Objective: To enable clinicians to evaluate and interpret results of studies which utilize multivariate regression analytic techniques.

Introduction: Medical journals increasingly publish clinical observational studies which utilize complex statistical analyses. Because treatment and prevention recommendations may be founded on such studies, clinicians need to understand the basic principles of multivariate regression to appropriately evaluate and interpret these results.

Course: This workshop will utilize a case-based teaching approach to illustrate how multivariate regression techniques work, when they are appropriate, and how they are interpreted. The participant will understand how to evaluate and interpret studies which use multivariate analytic models. Minimal computer and math skills are necessary. We will begin by defining confounding in an example study and how it is distinguished from bias. Next we will define/calculate crude odds ratios and confidence intervals using data sets from studies of common pediatric emergency diagnoses as examples. We will then explain and demonstrate the results from the same datasets using logistic regression to adjust for confounding. This case based approach will be repeated using an example of simple linear and multiple linear regression. In the final portion of the workshop, we will discuss how to detect inappropriate/inadequate presentations of multivariate regression, including lack of model diagnostics, unstable models, lack of model validation and co-linearity of variables.

Robert Wright, Jacqueline Grupp-Phelan, Nathan Kuppermann. Divisions of Emergency Medicine, Brown University, Providence RI, University of Cincinnati, Cincinnati OH, University of California-Davis, Davis, CA

Providing pediatric subspecialty care to underserved rural communities has been a long-standing challenge. Several different health care delivery models have addressed this issue, the most common being outreach outpatient clinics with traveling Pediatric subspecialty teams based at tertiary children's centers. This workshop will expose the participant as to how we have utilized telemedicine, live 2 way video linkages, between the rural health care centers and our children's center. We provide 3 types of pediatric patient care: 1) Outpatient subspecialty consultation and follow-up, 2) PICU to rural adult ICU/adult intensivists for critically ill children, 3) Rural emergency rooms to our pediatric emergency room for acutely ill and injured children. Although the model of outpatient consultation is well established, the PICU to Adult ICU is a new model and has many challenges and opportunities that will be discussed. These include acceptance of a rural partner, ability of the PICU to triage and control all patient flow and beneficial relationships derived by supporting appropriate patients to remain in the local community. The rural ER to pediatric emergency room model involves a complex electronic network but is well received in rural communities. A limited discussion of technical details will also be available.

R. J. Dimand, J. P. Marcin, H. J. Kallas, Department of Pediatrics, University of California Davis, Sacramento, CA

This session will provide an introduction to HCUP, a family of databases and tools maintained by the Agency for Healthcare Research and Quality (AHRQ). The objectives of HCUP are to provide data and tools for research on use of health care services in the U.S. HCUP is a partnership among states, industry, and the federal government; currently 26 states contribute data to the project. HCUP includes uniformly formatted administrative databases including state inpatient data, a nationwide sample of inpatient data, a children's inpatient dataset, and ambulatory surgery data. HCUP data provide many advantages for research: (1) all payers including uninsured, (2) all stays within participating institutions, (3) permit state-level analyses, (4) large sample sizes permit studies of subgroups (e.g., minorities, rare diseases or procedures), and (5) charge data.

Recent innovations in dissemination make these data even more accessible. HCUP also includes software and Internet tools to better use these databases, for example, HCUPnet (interactive web-based query tool for HCUP data), free software (HCUP Quality Indicators, performance measures), and clinical groupers (Clinical Classification Software). HCUP has been infrequently used for pediatric research, but recent advances - the Children's Nationwide Inpatient Sample and the Child and Adolescent Clinical Classification Software - will make the databases and tools even more useful to study children's health care issues. Examples of research studies using HCUP data and tools will be used to illustrate applications.

A. Elixhauser, C. Steiner, J. Thompson, Agency for Healthcare Research and Quality, Rockville MD and University of Arkansas for Medical Sciences, Little Rock, AR

Recent guidelines for health supervision include Bright Futures, AAP GHS III, and DSM-pc codes of psychiatric risk. Problems with implementation include: lack of evidence, time required, lack of economic reinforcement, complexity of DSM-pc coding, and lack of training for management, particularly of mental health concerns. Our Child Health and Development Interactive System (CHADIS) facilitates implementation of guidelines using computerized parent questionnaires, e.g., all DSM-pc categories, child strengths, health risks. Computer analysis of responses results in suggested diagnoses with specifically related prompts and guides (e.g., trigger questions, differential diagnoses, potential recommendations, menus of appropriate products and resources such as tutors, etc.) on the electronic medical record worksheet the clinician uses during the visit. The clinician's interaction with the system results in reports (health passport and memory book) for parents. Products and local resources are provided with reviews by experts and without advertising. CHADIS will include validated tools for some aspects of developmental surveillance (e.g., START; Communication Screening System) and for further evaluation of problems (e.g., autism checklists, depression inventories). CHADIS will be free, initially offered to faculty interested in training and collaborative research. We will: demonstrate how CHADIS addresses common problems such as encopresis, sleep problems, ADHD; present data on parent response to CHADIS and pilot prevalence data for DSM-pc; elaborate on its use for collaborative research; and provide an opportunity for joining the network.

