Examining patient reported outcome measures for phantom limb pain: Measurement use in a sample of Veterans with amputation

Abstract

Phantom limb pain (PLP) is treated with medications and non-drug treatments. Best clinical practices for measuring treatment outcomes have not been defined.

Purpose:

The objective of this study was to evaluate the internal consistency of patient-reported outcomes measures (PROMs) in a sample of Veterans with lower limb amputation.

Materials and Methods:

The Veteran phone survey included administering PROMs [1) PLP numeric rating scale (NRS), 2) general pain NRS, 3) Pain, Enjoyment, and General Activity (PEG) scale, 4) Patient-Reported Outcomes Measurement Information System (PROMIS) Pain Interference Short Form 6b Replacement, 5) PROMIS Short Form Depression 4a and 6) PROMIS Short Form Anxiety 4a].

Results:

Fifty Veterans (48 male, 2 female; average age: 66 years) completed PROMs. In our sample, 40 Veterans (80%) experienced PLP with an average PLP NRS of 5 (±3.4). Internal consistency of each measure was good to excellent based on Cronbach’s alpha co-efficient of >0.80. Correlations were moderate between PLP NRS and all other measures (≤0.32). Although many Veterans expressed bothersome PLP, the scores reflecting pain interference and impact on function were lower than pain intensity. Consistent use of outcome measures is needed to determine the effect of interventions for amputation-related pain.

Introduction

In the United States alone, the number of individuals living with an amputation is estimated to reach 3.6 million by 2050.¹ Chronic pain associated with amputation is highly prevalent.² Individuals with amputation experience multidimensional pain including phantom limb pain (PLP) that can vary in etiology, intensity, duration, and frequency throughout their lifetime.³ Gaps exist in understanding the current best practices for measuring the effect of treatment. By measuring the impact of clinical care and research components, we can better identify the “active ingredients” of effective treatment to optimize patient engagement and treatment effects given the range of baseline clinical presentations.⁴

Using reliable and valid measures to evaluate PLP and its effects on daily life is critical both to clinicians providing PLP intervention and researchers pursuing novel intervention design and testing. Prior pain rehabilitation intervention studies in the amputation population have primarily focused on a change in the average pain intensity numeric rating scale (NRS; general pain NRS), not to be confused with the PLP NRS, as the primary outcome measure.^5-7 Although the general pain NRS is a common, easily collected rating for clinical populations, measuring additional information about the impact of pain on function and other aspects of pain such as frequency, duration, and factors (aggravating and alleviating) can help guide pain treatment. Further evaluation of the measures in individuals with amputations is indicated as often these individuals experience other sources of problematic pain (e.g., back, knee, shoulder, residual limb pain), potentially confounding the clinician’s assessment and treatment of pain conditions.⁸ There is a lack of clarity on how well general pain measures capture PLP-specific pain.

Further work is warranted to examine the different types of validity and reliability of the measures when used in a population outside of the originally studied population.⁹ The objective of this exploratory, pilot study was to explore use of standardized pain patient-reported outcomes measures (PROMs) in a sample of Veterans with amputations. Our goal is to discern whether the current measures could be improved upon as we embark on a research trajectory, testing both the measures and interventions with the overall long-term goal of coming up with best practices for PLP treatment. Our selection of PROMs reflects several domains to characterize our population (e.g., pain intensity, pain interference, pain self-efficacy, and mental health). Of the measures we used for this study, we hypothesize that the PLP NRS will be more highly correlated with pain interference measures than with other measures as supported by prior work in the amputation population.⁸ Additional exploratory hypotheses included examining if depression and anxiety were associated with higher PLP NRS in our Veteran sample. We wanted to evaluate the internal consistency of PROMs in an amputation population and examine relationships among PLP intensity and measures commonly used in general pain populations.

Materials and Methods

Study Design

This study was an observational study in a convenience cross-sectional sample of VA patients with amputations. The project was assessed through Minneapolis VA Health Care System Institutional Review Board’s review procedures and documented to be Exempt from the requirements 45 CFR 46 under the 2018 Requirements of the Common Rule, Category 2iii. Therefore, written consent was not required.

