- Email: [email protected]
Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes
The American Journal of Surgery xxx (2017) 1e7
Contents lists available at ScienceDirect
The American Journal of Surgery journal homepage: www.americanjournalofsurgery.com
Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes Ian Solsky a, Bruce Rapkin b, Kristen Wong a, Patricia Friedmann a, Peter Muscarella a, Haejin In a, * a b
Montefiore Medical Center/Albert Einstein College of Medicine, Department of Surgery, Bronx, NY, USA Montefiore Medical Center/Albert Einstein College of Medicine, Department of Epidemiology and Population Health, Bronx, NY, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 19 August 2017 Received in revised form 4 October 2017 Accepted 10 October 2017
Background: The impact of diagnosis location on gastric cancer (GC) outcomes is poorly defined. Methods: Detailed chart review was conducted to identify presenting location leading to diagnosis and treatment for GC patients at a single institution (2009e2013). Patients treated non-emergently following a diagnosis prompted by an ED visit (EDdx) were compared with those diagnosed at other locations (non-EDdx). Results: EDdx patients comprised 52% of 263 GC patients. They were older, had later cancer stages (stage IV: 50% vs. 24%), more comorbidities (3: 68% vs. 47%), and presented with non-specific symptoms like bleeding (21% vs. 5%). Both groups were of similar race and insurance status. In a model adjusted for stage, EDdx was associated with increased mortality (aHR 1.9; 95% CI: 1.2e2.9). Conclusion: Half of GC patients had an ED visit prompting diagnosis, which is independently associated with increased mortality. Efforts should focus on reducing EDdx rates to improve GC outcomes. Summary: Demographics, treatment, and outcomes of gastric adenocarcinoma patients at a single center whose diagnosis was prompted by an emergency department (ED) visit were compared to that of patients diagnosed in non-ED settings. 52% of patients had an ED visit prompting their diagnosis, which was associated with an increased mortality risk. © 2017 Elsevier Inc. All rights reserved.
Meeting Presentation: Results in this manuscript were initially presented at the 69th Society of Surgical Oncology (SSO) Annual Cancer Symposium, March 2e5, 2016 in Boston, Massachusetts, USA. Keywords: Gastric cancer Emergency department Cancer diagnosis
1. Introduction Although emergency departments (ED) are intended to manage acute medical problems,1 they frequently serve as the entry point into the healthcare system for patients who are given a first-time diagnosis of cancer. In 2006, approximately 204,000 cases of cancer were identified following presentation to EDs in the United States (US) with 187,000 malignant neoplasms requiring hospital admission.2 Although there are limited studies that explore the prevalence of cancer diagnosis in US emergency departments, it may be a common phenomenon globally as it has been reported that in England approximately 13%e24% of all cancers are diagnosed through an emergency route.3e6 While these studies highlight that cancer patients are commonly diagnosed following presentation to the ED, the
* Corresponding author. 1300 Morris Park Avenue, Block Building Room 112, New York, NY 10461, USA. E-mail address: hin@montefiore.org (H. In).
significance of this finding is that they may also have more advanced disease and worse outcomes than those who initially present in non-emergency settings.7e13 An analysis of all cancer cases registered in England between 2006 and 2008 found that every tumor type examined had a significantly lower 1-year relative survival rate for emergency presentations as compared to nonemergency presentations with a magnitude of difference in survival typically in the range of 20e40%.4 Similarly poor results were seen in one study that focused on all gastric cancer cases managed at a single center. Patients given a diagnosis following ED presentation were more likely to present as stage IV cancers (45% vs. 25.42%, p < 0.005) and have worse one-year (48.3% vs. 63.4%, p < 0.05) and two-year overall survival (25% vs. 67.44%, p < 0.001) following surgical intervention.14 Nationally, it was found that 12% of gastric cancer surgeries are performed following an emergency department admission, which also results in worse outcomes.15 In order to better understand the nuanced role of the emergency department in these poor outcomes, further research is needed that specifically explores the impact of location of diagnosis on
https://doi.org/10.1016/j.amjsurg.2017.10.030 0002-9610/© 2017 Elsevier Inc. All rights reserved.
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030
2
I. Solsky et al. / The American Journal of Surgery xxx (2017) 1e7
subsequent treatment and outcomes for gastric cancer. Therefore, our objective was to provide a comprehensive analysis of gastric cancer patients diagnosed in a US ED in order to obtain insight into interventions that could potentially improve gastric cancer patient survival. Our study was designed to overcome some of the limitations of administrative data sets, including the underreporting of number of ED visits, difficulty in accurately capturing presenting symptoms and treatment intent (curative versus palliative). Our study aimed to 1) determine the percent of gastric cancer patients diagnosed following a presentation to the emergency department (EDdx) compared to those diagnosed in other locations (non-EDdx), 2) examine differences in presenting symptoms, and 3) examine differences in outcomes and treatment for these patients. 2. Methods 2.1. Database and patient selection Data for all gastric adenocarcinoma patients diagnosed and treated at Montefiore Medical Center over five years (2009e2013) was analyzed. Montefiore Medical Center is a large urban medical center in New York City that primarily serves the residents of the Bronx, one of the most socioeconomically challenged counties in the US.16 In 2009, Montefiore had 93,752 hospital discharges, 303,342 emergency room visits, and 1,970,920 visits to its ambulatory care centers.17 Patients were included if they were diagnosed with gastric cancer and were older than 18 years of age. Patients were excluded if their gastric cancer diagnosis was not their first cancer diagnosis, if they had stage 0 cancer, or if they had a gastric cancer diagnosis other than adenocarcinoma. The final analysis was limited to patients who underwent elective gastric cancer surgery; hence patients were excluded if they underwent an emergency gastric cancer operation on the same admission, which is known to be associated with poor outcomes.14,15 We considered this to be an understudied cohort in which it would be meaningful to examine how an ED presentation leading to the diagnosis of cancer impacted subsequent care received (Fig. 1).
