Development of the barriers to opioid access scale among individuals with chronic pain

Abstract

Background

Measurement of accessibility is a crucial pillar in assessing equity of access to pain treatment, particularly in the context of reducing opioid prescribing in response to rising overdose deaths in the United States.

Objectives

The aim of this study was to develop an instrument to measure barriers to prescription opioid access among individuals with chronic pain and test its psychometric properties.

Methods

This study used a cross-sectional online survey of a convenience sample of adults (>18 years) who reported any type of pain for at least 45 days or more in the previous 3 months. The survey captured demographic characteristics, self-reported medication use characteristics, and measures such as the Brief Pain Inventory-Short Form and the PROMIS Global Health measure, along with an item pool of potential questions that measure barriers to opioid access.

Results

Respondents (N = 200) were 89 % women, 86 % White, averaging 45.32 years old (SD:11.79), and reported poor quality life. Two subscales, Access to Care and Patient Concerns, were identified for the Barriers to Opioid Access Scale with good internal consistency reliability (α = 0.909 and 0.835, respectively). In multivariable analyses, the Access to Care subscale was associated with the PROMIS mental health score (−2.44; 95 % CI: −3.77, −1.11), and the Patient Concerns subscale was associated with self-reported frequency of opioid use (−0.70; 95 % CI: −0.99, −0.40).

Conclusions

The newly developed BOAS has the potential to serve as a tool for capturing quality of pain treatment as well as measuring the impact of policy changes on the quality of treatment provided to patients with chronic pain.

Highlights

• •This study developed the Barriers to Opioid Access Scale (BOAS) for chronic pain.
• •The BOAS includes two subscales: Access to Care and Patient Concerns.
• •BOAS captures significant barriers and shows satisfactory psychometric properties.
• •Access to Care was associated with mental health after adjusting for confounders.
• •BOAS was not associated with physical health after adjusting for confounders.

1. Introduction

Although there exist a variety of treatment approaches to mitigate pain among patients,¹ opioids remain a relevant pharmacological intervention in chronic pain management.^2,3 While there is earnest debate over benefits and safety of the use of prescription opioids for long-term pain,⁴ available guidelines and federal declarations recognize opioids will be required to manage pain for some individuals, for three reasons. First, prescribed opioids have some efficacy for chronic pain, according to the review commissioned by the Agency for Healthcare Research & Quality.⁵ Second, with cancer or palliative care needs, there is broad ethical acceptance that opioids are a crucial option. Third, among the estimated 8–10 million Americans who receive opioids long-term, federal agencies have recommended individualized decision-making and denounce across-the-board reductions.^6,7 in part because dose reductions and stoppages have been associated with adverse effects including suicidal events,⁸ death by overdose,^9,10 mental health crisis,^11,12 and medical deterioration.¹³ In light of such data, the 2022 Centers for Disease Control and Prevention (CDC) Clinical Practice Guideline for Prescribing Opioids for Pain acknowledges that some patients will require initiation or continuation of opioid therapy, when “expected benefits for pain and function are anticipated to outweigh risks to the patient.”⁶

Nevertheless, the use of prescription opioids for pain management has drawn scrutiny because of the North American overdose crisis.¹⁴ Concerns about this crisis have drwn an extensive national response from governmental and private institutions,¹⁵ and from clinicians attempting to recalibrate their practices. By the end of 2017, 26 states implemented laws on opioid prescriptions, while others regulated pain clinics and prescription drug monitoring programs.¹⁶ Private health plans have limited prescriptions as well.¹⁷ The responses achieved a reduction, with prescription rates in 2022 less than one-third of those observed in 2006, according to the Centers for Disease Control & Prevention (CDC).18., 19., 20.

As a result, prescribed opioids have become more difficult to obtain among individuals with chronic pain. For example, changes in Medicare Part D formulary restrictions have influenced prescribing rates.²¹ A survey conducted by the National Fibromyalgia & Chronic Pain Association (NFMCPA) found that 18 % of respondents faced at least one occasion where an opioid prescription was not filled, either because of a lack of stock, or a pharmacist's disagreement with the dose.²² An audit study found that 40 % of primary care clinics in Michigan refused to accept new patients on opioids.²³ In other cases, prescribers were uncomfortable with pain patients or unwilling to prescribe opioids.²⁴ Other barriers may include prior authorization, pill limits, or dosage caps, and the implementation of quality indicators that measure both opioid prescribing rates and dosages.²⁵

