Currently Viewing:
The American Journal of Managed Care February 2015
A Multidisciplinary Intervention for Reducing Readmissions Among Older Adults in a Patient-Centered Medical Home
Paul M. Stranges, PharmD; Vincent D. Marshall, MS; Paul C. Walker, PharmD; Karen E. Hall, MD, PhD; Diane K. Griffith, LMSW, ACSW; and Tami Remington, PharmD
Quality’s Quarter-Century
Margaret E. O'Kane, MHA, President, National Committee for Quality Assurance
How Pooling Fragmented Healthcare Encounter Data Affects Hospital Profiling
Amresh D. Hanchate, PhD; Arlene S. Ash, PhD; Ann Borzecki, MD, MPH; Hassen Abdulkerim, MS; Kelly L. Stolzmann, MS; Amy K. Rosen, PhD; Aaron S. Fink, MD; Mary Jo V. Pugh, PhD; Priti Shokeen, MS; and Michael Shwartz, PhD
Did Medicare Part D Reduce Disparities?
Julie Zissimopoulos, PhD; Geoffrey F. Joyce, PhD; Lauren M. Scarpati, MA; and Dana P. Goldman, PhD
Health Literacy and Cardiovascular Disease Risk Factors Among the Elderly: A Study From a Patient-Centered Medical Home
Anil Aranha, PhD; Pragnesh Patel, MD; Sidakpal Panaich, MD; and Lavoisier Cardozo, MD
Employers Should Disband Employee Weight Control Programs
Alfred Lewis, JD; Vikram Khanna, MHS; and Shana Montrose, MPH
Race/Ethnicity, Personal Health Record Access, and Quality of Care
Terhilda Garrido, MPH; Michael Kanter, MD; Di Meng, PhD; Marianne Turley, PhD; Jian Wang, MS; Valerie Sue, PhD; Luther Scott, MS
Leveraging Remote Behavioral Health Interventions to Improve Medical Outcomes and Reduce Costs
Reena L. Pande, MD, MSc; Michael Morris; Aimee Peters, LCSW; Claire M. Spettell, PhD; Richard Feifer, MD, MPH; William Gillis, PsyD
Decision Aids for Benign Prostatic Hyperplasia and Prostate Cancer
David Arterburn, MD, MPH; Robert Wellman, MS; Emily O. Westbrook, MHA; Tyler R. Ross, MA; David McCulloch, MD; Matt Handley, MD; Marc Lowe, MD; Chris Cable, MD; Steven B. Zeliadt, PhD; and Richard M. Hoffman, MD, MPH
Faster by a Power of 10: A PLAN for Accelerating National Adoption of Evidence-Based Practices
Natalie D. Erb, MPH; Maulik S. Joshi, DrPH; and Jonathan B. Perlin, MD, PhD, MSHA, FACP, FACMI
Currently Reading
Differences in Emergency Colorectal Surgery in Medicaid and Uninsured Patients by Hospital Safety Net Status
Cathy J. Bradley, PhD; Bassam Dahman, PhD; and Lindsay M. Sabik, PhD
Patients Who Self-Monitor Blood Glucose and Their Unused Testing Results
Richard W. Grant, MD, MPH; Elbert S. Huang, MD, MPH; Deborah J. Wexler, MD, MSc; Neda Laiteerapong, MD, MS; E. Margaret Warton, MPH; Howard H. Moffet, MPH; and Andrew J. Karter, PhD
The Use of Claims Data Algorithms to Recruit Eligible Participants Into Clinical Trials
Leonardo Tamariz, MD, MPH; Ana Palacio, MD, MPH; Jennifer Denizard, RN; Yvonne Schulman, MD; and Gabriel Contreras, MD, MPH
A Systematic Review of Measurement Properties of Instruments Assessing Presenteeism
Maria B. Ospina, PhD; Liz Dennett, MLIS; Arianna Waye, PhD; Philip Jacobs, DPhil; and Angus H. Thompson, PhD
Emergency Department Use: A Reflection of Poor Primary Care Access?
Daniel Weisz, MD, MPA; Michael K. Gusmano, PhD; Grace Wong, MBA, MPH; and John Trombley II, MPP

Differences in Emergency Colorectal Surgery in Medicaid and Uninsured Patients by Hospital Safety Net Status

Cathy J. Bradley, PhD; Bassam Dahman, PhD; and Lindsay M. Sabik, PhD
Safety net hospitals reduce emergency surgeries among Medicaid and uninsured patients, and provide a benefit to these populations relative to other providers.
We examined whether safety net hospitals reduce the likelihood of emergency colorectal cancer (CRC) surgery in uninsured and Medicaid-insured patients. If these patients have better access to care through safety net providers, they should be less likely to undergo emergency resection relative to similar patients at non– safety net hospitals.

Study Design
Using population-based data, we estimated the relationship between safety net hospitals, patient insurance status, and emergency CRC surgery. We extracted inpatient admission data from the Virginia Health Information discharge database and matched them to the Virginia Cancer Registry for patients aged 21 to 64 years who underwent a CRC resection between January 1, 1999, and December 31, 2005 (n = 5488).

