Difference between revisions of "Regenstrief Medical Record System (RMRS)"

From Clinfowiki
Jump to: navigation, search
(Rule-based reminders/Decision Support)
(29 intermediate revisions by 4 users not shown)
Line 1: Line 1:
Orginally developed in 1972 by Clem Mcdonald, Regenstrief Medical Record System (RMRS) is a complex system that involves manipulation, transfer, and storage of data in and associated with electronic medical records. It is the first EMR to generate rule-based reminders to physicians about its own content. Using an internally-developed decision support language - CARE, providers could develop protocol-specific reminders to streamline clinical decision making. One study found that physician compliance with preventative care rose to 51% from 22% when RMRS suggestions were used. Physicians did not seem to have learned from the suggestions, because compliance dropped back down to baseline when the suggestions from RMRS were removed.  This finding is particularly important because it indicates that physician knowledge is less important for patient care than providing the physician with the appropriate intellectual artifacts.
+
The '''Regenstrief Medical Record System (RMRS)''' is a comprehensive [[EMR|electronic health record (EHR)]] used at the [[Indiana University Medical Center]] and the surrounding Indianapolis area [http://www.ncbi.nlm.nih.gov/pubmed/10405881/]. The Regenstrief Medical Record System handles laboratory results, orders, medications, registration information, nursing assessments, EKGs and other clinical data.
  
The goal was to:
 
  
* eliminate the logistic problems of the paper record by making clinical data immediately available to authorized users wherever they are—no more unavailable or undecipherable clinical records;
+
== History of RMRS ==
* to reduce the work of clinical book keeping required to manage patients—no more missed diagnoses when laboratory evidence shouts its existence,no more forgetting about required preventive care;
+
* to make the informational ‘gold’ in the medical record accessible to clinical, epidemiologic, outcomes and management research. The system was to complement, not replace, the paper medical record.
+
  
It began in 1972 in a diabetes clinic with only 35 patients.  Rule-based physician reminders were implemented in 1974. Physicians began entry of outpatient test orders in 1984, and in 1990, this was extended to outpatient test orders. In 1994, the Institute extended RMRS to the Indiana Network for Patient Care (INPC) in order to link Indianapolis' five major hospital systems. The majority of all hospital discharge notes have been entered by physicians since 1997.
+
In 1972, [[Regenstrief_Chair_in_Health_Services_Research_--_Indiana_University_School_of_Medicine|Clement J. McDonald]] and [[Charles Clark]] began collecting data on 35 diabetes patients in Marion the County General Hospital. McDonald and Clark built a hard coded program to enter patient data, store it in a data struture, and print flowsheet reports. Difficulties arose because data was not always collected. Clinical personnel often did not mark patient replacement cards for manual data re-entry. Laboratory results from different departments were not shared. The hard coded data structure was also unsustainable, and soon replaced by a data base system.
  
As of 1999, the RMRS carried 200 million separate coded observations,3.25 million narrative reports, 15 million prescriptions
+
== Overview of RMRS ==
and 212,000 electrocardiographic (EKG) tracings. The RMRS carried records for 1.3 million patients in addition to all data generated from several thousand ambulatory and inpatient encounters per year,
+
  
As of now, the RMRS is still operating, which carries 660 million distinct observation. This makes RMRS one of the longest operating EMRs in the world. The RMRS is being accessed more than 10 million times every year by Wishard Health Services and 20 million times per year at Clarian Health in Indianapolis.
+
=== Organizational and geographic scope ===
  
The RMRS has been best studies and it is well known nationally and internationally to be the model for a number of commercial and academic EMR systems.
+
The Regenstrief Medical Record System (RMRS) serves 1.3 million patients, records 200 million separate coded observations, 15 million prescriptions and 212,000 [[electrocardiographic (EKG)]] tracings. 1300 nurses, 1000 physicians, and 220 medical students access patient data over 628,000 times per month. The [[Wishard Health Services]] accesses the RMRS 10 million times per year. The RMRS is used by the Indiana University Medical Center, 40 local clinics, and 30 outreach clinics.  
  
