The Dartmouth Atlas Memo: Diabetes Discharges in New Jersey

Erica Humphrey | Steinhardt 2014

The Dartmouth Atlas Project has documented variations in the distribution of medical care and resources throughout the United States for more than 20 years. [1] Researchers at The Dartmouth Institute for Health Policy and Clinical Practice use data to reveal differences across states, regions, counties, and from one hospital to the next, controlling for factors related to a lack of insurance by focusing exclusively on Medicare patients. The project aims to understand why such differences exist, as well as how physicians, hospitals, and policy-makers can change the way the healthcare system functions in order to increase both its efficiency and effectiveness. A closer look at the Health Service Areas (HSAs) of the state of New Jersey, for example, shows wide differences in diabetes discharges for ambulatory-sensitive care per 1,000 Medicare enrollees from 2008-2010. Because diabetes affects an estimated 25.8 million people in the U.S.—8.3% of the population—and has direct annual costs estimated at $116 billion dollars, understanding how patients with the disease can be treated most effectively and efficiently is crucial to a well-functioning and lower-cost health care system. [2]

Figure 1 (see Appendix) represents the Dartmouth Atlas data regarding diabetes discharges per 1,000 Medicare enrollees in the state of New Jersey by HSA. The Dartmouth Atlas defines an HSA as local health care markets for hospital care; “a collection of ZIP codes whose residents receive most of their hospitalizations from the hospitals in that area.” This distinction divides the United States into 3,436 HSAs, with 65 in the state of New Jersey. The clear outlier in the state is Perth Amboy, with 9.22 discharges, compared to the U.S. national average of 2.62, and the state’s county with the lowest rate of discharges, Princeton, at 2.00 per 1,000. Perth Amboy even sits 3.12 discharges per 1,000 than the second highest HSA, Hoboken. This could suggest a number of scenarios. Perhaps diabetics in Perth Amboy control their disease better than do those in Princeton, and therefore they present at the hospital with less severe diabetic complications and do not require admission. Or perhaps, doctors in Princeton hospitals are overly cautious and unnecessarily admit patients, either because they do not understand the severity of diabetes and its complications or because they have a greater number of available beds; or maybe the situation is that doctors in Perth Amboy are in fact not cautious enough or do not have the facilities required to admit all of the patients that do actually need inpatient care. It is difficult to know which of these discharge rates hospitals should be striving for, and even whether the U.S. average is a good indicator of proper procedure. All of these unknowns and possibilities present different implications for improved medical practice and how that can be achieved through policy change. For this reason it is essential to analyze and understand these variances.

Differences Among Patients
It is possible that wide variance in discharge rates for diabetes is due to similarly wide variance among patients—broad factors relating to socioeconomic status might affect their prioritization of health or their economic means to eat well, exercise, and control their disease; the existence of other conditions might complicate their health; and perhaps random factors such as an individual’s personal diligence in managing diabetes would affect how thedisease impacts one’s overall health. According to the U.S. Census Bureau, the median household income in Perth Amboy from 2007 to 2011 was $45,369, whereas in Princeton it was $112,808. [3,4] Similar disparities are reflected in percent of persons living below the poverty line, with 19.9% in Perth Amboy and 5.4% in Princeton. 3,4 HSAs do encompass more than the city limits that confine the collection of this Census data, but the differences in economic status between the two areas are clear. While individuals are eligible for Medicare despite income or poverty level, it is true that a vastly different socioeconomic status in Perth Amboy could potentially affect an individual’s ability to purchase healthy food, as well as find a time and a place to exercise. This, however, would suggest that residents of Perth Amboy are less healthy than those of Princeton, and therefore more likely to require hospitalization. This particular theory of personal differences among patients, then, must not be true, because the data shows the opposite.

Despite socioeconomic differences, the data might be suggesting that diabetics in Perth Amboy are in fact drastically healthier and have better control over their disease than the rest of New Jersey and even the rest of the country and are therefore less likely to require admission when presenting at a hospital with a diabetic incident or complication. But the Dartmouth Atlas data that indicates the average annual percent of diabetic Medicare enrollees age 65-75 having a hemoglobin A1c test in 2010, as a measurement of how closely an individual’s disease is being monitored, indicates that this is unlikely to be true. According to the American Diabetes Association’s (ADA) Standards of Care, the A1C is thought to reflect average glycemia over several months and has strong predictive value for diabetes complications. For this reason it “should be performed routinely in all patients…as part of continuing care.” [5] An HSA’s percentage of A1C testing can therefore be associated with that area’s quality of physician attention and patient management, which would also be associated with better control over diabetes. In this measure, Perth Amboy ranks in the 14 percentile, while Princeton ranks in the 92 percentile throughout the state. With Perth Amboy on the low end of diabetes control and Princeton on the upper end, the theory that Perth Amboy patients are less likely to require hospitalization due to initial better rates of health is most likely untrue.