review, data on patients' primary presenting symptom (pain, bleeding/anemia, GI symptoms: nausea, vomiting, dyspepsia, other) were collected, as well as treatment information (type of initial treatment, receipt of definitive cancer surgery, receipt of treatment with curative intent). The value of performing a chart review is that it allowed for the ED group to be more accurately identified. For example, some patients received a cancer diagnosis following presentation to an ED outside the institution and this would not be identified using administrative data alone. 2.4. Statistical analysis Outcomes of interest included receipt of definitive surgery, receipt of treatment with curative intent, cancer recurrence, and survival. Survival distribution for mortality was estimated using the Kaplan-Meier method. Cox proportional hazards regression was used to estimate the Hazard Ratio (HR) for mortality given a GC diagnosis in the ED after adjusting for relevant confounding variables. A logistic regression model with backwards selection was also used to determine predictors for receipt of curative intent surgery. Patient and cancer-specific variables and symptoms were compared between the two groups. Statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC). Statistically significant p-values were set at 0.05. 3. Results 3.1. Demographics Fifty-two percent of 263 patients identified with gastric cancer received their diagnosis following presentation to the ED. Ten patients who required surgery on the same admission after EDdx
2.2. Patient-specific variables, cancer-specific variables, and symptoms Patient specific covariates included sex, age (18e50, 50e65, 65e80, 80), race (non-Hispanic white, non-Hispanic black, Hispanic, other), and insurance (private, Medicare, Medicaid, other). Cancer-specific covariates included AJCC Stage and grade (III: well or moderately differentiated, III: poorly differentiated, cell type not determined). These variables were collected from the Montefiore Medical Center's Cancer Registry and combined with other pertinent demographic and clinic data obtained from the Montefiore clinical data repository, which is maintained by the Albert Einstein College of Medicine Department of Epidemiology's Study Management Unit. Comorbidity categories were created based on the Elixhauser Comorbidity Software, which uses ICD-9CM and DRG/MS-DRG to create 29 Elixhauser comorbidity categories and is available for public use on the Healthcare Cost and Utilization Project website.18 After excluding cancer and cancer metastasis as possible comorbidity categories, patients were then categorized as having 0, 1, 2, or 3 or more comorbidities. 2.3. Chart review A chart review was performed by the lead author to determine whether patients received their gastric cancer diagnosis following presentation to an ED or in a non-ED setting. By means of this chart
Fig. 1. Inclusion and exclusion criteria.
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030
I. Solsky et al. / The American Journal of Surgery xxx (2017) 1e7 Table 1 Characteristics of ED and non-ED diagnosed patients.
Age
Sex Race
Insurance
Stage
Grade
Comorbidities
Primary symptom
18e50 50e65 65e80 80 Female Male Non-Hispanic White Non-Hispanic Black Hispanic Other Private Medicare Medicaid Other 1 2 3 4 Unknown Well or moderately differentiated Poorly differentiated Cell type not determined 0 1 2 3 Bleeding (melena, emesis, etc.) Anemia (weakness/fatigue, low Hg) Pain Incidental finding Weight loss Surveillance Nausea/vomiting Dyspepsia Unknown/other
EDdx (%) (n ¼ 136)
non-EDdx (%) (n ¼ 127)
p-value
9.6 26.5 34.6 29.4 41.9 58.1 14.0 37.5 42.7 5.9 17.7 57.4 22.1 2.9 15.4 9.6 8.8 50.0 16.2 17.7
10.2 24.4 48.8 16.5 50.4 49.6 15.8 32.3 39.4 12.6 24.4 51.2 20.5 3.9 42.5 13.4 10.2 23.6 10.2 28.4
<0.05
61.0 21.3 2.9 9.6 19.9 67.7 21.3
52.8 18.9 7.1 22.8 23.6 46.5 4.7
19.9
12.6
27.9 2.2 11.0 0.0 1.5 4.4 11.8
31.5 3.9 9.5 5.5 1.6 21.3 9.5
3
be older than 80 (29% vs. 17%, p < 0.05), but were of similar race and insurance status. They also presented with later cancer stages (stage IV: 50% vs. 24%, p < 0.05) and more comorbidities (3: 68% vs. 47%). EDdx patients were more likely to present with a primary symptom of bleeding (21% vs. 5%), anemia (20% vs. 13%), or weight loss (11% vs. 10%) but were less likely to present with dyspepsia (4% vs 21%) or pain (28% vs. 32%).
0.2
3.2. Survival
0.3
Unadjusted survival can be seen in Fig. 2. Statistically significant shorter survival time was seen in the EDdx group. EDdx patients had a median survival of 12 months versus 48 months for nonEDdx patients. The trend for decreased survival in the EDdx group remained following subset analysis when groups were stratified by stage (Fig. 3), reaching statistical significance for stages I and IV. This suggests that stage alone cannot account for the decreased survival seen in the EDdx group.
0.5
<0.05
0.1
<0.05
<0.05
3.3. Cox proportional hazards regression model using backwards elimination Table 2 summarizes the findings of the regression model of factors affecting survival. After adjusting for patient and tumor factors, GC diagnosis after ED presentation was associated with an increased mortality risk (aHR 1.9; 95% CI: 1.2e2.9). Age 18e50 (aHR 2.3; 95% CI: 1.2e4.2) or 80 (aHR 2.6; 95% CI: 1.6e4.3), male gender (aHR 1.6; 95% CI: 1.0e2.3), and higher cancer stage (Stage 4: aHR 4.0, 95% CI 2.3e7.0) were also associated with an increased mortality risk. 3.4. Treatments and outcomes for stage I-III gastric cancer
were excluded from analysis. Patient characteristics, including symptoms and cancer-specific characteristics, are summarized in Table 1. Compared to non-EDdx, EDdx patients were more likely to
A subset analysis of the treatment and outcomes for stage I-III gastric cancer patients can be seen in Table 3. Stage IV cancers were excluded from this analysis as they could not receive curativeintent treatment. Although not reaching statistical significance, EDdx patients were less likely to be initiated on cancer treatment with curative intent (63% vs. 77%) and less likely to receive definitive cancer surgery (59% vs. 74%). In a logistic regression model created from backward selection to determine predictors of receiving curative intent surgery, no variable reached statistical
Fig. 2. Survival for EDdx and non-EDdx.
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030
4
I. Solsky et al. / The American Journal of Surgery xxx (2017) 1e7
Fig. 3. Survival for EDdx and non-EDdx stratified by stage.
Table 2 Cox regression model for mortality (Backwards selection).
Location of diagnosis Age
Gender Stage
Non-EDdx EDdx 18 - 50 50e65 65e80 80 Female Male 1 2 3 4
Hazard ratio (95% CI)
p-value
Ref 1.9 (1.2e2.9) 2.4 (1.3e4.7) Ref 1.3 (0.8e2.1) 2.7 (1.5e4.8) Ref 1.6 (1.0e2.3) Ref 0.7 (0.3e1.7) 1.5 (0.7e3.1) 4.0 (2.3e7.0)
Ref <0.05 <0.05 Ref 0.4 <0.05 Ref <0.05 Ref 0.4 0.3 <0.05
Table 3 Treatment and outcomes for stage I-III gastric cancer patients.
Treatment Intent Definitive surgery
Non-curative intent Curative intent No Yes
EDdx (%) (n ¼ 46)
non-EDdx (%) (n ¼ 84)
p-value
37.0 63.0 41.3 58.7
22.6 77.4 26.2 73.8
0.2 0.2
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030
I. Solsky et al. / The American Journal of Surgery xxx (2017) 1e7
5
Table 4 Logistic regression model (backward selection) for curative intent surgery, limited to stage I-III gastric cancer patients (n ¼ 130).
