Our findings show uniformly worsened outcomes of all patients during a period of five years, especially in minority and low-income patients, diagnosed with new cancer diagnosed in temporal proximity to an ED visit. These data infer that patients with undiagnosed cancer who rely on the ED as a source of primary medical care will have a worsened outcome compared with patients who have organized medical care.6 Prior work from the United Kingdom’s “Routes-to-Diagnosis” conducted by the Public Health England, suggested 23% of newly diagnosed cancer patients presented emergently and survival rates were much lower for those emergent presenters, which is similar to what we discovered in this work.9
In the US, a study from Michigan, lung and colorectal cancer patients with ED-associated cancer diagnosis had more advanced staged cancer and were more likely to be Black than those diagnosed without a recent ED visit, but no data on mortality were presented.10 In this study, the authors chose a conservative approach by defining an ED-diagnosed cancer as one that had an ED visit in the month of or month before the diagnosis of cancer, however, prior to this publication there is no precedent for defining an ED-associated cancer. Instead, we chose 6 months as our conservative time point as to define ED-associated cancer, as there is substantial evidence that ambulatory follow-up compliance after ED discharge is poor, and has been estimated to be between 26% and 56% depending on the ED population.11 Additionally, there is some evidence that suggests almost one third of patients with new onset cancer experience delays of over 90 days, and thus there is no precedent for establishing the optimal time from ED visit to cancer diagnosis.12 Among 9,470,626 ED visits among Medicare beneficiaries aged 65 and older, nearly 30% lacked ambulatory follow-up at 30 days, with lower rates of ambulatory follow-up observed among those of Black race and seen in rural EDs.13 Further, a study from Western Australia demonstrated that of 1358 people with incident breast, prostate, colorectal, and lung cancers the diagnostic interval from symptom to diagnosis ranged from 92-108 days, further suggesting the time to cancer diagnosis can exceed several months.14 Rural health disparities exacerbate the time to cancer treatment and ultimately mortality, due to long travel times, low availability of clinical trials, and additional health burdens, an issue that is well observed in the state of Indiana.15 Thus, additional research is clearly needed among US populations to delineate and improve the time to cancer diagnosis and treatment for ED-suspected cancer patients.
It has been demonstrated that emergent presentations of cancer are associated with lower curative rates and treatments, even when compared to cancers diagnosed “electively” (even at the same stage).16 The present data in Figure 2, together with prior findings support the inference that patients who have a diagnosis of cancer temporally connected with an ED visit, suffer from a disparity in their odds of survival. From a health services standpoint, potentially modifiable causes for this disparity include lack of access to primary medical care and cancer screening before diagnosis, increased rate of tobacco and alcohol use, worsened uncontrolled comorbidities at the time of diagnosis, and lack of access to specialty cancer care after diagnosis. Thus, additional work should focus on reducing the time to cancer diagnosis as expedited diagnosis of symptomatic cancer likely benefits patients’ survival and improved quality of life as demonstrated in a systematic review of over 200 studies.17
This work employed a state-wide assessment of cancer diagnoses in the state of Indiana. From this, numerous patient factors appear to contribute to the observed mortality and worse outcomes, namely race and socioeconomic status. Meanwhile, factors such as age and sex appear to have little relationship between the association of diagnosis and mortality. This is compared to other works where older age (≥85 years old) are 2.5 times more likely to present with an emergent diagnosis of cancer, when compared to a 65–74-year-old cohort.18 Those authors conclude that cancer and age are likely to reflect disease specific factors. Further, our work we excluded pediatric patients (<18 years old), and it is well known that more than half of patients that present with de novo cancer diagnoses in the ED are emergently diagnosed.19
What limited evidence exists from data obtained in the United States, has demonstrated associations between socioeconomic status and the diagnosis of cancer as an emergency. African Americans in one study had increased odds of emergently diagnosed colorectal cancer (AOR 1.5, 95% CI 1.38-1.63) as compared to a similar white cohort.20 This disparity not only exists among the diagnosis of cancer but in the primary and secondary preventions among lower SES populations. Colorectal screening, despite its efficacy and recommendations, has been shown to be low among African Americans and those with low SES, demonstrating an opportunity or intervening on this high risk population in the ED.21 Furthermore, the low SES population have inequities that result in poor lifestyle choices, some of which (smoking status, diet, physical activity) are preventable and modifiable if this population had access to equitable opportunities. Cancer mortality among this population has an association between mortality and modifiable risk factors, again at current date are not routinely performed in the ED.22 Blacks are diagnosed with breast and lung cancer in the cases versus controls and their outcomes appear to be worse. We speculate that blacks are more dependent upon EDs than whites for unscheduled, emergent care, and thus are more likely to present to an ED emergently for their undiagnosed malignancy.23 Similar phenomenon can be applied to whites with low SES, where patients of low SES are more reliant on the ED for their care and are more likely to present emergently with their undiagnosed cancer.24,25 These data suggest race and socioeconomic status are more important than other factors, such as comorbidities, since the CCI was equal. While not examined in this large data analyses, there is growing evidence that African Americans have more aggressive tumor biology, such as in several breast cancer studies, which can be speculated that this also contributes to these patients presenting more emergently for their undiagnosed cancer.26