Participants and Recruitment

From April 2020 to March 2021, we recruited Veterans via mail and followed up with a phone call to determine interest level in participating. We recruited Veteran participants who were receiving amputation care at the Minneapolis VA Health Care System (MVAHCS) Regional Amputation Center (RAC). The MVAHCS RAC serves Veteran patients throughout the Midwest region and is part of the national VA Amputation System of Care (ASoC). Nationally, the population served by the VA ASoC is Veterans with a lower limb amputation from vascular causes and is elderly.¹⁰ Locally, Veterans with lower limb amputation(s) and elderly are the largest population served in our RAC. Potential participants were identified by clinicians (J.K., M.M.) who referred them from a clinical database of Veterans with amputation. Inclusion criteria were 1) current patient of the MVAHCS RAC; 2) at least 18 years old; and 3) prior major lower limb amputation (e.g., above or below knee, bilateral or unilateral). We completed chart reviews on referred Veterans. We excluded those living in long-term care as they may have a surrogate power of attorney. We excluded those receiving hospice care as we wanted to learn about the impact of PLP on daily activities not affected by potential declines. We excluded those who may not be appropriate for this phone survey based on an electronic medical record behavioral flag (placed for prior history of physical or verbal aggression). We excluded those with untreated mental health or cognitive problems that might influence the reliability of their responses.

Measures

To comprehensively describe the Veterans’ status, interviewers administered a battery of measures to assess amputation-specific pain, general pain, pain self-efficacy, and mental health symptoms. A total of nine scales were administered and interviews were scheduled for 60 minutes and took 45-60 minutes.

For amputation-specific measures, we included the Phantom Phenomena Questionnaire and portions of the Trinity Amputation and Prosthesis Experience Scales-Revised (TAPES-R).^11,12 The Phantom Phenomena Questionnaire was administered as an interview, as directed by the developers of the measure. This measure has 23 questions if pain in the missing limb is reported and 9 questions if pain in the missing limb is not reported. This questionnaire guided the discussion with the Veteran in differentiating between residual limb pain, phantom sensation, and PLP. To characterize amputation-specific pain, the Phantom Phenomena Questionnaire includes a 0-10 NRS with no specified timeframe to measure the average PLP NRS, worst PLP NRS, and best PLP NRS and average residual limb pain (RLP) NRS. From the TAPES-R, we included demographic questions such as the duration of time since amputation and a portion of Part II including questions 1-6e. The TAPES-R Part II has questions on use of prosthesis, perception of general health and well-being, RLP, PLP, and other medical problems. The pain intensity and pain interference questions of the TAPES-R have a timeframe of the past 1 week.

For general pain measures, we included the Pain, Enjoyment, and General activity (PEG) scale, the PROMIS-Pain Interference Short Form V1.0 6b Replacement (PROMIS-PI), and Pain Self-Efficacy Questionnaire (PSEQ-4).^13-15 The PEG includes three 0-10 NRS items assessing average pain intensity, pain interference with enjoyment of life, and pain interference with general activity in the past week. Question 1 of the PEG is average general pain NRS in the past week. The PEG is scored as an average of the 3 items.¹⁴ In the original validation paper, reliability (internal validity) of the PEG was α=0.73 in a sample of patients with chronic musculoskeletal pain and 0.89 in a sample of VA clinic patients.¹⁴

The PROMIS-PI short form includes 6 questions that assess pain interference with enjoyment of life, concentration, daily activities, recreational activities, tasks away from home, and socializing. Ratings refer to the past 7 days and use an ordinal rating scale of 1 (not at all), 2 (a little bit), 3 (somewhat), 4 (quite a bit), and 5 (very much).¹³ The PROMIS-PI short form has a reported reliability of 0.94 in individuals with chronic musculoskeletal pain.¹⁶ For the PROMIS-PI and all other PROMIS measures used in this study, the raw score can be converted to a t-score for comparison to reference populations. For the PROMIS-PI, the reference population represents the US general population and clinical populations for comparison. A t-score of 40-60 represents the majority (68%) of the population.¹³

