1 INTRODUCTION

While cancer is a leading cause of death, little is known about the epidemiology and health care utilization prior a diagnosis of cancer, including emergency department (ED) use.¹ In 2017, approximately 2 million adults nationwide reported the ED as their usual course of health care, with 18 million adults reporting no access to routine medical care.² While cancer is routinely and ideally diagnosed through elective screening or by primary care physicians, 20%–50% of global cancer diagnoses originate from an index ED visit. Higher frequencies of ED-associated diagnoses are among racial/ethnic minorities and patients of lower socioeconomic status.³ Prior work from the United Kingdom's “Routes-to-Diagnosis” work suggested that 23% of newly diagnosed cancer patients presented emergently and survival rates were much lower for emergent presenters.⁴ When people present to the ED with life threatening manifestations of their undiagnosed cancer they are more likely advanced stages and are less likely to be curative.^{5, 6}

Safety-net hospital systems, including safety-net EDs, disproportionately serve vulnerable patient populations.⁷ Safety-net hospitals generally treat greater percentages of racial/ethnic minorities, Medicaid and uninsured patients, patients from communities in the lowest income quartile, and provide large amounts of uncompensated care.⁷ The populations that rely on safety-net hospital systems for care are the same populations that experience cancer care disparities, including disparities in equitable cancer screening opportunities and less access to the most advanced cancer treatments.⁸ Little descriptive data exists regarding the prevalence of cancer types treated at safety-net hospitals.⁹ Less is known about the frequency at which patients seek medical care in safety-net EDs prior to their diagnosis of cancer. We hypothesize that patients diagnosed in proximity to an ED visit have worse outcomes than patients without prior ED use, and these outcomes will be exacerbated by health disparities. This will suggest that the ED visit serves as a proxy for either inadequate health care maintenance/screening, limited access to health care services, or patients will present with more advanced and symptomatic disease.

While increasing evidence and factors associated with emergency presentations of cancer are available^{5, 10-13} (and primarily based out of the United Kingdom [UK]), to our knowledge, there are no administrative-database studies conducted in the United States (USA) that have examined the characteristics, eventual cancer types, and outcomes of ED utilization during the pre-diagnosis phase. This context is important, as studies from the UK cannot necessarily be generalized to the USA. This information may help illuminate potential gaps in access, quality, and equity of both cancer screening and the state-of-the-art cancer treatment, especially as they pertain to patients that receive a suspected cancer diagnosis in the ED.

2 MATERIALS METHODS

3 RESULTS

In total, over the 3-year study period, 3699 patients with cancer were identified in the EHR, with 1239 (33.50%) cases having had an ED visit within 6 months prior to the diagnosis of cancer. Figure 1 provides a flowchart of the assembly of the study cohort.

4 DISCUSSION

Patients who present to the ED with life threatening manifestations of their undiagnosed cancer are known to present with advanced stage malignancies and ultimately suffer higher mortality rates than those diagnosed through non-emergent pathways.⁵ Most literature in the space of emergency presentations of undiagnosed cancer is based out of the United Kingdom.⁵ In this work we begin to investigate the association between ED visits and a subsequent diagnosis of cancer by describing and comparing those patients with an ED visit prior to the diagnosis of cancer to those that have no ED visits within our safety-net health network. We demonstrate gender, age, insurance, social history (tobacco use, alcohol use) differences among those that use the ED and are subsequently diagnosed with cancer. Additionally, we demonstrate the most common cancers collectively seen at our safety-net hospital, with significant differences among cases and controls for many screening-preventable malignancies. Little to no literature exists examining the association with eventual cancer diagnosis in association with an ED visit, offering opportunities for addressing cancer prevention and cancer screening in the context of an ED visit.¹⁹

Describing post-cancer diagnosis ED utilization has been studied, and anywhere from 2% to 4% of all ED visits are attributable toward persons with a diagnosis of cancer.²⁰ As cancer survival rates increase, the number of cancer-related ED visits also increases, and future work is needed to improve cancer management in the ED and potentially reduce unnecessary ED visits.^{20, 21} Prior ED use among patients with cancer is the strongest independent predictor for future ED use after a new diagnosis.²² Since ICD-coded cancer diagnoses require tissue sample and biopsy, little to no confirmed cancer diagnoses originate from an ED visit. Instead, many patients in the ED are discharged with symptom-based discharge diagnoses, and thus linking a subsequent cancer diagnosis to an exact ED-visit is challenging in administrative databases.²³ Thus, our work adds to the current literature by describing the population (in a safety-net setting) that is diagnosed with cancer, and what major cancer types are diagnosed shortly after an ED visit.