While many hold that overall opioid prescribing reductions in the United States represent a favorable recalibration in care²⁶ this paper approaches access to prescribed opioids for pain as a phenomenon worth measuring. Measurement of prescription opioid accessibility is necessary to understanding whether these changes have played out equitably. The equity concern commands attention even for treatments that are suitable only for some patients, if not all, as would be the case with most therapies, ranging from biologics to chemotherapy.^27,28 No valid survey measures are available for measuring equity of access to opioid medications, or the impacts of that access on patients. Further, we are aware of no effort to profile accessibility and barriers opioid pain medication from the point-of-view of patients with long-term pain. Therefore, this study aims to quantify barriers to prescription opioid access among adults with chronic pain. The specific objectives were to (1) develop an instrument to quantify barriers to opioid access among individuals with chronic pain and (2) identify the impact of barriers to opioid access on patterns of medication use and quality of life.

2. Methods

This study used a cross-sectional online survey of a convenience sample of individuals with chronic pain living in the United States. The study population included individuals over 18 years of age who reported any type of pain within the past 6 months, and experienced pain for at least 45 days or more in the previous 3 months, consistent with existing literature.²⁹ The panel of potential respondents was obtained from Rare Patient Voice, a patient panel vendor,³⁰ and online forum used by over 200,000 patients and caregivers representing over 600 medical conditions. The panel vendor conducts recruitment of patients and caregivers at disease-specific conferences and through social media and offers them opportunities to participate in research. Individuals who met the inclusion criteria in the current study were invited to complete the survey and were offered $30 as an incentive. All study procedures and analyses were approved by the University Institutional Review Board (protocol #22×-020).

2.1. Study measures

The survey was programmed into the Alchemer survey platform for distribution and included demographic characteristics, clinical measures, and medication use characteristics. Demographic characteristics included patient age, race, ethnicity, education, smoking status, marital status, employment status, receipt of any palliative care, and housing status. In addition, characteristics specific to their pain, such as the frequency and types of pain experienced in the previous 6 months were also captured. A 6-month time frame was chosen in order to enhance clinical relevance of the condition and minimize recall bias. Use of opioid and non-opioid prescription medications, over-the-counter medications, and alternative therapies for the management of pain, including marijuana, were also queried. Respondents self-reported other diagnosed chronic conditions (including arthritis, asthma, cancer, depression, diabetes, heart disease, high blood pressure, insomnia, kidney disease, liver disease, lung disease, stroke, ulcer or stomach disease, or others).

Individual characteristics of pain were captured using the Brief Pain Inventory-Short Form (BPI-SF).^31,32 The BPI-SF has been widely used in several populations of individuals with pain, including in clinical trials, and has demonstrated adequate validity.33., 34, 35, 36, 37 It includes four items that capture pain severity (ratings of pain at its “worst,” “least,” “average,” and “now” or “current pain”) and seven items that capture how much pain has interfered with daily activities (general activity, walking, work, mood, enjoyment of life, relations with others, and sleep). Test-retest reliability for the BPI-SF has shown reliabilities ranging from 0.72 to 0.97 for both subscales, and there is good support for the factor structure of the scale.33., 34, 35, 36, 37

Respondent ratings of quality of life were captured using the Patient-Reported Outcomes Measurement Information System (PROMIS) Global Health Scale version 1.2.³⁸ The PROMIS Global Health measure is a commonly-used measure of quality of life and includes a 4-item physical health summary score and a 4-item mental health summary score. Scoring for the PROMIS measure is conducted by uploading an anonymized dataset of survey responses to HealthMeasures Scoring Service website.³⁹ PROMIS Global Health scores are standardized such that the mean quality of life score for the population is 50 and the standard deviation is 10. The PROMIS measure has demonstrated satisfactory psychometric properties and also been validated specifically in patients experiencing chronic pain.^40,41

2.2. Barriers to Opioid Access Scale (BOAS)

The preliminary development of the instrument to measure barriers to opioid access included a review of literature concerning access to care, barriers to prescription medications, and other research related to the management of chronic pain. Based on input from clinicians, patient advocates, and researchers, and a literature review, an initial item pool of 120 questions rated on a Likert response format from 1 through 5 was developed, where 1 was strongly disagree and 5 was strongly agree. Redundant items were retained at this stage for future refinement. A series of in-depth cognitive interviews were then conducted with a convenience sample of five individuals with chronic pain known to study investigators. The goal of the cognitive interviews was to ascertain item readability, clarity, face validity and to prune redundant items. As the original item pool included several iterations of each question, this step was critical in retaining only the most relevant and clearly worded questions. The resultant refined item pool included 32 questions (Appendix A) and was administered to the online study sample, retaining the 5-point Likert response format.