The Pain Self-Efficacy Questionnaire (PSEQ)-4 is a 4-question measure assessing beliefs and confidence to achieve desired outcomes (e.g., doing activities one enjoys, accomplishing goals, and living a normal lifestyle).^17,18 The PSEQ-4 uses an ordinal scale of 0 (not at all confident) to 6 (completely confident). There is no timeframe specified for the PSEQ-4. The internal consistency of the PSEQ-4 is 0.81 in a population of people with chronic musculoskeletal pain.¹⁵

For mental health measures, specifically depression and anxiety, we included the screening tools of the PROMIS Short Form V1.0 – Depression 4a (PROMIS-Depression) and PROMIS Short Form Anxiety 4a (PROMIS-Anxiety).^19-21 Both measures have a timeframe of the past 7 days. The PROMIS-Depression includes four questions to rate their mood, views of self, and loneliness.²⁰ The PROMIS-Anxiety includes four questions about fear, worrying, and hyperarousal.²⁰ Both measures use a 5-point ordinal scale of 1 (never), 2 (rarely), 3 (sometimes), 4 (often), and 5 (always). The reliability of the PROMIS-Depression and PROMIS-Anxiety are 0.93 and 0.86 respectively in a population of patients with chronic musculoskeletal pain.²⁰ For the PROMIS-Depression and PROMIS-Anxiety, the reference population represents the US general population.

Data Collection

All data were collected during a telephone interview. Data collection commenced with the Phantom Phenomena Questionnaire followed by the TAPES-R, PROMIS measures (anxiety, depression, pain interference), PEG, and PSEQ-4. A printed questionnaire was mailed in advance to assist with telephone-based data collection. All Veteran participants verbally consented to participate in the study.

Statistics

The sample size of 50 participants for this study was chosen as reasonable and feasible to gather valuable descriptive information, while also taking into consideration time and cost constraints.²² Additionally, several of our selected measures (i.e., PEG, PSEQ-4) have not yet to be tested in persons with amputation. Taken together, a sample size of 50 was thought to be sufficient for pilot purposes and it was feasible to better inform outcome measurement selection in future, larger clinical trials. We characterized the sample using descriptive statistics. We categorized patients according to their PLP NRS as pain-free (NRS 0), mild pain (NRS 1-3), moderate pain (NRS 4-6), or severe pain (NRS 7-10).²³

We calculated Cronbach’s alpha to assess internal consistency as a form of reliability for all the multi-item measures. Internal consistency reflects the extent that all items on the instrument measure the same construct.²² To assess the relationship of PLP NRS with other measures, we assessed correlations between the PLP NRS and PEG, and PROMIS-PI. To test our primary hypothesis, we used a Pearson correlation coefficient to evaluate if PLP NRS would be more highly correlated with the PEG and PROMIS-PI scales than with the PSEQ-4. Additionally, we secondarily tested our exploratory hypothesis also using a Pearson correlation coefficient that depression and anxiety were associated with higher PLP NRS in the Veteran population.^24,25 We used the published values of 0.10, 0.30, and ≥0.50 to interpret small, medium and large correlations respectively.²⁶ All statistics were computed using R (Version 4.0.3) and SAS^® 9.4 (SAS Institute Inc, 2013).

Results

A total of 187 Veterans were screened for eligibility. Forty-three were excluded based on medical record review. Of 144 Veterans invited, 51 Veterans declined, 35 Veterans did not respond, 7 Veterans scheduled for the survey but did not attend, and 1 Veteran did not have an accurate phone number on file. Of those who did not respond, we followed our protocol with reaching out to the Veteran via mail and two follow-up phone calls. We enrolled 50 Veterans with amputations (35% response rate). Table 1 shows the demographics of our sample. We observed that in these Veterans, 80% reported PLP and 76% reported RLP. If data were missing, the participant was contacted by phone a second time to complete the measure. In the first 16 interviews, one question of the PROMIS-PI was missed by the interviewer and all Veterans were contacted via phone to complete the question.

Table 1.

