Distribution of Genomic Testing Resources by Oncology Practice and Rurality: A Nationally Representative Study

Abstract

PURPOSE

Oncologists are increasingly using molecular profiling to inform personalized patient treatment decisions. Despite its promising utility, the integration of genomic testing into diverse clinical health care settings across geographic settings has been understudied.

METHODS

We used data from the National Survey of Precision Medicine in Cancer Treatment, a nationally representative sample of practicing US oncologists, to assess the availability of six genomic testing resources, including on-site pathology, contracts with outside laboratories, on-site genetic counselors, internal policies or protocols for using genomic and biomarker testing, electronic medical record alerts, and genomic or molecular tumor boards. We used multivariate logistic regression models to examine differences in the availability of each genomic testing resource by practice type and rurality while adjusting for payer mix and patient volume.

RESULTS

A larger proportion of multispecialty group and academic practices had genomic testing resources available compared with solo and nonacademic practices. Electronic medical record alerts were the least available resource, whereas contracts with outside laboratories were the most available resource. Compared with urban practices, there were significantly fewer practices located in rural areas that had on-site pathology, on-site genetic counselors, protocols for genomic tests, and molecular tumor boards.

CONCLUSION

Genomic testing resources varied by practice type and geography among a nationally representative sample of practicing oncologists. This variation has important implications for the development of interventions and policies to support the more equitable delivery of precision oncology to patients with cancer.

INTRODUCTION

Oncologists are increasingly using next-generation sequencing (NGS) and gene-expression profiling to inform personalized treatment decisions.¹ Often referred to as precision oncology, these genomic tests include single-gene and multigene tumor panels that require access to advanced pathology, laboratory, personnel, and organizational resources.^2,3 For example, precision oncology approaches can use NGS on molecularly heterogeneous tumor specimens to inform the use of molecularly targeted therapy, assist in identifying patients who may benefit from less invasive treatments, and reduce overtreatment of others.^4,5 However, despite the promising utility of precision oncology, the integration of genomic testing into health care settings is challenging.^3,6

CONTEXT

• Key Objective

• Next-generation sequencing is increasingly being used to inform personalized cancer treatment decisions; however, its availability across diverse health care settings and geographic settings is understudied. Does the availability of genomic testing resources—on-site pathology, contracts with outside laboratories, on-site genetic counselors, internal genomic testing policies and protocols, electronic medical record alerts, and genomic or molecular tumor boards—vary by practice type and rural geographic location?

• Knowledge Generated

• Using a nationally representative sample of oncologists, electronic medical record alerts were the least available resource (16.8%) and contracts with outside laboratories (85.6%) were the most available resource. Overall, a smaller proportion of solo and nonacademic practices and practices located in rural settings had these genomic testing resources compared with multispecialty group and academic practices and practices in urban settings.

• Relevance

• These findings highlight opportunities to advance the equitable delivery of personalized medicine for patients with cancer by understanding and addressing the health care delivery context.

Patients with cancer are diagnosed and treated in various health care settings that are diverse in organizational structure.^7-9 Growing evidence suggests that the delivery of precision oncology varies by practice structure and other characteristics. For example, oncologists practicing in nonprofit integrated health systems used genomic testing for colorectal and lung cancer at lower rates than those in hospital-based or single-specialty practices.¹⁰ Providers that treat more than 50 unique patients per month and have access to a molecular tumor board were more likely to report using NGS tests than providers who did not.² Additionally, community settings may be more likely to experience challenges in delivering guideline-recommended genomic testing, including long turnaround times to receive genomic test results and delayed coordination in the handling of tissue samples.¹¹ These findings may be explained by the clinical, financial, and administrative challenges of implementing genomic services into practice.^12,13 Understanding the availability of resources at the organizational level is a first step to identifying inequities in the utilization of precision oncology.

Rural oncology practices may also face particular challenges in the delivery of precision oncology.^14,15 Additionally, rural hospitals face closures, aging infrastructures, payment challenges, workforce shortages, quality challenges, and limited access to capital.¹⁶ Patients with cancer residing in rural areas are also more likely to have limited access to specialty providers and targeted cancer treatments that may be informed by genomic testing.¹⁷ Although few studies have documented challenges to precision oncology in rural settings, it is unknown whether genomic testing resource availability varies across geographical areas.

As the field of precision oncology moves forward, understanding how genomic testing resources and infrastructure vary across diverse clinical contexts and settings (eg, rural and urban and academic and nonacademic) will be informative for building capacity for delivering equitable and evidence-based cancer care.¹⁸ To begin to address the limited knowledge about the organizational contexts that influence the delivery of precision medicine, this study examined the availability of genomic testing resources, including on-site pathology, protocols for genomic test use, electronic medical record (EMR) alerts for genomic tests, and molecular tumor boards, across diverse oncology practice settings using a nationally representative sample of oncologists.

