Behavioural Economic Perspectives on Opioid Prescribing Patterns in General Practice

Past decades’ liberal prescribing of opioids has led to the development of a global reaching crisis of opioid overuse and abuse due to the strongly addictive nature of this class of drugs. Although the problem, which has been referred to as an epidemic, is most advanced in North America, health authorities worldwide have had to increase their focus on the management of opioid prescribing. There is now increasing recognition that opioids are both less safe and effective against chronic non-malignant types of pain than previously thought, which has led to stricter regulations and clinical guidelines generally recommending against the use of opioids for long term treatment for non-malignant pain. Despite this, over prescribing of opioids remain a persistent problem in many places, and health authorities are in need of additional and better tools to help drive prescribing practices towards guideline concordance. At onset of this thesis, Denmark had the highest rate of opioid prescribing amongst the Scandinavian countries, and despite reductions in overall prescribing, some aspects of the prescribing patterns in Denmark have room for improvement.Approximately 90 % of maintenance prescriptions in Denmark stem from general practice, making this the natural focal point of interest.

The aims of this thesis were first to explore heterogeneity in opioid prescribing practices of Danish general practitioners (GPs) and which factors may be driving it. Secondly, to explore whether nudging can be an effective and feasible tool for regulating prescribing practices in general, and opioid prescribing specifically, in the context of the danish general practice sector, with a focus on heterogeneity in effects and exploring the driving mechanisms.


Paper 1; Estimating prescriber related variation in inappropriate opioid prescribing: A two stage fixed effects vector decomposition model

Understanding the variation in prescribing practices amongst GPs and its associations with GP and general practice organisational characteristics may provide valuable insight needed to design and target interventions aiming to help physicians optimise prescribing practices. In this paper I used a two stage fixed effects vector decomposition model to assess the variation in inappropriate opioid prescribing for chronic non-malignant pain not attributable to practices patient case mix, and its associations with general practice characteristics that may explain this variation. I find that the variation in prescribing is largely unexplained by patient-related factors. Practices with higher average age, higher overall opioid prescribing, with closed patient lists and located in a densely populated areas issue a higher proportion of guideline-inconsistent prescriptions.


Paper 2; The use of nudge strategies in improving physicians’ prescribing
behaviour: A systematic review.

Sub optimal prescribing remains a problem which causes negative consequences in terms of poorer patient outcomes and increased healthcare costs. Through the past decade, nudging has increasingly gained popularity as a tool for influencing the decisions of medical professionals, including decisions around prescribing. There is some evidence that nudging appears to be an effective tool to influence medical professionals’ decisions in broad terms. In this review we collected and analysed the literature on nudges used to influence prescribing decisions of health care professionals. I assessed the effectiveness and cost-effectiveness of the interventions, and tested whether the effectiveness was associated with key characteristics related to study design, intervention design, or setting. The review showed that nudges may be an effective tool for influencing prescribing decisions - possibly more often compared to other problem types. However, the effects sizes are small and the derived health effects are unknown. The interventions seem to be cost-effective, however the numbers of studies which do economic analyses is low and all apply a limited perspective.


Paper 3; Effects of reminders on GPs intentions to reduce opioid prescribing? A randomised online survey experiment.

Guidelines typically recommend against using opioids for long term treatment of non-malignant pain. Despite this, over prescribing of opioids remains a persistent issue. Nudging has been suggested as a method to increase guideline concordance in prescribing, and there are examples of positive results with reminders and other forms of nudge-type messaging to reduce opioid prescribing. Little is however known about the heterogeneity and mechanism of effects. I designed a three arm randomised experiment, in which two messages, designed to draw attention to current Danish guidelines on opioid prescribing and data driven tools for prescription monitoring, respectively, was nested in a survey sent to all Danish GPs. As the outcome I asked GPs how likely they were to reduce their opioid prescribing on four key parameters in the following six months. In addition, I asked questions about their opinions on prescribing, experience with reducing opioid prescribing, and motivational styles. I tested the effect of the interventions on the self-reported likelihood of reducing opioid prescribing, controlling for age, sex, practice style, experience, opinions, and motivation styles, and tested for heterogeneity in effects according to the same parameters, and for each of the messaging questions’ impact on the outcomes. The results showed that the messages reminding GPs about guidelines had no overall effect. The message about data driven prescription monitoring tools on the other hand significantly lowered GPs’ perceived likelihood to engage in reducing prescribing. GPs who are younger, work in shared practice, disagree there is prescribed too much unnecessary medicine, and had lower levels of user orientation were relatively more negatively affected. Questions relating to positive views on data driven tools and the importance of prescription monitoring were most strongly correlated with the outcome questions.

On the basis of these studies I conclude that a substantial part of the inappropriate prescribing is attributable to the prescribing behaviour of providers and give recommendations to where policymakers and administrators should focus further initiatives to optimise prescribing practices. Further, using nudge type behavioural economic interventions for directing the prescribing behaviour of health care professionals, although rising in popularity, is still relatively unexplored scientific territory. This thesis also clearly demonstrates that there are risks of unintended effects and heterogeneous effects of nudge interventions, something which is not well reflected in the current body of literature pertaining to their effects on health-care professionals. Decision makers should keep these things in mind if considering to use nudges to direct prescribing behaviours.