2.3. Data analysis

All data management and analysis were conducted using IBM SPSS version 27.0. Descriptive statistics were estimated for the study sample and established scoring methods were used to calculate measure scores for the BPI-SF and the PROMIS Global Health measure. Responses to the BOAS were subject to exploratory factor analysis (EFA) with varimax rotation, with the aim of identifying underlying factors. The Kaiser-Meyer-Olkin (KMO) measure was used to assess whether the sampling of items was adequate for factor analysis and Bartlett's test of sphericity was used to determine if there was sufficient correlation among the items. The number of factors was identified using both a scree plot and parallel analysis. Items with poor communalities (less than 0.5), weak loadings (below 0.4), or cross-loading (loadings above 0.3 for more than one factor) were considered for deletion from the final scale.⁴² Thresholds for communalities and factor loadings were obtained from previously published research. The reliability of the resulting scales was estimated using Cronbach's alpha. Tests for face validity and content validity were considered to be completed through the use of pre-testing prior to deployment of the item pool. A test for convergent validity was proposed based on Andersen's model of access to care.⁴³ In this model, Andersen contends that the use of healthcare services is directly related to consumer satisfaction as well as improvement in perceived health status. This model also argues that effective access to care is the use of healthcare services that result in improvement in health status, satisfaction, or both. Therefore, it was hypothesized that patients who faced barriers for accessing needed opioid medications were likely to have decreased use of opioid medications as well as worse health related quality of life. In order to test for convergent validity, the sum scores of the resulting subscales derived from the results of the EFA model were used as predictors in multivariable regression models. Separate models were estimated using the PROMIS Global physical health and mental health scores as well as self-reported frequency of opioid use in the previous 6 months as dependent variables. Linear regression was used in predicting physical and mental health scores. Frequency of opioid use in the previous 6 months, a 6-level item ranging from no use to daily use, was treated as a continuous variable and linear regression was also used. In all regression models, demographic and clinical characteristics measured in the survey were included as covariates.

3. Results

A total of 1760 individuals were invited to participate in the survey and 321 initiated the survey, a response rate of 18.24 %. Of the 321 initiators, five were terminated because they were ineligible and 200 respondents completed the survey in full. Of these 200 respondents, 89 % (n = 178) were women and 86 % (n = 172) were White individuals. They were 45.32 years old on average (SD: 11.79). A majority of the sample reported being unemployed as a result of medical disability (59.5 %; n = 119). Other demographic characteristics of respondents are provided in Table 1. The burden of these conditions among respondents was also demonstrated through their quality of life scores. On average, respondents reported poor quality life with a physical health score of 30.23 (SD: 6.12) and mental health score of 37.75 (SD: 8.57), well below the population norm of 50.

Table 1.

Demographic characteristics of study sample.

Characteristics	N	%
Age [Mean, SD]	45.32	11.79
Gender
Male	19	9.5
Female	178	89
Race
Black	18	9.0
White	172	86.0
Other	8	4.0
Ethnicity
Not Hispanic or Latino	188	94.0
Hispanic or Latino	10	5.0
Prefer not to answer	2	1.0
Education
Some high school or less	3	1.5
High School /GED	17	8.5
Trade or other technical school degree	11	5.5
Some college	79	39.5
Bachelor's degree	52	26.0
Graduate or professional degree	38	19.0
With whom do you currently live
Spouse	94	47.0
Relative	91	45.5
Friend	11	5.5
Pet	79	39.5
Alone	26	13.0
Domestic or health-related help	2	1.0
Other	5	2.5
Smoking or Vaping
Every day	28	14.0
Some days	12	6.0
No	158	79.0
Prefer not to answer	2	1.0
Marital status
Never married	61	30.5
Married	71	35.5
Cohabitation/ domestic partner	14	7.0
Separated	7	3.5
Divorced	41	20.5
Widowed	5	2.5
Prefer not to answer	1	0.5
Employment status
Employed	53	26.5
Unemployed	16	8.0
Retired	10	5.0
Disabled due to my medical condition	119	59.5
Prefer not to answer/ do not know	2	1.0
Reporting receiving palliative care	18	9.0
PROMIS Quality of Life [Mean, SD]
Physical Health T score	30.23	6.12
Mental Health T score	37.75	8.57

NOTE: The frequencies for some questions do not add up to the full sample (N = 200) because of missing values.

