Pain measurement in research and practice
Pain measurement in research and practice
Kurt Kroenke 0 1
0 Indiana University School of Medicine , Indianapolis, IN , USA
1 VA HSR&D Center for Health Communication and Information, Regenstrief Institute, Inc. , Indianapolis, IN , USA
P health conditions in clinical practice as well as the general
ain is among the most prevalent, persistent, and costly
population.1 Moreover, musculoskeletal pain conditions
account for four of the nine most disabling diseases.2 Unlike
medical disorders such as diabetes and hypertension which
can be evaluated and monitored with tests or devices
independent of the patient, pain is reliant on patient report to determine
its severity as well as response to treatment. Patient-reported
outcomes (PROs) are central to assessing symptoms,
functional status, and other domains of health-related quality of life.
Measurement-based care using PROs has proven an essential
component of improving depression outcomes and has a
similar salience for pain. Ideally, pain and other PRO measures
used in research would not differ from those applied in patient
care. After all, the metrics for A1C, blood pressure, and
cholesterol are identical for investigators and clinicians,
making the findings from clinical trials and other research easily
interpretable and directly transferrable to clinical practice.
The Veterans Health Administration (VHA) Health
Services Research and Development Service sponsored a
state-ofthe-art research (SOTA) conference in November 2016
focused on nonpharmacological treatments for chronic
musculoskeletal pain. Although numerous pain measures have been
used in clinical research, there has been insufficient evidence
to recommend one measure over another. Recognizing the
advantages of using a common set of measures across studies,
the SOTA suggested that a process be initiated to identify core
measures for VA research. Therefore, the VHA convened a
work group for which the principal charge was Bto recommend
core outcome measures for pain intensity and interference to
be used in all VHA prospective clinical research studies of
chronic musculoskeletal pain (both interventional and
observational).^ Musculoskeletal pain (e.g., low back and neck
pain, arthritis, fibromyalgia, regional pain syndromes)
represents 75–80% of all pain disorders in clinical practice and the
majority of pain-related disability. Several streams of evidence
have informed the work group’s recommendations including a
systematic review in this issue of the journal,3 an overview of
recommendations formulated by other consensus groups, and
the several Delphi surveys of work group members.
The systematic review, conducted by Goldsmith and
colleagues,3 found for 17 different pain intensity and interference
measures, only 43 articles that met the inclusion criteria which,
in particular, required reporting at least one of four
psychometric characteristics (minimally important difference,
responsiveness, validity, test-retest reliability). While the relatively
small number of articles covering so many measures may in
part reflect the stringency of inclusion criteria, it indicates to a
greater degree the fact that measures are often adopted by
clinical researchers based upon one or a few psychometric
studies. This in turn might be attributable to the paucity of
funding for development and validation of scales; one
prominent exception is the NIH investment of more than $100
million dollars in the Patient-Reported Outcomes
Measurement Information System (PROMIS) scales. Many other
legacy scales have been developed and validated by individual
investigators with either a minimal amount of funding or as
secondary data analyses from larger studies funded for other
reasons (e.g., clinical trials or cohort studies).
Goldsmith and colleagues found that five measures had data
reported on all four psychometric criteria, and seven measures
had data reported on three of the criteria. Based upon this
review and other studies, the VHA measurement work group
has concluded there is no clear psychometric winner among the
pain measures for several reasons: (
) the review found only a
small number of studies for any given measure; (
) there were
substantial methodological and population differences across
these studies; (
) studies comparing the same measures across
different populations can show substantial psychometric
differences4; and (
) studies comparing different measures within the
same population tend to show comparable responsiveness.5, 6
Besides strong psychometrics, other characteristics
considered important, particularly for uptake into clinical practice
and also for research studies where many domains besides
pain are being assessed, include the measure being brief,
selfadministered, freely available (i.e., public domain), easy to
score, widely accessible (e.g., downloadable from websites),
and having actionable scores that can guide diagnostic or
therapeutic decision-making.7 Three characteristics ranked
highly in a Delphi survey of VHA work group members
include the measure being brief, public domain, and feasible
for clinical use including integration into the electronic health
records. Notably, studies have shown that the performance of
brief (less than ten items) or ultra-brief (two to four items) pain
measures is similar enough to longer measures to justify the
use of short measures in most clinical practice and many
research settings.4–6 For example, a reliable and valid
assessment of pain intensity and interference can be done with three
items (PEG scale), five items (PROMIS four-item interference
scale plus a single item numerical rating scale [NRS] assessing
average pain in the past week), or eight items (Brief Pain
Inventory seven-item interference scale plus the NRS item).
One debate in pain measurement is whether separate scores
for pain intensity and interference are necessary or whether a
composite measure combining the two domains into a single
score is sufficient. A composite score is supported by recent
research,5, 6 which is helpful since a single score for either
clinical decision-making or as a primary research outcome is
desirable. The PEG which consists of one pain intensity and
two pain interference items is one such composite measure that
has been recommended by the US Surgeon General’s Turning
the Tide campaign to reduce opioid use and by the National
Pain Strategy as an ultra-brief and public domain measure.
