Medical oncology referral and systemic therapy of patients with advanced stage urothelial carcinoma

Prior literature has documented the benefits of multidisciplinary management of UC patients, which is associated with improvement in informed and collaborative decision making and possibly earlier initiation of definitive treatments [17–19]. Using the linked SEER-Medicare dataset, we provide a detailed analysis that fills knowledge gaps related to patients with newly diagnosed advanced UC in terms of physician visits, patient referrals and treatment. In this study, we used chemotherapy as the ‘anchor’ to gauge receipt of systemic treatment for the aggregate stage IV UC patient population as well as the M0 and M1 subpopulations that together define stage IV disease. Our aim is to provide more granular information about these aspects of UC management than are generally reported, particularly since such an analysis may help provide a model to further study and understand UC management and outcomes as the treatment landscape continues to evolve with the advent of other systemic therapies beyond the traditional chemotherapy which remains an important backbone of treating advanced UC.

This study found that referral among physician specialties was common, consistent with reported literature [18–20]. An important aspect of the study was to not only look at the urologist but also the relatively understudied specialist, the PCP, within the context of their potential ‘gatekeeper’ role as they may relate to UC. Interestingly, patients who presented to a PCP initially were more likely to be older, have higher comorbidities and poor performance status compared with those presenting to a urologist, suggesting that PCPs may be interacting with the sicker UC population at the index visit (Supplementary Table 1). Although physician referrals varied by the index physician, the role of the PCP in the management of UC patients was prominent. Remarkably, 29% of patients with an index PCP visit were not referred to another specialist, and 50% never saw a medical oncologist over their disease course, which is surprising given that PCPs generally do not provide UC-directed care. Among the patients with index PCP visits who were referred out, the referrals tended to occur rapidly and were predominantly channeled to a urologist, with only a minority of such patients being directly referred to a medical oncologist.

Our descriptive and adjusted models show that urologists were more likely, compared with PCPs, to refer UC patients to a medical oncologist. Interestingly, the median time to visiting a medical oncologist is longer for patients who see an index urologist (52 days) compared with an index PCP (29 days). Seventy percent of the referrals to the medical oncologist stemmed from the urologist, highlighting the prominent role of the latter in channeling patients to systemic treatment. In exploratory descriptive analyses of overall survival by first specialty visits, patients who were initially managed by a urologist had the highest median survival while those initially seen by a PCP had the lowest survival; this trend was observed for both M0 and M1 patients (Supplementary Figure 3). Our results for stage IV UC also mirror findings of a study of muscle invasive bladder cancer, which found that more than 80% of medical oncologists had their patients referred from a urologist, with very few receiving referrals from PCPs or radiation oncologists [21], underscoring the gatekeeper role of the urologist in mediating patient flow and receipt of systemic therapy in both advanced UC and earlier stages of UC. Despite their key role in UC management, our analysis revealed that 30% of stage IV UC patients with index urologist visit did not get any subsequent medical oncologist referral during their entire disease course. However, this is still a lower proportion of patients not being referred out compared with those who had index PCP visit where 50% never had a medical oncology follow-up. Our model also shows that patients had lower relative likelihood of being referred to medical oncology if they were M0, aged 80 years or older, had more comorbidities or underwent cystectomy. It has been shown previously that patients who are older and have comorbidities are less likely to be suitable for chemotherapy [8,22], and hence not referred as quickly to medical oncology. Patients undergoing cystectomy may also be less likely to be referred to medical oncology as they could be receiving peri-operative and postoperative care through the urologist [23].

Looking at patient–physician flow patterns and chemotherapy receipt, certain demographic characteristics emerge that predict for the likelihood of receiving systemic therapy. Consistent with prior studies, older patients have a lower relative likelihood of seeing a medical oncologist or receiving chemotherapy post urologist visit, perhaps in part due to potential comorbidities/relative performance status [6,8,24,25]. Further, being Hispanic, non-Hispanic black, widowed/separated/divorced/never married or being treated by a physician in the Southern US are predictors of lower relative likelihood of receiving chemotherapy. The role of race in the context of delay in chemotherapy receipt or less likelihood of receiving multimodal therapy, or not receiving chemotherapy altogether, as well as poorer survival outcomes, have been well documented by others [26–29]. Better social support, as reflected by marital status, has been shown to lead to greater relative likelihood of UC treatment and overall survival, particularly among males [27,28]. The role of geography also needs further clarification but may reflect physician availability and proximity to treatment centers, keeping in mind that the Southern US houses some of the most socioeconomically deprived counties in the country [26,30,31]. Outlining the referral patterns could be helpful in informing patients, payers and specialists as it provides an understanding around the patient journey, characteristics and the time it takes for a stage IV patient to be referred to medical oncology and to receive systemic chemotherapy. Our analysis suggests that we should educate on the importance of timely referral from PCP/Urologist to medical oncology for timely receipt of systemic chemotherapy.

