Fung V, Cheung NT, Szeto K, Ngai L, Lau M, Kong JHB
Hospital Authority, Hong Kong Special Administrative Region
The use of standard clinical vocabulary to support the development of clinical information systems is well established. The ICD-9-CM was adopted locally to form the basis of the Hospital Authority Master Disease Code Table (HAMDCT) and incorporated for use into the Clinical Management System at all Hong Kong public hospitals since 1996. The code-based HAMDCT was migrated to a term-based vocabulary table: Hospital Authority Clinical Vocabulary Table (HACVT), in order to enhance communications between various clinical information systems. An ongoing maintenance process with the establishment of a formal maintenance network was developed to ensure that the HAMDCT and subsequent HACVT are able to meet the clinician's need. As clinical information systems advance, future development will be focused on knowledge-based terminology, terminology tools, and the uses of HACVT.
Terminology is considered the second most critical issue (after confidentiality) to be addressed in the development of healthcare information today.1 There is no doubt to the importance of adopting a standard clinical vocabulary to ensure data consistency, improve data quality, and also enhance the workflow in healthcare delivery. The need for a standard clinical vocabulary to facilitate the development of various health informatics applications, and ultimately, improving the efficiency and effectiveness of the healthcare services is well recognized.2
Standard clinical vocabulary should contain unambiguous terms that are uniquely identified with appropriate synonyms and retained permanently.3, 4 There are three major issues to be addressed in the development of terminology to support electronic medical records: facilitation of direct data entry, unambiguous understandability of data, and improvement of data presentation.5 The following paper describes the experience that we have in developing and maintaining a standard clinical vocabulary table, which is currently used at all Hong Kong public hospitals' clinical information systems.
The Hospital Authority Master Disease Code Table
Development of HAMDCT
When the Clinical Management System (CMS) was first implemented in 1996, clinicians started to enter data into the computer using the Hospital Authority Master Disease Code Table (HAMDCT).6 All of the 39 public hospitals under the Hospital Authority's management were using the same HAMDCT to report diagnoses and procedures into various clinical information systems.
The HAMDCT was derived from the International Classification of Diseases, 9th revision, Clinical Modification (ICD-9-CM). The limitation of ICD-9-CM, a statistical language to represent clinical content7, 8 was addressed by adding local terms. Clinicians of various specialties identified clinical terms that were commonly used in their daily practice. These terms were added to the HAMDCT as an extension to the ICD-9-CM code. Extensions were also added so as to incorporate coding rules in the HAMDCT. Aliases were introduced to facilitate searching for various terms. All added terms were coded to ICD-9-CM and double checked by the Health Information and Records Managers (HIRMs). To start with, around 4,300 diagnoses terms were added above the original 12,653 diagnosis terms, and nearly 1,600 procedure terms were added to the original 3,654 procedure terms.
Using the HAMDCT
Clinicians using the CMS can search for specific term using keyword, truncated word, ICD code, or their own preference list. When clinicians were unable to locate their required term, an option was provided for the hospital to use a dummy code to temporarily capture the data, which would be subsequently screened by the HIRMs for appropriate follow-up actions.
Specific guidelines for the clinicians on how to use the HAMDCT to report diagnoses and procedures were developed. At the hospital level, the HIRMs provide briefing to clinicians as part of the CMS orientation program. A HAMDCT Web site was set up so that hospital users could inquire about terms being included in the HAMDCT and also other related information--for example, guidelines on reporting diagnoses and procedures.
Maintenance of HAMDCT
An audit of the HAMDCT was conducted in 1999 aiming to ensure the quality of the table. In total, 3569 added terms were reviewed, including terms under codes that have more than 5 extensions, same description with different codes, and terms on the clinician's preference list. A trained coder rechecked the descriptions and the assigned codes. The results identified that there were approximately 2 percent of audited terms with code errors and 3 percent of the sample with ambiguous descriptions.
