Johns Hopkins University Student Paper Competition:
Exploration of Massive Healthcare Databases (Yasmin Said, organizer)

Ryan Archdeacon (Johns Hopkins University)
Andrew Stirn (Johns Hopkins University)
U. S. Hospitals Most Vulnerable to Terrorist Attacks

Saturday 8:30-8:50, San Marino

Abstract:

In this paper we propose an analytical method for determining U.S. hospital vulnerability to terrorist attacks based on criteria available in the 2003 Healthcare Cost and Utilization Project (HCUP) National Inpatient Survey (NIS) and to use this methodology to determine the top 20 at risk U.S. hospitals and the regions of the country most prone to terrorist attack based on concentration of these vulnerable hospitals. From a pool of 994 hospitals from around the country participating in the 2003 HCUP NIS, hospitals were assigned a vulnerability index from ranging from 0-2, 2 being most prone to terrorist attack. Criteria for assigning such an index to each hospital is based on hospital bed size, urban/rural setting, teaching/non-teaching status, and total patient discharges in 2003, each weighted differently. From this value, the top 20 most at risk hospitals were chosen as well as the most vulnerable region and state based on the number and concentration of at risk hospitals in these areas. New York State possesses both the highest occurrence of vulnerable hospitals and highest ratio of at risk to total hospitals by state, the state and city of New York would be most benefited by the development of a social network among its hospitals.


Devin O'Connor (Johns Hopkins University)
A Crime Study of Fairfax County, Virginia

Saturday 8:50-9:10, San Marino

Abstract:

Data pertaining to crime in the County of Fairfax was studied in the following report. Statistical analysis discerned a yearly trend in levels of crime rate. It has been found that crime level usually starts at low levels as the new years commence. A subsequent rise in crime rate culminates with a March peak, generated by extended economic markets. The March peak is followed by an initial drop in criminal activity. The year terminates with yet another peak in the month of December, seemingly caused by the holiday season. Furthermore, it has been found that crime levels within the different districts of Fairfax correlate very closely to the geospatial demographics, including population size, setting, and wealth.


Suraj Kabadi (Johns Hopkins University)
Eric Kim (Johns Hopkins University)
Eric Lin (Johns Hopkins University)
Effectiveness of Chemotherapy as a Treatment for Leukemia

Saturday 9:10-9:30, San Marino

Abstract:

Chemotherapy is one of the most widely used methods of treating leukemia and other cancers. This paper investigates the effect of chemotherapy treatment on mortality rates for leukemia patients during the years 1998-2003. Based on statistical differences found in preliminary analyses using a 2-proportions test and chi-square test on 2002 sample data, it was concluded that administration of chemotherapy, in general, decreases mortality rate; chemotherapy patients had a mortality rate of 5.6 percent while non-chemotherapy patients had a mortality rate of 8.3 percent. A prediction model was then generated, which describes the effect of age and chemotherapy administration on mortality rate. A full factorial design analysis using '00, '02 and '03 data produced a model, where Age and Chemo are 2-level variables. An interesting implication of this model is that chemotherapy treatment for older patients actually increases mortality rate when compared with patients not receiving chemotherapy treatment; however, for younger patients the opposite is true.


Edward Wei (Johns Hopkins University)
Adam Sifuentes (Johns Hopkins University)
Sarah Leismer (Johns Hopkins University)
Trends in Age Populations with High Occurrences of Meningitis

Saturday 9:30-9:50, San Marino

Abstract:

This paper develops a statistical model for meningitis that isolates at risk population groups and specific times of the year when meningitis is most prevalent, thus enabling medical professionals to have a heightened awareness when considering meningitis-like symptoms. Using the HCUP data set, age-based population models were developed, along with monthly basic incidence models. A lumbar puncture (LP) is a technique for diagnosing meningitis. The frequency of LP and corresponding diagnosis were found for the highest incidence group. Infants have the highest number of cases of diagnosed meningitis, yet LPs are not most often performed for meningitis, which is only 5.6 % of all LP. This suggests that meningitis is not diagnosed as frequently as it actually occurs. Lastly, it was shown that meningitis cases peak in the late summer and early fall, but are much lower in the winter months.