Table 2 shows information about clinical characteristics and opioid use patterns among respondents. Nearly all respondents reported muscle pain (99 %; n = 198), migraine pain (98 %; n = 196), or other types of joint pain. Respondents self-reported an average score of 6.93 (SD: 1.97) on the pain interference subscale of the BPI and an average score of 6.01 (SD: 1.45) on the pain severity subscale. Respondents reported having an average of 5.03 chronic conditions (SD: 1.96) and 72 % (n = 144) reported using opioid medications for pain relief within the past 6 months. Among those who reported opioid use within the previous 6 months, a majority reported using opioids for more than 2 years (72.6 %; n = 106), used opioids every day (75.3 %; n = 110), and were on a stable dose (56.2 %; n = 82). Nearly all (95.5 %; n = 191) respondents reported using non-opioid prescription medications for pain relief within the previous 6 months, with NSAIDS (65.4 %; n = 125) and muscle relaxants (64.9 %; n = 124) being the most commonly reported medication classes. Use of over-the-counter pain relief (71.5 %; n = 143), and other alternative therapies was also common. While most respondents did not use any marijuana in the past year (64.5 %; n = 129), among those who did report marijuana use, frequency of past-year use was fairly high with 28.2 % (n = 20) and 39.4 % (n = 28) reporting use between 91 and 270 days (∼3 to ∼4 months), and more than 270 days (more than ∼4 months), respectively. Most respondents (80 %; n = 162) reported not carrying naloxone. These patterns are depicted in Table 3.

Table 2.

Characteristics of pain and opioid use.

Characteristics	N	%
Types of pain experienced in previous 6 months
Muscle pain	198	99.0
Migraine	196	98.0
Knee pain	195	97.5
Lower abdomen pain	191	95.5
Hip pain	190	95.0
Other joint pain	188	94.0
Back pain	187	93.5
Neuropathy	131	65.5
Neck pain	129	64.5
Headache	127	63.5
Fibromyalgia	79	39.5
Osteoarthritis pain	74	37.0
Tooth pain	61	30.5
Rheumatoid arthritis pain	35	17.5
Radiculopathy	28	14.0
Chest pain	26	13.0
Other pain	20	10.0
Cancer pain	9	4.5
Pancreatitis	3	1.5
Frequency of pain in previous 90 days
45 to 59 days	20	10.0
60 days or more	180	90.0
Brief Pain Inventory [Mean, SD]
Pain interference	6.93	1.97
Pain severity	6.01	1.45
Number of chronic conditions [Mean, SD]	5.03	1.96
Opioid use in previous 6 months
Yes	144	72.0
No	54	27.0
Unsure	2	1.0
Duration of use of opioids
Less than 1 month	8	5.5
3 to 6 months	5	3.4
6 months to 1 year	15	10.3
1 to 2 years	12	8.2
More than 2 years	106	72.6
Frequency of use of opioids
Daily	110	75.3
5 to 6 times per week	4	2.7
3 to 4 times per week	5	3.4
1 to 2 times per week	11	7.5
Less than once a week	16	11.0
Change in dose of opioids within the previous 6 months
Increased	39	26.7
Decreased	25	17.1
Stayed the same	82	56.2

Table 3.

Medication use characteristics among respondents.

