Clinical instruments assessing harms and benefits of long-term opioid therapy for chronic pain: a scoping review

Abstract

Background:

While rates of opioid prescribing for chronic non-cancer pain have declined, long-term opioid therapy (LTOT) remains common. Current clinical guidelines emphasize the importance of continuous reassessment and weighing of harms and benefits associated with LTOT to inform treatment decisions. However, specific guidance on which harms and benefits to assess and how to weigh these factors are generally lacking. This scoping review aimed to identify and characterize clinical instruments used to assess opioid-related harms and benefits in adults receiving LTOT for chronic non-cancer pain with a focus on instrument content domains and weighing of harms and benefits.

Methods:

Study selection was guided by the Population, Concept, Context (PCC) framework. Studies were included if they described clinical assessment instruments used in adults prescribed LTOT (≥90 days) for chronic, non-cancer pain. We searched multiple databases and screened records using a multi-coder process. Data were extracted on instrument characteristics, content domains, individual items, and scoring.

Results:

From 8,738 records, 37 studies describing 25 instruments were included. Most instruments were self-administered by patients. All instruments assessed harms (either current or potential). Of the 448 total items contained in the 25 included instruments, 12% assessed current or potential benefits (including efficacy of analgesia, functional improvement, and quality of life) and 88% assessed harms (including misuse and safety concerns). Instruments prioritized assessment of current or potential harms over assessment of current or potential benefits of LTOT.

Discussion:

Existing instruments for monitoring LTOT predominantly assess harms—particularly misuse—and may underrepresent current or potential therapeutic benefit. The lack of instruments to guide prescribers on harms/benefits weighing leaves a major clinical research gap.

Primary Funding Source:

NIH-HEAL-U24DA058673.

Registration:

Our protocol was registered online via the Open Science Framework at https://osf.io/3ckq6.

Introduction

While opioid prescribing for chronic pain has declined since its peak in 2012, the practice remains common.¹ Given myriad potential harms and often limited benefit of opioid therapy, guidelines recommend continuous reassessment and weighing of the harms and benefits to guide treatment decisions, including tapering or discontinuation.² Continuous reassessment, which involves identifying, monitoring, and weighing opioid-related harms and benefits over time, is considered a critical tool in mitigating harms and maximizing benefits of long-term opioid therapy (LTOT) for patients with chronic pain.² A variety of clinical instruments, including questionnaires and machine learning applications, have been developed to support the continuous reassessment of LTOT for chronic pain management.³

In addressing the topic of continuous reassessment, there is some important nuance to note regarding terminology. Guidelines recommend “regularly reevaluating benefits and risks of continued opioid therapy with patients.”⁴ Practically speaking, clinicians must weigh current benefits against current harms of LTOT and consider dynamic changes to potential benefits and potential harms that may have occurred since initiation. Potential benefits of LTOT include the possibility and likelihood of improved daily functioning, meaningful pain relief, and improved quality of life,^5,6 while potential harms include the possibility or likelihood of experiencing safety concerns (e.g., respiratory depression, falls, brain fog) or misuse ranging from mild to fatal.⁶ Per current guidelines, assessment and reassessment of LTOT within the benefits and risks framework are intended to help providers optimize pain management over time;² however, there is little guidance on how to appropriately weigh harms against benefits and there is currently no widely accepted instrument or protocol to support regular reassessment.

The present study aimed to map the clinical instruments developed to monitor benefits and harms of LTOT for chronic pain and explore whether and how instruments guided the user in weighing harms against benefits. Mapping these extant clinical instruments and characterizing whether and how they assist clinicians in weighing harms against benefits is an important step in determining the degree to which these instruments reflect current guidelines in opioid-related assessment and reassessment. This review addresses a notable gap in the literature, as previous reviews did not attempt to focus on instrument content nor on weighing of harms against benefits.

Methods

The purpose of a scoping review is to identify and map available evidence rather than to answer a specific research question.⁷ We referenced the Arksey and O’Malley methodological framework, enhanced by the Joanna Briggs Institute, to conduct this scoping review. Our protocol was registered via the Open Science Framework (https://osf.io/3ckq6).