Location of diagnosis Race
Non-EDdx EDdx Non-Hispanic White Non-Hispanic Black Hispanic Other
significance. However, the EDdx group trended towards having half the odds of receiving curative intent surgery as compared to the non-EDdx group (OR: 0.5, 95% CI: 0.2e1.1, p ¼ 0.07) (Table 4). 4. Discussion From these results, there are two striking findings: 1) over half of all gastric cancer patients at a single urban institution were diagnosed with gastric cancer following presentation to an ED; 2) these patients had nearly a two-fold increased mortality risk. We believe that ED diagnosis can be considered to be a marker of poor outcomes for gastric cancer. The worse outcomes and higher mortality seen in the ED group can be partially explained by the higher percentage of stage IV cancers in the ED group. Our finding that 50% of the gastric cancers in the ED group were stage IV was similar to a study by Vases et al. that reported that almost half of gastric cancer patients that present as emergencies have stage IV disease (45%).14 This explanation alone appears to be inadequate, as noticeable survival differences persisted even after patients were stratified by stage, suggesting that stage alone cannot explain this difference. It was notable that the separation of survival curves was more pronounced for stage I than stages II & III. To explore what may be driving this difference, we performed an exploratory analysis of patient characteristics stratified by stage I vs. stages II & III. We found that stage I patients had greater separation of patient characteristics including 2 comorbidities (20.1% vs. 6%), nonprivate insurance (11.6% vs. 7.3%), and race other than nonHispanic White (9% vs. 7.3%) between ED and non-ED patients compared to stage II & III patients. This suggests that the diagnosis of early cancer in the ED may be driven more by patient characteristics than for later stages of cancer (Appendix 1). With these findings, we believe that there are three possible explanations for why so many gastric cancers are being diagnosed in the ED as well as why an ED diagnosis is associated with a higher mortality risk: 1) the patients who present to the ED for evaluation have different characteristics than those who go to other locations such as ambulatory primary care centers; 2) the non-specific symptoms of gastric cancer do not lend themselves well to an early and non-ED diagnosis; and 3) initial management in the ED results in poor cancer coordination related to systems issues that result in less than ideal care. While a study of initial diagnosis of all cancers at a single safetynet hospital in Florida found similarly that a high proportion of all cancer diagnoses occur via the ED (32%) and are associated with an increased mortality risk (176% higher risk of dying during the 2year study period), the study also found that African Americans and impoverished residents were disproportionately affected.19 The high prevalence of ED diagnosis may be related to issues related to healthcare disparities, access, and literacy with patients relying on the ED for services typically attained through a primary care physician. Other studies have found that cancer patients generally considered at higher risk such as those of older age,19,20 female gender,20,21 and low socioeconomic status6,19e22 were more likely to present emergently via the ED. However, in our study's total
Odds ratio (95% CI)
p-value
Ref 0.5 (0.2e1.1) Ref 1.9 (0.6e5.9) 1.3 (0.4e3.7) 0.2 (0.1e1.1)
Ref 0.07 Ref 0.29 0.67 0.06
cohort, both ED and non-ED groups were of similar race and insurance status. Interestingly, although not reaching statistical significance, non-Hispanic black or Hispanic patients had higher odds or receiving curative intent treatment as compared to white patients. Thus, our findings suggest that the poor outcomes associated with GC diagnosis following ED presentation likely cannot be accounted for based on patient characteristics alone. Our study shows that the symptoms of gastric cancer are varied and can often be non-specific. Although pain was the primary symptom in nearly a third of the gastric cancer patients in our study and bleeding occurred in 21% of the EDdx patients, many patients presented with less dramatic symptoms such as dyspepsia (21% of non-EDdx patients) and anemia-related weakness/fatigue (20% of EDdx patients). Where patients present with these symptoms may dictate their subsequent management. The non-specific symptoms of gastric cancer may not prompt a primary care provider to initiate a cancer work-up but they may prompt an ED physician to obtain a CT scan that could lead to a diagnosis. Unfortunately, by the time patients are symptomatic, they usually have advanced disease. Unlike countries such as Korea and Japan that have national screening programs designed to detect gastric cancer at early, asymptomatic stages,23,24 the US lacks such programs. This may result in a higher frequency of symptomatic gastric cancer presentations prompting ED visits and worse outcomes. Although the feasibility of a US national gastric cancer screening program is debated, the literature recognizes the importance of selective screening programs for high-risk individuals such as immigrants from regions associated with a high risk of gastric cancer and those who have a family history of gastric cancer.24 Continued research is needed to assess the effects of selective screening on mortality in Western countries. The ED is a poor location to coordinate complex cancer care that often relies on management by multiple specialists. It is possible that an ED diagnosis may cause delays in referral to appropriate specialists, or even for patients to be lost to follow-up completely. Furthermore, as gastric cancer care continues to evolve with neoadjuvant therapy believed to offer benefit for certain patients,25,26 it is important that appropriate referral pathways are established to ensure that patients are considered for neoadjuvant therapy before they are referred for surgery. In order to identify potentially curable early tumors in asymptomatic patients, screening programs that do not rely on symptoms should be developed. ED presentation should be used as a red-flag to signal newly diagnosed gastric cancer patients who will be in greater need of social services throughout their care and gastric cancer education for providers should be improved so that appropriate referrals are made. Although this study is an important first step in identifying a potentially modifiable risk factor for gastric cancer patients, it is not without its limitations. The primary limitation of this study is that it is a single center study and its applicability to other healthcare systems is unknown. The relatively homogenous population that a single urban hospital services may account for the lack of differences in socioeconomic status that were observed in this study.
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030
6
I. Solsky et al. / The American Journal of Surgery xxx (2017) 1e7
Furthermore, in collecting data from a single center, the patient population studied is relatively small, which may have contributed to the lack of statistical difference in outcomes and treatment for stage I to III cancers. Lastly, our study relied on chart review, which is a method that can be influenced by abstractor bias and information bias.
product mentioned or concept discussed in this article. Conflict of interest The authors report no proprietary or commercial interest in any product mentioned or concept discussed in this article. The authors have no conflicts of interest to report.