This system serves four hospitals on the Indiana University Medical Center campus and forty local and 30 outreach practices in the city of Indianapolis.
+
=== Mission statement ===
  
The systems success can be attributed to it's strong foundation in 3 areas
+
The RMRS hopes to lead in three areas:
  
* Physician leadership in the informatics effort is vital. It is their intelligence, self-confidence, high energy, and clinical knowledge.   
+
# Physician leadership in the informatics effort. Physicians are intelligent, confident, energetic, and possess important clinical knowledge.   
* Commitment to the mission and vision of high quality and excellence in health care driven and believed in from the highest echelon within the hospital, and
+
# Commitment to high quality health care that is driven by the best in the hospital
* Continuous quality improvement and incorporation of user feedback to guide this improvement.
+
# Continuous quality improvement via user feedback [https://www.himss.org/content/files/davies_1997_regenstrief.pdf]
  
Quickly fixed mistakes are tolerated by Physicians, and this had driven an evolutionary approach.  The software is updated incrementally and feedback from the users is sought early and often.  Feature's are added based upon which is easiest to fix or implement. 
+
=== Goals ===
  
The data repository is the key component of the system. It originally called for no data entry for physicians' observations. Creator Clement McDonald, M.D. and his co-authors note that physicians were reluctant to perform data entry. The system developers first enlisted physicians for data entry with a physician orders system because orders are more easily structured for data input than observations.
+
The goals of the RMRS are to:
  
Significant effort is required to create mechanisms to capture clinical data.  Universal or standardized codes such as LOINC and SNOMED through HL7 interfaces have eased the strain.
+
# Eliminate the logistic problems of a paper recording system. An electronic recording system makes clinical data immediately available to authorized users. There will no longer be lost or misplaced papers, or undecipherable notes.
What data to capture is an important decision. Multiple systems, types of data, formats, and the large number and variety of sources require a great deal of effort and thus the need for planning and prioritizing.
+
# Reduce the manpower needed in clinical bookkeeping and managing patients. An electronic recording system automatically reminds patients of their laboratory results and diagnoses.
 +
# Create informational 'gold' in the medical record. All the information would be accessible to clinical, epidemiological, outcomes and management research.
  
Direct capture by electronic interface based on HL7 can be facilitated from beside electronic instruments, patient registration system,  laboratory, pharmacy, appointment scheduling, dictation/transcription radiology, nurse telephone triage and billing systems.
+
=== The Three-Legged Stool ===
 +
The RMRS is built upon three crucial foundations- each of which may be considered a three-legged stool:[http://himss.files.cms-plus.com/FileDownloads/1997%20OrganizationRegenstrief%20Institute%20for%20Health%20Care.pdf]
  
Clerk data entry is used to code the impressions of most diagnostic reports not already coded by the system.  This allows the system can understand the diagnostic content for patient retrievals and reminders.  They enter standardized phrases and abbreviations instead of numeric codes that would add to the training.  The data entry clerks also enter the number values of predefined questions, such as blood pressure or finger stick glucose.
+
Foundation I: Committed support of the leadership and close collaboration between-
 +
# The Regenstrief Institute
 +
# The Wishard Memorial Hospital
 +
# The Indiana University School of Medicine, with it's faculty and the affiliated clinical practices
  
Physician workstation entry eliminates delays, costs and potential errors due to transcription. . For example, the system will trigger an alert response if a prescription is not covered by the patient’s insurance. In this case, the system suggests a more cost effect alternative.  Direct entry by physicians also validates the person who is most knowledgeable of the information and is in the best position to act on “smart” computer feedback.
+
Foundation II: The tripartite a mission of an academic medical center-
 +
# Teaching
 +
# Research
 +
# Public service
  
Today, most nursing home orders, as well as ED release and inpatient orders are entered directly into the computer system by physicians using the Medical Gopher CPOE workstation. Dictated discharge summaries have largely been replaced by physician-generated discharge notes entered directly into the computer. Similarly, most of the outpatient clinic notes are entered directly into the computer using the Medical Gopher system.
+
Foundation III:
 +
# Careful analysis and improvement of processes
 +
# Technical excellence
 +
# Responsiveness to user need
  
The Gopher workstation also allows the printing of personalized patient information handouts, permits doctors and nurses to communicate via confidential email and can even display satellite weather photos. . The system also allows for printing of hard copy inpatient paper reports, referred to as ‘scut’ cards, that fit neatly into a physician’s lab coat pocket in the form of a small booklet that provides a compact overview of the patient’s state. Physicians can also access past issues of leading medical journals and the American Hospital Formulary drug monographs to research specific topics or learn more about a certain medication. For clinical questions, they can consult the Journal of the American Medical Association, New England Journal of Medicine, Annals of Internal Medicine, the Yearbook of Medicine, NLM’s Medline, and full bibliographic references for any numbered citation on the user interface.
 