Data Sources and Searches.

In collaboration with a research librarian and key stakeholders (including people with lived experiences and coinvestigators from medicine and behavioral science), we developed a comprehensive search strategy that involved searching bibliographic databases from 2007 to November 22, 2023 (later updated to include articles published through August 4, 2025). Literature was obtained from the following databases: MEDLINE, Embase, PsycInfo, Web of Science, Central and Cochrane. Where available, we used controlled vocabulary terms to maximize the number of potentially relevant studies identified (see Supplementary Table 1). To identify additional relevant studies, we conducted hand searches of reference lists for potentially eligible full text articles.

Study Selection.

We used the Population, Concept, Context (PCC) framework for eligibility criteria (Table 1). Eligible studies included adults with chronic non-cancer pain (pain on most days for ≥3 months) prescribed LTOT (≥90 days continuous prescription) and described clinical instruments assessing opioid-related harms and benefits. There were no restrictions related to publication year or geographic location. Exclusion criteria included studies related to acute and palliative pain as well as cancer-related and post-operative pain.

Table 1.

PCC criteria

PCC category	Inclusion criteria	Exclusion criteria
P (Population)	Adult (18 years and older) Inpatient and outpatient populations Chronic, non-cancer pain diagnoses (or being treated for chronic, non-cancer pain, as many studies define the dx differently) Long-term opioid therapy	Acute pain Pain related to cancer Palliative care Non-patient populations
C (Concept)	Clinical instruments developed to assess harms and benefits, safety, efficacy, or misuse of ongoing (vs initiating) long term opioid therapy within the patient encounter, including diagnostic and risk stratification tools.	Clinical instruments developed for use prior to initiation of long-term opioid therapy, instruments focused primarily on assessing acute withdrawal, or approaches involving genetic screening or imaging. System-level frameworks/workflows involving the use of multiple individual instruments were also excluded.
C (Context)	Studies conducted in inpatient and outpatient healthcare facilities, including hospitals, medical centers, outpatient clinics, and long-term care facilities. Studies from any geographic location will be eligible for inclusion.

Data Synthesis and Analysis.

We used EndNote 21 bibliographic software and the Yale Reference Deduplicator to store the retrieved records and remove duplicates. We used Covidence, a review software, to evaluate records. The coding team included AG, clinical psychologist, MA, research assistant and MPH, and AC, primary care physician. PCC criteria (Supplementary Table 2) were applied to titles and abstracts of records. After each record was assessed for inclusion by two coders, the team met to discuss discrepancies, which were resolved by the third coder. This process was repeated with full-text articles to determine inclusion. The team then met to discuss guiding definition for coding and coded two instruments together to facilitate agreement on interpretation of guiding definitions. A data extraction form was used and revised throughout the full-text extraction process. The following data were extracted from each full text: journal, publication year, first author, location, clinical setting, population, sample size, study aim, administration method, purpose, number of items and item topic (i.e., harms or benefits). Items were assigned subcodes pertaining to their content (e.g., functioning, misuse, etc.). The coding team met three times to build consensus on application of codes. Following item coding, instruments were assessed for whether they guided the user in weighing harms against benefits. An operationalized definition of weighing harms against benefits was devised by authors AG, SE (clinical psychologists), and WB (internist) via a review of harm-benefit assessment literature.^8–10 To refine the guiding definition, AG coded five instruments according to the drafted definition, then met with WB and SE to discuss and refine the definition. This cycle was repeated three times until consensus was reached. Findings, key concepts, and knowledge gaps were synthesized and presented with narrative descriptions and in tables and figures.

Guiding definitions.