Other outcomes considered important in pain research by
both the VHA measurement work group and other consensus
panels8–10 include physical functioning and depression (ranked
most highly) followed by sleep and anxiety. Although the VHA
work group final report will propose a recommended measure
for both pain intensity and interference as well as these
secondary domains, there is emerging research on how to cross-walk
scores among measures which will facilitate the comparison of
results across studies that use different measures.4, 11
The existence of valid pain and other PRO measures does not
guarantee their use in clinical practice. Completion by interview
and self-administration using paper forms followed by clinician
scoring are relatively inefficient approaches. Computers and
other technology can facilitate measurement in several ways.
PRO administration and automated scoring using iPads or
kiosks minimizes clinic personnel time. Web-based or
interactive-voice response (IVR) phone administration enables
patient completion of PROs at home before or after clinic visits.
Embedding PRO scores into the electronic health record makes
the information available at the point of care for clinical
decision-making as well as longitudinal monitoring. The
development of phone-based apps can foster more sophisticated pain
tracking (e.g., ecological momentary assessment) as well as
prompted self-management and treatment adherence. Pragmatic
trials using collaborative or telecare models as well as vanguard
clinical practices have used one or more of these strategies to
enhance the feasibility and utility of PRO assessment.
Measurement is necessary but not sufficient for improving
the care of patients with chronic pain. Pragmatic trials have
shown that a stepped care approach using non-opioid
analgesics sequentially or in combination can improve pain and
reduce opioid use. Nonpharmacological approaches such as
exercise, cognitive-behavioral therapy, yoga, acupuncture,
chiropractic, and massage can be beneficial, although
strategies to enhance patient adherence as well as policies to
increase the accessibility to and payment for these treatments
are needed. Similar to depression, studies in patients with
chronic pain have shown that collaborative care or telecare
models using nurse care managers are effective and efficient
interventions; however, reimbursement policies need to be
realigned in order to implement these evidence-based models.
In short, measurement must be coupled with delivery of
costeffective treatments in order to optimize pain outcomes.
Accurate and longitudinal measurement helps clinicians and
patients work as partners to achieve target goals for diabetes,
hypertension, hyperlipidemia, weight loss, and many other
medical conditions. Tangible metrics provide a lingua franca for
communicating with patients as well as with other clinicians
who are co-managing a particular patient. Just as the numbers
attached to A1C levels, blood pressure, cholesterol, and body
weight all inform clinical decisions, the scores obtained from pain
measures and other PROs can guide symptom management.
Compliance with ethical standards:
Conflict of interest: The author declares that he does not have a
conflict of interest.
1. Institute of Medicine. Relieving pain in America: a blueprint for transforming prevention, care, education, and research . Washington, DC: National Academies Press; 2011 .
2. US Burden of Disease Collaborators. The state of US health , 1990 - 2010 : burden of diseases, injuries, and risk factors . JAMA . 2013 ; 310 ( 6 ): 591 - 608 .
3. Goldsmith ES , Taylor BC , Greer N , et al. Focused evidence review: psychometric properties of patient-reported outcome measures for chronic musculoskeletal pain . JGIM . 2018 . https://doi.org/10.1007/ s11606-018-4327-8
4. Chen CX , Kroenke K , Stump T , et al. Estimating minimally important differences for the PROMIS pain interference scales: results from three randomized clinical trials . Pain . 2018 , (in press).
5. Krebs EE , Bair MJ , Wu J , Damush TM , Tu W , Kroenke K. Comparative responsiveness of pain outcome measures among primary care patients with musculoskeletal pain . Med Care 2010 ; 48 : 1007 - 1014 .
6. Kean J , Monahan PO , Kroenke K , et al. Comparative responsiveness of the PROMIS pain interference short forms, brief pain inventory, PEG, and SF-36 bodily pain subscale . Med Care 2016 ; 54 ( 4 ): 414 - 421 .
7. Kroenke K , Monahan PO , Kean J . Pragmatic characteristics of patientreported outcome measures are important for use in clinical practice . J Clin Epidemiol 2015 ; 68 ( 9 ): 1085 - 1092 .
8. Dworkin RH , Turk DC , Farrar JT , et al. Core outcome measures for chronic pain clinical trials: IMMPACT recommendations . Pain 2005 ; 113 ( 1-2 ): 9 - 19 .
9. Deyo RA , Dworkin SF , Amtmann D , et al. Report of the NIH Task Force on research standards for chronic low back pain . J Pain 2014 ; 15 ( 6 ): 569 - 585 .
10. Chiarotto A , Terwee CB , Ostelo RW . Choosing the right outcome measurement instruments for patients with low back pain . Best Pract Res Clin Rheumatol 2016 ; 30 ( 6 ): 1003 - 1020 .
11. Cook KF , Schalet BD , Kallen MA , Rutsohn JP , Cella D. Establishing a common metric for self-reported pain: linking BPI Pain Interference and SF-36 bodily pain subscale scores to the PROMIS Pain Interference metric . Qual Life Res 2015 ; 24 ( 10 ): 2305 - 2318 .