5. Conclusion The results of this study indicate that the ED is a common point
Appendix 1. Characteristics of ED and non-ED diagnosed patients stratified by stage I vs. stage II & III
Stage I (n ¼ 75)
Age
Sex Race
Insurance
Comorbidities
18e50 50e65 65e80 80 Female Male Non-Hispanic White Non-Hispanic Black Hispanic Other Private Medicare Medicaid Other 0 1 2 3
Stage II & III (n ¼ 55)
EDdx (%)
Non-EDdx (%)
D%*
EDdx (%)
Non-EDdx (%)
D%*
4.8 14.3 47.6 33.3 38.1 61.9 9.5 33.3 47.6 9.5 14.3 76.2 9.5 0.0 9.5 0.0 19.1 71.4
11.1 27.8 42.6 18.5 44.4 55.6 18.5 22.2 44.4 14.8 25.9 48.2 18.5 7.4 7.4 22.2 16.7 53.7
6.3 13.5 5.0 14.8 6.3 6.3 9.0 11.1 3.2 5.3 11.6 28.0 9.0 7.4 2.1 22.2 2.4 17.7
8.0 28.0 32.0 32.0 24.0 76.0 24.0 40.0 32.0 4.0 16.0 52.0 24.0 8.0 4.0 20.0 16.0 60.0
3.3 23.3 60.0 13.3 56.7 43.3 16.7 40.0 36.7 6.7 23.3 56.7 16.7 3.3 10.0 20.0 20.0 50.0
4.7 4.7 28.0 18.7 32.7 32.7 7.3 0.0 4.7 2.7 7.3 4.7 7.3 4.7 6.0 0.0 4.0 10.0
* D%: EDdx non-EDdx.
of entry for the diagnosis of gastric cancer and is independently associated with increased mortality risk for these patients. These findings may be related to unaccounted patient-level factors such as poor health behaviors or systems-level factors such as lack of patient navigation or multi-disciplinary cancer care. The varied and often non-specific symptoms of gastric cancer underscore the need to develop cancer detection methods that do not rely on symptoms. While population-based mass screening in the US is not feasible for this low incidence disease, developing methods to identify highrisk individuals for a targeted screening program is imperative to promote early detection of gastric cancer and reduce poor outcomes. Author contributions Analysis and interpretation of data e HI, BR, PF, PM, IS. Editing and drafting of the manuscript - ALL. Conception and design of study, and/or acquisition of data e HI, IS, KW. All authors gave final approval for the manuscript to be published. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Disclosure The authors report no proprietary or commercial interest in any
References 1. Hargarten SW, Richards MJ, Anderson AJ. Cancer presentation in the emergency department: a failure of primary care. Am J Emerg Med. 1992;10: 290e293. 2. Pitts SR, Niska RW, Xu J, Burt CW. National Hospital Ambulatory Medical Care Survey: 2006 Emergency Department Summary (Rep. No. 7); 2008 [On-line]. Available http://www.cdc.gov/nchs/data/nhsr/nhsr007.pdf. 3. National Cancer Intelligence Network. Routes to Diagnosis-data Briefing. London: National Cancer Intelligence Network; 2010. 4. Elliss-Brookes L, McPhail S, Ives A, et al. Routes to diagnosis for cancerdetermining the patient journey using multiple routine data sets. Br J Cancer. 2012;107(8):1220e1226. 5. Bottle A, Tsang C, Parsons C, Majeed A, Soljak M, Aylin P. Association between patient and general practice characteristics and unplanned first-time admissions for cancer: observational study. Br J Cancer. 2012;107(8):1213e1219. 6. Tsang, et al. Cancer diagnosed by emergency admission in England: an observational study using the general practice research database. BMC Health Serv Res. 2013;13:308. 7. Oliphant R, Mansouri D, Nicholson GA, McMillan DC, Horgan PG, Morrison DS. Emergency presentation of node-negative colorectal cancer treated with curative surgery is associated with poorer short and longer-term survival. Int J Colorectal Dis. 2014 May;29(5):591e598. 8. Mallol M, Sabate A, Dalmau A, Koo M. Risk factors and mortality after elective and emergent laparatomies for oncological procedures in 899 patients in the intensive care unit: a retrospective observational cohort study. Patient Saf Surg. 2013 Sep;7(1):29. 9. Tataru D, Jack RH, Lind MJ, Moller H, Luchtenborg M. The effect of emergency presentation on surgery and survival in lung cancer patients in England, 20062008. Cancer Epidemiol. 2015 May. Epub. 10. Porta M, Fernandez E, Belloc J, Malats N, Gallen M, Alonso J. Emergency admission for cancer: a matter of survival? Br J Cancer. 1998;77(3):477e484. 11. McPhail S, Elliss-Brookes L, Shelton J, et al. Emergency presentation of cancer and short-term mortality. Br J Cancer. 2013;109(8):2027e2034. 12. Kim J, Mittal R, Konyalian V, King J, Stamos MJ, Kumar RR. Outcome analysis of patients undergoing colorectal resection for emergent and elective indications. Am Surg. 2007;73(10):991e993.
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030
I. Solsky et al. / The American Journal of Surgery xxx (2017) 1e7 13. Amri R, Bordelanou LG, Sylla P, Berger DL. Colon cancer surgery following emergency presentation: effects on admission and stage-adjusted outcomes. Am J Surg. 2015;209(2):246e253. 14. Vasas P, Wiggins T, Chaudry A, Bryant C, Hughes FS. Emergency presentation of the gastric cancer; prognosis and implications for service planning. World J Emerg Surg. 2012 Sep;7(1):31. 15. Solsky I, Friedmann F, Muscarella P, In H. Poor outcomes of gastric cancer surgery after admission through the emergency department. Ann Surg Oncol. 2016 Dec;24(5):1180e1187. 16. Foreman S. Montefiore Medical Center in the Bronx, New York: improving health in an urban community. Acad Med. 2004;79(12):1154e1161. 17. Bellin E, Fletcher D, Gerberer N, Islam S, Srivastava N. Democratizing information creation from health care data for quality improvement, research, and education-the Montefiore Medical Center Experience. Acad Med. 2010;85(8): 1362e1368. 18. Elixhauser et al. Healthcare Cost and Utilization Project (HCUP) Elixhauser Comorbidity Software. Agency for Healthcare Research and Quality, Rockville, MD. Available at:: www.hcup-us.ahrq.gov/toolssoftware/comorbidity/ comorbidity.jsp. Accessed June 2016. 19. Livingwood WC, Smotherman C, Lukens-Bull K, et al. An elephant in the emergency department: symptom of dispariteis in cancer care. Popul Health Manag. 2016 Jan 13 (Epub ahead of print). 20. Healthcare Cost and Utilization Project. Overview of the Nationwide Inpatient
21.
22.
23.
24.
25.
26.
7
Sample (NIS).. Accessed December 16 2015. Cauley CE, Panizales MT, Reznor G, et al. Outcomes after emergency abdominal surgery in patients with advanced cancer: opportunities to reduce complications and improve palliative care. J Trauma Acute Care Surg. 2015;79(3): 399e406. Raine R, Wong W, Scholes S, Ashton C, Obichere A, Ambler G. Social variations in access to hospital care for patients with colorectal, breast, and lung cancer between 1999 and 2006: retrospective analysis of hospital episode statistics. BMJ. 2010;340:b5479. Kim YG, Kong SH, Oh SY, et al. Effects of screening on gastric cancer management: comparative analysis of the results in 2006 and in 2011. J Gastric Cancer. 2014;14:129e134. Kim GH, Liang PS, Bang SJ, Hwang JH. Screening and surveillance for gastric cancer in the United States: is it Needed? Gastrointest Endosc. 2016 (Epub ahead of print). Cunningham D, Allum WH, Stenning SP, et al. Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer. N Engl J Med. 2006;355:11. Ychou M, Boige V, Pignon JP, et al. Perioperative chemotherapy compared with surgery alone for retractable gastroesophageal adenocarcinoma: an FNCLCC and FFCD multicenter phase III trial. J Clin Oncol. 2011;29:1715.