Veteran Demographic characteristics

Demographic Variables	All Participants n=50 (%)
Age Range (y)
18-39	1 (2%)
40-54	6 (12%)
55-69	15 (30%)
70-84	27 (54%)
85+	1 (2%)
Sex
Male	48 (96%)
Female	2 (4%)
Race
White	46 (92%)
Black or African American	2 (4%)
Asian or Pacific Islander	1 (2%)
Declined to respond	1 (2%)
Ethnicity
Not Hispanic or Latino	49 (98%)
Hispanic or Latino	0 (0%)
Declined to respond	1 (2%)
Amputation Side and Level
Right BKA	24 (48%)
Left BKA	17 (34%)
Bilateral (Varied levels)	9 (18%)
TAPES-R
Time since amputation, average (IQR), in years	15 years (0.3, 53 years)
Unable to recall (n=1)
Prosthesis Use
Average years wearing a prosthesis	14.9 years
Average time wearing a prosthesis per day	11.9 hours/day
Does not use a prosthesis	2 (4%)
Health status+
Very good	9 (18%)
Good	21 (43%)
Fair	15 (31%)
Poor	4 (8%)
Very Poor	0 (0%)
Physical capabilities+
Very good	11 (23%)
Good	21 (43%)
Fair	9 (18%)
Poor	7 (14%)
Very Poor	1 (2%)
Description of RLP among those reporting RLP in the past week (n=30),
Mild	5 (17%)
Discomforting	15 (50%)
Distressing	7 (23%)
Horrible	3 (10%)
Excruciating	0 (0%)
Description of PLP among those reporting PLP in the past week (n=26),
Mild	5 (19%)
Discomforting	10 (39%)
Distressing	6 (23%)
Horrible	3 (12%)
Excruciating	2 (7%)
Phantom Phenomena Questionnaire
Presence of PLP (%)	40 (80%)
Average frequency of PLP among those reporting PLP:
Continuous	3 (7%)
More than 10x/day	0 (0%)
More than 1x/day but less than 10x/day	10 (25%)
Once a day	3 (7%)
Less than 7x/week	7 (18%)
More than 10x/month	2 (5%)
Less than 10x/month	11 (28%)
Other: (i.e., Few times/year, 1-2x/every other month)	4 (10%)
Average duration of PLP among those reporting PLP:
Continuous	1 (2%)
More than 10 hours at a time	3 (7.5%)
Between 5-10 hours at a time	3 (7.5%)
Between 1-5 hours at a time	8 (20%)
1 hour at a time	0 (0%)
30-60 minutes at a time	3 (7.5%)
10-30 minutes at a time	5 (13.5%)
1-9 minutes at a time	10 (25%)
Less than a minute each time	7 (17%)
Presence of RLP (%)	38 (76%)
Presence of non-painful sensations (e.g., itching, sweating, etc.)	26 (52%)
Self-reported use of prescribed pain medications (%Yes / % No)	24 (48%)/26 (52%)

NOTE. Characteristics were identified using portions of the Trinity Amputation and Prosthesis Experience Scales-Revised (TAPES-R) and the Phantom Phenomena Questionnaire. Summary statistics are presented as frequency count (%) for categorical data. Self-reported pain medications may be prescribed or self-administered over-the-counter medications due to other conditions (e.g., arthritis, musculoskeletal pain, other medical conditions). ⁺These questions were omitted due to time in one participant. Abbreviations: BKA: Below knee amputation; PLP, phantom limb pain; RLP, residual limb pain.

Internal Consistency and Group Scores

Using the Phantom Phenomena Questionnaire, the mean and standard deviation of the sample was 7 (3.7), 5 (3.4), and 4 (3.5) of PLP worst intensity NRS, PLP average intensity NRS, and RLP average intensity NRS respectively. We report the distribution of worst PLP NRS scores (figure 1a) and average PLP NRS (figure 1b) ratings. We categorized our sample using the PLP NRS with Veterans reporting 20% (n=10) none, 20% (n=10) mild, 24% (n=12) moderate, and 36% of patients (n=18) were severe, on the PLP NRS. Sample means and internal consistency as measured by Cronbach’s alpha on all other pain measures are reported in Table 2 for the single timepoint. We present scatterplots for PEG and PROMIS-PI scores for Veterans by category of PLP NRS (figure 2A-B). Veterans scores on the PSEQ-4 were consistent with a high level of perceived self-efficacy. Thirty-six percent (n=18) had a t-score >60 on the PROMIS PI, meaning their pain interference with daily activities was greater than the majority of the normative comparison data for the PROMIS PI.