The following established definitions were used to guide this review: Harm was defined as “damaging to one’s health or welfare; having an adverse effect”; benefit was defined as “to the advantage or enhancement of one’s health or welfare.” Safety was defined as “adverse effects, harmful interactions, and opioid-related problems”; efficacy was defined as “ability of treatment or therapy to produce desired outcomes, including pain relief and improved quality of life or functioning”; misuse was defined as “use not in accordance with prescriber’s directions, regardless of the presence or absence of harm resulting from use.” In extracting characteristics of instrument items, items were coded exclusively as assessing either current or potential harms or benefits. Items were coded as pertaining to current harms or benefits if they pertained to dynamic, current characteristics, experiences, or behaviors. Items were coded as potential harms or benefits if they pertained to the likelihood of a future event, behavior, or experience (Supplementary Figure 1). Measures were considered to weigh harms against benefits that a)included items assessing harms and benefits and b)provided a qualitative (e.g., if/then contingency recommendations) or quantitative (e.g., numeric score) schema for weighing harms and benefits.

Results

A total of 8,738 records were identified (Figure 1). After removing duplicates and screening for relevance, we included 37 studies published between 1998 and 2023 describing 25 clinical instruments (Table 2). Nine clinical instruments were described in more than one study. Most studies were carried out in the United States (n=34; 91%).

Figure 1.

PRISMA diagram of systematic search and included studies

Table 2.

Included instruments and studies

Measure	Purpose of Instrument	Method of Administration	Studies	# Items total
Addiction Behaviors Checklist (ABC)	To track behaviors characteristic of addiction related to prescription opioid medications in chronic pain populations	Clinician administered	Wu 200619	20
Coping Strategies Questionnaire (CSQ) catastrophizing item ONLY	To provide a clinically useful tool for assessing risk for opioid misuse from a single item measure of pain catastrophizing	Patient self-administered	Lutz 201720	1
Current Opioid Misuse Measure (COMM 11 - PWDA)	To screen for harmful opioid use in people with disability and chronic pain due to arthritis	Patient self-administered	Lankford 202221	11
Current Opioid Misuse Measure (COMM - 9)	To monitor aberrant medication-related behaviors of chronic pain patients	Patient self-administered	McCaffrey 201922	9
Current Opioid Misuse Measure (COMM-17)	To monitor aberrant medication-related behaviors of chronic pain patients	Patient self-administered	Butler 2007;23 Barth 2014;24 Meltzer 201125	17
Diagnosis, Intractability, Risk, and Efficacy Score (DIRE)	To predict efficacy of analgesia and patient compliance with long-term opioid analgesic treatment	Clinician administered	Belgrade 200626	7
Empower Study Expert Consensus Protocol	To assess for opioid use disorder among patients receiving long term opioid therapy for chronic pain	Patient self-administered	You 202127	14
Opioid Abuse Risk Screener (OARS)	To support well-informed decision-making in opioid analgesic prescribing by extending the breadth of psychiatric risk factors evaluated relative to other non-clinician-administered measures	Patient self-administered	Henrie-Barrus 201611	28**
Opioid Compliance Checklist (OCC)	To help detect current and future aberrant drug-related behavior and non-adherence among chronic pain patients in primary care	Patient self-administered	Jamison 201628	8
Opioid Risk Tool (ORT)	To predict the probability of a patient displaying aberrant behaviors when prescribed opioids for chronic pain	Patient self-administered	Webster 200529 Jones 201130 Cheatle 201931	10* 9 (Cheatle 2019)
Opioid-Related Behaviors in Treatment (ORBIT)	To quantify aberrant behavior and assess change over time	Patient self-administered; Non-clinician administered	Larance 201632	10
Pain Assessment and Documentation Tool (PADT)	To focus on key outcomes and provide a consistent way to document progress in pain management therapy over time	Provider self-administered; Other: Physician-completed document for use while charting. Template for chart notes.	Passik 200433	39
Pain Medication Questionnaire (PMQ)	To assess risk for opioid medication misuse in chronic pain patients	Patient self-administered	Adams 200434 Holmes 200635	26
Patient Reported Indications for Opioid Reassessment (PRIOR)	To be sensitive to incipient or developing harms and low or absent benefit	Patient self-administered	Becker 201636	37
Prescribed Opioids Difficulties Scale (PODS)	To assess difficulties patients attribute to chronic opioid therapy	Non-Clinician Administered	Sullivan 2010;37 Banta-Green 201038	16
Prescription Drug Use Questionnaire (PDUQ)	To evaluate the pain condition, opioid use patterns, social and family factors, family history of pain and substance abuse syndromes, patient history of substance abuse, and psychiatric history	Non-clinician administered; Incorporates electronic medical record data	Compton 1998;39 Banta-Green 200940	42 (Compton 1998)* 29 (Banta-Green 2009)
Prescription Drug Use Questionnaire - Patient Version (PDUQp)	To help clinicians better identify the presence of opioid abuse or dependence in patients with chronic pain	Clinician and Non-Clinician Administered	Compton 200841	31
Prescription Opioid Misuse and Abuse Questionnaire (POMAQ)	To identify prescription opioid abuse and misuse among patients with chronic pain	Patient self-administered	Coyne 2021;42 Coyne 202343	19
Prescription Opioid Misuse Index (POMI)	To assess prescription opioid misuse	Patient self-administered under the supervision of a trained psychologist; Non-Clinician Administered	Coloma-Carmona 2023;44 Knisely 200845	6 (Coloma-Carmona 2023) 8 (Knisely 2008)*
Routine Opioid Outcome Monitoring tool (ROOM)	To assess the “4 A’s” outcomes (Analgesia, Activity, Adverse Effects, and Aberrant drug-related behaviors)	Patient self-administered	Nielsen 202046	12
Screener and Opioid Assessment for Patients with Pain (SOAPP)	To assess suitability of long-term opioid therapy for chronic pain patients	Patient self-administered	Butler 2004;47 Akbik 200648	24 (Butler 2004) 14 (Akbik 2006)*
Screener and Opioid Assessment for Patients with Pain-Revised (SOAPP-R)	To predict aberrant medication-related behaviors among chronic pain patients considered for long-term opioid therapy	Patient self-administered	Black 2018;49 Butler 200850	8 (Black 2018) 24 (Butler 2008)*
Screening Tool for Addiction Risk (STAR)	To identify pain patients at risk before they receive opioid treatment	Patient self-administered	Friedman 200351	14
Stratification Tool for Opioid Risk Mitigation (STORM)	To (a) estimate patient risk for an overdose-orsuicide-related event and (b) provide actionable information for risk-stratified intervention, thus serving as a platform to help providers visualize and implement a much-needed patient-centered approach to address opioid safety	Incorporates electronic medical record data	Oliva 201712	50