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030
Contents lists available at ScienceDirect
The American Journal of Surgery journal homepage: www.americanjournalofsurgery.com
Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes Ian Solsky a, Bruce Rapkin b, Kristen Wong a, Patricia Friedmann a, Peter Muscarella a, Haejin In a, * a b
Montefiore Medical Center/Albert Einstein College of Medicine, Department of Surgery, Bronx, NY, USA Montefiore Medical Center/Albert Einstein College of Medicine, Department of Epidemiology and Population Health, Bronx, NY, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 19 August 2017 Received in revised form 4 October 2017 Accepted 10 October 2017
Background: The impact of diagnosis location on gastric cancer (GC) outcomes is poorly defined. Methods: Detailed chart review was conducted to identify presenting location leading to diagnosis and treatment for GC patients at a single institution (2009e2013). Patients treated non-emergently following a diagnosis prompted by an ED visit (EDdx) were compared with those diagnosed at other locations (non-EDdx). Results: EDdx patients comprised 52% of 263 GC patients. They were older, had later cancer stages (stage IV: 50% vs. 24%), more comorbidities (3: 68% vs. 47%), and presented with non-specific symptoms like bleeding (21% vs. 5%). Both groups were of similar race and insurance status. In a model adjusted for stage, EDdx was associated with increased mortality (aHR 1.9; 95% CI: 1.2e2.9). Conclusion: Half of GC patients had an ED visit prompting diagnosis, which is independently associated with increased mortality. Efforts should focus on reducing EDdx rates to improve GC outcomes. Summary: Demographics, treatment, and outcomes of gastric adenocarcinoma patients at a single center whose diagnosis was prompted by an emergency department (ED) visit were compared to that of patients diagnosed in non-ED settings. 52% of patients had an ED visit prompting their diagnosis, which was associated with an increased mortality risk. © 2017 Elsevier Inc. All rights reserved.
Meeting Presentation: Results in this manuscript were initially presented at the 69th Society of Surgical Oncology (SSO) Annual Cancer Symposium, March 2e5, 2016 in Boston, Massachusetts, USA. Keywords: Gastric cancer Emergency department Cancer diagnosis
1. Introduction Although emergency departments (ED) are intended to manage acute medical problems,1 they frequently serve as the entry point into the healthcare system for patients who are given a first-time diagnosis of cancer. In 2006, approximately 204,000 cases of cancer were identified following presentation to EDs in the United States (US) with 187,000 malignant neoplasms requiring hospital admission.2 Although there are limited studies that explore the prevalence of cancer diagnosis in US emergency departments, it may be a common phenomenon globally as it has been reported that in England approximately 13%e24% of all cancers are diagnosed through an emergency route.3e6 While these studies highlight that cancer patients are commonly diagnosed following presentation to the ED, the
* Corresponding author. 1300 Morris Park Avenue, Block Building Room 112, New York, NY 10461, USA. E-mail address: hin@montefiore.org (H. In).
significance of this finding is that they may also have more advanced disease and worse outcomes than those who initially present in non-emergency settings.7e13 An analysis of all cancer cases registered in England between 2006 and 2008 found that every tumor type examined had a significantly lower 1-year relative survival rate for emergency presentations as compared to nonemergency presentations with a magnitude of difference in survival typically in the range of 20e40%.4 Similarly poor results were seen in one study that focused on all gastric cancer cases managed at a single center. Patients given a diagnosis following ED presentation were more likely to present as stage IV cancers (45% vs. 25.42%, p < 0.005) and have worse one-year (48.3% vs. 63.4%, p < 0.05) and two-year overall survival (25% vs. 67.44%, p < 0.001) following surgical intervention.14 Nationally, it was found that 12% of gastric cancer surgeries are performed following an emergency department admission, which also results in worse outcomes.15 In order to better understand the nuanced role of the emergency department in these poor outcomes, further research is needed that specifically explores the impact of location of diagnosis on
https://doi.org/10.1016/j.amjsurg.2017.10.030 0002-9610/© 2017 Elsevier Inc. All rights reserved.
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030
2
I. Solsky et al. / The American Journal of Surgery xxx (2017) 1e7
subsequent treatment and outcomes for gastric cancer. Therefore, our objective was to provide a comprehensive analysis of gastric cancer patients diagnosed in a US ED in order to obtain insight into interventions that could potentially improve gastric cancer patient survival. Our study was designed to overcome some of the limitations of administrative data sets, including the underreporting of number of ED visits, difficulty in accurately capturing presenting symptoms and treatment intent (curative versus palliative). Our study aimed to 1) determine the percent of gastric cancer patients diagnosed following a presentation to the emergency department (EDdx) compared to those diagnosed in other locations (non-EDdx), 2) examine differences in presenting symptoms, and 3) examine differences in outcomes and treatment for these patients. 2. Methods 2.1. Database and patient selection Data for all gastric adenocarcinoma patients diagnosed and treated at Montefiore Medical Center over five years (2009e2013) was analyzed. Montefiore Medical Center is a large urban medical center in New York City that primarily serves the residents of the Bronx, one of the most socioeconomically challenged counties in the US.16 In 2009, Montefiore had 93,752 hospital discharges, 303,342 emergency room visits, and 1,970,920 visits to its ambulatory care centers.17 Patients were included if they were diagnosed with gastric cancer and were older than 18 years of age. Patients were excluded if their gastric cancer diagnosis was not their first cancer diagnosis, if they had stage 0 cancer, or if they had a gastric cancer diagnosis other than adenocarcinoma. The final analysis was limited to patients who underwent elective gastric cancer surgery; hence patients were excluded if they underwent an emergency gastric cancer operation on the same admission, which is known to be associated with poor outcomes.14,15 We considered this to be an understudied cohort in which it would be meaningful to examine how an ED presentation leading to the diagnosis of cancer impacted subsequent care received (Fig. 1).
