Unraveling Racial Disparities in Supportive Care Medication Use among End-of-Life Pancreatic Cancer Patients: Focus on Pain Management and Psychiatric Therapies

Pancreatic cancer is a significant public health issue, as despite being only the 10th leading cause of cancer in the United States, it is the third leading cause of cancer death. Studies have noted a 50% to 90% higher incidence in Black patients as compared with other racial groups (1). Pancreatic cancer has a particularly poor prognosis with a 5-year survival rate of 12% (1). Pancreatic cancer accounts for approximately 50,000 deaths annually, with Black patients with pancreatic cancer having higher mortality rates (1, 2).

In addition to its well-documented lethality, pancreatic cancer is associated with significant neurologic and psychiatric symptoms, with severe pain, anxiety, depression, agitation, and delirium most often described, particularly in late-stage cancer (3, 4). These symptoms can be driven by the disease itself, or conversely, as a side effect of cancer-related treatment. Timely recognition and effective symptom management are a crucial component of cancer survivorship, and for optimizing health-related quality of life (HRQoL). Several studies have demonstrated that strategies promoting improved symptom management in patients with advanced cancer yield improvements in survival (5–7). In addition, HRQoL has been reported to have a high prognostic value, with lower HRQoL associated with worse response to cancer treatment (8). In pancreatic cancer, neurologic and psychiatric symptoms are associated with reduced HRQoL. Central to managing the neurologic and psychologic symptoms of pancreatic cancer are the effective use of supportive care medications, such as opioid and non-opioid analgesics, antidepressants, anxiolytics, and antipsychotics.

Pancreatic cancer-related pain occurs in up to 75% of patients at diagnosis, 90% of patients with late-stage disease, and overall is the most frequently reported adverse event in this patient population (9). Pancreatic cancer-related pain is multifactorial and includes visceral, somatic, and neuropathic origins. Furthermore, pancreatic cancer-related pain can be disease-induced (i.e., pancreatic duct obstruction, pancreatic neuropathy) or adverse side effects of treatment (i.e., chemotherapy, radiation, or surgery). Pancreatic cancer-related pain is primarily managed by supportive care medications, including opioid analgesics, non-opioid analgesics, and skeletal muscle relaxants. Among these, opioids are by far the most used treatment. Recent evidence suggests that opioid use in patients with pancreatic cancer is associated with increased survival (10). International clinical practice guidelines for managing cancer-related pain affirm the importance of opioids and other supportive care medications and recommend a stepwise management approach (11). Pancreatic cancer has also been associated with detrimental impacts on psychologic well-being. Pancreatic cancer-related depression and anxiety have been reported to occur in a substantial segment of the population, with the literature indicating approximately 15% of all patients with pancreatic cancer experience these symptoms (3). The reported anxiety and depression rates in patients with pancreatic cancer are higher than in other types of cancer (12, 13). Furthermore, undertreatment of cancer-related pain and other symptoms are common causes for treatment interruptions (14) and are associated with worse outcomes (15), thus underscoring the need for effective symptom management.

Multiple cancer studies exploring HRQoL have demonstrated worsened outcomes in racial and ethnic minorities, particularly at end-of-life (16). Reasons for these disparities are likely multifactorial, with differences in access to cancer treatment (17), less referrals to cancer specialists (18), and end-of-life hospice utilization (19) having been reported as possible drivers. Specific to end-of-life HRQoL, Black patients with cancer are more likely to have multiple emergency department visits and undergo intensive treatment in the last 6 months of life than White patients with cancer, indicating lower HRQoL at end-of-life (20). Recent clinical practice guidelines on palliative care in oncology acknowledge these observed disparities, but also highlight a paucity of health disparities research on palliative care, with an urgent need for more research in this area (21). Previous research has identified the presence of racial disparities in supportive care medication use in patients with multiple types of cancer (22, 23). Supportive care medications used to manage neurologic and psychiatric symptoms are cornerstones of care for patients receiving end-of-life care, and equitable access to this vital component of symptom management is desired.

However, little is known about the presence and magnitude of supportive care medication use disparities, and their contribution to existing racial and ethnic disparities in patients with pancreatic cancer, particularly those receiving end-of-life care. The objective of this study was to characterize supportive care medication use patterns among patients with pancreatic cancer during end-of-life care by race and ethnicity.