METHODS

Data Source and Study Population

We used data from the 2017 National Survey of Precision Medicine in Cancer Treatment, a nationally representative survey of oncologists, hematologists-oncologists, and hematologists. The survey was sponsored by the National Cancer Institute, the National Human Genomic Research Institute, and the American Cancer Society. Survey development, design, weighting, and methods were described in previous work,² and all survey questions were self-reported by oncologists.

Outcome Measure—Availability of Genomic Testing Resources

Availability of genomic testing resources was measured using six questions asking about the availability of various genomic testing resources in an oncologist's primary practice: (1) on-site pathology, (2) contracts with outside laboratories to perform tests not available on-site, (3) on-site genetic counselors, (4) internal policies or protocols for use of genomic and biomarker testing, (5) an EMR that alerts providers when a genomic test is recommended for a particular patient or before ordering a particular drug, and (6) genomic or molecular tumor boards. Responses for each question included yes, no, or don't know. We categorized responses as a binary outcome of yes and no (including don't know and missing responses).

Main Predictors—Practice Type and Rurality

Practice type was measured with two survey items. The first question asked oncologists to indicate if their main practice was a solo practice, single-specialty group, multispecialty group, or other type of practice. The second question asked whether their main practice was academically affiliated (yes or no). We then categorized the two responses into six mutually exclusive practice types: (1) solo office academic, (2) single-specialty group academic, (3) multispecialty group academic, (4) solo office nonacademic, (5) single-specialty group nonacademic, and (6) multispecialty group nonacademic. Oncologists whose main practice was other academic or other nonacademic were excluded from the analyses since we could not ascertain the exact type of practice setting. There is no framework for categorizing health systems and practice types.¹⁹ However, we operationalized practice type to capture the potential influence that academic affiliation and practice complexity may have on the availability of cancer resources on the basis of prior literature and definitions from the Compendium of U.S. Health Systems (2016).²⁰ These definitions are presented in Table 1.

TABLE 1.

Oncologists' Practice Types and Operationalized Definitions

Rurality was ascertained using each respondent's practice zip code linked to the 2013 Rural-Urban Continuum Codes to identify RUCC (1-9). Each practice address, including zip code, was obtained from the American Medical Association file and verified by phone. RUCC were categorized as urban (1-3) and rural (4-9).²¹

Covariates—Patient Volume and Payer Mix

Patient volume was identified from one item asking oncologists to indicate how many unique patients they see for evaluation or treatment each month. We dichotomized patient volume as < 100 and ≥ 100 patients consistent with the literature.²² Payer mix was identified from one item asking respondents to indicate what proportions of their patients were insured by Medicaid, or self-pay or uninsured. We dichotomized payer mix into < 10% and ≥ 10% Medicaid, or self-pay or uninsured.²²

Statistical Analysis

Descriptive statistics are presented to describe the sample and weighted percent of practicing oncologists. We examined the weighted percent of practicing oncologists who reported genomic resources in their practice by their primary practice type (Fig 1). We also examined differences in the proportion of practices with genomic testing resources by geographic area on the basis of RUCC (Fig 2). Chi-square tests were used to determine statistical significance for both figures. Multivariate logistic regression models were conducted to examine differences in each genomic testing resource by practice type and rurality while adjusting for known practice-level confounders (ie, patient volume and payer mix).^23-25 Adjusted odds ratios (aORs) and 95% CIs for the survey weighted population estimates are presented, and statistical significance for all analyses was determined at the P < .05 level (two-sided). Analyses were weighted to account for the probability sampling approach and survey nonresponse. All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).

FIG 1.

Percentage of practices with genomic testing resources by oncology practice type. On the basis of chi-square test: **P < .01, ***P < .001. The y-axis represents the weighted percent of oncologists with availability of each resource at their primary practice. EMR, electronic medical record.

FIG 2.

Percentage of practices with genomic testing resources by rurality. On the basis of chi-square test: *P < .05, ***P < .001. The y-axis represents the weighted percent of oncologists with availability of each resource at their primary practice. EMR, electronic medical record.

RESULTS

Characteristics of the 1,281 oncologists who completed the survey are summarized in Table 2. The mean age of responding oncologists was 50.2 years, and most were male (66.6%) and non-Hispanic White (62.1%). About two thirds (63.1%) of the oncologists' primary practices were academically affiliated. Most respondents practiced in single-specialty (41.8%) and multispecialty (44.8%) group practices. Geographically, almost all (95.7%) practiced in urban settings.

TABLE 2.