Figure 1 (A-B).

Frequency distribution of the 0 to 10 average general pain numeric rating scale (NRS) where 0 indicates no pain and 10 is worst imaginable for (a) worst phantom limb pain (PLP) intensity NRS and (b) average PLP NRS.

Table 2.

Outcome measures

Outcome Variables (n=50)	Possible Range of Measure	Sample Mean (SD)	Cronbach’s Alpha
PROMIS –Anxiety 4a	40.3-81.6	51 (8.9)	0.81
PROMIS –Depression 4a†	41-79.4	51 (9.6)	0.93
PROMIS – Pain Interference with 6b Replacement	41-78.3	57 (8.5)	0.92
PEG	0-10	3.4 (2.9)	0.96
Average general pain NRS	0-10	3.7 (2.9)	N/A
PSEQ-4	0-24	17 (5.9)	0.88

NOTE. Summary statistics are presented as possible range of the measure, sample mean, and sample standard deviation for normally distributed data. A high score on the PEG indicates more pain intensity and interference. T-scores are reported for PROMIS measures where a t-score of 50 is average. A high score on the PSEQ-4 indicates high confidence in the ability to carry out goals despite the pain. For measures with >1 question, Cronbach’s alpha coefficients are reported and is not applicable for measures with 1 question. ^†One participant had a missing score and was excluded from the PROMIS-4a Depression data. Abbreviations: PEG, Pain, Enjoyment, and General Activity Scale; PLP, phantom limb pain; PROMIS, Patient-Reported Outcomes Measurement Information System; PSEQ, Pain Self-Efficacy Questionnaire-4.

Figure 2 (A-B).

Distribution of participants by average PLP NRS [i.e., none (green triangle), mild (yellow circle), moderate (purple square), or severe (red diamond)] and scores on (a) PEG, (b) PROMIS PI with 6b replacement. Note: NRS: Numeric rating scale, PEG: Pain, Enjoyment, and General Activity Scale, PLP: Phantom limb pain, PROMIS-PI: Patient-Reported Outcomes Measurement Information System-Pain Interference. The x-axis represents the range of scores on each measure. The PEG scale score range is 0 to 10, with higher scores indicating worse pain severity; PEG scores of < 4 are considered to be mild pain (shaded in gray). The PROMIS-PI is reported using population-normed t-scores; t-scores of 40-60 represent 68% of the population and fall into the typical range.

Associations between PLP NRS and other measures

Table 3 displays the Pearson correlations of the PLP NRS score with each PROM. The general pain NRS and the PLP NRS were moderately correlated (r=0.32). The correlation between PLP NRS and depression (r=0.15) or anxiety (r=0.15) were small but positive. The general NRS was strongly positively correlated with depression (r=0.56) and anxiety (r=0.54). The correlation of pain interference as measured by PROMIS PI with average PLP intensity NRS was small (r=0.18, 0.26) and with general pain NRS was strong (r=0.71) The correlation of PEG, which includes both pain intensity and interference items, was moderate with PLP intensity NRS (r=0.26) and very strong with general pain intensity NRS (0.93).

Table 3.

Correlation of outcome measures to PLP NRS

Outcome Variables (n=50)	Corr (r) with the Average PLP NRS	Corr (r) with the Average general pain NRS
PEG	0.26	0.94
Average general pain NRS	0.32	1.0
Phantom Phenomena Questionnaire
PLP Worst intensity NRS	0.81	0.24
PLP Average intensity NRS	1.0	0.32
RLP Average intensity NRS	0.44	0.55
PROMIS – 4a Anxiety	0.15	0.54
PROMIS – 4a Depression†	0.15	0.56
PROMIS – Pain Interference with 6b replacement	0.18	0.71
PSEQ-4	−0.11	−0.61

NOTE. The average general pain NRS was collected from Question 1 of the PEG. Correlations are reported using Pearson’s correlation coefficients. ^†One participant had a missing score and was excluded from the PROMIS-4a Depression data. Abbreviations: NRS, 0-10 numeric rating scale for average PLP; PEG, Pain, Enjoyment, and General Activity Scale; PLP, phantom limb pain; PROMIS, Patient-Reported Outcomes Measurement Information System; PSEQ, Pain Self-Efficacy Questionnaire-4, RLP, residual limb pain.