Strengths of the study include the fact that the SEER-Medicare database is a large nationally representative database that provides relevant longitudinal information on patients. Since the median age at diagnosis for all (not only Medicare) patients with UC is about 73 years, the SEER-Medicare dataset captures most of the UC patients. Further, the availability of tumor, nodes and metastases (TNM) staging within SEER allows for analyzing advanced stage UC patient data, which would not be possible with claims data alone. Finally, linking outpatient and carrier files to the AMA Physician Masterfile allowed us to identify the specialties and demographics of the treating physicians. The integration of these three data sources provide robust insights into patient and provider demographics and treatment patterns.

Our study also has some inherent limitations. As referral patterns were captured over the duration of study follow-up from the time of the index visit, some patients with poor prognosis and shorter overall survival may not have enough time to enable referrals to different specialists. Similarly, patients diagnosed in the last year of the data cut may be administratively censored in other words, they may have seen specialists, but we were not able to capture such data since the most updated data available would be too close to their diagnosis. Further, UC patients may see specialists beyond those identified in this analysis (e.g., physician assistants, oncology nurses, nutrition specialists, pathologists and others). Clinical characteristics, including M stage, CCI score and performance status were measured only at diagnosis, and therefore disease improvement or progression during follow-up that could possibly affect patient treatment or referral patterns was not captured. Although the data used for this analysis was from 2003 through 2014, which precedes the recent integration of ICIs and other newer therapies in advanced UC, our study provides a relevant benchmark to further assess patient–physician dynamics in the context of emerging therapies, keeping in mind that up to very recently, chemotherapy has been the mainstay of systemic management of UC [32,33].

This is one of the few studies to look at physician specialty visits and referral patterns, as well as the determinants of time to medical oncologist visit and chemotherapy receipt among patients diagnosed with incident stage IV UC, including the M0 and M1 subpopulations that define stage IV disease. Future studies should continue to explore which configurations of multispecialists result in the most coordinated, timely and optimal care and improved survival for these patients, particularly as some of the newer therapies become increasingly available for the UC patient. Although younger age, good performance status, social support and race/ethnicity were important determinants of time to seeing a medical oncologist or receiving chemotherapy, it will be relevant in future work to monitor whether the newer therapies can help bridge some of the gaps observed in the management and outcomes of advanced UC patients.

A Aly, MF Botteman, Y Doleh and R Shenolikar conceptualized the research question and study design. MF Botteman was responsible for data acquisition. A Hussain was the lead clinician reviewing methods and results.

A Aly and C Johnson were responsible for data analysis and C Johnson drafted a first version of the manuscript. All authors reviewed and approved the final version of the manuscript. The authors on this manuscript have no other relevant financial or other relationships to disclose.

The authors thanked C Mirvis and V Chirikov for medical writing support and Z Chen and V Chirikov for analytical support.

A Aly was an employee of Pharmerit International during the conduct of this study and is currently an employee of AstraZeneca (200 Orchard Ridge Drive, Gaithersburg, MD 20878).

AstraZeneca provided funding for this research. A Aly was an employee of Pharmerit International during the conduct of this study; C Johnson and MF Botteman are current employees of Pharmerit International; MF Botteman is a shareholder of Pharmerit International. Y Doleh, A Aly and R Shenolikar are employees of and have stock options in AstraZeneca. Part of A Hussain’s time was supported by a Merit Review Award (I01 BX000545), Medical Research Service, Department of Veterans Affairs. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Writing assistance was provided by C Mirvis and V Chirikov from Pharmerit.

This retrospective longitudinal analysis used de-identified data from SEER-Medicare database, and was granted Institutional Review Board exemption. All results were reported per the data use agreement with the National Cancer Institute.