Characteristics of Oncologists and Their Primary Practice Setting in the United States (N = 1,281)^a

Genomic testing resources differed by practice setting. In general, a higher percentage of oncologists in multispecialty group practices and academically affiliated practices reported having genomic testing resources available than oncologists in solo, single-specialty, or nonacademic types of practices (Fig 1). Across all practice types, EMR alerts were the least available type of genomic testing resource and contracts with outside laboratories were the most available resource. Compared with practices located in urban areas, there were significantly fewer practices located in rural areas that had on-site pathology, on-site genetic counselors, protocols for genomic tests, and molecular tumor boards (Fig 2). There were no significant differences between urban and rural settings in the availability of contracts with outside laboratories and EMR alerts.

Table 3 presents the multivariate regression models of the six genomic testing resources by primary practice type and rurality adjusting for patient volume and payer mix.

TABLE 3.

Multivariate Regression Models of Genomic Testing Resources by Primary Practice Type and Rurality

On-Site Pathology

A total of 69.9% of oncologists reported that their primary practice had on-site pathology. In the adjusted model, multispecialty group academic practices were significantly more likely to have on-site pathology compared with other practice types. Onsite pathology did not differ by rurality (aOR, 1.14; 95% CI, 0.56 to 2.32).

Contracts With Outside Laboratories

A total of 85.6% of oncologists reported that their primary practice had contracts with outside laboratories. Single-specialty group nonacademic practices (aOR, 0.58; 95% CI, 0.40 to 0.90) and solo nonacademic practices (aOR, 0.29; 95% CI, 0.13 to 0.65) were significantly less likely to have contracts with outside laboratories compared with multispecialty group academic practices. There were no significant differences in contracts with outside laboratories by rurality (aOR, 2.10; 95% CI, 0.83 to 5.31).

On-Site Genetic Counselors

A total of 67.3% of oncologists reported that their primary practice had on-site genetic counselors. Multispecialty academic groups were significantly more likely to have on-site genetic counselors compared with the other five practice types. The odds of having on-site genetic counselors were 70% lower among rural practices compared with urban practices (aOR, 0.30; 95% CI, 0.14 to 0.63), which was statistically significant.

Internal Protocols for Genomic Tests

A total of 48.1% of oncologists reported that their primary practice had internal protocols for genomic tests. Multispecialty group academic practices were significantly more likely to have internal protocols for genomic tests compared with all other practices, including multispecialty group nonacademic practices (aOR, 0.24; 95% CI, 0.15 to 0.36). There were no significant differences in internal protocols for genomic tests by rurality (aOR, 0.54; 95% CI, 0.27 to 1.07).

EMR Alerts for Genomic Tests

A total of 16.8% of oncologists reported that their primary practice had EMR alerts for genomic tests. The odds of having EMR alerts in single-specialty group academic practice were significantly lower compared with multispecialty group academic practices (aOR, 0.52; 95% CI, 0.32 to 0.83). There were no significant differences in EMR alerts for genomic testing by other practice types and rurality (aOR, 0.46; 95% CI, 0.16 to 1.31).

Molecular Tumor Board

A total of 36.3% of oncologists reported that their primary practice had a molecular tumor board. Multispecialty group academic practices were significantly more likely to have molecular tumor boards compared with all other practice types, including multispecialty group nonacademic practices (aOR, 0.14; 95% CI, 0.08 to 0.24). Rural practices had 80% lower odds of having molecular tumor boards compared with practices in urban settings (aOR, 0.20; 95% CI, 0.06 to 0.64), which was statistically significant.

DISCUSSION

Oncologists are increasingly using molecular profiling to inform cancer treatment decisions. Our study suggests that there are variations in the availability of genomic testing resources by the oncology practice type and geographic setting. Across practice types, contracts with outside laboratories and on-site pathology were the most commonly available genomic testing resources. Yet on-site pathology was notably less available in rural practice settings. These resources are often used to provide standard of care in most health care settings, with the latter being a cost-effective approach for providing immediate and accurate identification of tumor staging that can be used for the development of cancer treatment plans.^26,27 Overall, our findings highlight an opportunity to advance the delivery of personalized medicine for patients with cancer by understanding the health care setting context of genomic testing resource availability.

