Search DBMI   
 

Clinical Computing Systems
at the Health Sciences Campus of Columbia University

The recent history of clinical computing at the health sciences campus dates to the early 1980s when Rachael Anderson, MS, director of the health sciences library, embarked on the Integrated Academic Information Management Systems (IAIMS) project funded by the National Library of Medicine. A number of successful prototype clinical systems led Dr. Henrik Bendixen, vice president of the health sciences campus, and Dr. Thomas Q. Morris, president of the Presbyterian Hospital, to create a unique unit, initially known as the Center for Biomedical Informatics. They recruited a well-known leader in the field, Dr. Paul Clayton, who had helped to develop the highly regarded and innovative clinical systems at LDS Hospital in Salt Lake City. Professor Clayton was simultaneously named head of the academic unit in P&S (where he was initially a Professor in the Department of Medicine) and the Director of Medical Informatics Services in the hospital. In the latter role he recruited a talented staff and oversaw the creation of new clinical systems, most of which were developed locally, often with funding from government grants. In the former role he developed a fine academic unit which by the mid-1990s had been transformed into a full department at Columbia. Today the Department of Biomedical Informatics (DMI) has 14 full-time faculty and almost 50 graduate students who are pursuing MA or PhD degrees in biomedical informatics. The chair of the department continues to have a senior management role in the hospital (Director of Biomedical Informatics) while sitting with both the clinical chairs and the basic science chairs in health sciences. With Professor Clayton’s return to Utah in 1998, Dr. Edward Shortliffe was recruited to chair the department and he has served in that role since early in 2000. The vice-chair is Dr. George Hripcsak, who served as acting chair after Dr. Clayton’s departure and has play a key role in the development and oversight of current clinical systems, as described below. To this day, out of clinical necessity, efficiency, and tradition, the university and hospital are intimately tied together in their implementation of centralized clinical information systems at the health sciences campus.

In addition to institutional support from the university and hospital, Professor Clayton and his staff turned to the federal government for support of their innovative systems. An early recipient of one of the coveted IAIMS phase III implementation grants from the National Library of Medicine the biomedical informatics unit also attracted many other grants both from government agencies and from industry. A joint development effort with IBM led to a significant infusion of external dollars that assisted in the development of their earliest systems for the hospital. Thus external grant dollars, generally awarded through the University, have played a significant role in the creation of the clinical systems currently in use at Columbia-Presbyterian Medical Center. The quality and novel capabilities of the resulting systems also brought great attention internationally to the DBMI and to the clinical computing environment at the campus. Today the clinical systems at the campus attract visitors from throughout the world and the Department is clearly identified as among the top two or three academic units. Its training programs are large, its students are top notch, and its faculty continue to provide an unusual blend of cutting edge research with superb management of a variety of practical systems that are in use by Columbia physicians on a daily basis.

The management of clinical systems changed substantially, first with the merger that created New York-Presbyterian Healthcare System (NYP) and subsequently with the decision, in late 1999, to outsource routine system management functions to First Consulting Group’s Management Services group (FCGMS). Routine support for the clinical systems created by the DBMI has now been turned over to FCGMS, although in several cases the FCGMS staff are supervised by informatics faculty who receive significant portions of their salaries from NYP. This unusual interweaving of university, hospital, and FCGMS staff and funds has evolved naturally with the changes in local systems and management structures, and it works remarkably well. There is perhaps no other academic department in biomedical informatics that has achieved such a fine and effective balance between routine operational roles in the clinical environment, funded by the health system, and simultaneous academic roles with extensive grant funding, a significant faculty-taught curriculum, and large numbers of graduate students. Columbia’s department and clinical systems are the envy of comparable academic medical centers in the US and abroad. Today’s clinical information systems at the health sciences campus are a complex mix of locally-developed and vendor-supplied products. The key to their success is the overall systems architecture, created by the DBMI and overseen by faculty who play significant roles in the hospital.

In parallel to the development of centralized clinical systems administered by DBMI, university administration, university departments, practitioners, and individual hospital services have been pushing forward to implement clinical systems essential to their respective areas. These systems, while not generally shared across the campus, are indispensable to the function of the campus. Similarly, units and departments at the health sciences campus have grown expertise in clinical information systems in their areas, in many cases achieving national renown for their work.