review, data on patients' primary presenting symptom (pain, bleeding/anemia, GI symptoms: nausea, vomiting, dyspepsia, other) were collected, as well as treatment information (type of initial treatment, receipt of definitive cancer surgery, receipt of treatment with curative intent). The value of performing a chart review is that it allowed for the ED group to be more accurately identified. For example, some patients received a cancer diagnosis following presentation to an ED outside the institution and this would not be identified using administrative data alone. 2.4. Statistical analysis Outcomes of interest included receipt of definitive surgery, receipt of treatment with curative intent, cancer recurrence, and survival. Survival distribution for mortality was estimated using the Kaplan-Meier method. Cox proportional hazards regression was used to estimate the Hazard Ratio (HR) for mortality given a GC diagnosis in the ED after adjusting for relevant confounding variables. A logistic regression model with backwards selection was also used to determine predictors for receipt of curative intent surgery. Patient and cancer-specific variables and symptoms were compared between the two groups. Statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC). Statistically significant p-values were set at 0.05. 3. Results 3.1. Demographics Fifty-two percent of 263 patients identified with gastric cancer received their diagnosis following presentation to the ED. Ten patients who required surgery on the same admission after EDdx
2.2. Patient-specific variables, cancer-specific variables, and symptoms Patient specific covariates included sex, age (18e50, 50e65, 65e80, 80), race (non-Hispanic white, non-Hispanic black, Hispanic, other), and insurance (private, Medicare, Medicaid, other). Cancer-specific covariates included AJCC Stage and grade (III: well or moderately differentiated, III: poorly differentiated, cell type not determined). These variables were collected from the Montefiore Medical Center's Cancer Registry and combined with other pertinent demographic and clinic data obtained from the Montefiore clinical data repository, which is maintained by the Albert Einstein College of Medicine Department of Epidemiology's Study Management Unit. Comorbidity categories were created based on the Elixhauser Comorbidity Software, which uses ICD-9CM and DRG/MS-DRG to create 29 Elixhauser comorbidity categories and is available for public use on the Healthcare Cost and Utilization Project website.18 After excluding cancer and cancer metastasis as possible comorbidity categories, patients were then categorized as having 0, 1, 2, or 3 or more comorbidities. 2.3. Chart review A chart review was performed by the lead author to determine whether patients received their gastric cancer diagnosis following presentation to an ED or in a non-ED setting. By means of this chart
Fig. 1. Inclusion and exclusion criteria.
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030
I. Solsky et al. / The American Journal of Surgery xxx (2017) 1e7 Table 1 Characteristics of ED and non-ED diagnosed patients.
Age
Sex Race
Insurance
Stage
Grade
Comorbidities
Primary symptom
18e50 50e65 65e80 80 Female Male Non-Hispanic White Non-Hispanic Black Hispanic Other Private Medicare Medicaid Other 1 2 3 4 Unknown Well or moderately differentiated Poorly differentiated Cell type not determined 0 1 2 3 Bleeding (melena, emesis, etc.) Anemia (weakness/fatigue, low Hg) Pain Incidental finding Weight loss Surveillance Nausea/vomiting Dyspepsia Unknown/other
EDdx (%) (n ¼ 136)
non-EDdx (%) (n ¼ 127)
p-value
9.6 26.5 34.6 29.4 41.9 58.1 14.0 37.5 42.7 5.9 17.7 57.4 22.1 2.9 15.4 9.6 8.8 50.0 16.2 17.7
10.2 24.4 48.8 16.5 50.4 49.6 15.8 32.3 39.4 12.6 24.4 51.2 20.5 3.9 42.5 13.4 10.2 23.6 10.2 28.4
<0.05
61.0 21.3 2.9 9.6 19.9 67.7 21.3
52.8 18.9 7.1 22.8 23.6 46.5 4.7
19.9
12.6
27.9 2.2 11.0 0.0 1.5 4.4 11.8
31.5 3.9 9.5 5.5 1.6 21.3 9.5
3
be older than 80 (29% vs. 17%, p < 0.05), but were of similar race and insurance status. They also presented with later cancer stages (stage IV: 50% vs. 24%, p < 0.05) and more comorbidities (3: 68% vs. 47%). EDdx patients were more likely to present with a primary symptom of bleeding (21% vs. 5%), anemia (20% vs. 13%), or weight loss (11% vs. 10%) but were less likely to present with dyspepsia (4% vs 21%) or pain (28% vs. 32%).
0.2
3.2. Survival
0.3
Unadjusted survival can be seen in Fig. 2. Statistically significant shorter survival time was seen in the EDdx group. EDdx patients had a median survival of 12 months versus 48 months for nonEDdx patients. The trend for decreased survival in the EDdx group remained following subset analysis when groups were stratified by stage (Fig. 3), reaching statistical significance for stages I and IV. This suggests that stage alone cannot account for the decreased survival seen in the EDdx group.
0.5
<0.05
0.1
<0.05
<0.05
3.3. Cox proportional hazards regression model using backwards elimination Table 2 summarizes the findings of the regression model of factors affecting survival. After adjusting for patient and tumor factors, GC diagnosis after ED presentation was associated with an increased mortality risk (aHR 1.9; 95% CI: 1.2e2.9). Age 18e50 (aHR 2.3; 95% CI: 1.2e4.2) or 80 (aHR 2.6; 95% CI: 1.6e4.3), male gender (aHR 1.6; 95% CI: 1.0e2.3), and higher cancer stage (Stage 4: aHR 4.0, 95% CI 2.3e7.0) were also associated with an increased mortality risk. 3.4. Treatments and outcomes for stage I-III gastric cancer
were excluded from analysis. Patient characteristics, including symptoms and cancer-specific characteristics, are summarized in Table 1. Compared to non-EDdx, EDdx patients were more likely to
A subset analysis of the treatment and outcomes for stage I-III gastric cancer patients can be seen in Table 3. Stage IV cancers were excluded from this analysis as they could not receive curativeintent treatment. Although not reaching statistical significance, EDdx patients were less likely to be initiated on cancer treatment with curative intent (63% vs. 77%) and less likely to receive definitive cancer surgery (59% vs. 74%). In a logistic regression model created from backward selection to determine predictors of receiving curative intent surgery, no variable reached statistical
Fig. 2. Survival for EDdx and non-EDdx.
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030
4
I. Solsky et al. / The American Journal of Surgery xxx (2017) 1e7
Fig. 3. Survival for EDdx and non-EDdx stratified by stage.
Table 2 Cox regression model for mortality (Backwards selection).
Location of diagnosis Age
Gender Stage
Non-EDdx EDdx 18 - 50 50e65 65e80 80 Female Male 1 2 3 4
Hazard ratio (95% CI)
p-value
Ref 1.9 (1.2e2.9) 2.4 (1.3e4.7) Ref 1.3 (0.8e2.1) 2.7 (1.5e4.8) Ref 1.6 (1.0e2.3) Ref 0.7 (0.3e1.7) 1.5 (0.7e3.1) 4.0 (2.3e7.0)
Ref <0.05 <0.05 Ref 0.4 <0.05 Ref <0.05 Ref 0.4 0.3 <0.05
Table 3 Treatment and outcomes for stage I-III gastric cancer patients.