Lung, pancreatic, colorectal, liver, and kidney cancers are all seen at higher rates in the ED prior to the diagnosis of cancer, suggesting that some undiscovered moderators may be integral to this problem, such as tobacco use among underserved populations.²⁴ Tobacco use is linked to many cancers and in this work there is a higher observed frequency of tobacco use among cases compared with controls (42.0% vs. 30.9%, p < 0.0001).²⁵ EDs frequently provide care to populations that use higher tobacco at higher rates than the general public.²⁶ Thus, it is not surprising that the most common cancers seen in our cases cohort are those that are heavily mediated by tobacco use, namely lung, HENT, pancreatic, colorectal, liver, and kidney. These results are similar compared with a statewide epidemiologic assessment we previously performed, again demonstrating that tobacco-related cancers appear to be associated with ED visits.^{10, 24} Randomized clinical trials have demonstrated success at improving tobacco abstinence in low-income ED smokers, and our work demonstrates further need for wide spread tobacco cessation programs and cancer prevention education programs which could possibly be facilitated out of the ED.^{19, 27} Additionally, this work provides for the data needed to selectively target those cancers that are more likely to be suspected in an ED visit, namely lung, breast, prostate, colorectal, and cervical cancers due to their prevalence of being seen in the ED prior to their diagnosis.

The presence of an ED visit prior to a cancer diagnosis in this work does not necessarily represent causation of an eventual diagnosis, but instead confers an opportunity for an ED physician to help transition the care of a patient with likely cancer. For example, lung cancer was the most common cancer seen in the ED prior to an eventual diagnosis, and whether that ED visit was associated with the diagnosis is unclear. One year survival rates for lung cancer can be as low as 15% when stage IV (compared to 85% stage I), and earlier detection is known to be associated with improved survival rates.²⁸ Delays in cancer treatment are common, and even a 4-week delay of cancer treatment is associated with increased mortality across surgical, systemic treatment, and radiotherapy indications for seven cancers.²⁹ Patients with lung cancer can experience delays upwards of 4–5 months from first diagnostic test until a definitive diagnosis.³⁰ It remains unclear how the involvement of the ED physician, ED-associated diagnostic test, and transitions of care from the ED visit are associated with these delays. It stands to reason that many ED patients observe delays in their care that could be prevented by improving and addressing the transitions of care for a newly suspected cancer diagnosis.

Cancer screening exists for many of the most prevalent cancer types, such as breast, colorectal, lung, cervical, and prostate cancers, with other focused screenings for various genetic-based cancers.³¹ The primary goal of cancer screening is to detect a malignancy before it is large enough to result in symptoms, which usually occur at more advanced stages. Breast cancer screening has a high rate of screening attendance in the United States, with over 76% of females having reported a mammogram within the last 2 years.³² This likely explains why a higher percentage of the controls are diagnosed with breast cancer than are cases. The uptake of lung cancer screening with yearly low-dose computed tomography (LDCT) is much lower than observed for breast cancer, with only <6% of screening-eligible smokers being screened.³³ Recent guideline changes in lung cancer screening eligibility were made in an attempt to ameliorate disparities in lung cancer screening access for many populations. It is hypothesized that the lung cancer screening attendance rate will decrease due to the large number of newly expected lung cancer screening eligible Americans.³⁴ Herein, a large number of patients with subsequently diagnosed lung cancer, and with ongoing tobacco use, utilize the ED in some capacity. Nationally, millions of Americans in need of recommended cancer screening are cared for in EDs, many of whom are hard to reach through other healthcare settings.¹⁹ Thus, the ED may potentially serve as a novel location to help addresss disparities in cancer screening uptake by improving access to cancer screening and prevention.¹⁹

Compared with private insurance, patients with Medicaid have been shown to present with more advanced cancer at time of diagnosis and are less likely to receive curative therapies, such as for prostate cancer.³⁵ ED usage is high among uninsured and Medicaid patients.³⁶ Lack of insurance is associated with lower rates of preventive care, delays in necessary care, and overall mortality in many studies.^{36, 37} Thus, it is not unexpected that insurance status, when compared with commercial insurance as a usual payer source, is associated with both mortality and the odds of being seen in the ED within 6 months of diagnosis. Patients with non-private insurance have been previously shown to be more likely diagnosed with cancer at advanced stages and have worse survival, possibly due to not receiving appropriate and timely cancer screening, diagnosis, and quality care.³⁸

Many variables were determined to be confounders in both predicting mortality and the likelihood of having an ED visit prior to the diagnosis of cancer. None of the studied variables had statistically significant associations with each other, and thus no effect modification was observed. However, many unobserved confounders were not included in this study, such as access to prior health care, access to prior cancer screening, socioeconomic status, access to transportation, education, and metabolic syndrome. Many of these unobserved variables are likely influential in the diagnosis of cancer through an ED visit, such as access to prior health care use. Medicaid patients have greater challenges in healthcare access compared with patients with private insurance, and thus knowing many of the patients in this study have Medicaid insurance it is likely that many patients do not have reliable primary care.³⁹ Access to equitable cancer screening opportunities varies depending on race, socioeconomic status, and other disparities, and many underserved populations do not have equitable access to cancer screening.⁴⁰ Eskenazi Health serves a primarily underserved population as the main safety-net hospital system for the greater Indianapolis metropolitan area. In our results, many are not generalizable toward all EDs, however, our population represents the poorest areas of Indiana (both rural and urban Indiana), and thus likely are representative of many rural and urban EDs across the United States.