Although a complete listing of all clinical systems would be beyond the scope of this document, the following elements are particularly important to understand. The following are centralized systems, administered by DBMI and, for the most part, represent joint university-hospital ventures:

  • At the core of the clinical environment is the clinical database, a large, longitudinal collection of patient data that is among the most comprehensive available anywhere. The database contains information on millions of patients, dating back over 12 years, and it now is the official medical record for the Columbia-Presbyterian campus of NYP. Many kinds of clinical data exist only in the clinical database, without traditional paper backup in a hard copy medical record. With the addition of more and more electronic “feeds” into the database over the years, it now provides a remarkably complete set of demographic, financial, and clinical data for patients who are seen at the Columbia campus.
  • The clinical data viewer (WebCIS) is the window into the clinical database. It is used by clinicians to access data on their patients and supports over 4,000 users of whom over 2,000 log into the system in any given day. Implemented for access via the Netscape web browser, WebCIS has achieved high level of user satisfaction and its name is now the vernacular for the entire clinical computing environment at the Columbia campus. Because WebCIS is accessed using the World Wide Web, patient data are potentially available to clinicians from any location where they have access to the Internet and to the Netscape browser. This is a great convenience for clinicians who wish to check patient data from their homes, for example, but it means that the DBMI has needed to create authentication protections that prevent access unless individuals have both recognized passwords and physical tokens that they carry with them (in our case, cryptocards that provide a new secondary password every 60 seconds).
  • The clinical data warehouse is a reflected version of the clinical database, designed to support clinical trials and a variety of management reporting functions. The clinical database is optimized for access to current information on a specific patient, whereas the warehouse allows queries and produces reports that summarize data across sets of patients, providers, or other organizational entities (such as clinical departments, wards, or clinics). Very few academic medical centers have comprehensive clinical data warehouses, and this developing facility at Columbia is expected to be a major boon to clinical research as well as management analyses. Many of its features are already available to the Columbia faculty and staff. The warehouse is funded by and overseen by the Clinical Trials Office.
  • The medical entities dictionary (MED) is a controlled vocabulary, developed by DBMI faculty, that allows the translation of data elements between systems and the standardization of data when they are stored in the clinical database. This “behind the scenes” capability is crucially important in the integration of diverse systems and the system developed at Columbia is a model for similar efforts at other institutions. In fact, there are current efforts to adapt the MED for use on the Cornell campus of NYP.
  • The security infrastructure is among the most advanced in the world. Columbia has taken issues of data privacy, protection, and user authentication very seriously, and the systems developed and overseen by DBMI faculty are especially secure and private. In a complex environment, however, some of the greatest challenges occur when vendor-supplied systems were not created with the same attention to detail in the area of security. A networked environment is only as secure as its weakest link. The security officer for NYP is a faculty member in DBMI, Professor Soumitra Sengupta, and he plays a crucial ongoing role in the oversight of system and network security in NYP (at both campuses).
  • The interface engine is the system that handles data from ancillary computer systems (e.g., from the clinical laboratories or radiology), transforms them appropriately, and passes them along for storage in the clinical database. The initial interface engine was built by DBMI faculty and staff, but when high quality commercial versions appeared on the market some years later, the home-grown system was replaced by a commercial product.
  • The event monitor is a novel decision-support capability that exists only at a handful of institutions at this time. The system supports the encoding of decision rules that can be triggered when a patient’s data (in the clinical database) meet a set of defined criteria. The rule, once triggered, then allows the computing environment to take an appropriate action (such as to send a warning to a clinician’s email or beeper).
  • The text processing engine (MedLEE) is a system developed with grant support by one of the DBMI faculty, Professor Carol Friedman, and her colleagues. This system processes raw text (e.g., from dictated radiology reports or admission histories and physical exams) in an effort to interpret their content and generate suitable summary data or codes. Automated coding is an extremely important emerging concept because of the significant cost associated with manual coding and the financial exposures if such coding is not done well (e.g., for a variety of reimbursement purposes). The MEDLEE system is well known in the informatics community and has attracted substantial interest from companies that wish to license the technology.
  • The clinical information system for patients (PatCIS) is an adaptation of WebCIS intended to allow individual patients to access both their own clinical information, as appropriate, and educational materials that their physicians prescribe for their use. Although PatCIS is a grant funded experiment at this time, it shows nicely how the existing clinical systems infrastructure can be leveraged for innovative applications developments that begin as research projects.
  • The immunization registry is another DBMI-developed project that seeks to develop, in collaboration with the Department of Pediatrics, a city-wide registry for childhood immunization data.
  • The clinical cancer system is under development as part of New York City’s AmDEC cancer project, an effort by multiple institutions in the city to monitor city-wide statistic on cancer and its care. The computer-based component of this project was awarded to Columbia and is being carried out by a DBMI faculty member, Professor Stephen Johnson.
  • The IDEATel telemedicine project is funded by a $28 million grant from HCFA to assess the cost effectiveness of telemedicine. The effort is a collaboration between researchers at Columbia and researchers in Central and Upstate New York.