We differentiated between medically defined emergencies and those that originated in the emergency department (ED). For each definition of emergency surgery, we estimated the linear probability models of the effects of being treated at a safety net hospital on the probability of having an emergency resection.

Safety net hospitals reduce emergency surgeries among uninsured and Medicaid CRC patients. When defining an emergency resection as those that involved an ED visit, these patients were 15 to 20 percentage points less likely to have an emergency resection when treated in a safety net hospital.

Our results suggest that these hospitals provide a benefit, most likely through the access they afford to timely and appropriate care, to uninsured and Medicaid-insured patients relative to hospitals without a safety net mission. Am J Manag Care. 2015;21(2):e161-e170
Safety net institutions reduce emergency colorectal cancer surgery among uninsured and Medicaid insured patients:
  • There is a lower percentage of resections following an emergency department (ED) visit.
  • Findings may be due to improved access to outpatient care for these patients who can only access care through the ED outside of the safety net.
  • Suggests an ongoing and vital role for safety net providers following the implementation of Medicaid expansions.
Nationwide, hospital emergency department (ED) use has increased due to patients who are Medicaid-insured or uninsured with limited access to primary care.1 Patients admitted through the ED often present with more severe illness,2 and their care is likely to be poorly coordinated following discharge. These emergency presentations are detrimental for patients, costly to society, create a burden for hospitals,3 and occur more often in safety net hospitals.4,5

Safety net hospitals are institutions that, by legal mandate or explicit mission, offer access to services regardless of patients’ ability to pay.5 Safety net hospitals are often located in underserved communities6 and they receive financial compensation from the state and federal government for providing care to underserved populations.7 Recent evidence suggests that safety net providers deliver lower quality care,8,9 calling into question the adequacy of these providers to deliver healthcare to the populations they serve.

We examined whether safety net hospitals are associated with emergency colorectal cancer (CRC) surgery, which serves as an indicator of poor access to outpatient cancer care. Because safety net institutions have a mission to serve the uninsured and Medicaid-insured, these hospitals may provide better access to timely and appropriate care for medically underserved populations compared with what these patients receive outside the safety net. For example, faculty associated with academic health centers, which are often core safety net providers, give a considerable amount of care to underserved patients in outpatient settings,5 possibly alleviating the need for emergency care and the use of the ED as a portal for symptom appraisal. In contrast, the ED may be the only point of access for uninsured and Medicaid patients in non–safety net settings. For these reasons, emergency CRC resection is an informative signal of access to care, making it an ideal condition to investigate the differential effects of safety net hospitals on access to care for complex and costly conditions such as cancer.

CRC is the third most common cancer in the United States,10 resulting in approximately 142,000 new cases annually,11 and spending on CRC was estimated to be $14.14 billion in medical care costs in 2010.12 Surgical resection is standard treatment for all stages of CRC, and is generally conducted on an elective basis, although patients may present acutely and require emergency surgery.13 Emergency presentation of CRC is associated with increased morbidity and mortality, including diminished 5-year survival.14 About 15% to 30% of CRC patients require an emergency resection for several reasons, including bowel perforation, peritonitis, obstruction, or hemorrhage.14 Uninsured and Medicaid patients aged less than 65 years are 2 to 2.5 times more likely to require an emergency resection than their privately insured counterparts.15 These emergency resections are associated with longer inpatient stays, higher costs, and higher inpatient mortality.15 Given the rising demand for care16 and increasing evidence of poor health outcomes in safety net hospitals,8,9,17 our investigation is timely. Although the safety net may underperform on some outcomes, safety net hospitals deliver care that might be otherwise forgone—the lack of which would result in increased morbidity and mortality among low-income uninsured and Medicaid patients.18


Inpatient admission and discharge status were extracted from the Virginia Health Information (VHI) discharge database, which contained discharge abstracts on all Virginia civilian hospital admissions that exceeded 23 hours. Since 1993, VHI has collected information on all licensed hospital discharges (more than 870,000 per year) under contract with the Virginia Department of Health. Discharge abstracts included patient information, International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis and procedure codes, payer information, dates of admission and discharge, source and type of admission, and discharge status. CRC resections were identified by one of these ICD-9-CM diagnosis codes—153.0-154.3, 154.8, V10.05, and V10.06—in conjunction with at least 1 of the following ICD-9-CM procedure codes—45.4X, 45.5X, 45.6X, 45.7X, 45.8X, 46.10, 46.43, 46.52, 46.81, 46.82, 48.3X, 48.4X, 48.5X, and 48.6X.19