  
A 2001 study from Wishard Memorial Hospital demonstrated that computerized reminders significantly improved the use of preventative measures in eligible patients admitted to the hospital. Compared to a control group, those patients whose physicians received reminders received more influenza and pneumococcal vaccinations, subcutaneous heparin prophylaxis and aspirin at the time of discharge. However, even with the reminders, compliance with the recommendations was still far from being universally accepted by the physicians.
+
== Technology ==
  
One of the keys behind the system is that the numerous forms and reports are not pre-generated.  Instead, they have a generalized template ("schema") which deals with high-level components like "notes in this section" and "vitals in this section".  And then based upon the data for a patient, the display is materialized.
+
The RMRS has served as the laboratory for many high quality studies of the application of clinical information systems to hearth care, including a landmark study of the impact of [[CPOE|physician order entry]] within a hospital.
  
The general layout of the system is with the problem list on the upper left side with a physician specified and chosen set of observation variables like vitals on the lower left portion of the screen.  On the right side is a section for notes and below that is where orders are entered and reviewed.  As far as reports, a great deal of them can be created and customized to the encounter.  Such reports can be as wordy or succinct as the doctor desires and ready prior to the patient encounter.  However, despite the utility of the electronic medical record, the paper chart has not been completely eliminated. Although its use is scarce, and use of the paper chart does not include inpatient orders or laboratory tests.
 
  
===Data capture===
+
=== Hardware ===
* [https://www.himss.org/content/files/davies_1997_regenstrief.pdf Data capture] obviously is an important component in any electronic patient record system. Accuracy, security, portability, accessibility are just some of the important but extremely difficult aspects of it that need to be addressed in every design. Add to this the multitude of touch points a patient may interact with in their medical encounter and the many data acquisition devices each department use - ECG prods, MRI, etc. - that all interface and send data to a central repository for storage. Some of the data types and mechanisms used to capture them are shown in this figure:
+
 
 +
The RMRS operates on Digital Equipment’s Alpha computers and a Novell-based PC network. The central system communicates via 10 megabit Ethernet links, 1.6 megabit T1 lines, or 56-Kb modems. The system’s software uses web browser technology that supports voice dictation and voice understanding features. Initial assessments of hospitalized patients are recorded on portable PCs that utilize Proxim’s spread spectrum technology.
 +
 
 +
=== Unique Patient Identification Numbers ===
 +
 
 +
A global patient number identifies patients. This allows the RMRS to identify patients, even when they go to a different doctor or hospital. However, one patient could be given more than one identifier; RMRS has a mechanism to combine the data from two different chart numbers into one, and for reversing when the combining was a mistake. <ref name=”mcdonold 1999”> McDonald, C. J., Overhage, J. M., Tierney, W. M., Dexter, P. R., Martin, D. K., Suico, J. G., . . . Wodniak, C. (1999). The regenstrief medical record system: A quarter century experience. International Journal of Medical Informatics, 54(3), 225-253. http://www.sciencedirect.com/science/article/pii/S138650569900009X</ref>  To account for variations in name, the identifying index uses a sophisticated matching algorithms to determine if registration records from different hospitals are from the same patient. If the registry records match, all medical records from all locations can be produced for one patient. This ‘cross enterprise’ access occurs at a very low level and allows most of the RMRS texts to function in this multiple enterprise environment with minimal modification.
 +
 
 +
=== Data Capture ===
 +
 
 +
RMRS [[Electronic data capture|data capture]] aims to be comprehensive and efficient. The RMRS faces a challenge to capture data from multiple systems, formats, types of data, and from a large number of people. Universal or standardized codes such as [[LOINC]] and [[SNOMED]] through [[HL7]] interfaces have made it easier to capture clinical data.
 +
 
 +
==== Types of data captured ====
 +
 
 +
* Registration information: demographics, insurance information
 +
* Scheduling and visit activity: site, charges and diagnoses
 +
* Outpatient prescription data: prescriptions filled at the Regenstrief pharmacy
 +
* Diagnostic test results: from clinical laboratory, pathology, radiology, diagnostic cardiology (e.g. EKGs)
 +
* Imaging studies, procedures and diagnoses
 +
* Vital signs: blood pressure, pulse, temperature, weight
 +
* Death certificates
 +
* Charges and billing
 +
 