Professional Uncertainty
If patient control over diabetes cannot explain the wide variation between Princeton and Perth Amboy in diabetes discharges, then doctors’ decision- making might warrant consideration as a cause. The ADA’s “Hospital Admissions Guidelines for Diabetes Mellitus” that are relevant to those of Medicare age dictate that “inpatient care may be appropriate for the following”:

– Life-threatening acute metabolic complications of diabetes.
– Substantial and chronic poor metabolic control that necessitates close monitoring of the patient to determine the etiology of the control problem, with subsequent modification of therapy.
– Severe chronic complications of diabetes that require intensive treatment or other severe conditions unrelated to diabetes that significantly affect its control or are complicated by diabetes.
– Institution of insulin-pump therapy or other intensive insulin regimens. [6]

Largely due to the nature of the disease and its potential to cause any number or severity of complications, these guidelines are vague and leave much discretion to doctors. Even the circumstances under which a physician should consider a patient to have “uncontrolled diabetes,” which would imply a generally poor state of health due to poor blood sugar control, are unclear—the ADA recommends categorizing diabetics as uncontrolled if they exhibit any combination of seven criteria, which include anything from recurring episodes of severe hypo- or hyperglycemia to experiencing psychosocial issues that interfere with work. Even less clear is whether such a state should automatically necessitate hospitalization.

Ultimately, individuals’ attention to their care for diabetes and personal experience of the disease are widely varied, and where there exists so much difference there is bound to exist inconsistency in practice from one hospital to the next and even from one doctor to the next. In contrast to a simple situation such as a hip injury, which has a single agreed upon treatment and thus sees little variation across hospitals, the complexity of a chronic disease such as diabetes does not allow for the same universal course of treatment, or an ability to easily predict or understand complications.

Resource Availability
The Dartmouth Atlas observes a trend of supply- sensitive care across the nation’s Medicare data—“in regions where there are more hospital beds per capita, patients will be more likely to be admitted to the hospital” and “in regions where there are relatively fewer medical resources, patients get less care.” [7] According to Fisher et al, however, additional utilization of care in high- spending regions, including treating patients with chronic disease in inpatient units, has “been demonstrated to be associated with the local supply of physicians and hospital resources.” [8] Looking at the range of discharge rates across New Jersey, it might then be worth questioning whether the numbers of discharges in the lower regions is actually warranted, or whether areas such as Princeton are more likely to admit patients, simply because they have the space, for a situation that could be handled through outpatient care. Perth Amboy is the sole outlier far above the data of the rest of the state, and it is hard to believe that this HSA would be the only area not unnecessarily admitting patients. But perhaps in fact it faces the opposite problem hypothesized to be taking place in Princeton, and has resource limitations that lead to increased discharges. The question then must be where the
ideal level lies and what can be done to reach that more universally across the United States.

Variation in diabetes discharges is most likely related to the combined factors of professional uncertainty regarding a disease as complex as diabetes and resource availability. Efforts to rectify these issues, which have the potential to improve patients’ health as well as decrease Medicare spending, must therefore focus on an effort for streamlined and enforceable science-based guidelines by which physicians can approach their work. Creating a universally accepted authority on best practices of care and implementing it through Electronic Health Records (EHRs) would be efficient and effective in addressing many issues of hospital variance.

David Eddy writes in Health Affairs that Evidence- Based Medicine (EBM) is “the conscientious, explicit, and judicious use of current best evidence in making decisions about the care of individual patients.” [9] Eddy notes that many of the guidelines that physicians currently follow are informal, with origins unrelated to medical evidence. The integration of current science and research with individual clinical expertise is in fact crucial to eliminating personal decision-making factors among doctors that are not based in science but cause overutilization of health care. Such standards can be disseminated and enforced through evidence-based decision support systems operating through EHRs. The Patient Protection and Affordable Care Act (ACA) mandates health plans’ adoption of EHRs for the purposes of reducing paperwork and administrative burdens, cutting costs, reducing medical errors and improving quality of care. [10] Additionally and more specifically, however, if implemented with EBM in mind, the policy has the potential to reduce unnecessary variations in care.