^*The Opioid Abuse Risk Screener¹¹ is not publicly available as the measure is trademarked and available for purchase. Thus, individual item content for the OARS was not included in analyses.

One instrument, the Opioid Abuse Risk Screener¹¹ (OARS) was not publicly available and, thus, OARS items were not included in item-level analyses. All instruments assessed opioid-related harms. Safety was assessed in 76% of instruments (n = 19). Approximately two-thirds of all instruments (n=18; 72%) included any item pertaining to benefits (or lack thereof). Most instruments (n=22; 88%) specified a time interval over which the patients’ behaviors, experiences, or characteristics should be considered (as in “Over the past three months, how often…”). Of these, twelve instruments specified timeframes within twelve months; seven instruments assessed an interval of over a year.

Instrument formats and variants.

A variety of formats were employed to assess harms and benefits of LTOT. Instruments were primarily patient (n=22; 88%) and clinician (n=2; 8%) report questionnaires and one was a machine learning algorithm. The algorithm, STORM (Stratified Tool for Opioid Risk Mitigation)¹², was developed to model overdose and suicide risk using individual and population-level data extracted from the electronic medical record (EMR).

Most of the identified instruments were intended for use during ongoing opioid therapy (n=17; 68%); the remaining instruments were intended to be used both before initiation of LTOT and during ongoing treatment (n=8; 32%).

Instrument content.