Conversely, protocols for genomic testing, molecular tumor boards, and EMR alerts were the least available genomic resources across all practice types. Established policies and protocols for the use of genomic tests are a critical step for assisting providers with testing platform selection and decision making. However, research has shown that the establishment of such protocols and processes is challenging, requiring organizational buy-in, capital, resources, and involvement of multiple levels of clinical and nonclinical stakeholders.^3,28 Although multidisciplinary molecular tumor boards can optimize the translation of complex NGS information into appropriate targeted therapy options for patients with cancer,¹² their development and integration within health care systems require additional institutional resources and capacity, which may not be available in smaller practices.^29,30 In clinical settings, information technologies and software are often used to offer automated clinical decision support to providers, but in our study, EMR alerts were the least available resource across all practice types. This could reflect the possible lack of interoperability between EMR software used across practice settings.³¹ As more clinical practices move toward interoperable and standardized electronic records, EMRs offer a modifiable point of intervention to efficiently integrate genomic testing processes (eg, provider reminders to use genomic testing and standardized genomic testing information) and facilitate shared decision making.^32,33

Our study found that the availability of genomic testing resources varied by practice type. A larger proportion of practices that were a specialty group and/or academically affiliated had genomic testing resources compared with solo oncology practices and nonacademic practices. A smaller proportion of solo practices and nonacademic practices had any of the six genomic testing resources compared with specialty group and academic practices. For example, solo nonacademic practices had 89% lower odds of having protocols for genomic tests compared with multispecialty group academic practices. These findings may suggest that solo, small practices face limited infrastructure, capital, staff, and other resources, whereas specialized clinical expertise and advanced technology may be concentrated in academic medical centers.³⁴ To address this gap, academic medical centers and larger health systems have been actively involved in targeted efforts to bring precision medicine to diverse health care settings.^35,36 For example, collaboration through community-academic partnerships and the development of decentralized, online resources have provided infrastructure for many patients in nonacademic practice settings to benefit from precision oncology advances.^35-37

Genomic testing resources also varied between urban and rural practices, such that a smaller proportion of rural practices had on-site genetic counselors, internal protocols for genomic tests, on-site pathology, and molecular tumor boards compared with urban practices. These findings are consistent with previous studies indicating limited availability of cancer care services in rural settings.¹⁷ Further exacerbating limited access to genomic testing resources in rural areas are the financial pressures of the competitive market and lower patient volumes that have led to the closure of many small and/or rural hospitals.³⁸ Interventions that leverage policy changes that expand access to care may help address this critical gap in access to specialized care for rural patients with cancer. For example, recent changes to Medicare increase reimbursement for telehealth services, which expands access to much needed specialty care in rural communities.³⁹ Telehealth technologies, including virtual molecular tumor boards and appointments with genetic counselors, can be leveraged to transcend the travel barriers that rural patients with cancer face in accessing genomic testing resources.

This study is not without limitations. The cooperation rate (38%) for the National Survey of Precision Medicine in Cancer Treatment is lower than that of previous physician surveys on the topic of precision medicine.^40,41 Additionally, our findings reflect the availability of genomic testing resources in 2017 but the state of precision medicine is rapidly evolving. For example, in 2018, the Centers for Medicare & Medicaid Services announced the coverage of Food and Drug Administration–approved NGS tests for patients with advanced-stage cancer.⁴² These broadscale policy decisions are ongoing and affect the cost, value, and availability of precision medicine resources.^43,44 Nonetheless, this study uses the most currently available data. Additionally, for the outcome measure, availability of genomic testing resources, don't know responses were included in the same category as no responses because they could not confirm the availability of a resource. This may cause the no category to be artificially inflated. The survey did not collect patient-level data (eg, clinical outcomes and cancer types for which oncologists used genomic testing resources), which is an area for future investigations. Regarding the statistical analysis, the number of oncologists practicing in rural areas was low, which can limit conclusions that can be drawn from this analysis. Future studies should oversample oncologists in rural areas and other underserved areas to better understand care delivery in underserved settings.

Despite these limitations, the survey used in this study is a unique resource for understanding access to precision oncology, especially for underserved populations. To our knowledge, this is the first study to examine the availability of genomic testing resources by practice type and geography among a nationally representative sample of oncologists. Our findings suggest that the availability of genomic testing resources vary by organizational structure and geographic location of where oncologists practice. This variation may affect the equitable access to and utilization of precision medicine for patients with cancer. These findings warrant further investigation to delineate the extent to which genomic testing utilization and patient outcomes are influenced by the availability of genomic resources at the health system level. Additionally, exploring ways to facilitate effective integration between smaller oncology practices and academic or integrated health systems will increase the accessibility of targeted cancer treatments to patients in diverse populations and ensure more equitable distribution of precision oncology resources.

DATA SHARING STATEMENT

The data underlying this article cannot be shared publicly due to the privacy of individuals who participated in the study.

AUTHOR CONTRIBUTIONS

Conception and design: Brittany Gardner, Michelle Doose, Janeth I. Sanchez, Janet S. de Moor

Provision of study materials or patients: Andrew N. Freedman, Janet S. de Moor

Collection and assembly of data: Andrew N. Freedman, Janet S. de Moor

Data analysis and interpretation: All authors

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

The following represents disclosure information provided by the 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/po/author-center.

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