 +
Table 1 shows a sample of the kinds of data captured by the RMRS at the [[Clarian/IU/Riley hospitals]] and [[Wishard hospitals]]. 
 
http://img16.imageshack.us/img16/3250/rmrs.png
 
http://img16.imageshack.us/img16/3250/rmrs.png
  
===Patient Identification===
+
==== Direct capture by electronic interface ====
Since the RMRS has developed the capacity to span multiple enterprises, it is necessary to use a global patient index to identify patients.  This index relies on sophisticated matching algorithms to determine which registration records from the various enterprises represent the same patient.  Once matching registry records are identified, virtual medical records about one patient from many enterprises can be produced at display time. This ‘cross enterprise’ access occurs at a very low level and allows most of the RMRS texts to function in this multiple enterprise environment with minimal modification.
+
  
===Technology===
+
Data capture can be made directly from instruments, such as a bedside measuring devices recording inpatient blood pressure, pulse and temperature, and EKG measurements and tracings. Data can also be collected by three patient registration systems, three laboratories, three pharmacies, two appointment scheduling, two dictation transcriptions, one radiology, one nurse telephone triage and five billing systems. The dictation systems produce an unstructured raw word processing file, which requires an algorithm to find key information, such as patient ID, date and report type. Manual effort must be expended as 1-5% of dictated reports have key data errors.  
The many intertwined local and outreach clinics that comprise the RMRS operate on Digital Equipment’s Alpha computers and a Novell-based PC network. The central system communicates via 10 megabit Ethernet links, 1.6 megabit T1 lines, or 56-Kb modems. The system’s software uses web browser technology that supports voice dictation and voice understanding features. Initial assessments of hospitalized patients are recorded on portable PCs that utilize Proxim’s spread spectrum technology.  
+
  
==References==
+
==== Clerk data entry ====
1. McDonald, C.J., Tierney, W.M., Overhage J.M., et al. (1997). [https://www.himss.org/content/files/davies_1997_regenstrief.pdf The three-legged stool: Regenstrief institute for health care.] In J.M. Teich (Ed.), The third annual Nicholas E. Davis award: Proceedings of the CPR recognition symposium: (pp. 131-158). Shaumburg, IL.
+
  
2. [http://www.medicine.indiana.edu/news_releases/archive_00/regenstrief_mrs_00.html Indiana School of Medicine]
+
Specially trained personnel code and enter most diagnostic reports that are not coded by the source system, such as EEG, EMG, colonoscopy, radiology and endoscopy reports. The coders enter standardized phrases and abbreviations rather than numeric codes, that would add to the training. This allows the system can understand the diagnostic content for patient retrievals and reminders. The computer translates these standardized text phrases into internal codes for storage, and back into human readable language. The data entry clerks also enter the number values of predefined questions, such as blood pressure or finger stick glucose.
  
3. [http://www.regenstrief.org/medinformatics/rmrs RMRS Introduction]
+
==== Physician entry ====
  
4. McDonald CJ. The Regenstrief Medical Record System: a quarter century experience. International Journal of Medical Informatics. 1999. 54(1999)225-53.
+
Originally, the RMRS did not require data entry for physician observations because physicians were reluctant to do so. The system developers first enlisted physicians for data entry with a physician orders system because orders are more easily structured for data input than observations.
  
5. Dexter PR, et al. A Computerized Reminder System to Increase The Use of Preventative Care for Hospitalized Patients. NEJM 2001; 345 (13):965-70.
+
Physicians began entry of outpatient test orders in 1984, and in 1990, this was extended to outpatient test orders.  In 1994, the Institute extended RMRS to the Indiana Network for Patient Care (INPC) in order to link Indianapolis' five major hospital systems.  The majority of all hospital discharge notes have been entered by physicians since 1997.
 +
 
 +
=== Rule-based reminders/Decision Support ===
 +
 
 +
In 1974, the RMRS is the first EHR to implement rule-based reminders to make physicians make better clinical decisions. Reminders can be generated by the system called CARE; a rule based system which can generate reminders or retrieve patient records base on pre-specified criteria. <ref name= "McDonald 1997" McDonald, C.J., Tierney, W.M., Overhage J.M., et al. (1997). [https://www.himss.org/content/files/davies_1997_regenstrief.pdf The three-legged stool: Regenstrief institute for health care.] In J.M. Teich (Ed.), The third annual Nicholas E. Davis award: Proceedings of the CPR recognition symposium: (pp. 131-158). Shaumburg, IL.
 +
 