The 25 clinical instruments identified in this review contained a total of 448 items, a majority of which addressed current or potential opioid-related harms (n=393; 88%). Approximately half of all items assessed current harms (n= 233; 52%) and approximately one-third assessed potential harms (n= 162; 36%). Items describing present harms were coded as current harms while items predicting the likelihood of future harm were coded as potential harms. Only 12% of all items (n=55) assessed current or potential benefits or lack of benefits, including pain relief, functional improvements, and quality of life. Items describing present benefits were coded as current benefits while items predicting the likelihood of future benefit were coded as potential benefits. Among items assessing benefits or lack of benefits, 7% assessed current benefits (n=31) and 5% assessed potential benefits (n=24; see Figure 2 for hierarchical visualization of instrument items by harms/benefits and current/potential).

Figure 2.

Sankey Graph of Content domain distribution among all instruments

Harm: Safety and misuse.

All instruments assessed opioid-related harms. Safety and misuse were coded as dimensions of harm. Harms related to safety were assessed in 31% of all items (n=138). More than half of all instrument items (n=255; 57%) assessed harms related to misuse. Items assessing harm were further coded into current and potential harms (see Guiding definitions).

Current harms.

“Current harms” items assessed the following factors: current use of illicit drugs, problem drinking, taking opioids not as prescribed (i.e., changing dose, altering method of administration, visiting other providers for scripts, diversion, obtaining medications from nonmedical sources, etc.), misrepresenting pain severity or pain management behaviors to clinician, current opioid use disorder symptoms (i.e., cravings, difficulty tapering, continued use despite harms), and safety-related side effects (e.g., falls fatigue, constipation, mental cloudiness, motor vehicle accidents).

Potential harms.

Items assessing “potential harms” (also defined as “risks” or “predictors of future harms”) assessed the following factors: psychological factors interfering with care, emotional dysregulation (including anger, overwhelm, impatience, boredom, mood swings, anxiety, and depression), adverse childhood experiences (e.g., substance use in immediate family, preadolescent sexual abuse), prior substance use, prior engagement in substance use treatment, demographic and medical factors broadly associated with increased likelihood of substance misuse (e.g., patient sex, age), past experience of side effects from similar medications, and personality factors (i.e., “How often have others kept you from getting what you deserve?”).

Benefits: pain relief, functioning, and quality of life.

Pain relief, functioning, and quality of life were coded as dimensions of opioid-related benefits. Many items assessed lack of potential benefit; these items assessed factors associated with low likelihood of future pain relief, improved functioning, or improved quality of life as a result of LTOT. About two-thirds of the identified instruments (n=18; 72%) included at least one item pertaining to opioid-related benefits (or lack thereof); about one-third of included instruments did not assess opioid-related benefits.

Current benefits.

“Current benefits” was assessed via the following factors: degree of pain relief and meaningful improvements to daily functioning (e.g., improved social/role or physical functioning). Lack of current benefits was assessed via the following factors: inadequacy of pain relief, signs of tolerance (i.e., need for higher doses to obtain the same analgesic effect), and absence of meaningful improvements to daily functioning.

Potential benefits.

“Potential benefits” were assessed via the following factors: strength of working relationship with provider, patient satisfaction with prior pain care, and patient openness to multimodal pain management. Lack of potential benefits (also defined as “a reduced likelihood of the patient benefitting” from LTOT), was assessed via the following the factors: number of therapies trialed without meaningful pain reduction, family dynamic of solicitousness or other social factors assumed to sustain the patient’s illness behavior or pain symptoms, ongoing litigation around pain-precipitating incident, evidence of psychosomatic influence on pain symptoms or diagnosis with a somatoform disorder, severity of pain diagnosis (benign, progressive, advanced), number of pain-related diagnoses, and family history of intractable chronic pain.

Weighing harms against benefits.

We coded whether instruments weighed harms against benefits (Y/N). Twelve of the identified instruments (48% of all instruments) met operationalized criteria (see Guiding Definitions above) as weighing harms against benefits. Among these twelve instruments, seven included only negatively framed benefits items (i.e., items assessing lack of current or potential benefit). Among these instruments with only negatively framed benefits items, four included only a single, negatively framed benefits item (see Figure 3 for item distribution by instrument).

Figure 3.

