Racial, Ethnic, and Socioeconomic Discrepancies in Opioid Prescriptions Among Older Patients With Cancer

PURPOSE:

Minority race and lower socioeconomic status are associated with lower rates of opioid prescription and undertreatment of pain in multiple noncancer healthcare settings. It is not known whether these differences in opioid prescribing exist among patients undergoing cancer treatment.

METHODS AND MATERIALS:

This observational cohort study involved 33,872 opioid-naive patients of age > 65 years undergoing definitive cancer treatment. We compared rates of new opioid prescriptions by race or ethnicity and socioeconomic status controlling for differences in baseline patient, cancer, and treatment factors. To evaluate downstream impacts of opioid prescribing and pain management, we also compared rates of persistent opioid use and pain-related emergency department (ED) visits.

RESULTS:

Compared with non-Hispanic White patients, the covariate-adjusted odds of receiving an opioid prescription were 24.9% (95% CI, 16.0 to 33.9, P < .001) lower for non-Hispanic Blacks, 115.0% (84.7 to 150.3, P < .001) higher for Asian–Pacific Islanders, and not statistically different for Hispanics (−1.0 to 14.0, P = .06). There was no significant association between race or ethnicity and persistent opioid use or pain-related ED visits. Patients living in a high-poverty area had higher odds (53.9% [25.4 to 88.8, P < .001]) of developing persistent use and having a pain-related ED visit (39.4% [16.4 to 66.9, P < .001]).

CONCLUSION:

For older patients with cancer, rates of opioid prescriptions and pain-related outcomes significantly differed by race and area-level poverty. Non-Hispanic Black patients were associated with a significantly decreased likelihood of receiving an opioid prescription. Patients from high-poverty areas were more likely to develop persistent opioid use and have a pain-related ED visit.

INTRODUCTION

Opioid prescribing patterns for patients with cancer are under increased scrutiny in the setting of the ongoing opioid crisis.^1,2 As in the general population, rates of opioid-related deaths and emergency department (ED) visits have increased among patients with cancer over the past decade.^3,4 There is also concern, however, that in response to the opioid epidemic, restrictive policies or fears over addiction could lead to systematic undertreatment of cancer pain.⁵ Prior studies suggest that up to 50% of patients with cancer have inadequate pain management and that minority patients are especially at risk for undertreatment.^1,6,7

Disparities in opioid prescribing exist in the noncancer acute care setting, with non-Hispanic White (NHW) patients being more likely to receive opioid analgesics compared with non-Hispanic Black (NHB), Hispanic, or Asian–-Pacific Islander (API) patients.^7-9 Patients from more affluent and educated neighborhoods are also more likely to receive an opioid prescription when presenting to the ED.¹⁰ It is not clear whether discrepancies in opioid prescribing exist among patients with cancer, who often undergo more protracted treatment regimens with multidisciplinary care. It is also not known if discrepancies in initial opioid prescribing by race, ethnicity, or socioeconomic status correlate with persistent opioid use or inadequate pain management among patients with cancer.

The purpose of this study is to determine if race and poverty are associated with the likelihood of receiving opioid analgesics during definitive cancer treatment. Understanding potential disparities in opioid prescriptions can highlight a need for additional research or a change in practice patterns to ensure equitable care. We evaluated rates of new opioid prescriptions in patients with breast, colon, lung, or prostate cancer among Medicare beneficiaries. Opioid prescription rates during cancer treatment were compared by race, ethnicity, and neighborhood poverty levels. To determine if initial discrepancies in prescribing affect long-term opioid behavior, we also compared rates of persistent opioid use one to 2 years after treatment. Rates of pain-related ED visits were also evaluated as a potential surrogate for inadequate pain management.

METHODS AND MATERIALS

Data Source and Patient Selection

This retrospective cohort study evaluated patients in the SEER registry linked to Medicare claims data diagnosed with cancer from 2008 to 2013. SEER registry linked to Medicare claims data was chosen as a nationwide population-based data set with prescription data coupled with a cancer registry and demographic data. We included patients with one of the four most common noncutaneous solid malignancies (breast, colorectal, lung, and prostate), of age > 65 years, and with continuous Medicare Part A and B coverage, without enrollment in Medicare Part C (managed care) throughout the study period. Patients listed as other were not included because of low sample sizes. We included only patients with Medicare Part D coverage to capture opioid prescription data. We restricted the analysis to patients treated with definitive local therapy including surgery, radiation therapy, or both. Metastatic and palliative intent patients were excluded to analyze longer-term opioid prescribing patterns. This analysis only included opioid-naive patients who were defined as not having received an opioid prescription from one to twelve months prior to the cancer diagnosis.^11-13 Patients were also excluded for metastatic disease, death within the first 2 years since starting treatment, or missing follow-up data. The final study cohort included 33,057 patients that met eligibility criteria. Appendix Figure A1 (online only) demonstrates the patient selection process.