A manual maintenance mechanism was set up to update the table, aiming to bridge the gap between clinician terms and those in the HAMDCT. Requests from clinicians and HIRMs at the hospital level was forwarded to the central HAMDCT administrator who was trained in clinical classification. Updates on the HAMDCT were sent to the technical team in table file format (Excel) on a monthly basis. Amendment history to the HAMDCT is kept in a separate file, and users are able to check this through the HAMDCT Web site.
The Hospital Authority Clinical Vocabulary Table
Aims of HACVT
The requirement of a term-based vocabulary instead of a code-based one became more significant when the Hospital Authority started reporting clinical data in ICD-10 to the Department of Health in 2001. There was concern that when the coding system was changed, it would increase the workload in retrieving previously recorded data for subsequent data analysis. In addition, more clinical modules serving various disciplines had been developed in CMS over the years and their terms were not included in the HAMDCT, thus, making it difficult to analyze data across disciplines.
In view of the above, the HAMDCT was migrated to the Hospital Authority Clinical Vocabulary Table (HACVT) with the following aims.
- Adopt standardized clinical terms to:
- Assist in developing clinical systems to standardize and expedite clinical documentation
- Facilitate interpretation by clinicians across disciplines
- Serve as a reference for future data analysis
- Allow extension of the vocabulary to domains other than diseases and symptoms (for example, outcomes)
- Match the locally used clinical terms with the international classification systems so that the collected data can be reported and compared internationally using any of the supported systems
- Avoid loss of data due to mapping between coding systems by removing the dependency of the HA clinical vocabulary upon any given coding system
Characteristics of HACVT
A unique term identifier is added to each HACVT term. Each term is coded to the relevant classification systems. This avoids loss of data due to change in version of the coding system. At present, diagnoses are coded to ICD-9-CM and ICD-10. Procedures are coded to ICD-9-CM.
With its origin from the ICD system, the HAMDCT inherited nonspecific terms being used in the statistical classification system--for example, not elsewhere classified, not otherwise specified, unspecified, etc. A cleaning exercise was performed to remove terms with ambiguous descriptions. Duplicated terms, as a result of the cleaning exercise, were inactivated. Although clinicians can no longer choose these inactivated terms when recording patient data, they can still use these terms when performing data analysis on the already recorded data.
The HACVT has also incorporated additional information that is related to a specific term, for example, terms that are being used in Clinical Data Framework (a mechanism collecting specific information relating to a disease in a structured format).9 This will alert the Hospital Disease Classification Coordinators (HDCC)s and HACVT administrators if there is any amendment to the term.
At the CMS front end, the searching mechanism was improved with term(s) that exactly matched the input by the clinician ranked at the top of the returned list.
HACVT Maintenance Network
To help clinicians report problems when they are using the HACVT, a formal network on HACVT maintenance was established. Each hospital nominates a Hospital Disease Classification Coordinator (HDCC) to serve as a bridge between the central HACVT administration team and the clinicians. The CMS was modified to collect clinicians' feedback via email to the HDCC. The HDCC is responsible for screening the problems and identifying the actions to be taken.
The table is still updated on a monthly basis, except in urgent situations. During the SARS (severe acute respiratory syndrome) outbreak in 2003, this term was added to the table within 24 hours to help the frontline clinicians report this disease.
A new HACVT Web site has developed, which in addition to the query function in HAMDCT, allows the users to submit requests for updating the table. Users can also view the related amendment history of individual terms via the Web site. Requests for updating the HACVT can be entered directly into the HACVT maintenance Web site. The maintenance Web site provides additional checking rules so that the HDCC and HACVT administrators can detect related issues when a request is submitted, for example, alert users if the newly created term duplicates an existing one. This reduces the turnaround time in the maintenance process. The central HACVT administration team receives a monthly report on the feedback. Where required, the central team will discuss with the HDCC issues raised by the clinicians. All requests are reviewed by two independent coders before being accepted for updating.