The following are clinical systems not administered by DBMI. The list is large, so it is not possible to mention every fine initiative on this campus. Representative examples are listed:

  • The major hospital ancillary systems, including clinical laboratory, pathology, and radiology, are administered by a hospital unit run by Gil Sofer. The unit is separate from the hospital information services department and from DBMI, although there is constant collaboration. The unit has had great success in implementing these systems.
  • A large number of departmental systems are essential to the operation of this campus. Virtually every clinical department has some initiative in clinical systems. The Department of Surgery has a large group run by Dr. Henry Spotnitz, and Cardiology, Radiology, and Pathology have similar initiatives. One recent success was Radiology’s implementation of a PACS system that replaces film with electronic imaging.
  • A number of research systems also contribute to clinical care. For example, the Cancer Center and the Irving Clinical Research Center have computer units that support active clinical research. The basic sciences departments and centers have strong computing facilities that support basic and clinical research. Included are the Genome Center and the Hughes Center. Similarly, all the schools of the health sciences campus have created initiatives related to clinical care and research. For example, the School of Dental and Oral Surgery’s informatics initiative, run by Dr. John Zimmerman, has national recognition, and the Division of Biostatistics of the Mailman School of Public Health has a strong computing component.
  • Members of Columbia’s practicing faculty have installed clinical practice systems in their own offices. One system that is used in several practices as well as the student health service, Probity, was developed by a Columbia faculty member.
  • There are a number of information systems to support education on the campus, including Dr. Pat Molholt’s Office of Scholarly Resources.
  • While not a clinical system in itself, the IDX outpatient administrative system is critical to support ambulatory care on the campus. The system is administered by the dean’s office and has rationalized billing and helped negotiate fees with outside payers.
  • The networking infrastructure is undergoing major upgrades at the health sciences campus. It is crucial that the infrastructure be upgraded so that current and future applications can rely on sufficient capacity and reliability from the network.

Two major projects are just getting underway, the details of which remain to be defined:

  • The ambulatory record system will support the clinical practices, both uptown at Columbia and downtown at Cornell. The current plan is to implement the EpiCare system (from a commercial vendor), carefully integrating it at Columbia with the clinical database so that WebCIS will be able to access both outpatient and inpatient data and the Epic system will have access to inpatient data as well. The Columbia implementation will be handled as a separate project from the downtown effort since the integration requirements are significantly different on the two campuses.
  • The physician order entry system is long overdue at Columbia and needs to be implemented as soon as resources can be made available and a suitable integration plan defined. Order entry is now in place on many wards at the Cornell campus, but it has been implemented using a commercial vendor and building on a very different computing infrastructure that exists on that campus. Talks are underway to explore how best to leverage the positive experience downtown in constructing a physician-order-entry solution at Columbia that will integrate well with the clinical computing environment uptown.

In summary, the Columbia University Health Sciences campus is clearly a leader in several clinical areas: a comprehensive clinical database, a World Wide Web-based viewer that offers ubiquitous access to clinical information to those who need it yet conforms to upcoming HIPAA regulations, text processing, telemedicine, biomedical informatics research, and dental informatics. The main areas for immediate expansion are comprehensive ambulatory systems and inpatient physician order entry.