This was a retrospective cohort study using data from the Surveillance, Epidemiology, and End Results (SEER)-Medicare-linked database from 2005 to 2017. The SEER-Medicare linked database is a collection of cancer registries consisting of clinical and demographic information for patients with cancer and their respective Medicare claims for health services. Prescription medication data were collected via Medicare Part D claims. This study was conducted using publicly available data and approved as exempt by the UF Institutional Review Board (study #IRB201701945) and is compliant with SEER reporting guidelines.

Participants were identified by using ICD-9 and ICD-10 codes specific to pancreatic cancer. In the NCI SEER cancer registry, race and ethnicity categories were identified on the basis of medical records and administrative information and included non-Hispanic White, non-Hispanic Black/African American, non-Hispanic Asian/Pacific Islander, and non-Hispanic American Indian/Alaska Native, and Hispanic (of any race). Racial and ethnic minorities were classified as individuals who identified as Black/African American, Asian/Pacific Islander, American Indian/Alaska Native, and Hispanic (of any race). Consistent with previous methodology evaluating the use of supportive care medications in patients with end-of-life cancer, patients were excluded if they died more than 1 year after the initial cancer diagnosis (22). In addition, patients were excluded if they did not have Medicare coverage in the 3 months prior to death, or had missing race/ethnicity demographic data. To comply with the NCI SEER- Medicare data use agreement, data was coarsened to prevent cell values less than 11. When applicable, data suppression was completed to prevent deriving cell values less than 11 (24).

Given the unique burden of pain and psychiatric symptoms in patients with pancreatic cancer during end-of-life care, our study was focused on the analysis of supportive care medications used to alleviate these symptoms. The end-of-life time period was defined as the 12-month period before death. Supportive care medication use was defined as at least one Medicare claim for a medication of interest during the end-of-life time period. Only outpatient use of supportive care medications was considered in the analysis. Pain medications were identified as those in the following drug classes: opioid analgesics, non-opioid analgesics, and skeletal muscle relaxants. Supportive care medications used to manage psychiatric symptoms included those from the following drug classes: antidepressants, anxiolytics, and antipsychotic agents. The full medication list was adapted from previously reported literature (22). The complete medication list by drug class is provided in Supplementary Table S1.

Baseline categorical variables were compared across racial subgroups with the χ² test. Normally distributed and non-normally distributed continuous covariates were compared by race using ANOVA and the Kruskal–Wallis tests. Individual logistic models were created for each class of supportive care medications. We also compared the receipt of any supportive care medication in each respective treatment category (i.e., pain medication, psychotropic medication). Multivariable logistic regression models characterized the relationship between supportive care medication use and race/ethnicity. In our primary analysis model (Adjustment Model 1), results were adjusted for age, sex, marital status, ZIP code-level high school completion rate, ZIP code-level median household income, rurality of patient residence, year of pancreatic cancer diagnosis, Charlson comorbidity index at cancer diagnosis, cancer stage, and Medicare hospice care claims. In addition, we performed a secondary analysis (Adjustment Model 2), that did not adjust for socioeconomic status (i.e., education, income), which is consistent with the National Academies of Medicine (previously Institute of Medicine) definition of healthcare disparities (25, 26). All analyses were performed using SAS v9.4. A two-sided P < 0.05 was considered statistically significant.

The data analyzed in this study are available from the SEER-Medicare Linked Database, which is available from the NCI (https://healthcaredelivery.cancer.gov/seermedicare/).

Upon initial query, a total of 145,818 patients with pancreatic cancer were identified. After application of exclusion criteria, a total of 71,509 patients were excluded, leaving 74,309 patients included in the final study analysis. The most common reason for exclusion was death more than 1 year after initial pancreatic cancer diagnosis. Among included subjects, 55.8% were women (n = 41,494), 92.7% (n = 68,868) were over the age of 65, and nearly half had stage IV pancreatic cancer (46.6%, n = 34,611).

When considering the race and ethnicity of included subjects, the breakdown was as follows: Non-Hispanic White, 72.5% (n = 53893), Non-Hispanic Black/African American, 11.0% (n = 8185), Hispanic (any race), 9.6% (n = 7131), Non-Hispanic Asian/Pacific Islander, 6.6% (n = 4889), and Non-Hispanic American Indian/Alaska Native, 0.28% (n = 211). Comparisons of clinical and demographic characteristics of pancreatic cancer by race and ethnicity are provided in Table 1. Supplementary Table S2 includes a comparison of clinical and demographic characteristics by gender and race/ethnicity.