Of the cancers that have proven screenings that demonstrate success, lung and colorectal cancers do markedly worse when associated with an ED-visit, 54% vs 18% mortality for lung cancer (P<.0001), and 26.4% vs 9% for colorectal cancer (P<.0001). Breast cancer also has a successful screening modality, namely mammography, and suffers from similar poor outcomes when associated with an ED visit, 11.9% mortality vs 3.4%, ED to non-ED associated.27 Prostate cancer screening is controversial and recommendations vary by organization and country, regardless, a screening modality is available and those diagnosed with prostate cancer associated with an ED visit have higher mortality than those that don’t (13.2% vs 3.5%).28 Lastly, cervical cancer also is frequently screened for as outpatients and 22% of those seen in the ED were dead, versus 7.4% of those not seen in the ED. This evidence is supported by known disparities in cancer screening, with minority patients experiencing greater delays in evaluation and screening for cancer, leading to suboptimal treatment among those patients subsequently diagnosed with cancer.29
Previous research has called for both improving the outcomes of patients that are diagnosed with cancer through an emergent presentation, as well as helping to reduce the burden of emergent diagnoses by improving cancer screening.6,30 This is likely a systems issue, but plausible future steps is utilizing the ED space for more than just emergent care. Average length of stays in ED has multiple variables that impact exact time frame which patients sit in the ED, but in one paper an average a patient can expect a wait of 4 hours.31 As the trend of increasing length of stay continues to increase nationwide, EDs are experiencing the unfortunate phenomena of ED crowding which has been well demonstrated to be associated with increased hospital death.32 We propose future work in developing interventions for this at risk population while in the ED, similar to what has been performed for rapid hepatitis screening and cervical cancer screening in urban EDs, demonstrating proof of concept and utilizing the ED space for more than emergent care.33,34 Similar revolutionary changes in ED workflow for improving the overall health of ED patients has been adopted with universal HIV screening in the ED, as well as universal suicide screening in EDs.35,36 Intervening on this population is challenging but supported by a Cochrane Review, guaiac fecal immunochemical test (FIT) can reduce colorectal mortality by 15%, and providing appropriately chosen patients with home use FIT tests through primary care has improved screening rates, sustaining a screening rate of 75%.37,38 Currently screening routinely for cancer does not occur in the emergency setting, but for many vulnerable patients (uninsured, lower SES, racial minorities), the emergency room serves as the only opportunity for routine care and we should begin to explore alternative strategies to maximally improve the care of these patients.39 Removing the barriers to cancer screening, such as providing patients with FIT cards prior to discharge may represent an opportunity to increase adherence to CRC screening and reduce the burden of emergently diagnosed CRC.40 Novel approaches need be undertaken at the systems and health policy level to address the disparities that are well demonstrated among ED-associated cancer diagnoses.
There are limitations to this study, namely the retrospective methodology to obtaining the administrative data. A diagnosis of cancer, while suspected in the ED, usually occurs weeks to months after the actual ED encounter, and thus confirming linkage from a suspected ED visit and a diagnosis is challenging. Further, patient stage and tumor specific biology are not included on the ICD-coded diagnosis and thus knowing the stage and extent of disease is not available without retrospective chart review. Regardless, we made an inference that being diagnosed with cancer within 6 months of a recorded ED visit, meant those ED physicians had an opportunity to diagnose asymptomatic cancer, if that ED visit wasn’t directly related to the presentation of emergent cancer diagnosis. The relationship between the ED visit and the diagnosis of cancer are unclear, due to lack of detailed ED visit information as well as most patients that are discharged from the ED have symptom-based discharge diagnoses as opposed to a definitive discharge diagnosis.41 Additionally, no knowledge is known about previous screenings and primary care follow up, thus no inference can be made to know whether or not screening may have reduced the likelihood of the found associations. The lack of follow-up knowledge means we can’t examine the relationship between mortality and inadequate access to expert care, but the correction for clustering on logistic analyses suggests this is not just a result of location. Ultimately, it is difficult to determine if the observed outcome differences primarily reflect lead-time bias, health care disparities, or other factors, but comparing to the limited available data primarily from Europe, these data clearly present a concerning trend for patients with a new diagnosis of cancer.
Further limitations include those cancers that are so advanced that tissue biopsy is not obtained, or patients prefer to not seek treatment. Lastly, pediatric cancers and their association with being emergently diagnosed in the ED were not explored in this study. No pediatric cancers have preventable or modifiable risk factors, and as such the goal of this work is to find interventions on cancers that can be prevented with lifestyle modification (smoking cessation, weight loss), or caught earlier with age-appropriate screening.