Development of a brief tool for monitoring aberrant behaviours among patients receiving long-term opioid therapy: The Opioid-Related Behaviours In Treatment (ORBIT) scale

Highlights

• The unintended consequences of long-term opioid therapy have been well-documented.

• Early identification of medication-related problems in treatment is essential.

• A 10-item Opioid-Related Behaviours In Treatment (ORBIT) scale was developed.

• The ORBIT scale is brief, reliable and has strong face validity.

• The ORBIT scale is suitable for use in clinical and research settings.

Abstract

Background

Early identification of problems is essential in minimising the unintended consequences of opioid therapy. This study aimed to develop a brief scale that identifies and quantifies recent aberrant behaviour among diverse patient populations receiving long-term opioid treatment.

Method

40 scale items were generated via literature review and expert panel (N = 19) and tested in surveys of: (i) N = 41 key experts, and (ii) N = 426 patients prescribed opioids >3 months (222 pain patients and 204 opioid substitution therapy (OST) patients). We employed item and scale psychometrics (exploratory factor analyses, confirmatory factor analyses and item-response theory statistics) to refine items to a brief scale.

Results

Following removal of problematic items (poor retest-reliability or wording, semantic redundancy, differential item functioning, collinearity or rarity) iterative factor analytic procedures identified a 10-item unifactorial scale with good model fit in the total sample (N = 426; CFI = 0.981, TLI = 0.975, RMSEA = 0.057), and among pain (CFI = 0.969, TLI = 0.960, RMSEA = 0.062) and OST subgroups (CFI = 0.989, TFI = 0.986, RMSEA = 0.051). The 10 items provided good discrimination between groups, demonstrated acceptable test–retest reliability (ICC 0.80, 95% CI 0.60–0.89; Cronbach's alpha = 0.89), were moderately correlated with related constructs, including opioid dependence (SDS), depression and stress (DASS subscales) and Social Relationships and Environment domains of the WHO-QoL, and had strong face validity among advising clinicians.

Conclusions

The Opioid-Related Behaviours In Treatment (ORBIT) scale is brief, reliable and validated for use in diverse patient groups receiving opioids. The ORBIT has potential applications as a checklist to prompt clinical discussions and as a tool to quantify aberrant behaviour and assess change over time.

Introduction

Opioids are essential medicines in the treatment of pain and of opioid dependence (World Health Organization, 2013). Yet there is global concern regarding the availability and regulation of opioid medications, largely as a result of the associated risks: diversion (Compton and Volkow, 2006, Manchikanti et al., 2012), opioid use disorders (Manchikanti et al., 2012) and mortality (McCarthy, 2013, Paulozzi, 2012). It is, therefore, important to maximise adherence to opioid therapy and the subsequent treatment benefits, while minimising unintended consequences.

Opioid-related (or ‘aberrant’) behaviours are divergent, or unexpected, behaviours among patients receiving opioid therapy (Passik and Kirsh, 2005). These behaviours are usually taken as indicating problems in the course of opioid therapy, such as issues with non-adherence, unsanctioned diversion of medication to others, hazardous use or misuse (Michna et al., 2004). In this context, prescribers should monitor aberrant medication behaviours when reviewing patients, including the safety and effectiveness of opioid treatment, for example as part of the ‘four A's of monitoring opioid treatment in pain management: (i) analgesia, (ii) activities of daily living, (iii) adverse side effects, and (iv) aberrant (medication-related) behaviours (Coluzzi et al., 2005, Nicholson and Passik, 2007).

Despite a growing literature of screening/review tools to identify or predict problems in opioid treatment, there is little agreement regarding which individual opioid-related behaviours are clinically the most significant (Turk et al., 2008). The existing tools are varied in terms of their aims, administration and scope. Most tools are designed as ‘screening tools’ that aim to predict those at risk of misuse of medication or of developing opioid use disorders (e.g., Adams et al., 2004, Butler et al., 2007, Butler et al., 2004, Chabal et al., 1997, Chou et al., 2009, Coambs et al., 1996, Compton et al., 1998, Friedman et al., 2003, Michna et al., 2004, Miotto et al., 1996, Passik and Kirsh, 2003, Passik et al., 2008, Passik et al., 2006, Sees and Clark, 1993, Webster and Webster, 2005, Wu et al., 2006) and often include patient characteristics such as age, gender and history of substance use or mental health. Some are intended as clinical review tools for monitoring patient progress (e.g., Belgrade et al., 2006, Passik et al., 2004) and others were developed to validate other tools (e.g., Adams et al., 2004, Butler et al., 2008). There is also variety in how the tools are administered. Modes of administration include: patient self-complete (e.g., Adams et al., 2004, Butler et al., 2010, Butler et al., 2007, Elander et al., 2003, Fleming et al., 2008, Holmes et al., 2006, Passik et al., 2000); clinician observations (e.g., Adams et al., 2004, Atluri and Sudarshan, 2004, Belgrade et al., 2006, Chabal et al., 1997, Compton et al., 1998, Miotto et al., 1996, Webster and Webster, 2005); and clinician-guided questioning of the patient (e.g., Cowan et al., 2001, Cowan et al., 2003a, Friedman et al., 2003, Passik et al., 2004). Existing tools also vary widely in terms of their scope: some include a focus on social/structural determinants of health and wellbeing (such as legal problems, employment, age, and social environment; Belgrade et al., 2006, Butler et al., 2009a, Compton et al., 1998, Friedman et al., 2003, Miotto et al., 1996, Webster and Webster, 2005). Other instruments include items examining comorbidity (such as past/current substance use, mental health problems) (Adams et al., 2004, Butler et al., 2010, Butler et al., 2009a, Coambs et al., 1996, Compton et al., 1998, Compton et al., 2008, Friedman et al., 2003, Miotto et al., 1996, Passik et al., 2000, Webster and Webster, 2005), personality or mood (such as anger, social functioning, etc.; Butler et al., 2009a, Butler et al., 2010, Friedman et al., 2003, Passik et al., 2000, Webster and Webster, 2005). Many scales include some items identifying ‘aberrant’ opioid-related behaviour, such as stockpiling medication, emergency department visits for medication or requesting early script renewals (e.g., Adams et al., 2004, Atluri and Sudarshan, 2004, Banta-Green et al., 2010, Butler et al., 2010, Butler et al., 2009a, Chabal et al., 1997, Compton et al., 1998, Cowan et al., 2001, Cowan et al., 2003a, Fleming et al., 2008, Holmes et al., 2006, Knisely et al., 2008, Manchikanti et al., 2003, Michna et al., 2004, Passik et al., 2004), although the aberrant medication behaviours included vary across different scales.

A recent review has highlighted the important limitations of the existing instruments (Smith et al., 2015). Many have been developed in very small sample sizes, and typically have not been well validated (Smith et al., 2015, Turk et al., 2008). The psychometric data across existing tools is limited in scope (Smith et al., 2015). The vast majority of scales also do not produce a score that enables changes over time to be monitored (Smith et al., 2015), and many do not specify timeframes for behaviours. In addition, many tools are lengthy and can contain up to 42 items (e.g., Compton et al., 1998), rendering them unlikely to be used by clinicians in routine practice, particularly in primary care settings.

The main aim of the current study was to develop an opioid-related behaviour scale based on the existing literature, with the specific aim of identifying and quantifying recent aberrant behaviour by patients currently prescribed opioids, in order to assist ongoing clinical review. Additional aims of the scale were to: (i) be as brief as possible; (ii) have additional utility as a research tool; (iii) be useful in monitoring diverse patient populations receiving long-term opioid therapy; and (iv) enable measurement of change over time.