In our primary analysis model (Model 1), when considering the use of pain medications, Black [Adjusted OR (aOR), 0.84; 95% confidence interval (CI), 0.79–0.88] and Asian (aOR, 0.84; 95% CI, 0.79–0.90) patients had lower utilization of opioids compared with non-Hispanic White patients. Hispanic patients had higher opioid utilization compared with non-Hispanic Whites (aOR, 1.07; 95% CI, 1.01–1.14). Regarding non-opioid analgesics, Hispanic (aOR, 1.16; 95% CI, 1.08–1.25), and Asian (aOR, 1.37; 95% CI, 1.26–1.49) patients had increased utilization compared with non-Hispanic White patients. There were no differences in non-opioid analgesic use among Black patients. The use of musculoskeletal relaxants was decreased across all racial groups compared with non-Hispanic White patients with pancreatic cancer; Black: aOR, 0.90; 95% CI, 0.82–0.99; Hispanic: aOR, 0.82; 95% CI, 0.74–0.91; Asian: aOR, 0.59; 95% CI, 0.51–0.68. In our secondary analysis model (Model 2), opioid utilization remained lower in Black and Asian patients, while Hispanic patients had no difference in the use of opioids (aOR, 1.05; 95% CI, 0.99–1.12). Utilization of non-opioids and musculoskeletal relaxants were consistent with the results observed in Model 1.

When considering the use of psychiatric medications, in the primary analysis model, all race/ethnic groups had lower antidepressant (Black: aOR, 0.56; 95% CI, 0.53–0.59; Hispanic: aOR, 0.77; 95% CI, 0.73–0.812; Asian: aOR, 0.47; 95% CI, 0.44–0.51), and anxiolytic (Black: aOR, 0.47; 95% CI, 0.43–0.50; Hispanic: aOR, 0.66; 95% CI, 0.62–0.71; Asian: aOR, 0.52; 95% CI, 0.48–0.57) utilization as compared with non-Hispanic White patients. When considering antipsychotic use, Hispanic and Asian patients had lower use compared with non-Hispanic White patients (Hispanic: aOR, 0.90; 95% CI, 0.82–0.99; Asian: aOR, 0.84; 95% CI, 0.74–0.95). There were no differences in the use of antipsychotics among Black patients. In our secondary analysis model, the use of antidepressants and anxiolytics were consistent with the results from Model 1. When considering antipsychotics, the only difference in utilization was observed in Asian patients (aOR, 0.88; 95% CI, 0.78–0.99). Due to small sample size, non-Hispanic American Indian/Alaska Native patients were excluded from the multivariable logistic regression analysis. The prevalence of supportive care medication use by race and ethnicity are shown in Fig. 1. No differences or trends were observed on the basis of year of pancreatic cancer diagnosis.

Regarding the use of any pain medication, in our primary analysis (Model 1), Black (aOR, 0.84; 95% CI, 0.80–0.99), and Asian patients (aOR, 0.93; 95% CI, 0.87–0.99) were less likely to receive any pain medication, while Hispanic patients were more likely to receive any pain medication (aOR, 1.11; 95% CI, 1.04–1.17), as compared with non-Hispanic White patients. When considering the use of any psychotropic medication, in our primary analysis (Model 1), all groups had less use compared with non-Hispanic White patients (Black: aOR, 0.54; 95% CI, 0.52–0.57; Hispanic: aOR, 0.74; 95% CI, 0.70–0.78; Asian: aOR, 0.49; 95% CI, 0.46–0.52). Similar results were observed in the secondary analysis model (Model 2) for the receipt of any pain medication or psychotropic medication, respectively. Table 2 describes the unadjusted and adjusted utilization of supportive care medications, and receipt of any supportive care medication by treatment category by race and ethnicity.