Covariates, Exposures, and Outcomes

Baseline patient and cancer-related factors including age, sex, race, ethnicity, marital status, stage, and cancer type were extracted from the SEER database. Patients were categorized as NHW, API, NHB, or Hispanic. Patients were categorized as Hispanic if they were of Hispanic ethnicity irrespective of their race. Poverty rate was estimated using median household income data at the census-tract level and represents the percent of the local population living below the poverty level.¹⁴ Rural zip codes were defined as those with a population center < 20,000.¹⁵ The non–age-adjusted Charlson Comorbidity Index was calculated from Medicare claims data using the International Classification of Diseases, Ninth Edition diagnostic codes, as previously described.¹⁶ Similarly, claims and diagnostic codes were used to identify prior diagnoses of depression; alcohol abuse; and high-risk psychiatric conditions including schizophrenia, attention deficit disorder, and obsessive compulsive disease.^17,18 The primary end point of a new opioid prescription during treatment was defined as receiving any opioid analgesic from one month prior to 3 months after the date of cancer diagnosis.^11,12 Persistent opioid use was identified as having filled ≥ 120 days' supply or 10 or more opioid prescriptions in the window between 1 and 2 years after the start of treatment.^11-13 Pain-related ED visits within 2 years of starting treatment were identified using International Classification of Diseases, Ninth Edition claims data.

Statistical Methods

The median and interquartile ranges were calculated for continuous variables by opioid prescription status and compared using a Mann-Whitney U Test. The absolute number and percentage were calculated for categorical variables and cohorts were compared using a chi-squared test.

Rates of new opioid prescriptions and persistent opioid use were calculated by race or ethnicity and area-level poverty subgroups; 95% CIs were estimated using a Pearson-Clopper interval.¹⁹ In an exploratory analysis, we also tested for differences in outcomes by poverty level when stratified by race. Multivariable logistic regression was used to estimate the odds of a new opioid prescription or persistent use when accounting for baseline patient, cancer-related, and treatment factors. Model covariates were considered significant with a two-sided P value < .05. Statistical analyses were performed using R version 3.5.3.²⁰

RESULTS

Patient Characteristics and Outcomes

This cohort of opioid-naive patients undergoing definitive cancer treatment included 28,779 (84.9%) NHWs, 815 (2.4%) APIs, 2,559 (7.6%) NHBs, and 1,719 (5.1%) Hispanics (Table 1). There were 5,562 (16.4%) patients identified as living in areas with poverty levels ≥ 20%. When separated by race and ethnicity, rates of living in a high-poverty area were 13.0%, 17.1%, 46.0%, and 28.6% for NHW, API, NHB, and Hispanic patients, respectively. Rates of new opioid prescriptions during cancer treatment differed significantly (P < .001) among NHW (24.3%), API (38.0%), NHB (18.2%), and Hispanic (28.1%) patients (Fig 1A). Patients living in high-poverty areas had increased rates of new opioid prescriptions (25.2% v 23.9%, P = .03) (Fig 1B). On average, patients prescribed an opioid during treatment were younger than those not given a prescription (73.8 v 74.6, P < .001). Prescription rates also varied significantly by sex, marital status, cancer type, tumor stage, nodal stage, and local and systemic treatment (Table 1).

TABLE 1.

Baseline Characteristics of Patients Undergoing Curative-Intent Cancer Treatment Stratified by Opioid Prescription During Treatment

FIG 1.

Opioid prescription rates during cancer treatment by (A) race and (B) area-level poverty. API, Asian–Pacific Islander; NHB, non-Hispanic Black; NHW, non-Hispanic White; Rx, prescription.

Rates of persistent opioid use did not differ significantly by race (P = .427) (Fig 2A). In contrast, patients living in communities with high poverty levels had a significantly increased rate of persistent opioid use (2.6% v 1.5%, P < .001) (Fig 2B). For the overall cohort, 2.4% of patients had a pain-related ED visit within 2 years since starting treatment. This rate did not vary significantly by race (P = .75) but did vary between patients living in high- and low-poverty areas, respectively (3.2% v 2.3%, P < .001) (Fig 3).

FIG 2.

Rates of persistent opioid use after cancer treatment by (A) race and (B) area-level poverty. API, Asian–Pacific Islander; NHB, non-Hispanic Black; NHW, non-Hispanic White; Rx, prescription.

FIG 3.

Rates of pain-related ED visits after cancer treatment by (A) race and (B) area-level poverty. API, Asian–Pacific Islander; ED, emergency department; NHB, non-Hispanic Black; NHW, non-Hispanic White.

