Factors Associated with Black Cancer Patients' Ability to Obtain Their Opioid Prescriptions at the Pharmacy

Abstract

Background: Black cancer patients experience pain. Barriers to opioid medications for pain may include geographic factors. This study examines neighborhood factors associated with difficulties receiving prescription opioids from pharmacies for black cancer patients.

Design: A secondary data analysis of a study on opioid adherence was used to examine neighborhood-level and individual factors related to difficulties filling prescriptions for opioids.

Setting/Subjects: Patients being treated for cancer pain with opioids (n = 104) were recruited. All self-identified as black, were 21 years or older, had cancer diagnoses, and had been prescribed extended release opioids.

Measurements: A seven-item survey to identify problems filling opioids was completed by 98 participants along with a nine-item scale to assess perceived neighborhood characteristics. Scales of neighborhood amenities and neglect were created from the perceived neighborhood characteristics scale using principal components analysis. The 2009–2013 American Community Survey data were used to estimate the census tract percentage of non-Hispanic black residents, residents ≥25 years of age without a high school degree or equivalent, and households earning below the federal poverty level within the past 12 months.

Results: Nearly 51% reported problems getting their opioids filled: 28% had to wait days and 24% had to return to the pharmacy multiple times. The main theme identified in analysis of an open-ended question was pharmacies not stocking medication. Neighborhood locations that rated higher on the amenities scale were protective for pharmacies sufficiently stocking opioids.

Conclusions: Additional research on pharmacies sufficiently stocking opioid pain and neighborhood perceptions is warranted.

Background

Many cancer patients do not receive proper pain treatment,^1–6 and although black persons are more likely to visit health care providers for pain than non-Hispanic white people,⁷ they are less likely to receive adequate pain management.^8,9 Several factors, such as reduced access to pain specialists,¹⁰ receiving less aggressive pain treatments,^8,10 and pharmacy practices,^11,12 might contribute to this disparity. Both Green et al.¹¹ and Morrison et al.¹² found that pharmacies in predominately black neighborhoods were less likely than those in predominately white neighborhoods to sufficiently stock opioid pain medications.^11,12 Reasons pharmacies gave for not stocking opioid medication sufficiently included concerns about illicit drug use,^11,12 fear of robbery,¹² disposal concerns,¹² and little demand for opioid pain medications.^11,12 In this article, we examine patient and neighborhood factors related to problems filling opioid prescriptions among black cancer patients.

Design

This secondary data analysis comes from a parent study (NIH K01NR014673) that explored adherence to daily opioids in black patients being treated for cancer pain. Data collection occurred during 2014–2016. Institutional review board approval was obtained from the involved institutions.

Methods

Patients being treated for cancer pain with opioids (n = 104) were recruited from medical oncology, radiation oncology, and palliative care clinics of an urban safety net hospital and a cancer center (Atlanta, GA). All participants self-identified as black were 21 years or older, had cancer diagnoses, and had been prescribed an extended release opioid.

Measures

Pharmacy problems

As data collection began for the parent study, participants anecdotally reported difficulty getting opioid prescriptions filled at pharmacies, despite having valid prescriptions and cancer diagnoses. No research instrument to investigate pharmacy barriers was found; therefore, we developed a seven-item survey based on barriers identified in the literature and input from a palliative care clinician. Our primary outcome variable was any problem filling opioids, measured by participants responding affirmatively to at least one of three questions (Table 1). An open-ended survey question enabled participants to describe their difficulties attempting to fill their opioid prescription.

Table 1.

Demographics and Clinical Characteristics of Participants^a

Variable	Value
Gender, n (%)
Female	63 (61)
Male	41 (39)
Age at study enrollment, mean (SD)
Age	55 (10)
Education levels, n (%)
Less than high school	19 (18)
High school graduate	29 (28)
Some college/college graduate	46 (44)
Graduate school	10 (10)
Marital status, n (%)
Living with a partner/married	36 (35)
Single	28 (27)
Divorced/separated	28 (27)
Widowed	12 (12)
Insurance status, n (%)
Has insurance	94 (90)
No insurance	10 (10)
Total income (n = 83), mean (SD)
Total income	25,626 (33,031)
Site of study enrollment, n (%)
Safety net hospital	46 (44)
Cancer center	58 (56)
Cancer diagnosis (n = 82), n (%)
Myeloma	25 (24)
Breast	19 (17)
Lung	15 (13)
Colon, rectal, prostate, anal, and pancreatic	23 (22)
Other	22 (22)
Metastatic disease (n = 94), n (%)
Yes	54 (57)
Opioid type, n (%)
Extended release morphine	69 (66)
Extended release oxycodone	30 (29)
Methadone	5 (5)
Type of pharmacy used, n (%)
Hospital or clinic	62 (60)
Nonhospital pharmacy	42 (40)
Pharmacy close to home (n = 96), n (%)
Yes	52 (54)
Pharmacy problems filling opioids, n (%)
Yes to at least one of the following three questions	69 (66)
(1) Do you have to wait any number of days before your pharmacy has your med ready for you?	29 (28)
(2) Do you have to go back to the pharmacy before they have your med ready for you?	25 (24)
(3) Have you experienced any problems in getting your pain medicine filled?	53 (51)
Presence of neighborhood amenities, n (%)
Sidewalks or walking paths (n = 102)	75 (74)
Parks or playground	82 (79)
Recreation or community center or boys/girls club (n = 102)	65 (64)
Library or bookmobile (n = 102)	72 (71)
Neighborhood amenities component, mean (SD)	0 (1.30)
Presence of neighborhood neglect, n (%)
Litter or garbage on the street or sidewalk	24 (23)
Rundown housing or mess homes	22 (21)
Vandalism such as broken windows or graffiti (n = 102)	12 (12)
Abandoned homes or apartments properties	18 (17)
Homes with bars on the windows or doors (n = 99)	36 (36)
Neighborhood neglect component, mean (SD)	0 (1.74)