Characteristics	N	%
Use of non-opioid prescription medication for pain relief within previous 6 months	191	95.5
Type of non-opioids used
Gabapentin	81	42.4
Benzodiazepines	40	20.9
Muscle relaxants	124	64.9
Antidepressants	98	51.3
Antipsychotics	7	3.7
NSAIDs	125	65.4
Antimigraine	2	1.0
Other anticonvulsants	20	10.5
Other sedatives	18	9.4
Other non-opioid	35	18.3
Frequency of use of non-opioids
Daily	129	67.5
5 to 6 times each week	23	12.0
3 to 4 times each week	16	8.4
1 to 2 times each week	8	4.2
Less than once a week	15	7.9
Use of over the counter medications to treat pain in the previous 6 months
Yes	143	71.5
No	57	28.5
Frequency of use of over the counter medications
Daily	43	30.1
5 to 6 times each week	30	21.0
3 to 4 times each week	33	23.1
1 to 2 times each week	19	13.3
Less than once a week	18	12.6
Alternative therapies used within previous 6 months
Acupuncture	17	8.5
Chiropractor	37	18.5
Exercise	106	53.0
Massage	94	47.0
Meditation	67	33.5
Mindfulness	103	51.5
Nerve stimulation	41	20.5
Occupational/physical therapy	70	35.0
Weight loss	45	22.5
Yoga	43	21.5
Carries Naloxone
Always	20	10.0
Sometimes	18	9.0
No	162	81.0
Frequency of previous year marijuana use
None	129	64.5
90 days or less	23	11.5
91 to 270 days	20	10.0
> 270 days	28	14.0

3.1. Barriers to opioid access scale

The EFA model for the item pool was found to have a KMO test statistic of 0.878 and a significant Bartlett's test of sphericity (p < 0.0001). Seven items were deleted for having communalities less than 0.5. When the remaining 25 items were inserted into the EFA model, seven factors were found to have an eigenvalue greater than 1. Based on the sharp cutoff found in the scree plot as well as the results from the parallel analysis, it was decided that a two-factor solution might provide the best combination of variance explained and parsimony. The resulting two-factor solution was found to have minimal correlation (0.143) between factors when using oblimin rotation. Therefore, varimax rotation was used to obtain a final two-factor solution. Finally, three items were deleted for cross-loading on both factors and one final item was deleted for having poor loading. A two-factor solution consisting of 20 items was extracted as part of the BOAS. Based on the items loading on the factor, the two subscales were named the ‘Access to Care’ subscale and the ‘Patient Concerns’ subscale and included 15 items and 5 items, respectively. The specific wording of each item along with the standardized factor loading and item means are presented in Table 4. Both subscales demonstrated adequate reliability (Access to Care subscale: 0.909; Patient Concerns subscale: 0.835; Table 5).

Table 4.

Barriers to prescription opioid access – item means and factor loadings.

Item	Name of subscale	Standardized loadings	Mean (SD)
I worry that if I complain of inadequate pain management with opioids my doctor will think I am drug-seeking	Access to Care	0.639	3.34 (1.45)
My physician that provides me the bulk of my pain management will not prescribe me the amount of opioid medication that I need	Access to Care	0.750	2.46 (1.30)
My insurance has policies that prevent me from getting the opioid medications that I need	Access to Care	0.636	2.58 (1.32)
Laws restricting the number of opioid medications that can be dispensed at one time make getting pain medication difficult for me	Access to Care	0.777	2.85 (1.46)
I do not get the treatment I need to control my pain	Access to Care	0.683	3.25 (1.37)
My daily life is affected severely because of my untreated pain	Access to Care	0.519	3.85 (1.17)
I feel comfortable talking with my healthcare provider about my opioid medications1	Access to Care	0.552	2.29 (1.28)
My physician has denied me opioid medications because of prescribing regulations	Access to Care	0.761	2.50 (1.44)
I have difficulty finding a pharmacy that can fill my opioid prescription	Access to Care	0.645	1.96 (1.05)
I do not take my opioid medication as often as I need to because I am not sure I can get a refill	Access to Care	0.622	2.31 (1.29)
I cannot get a prescription for opioids because my doctor or my pharmacist thinks I may be abusing them	Access to Care	0.547	1.61 (0.89)
It is easy to get a prescription for opioids medications whenever I need it1	Access to Care	0.527	3.50 (1.31)
I have to travel very far to get my opioid medication	Access to Care	0.475	1.75 (0.94)
I cannot access my opioid medication because my pharmacy does not allow someone else to pick up my medication	Access to Care	0.465	1.89 (0.98)
I have difficulty finding a doctor that can prescribe me opioid medications	Access to Care	0.812	2.60 (1.42)
I do not use as much opioid medicine as I need to because I do not want to become addicted	Patient Concerns	0.542	3.10 (1.41)
I am worried that if I use my opioids as prescribed then I will have an overdose	Patient Concerns	0.585	1.68 (0.98)
I delay taking my opioid medications because I worry about their impact on some areas of my ability to function	Patient Concerns	0.765	2.55 (1.39)
I do not take my opioid medication when I need it because I am afraid it will interact with my other medications to harm me	Patient Concerns	0.811	2.19 (1.17)
I do not take my opioid medication when I need it because I do not like the side effects of the medications	Patient Concerns	0.861	2.47 (1.39)

Note: All items were on a 5-point Likert scale from 1 = strongly disagree to 5 = strongly agree

1: Item was reverse coded prior to scoring

Table 5.