When stratified by race and ethnicity, living in a high-poverty area was associated with significantly increased rates of new opioid prescriptions for NHB (21.1% v 15.9%, P = .001) and Hispanic (32.9% v 26.2%, P = .006) patients but not NHWs (25.6% v 24.2%, P = .075) or APIs (40.3 v 38.2, P = .71) (Appendix Fig A2, online only). Living in a high-poverty area was associated with increased rates of persistent use among NHW (2.6% v 1.5%, P < .001) patients (Appendix Fig A2). There was no significant association between poverty and increased persistent opioid use within NHB (2.2% v 1.7%, P = .48), Hispanic (2.9% v 1.6%, P = .11), or API (2.9% v 1.0%, P = .19) patients.

Multivariable Logistic Regression

Compared with NHWs, the covariate-adjusted odds ratio for receiving a new opioid prescription was 0.75 (95% CI, 0.67 to 0.84, P < .001) for NHBs, 2.15 (1.85 to 2.50, P < .001) for APIs, and 1.14 (0.99 to 1.30, P = .06) for Hispanics (Table 2). There was no statistically significant association between race or ethnicity and persistent opioid use or pain-related ED visits. Patients living in a rural community had increased odds of receiving a new prescription (odds ratio [95% CI], 1.54 [1.42 to 1.67], P < .001) and decreased odds of having a pain-related ED visit (0.72 [0.56 to 0.93], P = .01). Living in a higher-poverty area was not associated with an increased likelihood of receiving a prescription during treatment on multivariable analysis but was associated with a higher likelihood for persistent opioid use (1.54 [1.25 to 1.89], P < .001) and having a pain-related ED visit (1.39 [1.16 to 1.67], P < .001]). Additional patient, cancer, and treatment factors correlated with receiving an opioid prescription or becoming a persistent user on multivariable analysis (Table 2).

TABLE 2.

Multivariable Logistic Regression for New Opioid Prescription and Persistent Use

DISCUSSION

In this population-based cohort study, NHBs were less likely to receive an opioid prescription during definitive cancer treatment when compared with NHWs, with no difference in rates of persistent opioid use or pain-related ED visits. APIs were more likely to receive an opioid during treatment; however, this did not translate into increased rates of persistent use. Patients in rural communities had an increased likelihood of having an opioid prescribed during treatment and a decreased likelihood of presenting to the ED for pain. We found that patients living in areas of high poverty had an increased likelihood of becoming a persistent user and having a pain-related ED visit. Within racial or ethnic groups, high area-level poverty was associated with increased rates of new opioid use for NHB and Hispanic but not NHW or API patients. High area-level poverty was associated with increased persistent use for NHW but not API, NHB, or Hispanic patients. Subgroup analysis for persistent opioid use among the NHB and Hispanic patients had limited power because of smaller sample sizes and limited events.

Associations between race and both adequate pain management and opioid prescriptions have been previously described in multiple healthcare settings.^21-24 In an analysis of the National Hospital Ambulatory Medical Care Survey, NHW patients presenting to the ED for pain-related visits were more likely to receive an opioid compared with NHB, Hispanic, or API patients.⁸ NHB children were less likely to receive analgesia for a diagnosis of appendicitis and less likely to receive opioids for severe pain in a population-based study of ED visits.²⁵ To what degree race is associated with opioid analgesic use among patients with cancer is less clear, especially during the era of the ongoing opioid crisis. In a cross-sectional study of older patients with cancer living in nursing homes from the 1990s, minority race and advanced age were associated with undertreatment of daily pain.⁶ A more recent prospective cohort study found that among ambulatory patients with cancer, the odds of undertreatment of pain was twice as high for minority patients.⁷

The etiology of disparities in pain management by race and ethnicity is not fully understood. Hypothesized barriers to optimal pain management include factors related to healthcare systems, providers, implicit biases, communication, and individual patient beliefs and coping strategies.²² In a survey study of minority patients with cancer and their providers, inadequate pain assessment was identified as a primary barrier for optimal pain management.²⁶ Others have suggested that challenges in communication between NHW providers and minority patients may limit trust and result in discrepancies in the perceived levels of pain.^26-28

The association between socioeconomic status, pain control, and risk for adverse opioid outcomes in patients with cancer is less studied. We found that patients from lower-income areas had increased rates of prescriptions on univariable but not multivariable analysis. Rurality was a more dominant explanatory variable with increased rates of opioid prescriptions on both univariable and multivariable analyses (results not shown). Prior studies from the noncancer acute care setting have demonstrated decreased prescription rates among low-income patients.^10,29 An additional study of cancer survivors in Ontario, Canada, found a strong correlation between lower-income quantile and higher rates of opioid prescriptions among cancer survivors.³⁰