Using data from a large, national sampling of patients with pancreatic cancer, we identified differences in end-of-life utilization of multiple supportive care medication classes among racial and ethnic minorities compared with non-Hispanic White patients. These results were consistent in two different adjustment models which included several factors that could influence medication use including age, gender, marital status, ZIP code level high school completion rate and median household income, patient residence type, year of pancreatic cancer diagnosis, Charlson comorbidity index, cancer stage, and receipt of hospice care. An interesting finding from our results was the substantial intergroup heterogeneity in supportive care medication utilization among different racial minorities. Specifically, in our primary analysis, when considering pain medications, Black patients had lower odds of using opioids (−16%), and skeletal muscle relaxants (−10%) compared with Non-Hispanic Whites. Conversely, Hispanic patients had lower odds of using skeletal muscle relaxants (−18%) but increased use of opioid (+7%) and non-opioid analgesics (+16%). Asian patients had lower odds for the use of opioids (−16%), and skeletal muscle relaxants (−41%), but increased use of non-opioid analgesics (+37%) compared with non-Hispanic Whites. When considering the receipt of any pain medication, we continued to observe diverging results across minority groups, with Black (−16%) and Asian (−7%) patients had lower odds to receive any pain medication, while Hispanic patients (+11%) were more likely to receive any pain medication as compared with non-Hispanic Whites. Disparities in psychiatric medication use were more consistent, with utilization patterns generally lower than non-Hispanic White patients across racial and ethnic groups. The findings of heterogeneous supportive medication usage patterns, both across and within racial minority groups, suggest that sociocultural and psychosocial factors may influence the use of supportive care medications. Sociocultural differences among racial and ethnic minorities have been considered as factors in medication adherence. Among these, beliefs about disease and the necessity of medications, and approach to self-management are the most prominently described. These factors can all be influenced by race and ethnicity including history, culture, family experiences, and individual preferences (27). Our findings also highlight that racial and ethnic minorities as a group are not homogenous, and a deeper understanding of specific influences on medication use within specific populations are needed. Another interesting finding is the bi-directional nature of the observed differences in pain medication use. Historically, racial medication use disparities have been described as decreased medication use among racial and ethnic minority patients as compared with White patients. Our study showed increased use of non-opioid analgesics among Hispanic and Asian patients compared with non-Hispanic Whites. As opioid therapy is considered the gold standard for pain management in patients with cancer, the increased use of non-opioid analgesics in Asians and Hispanics may indicate lower quality of pain management. Furthermore, the lower receipt of any pain medication by Black and Asian patients may indicate potential undertreatment of pain symptoms. However, further research exploring potential disparities in patient-related outcomes related to pain symptoms are needed to validate these hypotheses.

Our study results are consistent with previous reports which describe supportive care medication use disparities in other cancer populations. In a retrospective, observational study of 881 women with stage IV breast cancer, Check and colleagues identified racial disparities in supportive care medication utilization, primarily in the use of antidepressants and sleep aids, with Black patients using them less often than White patients (Risk ratio, 0.56; 95% CI, 0.39–0.80; ref. 23). However, no racial disparities were noted in opioid pain medication use, nor the overall use of supportive care medications. Although this study provided the initial data on supportive care medication racial disparities, the study had some characteristics that limit its broad applicability to patients with pancreatic cancer. First, the investigation focused only on the first 90-days post-cancer diagnosis, which may not allow for sufficient patient follow-up. Furthermore, the study only evaluated Black-White disparities, so the impact on other racial and ethnic minorities was not assessed. In another study, Lamba and colleagues evaluated supportive care medication used in patients with brain metastases and found similar results to our study, reporting that racial and ethnic minorities were less likely to be prescribed psychiatric medications, with ambiguous results in the use of pain medications (22). The study was well-designed but lack of inclusion of patients with pancreatic cancer limits applicability in this population. In a recently published study of opioid access among Medicare descendants with multiple types of late-stage cancer, Enzinger and colleagues identified that Black and Hispanic patients were less likely to receive opioids, and when prescribed received lower-intensity dosing as compared with White patients (28). Conversely, Lu and colleagues evaluated the use of opioids among patients with pancreatic cancer and reported no difference in opioid use by race and ethnicity (29). However, their study did not include non-opioids, which are an important component of pain management in pancreatic cancer-related pain.