Posters:

Emma Morrison (Johns Hopkins University)
Neha Bajwa (Johns Hopkins University)
Labor Complications among Diabetic Pregnancies

Abstract:

The focus of this paper is to gather information about the type of complications in pregnant women due to diabetes. We analyze the 2002 HCUP dataset. Potential relationships between pregnancy and diabetes were explored using SAS v.9.1. Given only categorical datasets, Pearson chi-square analyses of contingency tables were implemented. In addition the data were analyzed using histogram plots and frequency tables. Both the phi and contingency coefficients were calculated in order to gather more insight into diabetic pregnancies. Based on the observed data, all complications associated with pregnancies, with the exception of C-sections, were related to diabetes. The strength of each condition was determined using the appropriate chi-square and alpha values.


Derrick M. Kuan (Johns Hopkins University)
Matthew A. Lalli (Johns Hopkins University)
Sarah E. Schaake (Johns Hopkins University)
Social Aspects and Attitudes Surrounding Tobacco Smoking

Abstract:

While medical research correlating smoking and cancer is commonly known, previous generations have not had such abundant information. As a result, many of the trends concerning tobacco products have changed over the past half-century. The purpose of this paper is to take a look at a large number of social aspects and changing attitudes surrounding the use of tobacco products. Examples include how people perceive the degree of danger in smoking, whether or not they would date a smoker, and how important it is to live or socialize in a non-smoking environment. As new generations continue to obtain more knowledge regarding the harmful effects of smoking, it is necessary to determine whether or not there are decreasing trends and changing attitudes in smoking.


Mike Bisogno (Johns Hopkins University)
Nik Ray-Mazumder (Johns Hopkins University)
A Comparison of Patient Mortality between Healthcare Maintenance Organizations (HMOs) and Private Insurance

Abstract:

HMOs, private insurance, and preferred provider organizations (PPOS) currently provide health insurance to most of the nation. This study is aimed at determining if there is a significant difference between the mortality rates of these two providers. This study analyzed 10-percent sample from the 1997 nationwide inpatient sample (NIS) to determine the top ten most lethal diseases. Based on a binomial model, the mortality rate and its significance level was determined using a one tail test. The results of this experiment show a significant increase in mortality in patients with HMO coverage for serious ailments that require treatment almost immediately such as myocardial infarction, respiratory arrest, septicemia, and stroke. The results also show a significant increase of mortality in patients with private insurance/PPO coverage in diseases that require prolonged treatment and medicines such as lung cancer and congestive heart failure. There are significant differences in many serious diseases between the two types of insurance analyzed in this study and the correlation seems to show that HMOs are worse for acute conditions while private insurance is worse for many more chronic conditions.


Liny John (Johns Hopkins University)
Thai Le (Johns Hopkins University)
Sandra Zoubovsky (Johns Hopkins University)
Degenerative Neurological Disorders: A Study of its Correlation with Race, Gender, Age, Admission Month, Length of Stay, and Geospatial Location of Hospital

Abstract:

The aim of this research was to analyze how different variables (race, gender, age, hospital state, admission month and length of hospital stay) correlated to each other in patients who were affected with degenerative neurological disorders. These variables were analyzed for the years 1998 and 2002 in order to compare trend relative to time. Degenerative neurological disorders are a class of disorders characterized by deterioration of nerve cells in the brain. This data was collected through hcup (healthcare cost and utilization project), which had data of hospital patients with various disorders and the characteristics of each of these patients. One of the main conclusions that the data and analysis suggested was the statistically significant difference between length of stay between 1998 and 2002. We explained this to be due to improvements in health care in the four-year span.