Treatment Intent Definitive surgery
Non-curative intent Curative intent No Yes
EDdx (%) (n ¼ 46)
non-EDdx (%) (n ¼ 84)
p-value
37.0 63.0 41.3 58.7
22.6 77.4 26.2 73.8
0.2 0.2
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030
I. Solsky et al. / The American Journal of Surgery xxx (2017) 1e7
5
Table 4 Logistic regression model (backward selection) for curative intent surgery, limited to stage I-III gastric cancer patients (n ¼ 130).
Location of diagnosis Race
Non-EDdx EDdx Non-Hispanic White Non-Hispanic Black Hispanic Other
significance. However, the EDdx group trended towards having half the odds of receiving curative intent surgery as compared to the non-EDdx group (OR: 0.5, 95% CI: 0.2e1.1, p ¼ 0.07) (Table 4). 4. Discussion From these results, there are two striking findings: 1) over half of all gastric cancer patients at a single urban institution were diagnosed with gastric cancer following presentation to an ED; 2) these patients had nearly a two-fold increased mortality risk. We believe that ED diagnosis can be considered to be a marker of poor outcomes for gastric cancer. The worse outcomes and higher mortality seen in the ED group can be partially explained by the higher percentage of stage IV cancers in the ED group. Our finding that 50% of the gastric cancers in the ED group were stage IV was similar to a study by Vases et al. that reported that almost half of gastric cancer patients that present as emergencies have stage IV disease (45%).14 This explanation alone appears to be inadequate, as noticeable survival differences persisted even after patients were stratified by stage, suggesting that stage alone cannot explain this difference. It was notable that the separation of survival curves was more pronounced for stage I than stages II & III. To explore what may be driving this difference, we performed an exploratory analysis of patient characteristics stratified by stage I vs. stages II & III. We found that stage I patients had greater separation of patient characteristics including 2 comorbidities (20.1% vs. 6%), nonprivate insurance (11.6% vs. 7.3%), and race other than nonHispanic White (9% vs. 7.3%) between ED and non-ED patients compared to stage II & III patients. This suggests that the diagnosis of early cancer in the ED may be driven more by patient characteristics than for later stages of cancer (Appendix 1). With these findings, we believe that there are three possible explanations for why so many gastric cancers are being diagnosed in the ED as well as why an ED diagnosis is associated with a higher mortality risk: 1) the patients who present to the ED for evaluation have different characteristics than those who go to other locations such as ambulatory primary care centers; 2) the non-specific symptoms of gastric cancer do not lend themselves well to an early and non-ED diagnosis; and 3) initial management in the ED results in poor cancer coordination related to systems issues that result in less than ideal care. While a study of initial diagnosis of all cancers at a single safetynet hospital in Florida found similarly that a high proportion of all cancer diagnoses occur via the ED (32%) and are associated with an increased mortality risk (176% higher risk of dying during the 2year study period), the study also found that African Americans and impoverished residents were disproportionately affected.19 The high prevalence of ED diagnosis may be related to issues related to healthcare disparities, access, and literacy with patients relying on the ED for services typically attained through a primary care physician. Other studies have found that cancer patients generally considered at higher risk such as those of older age,19,20 female gender,20,21 and low socioeconomic status6,19e22 were more likely to present emergently via the ED. However, in our study's total
Odds ratio (95% CI)
p-value
Ref 0.5 (0.2e1.1) Ref 1.9 (0.6e5.9) 1.3 (0.4e3.7) 0.2 (0.1e1.1)
Ref 0.07 Ref 0.29 0.67 0.06
cohort, both ED and non-ED groups were of similar race and insurance status. Interestingly, although not reaching statistical significance, non-Hispanic black or Hispanic patients had higher odds or receiving curative intent treatment as compared to white patients. Thus, our findings suggest that the poor outcomes associated with GC diagnosis following ED presentation likely cannot be accounted for based on patient characteristics alone. Our study shows that the symptoms of gastric cancer are varied and can often be non-specific. Although pain was the primary symptom in nearly a third of the gastric cancer patients in our study and bleeding occurred in 21% of the EDdx patients, many patients presented with less dramatic symptoms such as dyspepsia (21% of non-EDdx patients) and anemia-related weakness/fatigue (20% of EDdx patients). Where patients present with these symptoms may dictate their subsequent management. The non-specific symptoms of gastric cancer may not prompt a primary care provider to initiate a cancer work-up but they may prompt an ED physician to obtain a CT scan that could lead to a diagnosis. Unfortunately, by the time patients are symptomatic, they usually have advanced disease. Unlike countries such as Korea and Japan that have national screening programs designed to detect gastric cancer at early, asymptomatic stages,23,24 the US lacks such programs. This may result in a higher frequency of symptomatic gastric cancer presentations prompting ED visits and worse outcomes. Although the feasibility of a US national gastric cancer screening program is debated, the literature recognizes the importance of selective screening programs for high-risk individuals such as immigrants from regions associated with a high risk of gastric cancer and those who have a family history of gastric cancer.24 Continued research is needed to assess the effects of selective screening on mortality in Western countries. The ED is a poor location to coordinate complex cancer care that often relies on management by multiple specialists. It is possible that an ED diagnosis may cause delays in referral to appropriate specialists, or even for patients to be lost to follow-up completely. Furthermore, as gastric cancer care continues to evolve with neoadjuvant therapy believed to offer benefit for certain patients,25,26 it is important that appropriate referral pathways are established to ensure that patients are considered for neoadjuvant therapy before they are referred for surgery. In order to identify potentially curable early tumors in asymptomatic patients, screening programs that do not rely on symptoms should be developed. ED presentation should be used as a red-flag to signal newly diagnosed gastric cancer patients who will be in greater need of social services throughout their care and gastric cancer education for providers should be improved so that appropriate referrals are made. Although this study is an important first step in identifying a potentially modifiable risk factor for gastric cancer patients, it is not without its limitations. The primary limitation of this study is that it is a single center study and its applicability to other healthcare systems is unknown. The relatively homogenous population that a single urban hospital services may account for the lack of differences in socioeconomic status that were observed in this study.
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030
6
I. Solsky et al. / The American Journal of Surgery xxx (2017) 1e7
Furthermore, in collecting data from a single center, the patient population studied is relatively small, which may have contributed to the lack of statistical difference in outcomes and treatment for stage I to III cancers. Lastly, our study relied on chart review, which is a method that can be influenced by abstractor bias and information bias.
product mentioned or concept discussed in this article. Conflict of interest The authors report no proprietary or commercial interest in any product mentioned or concept discussed in this article. The authors have no conflicts of interest to report.
