’ treatment adherence as well as their own willingness to take their medication. Jarmolowicz told NeurologyLive’s sister publication that the analytic model is more nuanced than those which have previously been applied, adding that it is capable of accurately describing DMT choices in a larger sample across a wide and interactive parametric space. The model measures patients' propensity to devalue delayed rewards—dubbed "delay discounting” in earlier models—but includes other measures that they postulate may more straightforwardly reflect patients’ medication choices.

Ultimately, the questionnaire provided data on the likelihood of a patient taking a hypothetical DMT, on a scale of 0% to 100%, with the probability of the occurrence of AEs, which included 11 values from 0.1% to 99.99%, and the projected medication efficacy, which also included 11 values from 0.1% to 99.99%.

Participants responded to these queries across a full range of likely medication efficacies for different probabilities of mild, moderate, and severe AEs.

The investigators factored in a higher relevance for probability discounting compared to delay discounting, mainly because, for example, DMTs do not produce immediately discernable improvements, but do decrease the probability of symptoms while carrying a substantial probability of producing AEs. Additionally, Jarmolowicz and colleagues incorporated both the probability and the severity of AEs into the modeling, as well as measures of patients' psychophysical scaling of AEs and efficacy of DMTs.

They reported that a patient’s likelihood of taking medications systematically decreased as the probability of AEs increased or the efficacy decreased, and this pattern of choice remained despite an increase in the degree as the AEs became more severe.

Although this model is suggestive of an approach using probability discounting could advance efforts to predict a patient’s decision to initiate the recommended treatment, Jarmolowicz et al. acknowledged that the design could be even further refined. They noted that future designs with non-hypothetical medications could have a particular utility for specific patients. Additionally, while this study examined variables influencing the decision to initiate treatment, they anticipate future studies will seek to identify factors that influence adherence.

Jarmolowicz said that while the model is indicative of what might be considered instinctual—patients declining the initiation of treatment with a delayed benefit and immediate potential AEs—"intuition is seldom precise." He noted that behavioral economic models often allow for investigators to leverage observations to predict important patient behavior.

“Once we can predict behavior, we are halfway to improving it,” he said.

“The current study identified a behavioral-economic model that described multiple sclerosis patients’ medication choices across a wide range of [AE] probability, [AE]-severities, and medication efficacies,” Jarmoloqwicz and colleagues concluded. “These data should help determine personalized approaches to match patients with medications that they are likely to take.”

A version of this article first appeared on MDMag.com.

REFERENCES

1. Jarmolowicz DP, Reed DD, Bruce AS, Lynch S, Smith J, Bruce JM. Modeling effects of side-effect probability, side-effect severity, and medication efficacy on patients with multiple sclerosis medication choice. Exp Clin Psychopharm. 2018;26(6):599-607. doi: 10.1037%2Fpha0000220