 +
Developers made protocol-specific reminders to streamline clinical decision making.
 +
 
 +
The general layout of the system is with the problem list on the upper left side with a physician specified and chosen set of observation variables like vitals on the lower left portion of the screen.  On the right side is a section for notes and below that is where orders are entered and reviewed.  As far as reports, a great deal of them can be created and customized to the encounter.  Such reports can be as wordy or succinct as the doctor desires and ready prior to the patient encounter.  However, despite the utility of the electronic medical record, the paper chart has not been completely eliminated. Although its use is scarce, and use of the paper chart does not include inpatient orders or laboratory tests.
 +
 
 +
==== Studies on efficiency of rule based reminds ====
 +
 
 +
One study found that physician compliance with preventative care rose to 51% from 22% when RMRS suggestions were used.  Physicians did not seem to have learned from the suggestions, because compliance dropped back down to baseline when the suggestions from RMRS were removed.  This finding is particularly important because it indicates that physician knowledge is less important for patient care than providing the physician with the appropriate intellectual artifacts.
 +
 
 +
Quickly fixed mistakes are tolerated by Physicians, and this had driven an evolutionary approach.  The software is updated incrementally and feedback from the users is sought early and often.  Feature's are added based upon which is easiest to fix or implement. 
 +
 
 +
A 2001 study from Wishard Memorial Hospital demonstrated that computerized reminders significantly improved the use of preventative measures in eligible patients admitted to the hospital. Compared to a control group, those patients whose physicians received reminders received more influenza and pneumococcal vaccinations, subcutaneous heparin prophylaxis and aspirin at the time of discharge. However, even with the reminders, compliance with the recommendations was still far from being universally accepted by the physicians.
 +
 
 +
One of the keys behind the system is that the numerous forms and reports are not pre-generated.  Instead, they have a generalized template ("schema") which deals with high-level components like "notes in this section" and "vitals in this section".  And then based upon the data for a patient, the display is materialized.
 +
 
 +
=== Gopher System ===
 +
 
 +
Most [[CDS#Order_set|order sets]] are entered directly into the computer system using the Medical Gopher CPOE workstation. Dictated discharge summaries have largely been replaced by physician-generated discharge notes entered directly into the computer. Similarly, most of the outpatient clinic notes are entered directly into the computer using the Medical Gopher system.
 +
 
 +
Physician workstation entry eliminates delays, costs and potential errors due to transcription. For example, the system will trigger an alert response if a prescription is not covered by the patient’s insurance. In this case, the system suggests a more cost effect alternative. Direct entry by physicians also validates the person who is most knowledgeable of the information and is in the best position to act on “smart” computer feedback.
 +
 
 +
The Gopher physician workstation is used in a wide variety of settings by different types of practitioners.  To accommodate the specific documentation needs of these diverse users, Gopher provides many documentation menus tailored for different medical specialties and settings. <ref name=”mcdonold 1999”> McDonald, C. J., Overhage, J. M., Tierney, W. M., Dexter, P. R., Martin, D. K., Suico, J. G., . . . Wodniak, C. (1999). The regenstrief medical record system: A quarter century experience. International Journal of Medical Informatics, 54(3), 225-253. http://www.sciencedirect.com/science/article/pii/S138650569900009X</ref>
 +
 
 +
The Gopher workstation also allows the printing of personalized patient information handouts, permits doctors and nurses to communicate via confidential email and can even display satellite weather photos. . The system also allows for printing of hard copy inpatient paper reports, referred to as ‘scut’ cards, that fit neatly into a physician’s lab coat pocket in the form of a small booklet that provides a compact overview of the patient’s state. Physicians can also access past issues of leading medical journals and the American Hospital Formulary drug monographs to research specific topics or learn more about a certain medication. For clinical questions, they can consult the Journal of the American Medical Association, New England Journal of Medicine, Annals of Internal Medicine, the Yearbook of Medicine, NLM’s Medline, and full bibliographic references for any numbered citation on the user interface.
 +
 