5. Conclusion The results of this study indicate that the ED is a common point
Appendix 1. Characteristics of ED and non-ED diagnosed patients stratified by stage I vs. stage II & III
Stage I (n ¼ 75)
Age
Sex Race
Insurance
Comorbidities
18e50 50e65 65e80 80 Female Male Non-Hispanic White Non-Hispanic Black Hispanic Other Private Medicare Medicaid Other 0 1 2 3
Stage II & III (n ¼ 55)
EDdx (%)
Non-EDdx (%)
D%*
EDdx (%)
Non-EDdx (%)
D%*
4.8 14.3 47.6 33.3 38.1 61.9 9.5 33.3 47.6 9.5 14.3 76.2 9.5 0.0 9.5 0.0 19.1 71.4
11.1 27.8 42.6 18.5 44.4 55.6 18.5 22.2 44.4 14.8 25.9 48.2 18.5 7.4 7.4 22.2 16.7 53.7
6.3 13.5 5.0 14.8 6.3 6.3 9.0 11.1 3.2 5.3 11.6 28.0 9.0 7.4 2.1 22.2 2.4 17.7
8.0 28.0 32.0 32.0 24.0 76.0 24.0 40.0 32.0 4.0 16.0 52.0 24.0 8.0 4.0 20.0 16.0 60.0
3.3 23.3 60.0 13.3 56.7 43.3 16.7 40.0 36.7 6.7 23.3 56.7 16.7 3.3 10.0 20.0 20.0 50.0
4.7 4.7 28.0 18.7 32.7 32.7 7.3 0.0 4.7 2.7 7.3 4.7 7.3 4.7 6.0 0.0 4.0 10.0
* D%: EDdx non-EDdx.
of entry for the diagnosis of gastric cancer and is independently associated with increased mortality risk for these patients. These findings may be related to unaccounted patient-level factors such as poor health behaviors or systems-level factors such as lack of patient navigation or multi-disciplinary cancer care. The varied and often non-specific symptoms of gastric cancer underscore the need to develop cancer detection methods that do not rely on symptoms. While population-based mass screening in the US is not feasible for this low incidence disease, developing methods to identify highrisk individuals for a targeted screening program is imperative to promote early detection of gastric cancer and reduce poor outcomes. Author contributions Analysis and interpretation of data e HI, BR, PF, PM, IS. Editing and drafting of the manuscript - ALL. Conception and design of study, and/or acquisition of data e HI, IS, KW. All authors gave final approval for the manuscript to be published. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Disclosure The authors report no proprietary or commercial interest in any
References 1. Hargarten SW, Richards MJ, Anderson AJ. Cancer presentation in the emergency department: a failure of primary care. Am J Emerg Med. 1992;10: 290e293. 2. Pitts SR, Niska RW, Xu J, Burt CW. National Hospital Ambulatory Medical Care Survey: 2006 Emergency Department Summary (Rep. No. 7); 2008 [On-line]. Available http://www.cdc.gov/nchs/data/nhsr/nhsr007.pdf. 3. National Cancer Intelligence Network. Routes to Diagnosis-data Briefing. London: National Cancer Intelligence Network; 2010. 4. Elliss-Brookes L, McPhail S, Ives A, et al. Routes to diagnosis for cancerdetermining the patient journey using multiple routine data sets. Br J Cancer. 2012;107(8):1220e1226. 5. Bottle A, Tsang C, Parsons C, Majeed A, Soljak M, Aylin P. Association between patient and general practice characteristics and unplanned first-time admissions for cancer: observational study. Br J Cancer. 2012;107(8):1213e1219. 6. Tsang, et al. Cancer diagnosed by emergency admission in England: an observational study using the general practice research database. BMC Health Serv Res. 2013;13:308. 7. Oliphant R, Mansouri D, Nicholson GA, McMillan DC, Horgan PG, Morrison DS. Emergency presentation of node-negative colorectal cancer treated with curative surgery is associated with poorer short and longer-term survival. Int J Colorectal Dis. 2014 May;29(5):591e598. 8. Mallol M, Sabate A, Dalmau A, Koo M. Risk factors and mortality after elective and emergent laparatomies for oncological procedures in 899 patients in the intensive care unit: a retrospective observational cohort study. Patient Saf Surg. 2013 Sep;7(1):29. 9. Tataru D, Jack RH, Lind MJ, Moller H, Luchtenborg M. The effect of emergency presentation on surgery and survival in lung cancer patients in England, 20062008. Cancer Epidemiol. 2015 May. Epub. 10. Porta M, Fernandez E, Belloc J, Malats N, Gallen M, Alonso J. Emergency admission for cancer: a matter of survival? Br J Cancer. 1998;77(3):477e484. 11. McPhail S, Elliss-Brookes L, Shelton J, et al. Emergency presentation of cancer and short-term mortality. Br J Cancer. 2013;109(8):2027e2034. 12. Kim J, Mittal R, Konyalian V, King J, Stamos MJ, Kumar RR. Outcome analysis of patients undergoing colorectal resection for emergent and elective indications. Am Surg. 2007;73(10):991e993.
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030
I. Solsky et al. / The American Journal of Surgery xxx (2017) 1e7 13. Amri R, Bordelanou LG, Sylla P, Berger DL. Colon cancer surgery following emergency presentation: effects on admission and stage-adjusted outcomes. Am J Surg. 2015;209(2):246e253. 14. Vasas P, Wiggins T, Chaudry A, Bryant C, Hughes FS. Emergency presentation of the gastric cancer; prognosis and implications for service planning. World J Emerg Surg. 2012 Sep;7(1):31. 15. Solsky I, Friedmann F, Muscarella P, In H. Poor outcomes of gastric cancer surgery after admission through the emergency department. Ann Surg Oncol. 2016 Dec;24(5):1180e1187. 16. Foreman S. Montefiore Medical Center in the Bronx, New York: improving health in an urban community. Acad Med. 2004;79(12):1154e1161. 17. Bellin E, Fletcher D, Gerberer N, Islam S, Srivastava N. Democratizing information creation from health care data for quality improvement, research, and education-the Montefiore Medical Center Experience. Acad Med. 2010;85(8): 1362e1368. 18. Elixhauser et al. Healthcare Cost and Utilization Project (HCUP) Elixhauser Comorbidity Software. Agency for Healthcare Research and Quality, Rockville, MD. Available at:: www.hcup-us.ahrq.gov/toolssoftware/comorbidity/ comorbidity.jsp. Accessed June 2016. 19. Livingwood WC, Smotherman C, Lukens-Bull K, et al. An elephant in the emergency department: symptom of dispariteis in cancer care. Popul Health Manag. 2016 Jan 13 (Epub ahead of print). 20. Healthcare Cost and Utilization Project. Overview of the Nationwide Inpatient
21.
22.
23.
24.
25.
26.
7
Sample (NIS).
Please cite this article in press as: Solsky I, et al., Gastric cancer diagnosis after presentation to the ED: The independent association of presenting location and outcomes, The American Journal of Surgery (2017), https://doi.org/10.1016/j.amjsurg.2017.10.030