Reliability of subscales in the barriers to opioid access scale.

Subscale	Mean (SD)	Cronbach's alpha
Access to Care	2.58 (0.83)	0.909
Patient Concerns	2.40 (0.99)	0.835
Total scale score	2.49 (0.69)

As a descriptive matter, we note that among the 20 BOAS items, five obtained a mean response greater than 3 (more than ‘neutral’) on a 5-point scale, where endorsement signaled a difficulty in care. For example, the item ‘I do not get the treatment I need to control my pain’ obtained a mean response score of 3.25 (SD: 1.37). These descriptive data demonstrate the endorsement of these barriers, indicating the importance of measuring these characteristics among individuals with chronic pain.

3.2. Assessment of validity

Table 6 shows the relationship between the subscales of the BOAS and the PROMIS physical and mental health scores. After adjusting for hypothesized confounders, the Access to Care subscale was significantly associated with the PROMIS mental health score (−2.44; 95 % CI: −3.77, −1.11; p < 0.0005), but not the PROMIS physical health score (−0.85; 95 % CI: −1.77, 0.07; p = 0.070), and the Patient Concerns subscale was not significantly associated with either the PROMIS mental health score (−0.24; 95 % CI: −1.30, 0.81; p = 0.651) or physical health score (−0.23; 95 % CI: −0.96, 0.50; p = 0.532). Table 7 shows that the Patient Concerns subscale (−0.70; 95 % CI: −0.99, −0.40; p < 0.0005) but not the Access to Care subscale (−0.32; 95 % CI: −0.70, 0.05; p = 0.088) was significantly associated with self-reported frequency of opioid use, after accounting for hypothesized confounders.

Table 6.

Relationship between barriers to opioid access subscale and PROMIS quality of life scores.

Variable		PROMIS physical health		PROMIS mental health
		Estimate (95 % CI)	p-value	Estimate (95 % CI)	p-value
Barriers to Opioid Access Scale
	Access to Care subscale	−0.85 (−1.77, 0.07)	0.070	−2.44 (−3.77, −1.11)	<0.0005
	Patient Concerns subscale	−0.23 (−0.96, 0.50)	0.532	−0.24 (−1.30, 0.81)	0.651
Past year marijuana use frequency
	Never	Reference
	90 days or less	0.23 (−2.07, 2.53)	0.843	−1.43 (−4.75, 1.90)	0.398
	91–270 days	−0.85 (−3.28, 1.57)	0.488	−2.05 (−5.56, 1.45)	0.250
	>271 days	0.92 (−1.20, 3.03)	0.393	−0.09 (−3.14, 2.97)	0.954
Race
	White	Reference
	Black	−0.04 (−2.55, 2.47)	0.975	0.18 (−3.45, 3.80)	0.924
	Other	−3.68 (−7.28, −0.07)	0.046	−1.82 (−7.03, 3.40)	0.493
Age		0.03 (−0.03, 0.10)	0.282	0.16 (0.07, 0.25)	0.001
Brief Pain Inventory
	Pain interference	−1.20 (−1.66, −0.73)	< 0.0005	−1.57 (−2.24, −0.90)	< 0.0005
	Pain severity	−0.73 (−1.32, −0.15)	0.014	−0.23 (−1.08, 0.61)	0.584
Number of days with pain in previous 3 months
	45 to 59 days	0.71 (−1.71, 3.13)	0.563	−0.36 (−3.87, 3.14)	0.838
	≥ 60 days	Reference
Palliative care		0.91 (−1.58, 3.40)	0.473	0.776 (−2,83, 4.38)	0.671
Number of comorbidities		−0.56 (−0.97, −0.15)	0.007	−0.57 (−1.16, 0.02)	0.059

Table 7.

Relationship between barriers to opioid access subscale and self-reported frequency of opioid use.