 +
=== Search and retrieval capabilities— cross-patient reports ===
 +
Cross-patient searches can be performed for IRB-approved research and quality management purposes. CARE language can be used to search the entire data base, a subset of the data base, or several institutions’ databases for patients whose EMR contains particular patterns of data. [http://www.ncbi.nlm.nih.gov/pubmed/10405881/]
 +
Fast retrieval is a second way to search the data base. It uses direct indexes by clinical variable (e.g. glucose, chest X-ray) and by results within the medical record files to provide fast access to medical records based on the content of those records. Fast retrieval produces a list of patients who satisfy the search criteria. It can then retrieve specific data elements from each of those patient’s medical records. This functionality lends itself well to retrospective clinical research.
 +
 
 +
=== Future challenges ===
 +
*Complete capture of care provider notes
 +
*Dictation & Voice recognition (eg. Active X component that Dragon System's voice understanding)
 +
*Note capture system from Datamedic
 +
*Wireless provider carried devices
 +
*User identification (eg. Biometric identification)
 +
 
 +
==References==
  
6. Friedlin J., et al. Details of a Successful Clinical Decision Support System. AMIA Annu Symp Proc 2007; 2007:254-258.
+
<references />
  
7. [http://www.regenstrief.org/medinformatics/inpc Indiana Network for Patient Care]
+
# McDonald CJ. The Regenstrief Medical Record System: a quarter century experience. International Journal of Medical Informatics. 1999. 54(1999)225-53 [http://www.ncbi.nlm.nih.gov/pubmed/10405881]
 +
# McDonald, C.J., Tierney, W.M., Overhage J.M., et al. (1997). [https://www.himss.org/content/files/davies_1997_regenstrief.pdf The three-legged stool: Regenstrief institute for health care.] In J.M. Teich (Ed.), The third annual Nicholas E. Davis award: Proceedings of the CPR recognition symposium: (pp. 131-158). Shaumburg, IL.
 +
# McDonald CJ, Tierney WM, Overhage JM, Dexter PR. The Three Legged Stool: Regenstrief Institute for Health Care.  Proc 3rd Annual N.E. Davis CPR Recognition Symp. 1997; 101-147
 +
# [http://www.medicine.indiana.edu/news_releases/archive_00/regenstrief_mrs_00.html Indiana School of Medicine]
 +
# [http://www.regenstrief.org/medinformatics/rmrs RMRS Introduction]
 +
# Dexter PR, et al. A Computerized Reminder System to Increase The Use of Preventative Care for Hospitalized Patients. NEJM 2001; 345 (13):965-70.
 +
# Friedlin J., et al. Details of a Successful Clinical Decision Support System. AMIA Annu Symp Proc 2007; 2007:254-258.
 +
# [http://www.regenstrief.org/medinformatics/inpc Indiana Network for Patient Care]
 +
# Clement J McDonald, William M Tierney, J Marc Overhage, Paul Dexter, Blaine Takesure, Greg Abernathy. The Three-Legged Stool, in : Regenstrief Institute for Health Care
 
   
 
   
 
[[Category: EHR]]
 
[[Category: EHR]]
 
[[Category: UT-SHIS SP09]]
 
[[Category: UT-SHIS SP09]]

Revision as of 18:19, 7 September 2015

The Regenstrief Medical Record System (RMRS) is a comprehensive electronic health record (EHR) used at the Indiana University Medical Center and the surrounding Indianapolis area [1]. The Regenstrief Medical Record System handles laboratory results, orders, medications, registration information, nursing assessments, EKGs and other clinical data.


History of RMRS

In 1972, Clement J. McDonald and Charles Clark began collecting data on 35 diabetes patients in Marion the County General Hospital. McDonald and Clark built a hard coded program to enter patient data, store it in a data struture, and print flowsheet reports. Difficulties arose because data was not always collected. Clinical personnel often did not mark patient replacement cards for manual data re-entry. Laboratory results from different departments were not shared. The hard coded data structure was also unsustainable, and soon replaced by a data base system.

Overview of RMRS

Organizational and geographic scope

The Regenstrief Medical Record System (RMRS) serves 1.3 million patients, records 200 million separate coded observations, 15 million prescriptions and 212,000 electrocardiographic (EKG) tracings. 1300 nurses, 1000 physicians, and 220 medical students access patient data over 628,000 times per month. The Wishard Health Services accesses the RMRS 10 million times per year. The RMRS is used by the Indiana University Medical Center, 40 local clinics, and 30 outreach clinics.

