Achieving two-part harmony: standardizing pain-related phenotypes and outcomes

Introduction

Military service members and veterans are especially vulnerable to developing chronic pain conditions, given their high risk of musculoskeletal injuries and exposure to combat-related trauma. Approximately half of active-duty military service members experience significant pain, especially women.¹ Indeed, persistent back pain is among the most common drivers of health care use and interruption of combat duty.¹ These realities highlight the need to develop and deploy effective strategies for prevention and management of pain, particularly nonpharmacological treatments, which are effective first-line therapies for military and veteran personnel but are underutilized in clinical practice. Pragmatic trials can help to address this implementation gap.

In this Commentary, we describe the benefits and the process of harmonizing data collection efforts through use of standardized measures across a pragmatic trial network (the Pain Management Collaboratory, PMC). Though harmonization poses significant challenges (eg, possible delays in trial initiation as measures are harmonized, slight elevation of trial costs, potential increases in participant burden), its benefits are likely to outweigh the costs. As a large trial network, the PMC provides a model of how combining pragmatic trial design with collaborative tools and multidisciplinary expertise can advance the science and impact of nonpharmacological pain management research in real-world settings. We believe that harmonizing phenotypic and outcome data collection across similar trial networks has the potential to magnify the results emerging from those studies, support comparisons across populations, and promote the wide adoption of evidence-based care.

Advantages of harmonization

Data are maximally useful if collected in a standardized manner to facilitate reproducibility, subgroup analyses, data pooling, and cross-study comparisons. Moreover, variability in outcome measures across clinical trials hinders the field’s ability to conduct evaluations of treatment efficacy and effectiveness.² Harmonized data collection is especially important for research studies related to conditions such as chronic pain, in which many different types of participant data are collected, including many measurements that fall within the category of patient-reported outcomes.³ Collectively, the use of a standard set of validated outcome measures across network trials: (1) facilitates the process of developing individual study research protocols; (2) ensures that all trials within a network will provide potentially actionable clinical information; (3) simplifies the process of designing and reviewing research proposals, manuscripts, and published articles within the network; (4) enhances study reproducibility; (5) provides a basis for determining treatment outcomes that constitute clinically important differences; (6) permits pooling of data from different studies to enhance power and generalizability; and (7) aids future systematic reviews and meta-analyses.

In addition, standardizing a core set of outcome domains encourages investigation and reporting of outcomes that are relevant to research partners and end users, so that dissemination of trial-related data does not involve selective presentation of some outcomes while excluding others. Doing so is especially critical in trials for chronic pain, which affects numerous life domains including those rated as very important by individuals experiencing chronic pain (eg, physical and emotional functioning).⁴ Notably, the PMC identified strategies for optimizing pragmatic clinical trials for pain management for military service members and veterans; development and use of a harmonized set of outcome measures endorsed by pain experts is an important element of this ongoing process.

Together with harmonizing outcome data collection, standardizing the assessment of patient phenotypes provides robust advantages. It is widely recognized that there is significant variability in treatment outcomes across patients, contributing to difficulty interpreting clinical trial findings and their subsequent application to clinical practice.⁵ Although substantial heterogeneity among patients can obscure outcomes in certain subgroups of a study cohort, this heterogeneity reflects the need for personalized pain therapeutic strategies. A cornerstone of precision pain medicine is the need to identify characteristics that render an individual patient, or subgroup of patients, more or less responsive to a specific treatment. Many of the same benefits of harmonizing outcome measures apply equally to the development of a unified and standardized set of recommendations for phenotyping. Chief among these benefits is the potential for pooling phenotypic data across studies to achieve enhanced power for subgroup analyses. Such activities will contribute to advancing the science of personalized pain management by helping pain researchers and clinicians identify psychosocial and clinical pain phenotypes that are likely to respond to a given class of interventions.

The PMC phenotypes and outcomes work group

In practice, a common method for standardizing data collection and analysis is through the definition and use of common data elements (CDEs). For example, the NIH CDE Repository provides access to structured definitions of recommended or required data elements to be used in research studies. Previously, in 2005, the Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials (IMMPACT) consensus working group recommended the use of 6 core outcome domains in chronic pain clinical trials (pain intensity, physical and emotional function, global improvement, symptoms, and disposition).⁶ More recently, the NIH’s Helping to End Addiction Long-Term^® Initiative (HEAL Initiative^®) established requirements for HEAL-funded clinical trials and large scale observational studies to collect a core group of CDEs and patient-reported outcomes for 9 pain domains that were identified through partner consensus across the NIH and the pain research community.⁷

Similarly, the PMC relies on interdisciplinary expertise to identify key domains and best practices to harmonize data collection and analyses across trials. One unique characteristic of the PMC is a set of work groups comprised of collaborators with diverse expertise. Co-chaired by a pain psychologist and a health services researcher, the PMC Phenotypes and Outcomes (P&O) Work Group meets monthly to promote harmonization of measurement approaches across studies to examine treatment effect modifiers (ie, subgroups of responders). P&O recommendations have frequently been formulated in collaboration with the Patient Resource Group (PRG), which was developed to support and advise patients, investigators, Work Groups, and study leadership. The PRG is comprised of Veterans and Military Service Members and their dependents and/or family members, all of whom have received education in research principles and practices of clinical research. After being finalized in consultation with the PRG and other stakeholders, recommendations from the Work Group are submitted to the Steering Committee for approval, after which (if approved) they are propagated to the relevant end users (eg, trial PIs).