Item distribution by instrument

* The Opioid Abuse Risk Screener¹¹ (OARS) is not publicly available as the measure is trademarked and available for purchase. Thus, OARS items were not included in item-level analyses.

Discussion

This scoping review identified 37 studies reporting on 25 clinical instruments designed to assess opioid-related harms and benefits of LTOT for chronic pain. Instruments were primarily intended for direct patient administration. Instrument item content was notably heterogeneous: instruments assessed a diverse range of factors related to opioid-related harms and benefits. One commonality among identified instruments was prioritized assessment of opioid-related harms and minimal assessment of opioid-related benefits. Among the 25 identified clinical instruments, twelve were found to assist in weighing harms against benefits. Of these instruments, more than half only assessed lack of current or potential benefits.

Challenges and next steps

While guidelines have emphasized the need for prescribers to frequently reassess harms and benefits of LTOT, consensus is lacking on which instrument(s) should be used to execute this task. Notably, the present review found potential harms and benefits are frequently assessed as a dimension of opioid-related harms and benefits. While guidelines suggest opioid monitoring and prescribing rely primarily on assessment of current harms and benefits.^2,13,14, assessment of potential harms and benefits is appropriate during ongoing LTOT: the likelihood of future harms and benefits is dynamic as patients age, develop comorbidities, are prescribed other medications, etc. Thus, assessing current harms and benefits and monitoring potential harms and benefits is the implicit task of the identified instruments. There is little guidance available, however, on how to weigh current versus potential harms and benefits in prescribing decisions. On the one hand, current harms and benefits might assume primacy in these decisions: they are ostensibly more straightforward to assess, they are often observable/quantifiable within a constrained timeframe (e.g., weeks since prior assessment), and they are likely to be prioritized by patients due to their immediate impact.¹⁵ As current guidelines do not explicitly call out the dual tasks of assessing current and potential harms and benefits, the implications of these distinct dimensions are up for clinician interpretation. Future guidelines might provide guidance on how these distinct dimensions are best factored in prescribing decisions and future research might aim to characterize how current versus potential factors are distinctly weighed by clinicians, how they are perceived by patients, and whether clinicians find this to be a helpful framework for weighing harms and benefits of LTOT.

There are additional implications, from a psychometric standpoint, of assessing both current and potential harms and benefits as distinct dimensions. Items assessing potential harms or benefits may not be adequately sensitive due to the timeframe being either too broad or too vague, and thus fail to detect dynamic changes over the period of assessment.¹⁶ For example, the SOAPP item “Did any family member physically or verbally abuse you when you were a child?” will lack variability over time given the historical nature of the question.

In the assessment of harms and benefits of LTOT for chronic pain, it is up for debate whether specificity or sensitivity of instruments should be prioritized: typically, sensitivity is prioritized in measures of high risk situations with low stakes consequences, while specificity is prioritized in measures of low risk situations with high stakes consequences.¹⁶ Recent qualitative research suggests the stakes of this assessment—maintenance or changes to LTOT—are likely to be viewed differently by patients and providers.¹⁵ The breadth of domains in the identified instruments, including misuse, safety, functioning, quality of life, and effective relief of pain, may contribute to greater instrument sensitivity (i.e., more people screen positive), which induces a proportional loss of specificity (i.e., the percentage of positive screens that actually reflect the measure’s target construct), resulting in a higher rate of false positives. False positives, in this context, are cases in which harms of LTOT may be overestimated, which may be interpreted by patients as unfair or disadvantageous.

Several items assessing potential harms and benefits estimate the likelihood of future events indirectly (e.g., adverse childhood experiences, receipt of disability income, ongoing litigation around pain-precipitating incident, relationship with prior pain care providers, emotionality, personality factors, etc.). Indirect measurement of potential harms or benefits in the included instruments was frequently based on adverse childhood experiences (as the SOAPP item above), substance use among friends or family, or personal history of substance use, topics that may cause feelings of stigmatization or distress during disclosure. Distressing and stigmatizing items are known to contribute to dishonest responding¹⁷ and, in the context of pain management prescribing decisions, hazard escalation of an already emotionally charged clinical encounter. Moreover, there are ethical considerations to indirect measurement of potential harms and benefits in pain management prescribing decisions: many historical factors associated with risk for future harms are also associated with risk for the development of chronic pain (e.g., adverse childhood experiences),¹⁸ calling into question the fairness of using these factors to assess potential harm of LTOT, per se. Future research might explore clinician and patient perspectives on this practice. New or revised clinical instruments might prioritize more direct measurement along with transparent scoring, enabling patients and providers to intuit how instrument items relate to current and potential harms and benefits and how they inform prescribing decisions.