Mission statement

The RMRS hopes to lead in three areas:

  1. Physician leadership in the informatics effort. Physicians are intelligent, confident, energetic, and possess important clinical knowledge.
  2. Commitment to high quality health care that is driven by the best in the hospital
  3. Continuous quality improvement via user feedback [2]

Goals

The goals of the RMRS are to:

  1. Eliminate the logistic problems of a paper recording system. An electronic recording system makes clinical data immediately available to authorized users. There will no longer be lost or misplaced papers, or undecipherable notes.
  2. Reduce the manpower needed in clinical bookkeeping and managing patients. An electronic recording system automatically reminds patients of their laboratory results and diagnoses.
  3. Create informational 'gold' in the medical record. All the information would be accessible to clinical, epidemiological, outcomes and management research.

The Three-Legged Stool

The RMRS is built upon three crucial foundations- each of which may be considered a three-legged stool:[3]

Foundation I: Committed support of the leadership and close collaboration between-

  1. The Regenstrief Institute
  2. The Wishard Memorial Hospital
  3. The Indiana University School of Medicine, with it's faculty and the affiliated clinical practices

Foundation II: The tripartite a mission of an academic medical center-

  1. Teaching
  2. Research
  3. Public service

Foundation III:

  1. Careful analysis and improvement of processes
  2. Technical excellence
  3. Responsiveness to user need


Technology

The RMRS has served as the laboratory for many high quality studies of the application of clinical information systems to hearth care, including a landmark study of the impact of physician order entry within a hospital.


Hardware

The RMRS operates on Digital Equipment’s Alpha computers and a Novell-based PC network. The central system communicates via 10 megabit Ethernet links, 1.6 megabit T1 lines, or 56-Kb modems. The system’s software uses web browser technology that supports voice dictation and voice understanding features. Initial assessments of hospitalized patients are recorded on portable PCs that utilize Proxim’s spread spectrum technology.

Unique Patient Identification Numbers

A global patient number identifies patients. This allows the RMRS to identify patients, even when they go to a different doctor or hospital. However, one patient could be given more than one identifier; RMRS has a mechanism to combine the data from two different chart numbers into one, and for reversing when the combining was a mistake. [1] To account for variations in name, the identifying index uses a sophisticated matching algorithms to determine if registration records from different hospitals are from the same patient. If the registry records match, all medical records from all locations can be produced for one patient. This ‘cross enterprise’ access occurs at a very low level and allows most of the RMRS texts to function in this multiple enterprise environment with minimal modification.

Data Capture

RMRS data capture aims to be comprehensive and efficient. The RMRS faces a challenge to capture data from multiple systems, formats, types of data, and from a large number of people. Universal or standardized codes such as LOINC and SNOMED through HL7 interfaces have made it easier to capture clinical data.

Types of data captured

  • Registration information: demographics, insurance information
  • Scheduling and visit activity: site, charges and diagnoses
  • Outpatient prescription data: prescriptions filled at the Regenstrief pharmacy
  • Diagnostic test results: from clinical laboratory, pathology, radiology, diagnostic cardiology (e.g. EKGs)
  • Imaging studies, procedures and diagnoses
  • Vital signs: blood pressure, pulse, temperature, weight
  • Death certificates
  • Charges and billing

Table 1 shows a sample of the kinds of data captured by the RMRS at the Clarian/IU/Riley hospitals and Wishard hospitals. http://img16.imageshack.us/img16/3250/rmrs.png

Direct capture by electronic interface

Data capture can be made directly from instruments, such as a bedside measuring devices recording inpatient blood pressure, pulse and temperature, and EKG measurements and tracings. Data can also be collected by three patient registration systems, three laboratories, three pharmacies, two appointment scheduling, two dictation transcriptions, one radiology, one nurse telephone triage and five billing systems. The dictation systems produce an unstructured raw word processing file, which requires an algorithm to find key information, such as patient ID, date and report type. Manual effort must be expended as 1-5% of dictated reports have key data errors.

Clerk data entry

Specially trained personnel code and enter most diagnostic reports that are not coded by the source system, such as EEG, EMG, colonoscopy, radiology and endoscopy reports. The coders enter standardized phrases and abbreviations rather than numeric codes, that would add to the training. This allows the system can understand the diagnostic content for patient retrievals and reminders. The computer translates these standardized text phrases into internal codes for storage, and back into human readable language. The data entry clerks also enter the number values of predefined questions, such as blood pressure or finger stick glucose.