Guidelines for maintaining the HACVT are being developed. Apart from the basic introduction on various data being kept in the HACVT, the guidelines also stress the points to be considered when updating the HACVT under various scenarios--creating new terms, terms with error codes, ambiguous terms. All HDCCs will be briefed on the principles in HACVT maintenance.
In line with the development of healthcare, the complexity of service delivery challenges to health informaticians has increased significantly. Further development of the HACVT will focus on three areas: the HACVT content, the HACVT maintenance tool, and how the HACVT can improve clinical care.
As the clinical vocabulary standard in the Hospital Authority, the HACVT should incorporate various facets of terminology that are captured during the patient care process--rehabilitation outcome. Previous work has indicated the advantages of incorporating a knowledge-based terminology for the delivery of patient care and analysis of patient data for healthcare management.10 To move further to an intelligent system, the existing HACVT must migrate to a knowledge-based vocabulary table to support various engines that are required to build the future clinical information system in HA. Nevertheless, this would be considered as the reference terminology for the HACVT in order to maintain the responsiveness in updating the table.
Moreover, this would introduce greater challenges to the HACVT administrators in managing the table as it also involves maintaining the semantic network.11 The existing HACVT maintenance tool needs to be enhanced to incorporate such requirements.
More detailed discussion on the use of HACVT (implementation of the master problem list, implementation of information architecture) is required so that it can be incorporated into the clinical workflow to support the development of the electronic patient record.12
The HACVT has evolved with the development of our clinical information systems. The experience gained in the past few years has provided a solid foundation for the HACVT administrators to further their work in supporting the development of the next generation of clinical information systems. We will focus on both the development and maintenance processes so that the Hospital Authority Clinical Vocabulary Table can continuously keep abreast of the advancement of medical science.
The authors thank all HIRMs for their dedicated work in developing the HAMDCT and all HDCCs for their support in HACVT maintenance. The HAMDCT and HACVT would not be workable without the contribution of the following colleagues: Kelvin Law, Wilson Yu, Anthony Yu, Stephen Leung, and William Ho.
- Chute CG. Medical concept representation: from classification to understanding. 2002; lecture notes: AMIA Symposium 2003.
- Chute CG. A framework for a comprehensive health terminology system in the United States: development guidelines, criteria for selection, and public policy implications. JAMIA. 1998;5:503-510.
- Cimino JJ. Desiderata for controlled medical vocabularies in the twenty-first century. Methods Inf Med. 1998;37(4-5):394-403.
- Rector AL. Clinical terminology: why is it so hard? Methods Inf Med. 1999;38:239-252.
- Tange HJ, Hasman A, de Vries Robbe PF, Schouten HC. Medical narratives in electronic medical records. International Journal of Medical Informatics. 1997;46:7-29.
- Szeto K, Chu YC, Chong LC, Chiu HM, Chow MK, Chow YY, Chisholm K, Fung V. (1997). Development of Hospital Authority master disease code table In Hong Kong. MIC '97 Proceedings of the Fourth Hong Kong (Asia Pacific) Medical Informatics Conference, Hong Kong. pp.153-157.
- Chon S P, Chute GC. Clinical terminologies and computer-based patient records. Journal of AHIMA. 1997;68(2):41-43.
- Chute CG, Cohn SP, Campbell KE, Oliver DE, and Campbell JR. The content coverage of clinical classifications. JAMIA. 1996;3:224-33.
- Cheung NT, Fung V, Chow YY, Tung Y. Structured data entry of clinical information for documentation and data collection. 2001; Medinfo. 10(1):609-13.
- Cimino, JJ. From data to knowledge through concept-oriented terminologies: experience with the Medical Entities Dictionary. JAMIA. 2000;7(3):288-297.
- Rodriguez J, Maojo V, Crespo J, Fernandez I. A concept model for the automatic maintenance of controlled medical vocabularies. IMIA. 1998:618-622.
- Cheung NT, Fung V, Kong J. The Hong Kong Hospital Authority's Information Architecture. Medinfo. 2004. (forthcoming)
|Source: 2004 IFHRO Congress & AHIMA Convention Proceedings, October 2004|