Discussion

The purpose of this study was to explore use of standardized pain PROMs in a sample of Veterans with amputations.

Veteran Sample-Pain Intensity, Frequency, Duration, and Interference

In our Veteran population, a majority (60%) of participants reported scores in the moderate or high range on the PLP NRS, with an average rating across all subgroups of 5 on a 0-10 scale (Figure 1A). We observed diverse responses to questions about the frequency and duration of PLP episodes (Table 1). The frequency range was from a few times per year to continuous. The duration ranged from <1 min at a time to continuous. The timeframe of pain interference measures, often the past 7 days, may not capture the important variability in pain episodes reported in this study. PLP may be more variable than other types of chronic pain, so existing measures could under-identify individuals who might benefit from pain treatment. For example, some Veterans described how they experience PLP once per month, but it can be quite disruptive to their lifestyle for 1-2 days following the onset. Notably, one participant reported an average PLP of 0 out of 10 but a worst PLP of 10 out of 10. This participant reported he didn’t experience PLP on average and could go months without it. However, when PLP occurred, the pain intensity was “10 out of 10,” lasted >24 hours, and required use of prescribed pain medications. There are several factors that influence the impact of pain, such as frequency and duration. A person may report a low pain rating, but if the pain is constant, it may greatly affect the person’s activities and participation. These findings demonstrate the clinical complexity of patients with amputation and measuring their pain.

Additionally, patients with vascular-related amputations often have multiple comorbidities further complicating their care and potentially influencing the rating on general pain NRS. The comorbidities and associated pain may interfere more than PLP pain with function. The potential exists for an improvement in the Veteran’s overall pain management with intervention that reduces pain interference and positively impacts their quality of life.^27,28 Further research to understand the patient-specific factors of Veterans with amputation could guide future work for intervention design and responder analysis. For example, if a certain patient-specific factor (e.g., sleep, mobility status, pain qualities and contextual consideration [e.g., pain at rest vs. pain during movement]) is associated with reduced pain interference in daily life, interventions could target that patient-specific factor.

Psychometric considerations of selected pain measurement tools

The measures tested demonstrated high internal consistency in this study population of Veterans with amputations. Some measures (e.g., PROMIS) provided a timeframe to anchor the rating of responses; however, the Phantom Phenomena Questionnaire was constructed without a time-based anchor (e.g., within the past 7 days). Without a timeframe for the question, we are unable to discern if the Veteran is rating the experience of chronic pain in general (e.g., over the course of their life) or the experience of chronic pain in a set timeframe. Not having an anchor may skew the severity ratings, since more severe episodes are likely to be more memorable.²⁹ Our clinician collaborators subjectively report difficulty with what is the optimal timeframe for asking questions about PLP given some patients report infrequent PLP episodes. Potentially the lack of association observed with PLP NRS and pain interference may be related to the timeframe of measures and frequency of PLP experienced.

In our sample, we observed a moderate correlation between general pain NRS and PLP NRS which aligns with our clinical experience as musculoskeletal pain is common with up to 82% experience persistent low back pain and 42% experience osteoarthritis in the contralateral limb.^3,30 We observed an association between anxiety and depression screening measures and functional pain measures (PEG, PROMIS-PI, PSEQ-4) and general pain NRS but not with the PLP NRS. Prior studies exploring the relationships between depression, anxiety, and amputation pain have reported conflicting results.³¹ One systematic review reported that depression and anxiety declined to the rates observed in the general population within two years post-amputation.³² In contrast, a study of 124 active military service members who experienced traumatic lower limb amputation demonstrated an association between neuropathic pain of PLP and depression.²⁴ Morgan et al., 2017 completed a large-scale survey study (electronic or paper formats) and also observed that depression and anxiety were associated with pain interference (i.e., PROMIS-PI).⁸ Additional previous reports in the literature on the impact of persistent pain on mental health conditions wherein pain impacts physical functioning and overall well-being.^2,33,34 The impact of amputation and mental health is widely understudied research domain and reporting baseline values with respect to PLP NRS is a useful starting point.³⁵ Large scale, mixed methods are warranted using comprehensive measurements of anxiety and depression to examine the relationship between these aspects of mental health and amputation-related pain to better define treatment targets. These types of investigations could be accomplished using a national sample through the VA.