Raymond Sturner and Barbara Howard, The Johns Hopkins U School of Med., Baltimore, MD

As pediatricians are called upon to manage more complex clinical and administrative information, educators of residents should respond by providing relevant curricula in medical informatics (MI).

In this workshop, participants will relate their experiences with two different but long-standing MI electives for pediatric house officers. The workshop leaders will demonstrate templates, reading lists, web-based course administration, syllabi, and curricula used to facilitate MI experiences. Curricular content areas for these rotations include basic information technology literacy, the Internet, accessing medical literature, evidence-based medicine, practice management systems, hospital information systems, decision support, and health services research. The workshop includes an orientation to educational guidelines, a tutorial on web-based course administration, and strategies for integrating the experience into the residency program. Participants should have some familiarity with information technology but not necessarily special expertise in MI. Participants who desire to create an elective experience as described in this workshop should need no special resources other than internet access and as little as four hours a month to devote to supervising the residents. Participants will also be able to take away evaluation tools for pre-test and course evaluation purposes.

S. Andrew Spooner, University of Tennessee at Memphis, Memphis, TN, and Raymond G. Duncan, Cedars-Sinai Medical Center, Los Angeles, CA

Introduction: Practicing evidence-based decision-making (EBDM) is an essential skill for lifelong learning among practicing pediatricians and pediatric residents. The author has developed a set of workshops for general pediatricians designed to help build skills in EBDM and to teach these skills to residents in their clinics.

Goals: As a result of this workshop, participants will be able to 1) translate a clinical scenario into a well-built clinical question; 2) describe basic literature search techniques; 3) critically appraise an article; and 4) describe several tools useful for the daily practice of EBDM. This workshop will be useful for those just learning about EBDM and those want to teach this material to students or residents in their offices.

Format: Following a brief overview of EBDM, the participants will self-select small groups to focus on skills in the practice or teaching of EBDM. Each will have its own facilitator, to ensure that each participant comes away with the skills and comfort to be able to apply this knowledge in their daily practice. Special attention will be given to the newest sources available through the internet to facilitate the practice of EBDM. The session will conclude with discussion of useful pearls for the practice and teaching of EBDM generated by the participants. Participants will receive a workbook with handouts, pediatric case discussions, and teaching materials.

J. G. Frohna, Sheila Gahagan, Kenneth Pituch, Stephen M. Park, Department of Pediatrics, University of Michigan, Ann Arbor, MI

This workshop will enable participants to begin using public use files from the Medical Expenditure Panel Survey (MEPS), a large national study of health care use and expenditures, health status, access to care and insurance. The design of the survey will be reviewed and examples of past uses of the survey for pediatric research will be discussed to demonstrate the potential uses of the MEPS for children's health care issues and policy interventions. Participants will become familiar with the content of public use files, including person-level files, condition files, and event files, such as ambulatory care, hospital, emergency room, and prescribed medication files. The workshop will also cover the basics of setting up an analytic file for research from public use files and the use of appropriate software. Time will be allotted for questions. This workshop is designed for health services researchers who have a background or interest in using national surveys. Some familiarity with statistical software (SAS, SPSS, etc.) would be helpful. Analysts whose primary interest is in local area analysis or clinical research would not benefit from this workshop.

N. A. Krauss, R. M. Weinick, and J. S. Banthin Center for Cost and Financing Studies, Agency for Healthcare Quality and Research, Rockville, MD USA

Now in its third year, the Medical Informatics Special Interest Group will discuss research projects underway by our members. We will spend the bulk of the meeting presenting the results of a recent conference sponsored by the Agency for Healthcare Research and Quality that will result in requests for proposals to help create, evaluate, and disseminate information technology tools that enhance the health of children.

Each year there are thousands of injuries and deaths in US hospitals from medication errors. These blunders cost millions of dollars and result in loss of public confidence in our health care system. Medication errors are the second most frequent and the second most expensive event causing liability claims. Errant orders on the inpatient unit and incorrect prescriptions in the ED and clinics are common. Physicians, nurses, pharmacists, students, manufacturers and even caregivers all share responsibility. Many errors are preventable. This workshop will focus on common sources of medical errors and ways to prevent them. Workshop leaders will discuss specific issues including:

Root cause analysis looking at systems issues rather than individuals
Moving toward non-punitive approaches to reporting errors
Developing teaching programs for housestaff and students
The role of the pharmacist - how can they help?
New technology-are computer systems useful? Are the programs affordable?
Creating multidisciplinary medication use process improvement teams
Approaching the family after a medical error - legal and ethical implications

Participants will be asked to share their ideas and experiences. It is hoped that participants will be able to develop a multidisciplinary program in their own institutions to reduce and manage medical errors.

S. Selbst,* S. Levine,* J. Gould,* B. Taylor,* J. Fein,# K. Osterhoudt# A.I. duPont Hospital for Children,* Wilmington, DE and Children's Hospital of Philadelphia#, PA

The Human Genome Project is impacting every aspect of medicine. Dr. Craig Venter, President of Celera Genomics, one of the chief architects of this venture, will discuss the accomplishments of the human genome project and implications for future impact on health and disease in this special one-hour state of the art lecture.

Sequencing the Human Genome
J. Craig Venter, President, Celera Genomics, Rockville, MD

Supported in part by an educational grant from the Columbus Children's Hospital, Columbus, OH