In addition to the wide range of quality and efficiency issues that EHRs are said to address, variations in handling diabetes cases will specifically be addressed by the use of checklists, alerts, and predictive tools as well as embedded clinical guidelines that promote standardized, evidence- based practices. A 2012 study by the Commonwealth Fund reports high levels of consistency across multiple hospitals with the implementation of an EHR system, which presents prompts and alerts to set reminders and guide care based on clinical guidelines that gain consensus from staff.[11] Cebul et al’s 2011 report to the New England Journal of Medicine also found that across 46 practices, “achievement of composite standards for diabetes care was 35.1 percentage points higher at EHR sites, and achievement of composite standards for outcomes was 15.2 percentage points higher.”[12] Diabetes is a complex disease, and the complexities of managing it as a physician are not easy to overcome. The technology’s ability to consider every factor of a patient’s health against the guidelines for caring for a patient with diabetes can aid doctors in avoiding over- and under-utilization of care. And as the use of EHRs expands, evidence-based guidelines can span across states to cohesively guide doctors throughout the U.S. in their administration of care.

The widespread use of EHRs does present some barriers, such as economic feasibility in some hospitals as well as the compatibility of systems from hospital to hospital. The American Recovery and Reinvestment Act of 2009, however, addressed some of these issues by allocating $1.2 billion to aid hospitals in implementing the technology targeted at technical assistance and information sharing. [13] The ACA’s further push to implement this technology is necessary in a time when Medicare spending adds up to $556 billion a year, 26% of which went to hospital inpatient services in 2012. [14] If the U.S. as a whole can achieve spending levels comparable to those of the lowest-spending regions highlighted by the Dartmouth Atlas Project, annual Medicare expenditures could be reduced by up to 30% of Medicare expenditures could be achieved. [15] Further research on this topic should include investigating the association in these HSAs between number of beds and other hospital resources with the number of discharges. Additionally, the effects of EHRs on certain quality and efficiency measures have been studied, but further examination of their potential to prevent unnecessary use of resources would encourage the uptake of this method to alleviate the current situation.

Figure 1: Diabetes discharges per 1,000 Medicare enrollees (2008-10) by NJ HSA

Figure 1: Diabetes discharges per 1,000 Medicare enrollees (2008-10) by NJ HSA






Figure 2: Average annual percent of diabetic Medicare enrollees age 65-75 having hemoglobin A1c

Figure 2: Average annual percent of diabetic Medicare enrollees age 65-75 having hemoglobin A1c

1 The Dartmouth Atlas of Health Care. (2013). from

2 Centers for Disease Control and Prevention. National diabetes fact sheet: national estimates and general information on diabetes and prediabetes in the United States, 2011. Atlanta, GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, 2011.

3 U.S. Census Bureau: State and County QuickFacts. (2013). Perth Amboy (city), New Jersey.
4 U.S. Census Bureau: State and County QuickFacts. (2013). Princeton (borough), New Jersey.
5 American Diabetes Association. (2009). Standards of Medical Care in Diabetes—2009. Diabetes Care, 32(Supplement 1), S13-S61. doi: 10.2337/dc09-S013
6 Hospital Admission Guidelines for Diabetes Mellitus*. (2002). Diabetes Care, 25(suppl 1), s109. doi: 10.2337/diacare.25.2007.S109
7 Supply-Sensitive Care. (2013). from
8 Fisher, E. S., Wennberg, D. E., Stukel, T. A., Gottlieb, D. J., Lucas, F. L., & Pinder, E. L. (2003). The implications of regional variations in Medicare spending. Part 2: health outcomes and satisfaction with care. Annals of Internal Medicine, 138(4), 288-298.
9 Eddy, David M. (2005). Evidence-Based Medicine: A Unified Approach. Health Affairs, 24(1), 9-17. doi: 10.1377/hlthaff.24.1.9
10 Key Features of the Affordable Care Act, By Year. from
11 Silow-Carroll, S., Edwards, J. N., & Rodin, D. (2012). Using electronic health records to improve quality and efficiency: the experiences of leading hospitals. Issue Brief (Commonw Fund), 17, 1-40.
12 Cebul, Randall D., Love, Thomas E., Jain, Anil K., & Hebert, Christopher J. (2011). Electronic Health Records and Quality of Diabetes Care. New England Journal of Medicine, 365(9), 825-833. doi:10.1056/NEJMsa1102519
13 Vice President Biden Announces Availability of Nearly $1.2 Billion in Grants to Help Hospitals and Doctors Use Electronic Health Records. (2009).
14 Billings, John. (2013). Medicare: The basics and issues for reform [PowerPoint slides]. Retrieved from NYUClasses website: 4e62-a654- 2eff3989c1cc/page/2f88d928-13de-40c9-afa1-a412561883ea#
15 Fisher, E. S., Wennberg, D. E., Stukel, T. A., Gottlieb, D. J., Lucas, F. L., & Pinder, E. L. (2003). The implications of regional variations in Medicare spending. Part 2: health outcomes and satisfaction with care. Annals of Internal Medicine, 138(4), 288-298.


  1. You really know your stuff… Keep up the good work!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s