Our study has several strengths. To our knowledge, our study is the first to investigate the use of supportive care medications to manage the constellation of pain and psychiatric symptoms experienced by patients with end-of-life pancreatic cancer. Given the unique burden of pain and psychiatric symptoms experienced by this population, and the impact of successful symptom management on HRQoL, this study focused on a clinically relevant topic using large-scale population data. Identification of supportive care medication use disparities in racial minorities is important as it can inform future research to develop interventions to close gaps in supportive care medication use in racial and ethnic minorities. Another strength is that the racial demographics of our study cohort are similar to those reported in previous literature (30), suggesting that our sample was representative of patients with pancreatic cancer. Another strength was our study design, which accounted for factors known to impact medication use (i.e., age, gender, chronic comorbid conditions, cancer stage), and the use of previous methodology designed to evaluate supportive care medication use among patients with cancer at end-of-life. In addition, another strength of our study was the inclusion of an adjustment model that excluded socioeconomic factors, which is unique relative to other studies that evaluated supportive care medication use in patients with cancer. Socioeconomic factors are closely related to race and ethnicity and are routinely adjusted in studies to isolate the impact of race and ethnicity on outcomes. However, the National Academies of Medicine (previously Institute of Medicine) definition of healthcare disparities accounts for all racial and ethnic differences in care that are mediated through factors other than patient preferences and health status (25). This definition acknowledges that socioeconomic status is often a downstream result of race and ethnicity, which in turn can influence health care quality and use. As a result, their definition includes differences in socioeconomic status (e.g., income, education) in the accounting of total disparities. The National Academies of Medicine definition of racial and ethnic healthcare disparities suggests that adjusting for socioeconomic status may reduce or eliminate the estimated independent effect of race on care (26). We chose to include a model that adjusts for socioeconomic status (Model 1) and another model that did not adjust for socioeconomic factors (Model 2). In our study, the different adjustment models produced similar results. Previous studies evaluating supportive care medication use adjusted for socioeconomic factors but did not include models that are consistent with the National Academies of Medicine definition of healthcare disparities. Another strength of the current study was the use of a comprehensive medication list based on previous literature to evaluate for the presence of supportive care medication use disparities (22).

Despite these strengths, our study is not without limitations. Due to its quantitative research approach, our study was not designed to determine the influence of sociocultural or psychosocial factors on supportive care medication use. Future research is needed to determine the influence of these critical factors on supportive care medication use disparities in racial minorities with pancreatic cancer. In addition, we used Medicare claims data to identify supportive care medication use, which primary includes older patients, limiting the ability to investigate pancreatic cancer in patients younger than 65 years. However, our study is representative of the population, because pancreatic cancer is predominantly a disease of older adults, with approximately 90% of patients diagnosed after age 55, and the incidence progressively increases with age (1, 31). In addition, because previous research has identified that older patients are at the highest risk of experiencing pancreatic cancer treatment disparities (32), the current study population is at high-risk for treatment disparities and warranted special investigation. In addition, we chose to be as comprehensive as possible in our assessment of supportive care medication use in pancreatic cancer, we included all eligible patients including those younger Medicare beneficiaries. However, younger Medicare beneficiaries (e.g., those who enrolled before the age 65 years) typically receive coverage due to disability and are not typically considered to be representative of the general population aged less than 65 or the disabled population. As noted, our results are not applicable to younger patients, and more investigation is needed to evaluate supportive care medication use in younger patients with pancreatic cancer. Another limitation of our study is that because we used Medicare claims data, our study would not capture medication use outside of Medicare. However, approximately 94% of the patients aged >65 have Medicare coverage suggesting that a substantial portion of overall medication use in this population would be captured through Medicare claims (33). In addition, because of the reliance on Medicare Part D outpatient claims data there is the possibility of uncaptured confounding variables including inpatient supportive care medication use, the use of non-prescription therapies such as herbal supplements or non-pharmacologic therapies, such as acupuncture, physical therapy, or mental health counseling. Future research investigation of the use of these therapies may be warranted. Another limitation of this study is the potential overlap in medication use across treatment categories. Supportive care medications often have several indications for use. We attempted to address this by defining specific supportive care medication categories based on their most common indication for use; however, the use of supportive care medications for less common indications cannot be excluded. Another limitation of this study was that it was not designed to evaluate patient-reported symptoms and quality of life, as these data were unavailable, leaving us unable to assess whether the disparities in supportive care medication use resulted in worse symptom management. Another possible limitation of the study is that our research question focused on the end-of-life population. As such, our findings may not apply to patients outside of this phase of care. However, given the unique symptom burden associated with pancreatic cancer, particularly at end-of-life, we believe that our focus on supportive care medication use disparities is important and expands our understanding about cancer survivorship in this population. Furthermore, inclusion of all patients with pancreatic cancer without limiting supportive care medication use to a specific period (i.e., 12 months before death) has the high potential of introducing survivor bias (i.e., patients who lived longer would have more opportunity to receive medications, while those who died early would not). Given the widely reported increased rate of mortality among racial minorities with pancreatic cancer, introducing this would limit applicability and skew the study results. Finally, because we used secondary administrative data for our analysis, the possibility of misclassification of race and ethnicity cannot be excluded. However, previously reported literature evaluating the quality of race and ethnicity obtained from the SEER registry data has described moderate (Hispanic ethnicity) to excellent (race) agreement between SEER racial and ethnic data and self-identified race and ethnicity (34).