Census tract variablesb	Mean (SE)	Median	25th and 75th Percentiles
Percentage of non-Hispanic black residents	69.24 (26.40)	75.43	53.08–92.17
Percentage of residents ≥24 without a high school education	15.08 (8.39)	15.33	7.88–19.41
Percentage of households living below poverty level	21.60 (12.59)	20.18	11.60–29.58

^aN = 104, unless noted otherwise.

^bUntransformed variables reported, 103 participants reported living in 88 census tracts at study enrollment in/between year.

SD, standard deviation; SE, standard error.

Demographic and clinical information

A demographic form was used to collect age, gender, individual income, education level, and marital status for each patient. Clinical characteristics (i.e., cancer types, comorbidities, and type of opioid) were collected from a medical record review.

Neighborhood amenities and neglect

Patient-reported data on neighborhood physical characteristics were collected with a nine-item scale modified from the National Survey of Children's Health,¹³ which assessed neighborhood characteristics, such as the presence of parks or barred windows.^13,14 Response options were yes, no, or do not know. We divided the nine-item scale into two components, amenities and neglect, using principal components analysis (Appendix Table A1). The amenities component derived from questions regarding the presence of parks, recreation centers, libraries, and sidewalks within neighborhoods. The neglect component derived from questions regarding the presence of litter, rundown buildings, vandalism, abandoned buildings, and barred windows.

Census characteristics

For each census tract in which patients lived, we used 2009–2013 American Community Survey data to estimate the census tract percentage of non-Hispanic black residents, residents ≥25 years of age without a high school degree or equivalent, and households earning below the federal poverty level within the past 12 months.¹⁵ To assign census tract information to participants, we geocoded participants to census tracts using their address at study enrollment with ArcGIS (10.5.1).¹⁶ We ignored duplicity of participants within census tracts because most participants were the sole representatives of their census tract.

Data Collection and Analysis

All patient-report data were collected by trained research staff through face-to-face interviews at the participant's home or a private area of the clinic. Quantitative data were analyzed with standard descriptive and frequency analyses. Regression and correlation analyses were used to explore relationships between neighborhood characteristics and pharmacy problems. All continuous variables were median centered for analysis. We first conducted bivariate logistic regressions. Predictors that associated with experiencing pharmacy problems at a p-value ≥0.20 were included in a multivariable logistic regression. We used deviance divided by degrees of freedom to test overdispersion. We performed all analyses in SAS (9.4).¹⁷ We estimated models with maximum-likelihood estimation. Data from an open-ended survey question about pharmacy problems were analyzed with thematic analysis.¹⁸ Themes were identified independently by two authors (K.A.Y. and K.J.) and compared. Discrepancies were discussed until consensus was reached.

Results

One hundred and four participants completed the study (Table 1). Participants were largely low income (mean ± standard deviation = $25,626 ± $33,031) with a variety of cancers. Fifty-seven percent had metastatic disease. Patients had prescriptions for extended release morphine (N = 69; 66%), extended release oxycodone (N = 30; 29%), or methadone (N = 5; 5%); in addition, most also were prescribed as needed opioids for breakthrough pain. Nearly 66% (N = 69) experienced any difficulty getting their opioids filled: 28% (N = 29) had to wait days and 24% (N = 25) had to return to the pharmacy multiple times. Participants lived in census tracts where ∼22% of residents were at or below the poverty level and where ∼69% of residents were black.

Although several variables were significantly associated with problems filling opioids (p ≤ 0.20) in bivariate analyses (Table 2) in the multivariable model, only the neighborhood amenities component derived from the neighborhood perceptions scale remained significant (adjusted odds ratio 0.61).

Table 2.