The predominant focus on assessment of harms presents important considerations. One-third of identified instruments lacked any assessment of benefits. Among instruments with benefits items, most included only 1–2 items. Minimal assessment of opioid-related benefits may present a challenge to making patient-centered management decisions: instruments are needed that offer more balanced assessment of both opioid-related harms and benefits. Further, less than half of the included clinical instruments met our criteria for weighing harms against benefits. More than half of the instruments that met criteria for weighing harms against benefits included only negatively framed benefits items (items associated with lack of current benefits or reduced likelihood of future benefits). Negatively framed benefits items, which are scored in the same direction as harms items, result in simpler instruments (in that no reverse scoring is necessary); however, this simplification may obscure the process of weighing harms against benefits to the instrument user, which may impact response tendencies and clinical utility of the instrument as a decision-making support tool. That many instruments assessing harms and benefits of LTOT did not further aid prescribers in weighing harms versus benefits is not necessarily a critique of those instruments – indeed, “weighing” may simply have been beyond the scope of these instruments. Nonetheless, our review suggests a large gap in instruments providing explicit guidance for prescribers in the weighing of harm versus benefits, a core guideline-recommended task.

It was notable that all but one identified instrument included patient or clinician-reported outcomes. In clinical practice, data on harms and benefits of LTOT are likely to be gathered from several sources, including patient report/priorities, clinician assessment, and relevant patient history. STORM, the lone EMR algorithm, was the only instrument identified to make explicit use of EMR data. As machine learning models and other automated tools advance, their ability to integrate data on patient experience and self-report will be critical to their accuracy and utility as clinical support tools. Ideally, objective measures, clinician reports, and patient-reported outcomes can be triangulated to achieve the most accurate, patient-centered estimation of harms and benefits of LTOT.

Strengths, Limitations, and Conclusions

This review was strengthened by a broad search strategy codesigned with people with lived experience of chronic pain and LTOT, a multidisciplinary analysis and authorship team, and carefully developed guiding definitions for key terms (e.g., “weighing harms and benefits”). Our search was limited to clinical instruments intended for use in the context of LTOT for chronic pain; instruments designed for populations using opioids for other indications (e.g., OUD, cancer-related pain) were excluded, as there are distinct harms and benefits of LTOT and different guidelines for these indications. While this allowed for a more targeted, item-level analysis in the present review, it is possible that instruments designed for other populations may be useful in this population. Future research may investigate whether and how harms and benefits of LTOT are weighed in other populations, and how this compares to populations with chronic pain.

Given guideline recommendations for frequent reassessment of current harms/benefits during LTOT for chronic pain, extant instruments may be inadequate due to 1) lack of clarity on the target construct (assessment of both current and potential harms/benefits), 2) imbalance or absence of benefits assessment, and 3) lack of sensitivity to dynamic changes over the reassessment period. The heterogeneity of items used to assess opioid-related harms and benefits suggests that any new tool targeting this construct should reflect the spirit and intent of clinical guidelines for assessment of benefits and harms, have utility to providers making real-time decisions about ongoing opioid prescriptions, and be developed with the input of both patients and clinicians who can inform the scope of benefits and harms and ensure items have face validity. Rigorous assessment of item response characteristics should be conducted to ensure items are unbiased, valid, and sensitive to change over time.

Supplementary Material

Supplementary Material

Supplementary Figure 1: Guiding model of current and potential harms and benefits, characteristics of measurement

Supplementary Table 1. Comprehensive search strategies