In conclusion, supportive care medications are cornerstones of the management of pain and psychiatric symptoms in patients with pancreatic cancer. The effective use of supportive care medications have previously been reported to improve HRQoL. In this study of nearly 75,000 patients with end-of-life pancreatic cancer, racial and ethnic disparities were identified across multiple supportive care medication classes, including opioid and non-opioid analgesics, and psychotropic medications. However, the observed differences were heterogeneous among racial and ethnic groups, with significant intragroup differences noted, suggesting that sociocultural or psychosocial factors may influence supportive care medication use in racial minorities, but further inquiry is needed. Our findings are particularly important given the emerging importance of symptom management on HRQoL and survival. Eliminating disparities in supportive care medication use may help to close existing racial gaps in these important outcomes in patients with pancreatic cancer. Future research is needed to explore causal factors and design of potential interventions to close supportive medication use disparity gaps in patients with pancreatic cancer.

J.M. Allen reports grants from NIH during the conduct of the study. Y. Guo reports other support from Merck outside the submitted work. S.C. Rogers reports other support from Natera Oncology outside the submitted work. D.J. Wilkie reports grants from NIH/NCI during the conduct of the study; other support from eNursing LLC outside the submitted work. No disclosures were reported by the other authors.

The interpretation and reporting of these data are the sole responsibility of the authors and do not necessarily reflect the opinions of the NIH, State of California, Department of Public Health, the NCI, and the Centers for Disease Control and Prevention or their contractors and subcontractors.

J.M. Allen: Conceptualization, resources, formal analysis, supervision, funding acquisition, investigation, methodology, writing–original draft, project administration, writing–review and editing. M. Awunti: Investigation, writing–original draft, writing–review and editing. Y. Guo: Resources, data curation, software, formal analysis, validation, methodology, writing–review and editing. J. Bian: Resources, data curation, software, formal analysis, validation, methodology, writing–review and editing. S.C. Rogers: Conceptualization, writing–review and editing. L. Scarton: Conceptualization, writing–review and editing. D.L. DeRemer: Conceptualization, writing–review and editing. D.J. Wilkie: Resources, formal analysis, methodology, writing–review and editing.

This work was supported in part by the following grants (to J.M. Allen: NIH U54 CA233444; to Y. Guo: NIH R01 CA246418, NIH R21 CA253394, NIH R21 CA245858; to J. Bian: NIH R01 CA246418, NIH R21 CA253394, NIH R21 CA245858; to D.J. Wilkie: NIH U54 CA233444, NIH U54 CA233396, NIH U54 CA233465).

This study used the linked SEER-Medicare database. The authors acknowledge the efforts of the NCI; Information Management Services (IMS), Inc.; and the Surveillance, Epidemiology, and End Results (SEER) Program tumor registries in the creation of the SEER-Medicare database. The collection of cancer incidence data used in this study was supported by the California Department of Public Health pursuant to California Health and Safety Code Section 103885; Centers for Disease Control and Prevention's National Program of Cancer Registries, under cooperative agreement 1NU58DP007156; the NCI's SEER Program under contract HHSN261201800032I awarded to the University of California, San Francisco, contract HHSN261201800015I awarded to the University of Southern California, and contract HHSN261201800009I awarded to the Public Health Institute, and not to the authors.

The authors also wish to thank the Cancer Informatics Shared Resource at the University of Florida Health Cancer Center for support in accessing and analyzing the SEER-Medicare dataset.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.