The P&O Work Group has been instrumental in contributing to the harmonization of phenotypic and outcome data within the PMC. For PMC trials of patients with chronic pain, 100% of the trials assessed high-impact chronic pain and incorporated the PEG as a secondary outcome measure, allowing harmonization on the outcome domains of pain intensity and impact. In addition, over 90% of those trials are contributing harmonized data on phenotypes such as substance use disorders, depression, opioid utilization, impact of COVID, and complementary and alternative medicine. Additional P&O Work Group activities have included rapid development of a brief measure to assess PMC trial participants’ perceptions of COVID-related impacts.⁸ These recommendations derive from evaluation of best practices in the PMC and related networks, as well as literature reviews that summarize key domains in pain-related pragmatic clinical trials (Table 1). It is important to mention that not all assessment domains were amenable to full harmonization. For example, the assessment of physical function/disability was highly heterogenous across trials,³ with some using condition-specific measures such as the Oswestry Low Back Pain Disability Questionnaire, others opting for more general indices of physical function such as the PROMIS Physical Function Scale, and still others using even broader measures such as the EQ-5D questionnaire (which assesses mobility, self-care, etc.).

Table 1.

PMC recommendations for domains on which to consider harmonizing data collection in trial networks. A current list of recommended measures can be accessed here: https://painmanagementcollaboratory.org/research-researchers/research-tools-and-measures/.

Domain	Individual variables	PMC considerations
Sociodemographic characteristics	Age, Sex, Race/Ethnicity, Education, Employment Status, Relationship status	For DoD trials, employment status may not be an optimal framing for individual differences (this variable is omitted for DoD studies).
Pain outcomes	Pain intensity, Pain impact, High-impact chronic pain, Physical function, Quality of life, Healthcare utilization	Trials used a variety of outcome measures, but all incorporated the construct of high-impact chronic pain, as well as the PEG, as an outcome measure, allowing harmonization on the outcome domains of pain intensity and impact. Factors such as physical function showed more heterogeneity of assessment across PMC trials.
Psychosocial phenotype	Depression, Anxiety, PTSD, Resilience	PMC trials used brief, validated psychosocial measures such as the PROMIS Depression and PROMIS Anxiety scales. Such factors play an important role as both potentially crucial phenotypic variables and secondary outcome measures for many of the PMC trials.
Clinical pain phenotype	Pain Location, Pain Duration, Long-term opioid use, Use of non-pharmacologic pain management	PMC investigators developed a harmonized definition of long-term (ie, 90+ days) opioid use, an important pain phenotype that can be evaluated using either electronic health records or information from participant self-reports.
Additional biopsychosocial factors	Sleep disruption, Substance use disorder, COVID impact, Self-management approaches	In 2020, the P&O Work Group rapidly developed and implemented a brief measure of COVID-19 impact within the majority of PMC trials.8

Many biopsychosocial factors act as both phenotypes (when measured at baseline) and outcomes. The formation of mutually exclusive and exhaustive categories of phenotype vs outcome is not generally possible.

Conclusions

Harmonizing outcome and phenotype-related measures within a trial network is complex and challenging but has substantial benefits. We note here that the categories of “phenotyping variable” and “outcome” are overlapping; numerous factors such as pain intensity, psychosocial functioning, sleep quality, medication use, and many others may serve a phenotyping function when assessed at baseline in a treatment study and may also serve as important outcomes over the course of the intervention.^2,4,6 As encouraged by research initiatives such as HEAL, standardization of measurement across studies is increasingly common. Such practices help to accelerate the process of scientific discovery and also enhance progress toward achieving personalized pain medicine.² It is important to keep in mind the continued need to define measures that are clinically relevant, low-burden, broadly accessible, and publicly available.⁹ We acknowledge the need for flexibility and regular re-evaluation of these recommendations as the science and practice of pain medicine evolves. For example, researchers should routinely reassess the terminology and language they use to describe populations and their experiences, exemplified by ongoing efforts to promote the inclusion of individuals from understudied groups in pain research.¹⁰ Collectively, movement toward harmonization of measurement in pain trials has immense potential to benefit the full pain community—including organizations, clinicians, researchers, and individuals and families living with chronic pain.

Funding

Research reported in this publication was made possible by Grant Number U24 AT009769 from the National Center for Complementary and Integrative Health (NCCIH), and the Office of Behavioral and Social Sciences Research (OBSSR).

This work was supported by the Assistant Secretary of Defense for Health Affairs endorsed by the Department of Defense, through the Pain Management Collaboratory—Pragmatic Clinical Trials Demonstration Projects under Awards No. W81XWH-18–2-0003 (Burgess LAMP trial). The U.S. Army Medical Research Acquisition Activity, 820 Chandler Street, Fort Detrick MD 21702–5014 is the awarding and administering acquisition office.

Research reported in this publication was supported by the NIH under Award Numbers UG3/UH3-AT009767 (Heapy/Higgins COPES-ExTRA trial), UG3/UH3 AT012257 (Lovejoy/Morasco CORPs trial), and UG3/UH3AT009761 (Long/Goertz VERDICT trial).

This work was supported [or supported in part] by HSR&D Award # SDR-17–306 (Taylor/Zeliadt APPROACH trial) from the United States (U.S.) Department of Veterans Affairs Health Services Research and Development Service.

This (commentary) is a product of the NIH-DOD-VA Pain Management Collaboratory. For more information about the Collaboratory, visit https://painmanagementcollaboratory.org/.

Conflicts of interest: The authors have no disclosures to report.

Supplement statement

This article appears as part of the supplement entitled “Pain Management Collaboratory: Updates, Lessons Learned, and Future Directions.”

This manuscript is a product of the Pain Management Collaboratory. For more information about the Collaboratory, visit https://painmanagementcollaboratory.org/.