Variable		Frequency of use of opioids
		Estimate (95 % CI)	p-value
Barriers to Opioid Access Scale
Access to Care subscale		−0.32 (−0.70, 0.05)	0.088
Patient Concerns subscale		−0.70 (−0.99, −0.40)	< 0.0005
Past year marijuana use frequency
Never		Reference
<90 days		−0.66 (−1.59, 0.27)	0.160
91–270 days		−0.03 (−1.01, 0.95)	0.950
>271 days		−1.00 (−1.85, −0.14)	0.023
Race
White		Reference
Black		−0.28 (−1.29, 0.74)	0.592
Other		−0.46 (−1.92, 0.99)	0.531
Age		0.01 (−0.01, 0.04)	0.364
Brief Pain inventory
Pain interference		0.19 (0.00, 0.38)	0.048
Pain severity		0.160 (−0.07, 0.40)	0.179
Number of days with pain in previous 3 months
45 to 59 days		0.06 (−0.92, 1.03)	0.912
	≥ 60 days	Reference
Palliative care		1.48 (0.47, 2.48)	0.004
Number of comorbidities		0.00 (−0.16, 0.17cap)	0.971

4. Discussion

To our knowledge, this study is the first to attempt to establish a survey-based measure of access to opioids and to assess, preliminarily, the impact of barriers to opioid access among a self-selected respondent group of adults with chronic pain in the United States. Although opioids are rarely a first-line treatment for chronic pain, a need to access them exists, and is anticipated by the CDC's 2022 Clinical Practice Guideline for Prescribing Opioids for Pain.⁶ Cautionary statements about stoppage from the CDC and other authorities44, 45., 46. underscore that interrupted access is not considered a safe or desirable situation for many patients.

The study population permitted the development and initial statistical validation of a 20-item 2-factor scale termed the BOAS. It showed adequate internal consistency reliability and one of two subscales was associated with individual mental health and frequency of use of medications, providing some evidence of validity.

This study's effort to correlate each of the BOAS subscales with physical and mental health function must be considered tentative, since respondents to this online network were, by definition, self-selected. Acknowledging this caveat, the findings may lend some support to a burden of decreasing opioid prescribing indiscriminately. That burden is suggested by relatively high levels of agreement with items that connote fears, concerns, or adverse experiences in care. If such data can be replicated in more generalizable samples (see Limitations, below), then they would bolster arguments for a more patient-centered approach to prescribing at a time where changes reflect continuing concern about physicians' contributions to the overdose crisis in the United States.⁴⁷ These findings are also consistent with Andersen's model of access to care, demonstrating that access to pain management medication follow the same conceptual framework as barriers to any other needed healthcare services. This conceptual understanding of barriers for access reinforces the need for delineating patients who have untreated pain from the epidemic of substance use crisis and the overdose crisis currently plaguing the United States.

This study's findings also support previously-identified concerns regarding quality of life and safety of care among this patient population. In this study, patients with chronic pain demonstrated low quality of life scores, for both mental and physical health – as is shown in previous research involving individuals with chronic pain.48, 49, 50, 51, 52 Respondent ratings from the brief pain inventory were also comparable with previous literature.^53,54 Only about 72 % of survey respondents reported using opioid medications in the previous 6 months, but a majority of recipients of prescribed opioids reported being on a stable dose. The frequent use of non-opioid psychotropic medications such as gabapentin, benzodiazepines, muscle relaxants, and others observed in this study is also well-known from previous studies.55, 56, 57, 58, 59 This finding highlights a continuing concern for potential interactions between opioids and these medications.60, 61, 62, 63, 64 Further concern was noted as most respondents reported not having a prescription for naloxone, a medication that can help reverse overdoses. The CDC and others recommend prescribing take-home naloxone for individuals considered at high risk for overdose including those using opioids on a regular basis.65, 66, 67 Even though many states have issued a standing order for naloxone prescriptions and implemented other harm-reduction policies, these findings indicate that fills for take-home naloxone continue to be low among individuals with chronic pain.^68,69