The VHI and the American Hospital Association (AHA) survey supplied hospital tax status, teaching status, staffed beds, ownership, charity care, and Medicaid charges that were used to classify hospitals according to safety net status. The AHA Annual Survey of Hospitals profiles a universe of more than 6500 hospitals. Data are available at the hospital and system level for research, and the AHA survey is a primary reference for government agencies and industry reports. Using the proportion of charges for charity care, Medicaid, and for receipt of Disproportionate Share Hospital funds, 2 out of 61 hospitals were designated as safety net providers.20 In a sensitivity analysis, we expanded the definition to the top 10 hospitals in proportion of charges for charity care and Medicaid. These estimations were qualitatively unchanged, although the evidence that safety net hospitals reduced emergency resections for Medicaid patients became stronger (eAppendix, available at

The Virginia Cancer Registry (VCR) is a statewide registry of data on individuals diagnosed or treated in Virginia and on Virginia residents who received cancer care out of state. The North American Association of Central Cancer Registries certifies the VCR to be a provider of complete, accurate, and timely cancer incidence data. Using the VCR from January 1, 1999, to December 31, 2005, we identified 8666 CRC patients. The following exclusions were made: unknown gender (n = 1), unknown race (n = 93), insurance other than Medicaid or private, or uninsured (n = 1119). Among the 7453 remaining patients, 5488 (74%) had a claim for inpatient resection while 1965 (26%) did not. Of those 1965, we tried to distinguish between patients without a claim due to missing data and patients who legitimately did not have an inpatient surgery. Our assessment suggests that most of those without a claim did not have an inpatient surgery. In sum, the VCR also indicated an absence of surgery on 464 patients: 409 patients had stage IV cancer or unstaged cancer for which surgery is often not indicated, 324 had stage 0 disease for which outpatient surgery is indicated, and 90 had military insurance and likely had surgery in a military facility, leaving 678 (9%) patients who may have had outpatient surgery or no surgery, or were missing claims.

In addition to the 5488 patients identified in the VCR, we identified another 981 patients from the VHI discharge data set with inpatient surgical claim for CRC, but since they were not reported in the VCR, we could not be certain they had cancer. Therefore, in a separate analysis, we included these patients in the sample; the findings were unchanged (results not shown). The Figure traces the steps taken to select the study sample.

We considered 4 definitions of emergency resection. The first 3 are: 1) all admissions through the ED that also had an emergency ICD-9-CM procedure code for bowel perforation (569.83),21 peritonitis (567.2, 567.8, 567.9),21 obstruction (560.X)21 or hemorrhage (578.X)16; 2) all ED admissions; 3) ED admissions, but without an ICD-9-CM emergency code. By considering ED admissions without an emergency diagnosis code, we identified admissions related to access, but without a condition that required immediate medical attention. The fourth definition restricts the definition of emergency resection to patients with an emergency ICD-9-CM code, reflecting an immediate medical emergency, regardless of source of admission. Table 1 reports the number of emergency resections using each definition.

Statistical Analysis

We estimated the probability of an emergency resection using linear probability models to avoid challenges associated with the interpretation of interaction terms in nonlinear models,22 such as those between health insurance and safety net status. We account for clustering of patients within hospitals using multiple variance covariance structures. Bayesian Information Criterion was used to select the best fit, allowing for different correlations among patients in safety net hospitals and non–safety net hospitals. The models included random intercepts and accounted for compound symmetric correlations within hospitals.

We calculated the predicted percent of patients who underwent emergency resection using the expected values for patients in safety net and non–safety net hospitals and health insurance type under the same multivariate distribution of the other covariates in the sample. We used the bootstrap method to construct the nonparametric percentile and 95% CIs. One thousand random samples of the same size as the original analytical data set were drawn with replacement. Statistical significance was estimated between hospital types and insurance groups.

We were mindful of the possibility that unobservable characteristics that lead one to seek care in a safety net facility may also be the same characteristics that are associated with emergency surgery. These characteristics include delay in seeking care, poor preventive care, poor health status, etc. Observable patient characteristics included in the estimation were public insurance, no insurance, and racial/ethnic minority, which are also strongly associated with receiving care in a safety net hospital. If endogeneity were a problem, we might observe an effect attributable to the safety net hospital when in fact, the effect is attributable to the unobserv-able patient characteristics that led them to the safety net hospital.

We address the possibility of endogeneity using 3 approaches. First, we estimated an instrumental variables equation. We tested 2 specifications for distance as an instrument for hospital choice. The first specification used distance to the closest safety net hospital and the second used the difference between the distance to the closest safety net hospital and the distance to the closest non– safety net hospital. In both models, statistical tests for endogeneity were rejected (results not shown). Second, we added controls associated with the use of safety net hospitals to the linear probability models. In addition to insurance status, race, and socioeconomic status, we control for age, sex, distance traveled to the hospital where surgery was performed, comorbid conditions, cancer stage, and a summary measure of socioeconomic status for the zip code in which patients resided. Third, we used multiple definitions of “emergency surgery” to test the robustness of the findings under different definitions.

Copyright AJMC 2006-2019 Clinical Care Targeted Communications Group, LLC. All Rights Reserved.
Welcome the the new and improved, the premier managed market network. Tell us about yourself so that we can serve you better.
Sign Up