Various types of cancers are more commonly associated with ED visits than others, along with certain risk factors such as insurance status, tobacco use, male gender, and overall CCI being more likely to be a case than a control. Many of the cancers in this cohort do not have a recognized screening modalities, likely due to their rare occurrences, such as penis cancer. Thus, future work will begin to focus on the more common ED-associated cancers, such as lung and colorectal. Lung and colorectal cancers do not observe as high rate of screening attendances as breast, and thus future efforts are needed to expand access and awareness for these cancers screenings.⁴¹ Exploring prior heath care utilization by patients with cancer will offer opportunities to improve cancer care delivery in the ED.

5 LIMITATIONS

Several limitations exist in this study, including the retrospective nature by which the data were obtained. This method of data collection does not account for cancer stage at diagnosis, or if the prior ED visit was related to the eventual cancer diagnosis. Further, many patients present to the ED with undiagnosed cancer and are subsequently discharged from the ED with outpatient follow-up.²⁴ We chose 6 months as the time cutoff to be a case, as many patients experience significant delays (months) in cancer staging and treatment initiation. In our work, 33.5% of patients with cancer in our system were seen in our ED prior to their eventual diagnosis of cancer, which is in accordance with reported frequencies by which cancer diagnoses present emergently.³ Thus, it is likely that some, if not many (especially for lung cancer), cancer diagnoses originate from an ED visit, and we attempted to capture any and all cases by extending our case definition to 6 months. Additionally, we relied on the ICD-10 codes for obtaining our cohort, which may lead to bias in our sample selection. Patients with cancer frequently visit the ED with end-of-life presentations, many of whom go undiagnosed and never receive biopsy-confirmed cancer.⁴² The burden of end-of-life manifestations of undiagnosed cancer that present acutely to the ED is unknown; however, would likely worsen the outcomes in this study if we could include them in our cases cohort.

We did not account for previous cancer screenings and the impact that those screenings may have on the ultimate diagnosis. Many of the cancer types have multiple methods and opportunities for screening. For example, to screen for colorectal cancer patients can undergo a colonoscopy, flexible sigmoidoscopy, or even fecal occult blood testing. Lastly, due to the heterogeneity of cancer types studied, we did not analyze the treatment types (surgical, immunotherapy, chemotherapy) and its impact on survival, although we surmise that ED-associated cancer diagnoses would receive fewer curative therapies.

6 CONCLUSIONS

Patients with an eventual cancer diagnosis, in our safety-net health care setting, frequently utilize the ED prior to their diagnosis. Those who utilize the ED prior to their diagnosis of cancer are more likely to be black, are more likely to not have commercial insurance, and have worsened outcomes than those who do not use the ED. Among those who are seen in the ED prior to an eventual diagnosis, lung cancer is overwhelmingly the most common cancer seen in the ED. Lastly, having an ED visit prior to the diagnosis of cancer is likely a proxy for either inadequate health care maintenance/screening, limited access to health care services, and is associated with more advanced disease at time of diagnosis. Future work should explore the disparities observed among those who utilize the ED prior to the diagnosis of cancer, and interventions will be designed to help reduce the cancer care disparities observed in the ED. In conclusion, many people utilize the ED prior to a biopsy-confirmed cancer diagnosis, offering an opportunity to address cancer prevention, cancer diagnosis, and cancer screening in the ED setting.

AUTHOR CONTRIBUTIONS

Nicholas R Pettit conceived the study, designed the study, and obtained IRB approval. Nicholas R Pettit, Xin Li, Jeffrey Kline managed the data, assisted in drafting of the manuscript, and contributed to the revision. Nicholas R Pettit performed statistical analysis.

FUNDING INFORMATION

None.

CONFLICT OF INTEREST

The authors have no disclosures.

ETHICS APPROVAL

This study was approved by the Indiana University's Institutional Review Board (IU IRB), #2001716264 prior to commencing the study.

LAY SUMMARY

The emergency department serves as a common health care setting where patients with an eventual diagnosis of cancer are seen prior to their diagnosis. The association between an emergency department visit and the eventual diagnosis of cancer in terms of cancer type and mortality is unknown. In this retrospective cohort study involving 3699 patients with non-skin cancers at an urban safety-net hospital, 33.50% are seen in the emergency department prior to their diagnosis of cancer, with lung cancer being the most common cancer type seen in the emergency department. The emergency department visit is an independent risk factor for mortality.

PRECIS FOR USE IN THE TABLE OF CONTENTS

Over one third of patients with cancer (1239/3699) were seen in the ED within 6 months prior to their cancer diagnosis. Higher mortality rates were observed for those seen in the ED and those without commercial insurance.