Factors Associated with Black Cancer Patients' Problems Filling Opioid Prescriptions at Pharmacies: Bivariate and Multivariable Results of Logistic Regressions, n = 104

	Bivariates		Amenitiesamultivariable model
Variable	Unadjusted odds ratio	p	Adjusted odds ratio	p
Census tract variables, %
Non-Hispanic black residents	0.99	0.12	0.998	0.88
≥24 Years without high school education	0.97	0.17	0.96	0.34
Households below poverty level	0.97	0.04	1.01	0.71
Neighborhood amenities
Sidewalks	0.84	0.72
Park	0.37	0.09
Recreation center	0.59	0.24
Library	0.54	0.21
Neighborhood amenities component	0.63	0.03	0.61	0.03
Neighborhood neglect
Litter	1.31	0.60
Rundown	0.67	0.42
Vandalism	0.67	0.52
Abandoned properties	0.76	0.61
Barred windows	0.71	0.44
Neighborhood neglect component	0.91	0.47
Site of study enrollment (Ref. = safety net hospital)
Tertiary cancer center	5.71	0.0001	3.00	0.12
Type of pharmacy used (Ref. = hospital pharmacy)
Nonhospital pharmacy	3.28	0.01	1.56	0.50
Gender (Ref. = male)
Female	1.48	0.35
Age at study enrollment
Age	1.02	0.47
Insurance status (Ref. = has insurance)
No insurance	0.30	0.08	1.10	0.92
Total income	1.00	0.85

^aMultivariable model with amenities component portrayed; multivariable model results with park is available upon request.

Bold indicates p < 0.20.

Ref., referent.

The main theme identified in qualitative analysis of the open-ended question was pharmacies not stocking the medication. Participants reported going to six or more pharmacies to get their opioid filled and others reported needing to wait days before their opioid was available. These challenges were amplified by pharmacists refusing to tell patients over the phone whether the pharmacy stocked these medications. A secondary theme was the importance of relationships with pharmacists. One participant was initially told by a pharmacist tech that his pain medicine would take a week to fill, but then, a pharmacist who knew the patient saw him and filled it immediately. This relationship between the patient and the pharmacy staff is illustrated in this quote, “Some pharmacies give you third degree, act like I'm a criminal. I decided to just go to other pharmacy. … that pharmacy works for me because they know me.” One participant gave this advice, “Be persistent and establish a relationship” and noted that “sometimes pharmacists will persuade you to go someplace else. Some say no, they cannot fill it. Some say come back later.” Other problems reported included perceptions that pain medications were being abused. Overall, these findings demonstrate the presence of many barriers to filling opioid prescriptions.

Discussion

We found two-thirds of participants reported experiencing at least one pharmacy problem. In our multivariable model, the presence of neighborhood amenities was associated with fewer pharmacy problems, and the presence of neighborhood neglect indicators or other census tract neighborhood factors was not associated with pharmacy problems. Qualitative themes suggested pharmacy problems were related to pharmacies not stocking opioids and the quality of the relationship the patient had with the pharmacy staff. This concurs with recent research by Lal et al.,¹⁹ which found that pharmacies in communities were more likely than pharmacies in institutions to experience shortages of opioids and tell patients that they could not provide a refill. In this study, pharmacists reported that concerns about illicit use and diversion were big barriers to providing opioid medication.

Where a person lives and shops matters. Consistent with previous findings,^11,12 many respondents in our study reported that pharmacies did not carry opioids. Concerns about illicit drug use in predominately black neighborhoods influencing pharmacies' decisions not to stock opioid medications has been reported.^11,12 Our observations concur with this finding; some participants reported feeling criminalized when attempting to fill their prescription pain medications. It is also possible that some pharmacy staff may differentially discriminate against black patients if opioids are in stock. Future research should examine differences between the ways individuals of different races are treated at pharmacies when requesting opioids. Additional research examining pharmacists' perspectives on the situation, as well as insurance and legislative barriers that impact pharmacists, is necessary to better understand the problems reported by our participants. Such research should pay close attention to the relationship between pharmacists' perceptions of neighborhoods and their decisions to stock opioid medications.

This study has some limitations. For instance, our sample was a very specific group and we did not include a comparison group of nonblack patients or those filling a nonopioid prescription.

Clinical Implications

As our society deals with the crisis of prescription drug abuse, we must balance our actions to promote successful management of cancer pain. This study informs clinicians about barriers that cancer patients experience obtaining opioids. When prescribing opioids, clinicians might suggest patients attempt establishing a relationship with pharmacy staff.

Although evidence-based guidelines recommend opioids as appropriate for management of cancer pain,^20,21 receiving the opioid prescription is not the final hurdle patients in pain will most likely experience. The cancer care community along with pharmacy organizations can work to break down pharmacy barriers to provide our patients with timely pain relief. For example, educational materials may be developed for pharmacists about barriers minority patients experience filling medications and professional pharmacist conferences can host discussions about ways to overcome barriers.

Appendix Table A1.

Components Generated by Principal Components Analysis of Amenities and Neglect Variables

Component	Constituent variables	Correlations of constituent variables with component
Neglect
Eigenvalue: 3.02	Litter	0.37
Variance explained: 38%	Rundown	0.45
	Vandalism	0.46
	Abandoned	0.50
	Barred window	0.38
Amenities
Eigenvalue: 1.70	Park	0.50
Variance explained: 21%	Recreation center	0.61
	Library	0.54

Correlation matrix used.

Author Disclosure Statement

All the authors report no potential conflicts of interest.