Physician entry

Originally, the RMRS did not require data entry for physician observations because physicians were reluctant to do so. The system developers first enlisted physicians for data entry with a physician orders system because orders are more easily structured for data input than observations.

Physicians began entry of outpatient test orders in 1984, and in 1990, this was extended to outpatient test orders. In 1994, the Institute extended RMRS to the Indiana Network for Patient Care (INPC) in order to link Indianapolis' five major hospital systems. The majority of all hospital discharge notes have been entered by physicians since 1997.

Rule-based reminders/Decision Support

In 1974, the RMRS is the first EHR to implement rule-based reminders to make physicians make better clinical decisions. Reminders can be generated by the system called CARE; a rule based system which can generate reminders or retrieve patient records base on pre-specified criteria. [2]

The Gopher workstation also allows the printing of personalized patient information handouts, permits doctors and nurses to communicate via confidential email and can even display satellite weather photos. . The system also allows for printing of hard copy inpatient paper reports, referred to as ‘scut’ cards, that fit neatly into a physician’s lab coat pocket in the form of a small booklet that provides a compact overview of the patient’s state. Physicians can also access past issues of leading medical journals and the American Hospital Formulary drug monographs to research specific topics or learn more about a certain medication. For clinical questions, they can consult the Journal of the American Medical Association, New England Journal of Medicine, Annals of Internal Medicine, the Yearbook of Medicine, NLM’s Medline, and full bibliographic references for any numbered citation on the user interface.

Search and retrieval capabilities— cross-patient reports

Cross-patient searches can be performed for IRB-approved research and quality management purposes. CARE language can be used to search the entire data base, a subset of the data base, or several institutions’ databases for patients whose EMR contains particular patterns of data. [4] Fast retrieval is a second way to search the data base. It uses direct indexes by clinical variable (e.g. glucose, chest X-ray) and by results within the medical record files to provide fast access to medical records based on the content of those records. Fast retrieval produces a list of patients who satisfy the search criteria. It can then retrieve specific data elements from each of those patient’s medical records. This functionality lends itself well to retrospective clinical research.

Future challenges

  • Complete capture of care provider notes
  • Dictation & Voice recognition (eg. Active X component that Dragon System's voice understanding)
  • Note capture system from Datamedic
  • Wireless provider carried devices
  • User identification (eg. Biometric identification)

References

  1. McDonald, C. J., Overhage, J. M., Tierney, W. M., Dexter, P. R., Martin, D. K., Suico, J. G., . . . Wodniak, C. (1999). The regenstrief medical record system: A quarter century experience. International Journal of Medical Informatics, 54(3), 225-253. http://www.sciencedirect.com/science/article/pii/S138650569900009X
  2. McDonald, C. J., Overhage, J. M., Tierney, W. M., Dexter, P. R., Martin, D. K., Suico, J. G., . . . Wodniak, C. (1999). The regenstrief medical record system: A quarter century experience. International Journal of Medical Informatics, 54(3), 225-253. http://www.sciencedirect.com/science/article/pii/S138650569900009X
  1. McDonald CJ. The Regenstrief Medical Record System: a quarter century experience. International Journal of Medical Informatics. 1999. 54(1999)225-53 [5]
  2. McDonald, C.J., Tierney, W.M., Overhage J.M., et al. (1997). The three-legged stool: Regenstrief institute for health care. In J.M. Teich (Ed.), The third annual Nicholas E. Davis award: Proceedings of the CPR recognition symposium: (pp. 131-158). Shaumburg, IL.
  3. McDonald CJ, Tierney WM, Overhage JM, Dexter PR. The Three Legged Stool: Regenstrief Institute for Health Care. Proc 3rd Annual N.E. Davis CPR Recognition Symp. 1997; 101-147
  4. Indiana School of Medicine
  5. RMRS Introduction
  6. Dexter PR, et al. A Computerized Reminder System to Increase The Use of Preventative Care for Hospitalized Patients. NEJM 2001; 345 (13):965-70.
  7. Friedlin J., et al. Details of a Successful Clinical Decision Support System. AMIA Annu Symp Proc 2007; 2007:254-258.
  8. Indiana Network for Patient Care
  9. Clement J McDonald, William M Tierney, J Marc Overhage, Paul Dexter, Blaine Takesure, Greg Abernathy. The Three-Legged Stool, in : Regenstrief Institute for Health Care