We observed a high level of reported perceived pain self-efficacy as measured by the PSEQ-4 including those participants with severe PLP NRS. Prior work suggests that individuals with high pain self-efficacy may benefit from programs that focus on self-management approaches for pain³⁶ (e.g., exercises, mindfulness, cognitive-behavioral techniques). This work reinforces the potential for these self-management approaches to be beneficial in the amputation population when reporting high pain self-efficacy. Unlike those individuals with low self-efficacy where more intense pain programs with face-to-face sessions may be needed for adequate support. This finding is interesting as in other populations, such as chronic low back pain, higher self-efficacy is associated with decreased disability.³⁷ Our results suggest that some individuals with high PLP intensity also exhibit high pain self-efficacy. The possibility exists that high pain self-efficacy reduces pain interference, but this is yet to be extensively studied.

We expected PLP NRS to be correlated with pain interference as this is what we commonly hear reported in the clinical environment. We observed that general pain NRS was associated with pain interference as observed in other patient populations with chronic pain, but PLP NRS was not.^38,39 Additionally, a relatively small proportion (30%) of the Veterans reported pain interference scores greater than the normative population. Our pilot data suggests an association between general pain intensity and pain interference. The use of PEG ratings provides another perspective of pain interference in the enjoyment or and performance of daily activities that may be more sensitive to detect pain interference however this would need to be studied in larger samples. The PEG does include one question of pain intensity so the very strong correlation to general pain intensity NRS is expected.

Limitations

This study was exploratory in nature with limitations including our small sample size which may have been underpowered to identify the relationships among outcome measures. We attempted to minimize question fatigue with the interactive nature of an interview. We did not collect current medications, reason for amputation, or additional co-morbidities that could impact overall pain from our participants. Our sample of Veterans limits the generalization of these findings to the broader amputation population. We made every effort to recruit female Veterans and diverse Veterans into this study. Only having 2 female Veterans and 8% from non-white backgrounds participate limits the generalization of these findings. However, this study is a valuable contribution to the literature by evaluating and identifying pain outcome measures in this complex population and may help to generate future research.

Future Research

Veterans identified that PLP is a distinct pain experience. Future work is needed in diverse populations to examine: 1) how effectively the current pain PROMs reflect the lived experience of PLP from both the patient and clinician perspectives and solid recommendations for clinicians in terms of pain outcome measurement, 2) the scales’ test-retest and sensitivity to change following novel pain treatments, important for future intervention studies, 3) the influence of depression and anxiety on PLP, and 4) exploration of time-based anchors for optimizing PLP assessment. Large scale, multisite studies could evaluate the influence of age, sex, time since amputation, and the influence of co-morbidities. Such studies could be conducted within national healthcare systems such as the VA Health Care System. This work will allow for larger, more robust studies to be done to establish guidelines for researchers for the use of pain scales for PLP with specific recommendations.

Clinician Take-Away

The PLP NRS should be used with caution as it demonstrated no strong associations with other pain scales (PEG, general pain NRS, PROMIS-PI). While measuring PLP intensity is important, it may not give enough information on its own to evaluate how the PLP is affecting the person’s daily activities. Using one of the tested scales in isolation could result in a missed opportunity to intervene clinically. At this stage, it is recommended to administer more than one measure if a clinician or researcher were to select from any of the measures tested in this study. To best serve the long-term needs of the patient, PLP warrants evaluation separate from the measures that ask about general pain.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author, [T.R.], upon reasonable request.