The BOAS is self-reported, includes only 20 items, and captures barriers to opioid use, both due to inadequate prescribing and due to individual beliefs about opioid medications. Items such as ‘I have difficulty finding a pharmacy that can fill my opioid prescription’ and ‘My physician has denied me opioid medications because of prescribing regulations’ in the Access to Care subscale capture challenges faced by respondents in attempting to get treatment for their pain. This subscale was found to be significantly associated with a decrease in mental health scores, even after accounting for the severity of pain and other characteristics. Individual beliefs about side effects and consequences of opioid use, captured in the Patient Concerns subscale – using items such as ‘I do not take my opioid medication when I need it because I do not like the side effects of the medications’ and ‘I delay taking my opioid medications because I worry about their impact on some areas of my ability to function’ – were associated with frequency of opioid use, but not quality of life scores. These findings demonstrate that the combination of these two subscales captures meaningful constructs that can influence quality of life and medication use behavior among individuals with chronic pain. However, neither subscale was associated with physical health scores, for which more than one explanation might be offered. For example, it is plausible that our study respondents were more likely to be persons who overcame “difficulty” in obtaining prescribed opioids and were therefore protected from physical effects of such difficulty. It is also possible that the barriers captured in this scale are associated with reduction in opioid use but not actually leading to worsening of physical health.

Future research should aim to administer these scales in clinical populations to establish factor structure and construct validity. Nevertheless, this scale can be a valuable instrument to capture trends in barriers to opioid access, as prescribers, payers, and policy makers attempt to improve the standard of care for patients with chronic pain.

Data from several studies show that the rate of prescribing of opioid medications has decreased since 2012.^70,71 This decrease has been rather indiscriminate,⁷⁰ much of it occurring in the wake of the CDC's 2016 guideline,^72,73 through stricter restrictions.^17,74,75 The misapplication of these guidelines was acknowledged by the CDC, which published a revised guideline, in 2022.76, 77, 78, 79. The CDC was, in our view, correct to offer a course correction, given a steep rise in discontinuation of opioid therapy^80,81 and refusal of physicians to receive patients on prescribed opioids.^23,82, 83, 84, 85. The situation affecting patients with pain who might need opioids, or whose care has been interrupted, is unlikely to resolve itself in any short-term time frame. For that reason, we believe the BOAS represents a tool to assess some of the impediments to opioid medication access, by offering a direct window to the experience of patients.

4.1. Limitations

The findings of this paper have to be interpreted in the context of certain limitations. First, this study used a panel of chronic pain patients to meet the study objectives. The composition of this panel is likely to include patients with more severe chronic pain and those that are more willing to be actively managing their condition. Therefore, the study sample may not be representative of the chronic pain population in the United States and generalizability of study findings should be made with caution. Future studies should test the BOAS across broader samples with greater representativeness. Further, several questions in the survey – such as frequency of use of opioids and marijuana – could be considered sensitive, and were therefore susceptible to biases of self-report such as social desirability bias. Questions about use of other illicit substances were not considered in this study, but may warrant future examination. The study also included a relatively modest sample size and it is possible that larger sample sizes would have better powered some of the study analyses. Additionally, the scale development process can only be considered complete when the factor structure identified from the EFA can be confirmed in other independent samples. Despite these limitations, because this is a preliminary study aimed at developing a new instrument, we believe these findings are a valuable contribution to literature. However, future studies should be conducted in larger real-world clinical populations to replicate these findings and confirm the psychometric properties of the newly developed scale. Further, while the BOAS developed in this study showed potential for many future applications, further validation studies of the scale using confirmatory factor analysis in clinical populations are needed.

5. Conclusions

This study developed a self-reported 20-item scale to quantitatively assess the burden of lack of access to opioid medications among individuals with chronic pain in the United States. Findings indicate that barriers to access can lead to measurable impact on medication use patterns and mental health. This scale offered here can serve as a tool for capturing quality and access to pain treatment as well as measuring the impact of policy changes on the quality of treatment provided to patients with chronic pain.

Funding source

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

CRediT authorship contribution statement

Sujith Ramachandran: Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Software, Writing – original draft, Supervision. Stefan Kertesz: Conceptualization, Investigation, Writing – review & editing. Emily Gravlee: Data curation, Writing – review & editing. Prachi Prajapati: Data curation, Writing – review & editing. John P. Bentley: Formal analysis, Supervision, Writing – review & editing. Yi Yang: Investigation, Project administration, Supervision, Writing – review & editing.

Declaration of competing interest

Dr. Kertesz reports holding stock in Zimmer Biomet and Thermo Fisher and receiving royalties from UpToDate (Wolters-Kluwer).