Breaking News
Home / Medicine & Technology / Patient data warehousing on the Web

Patient data warehousing on the Web

By Joseph Warnicki

This paper describes an active database system which has been developed to meet the needs for photographic and diagnostic records as part of practice which includes ophthalmic care, education and research. This system embodies a number of strategies to maximize access to diverse clinical information while preserving patient confidentiality. It is available for demonstration to everyone with an Internet Web browser. In several places in the text there are Internet URL addresses that reference relative pages of the database.

The ODIE–Ophthalmic Data & Imaging Engine–is a system of collecting data from many sources and storing and displaying the data in common formats. This data is accessible by anyone with the proper passwords in the form of World Wide Web pages viewable by a web browser. Using data and images in electronic form reduces the number of photographic prints we need to produce and improves access of these images. Though there are costs in time and labor to converting information into the database, it eliminates the costs and time delays associated with records retrieval and filing, particularly when clinical information is generated from and required at multiple sites. ODIE offers improved file retrieval, because the chart can be viewed by several users at once and will never be misplaced. The system offers immediate access, and its ease of operation encourages its use.

We wanted a system that would supplement our existing record storage system. It was essential that this system be able to incorporate newer forms of digital data and be capable of handling older forms of images and records that are currently being stored as hard copy.

Design considerations

A system that could store and distribute the typical information used in the Ophthalmic department in an easily understood and consistent manner. This data can include: Fluorescein and ICG angiographies, Fundus photos, slit-lamp photos, endothelial cell images, x-rays, Cat scans, corneal topography, images of automatic and manual visual fields, typed word processing documents such as angiogram readings, and scanned images of chart pages with hand written notes or drawings. http://visualeyes.upmc.edu/odie/johndoem.html

Automatic methods of transfer of this data so information could be integrated into the system in a timely and economical fashion. This is particularly critical as we convert from older forms of photography, diagnostic testing and clinical records

A secure system that could handle the needs of the physicians and staff in a teaching environment without compromising patient confidentiality. Unlike a private practice, our clinical database serves as both a teaching resource for residents as well as material for clinical research. The system needed to be able to provide levels of access depending on the responsibilities of the client.

A system that could work with existing hardware currently being used by our secretaries and physicians.

Solution

We decided to use a WWW local Internet, or Intranet, system to distribute and display the information. This solved our problems and afforded us additional benefits:

A very quick learning curve since most users already know how to “surf” the web.

Since a web browser is being used as the client software, there is usually no charge for the software. Netscape, Explorer, and most of the other browsers all work well on the web server and some of the best are free.

Review stations could be either Mac or PC based, since operationally the browsers provide uniform access.

Common data formats

Web servers can distribute virtually any type of data that a database may store. However, it is quite another problem for the viewers web browser to interpret all of the different types of data correctly. The two browsers that the system was designed around, Netscape’s Navigator and Microsoft’s Internet Explorer read quite a large variety of file types but extra “plug-in” and “helper” programs would have to be added to encompass all the file types used for medical records. To avoid this complexity in the system, i.e., difficult installations of software and possible “in-the-field” upgrades, all the data was converted into standard formats common to both browsers. This enables any viewer using a version 2.0 or 3.0 browser to view all the data directly without additional program installations.

Automatic conversions

Each equipment manufacture stores information and images in different forms. It was not possible to have any agreement of format between them. It was found to be easier to develop a protocol that was independent from all the systems and convert all data to this new format. Currently, each vendor that has been asked to help program an interface has agreed to work with our project. The largest amount of data and compatibility problems involved our digital imaging system, so they were approached first. The resulting program works well. Single or multiple tests can be converted into our database and stored with only a few keystrokes for the entire data set. The time it takes to transfer an image and convert it to the standard format is 15 seconds. A typical fluorescein angiogram study compromising 20 images takes 5 minutes to be transferred and accessible on the Web database.

Security

Security is a major concern for every database that contains confidential data and that is accessible through the Internet. Data must be maintained in a manner that is not subject to tampering or modification. Security breaches have to be checked from two possible areas. Viewers without any authorization level have to be restricted to the guest areas, demonstration areas and web pages explaining the operation of the system. Secondly, authorized viewers need to be restricted to viewing only the patient information and images to which they need access. Security is maintained by: establishing the level of access that a specific client requires, masking the patient’s identifiers for records that are viewed by clients who are not the eyecare providers for that patient, limiting the level of resolution of images that can be accessed by the browser to prevent “pirating” of photographic records and monitoring the access of all clients to individual records. Thus, the system can establish if and how often a client has accessed a patient’s records at any time. The database records all the physicians each patient has seen and restricts access to all other viewers. Individual patient logs can be edited to add or remove physicians as necessary. Access can be given to physicians even across the country if a patient requests the files be made available for a second opinion.