Although the association between poverty and new opioid prescriptions differs across healthcare settings, the association between poverty and adverse opioid outcomes is more consistent. Several studies have described higher rates of adverse opioid outcomes among patients from lower-income communities, which could correspond to the higher rates of persistent use observed in our study. Our data also suggest that low-income patients are at increased risk for inadequate pain control as evidenced by the increased rate in pain-related ED visits. In a nationwide cross-sectional study of Medicare enrollees, poverty was associated with opioid-related mortality at the county level.²⁹ Further, a population-based study of opioid-related events requiring emergency room evaluation revealed that opioid-related harms occurred nearly 2.4 times more frequently in the lowest-income quintile when compared with the highest quintile.³¹ High levels of opioid-related morbidity and mortality in this population could be attributed to lower healthcare utilization, especially regarding mental health and substance abuse treatment.³² There may also be decreased access to nonpharmacologic therapies in economically disadvantaged communities.³³

Optimal treatment of cancer pain with opioid analgesics requires an assessment of risks and benefits for an individual patient.¹ A primary limitation of this study is the inability to measure the level of pain and adequacy of pain control from the available data. Without this information, it is challenging to determine whether opioids are being prescribed appropriately. Depending on the clinical scenario, the risks associated with opioid use may outweigh benefits and lower prescription rates may be more in line with best practices.³⁴ Given the history of undertreatment of pain in minority patients with cancer, however, our findings raise concern that NHB patients may be at risk for inadequate pain management with opioids. The higher rates of persistent use among patients from high-poverty areas has not been well-studied in the cancer population. Our findings suggest that this population may be at higher risk for adverse outcomes and could benefit from increased risk mitigation techniques such as integrating nonopioid analgesics or referrals into pain management or counseling services.³⁵ A lack of resources for pain management is further evidenced by the increased utilization of the ED for pain among lower-income patients.

The results of this study should be interpreted in the context of its limitations. This analysis was limited to patients of age > 65 years continuously enrolled on Medicare Part D. It is not known if our findings apply to younger patients with cancer or those with varying levels of health insurance coverage. The population-based nature of the study also makes it difficult to determine an etiology for the observed discrepancies in opioid prescribing. Racial and ethnic groups beyond NHBs, Hispanics, and NHWs were underrepresented in this cohort and require future research. This data set was also limited by a lack of information on prescribers and institutions. Future research is warranted to determine if discrepancies in opioid use can be explained by institutional or prescribing practice patterns. Of note, previously observed racial disparities in cancer outcomes have been found to be reduced or absent in equal-access health systems such as the Veteran's Affairs medical system.^36,37 Opioid prescribing patterns are dynamic and have been continuing to evolve over the past two decades.³⁸ It is possible that the practice patterns, including the presence of disparities, have changed since the time frame of this cohort study from 2008 to 2013.

In conclusion, there are discrepancies in opioid prescriptions by race and area-level income for patients undergoing cancer treatment. In this cohort, NHB patients were less likely to be prescribed an opioid analgesic during cancer treatment without an increased risk for persistent use. Patients from high-poverty areas were more at increased risk for persistent opioid use and presenting to the ED for pain. Prescribing opioids for patients with cancer can involve complicated decision making requiring a personalized assessment of risks and benefits that is further confounded by the ongoing opioid epidemic.^39,40 To ensure equitable, compassionate care, additional research is needed to identify the etiology of and solutions to disparities in opioid prescribing and risk for adverse outcomes.

APPENDIX

FIG A1.

Flow diagram showing patient inclusion and exclusion in observational cohort. NHW, non-Hispanic White; NHB, non-Hispanic Black; NHA, non-Hispanic Asian.

FIG A2.

Rates of (A) new opioid prescription during treatment and (B) persistent opioid use after treatment stratified by race and area-level poverty. API, Asian–Pacific Islander; NHB, non-Hispanic Black; NHW, non-Hispanic White; Rx, prescription.

AUTHOR CONTRIBUTIONS

Conception and design: Lucas K. Vitzthum, Paul Riviere, Timothy Furnish, James D. Murphy

Collection and assembly of data: Lucas K. Vitzthum, Vinit Nalawade, Paul Riviere

Data analysis and interpretation: Lucas K. Vitzthum, Vinit Nalawade, Whitney Sumner, Tyler Nelson, Loren K. Mell, Timothy Furnish, Brent Rose, María Elena Martínez

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Racial, Ethnic, and Socioeconomic Discrepancies in Opioid Prescriptions Among Older Patients With Cancer

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/op/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).