System Description

Database. The server is a combination of an Internet web server and a traditional database. The data is presently being kept on a Microsoft Foxpro database. Future plans will include switching to an SQL database to better handle the client/server environment.

The patients’ demographic information is stored in the database. There are supplemental databases linked to the primary data. These are:

A list of all referring physicians with multiple areas to allow for their satellite offices.

A link to reference which physicians have seen each patient. This is how the database checks access rights to each patient file.

A day log that shows the patients seen, by whom, and a short diagnosis.

Main diagnostic area. This is a free form searchable area that stores Dx and comments. The diagnostic area can also be used in a formatted manner or to store ICD-9 codes. There is an attached spell checker with an ophthalmic dictionary.

Server. The database server is a Windows NT 3.51 system running Website 1.1 web server software with custom “C” and Visual Basic programs to interface the database with the web server. The computer uses 18 CD-Rom drives in a juke box to store patient data. Copies of the newest and most recently reviewed patient records are stored in temporary hard disk cache files for faster access. Additionally, a subset of all the patients’ data is stored on the hard disk to speed access while the CD-Rom is loading additional data.

Software. After the user logs in, patient data can be accessed in several ways according to assigned access privileges.

Most recently tested patients.

Alphabetic name searches. Only your patients will be listed. If your access includes teaching you can search the full database. The color of the access icons on the name list indicate which patients are yours and which are teaching cases. A record is kept of all teaching accesses to check for possible misuse of patient information.

Diagnostic searches can be made for words or codes used in the patient Dx area. Boolean search parameters (and, or, not) can be employed. The resulting list will show all patients with that Dx but only yours will have names and identifying information. Again, physicians with teaching access can override this restriction, but a record is kept.

Physicians who read the FA or ICG angiograms have access to the files of recently tested patients. This access is removed when the reading is completed.

Capacity. Presently, the demographic and diagnostic data for our 43,000 patients and 120,000 visits occupies 60 meg. of space on the hard disk. We do not foresee any space limitations for growth in this area. Images are recorded onto two separate CD-Roms. One CD-Rom stores several resolutions of each image in “jpeg” compressed format. This disk is made for the juke box. The compressed images increase capacity and transfer rates. An additional CD-Rom is recorded in original, lossless format that is kept off-line for security. Capacity of on-line data can be a mix of:

50,000 images or about 2,500 fluorescein angiograms.

75,000 scanned chart pages or documents

1,500,000 word processing pages

DEMONSTRATION AREA – http://visualeyes.upmc.edu/odie/

The system is designed for secure use by the department, however a demonstration area has been added to show the system. There is a full database with fictitious names to enable you to operate with the exact software that is being used for our patient data. Everything is functional with these exceptions.

Real patient data and diagnostic information was used, however the identifiers have been erased and replaced using a list of common names.

Currently, we have 500 sets of patient tests on-line. This is approximately 8,000 images. ODIE displays the actual proofsheet (thumbnail) images for that patient. Normally from this screen you could request medium (300 line), or high (1000 line) resolution images for viewing by clicking the low or high portion of the image you want. For the demonstration, ODIE substitutes a test image of equal size and resolution. These test images will still allow the viewer to evaluate the detail of higher resolution images while allowing us to restrict access to our images and to reduce usage of the CD-Rom juke box. Should a client require a photographic reproduction of a particular patient record, this can be done by a formal request to the Photography division which maintains a maximum resolution copy that is not accessible through ODIE.

ODIE, while still in its infancy, offers a unique approach to clinical information in a fashion that preserves confidentiality and yet makes vast amounts of clinical material available for teaching and research. As physicians and institutions become increasingly decentralized, the access to a wide range of data from any potential site will be critical for providing rapid, integrated and effective care. Such a database eliminates the physical barriers among the sites of diagnostic testing, of the primary eyecare provider and of the university subspecialist who must all share in the care of patients. To our knowledge, ODIE is the first such database to specifically address the needs of an academic eyecare program.

Joseph Warnicki is a clinical instructor at the University of Pittsburgh, and director of ophthalmic photography, University of Pittsburgh Medical Center. He can be reached at jwar@vision.eei.upmc.edu.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.